Datasets:
link
stringlengths 41
45
| date
stringlengths 9
9
| paper
dict | reviews
listlengths 1
6
| version
int64 1
5
| main
stringlengths 38
42
|
|---|---|---|---|---|---|
https://f1000research.com/articles/11-565/v1
|
24 May 22
|
{
"type": "Systematic Review",
"title": "Assessing the role of vascular risk factors in dementia: Mendelian randomization meta-analysis and comparison with observational estimates",
"authors": [
"Liam Lee",
"Rosie Walker",
"William Whiteley",
"Rosie Walker",
"William Whiteley"
],
"abstract": "Background: Although observational studies demonstrate that higher levels of vascular risk factors are associated with an increased risk of dementia, these associations might be explained by confounding or other biases. Mendelian randomization (MR) uses genetic instruments to test causal relationships in observational data. We sought to determine if genetically predicted modifiable risk factors (type 2 diabetes mellitus, low density lipoprotein cholesterol, high density lipoprotein cholesterol, total cholesterol, triglycerides, systolic blood pressure, diastolic blood pressure, body mass index, and circulating glucose) are associated with dementia by meta-analysing published MR studies. Secondary objectives were to identify heterogeneity in effect estimates across primary MR studies and to compare meta-analysis results with observational studies. Methods: MR studies were identified by systematic search of Web of Science, OVID and Scopus. We selected primary MR studies investigating the modifiable risk factors of interest. Only one study from each cohort per risk factor was included. A quality assessment tool was developed to primarily assess the three assumptions of MR for each MR study. Data were extracted on study characteristics, exposure and outcome, effect estimates per unit increase, and measures of variation. Effect estimates were pooled to generate an overall estimate, I2 and Cochrane Q values using fixed-effect model. Results: We screened 5211 studies and included 12 primary MR studies after applying inclusion and exclusion criteria. Higher genetically predicted body mass index was associated with a higher odds of dementia (OR 1.03 [1.01, 1.05] per 5 kg/m2 increase, one study, p=0.00285). Fewer hypothesized vascular risk factors were supported by estimates from MR studies than estimates from meta-analyses of observational studies.\n\nConclusion: Genetically predicted body mass index was associated with an increase in risk of dementia.",
"keywords": [
"Dementia",
"Alzheimer's Disease",
"Mendelian Randomization",
"vascular risk factors",
"meta-analysis",
"systematic review"
],
"content": "Introduction\n\nHigher measured mid-life blood pressure, mid-life and late-life hyperlipidaemia, mid-life obesity and diabetes are associated with the development of dementia in observational cohort studies.1 If this association is causal, risk factors for vascular diseases (e.g. diabetes and hypertension) may be responsible for around 40% of the cases of dementia worldwide.2 However, the effect sizes seen in observational cohort studies are usually larger than those seen in randomized trials of vascular risk intervention to reduce cognitive decline or dementia.3 This may be because cohort studies are limited by residual confounding, reverse causation, differential loss to follow-up, or selection biases.4,5 More data from less biased designs are needed to triangulate the causal effects of vascular risk factors on the development of dementia.\n\nMendelian randomization (MR) uses genetic variants as proxies, or instrumental variables (IVs), to estimate a causal effect of an exposure on an outcome. MR is less susceptible to confounding and reverse causation than observational studies because genetic variants are assumed to be randomly assigned at meiosis.6 As such, MR can be thought of as a “natural” randomized control trial. MR studies are subject to different biases from observational studies. Their validity rests on three key assumptions: (i) the genetic variant has a known association with the risk factor of interest; (ii) the genetic variant is not associated with a known confounder; and (iii) the genetic variant affects the outcome only through the risk factor of interest. As most genetic instrumental variables are only modestly associated with their exposures of interest, MR gives an unbiased but imprecise estimate.7 Meta-analysis of MR studies could mitigate this imprecision. This approach has previously refined estimates of the effect obesity on vascular diseases.8\n\nIn this study, we meta-analysed MR studies of the association of modifiable vascular risk factors with dementia. Secondly, we estimated the heterogeneity between estimates from different MR studies for a given risk factor that used the same outcome cohort. Thirdly, we compared our meta-analysis estimates with estimates obtained from meta-analysis of observational studies.\n\n\nMethods\n\nWe used the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline to report this study.9 A protocol has been developed and made available online (Supplementary Material 3).10 Minor amendments in study methodology have been made since its publication. Specifically, 1) the quality assessment questionnaire was shortened from 11 questions to 10 questions; 2) we applied a Bonferroni correction to account for the assessment of multiple risk factors; 3) MR meta-analysis results were compared with observational studies. No ethical approval was required.\n\nWe searched on OVID, Scopus and Web of Science, covering 13 databases: Medline, Embase, AMED, PsycINFO, BIOSIS Citation Index, Web of Science core collection, Current Contents Connect, Data Citation Index, Derwent Innovations Index, KCI-Korean Journal Database, Russian Science Citation Index, SciELO Citation Index, and Zoological Record. The search looked for a combination of dementia, and Mendelian randomization (Supplementary Material 1, Table 1). No risk factors were specified in the search query. We forward searched by screening for all referenced articles in the retrieved articles using Google Scholar. The final search was performed on 22nd October 2021.\n\nWe included published or pre-print studies that used inverse-variance weighted (IVW) two-sample MR with a poly- or oligo-genetic instrument for type 2 diabetes mellitus, low density lipoprotein (LDL) cholesterol, high density lipoprotein (HDL) cholesterol, total cholesterol, triglycerides, systolic blood pressure, diastolic blood pressure, body mass index (BMI), or plasma glucose. Included studies reported a causal estimate value with an odds ratio (OR), hazard ratio (HR), risk ratio (RR) or β-coefficient by an absolute value of per unit increase, and associated 95% confidence interval (CI) or standard error. When interquartile range was reported, we estimated the standard deviation as interquartile range/1.35.11\n\nWe excluded studies that were duplicates, not written in English, or where no full text was available. We included only one estimate from each cohort per risk factor. Where more than one study had been carried out using a cohort, we included the highest quality study; where the quality of the studies was similar, we included the study with the most recent exposure GWAS. The outcomes were all cause dementia or late-onset Alzheimer’s disease (LOAD). Uncertainties were resolved by discussion with two other reviewers (WNW and RMW).\n\nA quality assessment tool was developed by synthesizing three published guidelines for assessing MR studies12–14 (Table 1). Studies that did not satisfactorily address items 3-5, which describe the three core assumptions of MR, were excluded.\n\n2SMR = 2-sample Mendelian Randomization, has independent outcome and exposure samples. 1SMR = 1-sample Mendelian Randomization.\n\nFor each study, we extracted: GWAS source, ethnicity, number of single nucleotide polymorphisms (SNPs) used as instrumental variables for each risk factor, case/control sample size, effect estimates and units, and measures of variation were extracted for each study. When multiple analysis approaches were used to generate an effect estimate, the value generated using the most SNPs without compromising pleiotropy was used (linkage disequilibrium r2 < 0.2). Effect estimates that included IVs mapping to the APOE locus, which has a known association with dementia were excluded. For studies with missing data, the authors were emailed twice and the study was excluded if no reply was received. The full details are available in Supplementary Material 2, S1 & S2.42\n\nEffect estimates and measures of variation were standardized into common units for each risk factor. The effect estimates were pooled using a fixed effects model to generate an overall estimate for each risk factor of interest. Cochrane Q and I2 statistics were used to assess heterogeneity across studies. Analyses and plots were executed with the Metafor package (version 2.4-0) in R (version 4.0.3).15 A Bonferroni correction was applied to maintain a 5% family-wise error rate, yielding a significance threshold of 0.05 divided by n risk factors assessed (p = 0.05/9).\n\nWe performed sensitivity analysis by meta-analysing using alternative eligible studies, which were excluded in our primary meta-analysis due to outcome cohort overlap or the use of a superseded exposure GWAS. We substituted the studies with IGAP (2013) as the outcome cohort, with another study with the same outcome cohort with either the lowest or highest estimate in attempt to determine any significant change in overall effect estimate. Statistical significance was determined by applying a Bonferroni correction, as described above.\n\nSearch strategy\n\nWe searched on OVID, Scopus and Web of Science for a meta-analyses of cohort studies estimating the association between vascular risk factors with later dementia (Supplementary Material 1, Table 2).\n\nStudy selection\n\nWe selected one representative meta-analysis of observational studies for each risk factor. We considered all articles in English which analysed cohorts, reported OR/HR/RR, and associated 95% confidence interval (CI) or standard error. If multiple meta-analyses were eligible for inclusion per given risk factor, we selected the study with the highest total number of participants.\n\n\nResults\n\nA search of three databases, OVID, Scopus and Web of Science, found 5211 unique articles. After applying inclusion and exclusion criteria to the abstracts, 29 articles were retained (Figure 1). Seventeen articles were not included for meta-analysis after full text review because either the results were not reported per unit increase in risk factor or there was an overlap of between outcome cohorts. Of those seventeen articles, ten studies were reserved for secondary outcome analyses to give a total of 22 articles included in this study (Supplementary Material 2, S1).\n\nArticle relevance was assessed from title and abstract. Subsequently, full text was read for confirmation or further exclusion and additional studies were identified by forward searching from selected articles.\n\nThere were three primary MR studies each for type 2 diabetes16–18 and LDL cholesterol19–21; two studies each for HDL cholesterol,19,22 total cholesterol,19,23 triglycerides,19,21 and systolic blood pressure20,24; and one study each for diastolic blood pressure,25 BMI26 and circulating glucose27 (Figure 1). The MR studies for diastolic blood pressure, BMI, and circulating glucose reported an overall estimate produced through the meta-analysis of two or more outcome cohorts. As such, the estimates from these studies were included in the present study.\n\nQuality assessment was performed on the 12 selected studies (Table 2). All studies addressed the three assumptions of MR. The MR results for each of the six vascular risk factors with more than one MR study were meta-analysed (Figure 2). BMI was significantly associated with dementia as stated in the original study by Li et al. (2021), with higher BMI increasing the odds of developing dementia (1.03 [1.01, 1.05] per 5 kg/m2 increase, p = 0.00285), and met the corrected significance threshold.\n\nYes = Reported, No = Not reported. Each question corresponds to the questions outlined in Table 1.\n\n* Ware et al. (2021) conducted a one sample MR.\n\nUKB = UK Biobank, IGAP = International Genomics of Alzheimer's Project, HRS = Health and Retirement Study, MRC-WTCCC2 = Medical Research Council (MRC)-Wellcome Trust Case Control Consortium, IOP+ = Institute of Psychiatry Plus, ADNI = Alzheimer's Disease Neuroimaging Initiative. Copenhagen Studies = Copenhagen General Population Study and the Copenhagen City Heart Study.\n\nComparison of different MR studies using the same cohort (IGAP) had similar estimates (I2 = 0%) for all risk factors, apart from for LDL-c (I2 = 65.2%) and systolic blood pressure (I2 = 25.2%) (Figure 3). The MR studies included in this analysis all fulfilled the three core assumptions of MR. Sensitivity analysis to replace MR studies using the IGAP (2013) outcome GWAS with another MR study with the most extreme values rendered the meta-analysed effect estimate for all risk factors to remain non-significant (Supplementary Material 1, Figure 1). Diastolic blood pressure, BMI and circulating glucose were excluded from sensitivity analysis because only a single study was reported for each risk factor in this study.\n\nThis figure includes studies that have not been selected in the meta-analysis and is shown to highlight the heterogeneity of results despite using the same outcome cohort. LOAD = Late onset Alzheimer’s disease; IGAP = International Genomics of Alzheimer's Project; CI = Confidence Interval.\n\nFor eight of the risk factors, we compared the effect estimate from MR studies with the effect estimate from the largest available meta-analysis of observational studies (Figure 4).28–33 One risk factor, circulating glucose, did not have an eligible meta-analysis; therefore, the primary study with largest study cohort was included as the comparator.34 The units of estimates from most meta-analyses of observational studies were not explicitly stated, so it was not possible to compare the magnitude of effect with that obtained from the meta-analysis of MR studies. There were significant associations between LDL cholesterol, diastolic blood pressure diabetes and circulating glucose with a higher risk of later dementia in observational studies, but neutral associations from studies using MR.\n\nSample size is the sum of control and case numbers.\n\n\nDiscussion\n\nWe performed a meta-analysis of MR studies assessing the effects of nine modifiable risk factors for vascular diseases on the odds of dementia. With the exception of for BMI, we did not obtain evidence for an association between genetically predicted levels of any vascular risk factor and the odds of developing dementia.\n\nFewer vascular risk factors were associated with dementia based on meta-analysed estimates from MR studies than from meta-analyses of observational studies. This may be because the estimates from cohort studies were limited by residual confounding, or by a selection bias that identified populations with particularly high risk of dementia due to vascular risk factors, or by publication bias towards reporting positive results. The MR studies may have been limited by weak genetic instruments that explained only a small percentage of the variation in the risk factors of interest. For example, in Andrews et al. (2021) and Malik et al. (2021) less than 6% of the variation in blood pressure was explained by all the SNPs identified through GWASs. Moreover, these MR studies are unable to distinguish age-dependent mechanisms. For instance, high systolic blood pressure has been linked with harmful effects during mid-life but protective effects during late-life (>75).35,36 A similar case has been made for cholesterol levels.37 There is additional uncertainty regarding the reliability of GWASs themselves in which genetic instruments were derived from. GWASs may have been underpowered, resulting in unreliable identification of IV SNPs.38 The populations included in the GWASs for exposure and outcome may differ, limiting their comparability. An attempt to mitigate this issue was, however, made by using GWAS studies that focused on populations of European descent. A further consideration is that differences in model formulation between the GWASs from which IVs are identified and observational studies has the potential to limit the comparability of the phenotype under consideration.\n\nAn exploration of the consistency between the findings of MR studies that used different exposure GWASs but the same outcome GWAS (citation) found consistency between the MR estimates for all risk factors, except for LDL cholesterol and systolic blood pressure. This may be because of differences in the SNPs used as IVs for these risk factors. Sometimes differences in the SNPs used as IVs are attributable to a decision to focus on specific pathways: for example, Benn et al. (2017) focused on PCSK9 and HMGCR variants for LDL cholesterol to tackle pathways that are therapeutically relevant based on PCSK9 inhibitors and statins.39 They reported a statistically significant causative effect of LDL-c, unlike other studies that included SNPs from other genomic regions (Figure 3). However, the choice of SNPs may also be simply due to differences in available data. For instance, Østergaard et al. (2015) obtained 24 SNPs to act as IVs for systolic blood pressure from an up-to-date GWAS at that time, while Larsson et al. (2021) obtained 93 IV SNPs from a study published in 2017. Østergaard et al. (2015) concluded that higher systolic blood pressure lowered the odds of dementia, while Larsson et al. (2021) found no significant association (Figure 3). Therefore, recognizing these differences is important when interpreting the estimates from MR studies.\n\nWe identified relatively few eligible studies. Although many potentially relevant primary MR studies were identified, many studies used data from the same source; for example, many studies of Alzheimer’s disease use IGAP (2013).40 Therefore, many otherwise eligible studies had to be excluded to ensure independence of estimates in our meta-analysis (Supplementary Material 2, S2). The limited number of eligible studies led us to pool our two outcomes of interest (Alzheimer’s disease and dementia). Although Alzheimer’s disease constitutes 60-80% of dementia cases, a significant proportion of dementia cases are associated with different diseases with distinct pathophysiology such as vascular dementia.41 Discrepancies in pathophysiology may potentially be reflected in the heterogeneity of estimates as seen in systolic blood pressure where outcomes Alzheimer’s disease and dementia were both present (I2 = 92.6%) (Figure 2).\n\nSecondly, although we meta-analysed studies with unique outcome cohorts, we did not have access to individual participant data. We were therefore unable to control any potential overlap between the cohorts.\n\nThirdly, we found heterogeneity between the MR estimates obtained using the same outcome cohort for two risk factors, systolic blood pressure and LDL cholesterol. This was in spite of key similarities in the study, such as only including population of European ancestry and using the same GWAS to derive the SNPs. The heterogeneity observed is likely to stem from differences in methods, such as discrepancy in rationale for selection of SNPs, p-value cutoff for the use of SNPs as IVs, and discrepancy in covariates when analysing exposure-outcome relationship. The exact effect these differences have on the estimate and its clinical implications remains yet to be characterized. There is a need to assess the effect of such discrepancies and the robustness of MR estimates through sensitivity analyses.\n\n\nConclusion\n\nOut of the nine vascular risk factors assessed in this study, only genetically predicted BMI showed evidence of being causally associated with dementia. Estimates from observational studies for many risk factors were significantly associated with dementia.\n\n\nData availability\n\nOpen Science Framework: Assessing the role of vascular risk factors in dementia: Mendelian randomization meta-analysis and comparison with observational estimates, https://doi.org/10.17605/OSF.IO/UA7Z6.42\n\nThis project contains the following underlying data:\n\n- Supplementary material 1. pdf (search strategy and sensitivity analysis)\n\n- Supplementary material 2. xlsx (raw data and analysis)\n\n\nReporting guidelines\n\nOpen Science Framework: PRISMA checklist for ‘Assessing the role of vascular risk factors in dementia: Mendelian randomization meta-analysis and comparison with observational estimates’, https://doi.org/10.17605/OSF.IO/UA7Z6.42\n\nData are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).",
"appendix": "References\n\nBaumgart M, Snyder HM, Carrillo MC, et al.: Summary of the evidence on modifiable risk factors for cognitive decline and dementia: A population-based perspective. Alzheimers Dement. 2015 Jun 1; 11(6): 718–726. PubMed Abstract | Publisher Full Text\n\nLivingston G, Huntley J, Sommerlad A, et al.: Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet 2020 Aug 8 [cited 2021 Sep 21]; 396(10248): 413–446. PubMed Abstract | Publisher Full Text Reference Source\n\nHughes D, Judge C, Murphy R, et al.: Association of Blood Pressure Lowering With Incident Dementia or Cognitive Impairment: A Systematic Review and Meta-analysis. JAMA 2020 May 19 [cited 2021 Oct 19]; 323(19): 1934. Publisher Full Text | Free Full Text\n\nPatel CJ, Burford B, Ioannidis JPA: Assessment of vibration of effects due to model specification can demonstrate the instability of observational associations. J. Clin. Epidemiol. 2015 Sep 1 [cited 2021 Jul 9]; 68(9): 1046–1058. PubMed Abstract | Publisher Full Text | Free Full Text\n\nCarlson MDA, Morrison RS: Study design, precision, and validity in observational studies. J. Palliat. Med. 2009 Jan 1 [cited 2020 Nov 12]; 12(1): 77–82. PubMed Abstract | Publisher Full Text | Free Full Text\n\nSmith GD, Hemani G: Mendelian randomization: Genetic anchors for causal inference in epidemiological studies. Hum. Mol. Genet. 2014 [cited 2020 Nov 12]; 23(R1): R89–R98. PubMed Abstract | Publisher Full Text\n\nBurgess S, Thompson SG; Collaboration CCG: Avoiding bias from weak instruments in Mendelian randomization studies. Int. J. Epidemiol. 2011 Jun 1 [cited 2021 Oct 19]; 40(3): 755–764. PubMed Abstract | Publisher Full Text\n\nRiaz H, Khan MS, Siddiqi TJ, et al.: Association Between Obesity and Cardiovascular Outcomes: A Systematic Review and Meta-analysis of Mendelian Randomization Studies. JAMA Netw. Open 2018 Nov 2 [cited 2021 Oct 19]; 1(7): e183788. Publisher Full Text | Free Full Text\n\nMoher D, Liberati A, Tetzlaff J, et al.: Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. Vol. 6, PLoS Medicine. PLoS Med. 2009 [cited 2020 Nov 12]; 6: e1000097. PubMed Abstract | Publisher Full Text\n\nSystematic review and meta-analysis of Mendelian randomisation studies on modifiable risk factors for dementia.[cited 2022 Feb 16].Reference Source\n\nWan X, Wang W, Liu J, et al.: Estimating the sample mean and standard deviation from the sample size, median, range and/or interquartile range. BMC Med. Res. Methodol. 2014 Dec 1 [cited 2021 Dec 22]; 14(1). PubMed Abstract | Publisher Full Text\n\nDavies NM, Holmes MV, Davey Smith G: Reading Mendelian randomisation studies: A guide, glossary, and checklist for clinicians. BMJ (Online) 2018 [cited 2020 Nov 12]: k601. PubMed Abstract | Publisher Full Text\n\nGrover S, Del Greco F, König IR: Evaluating the current state of Mendelian randomization studies: A protocol for a systematic review on methodological and clinical aspects using neurodegenerative disorders as outcome. Syst. Rev. 2018 Sep 24 [cited 2020 Nov 12]; 7(1): 145. PubMed Abstract | Publisher Full Text\n\nBurgess S, Davey Smith G, Davies NM, et al.: Guidelines for performing Mendelian randomization investigations. Wellcome Open Res. 2020 [cited 2020 Nov 12]; 4: 4. Publisher Full Text | Free Full Text\n\nViechtbauer W: Conducting meta-analyses in R with the metafor. J. Stat. Softw. 2010; 36(3). Publisher Full Text\n\nGarfield V, Farmaki A-E, Fatemifar G, et al.: The Relationship Between Glycaemia, Cognitive Function, Structural Brain Outcomes and Dementia: A Mendelian Randomisation Study in the UK Biobank. Diabetes 2021 Jun 15 [cited 2021 Jul 7]: db200895.Reference Source\n\nPan Y, Chen W, Yan H, et al.: Glycemic traits and Alzheimer’s disease: a Mendelian randomization study. Aging (Albany NY) 2020 Nov 1 [cited 2021 Jul 7]; 12(22): 22688. Free Full Text\n\nWare EB, Morataya C, Fu M, et al.: Type 2 Diabetes and Cognitive Status in the Health and Retirement Study: A Mendelian Randomization Approach. Front. Genet. 2021 Mar 25 [cited 2021 Jul 7]; 12: 634767. PubMed Abstract | Publisher Full Text | Free Full Text\n\nProitsi P, Lupton MK, Velayudhan L, et al.: Genetic Predisposition to Increased Blood Cholesterol and Triglyceride Lipid Levels and Risk of Alzheimer Disease: A Mendelian Randomization Analysis. PLoS Med. 2014 Sep 1 [cited 2021 Jul 7]; 11(9): e1001713. PubMed Abstract | Publisher Full Text | Free Full Text\n\nMalik R, Georgakis MK, Neitzel J, et al.: Midlife vascular risk factors and risk of incident dementia: Longitudinal cohort and Mendelian randomization analyses in the UK Biobank. Alzheimers Dement. 2021 [cited 2021 Jul 7]; 17: 1422–1431. Publisher Full Text\n\nBurgess S, Smith GD: Mendelian Randomization Implicates High-Density Lipoprotein Cholesterol–Associated Mechanisms in Etiology of Age-Related Macular Degeneration. Ophthalmology 2017 Aug 1 [cited 2021 Jul 7]; 124(8): 1165–1174. PubMed Abstract | Publisher Full Text | Free Full Text\n\nKjeldsen EW, Thomassen JQ, Juul Rasmussen I, et al.: Plasma HDL cholesterol and risk of dementia - observational and genetic studies. Cardiovasc. Res. 2021 May 8 [cited 2021 Dec 21]; 118: 1330–1343. PubMed Abstract | Publisher Full Text\n\nØstergaard SD, Mukherjee S, Sharp SJ, et al.: Associations between Potentially Modifiable Risk Factors and Alzheimer Disease: A Mendelian Randomization Study. PLoS Med. 2015 Jun 1 [cited 2021 Sep 4]; 12(6): e1001841. PubMed Abstract | Publisher Full Text\n\nAndrews SJ, Fulton-Howard B, O’Reilly P, et al.: Causal Associations Between Modifiable Risk Factors and the Alzheimer’s Phenome. Ann. Neurol. 2021 [cited 2021 Jul 7]; 89(1): 54–65. PubMed Abstract | Publisher Full Text | Free Full Text\n\nSproviero W, Winchester L, Newby D, et al.: High Blood Pressure and Risk of Dementia: A Two-Sample Mendelian Randomization Study in the UK Biobank. Biol. Psychiatry 2021 Apr 15 [cited 2021 Dec 21]; 89(8): 817–824. PubMed Abstract | Publisher Full Text\n\nLi X, Tian Y, Yang Y-X, et al.: Life Course Adiposity and Alzheimer’s Disease: A Mendelian Randomization Study. J. Alzheimer's Dis. 2021 [cited 2021 Nov 5]; 82(2): 503–512. PubMed Abstract | Publisher Full Text\n\nBenn M, Nordestgaard BG, Tybjærg-Hansen A, et al.: Impact of glucose on risk of dementia: Mendelian randomisation studies in 115,875 individuals. Diabetologia 2020 Mar 14 [cited 2021 Nov 5]; 63(6): 1151–1161. PubMed Abstract | Publisher Full Text\n\nRahmani J, Roudsari AH, Bawadi H, et al.: Body mass index and risk of Parkinson, Alzheimer, Dementia, and Dementia mortality: a systematic review and dose-response meta-analysis of cohort studies among 5 million participants. Nutr. Neurosci. 2020 [cited 2021 Dec 18]; 25: 423–431. PubMed Abstract | Publisher Full Text\n\nXue M, Xu W, Ou YN, et al.: Diabetes mellitus and risks of cognitive impairment and dementia: A systematic review and meta-analysis of 144 prospective studies. Ageing research reviews.2019 Nov 1 [cited 2021 Dec 18]; 55. PubMed Abstract\n\nSáiz-Vazquez O, Puente-Martínez A, Ubillos-Landa S, et al.: Cholesterol and Alzheimer’s disease risk: A meta-meta-analysis. Brain Sci. MDPI AG 2020 [cited 2021 Apr 21]; 10: 1–13. Publisher Full Text | Free Full Text\n\nPeters R, Xu Y, Antikainen R, et al.: Evaluation of High Cholesterol and Risk of Dementia and Cognitive Decline in Older Adults Using Individual Patient Meta-Analysis. Dement. Geriatr. Cogn. Disord. 2021 [cited 2021 Dec 18]; 50(4): 318–325. PubMed Abstract | Publisher Full Text\n\nFord E, Greenslade N, Paudyal P, et al.: Predicting dementia from primary care records: A systematic review and meta-analysis. PLoS One. 2018 Mar 1 [cited 2021 Dec 18]; 13(3): e0194735. PubMed Abstract | Publisher Full Text | Free Full Text\n\nOu YN, Tan CC, Shen XN, et al.: Blood Pressure and Risks of Cognitive Impairment and Dementia: A Systematic Review and Meta-Analysis of 209 Prospective Studies. Hypertension 2020 Jul 1 [cited 2021 Dec 21]; 76(1): 217–225. Publisher Full Text\n\nCrane PK, Walker R, Hubbard RA, et al.: Glucose levels and risk of dementia. Forschende Komplementarmedizin. 2013 Aug 7 [cited 2021 Dec 18]; 20(5): 386–387. Publisher Full Text\n\nKivipelto M, Helkala EL, Laakso MP, et al.: Midlife vascular risk factors and Alzheimer’s disease in later life: longitudinal, population based study. BMJ (Clinical research ed) 2001 Jun 16 [cited 2021 Sep 29]; 322(7300): 1447–1451. PubMed Abstract\n\nQiu C, von Strauss E , Winblad B, et al.: Decline in Blood Pressure Over Time and Risk of Dementia. Stroke 2004 Aug 1 [cited 2021 Sep 29]; 35(8): 1810–1815. Publisher Full Text\n\nvan Vliet P : Cholesterol and late-life cognitive decline. J. Alzheimer's Dis. 2012 [cited 2021 Sep 29]; 30(SUPPL.2): S147–S162. PubMed Abstract | Publisher Full Text\n\nTam V, Patel N, Turcotte M, et al.: Benefits and limitations of genome-wide association studies. Nat. Rev. Genet. 2019 May 8 [cited 2022 Jan 24]; 20(8): 467–484. PubMed Abstract | Publisher Full Text Reference Source\n\nBenn M, Nordestgaard BG, Frikke-Schmidt R, et al.: Low LDL cholesterol, PCSK9 and HMGCR genetic variation, and risk of Alzheimer’s disease and Parkinson’s disease: Mendelian randomisation study. BMJ 2017 Jun 29 [cited 2021 Sep 28]: j3170. https://www.bmj.com/content/357/bmj.j3170. Publisher Full Text\n\nLambert JC, Ibrahim-Verbaas CA, Harold D, et al.: Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer’s disease. Nat. Genet. 2013 Dec 1 [cited 2021 Jun 13]; 45(12): 1452–1458. PubMed Abstract | Publisher Full Text | Free Full Text\n\n2021 Alzheimer’s disease facts and figures: Alzheimer’s and Dementia.2021 Mar 1 [cited 2021 Apr 20]; 17(3): 327–406. PubMed Abstract\n\nLee LS: Assessing the role of vascular risk factors in dementia: Mendelian randomization meta-analysis and comparison with observational estimates. [Dataset] 2022, May 5. Publisher Full Text"
}
|
[
{
"id": "157914",
"date": "19 Jan 2023",
"name": "Daniel A Nation",
"expertise": [
"Reviewer Expertise Vascular factors contributing to dementia"
],
"suggestion": "Approved",
"report": "Approved\n\ninfo_outline\nAlongside their report, reviewers assign a status to the article:\n\nApproved The paper is scientifically sound in its current form and only minor, if any, improvements are suggested\n\nApproved with reservations\nA number of small changes, sometimes more significant revisions are required to address specific details and improve the papers academic merit.\n\nNot approved Fundamental flaws in the paper seriously undermine the findings and conclusions\n\nThis is a meta-analysis of mendelian randomization studies and a comparison of this meta-analysis with observational study estimates, to evaluate the potential causal nature of observed associations between 9 vascular risk factors and dementia. The authors conducted a thorough search and identified 12 out of 5211 studies meeting criteria for inclusion in the analysis. Findings indicated higher genetically predicted BMI is associated with slightly higher odds of dementia. Furthermore, fewer vascular risk factors were associated with dementia in mendelian randomization studies than in observational studies. The question of a causal role for modifiable vascular risk factors for dementia is critical to efforts to treat and prevent cognitive impairment, making the topic of this article particularly important. Strengths of the present study include examination of MR which may allow for causal inference, meta-analysis of MR studies in this area which is novel, and comparison with observational studies. The methods appear to be rigorous. Unfortunately, very little can be concluded from the study due to limitations of available data within MR studies themselves and across studies. Although the authors extensively outline these limitations in excellent detail, the limits curtail the ultimate impact of this study. Nevertheless, it was well written and perhaps serves as an initial step forward or a snapshot of field as it currently stands, including the many remaining questions.\n\nAre the rationale for, and objectives of, the Systematic Review clearly stated? Yes\n\nAre sufficient details of the methods and analysis provided to allow replication by others? Yes\n\nIs the statistical analysis and its interpretation appropriate? I cannot comment. A qualified statistician is required.\n\nAre the conclusions drawn adequately supported by the results presented in the review? Yes",
"responses": []
},
{
"id": "307125",
"date": "27 Sep 2024",
"name": "Ahmet Turan Isik",
"expertise": [
"Reviewer Expertise Cognitive Impairement in older adults",
"NPH",
"Dysautonomia in DLB"
],
"suggestion": "Approved With Reservations",
"report": "Approved With Reservations\n\ninfo_outline\nAlongside their report, reviewers assign a status to the article:\n\nApproved The paper is scientifically sound in its current form and only minor, if any, improvements are suggested\n\nApproved with reservations\nA number of small changes, sometimes more significant revisions are required to address specific details and improve the papers academic merit.\n\nNot approved Fundamental flaws in the paper seriously undermine the findings and conclusions\n\nI have reviewed the manuscript \"Assessing the role of vascular risk factors in dementia: Mendelian randomization meta-analysis and comparison with observational estimates\" with interest. The paper needs to reconsider a few flaws. The background and purpose of the study should be emphasized more clearly. The authors should explain the effects of any potential overlap between the cohorts on the results. It would be better for a statistician to recheck the results because I am not good enough at this topic. No competing interests were disclosed\n\nAre the rationale for, and objectives of, the Systematic Review clearly stated? No\n\nAre sufficient details of the methods and analysis provided to allow replication by others? Yes\n\nIs the statistical analysis and its interpretation appropriate? I cannot comment. A qualified statistician is required.\n\nAre the conclusions drawn adequately supported by the results presented in the review? Yes\n\nIf this is a Living Systematic Review, is the ‘living’ method appropriate and is the search schedule clearly defined and justified? (‘Living Systematic Review’ or a variation of this term should be included in the title.) Not applicable",
"responses": [
{
"c_id": "13263",
"date": "07 Feb 2025",
"name": "Liam Lee",
"role": "Author Response",
"response": "Dear Dr Isik, Many thanks for your valuable feedback on our manuscript. We have carefully considered your comments and made the following changes to address them. Comment 1: The background and purpose of the study should be emphasized more clearly. We have revised the background section to provide a clearer explanation of the rationale and purpose of the study. This includes highlighting the limitations of observational studies in inferring causality and the importance of using Mendelian randomization (MR) as a complementary approach. Comment 2: The authors should explain the effects of any potential overlap between the cohorts on the results. We have added a discussion on the effects of cohort overlap in the Limitations section of the manuscript. We acknowledge that cohort overlap could lead to bias, such as inflated precision and potential under- or overestimation of effect sizes. To address this, we emphasized our sensitivity analyses by substituting alternative estimates where possible and found that the overall pooled estimates remained stable. Comment 3: It would be better for a statistician to recheck the results because I am not good enough at this topic. We would like to reassure the reviewer that one of the co-authors, Dr. Walker, is a trained epidemiologist with extensive experience in statistical methods and MR studies. Dr. Walker has validated the statistical analyses and interpretation presented in the manuscript."
}
]
}
] | 1
|
https://f1000research.com/articles/11-565
|
https://f1000research.com/articles/13-277/v2
|
10 May 24
|
{
"type": "Research Article",
"title": "Impact of yoga on the central and peripheral vascular function among desk-based workers: A single-centered trial study",
"authors": [
"Poovitha Shruthi P",
"Koustubh Kamath",
"Vaishali K",
"Shivashankar K N",
"Suresh Sugumar",
"Sneha Ravichandran",
"Leena R David",
"Peter Hogg",
"Guruprasad V",
"Banumathe K R",
"Shovan Saha",
"Rajagopal Kadavigere",
"Poovitha Shruthi P",
"Koustubh Kamath",
"Vaishali K",
"Shivashankar K N",
"Sneha Ravichandran",
"Leena R David",
"Peter Hogg",
"Guruprasad V",
"Banumathe K R",
"Shovan Saha"
],
"abstract": "Background The aim of this study was to observe and analyze vascular function in ‘prolonged sitting’, followed by a yoga asana routine and pranayama intervention. Participants in this study include those who work from desks in offices. The study required the participants to attend on three separate days at random, and they had to finish a computerized test on each day. On the first day, participants were required to complete a computer test while sitting still for four hours (with the exception of washroom breaks). The next day, they underwent a computerized test along with a pranayama intervention. Finally, on the last day, they underwent a computerized test along with a yoga asana intervention. At the start of the study and after two and four hours, we measured the diameter and velocity of the common carotid artery (CCA) and superficial femoral artery (SFA).\n\nMethods The study was a within-subjects prospective single-center trial conducted in the Department of Radio-Diagnosis and Imaging, Kasturba Medical Hospital, Manipal, India, between September 2022 and January 2023. Participants were asked to do one of the following ‘activities’ over successive weeks: Week 1 – Prolonged sitting; Week 2 – Pranayama intervention; and Week 3 – Yoga asana intervention during prolonged sitting. The baseline and follow-up variables of pulse velocity, endothelial thickness, and shear rate were assessed for normality through a Shapiro-Wilk Test.\n\nResults Our sample included 11 participants with moderate physical activity, five with high physical activity and one with low physical activity. Yoga asana intervention comprised participants sitting continuously for four hours, with a yoga asana intervention being provided every hour, lasting for 10 minutes.\n\nConclusions Yoga asana improves vascular functions in prolonged sitting conditions. This routine can promote the concept of interrupted sitting and ways to reduce it with efficient yoga asana practice without changing the work culture and provide better physical relief.\n\nTrial registration\nClinical Trials Registry – India ( CTRI/2022/09/045628), date of registration: 19/09/2022(CTRI/2022/9/045628)https://ctri.nic.in/Clinicaltrials/main1.php?EncHid=16349.27799,",
"keywords": [
"Physical Activity",
"Vascular Function",
"FMD",
"Exercise",
"Sedentary lifestyle"
],
"content": "Introduction\n\nFor a growing proportion of the world’s population, the impact of the digital working era combined with a tendency to adopt an overall sedentary lifestyle has resulted in people sitting for prolonged periods of time. This can lead to low energy expenditure and lower physical activity (LPA) levels. This has many adverse effects, including early onset of metabolic diseases such as hypertension, type 2 diabetes mellitus, obesity, atherosclerosis, late-life cognitive decline, loss of alertness, and a reduced reaction time.1–8 For an adult, according to the World Health Organization (WHO), the average individual requires at least 150 min of exercise per week.9 However, meeting this demand might not be accessible due to the pressures of modern life. As a result, extended sitting has been one strategy used to assist people in adopting a variety of postural motions, mostly in office-based working contexts.1–8\n\nA root cause of the above disorders has been linked to vascular impairment; therefore, improving vascular blood flow becomes essential in minimizing the potential of developing adverse effects in later life.10 Blood vessels that are particularly affected include the superficial femoral artery (SFA) and common carotid artery (CCA).11 The seated position has a marked and negative impact on SFA blood flow, being confounded by a sedentary lifestyle and working at computers. SFA flow reduction can cause damage and acute distress12; similarly, CCA flow reduction can also have detrimental consequences.\n\nResearch has been published to investigate ways to reduce prolonged sitting and to offer interventions that can improve vascular function whilst minimizing disturbance to the working environment and work productivity.13–16 Additional problems of interventions within the working environment are related to space for physical activity and the cost of any exercise-related equipment. For the workplace, a need exists to identify even more productive, feasible, convenient, cost-limited and practical exercises that increase blood flow without negative impact and at an acceptable cost.\n\nIt is suggested that customized yoga asana, using the pranayama routine, can improve concentration and alertness.17–20 Yoga experts suggest this can improve vascular function and blood flow. However, yoga asana’s impact on CCA and SFA blood flow is unclear. Consequently, our study aimed to observe and analyse CCA and SFA blood flow in ‘prolonged sitting’, followed by a yoga asana routine and pranayama intervention.\n\n\nMethods\n\nThe study was a within-subjects prospective single-centre trial conducted in the Department of Radio-Diagnosis and Imaging, Kasturba Medical Hospital, Manipal, India, between September 2022 to January 2023. After approval from the Institutional Ethics Committee, Kasturba Medical College (KMC) and Kasturba Hospital (KH) (IEC1:108/2022) (approved on 22 July 2022) and Clinical Trial Registry of India (CTRI/2022/9/045628)https://ctri.nic.in/Clinicaltrials/main1.php?EncHid=16349.27799, with prior written informed consent, 17 participants (desk based workers) were recruited for the study. This included nine males (mean age 25, age range 24–28) and eight females, mean age 25 (age range 24–28). Flyers in campus stores and adverts in Manipal University Trading & Property helpline MUTC a facebook based group were used to recruit participants for the study, which was conducted at Kasturba Hospital in Manipal. The participants had no history of cardiovascular or metabolic disorders, the females were not menstruating, none had depression during the study period, and none were taking medication, which might alter vascular function. Self-reported physical activity was assessed using the International Physical Activity Questionnaire (IPAQ).1 The questionnaire assesses time (frequency and duration) spent on vigorous, moderate intensity, walking activities and sitting time in hours for the past seven days. Our sample included 11 participants with moderate physical activity, five with high physical activity and one with low physical activity; the IPAQ data is shown in Table 1. For this investigation, repeated measures ANOVA (parametric) or Friedman’s test (non-parametric) are used as statistical tests performed on JASP statistical software version 0.16.2 (https://jasp-stats.org/previous-versions/).\n\nOver three weeks, participants were requested to do one of the following ‘activities’ over successive weeks: Week 1 – prolonged sitting; Week 2 – pranayama intervention during prolonged sitting; and Week 3 – yoga asana intervention during prolonged sitting. Vascular function was assessed for each of the three conditions including the diameter and blood velocity of the SFA and CCA were measured at the start of the study (0), two, and four hours during sitting using a Doppler ultrasound (Philips, Epiq Elite). Ultrasound machine quality assurance test results fell within manufacturer specifications. Ultrasound was performed by a competent radiologist (see Figure 1).\n\nProlonged sitting comprised of participants sitting continuously for four hours (see Figure 2). Pranayama intervention comprised participants sitting continuously for four hours, with the pranayama intervention being provided every hour, lasting 10 minutes. The yoga asana intervention comprised participants sitting continuously for four hours, with a yoga asana intervention being provided every hour, lasting for 10 minutes.\n\nPranayama practice included Deep breathing, Nadisodhana Pranayama and Bhramari Pranayama.17–19 Yoga asana practices included Tadasana, Ardha Chakrasana, Uttana Mandukasana,20 Skandha Chakra Asana, Kati Chakrasana, Prasarita Padottanasan.\n\n\nResults\n\nThe study comprised 17 participants in the 25–35 age range. Nine males (mean age 25, age range 24–28) and eight females (mean age 25, age range 24–28) made up the participants.25 None of the participants experienced depression throughout the study period, none had a history of cardiovascular or metabolic diseases, and none were taking any medications that would affect vascular function. Additionally, none of the female participants were menstruating. The IPAQ was used to evaluate self-reported physical activity. For the previous seven days, the questionnaire measures the amount of time (frequency and duration) spent on walking activities that are vigorous or moderately intense as well as hours spent sitting down.\n\nFigure 3 illustrates how the results of the CCA reveal that the artery’s width decreases with each set of interventions. In comparison with time, the first and second hours exhibited statistical significance with a p value<0.001, whereas the third and fourth hours do not exhibit statistical significance (p value=0.020). Although the yoga asana intervention shows significance, it does not show statistical significance when compared to pranayama group and prolonged sitting (p value=0.014). Table 2 represents the data of CCA diameter at different time points.\n\n(In figure X axis represents Time, Y axis represents vascular diameter (cm).)\n\nIn addition, the yoga asana intervention did not show statistical significance when compared to the pranayama intervention and prolonged sitting (p value=0.017).\n\nIn the CCA, a reduction in velocity was observed after the pranayama and yoga asana intervention. However, when it comes to prolonged sitting, there was a decrease in velocity at the start of the study and at two hours, followed by an increase in velocity during the final hour as shown in Figure 4. With CCA velocity, neither time nor blood velocity were statistically significant. Table 3 represents the data of CCA velocity at different time points.\n\n(In figure X axis represents Time, Y axis represents velocity (m/s).)\n\nThe diameter of the SFA decreased each hour during prolonged sitting and pranayama intervention. However, during the yoga asana intervention, there was a reduction in diameter at the beginning of the study and at 2 hours, but an increase in diameter was observed in the 4th hour, shown in Figure 5. Neither time nor group showed any significance. Hence the group has not demonstrated any substantial response to any intervention. Tables 4 & 5 represents the data of SFA diameter and velocity at different time points respectively.\n\n(In figure X axis represents Time, Y axis represents diameter in cm.)\n\nIn the CCA, the shear rate increased each hour during prolonged sitting, whereas it decreased in both the pranayama and yoga asana interventions (Figure 6). All interventions showed a reduction in stress in CCA (Figure 7). However, in terms of shear and stress, neither the time nor the group showed any significant difference. Table 6 represents the data of CCA shear and stress at different time points.\n\n(In figure X axis represents Time, Y axis represents velocity (m/s).)\n\n(In figure X axis represents Time, Y axis represents velocity (m/s).)\n\n\nDiscussion\n\nOur study aimed to determine whether yoga asana affects vascular function in desk-based workers. Exercise will maintain and enhance both physical and mental health. By controlling the sympathetic nervous system and hypothalamus pituitary adrenal axis, yoga asana can improve physical and psychological health.21\n\nIn our study, yoga asana intervention improves the vascular function of the carotid and superficial femoral artery diameter. Comparing time, the first and second hours exhibited statistical significance with a p value of <0.001. Compared to the group, the yoga asana intervention showed effectiveness but no statistical significance in the carotid artery. Significance is shown in both the 0th and 2nd hours in the superficial femoral artery, with a p-value of 0.026 and 0.004*, respectively. Although they are not statistically significant. A study by Ross and Thomas (2010) found that both yoga asana and exercise significantly reduced fasting blood glucose levels at three and six months (29.48%, 27.43%; p<0.0001), with yoga asana showing better results in social and occupational functioning compared to the exercise group. The study also found that both yoga asana and exercises improved total serum cholesterol (p<0.0001) and low protein level (p=0.030).19 So, when we compare this to our study it proves that yoga asana shows significant changes.\n\nIn a study by Pearson and Smart (2017), participants underwent aerobic exercise with an intervention duration ranging from four weeks to six months, with the frequency of sessions varying from daily to two days per week, and the time of exercise sessions ranged from 10 to 60 minutes. The study concluded that aerobic exercise training significantly improved endothelial function.21 As compared to our study, the yoga asana intervention showed improvement in both carotid and superficial femoral artery diameter.\n\nA study conducted by Cortez-Cooper examined the effect of different types of intervention on cardiovascular health. The study included three groups: a strength training group, a combination group that received both strength training and aerobic exercise, and a stretching group. The results showed that the strength training and combination groups experienced significant changes in cardiovascular health markers, including carotid artery pulse pressure. The study also found that carotid artery compliance decreased slightly in the strength training and combination groups and increased in the stretching group. However, the study found no carotid-femoral pulse wave velocity changes in the strength training and combination groups. Overall, the combination of strength training and aerobic exercise may be more effective than stretching alone in improving cardiovascular health.22 When compared to our study, the yoga asana intervention gave significant changes.\n\nA study by Thosar et al. (2015) involved 12 men participating in two randomized three-hour sitting trials. The first trial was a sitting trial where the subject was seated in one position without interruption The second trial was a sit-walk trial where the subjects walked on treadmills for five minutes at two mph for 30 minutes, 1.5-hour and 2.5-hour intervals during the sitting period. The study assessed the flow mediated dilation (FMD) of the SFA at baseline, one hour, two hours, and three hours in each trial. The results showed that there was no significant decline in the sitting trial in the FMD at each hour from baseline. The mean shear rate also showed a substantial decrease in the baseline across time. The sit-walk trial, however, showed a significant difference in the FMD and mean shear rate compared to the sit trial, indicating that walking during prolonged sitting may benefit arterial health.23 As compared to our study, both CA and SFA are measured. In our study all interventions – aside from the prolonged sitting – showed a minor decrease in CA velocity over time. When the seated intervention reached the last four hours, it will rise. Compared to time, the first and second hours exhibit statistical significance with a p value of <0.001, whereas the third and fourth hours exhibit significance but do not exhibit statistical significance (p value=0.020). There is no significant difference in SFA. But in our study, yoga asana showed effectiveness in both CA and SFA.\n\nWhen people sit for lengthy periods, their CA shear increases, whereas pranayama and yoga asana interventions showed decreases in CA shear for each hour. In CA shear, both time and group will not show any significance.\n\nLong periods of sitting have been shown to reduce stress in SFA, and yoga asana and pranayama interventions have also been shown to reduce stress over time. Both time and group do not exhibit any significant differences when compared.\n\nThe study by Carter et al. (2018) recruited 15 office workers (10 male). The participants sat for hours with two minutes of light intensity treadmill walking every 30 minutes and four hours sitting with an eight-minute light intensity treadmill walking break every 120 minutes. Middle cerebral artery measures were taken, and it was found that there was no significant difference in cardiorespiratory and hemodynamic measures, but a significant main effect was observed in cerebral blood flow. Post hoc analysis revealed a greater change in middle cerebral artery (MCA) during sitting compared to walking, with a considerable reduction in MCA observed in both the sitting and walking conditions.24 Compared to our study, the diameter significantly decreases in the sitting intervention, but velocity will increase similarly in SFA diameter and velocity decreases during the sitting intervention. Compared to time, the first and second hours exhibit statistical significance with a p value of <0.001, whereas the third and fourth hours demonstrated significance but did not exhibit statistical significance (p value=0.020). The yoga asana intervention showed significance compared to pranayama and prolonged sitting, though it didn’t reach statistical significance (p-value=0.014). In contrast, the yoga asana intervention showed significance but not statistical significance when compared to the group in pranayama intervention and prolonged sitting (p value=0.017). While in CA velocity, neither time nor velocity were significant.\n\nThe yoga asana intervention demonstrated improvement in SFA diameter; however, there was no difference between time and group when compared side by side. Each hour of the intervention resulted in a drop in velocity. The time points at the study’s start (0th hour) and at the 2nd hour demonstrate significance (p-value=0.026). However, while both the 0th and 2nd hours show significance with a p-value of 0.004, they still do not reach statistical significance. The group did not demonstrate any substantial response to any intervention. Hence, it shows that yoga asana will improve vascular function.\n\n\nConclusions\n\nBased on the results, we can observe that yoga asana improved vascular functions in prolonged sitting conditions. Many organizations in India and globally are promoting the practice of yoga asana in the workplace. This routine can promote the concept of interrupted sitting and ways to reduce it with efficient yoga asana practice without changing the work culture and provide better physical relief.",
"appendix": "Data availability\n\nHarvard Dataverse: Underlying data for ‘Impact of Yoga on the central and peripheral vascular function among desk-based workers’, https://doi.org/10.7910/DVN/GLGPCZ. 25\n\nHarvard Dataverse: CONSORT checklist for ‘Impact of Yoga on the central and peripheral vascular function among desk-based workers’, https://doi.org/10.7910/DVN/GLGPCZ. 25\n\nData are available under the terms of the Creative Commons Zero “No rights reserved” data waiver (CC0 1.0 Public domain dedication).\n\n\nReferences\n\nMesas AE, Guallar-Castillón P, León-Muñoz LM, et al.: Obesity-related eating behaviors are associated with low physical activity and poor diet quality in Spain. J. Nutr. 2012 Jul; 142(7): 1321–1328. PubMed Abstract | Publisher Full Text\n\nPettit-Mee RJ, Ready ST, Padilla J, et al.: Leg Fidgeting During Prolonged Sitting Improves Postprandial Glycemic Control in People with Obesity. Obesity (Silver Spring). 2021 Jul; 29(7): 1146–1154. PubMed Abstract | Publisher Full Text | Free Full Text\n\nPaz-Krumdiek M, Rodriguez-Vélez SG, Mayta-Tristán P, et al.: Association between sitting time and obesity: A population-based study in Peru. Nutr. Diet. 2020 Apr; 77(2): 189–195. PubMed Abstract | Publisher Full Text\n\nBiddle SJH, Edwardson CL, Gorely T, et al.: Reducing sedentary time in adults at risk of type 2 diabetes: process evaluation of the STAND (Sedentary Time ANd Diabetes) RCT. BMC Public Health. 2017 Jan; 17(1): 80. PubMed Abstract | Publisher Full Text | Free Full Text\n\nHelmink JHM, Kremers SPJ, van Brussel-Visser FN , et al.: Sitting time and Body Mass Index in diabetics and pre-diabetics willing to participate in a lifestyle intervention. Int. J. Environ. Res. Public Health. 2011 Sep; 8(9): 3747–3758. PubMed Abstract | Publisher Full Text | Free Full Text\n\nDharmastuti DP, Agni AN, Widyaputri F, et al.: Associations of Physical Activity and Sedentary Behaviour with Vision-Threatening Diabetic Retinopathy in Indonesian Population with Type 2 Diabetes Mellitus: Jogjakarta Eye Diabetic Study in the Community (JOGED.COM). Ophthalmic Epidemiol. 2018 Apr; 25(2): 113–119. PubMed Abstract | Publisher Full Text\n\nDillon K, Prapavessis H: REducing SEDENTary Behavior Among Mild to Moderate Cognitively Impaired Assisted Living Residents: A Pilot Randomized Controlled Trial (RESEDENT Study). J. Aging Phys. Act. 2020 Jun; 29(1): 27–35. PubMed Abstract | Publisher Full Text\n\nLabonté-LeMoyne E, Jutras MA, Léger PM, et al.: Does Reducing Sedentarity With Standing Desks Hinder Cognitive Performance? Hum. Factors. 2020 Jun; 62(4): 603–612. PubMed Abstract | Publisher Full Text\n\nOkely AD, Kontsevaya A, Ng J, et al.: WHO guidelines on physical activity and sedentary behavior. Vol. 3, Sports Medicine and Health Science. Elsevier; 2020; pp. 115–118.\n\nRajendran P, Rengarajan T, Thangavel J, et al.: The vascular endothelium and human diseases. Int. J. Biol. Sci. 2013; 9(10): 1057–1069. Publisher Full Text\n\nSvensson C, Eriksson P, Zachrisson H, et al.: High-Frequency Ultrasound of Multiple Arterial Areas Reveals Increased Intima Media Thickness, Vessel Wall Appearance, and Atherosclerotic Plaques in Systemic Lupus Erythematosus. Front. Med (Lausanne). 2020 Oct 9; 7. Publisher Full Text\n\nWest CR, AlYahya A, Laher I, et al.: Peripheral vascular function in spinal cord injury: a systematic review. Spinal Cord. 2013 Jan 27; 51(1): 10–19. PubMed Abstract | Publisher Full Text\n\nHildebrand JS, Gapstur SM, Gaudet MM, et al.: Moderate-to-vigorous physical activity and leisure-time sitting in relation to ovarian cancer risk in a large prospective U.S. cohort. Cancer Causes Control. 2015 Nov; 26(11): 1691–1697. PubMed Abstract | Publisher Full Text\n\nGarten RS, Hogwood AC, Weggen JB, et al.: Aerobic training status does not attenuate prolonged sitting-induced lower limb vascular dysfunction. Appl. Physiol. Nutr. Metab. 2019 Apr; 44(4): 425–433. Publisher Full Text\n\nAguirre-Betolaza AM, Mujika I, Loprinzi P, et al.: Physical activity, sedentary behavior, and sleep quality in adults with primary hypertension and obesity before and after an aerobic exercise program: Exerdiet-hta study. Life. 2020 Aug 17; 10(8): 1–13. Publisher Full Text\n\nRavichandran S, Sukumar S, Chandrasekaran B, et al.: Influence of Sedentary Behaviour Interventions on Vascular Functions and Cognitive Functions in Hypertensive Adults—A Scoping Review on Potential Mechanisms and Recommendations. Int. J. Environ. Res. Public Health. 2022 Nov 16; 19(22): 15120. Publisher Full Text\n\nMuktibodhananda S: Hatha yoga pradipika. Sri Satguru Publications; 2012.\n\nIyengar BKS: Light on the Yoga Sūtras of Patañjali.359.\n\nRoss A, Thomas S: The Health Benefits of Yoga and Exercise: A Review of Comparison Studies. J. Altern. Complement. Med. 2010 Jan; 16(1): 3–12. PubMed Abstract | Publisher Full Text\n\nHagins M, Moore W, Rundle A: Does practicing hatha Yoga satisfy recommendations for intensity of physical activity which improves and maintains health and cardiovascular fitness? BMC Complement. Altern. Med. 2007 Nov; 7: 40. PubMed Abstract | Publisher Full Text | Free Full Text\n\nPearson MJ, Smart NA: Aerobic Training Intensity for Improved Endothelial Function in Heart Failure Patients: A Systematic Review and Meta-Analysis. Vol. 2017, Cardiology Research and Practice. Hindawi Limited; 2017.\n\nCortez-Cooper MY, Anton MM, Devan AE, et al.: The effects of strength training on central arterial compliance in middle-aged and older adults. Eur. J. Prev. Cardiol. 2008; 15(2): 149–155.\n\nThosar SS, Bielko SL, Mather KJ, et al.: Effect of prolonged sitting and breaks in sitting time on endothelial function. Med. Sci. Sports Exerc. 2015 Apr 25; 47(4): 843–849. PubMed Abstract | Publisher Full Text\n\nCarter SE, Draijer R, Holder SM, et al.: Regular walking breaks prevent the decline in cerebral blood flow associated with prolonged sitting. J. Appl. Physiol. 2018; 125: 790–798. PubMed Abstract | Publisher Full Text Reference Source\n\nSukumar S: Impact of yoga on the central and peripheral vascular function among desk-based workers. Harvard Dataverse. 2023; V1. Publisher Full Text"
}
|
[
{
"id": "351086",
"date": "23 Dec 2024",
"name": "Mansoor Rahman",
"expertise": [
"Reviewer Expertise Physical Activity",
"Cerebral Palsy. Adapted Physical education",
"traditional Indian games for Physical activity for individuals with neurodevelopmental disabilities."
],
"suggestion": "Approved With Reservations",
"report": "Approved With Reservations\n\ninfo_outline\nAlongside their report, reviewers assign a status to the article:\n\nApproved The paper is scientifically sound in its current form and only minor, if any, improvements are suggested\n\nApproved with reservations\nA number of small changes, sometimes more significant revisions are required to address specific details and improve the papers academic merit.\n\nNot approved Fundamental flaws in the paper seriously undermine the findings and conclusions\n\nIs the work clearly and accurately presented and does it cite the current literature? - There is mismatch in discussion references with duration of this study intervention with the existing evidences chosen to agree the study findings. Are all the source data underlying the results available to ensure full reproducibility? - Intervention procedures and descriptions were not mentioned in detail. Are the conclusions drawn adequately supported by the results? - With single session to conclude requires a strong evidences with theoretical background.\n\nIs the work clearly and accurately presented and does it cite the current literature? No\n\nIs the study design appropriate and is the work technically sound? Partly\n\nAre sufficient details of methods and analysis provided to allow replication by others? Partly\n\nIf applicable, is the statistical analysis and its interpretation appropriate?\nYes\n\nAre all the source data underlying the results available to ensure full reproducibility? No\n\nAre the conclusions drawn adequately supported by the results? Partly",
"responses": []
}
] | 2
|
https://f1000research.com/articles/13-277
|
https://f1000research.com/articles/12-118/v1
|
01 Feb 23
|
{
"type": "Software Tool Article",
"title": "Med-ImageTools: An open-source Python package for robust data processing pipelines and curating medical imaging data",
"authors": [
"Sejin Kim",
"Michal Kazmierski",
"Kevin Qu",
"Jacob Peoples",
"Minoru Nakano",
"Vishwesh Ramanathan",
"Joseph Marsilla",
"Mattea Welch",
"Amber Simpson",
"Benjamin Haibe-Kains",
"Sejin Kim",
"Michal Kazmierski",
"Kevin Qu",
"Jacob Peoples",
"Minoru Nakano",
"Vishwesh Ramanathan",
"Joseph Marsilla",
"Mattea Welch",
"Amber Simpson"
],
"abstract": "Background: Machine learning and AI promise to revolutionize the way we leverage medical imaging data for improving care but require large datasets to train computational models that can be implemented in clinical practice. However, processing large and complex medical imaging datasets remains an open challenge. Methods: To address this issue, we developed Med-ImageTools, a new Python open-source software package to automate data curation and processing while allowing researchers to share their data processing configurations more easily, lowering the barrier for other researchers to reproduce published works. Use cases: We have demonstrated the efficiency of Med-ImageTools across three different datasets, resulting in significantly reduced processing times. Conclusions: The AutoPipeline feature will improve the accessibility of raw clinical datasets on public archives, such as the Cancer Imaging Archive (TCIA), the largest public repository of cancer imaging, allowing machine learning researchers to process analysis-ready formats without requiring deep domain knowledge.",
"keywords": [
"medical imaging",
"deep learning",
"open source",
"data processing",
"dicom",
"nifti",
"nnunet"
],
"content": "Introduction\n\nRadiology is a powerful modality of data for clinical work — it gives clinicians the ability to see the inner workings of the human body, that cannot be seen from the outside.1 They can inspect in 2D or 3D, the anatomy surrounding the disease, enabling key information to make life-altering clinical decisions. While regular images taken on cameras or phones are stored in a variety of accessible formats, medical images are encoded in the Digital Imaging and Communications in Medicine (DICOM) standard file format.2\n\nThe DICOM standard was developed in the 1980s due to the increasing need for an interoperable standard for 3D medical images across various manufacturers.2 A key feature of DICOM is the plethora of metadata fields that store information beyond imaging data, such as patient information, clinical variables, and acquisition parameters. As modern medical practice evolved over the years, the DICOM standard has grown to accommodate more metadata fields and encompass new imaging modalities or therapy information.2 This type of data format is unsuitable for imaging analysis as the relevant voxel array must be manually accessed through DICOM hierarchy.2 Furthermore, 3D scans are acquired on a slice-by-slice basis; thus, researchers must stitch together data from multiple files to create one 3D image, adding delays caused by disk reads and consolidation processes.2\n\nSpecialties that rigorously use imaging data are heavily reliant on DICOM, one of which is radiation oncology. Images are used along every step of the clinical workflow: from deriving a precise diagnosis, to designing personalized radiation therapy plans, and delivering each radiation dose with the appropriate alignment and orientation with brief scans. While the defined standard serves as a good guideline, each manufacturer has slightly different implementations. This is especially the case for DICOM-RT (radiotherapy), a subset of modalities for communicating radiotherapy data.2 The DICOM-RT standard includes additional modalities such as RTStruct for contour data, and RTDose for radiotherapy dose maps and dose-volume histograms (DVH).2 While the broad adoption of the DICOM standard to accommodate for various use cases has allowed it to become the defacto standard for encoding, storing, and transferring of medical images, its comprehensive nature has made it difficult for researchers to navigate for the purposes of imaging projects.2\n\n\nCurrent workflows\n\nThe Cancer Imaging Archive (TCIA) (RRID:SCR_008927) is one of the largest public repositories of DICOM images available, with over 140 datasets consisting of more than 60,000 patients.3 The datasets undergo a quality assurance process to ensure the recorded clinical variables are coherent and the DICOM files are not missing any important metadata fields.3 These stringent processes and infrastructure have allowed TCIA to become one of the most comprehensive repositories for biomedical imaging datasets, inviting researchers from different fields to explore new ideas and methods on high quality datasets.3\n\nWhile the underlying data and its annotations are of clinical quality, processing the dataset for subsequent analysis requires a non-trivial amount of effort: manually reorganizing directories and matching radiation therapy structures (RTStruct), referred to as DICOM-RT contours, and radiation therapy plans (RTDose/RTPlan) to its corresponding images.4 This is partly due to the inherently complex nature of clinical datasets, as data is collected on the basis of need and iterative improvements, not structured scientific inquiries. It is also sometimes due to the lack of familiarity from machine learning (ML) and artificial intelligence (AI) researchers in handling the DICOM files for their analytical pipelines. Typical AI imaging datasets have pairwise associations of one image to a single ground-truth label.5 However, one patient in clinical datasets may have multiple RTStruct and RTDose files with one imaging acquisition, one RTStruct and RTDose with multiple images, or worst of all, multiple RTStruct and RTDose files with multiple images.6 In any of these cases, the directories are not always intuitively structured to help researchers understand which files correspond with another.\n\nOnce researchers have successfully curated the dataset into an organized structure for analysis, in order to process the raw dataset into analysis-ready format, they must choose from a variety of processing parameters, ranging from voxel spacing, RTStruct name parsing, and hounsfield unit (HU) window levels, based on the design of their analysis. While these implicit decisions for image processing are often arbitrary, they can greatly impact model training and performance, but are not transparently disclosed in publications.7 This leads to the difficulty of reproducibility of medical deep learning research, adding another deterrence to clinical adoption.\n\nFurthermore, there are a limited number of software packages that researchers can use to quickly parse DICOM-RT files into analysis-ready arrays (Table 1). Chief among those, SlicerRT,8 an extension of 3Dslicer9 (RRID:SCR_005619), an open-source DICOM visualization tool, has been widely used by the medical imaging community. Despite its broad adoption, it is not easily interfaced with Python, the most popular language for machine learning and deep learning.10 RT-Utils is a lightweight, open-source Python package designed to handle RTStruct files with relative ease and simplicity, allowing users to easily export contours into segmentation masks in arrays. However, the functionalities of RT-Utils are limited to the RTStruct modality. PlatiPy is a recent processing library and analysis toolkit for medical imaging, mainly designed for the context of radiation therapy. It features a comprehensive set of image manipulation functions such as registration and atlas-based segmentation methods, allowing researchers the flexibility to process imaging data into any format they need. However, PlatiPy does not solve the inherent complexity of clinical datasets, and researchers must spend hours reorganizing the data into a structured set of samples and labels. The current landscape of open-source medical imaging tools highlights the need for a native Python package that can parse large DICOM/DICOM-RT datasets to an analysis-ready format for ML/AI development in a consistent, reproducible workflow.\n\nRTSTRUCTs: DICOM-RT Contours; RTDOSEs: DICOM-RT Dose; CT: Computed Tomography; MRI: Magnetic Resonance Imaging; NifTI: Neuroimaging Informatics Technology Initiative; Nrrd: Nearly raw raster data; DICOM: Digital Imaging and Communications in Medicine.\n\nTo address the limitations of the current software packages used to process medical images, we developed Med-ImageTools,11 a new Python package designed to help researchers transform complex medical datasets into analysis-ready format with few lines of code. It is also focused on helping researchers develop transparent and reproducible medical image processing pipelines by addressing most of the boilerplate code required for image transformations and processing parallelization. While Med-ImageTools has many modular functions for image, contour, and dose input/output (IO) built on popular frameworks such as SimpleITK, TorchIO and PyDicom, such functionalities are redundant and available in other open-source packages as well. Our main contribution is in the development of AutoPipeline and will mainly discuss its functionalities and implementation.\n\n\nMethods\n\nAutoPipeline is the main feature of Med-ImageTools that allows users to easily process raw DICOM clinical datasets into analysis-ready Nrrd or NifTI files, which are commonly used file formats for 3D volumetric data. It is interfaced using the command line, so the user only needs to submit a single command into the terminal to execute the three core steps in the AutoPipeline process (Figure 1):\n\n1. Crawl: The crawler opens every DICOM file in the dataset, indexing important metadata, such as unique identifiers, modality information, and references to other modalities. This produces a database of every unique image and DICOM-RT modality.\n\n2. Connect: In this step, each patient’s indices of unique files of different modalities are connected to form one coherent sample. There are various heuristics the user may choose to connect samples. The default option is through DICOM metadata, as datasets derived from clinical practice are expected to have corresponding metadata that references unique identifiers of the parent image or RTPlan. Alternative heuristics allow users to deal with anomalies with corrupted or missing metadata.\n\n3. Process: All identified samples are processed according to imaging and transformation parameters defined by the user.\n\nRaw datasets are indexed by the Crawler module and automatically processed by AutoPipeline. DICOM: Digital Imaging and Communications in Medicine; RTSTRUCTs: DICOM-RT Contours; PET: Positron emission tomography. NifTI: Neuroimaging Informatics Technology Initiative; CSV: Comma-separated values.\n\nThe AutoPipeline feature can be directly interfaced once Med-ImageTools is installed. To install Med-ImageTools, the recommended method is via the PyPI package repository in a virtual environment. Running the ‘pip install med-imagetools’ command will install the latest version of Med-ImageTools and its associated dependencies. Now, whenever the user is in the virtual environment where Med-ImageTools is installed, they can directly interact with the AutoPipeline feature in the command line. The simplest way to use it is through ‘autopipeline input_directory output_directory’. This will automatically process the dataset located in ‘input_directory’ and process them at ‘output_directory’ using the default parameters. An extended tutorial of AutoPipeline and all its associated parameters are available here. At the present time, there are no minimum system requirements for Med-ImageTools as it will run regardless of number of processor cores or memory (RAM). However, if the researcher can leverage greater number of cores and RAM, it will allow the AutoPipeline to be parallelized and process the data faster.\n\nAs the crawl and process steps are computationally intensive, all steps in AutoPipeline are automatically parallelized to efficiently leverage all available computational resources. While the output of the AutoPipeline processing can result in terabytes of images, the crawl is limited to a few kilobytes of a metadata database, making it an ideal asset to share with other researchers as a detailed descriptor of a medical imaging dataset. We therefore propose to attach the crawled metadata spreadsheet to large TCIA datasets to allow Med-ImageTools users to process large datasets much faster and more efficiently. These databases are expected to save up to 1000 core-hours of crawling per dataset, accumulating over 2000 core-hours of computation saved per user. By standardizing a commonly repeated imaging processing pipeline into a single unified package, we hope to improve the reproducibility and transparency of future medical imaging research.\n\n\nUse cases\n\nWe showcase the value of the AutoPipeline implemented in Med-ImageTools v1.0.0 for processing three medical imaging datasets, namely Pancreatic-CBCT-SEG from TCIA, liver metastasis private dataset and RADCURE pending public release on TCIA, in order of complexity and sizes. In each use case, we initially describe the process.\n\n40 patients with abdominal CBCT scans and their associated contours of regions of interest and other organs, publicly available on TCIA\n\nThe Pancreatic-CT-CBCT-SEG dataset was processed twice using AutoPipeline: once from scratch, and once using the pre-crawled dataset, available on the tcia-crawls branch of Med-ImageTools. Processing from scratch, the dataset took 10.77 core-hours (10:46), whereas using a pre-crawled database allowed the processing to finish in 9.14 core-hours (9:08) (Figure 2a). The pre-crawled database reduced processing time by 1.63 core-hours, representing an 18% increase in total processing speed or 2 minutes 27 seconds per patient. The time saved from pre-crawled databases is not a substantial quantity for datasets with less than 100 patients. However, when scaled up to larger TCIA datasets such as OPC-Radiomics13 (n=606) and NLST14 (n=26254), it can save researchers 24.7 core-hours and 1072 core-hours, respectively (Figure 2b). While it may vary depending on the research infrastructure utilized, these databases can result in significant savings in billing.\n\n97 patients with abdominal CT scans and their associated contours of liver and gross tumour volumes (GTV), data access described in data availability section below.\n\nThe liver metastasis dataset was processed using the Slicer API to export a CT scan along with segmentations of the liver and GTVs for each patient. In the initial DICOM dataset, each patient had a single RTSTRUCT file along with one or more CT series, one of which was referenced by the RTSTRUCT. The first step of the process was to load the entire DICOM dataset into the Slicer DICOM database via the graphical user interface (GUI), to make the data available inside the Slicer scripting environment. Then, we wrote a script to export an initial set of candidate segmentations for the liver and GTVs for each patient, along with the referenced CT scan. This script leveraged the Slicer API to identify the CT series that was referenced by the patient’s RTSTRUCT and used an ad hoc string filtering function to identify candidate segmentations. Finally, we iteratively refined the set of exported RTSTRUCT contours on a patient-by-patient basis, based on visual checking, and physician feedback. Although this step was time-consuming, there is, at present, no substitute for manual verification to ensure data quality and correctness. On the other hand, the initial database construction and export script using the Slicer API achieved results roughly analogous to the automatic output of AutoPipeline. The export script is approximately 170 lines long and took 7.5 hours to run on the whole dataset-this script is available on our GitHub repository. In contrast, on the same 6-core machine (16GB RAM, Windows 10), AutoPipeline took only 2.3 hours with 1 command line, mainly accelerated by parallelization (Figure 2c).\n\n3,219 patients with head and neck CT scans and their associated contours of organs at risk (OAR) for radiotherapy, pending public release on TCIA.\n\nThe RADCURE dataset is a large dataset of 3,219 head and neck cancer patients and their radiotherapy planning data. The dataset was extracted from two separate treatment planning software systems, meaning the directories and DICOM metadata were structured differently. The directories from each system were restructured using general heuristics, and any abnormal cases were flagged and manually organized. We used SimpleITK and PyDicom to extract the imaging and contour data from the DICOM files, which underwent a similar iterative process as the liver metastatic dataset. The script is over 1000 lines long and takes 30-40 hours to run on the whole dataset using a single core-this script is available on our GitHub repository. AutoPipeline scales dramatically based on the number of cores available, which enables the entire dataset to be processed in 40 minutes using 32 cores, automatically managing the directories from different systems and extracting the contours (Figure 2d).\n\nOne caveat of these comparisons is that the iterative process of filtering appropriate contours, which can add weeks to months of cooperation between the researchers and the physicians, were already conducted in the original processing steps. Various data cleaning steps that require human intervention, such as sorting contour names or selecting specific subseries acquisitions, cannot be fully automated for the foreseeable future. The Med-ImageTools team aims to add features to the package that will assist researchers in these steps, such as adding a flag to visualize all unique contours without requiring code, adding subseries detection to the crawler, and publishing a set of regular expressions (regex) that can be used to automatically choose contours from prominent head and neck datasets on TCIA. Also, these comparisons do not take into account the time it takes for researchers to develop the manual processing scripts. Hence, the actual amount of time saved for the researcher may be greater than the reported times.\n\n\nDiscussion\n\nAlthough ML and AI promise to revolutionize the way we leverage medical imaging data for improving care, they require large datasets to train computational models that can be implemented in clinical practice. However, processing large and complex medical imaging datasets remains an open challenge. To address this issue, we developed Med-ImageTools, a new open-source software package to automate data curation and processing while allowing researchers to share their data processing configurations more easily, lowering the barrier for other researchers to reproduce published works. The AutoPipeline feature will improve the accessibility of raw clinical datasets on public archives, such as TCIA, allowing machine learning researchers to process analysis-ready formats without requiring deep domain knowledge.\n\nWhile our package aims to address challenges encountered across a few medical imaging labs, we acknowledge that there may be infinite other issues that may arise in DICOM datasets. This is one of the key reasons why our package is open-source for community involvement and contribution. Also, as stated previously, there are certain onerous tasks that cannot be automated and must undergo human supervision. These aspects of researcher-clinician collaboration are an inevitable part of medical imaging research and are subject to delay.\n\nNo single solution can completely solve the reproducibility crisis of medical deep learning research, due to a variety of issues ranging from ambiguous data processing techniques to stochasticity of model training. However, community-centered open-source solutions and increased clinical adherence to data standards, such as contour nomenclature,15 can incrementally improve research quality and reproducibility, and make medical deep learning research more accessible for everyone.",
"appendix": "Data availability\n\nThe Pancreatic-CT-CBCT-SEG dataset is available on The Cancer Imaging Archive at: https://doi.org/10.7937/TCIA.ESHQ-4D90.\n\nThe pre-crawled Med-ImageTools database of the Pancreatic-CT-CBCT-SEG dataset is available on GitHub at: https://github.com/bhklab/med-imagetools/raw/tcia-crawls/csvs/imgtools_Pancreatic-CT-CBCT-SEG.csv.\n\nThe liver dataset was collected as part of the study titled Radiomics artificial intelligence modelling for prediction of local control for colorectal liver metastases treated with radiotherapy 16 with Institutional Review Board and informed consent. Interested individuals should contact the senior author for access to the dataset, which will be made available following standard institutional rules for IRB and HIPAA.\n\nThe RADCURE dataset is pending public release on TCIA and the article will be updated upon release. Data published on TCIA are subject to TCIA data usage policies and restrictions.\n\n\nReferences\n\nBrant WE, Helms CA: Fundamentals of Diagnostic Radiology.2012.\n\nMildenberger P, Eichelberg M, Martin E: Introduction to the DICOM standard. Eur. Radiol. 2002; 12: 920–927. PubMed Abstract | Publisher Full Text\n\nClark K, et al.: The Cancer Imaging Archive (TCIA): maintaining and operating a public information repository. J. Digit. Imaging. 2013; 26: 1045–1057. PubMed Abstract | Publisher Full Text | Free Full Text\n\nLaw MYY, Liu B: DICOM-RT and Its Utilization in Radiation Therapy. RadioGraphics. 2009; 29: 655–667. PubMed Abstract | Publisher Full Text\n\nDeng J, et al.: ImageNet: A large-scale hierarchical image database.In 2009 IEEE Conference on Computer Vision and Pattern Recognition.2009; pp. 248–255. Publisher Full Text\n\nVallières M, Kay-Rivest E, et al.: Data from Head-Neck-PET-CT. The Cancer Imaging Archive. 2017. Publisher Full Text\n\nSabottke CF, Spieler BM: The Effect of Image Resolution on Deep Learning in Radiography. Radiol. Artif. Intell. 2020; 2: e190015.\n\nPinter C, Lasso A, Wang A, et al.: SlicerRT: radiation therapy research toolkit for 3D Slicer. Med. Phys. 2012; 39: 6332–6338. PubMed Abstract | Publisher Full Text\n\nFedorov A, et al.: 3D Slicer as an image computing platform for the Quantitative Imaging Network. Magn. Reson. Imaging. 2012; 30: 1323–1341. PubMed Abstract | Publisher Full Text | Free Full Text\n\nYli-Heikkilä M:Data Science Languages. DSNS '19. University of Helsinki;2019.\n\nKazmierski M, Kim S, Ramanathan V, et al.: bhklab/med-imagetools: v4.4.0. Zenodo. 2022. Publisher Full Text\n\nCamp B: Pancreatic-CT-CBCT-SEG - The Cancer Imaging Archive (TCIA) Public Access - Cancer Imaging Archive Wiki.Publisher Full Text\n\nKwan JYY, et al.: Radiomic Biomarkers to Refine Risk Models for Distant Metastasis in HPV-related Oropharyngeal Carcinoma. Int. J. Radiat. Oncol. Biol. Phys. 2018; 102: 1107–1116. PubMed Abstract | Publisher Full Text\n\nClark K: National Lung Screening Trial.Publisher Full Text\n\nMayo CS, et al.: American Association of Physicists in Medicine Task Group 263: Standardizing Nomenclatures in Radiation Oncology. Int. J. Radiat. Oncol. Biol. Phys. 2018; 100: 1057–1066. PubMed Abstract | Publisher Full Text | Free Full Text\n\nHu R, Chen I, et al.: Radiomics artificial intelligence modelling for prediction of local control for colorectal liver metastases treated with radiotherapy. Phys. Imaging Radiat. Oncol. 2022; 24: 36–42. PubMed Abstract | Publisher Full Text | Free Full Text\n\nCranmer K, Kreiss S: decouple software associated to arXiv:1401.0080. [Code] Zenodo. 2014. Publisher Full Text"
}
|
[
{
"id": "189015",
"date": "17 Oct 2023",
"name": "Rachel Sparks",
"expertise": [
"Reviewer Expertise Software development",
"open source software",
"medical image computing"
],
"suggestion": "Approved With Reservations",
"report": "Approved With Reservations\n\ninfo_outline\nAlongside their report, reviewers assign a status to the article:\n\nApproved The paper is scientifically sound in its current form and only minor, if any, improvements are suggested\n\nApproved with reservations\nA number of small changes, sometimes more significant revisions are required to address specific details and improve the papers academic merit.\n\nNot approved Fundamental flaws in the paper seriously undermine the findings and conclusions\n\nI think the motivation and justification for why a light weight Python module to support data loading and preprocessing of radiotherapy DICOM formats was very well articulated and clear. And I applaud the authors on making such a useful tool available to the community.\nHowever, I have two major concerns that must be addressed prior to this article being acceptable for indexing.\nFirst there are no details as to the design of Med-Image-Tools this includes important information such as:\n(1) dependencies on other python packages\n(2) what functions or other tools are being leveraged within the tool and how these are arranged.\n(3) the structure of the outputs of this pipeline specifically the csv file corresponding to the 'Dataset Index', the imaging and non-imaging feature outputs. For instance for imaging data how are pre-operative scans and segmentations saved and named (is there a convention is this something the user must write)\n(4) While there is a reference to the tutorial about the processing pipeline I think a high level overview listing the possible options would be appropriate to help the reader understand what this tool can offer.\nMy other concern is related to the use cases. While lines of code is an easy metric to report it is mostly meaningless, some lines of code may be boilerplate or easy to write while others may obviously be different. It is also easy to pad or alter depending on the level of skill of the programmer. I think reporting run time is far more meaningful to your argument and I would stick to these.\nFinally, specifically, for the RADCURE dataset, it is not clear to me why you are comparing run time on a single core versus run time on a 32 core machine. I suspect the reported run time for the comparison could be significantly reduced by using a very simple parallelization which should be achievable tools available in python. I would suggest a fairer comparison for this specific example, it wont detract from the best argument you have which is there you are presenting a nice package so that other developers and researchers do not have to spend time re-writing code. Highlighting where some time savings may be gained is in some ways a secondary advantage.\nFinally, as a minor note the statement that SlicerRT is not easily interfaced with python is not a fully justified statement. One can use the python interface in Slicer to run python scripts and even run from the command prompt by called s'licer.exe --python-code'. However, I think you can modify this to make a true and equally motivated statement which is that SlicerRT requires one to install Slicer/SlicerRT and stay within the slicer ecosystem to process data. Your tool lacks such dependencies so can be a more light weight and does not require running python through another software system.\n\nIs the rationale for developing the new software tool clearly explained? Yes\n\nIs the description of the software tool technically sound? No\n\nAre sufficient details of the code, methods and analysis (if applicable) provided to allow replication of the software development and its use by others? No\n\nIs sufficient information provided to allow interpretation of the expected output datasets and any results generated using the tool? No\n\nAre the conclusions about the tool and its performance adequately supported by the findings presented in the article? Yes",
"responses": [
{
"c_id": "11384",
"date": "07 May 2024",
"name": "Benjamin Haibe-Kains",
"role": "Author Response",
"response": "We thank Dr Sparks for her feedback regarding our manuscript titled ‘Med-ImageTools: An open-source Python package for robust data processing pipelines and curating medical imaging data’. We have added our responses to their comments here in subdivided sections of comments/questions (Q), in bold, our corresponding responses/answers (A), and any relevant sections of the manuscript as an italicized bullet point, with changes emphasized in bold. Q: I think the motivation and justification for why a light weight Python module to support data loading and preprocessing of radiotherapy DICOM formats was very well articulated and clear. And I applaud the authors on making such a useful tool available to the community. A: We thank the reviewer for their recognition of Med-ImageTools’ contribution to the medical imaging community and positively evaluating the authors’ works. Q: However, I have two major concerns that must be addressed prior to this article being acceptable for indexing. First there are no details as to the design of Med-Image-Tools this includes important information such as: (1) dependencies on other python packages A: We thank the reviewer for pointing out the need to communicate the technical dependencies of Med-ImageTools. We have updated our ‘AutoPipeline’ section as follows, to include information on how to find the list of package dependencies. Running the ‘ pip install med-imagetools’ command will install the latest version of Med-ImageTools and its associated dependencies defined by the requirements.txt file. The ‘Current workflows’ section also contains a concise description of what frameworks Med-ImageTools leverages such as follows: While Med-ImageTools has many modular functions for image, contour, and dose input/output (IO) built on popular frameworks such as SimpleITK, TorchIO and PyDicom… Q: (2) what functions or other tools are being leveraged within the tool and how these are arranged. A: We thank the reviewer for calling attention to the need to clarify how other tools/packages function within Med-ImageTools. We have added some information to our ‘AutoPipeline’ section to clarify major operations performed using external tools/packages as follows. For a granular understanding of the integration behind every external tool, we defer the reader to learn about the design choices made throughout the package and its implications by gaining a deeper understanding of the source code available through our GitHub repository. 1. Crawl: The crawler opens every DICOM file in the dataset using Pydicom, indexing important metadata, such as unique identifiers, modality information, and references to other modalities. This produces a database of every unique image and DICOM-RT modality. 3. Process: All identified samples are processed according to imaging and transformation parameters defined by the user. The user can configure parameters such as the pixel spacing in mm(s), which specific modalities to process, the number of cores they want to use for multiprocessing, and define nnU-Net specific flags as well. The images are manipulated using SimpleITK without requiring user intervention. As the crawl and process steps are computationally intensive, all steps in AutoPipeline are automatically parallelized using the joblib backend to efficiently leverage all available computational resources. Q: (3) the structure of the outputs of this pipeline specifically the csv file corresponding to the 'Dataset Index', the imaging and non-imaging feature outputs. For instance for imaging data how are pre-operative scans and segmentations saved and named (is there a convention is this something the user must write) A: We thank the reviewer for this question and for highlighting the lack of clarity of our description of AutoPipeline. The ‘Dataset Index’ is designed to be an autogenerated output of the Crawler and does not require any user interaction to be created; this is simply there to ensure convenient user experience. We have made this and other aspects of the outputs clearer by adding the following sentences to the ‘AutoPipeline’ section. Med-ImageTools also generates files that are not analysis-ready images, such as the autogenerated ‘Dataset Index’ (Figure 1), and stores them in a folder named “.imgtools” at the path of the ‘input_directory’. This is to ensure a convenient user experience and intuitive folder structure by hiding extraneous components. Q: (4) While there is a reference to the tutorial about the processing pipeline I think a high level overview listing the possible options would be appropriate to help the reader understand what this tool can offer. A: We agree with the reviewer that adding this information to our description of AutoPipeline can help readers quickly understand the tool’s strengths before interfacing with the GitHub repository or ReadTheDocs. We have updated the ‘AutoPipeline’ section accordingly. 3. Process: All identified samples are processed according to imaging and transformation parameters defined by the user. The user can configure parameters such as: the pixel spacing in mm(s), which specific modalities to process, the number of cores they want to use for multiprocessing, and define nnU-Net specific flags as well. The images are manipulated using SimpleITK without requiring user intervention. Q: My other concern is related to the use cases. While lines of code is an easy metric to report it is mostly meaningless, some lines of code may be boilerplate or easy to write while others may obviously be different. It is also easy to pad or alter depending on the level of skill of the programmer. I think reporting run time is far more meaningful to your argument and I would stick to these. A: We appreciate the reviewer’s consideration in the metrics reported in Figure 2. However, we believe this metric emphasizes Med-ImageTools’ strength allowing users to process DICOM datasets with a single command line interaction, in contrast to writing dozens of lines of Python/R/MATLAB scripts. One particular reason why we believe a single command line is useful to highlight is to express the ease-of-use for users that may not be comfortable writing their own code for DICOM to NifTI/Nrrd processing, especially if they lack deep domain knowledge. Another, perhaps more, exciting reason is to show the potential of building automated workflows on top of Med-ImageTools. If one were to build an end-to-end automated pipeline starting from clinical DICOM datasets to outputting an inference-ready deep learning model, they could easily develop a reproducible processing step by configuring only the command line interaction of Med-ImageTools, making debugging and custom configurations simpler since the developer would not have to rely on a static script. We have elaborated on Med-ImageTools’ contributions for improved scientific reproducibility in the ‘Discussion’ section as follows. The AutoPipeline feature will improve the accessibility of raw clinical datasets on public archives, such as TCIA, allowing machine learning researchers to process analysis-ready formats without requiring deep domain knowledge. Another exciting potential of Med-ImageTools lies in building automated workflows using AutoPipeline. For a researcher to build an end-to-end automated pipeline starting from clinical DICOM datasets to outputting an inference-ready deep learning model, they could easily develop a reproducible processing step by configuring only the command line interaction of Med-ImageTools, making debugging and custom configurations simpler since the developer would not have to rely on a static script. No single solution can completely solve the reproducibility crisis of medical deep learning research, due to a variety of issues ranging from ambiguous data processing techniques to stochasticity of model training. However, community-centered open-source solutions and increased clinical adherence to data standards, such as contour nomenclature,15 can incrementally improve research quality and reproducibility, and make medical deep learning research more accessible for everyone. Q: Finally, specifically, for the RADCURE dataset, it is not clear to me why you are comparing run time on a single core versus run time on a 32 core machine. I suspect the reported run time for the comparison could be significantly reduced by using a very simple parallelization which should be achievable tools available in python. I would suggest a fairer comparison for this specific example, it wont detract from the best argument you have which is there you are presenting a nice package so that other developers and researchers do not have to spend time re-writing code. Highlighting where some time savings may be gained is in some ways a secondary advantage. A: We thank the reviewer for their comment about Figure 2d. The goal of the comparison was not only to show a reduction in processing time based on parallelization, but to show that our package’s design does not bottleneck any multiprocessing backends and brings meaningful acceleration on large datasets. In other words, we wanted to show that a 32x increase in computational hardware was going to bring at least 32x improvement in performance as observed in Figure 2d. This is in contrast to Figure 2c which does not show similar levels hardware acceleration, but its usefulness in parsing private, unseen data. We have expanded on these ideas in the manuscript as follows: In contrast, on the same 6-core machine (16GB RAM, Windows 10), AutoPipeline took only 2.3 hours with 1 command line, mainly accelerated by parallelization ( Figure 2c). The authors involved in validating Med-ImageTools' effectiveness on this private dataset had no active involvement in the development of the package before its application. This highlights the package’s robustness on unseen data and its potential utility for multi-centre collaborations to ensure consistent processing. One use case might be to enable federated learning platforms to automatically process each node’s datasets without requiring any user intervention. AutoPipeline scales dramatically based on the number of cores available, which enables the entire dataset to be processed in 40 minutes using 32 cores, automatically managing the directories from different systems and extracting the contours ( Figure 2d). These results demonstrate Med-ImageTools' design does not bottleneck any multiprocessing backends and brings meaningful acceleration on very large datasets, such as RADCURE. Q: Finally, as a minor note the statement that SlicerRT is not easily interfaced with python is not a fully justified statement. One can use the python interface in Slicer to run python scripts and even run from the command prompt by called s'licer.exe --python-code'. However, I think you can modify this to make a true and equally motivated statement which is that SlicerRT requires one to install Slicer/SlicerRT and stay within the slicer ecosystem to process data. Your tool lacks such dependencies so can be a more light weight and does not require running python through another software system. A: We thank the reviewer for this point. We have revised the relevant section from ‘Current Workflows’ as follows to better justify our statement. Despite its broad adoption, batch data processing with Slicer requires custom scripting in Python, to be executed in the Slicer ecosystem. Rather than simply installing a package within their Python environment, users must install the Slicer application and add any other dependencies, not provided by Slicer, into the application environment. As a result, machine learning projects relying on Python for data preprocessing will have their code fragmented across multiple environments–the Slicer environment for data processing, and another Python environment for data analysis and machine learning."
}
]
},
{
"id": "189038",
"date": "17 Oct 2023",
"name": "Johann Faouzi",
"expertise": [
"Reviewer Expertise Open source software",
"Python programming",
"machine learning",
"medical applications"
],
"suggestion": "Approved With Reservations",
"report": "Approved With Reservations\n\ninfo_outline\nAlongside their report, reviewers assign a status to the article:\n\nApproved The paper is scientifically sound in its current form and only minor, if any, improvements are suggested\n\nApproved with reservations\nA number of small changes, sometimes more significant revisions are required to address specific details and improve the papers academic merit.\n\nNot approved Fundamental flaws in the paper seriously undermine the findings and conclusions\n\nThe authors present their Python package for data processing pipelines and curating medical imaging data in this manuscript. Overall the manuscript is well written and clearly presents the main contributions of this package to the community. Nonetheless, some points on the software side could be improved. Please find my comments below:\n* How does the output format used compare to other formats such as other popular formats such as BIDS (Brain Imaging Data Structure, which is maybe less general since it's intended for brain imaging data)?\n* Figure 2b (and the corresponding text): Raw saved running time, even though useful, is irrelevant without knowing the total running time. Please provide also the saved time as a percentage of the total running time.\n* Figures 2c and 2d: \"1 lines\" => \"1 line\"\n* The authors should consider renaming the \"master\" branch as the \"main\" branch (see https://sfconservancy.org/news/2020/jun/23/gitbranchname/).\n* Is there a reason for Python 3.8 not being tested in the continuous integration (CI) (maybe a limit on the number of jobs)? By the way, Python 3.7 is no longer supported and Python 3.11 has been released for over 9 months (https://devguide.python.org/versions/). The Python versions tested in the CI should probably be updated.\n* Flake8 is installed in the CI but never used. I ran the command locally (flake8 | wc -l) and obtained 2163 errors or warnings, which is not good. It would be nice to clean the source code a bit.\n* There is no code coverage associated to the tests, which is an important limitation. Please add code coverage to your project. There are \"only\" 16 tests for the moment, so I suspect that a substantial part of the source code is not tested.\n* I don't understand the point of using the \"-s\" option with pytest. I don't find the output (when using this option) to be clear or useful to identify tests that would fail for instance. In particular, I don't think that printing stuff when running the tests in the CI is relevant.\n\nIs the rationale for developing the new software tool clearly explained? Yes\n\nIs the description of the software tool technically sound? Yes\n\nAre sufficient details of the code, methods and analysis (if applicable) provided to allow replication of the software development and its use by others? Yes\n\nIs sufficient information provided to allow interpretation of the expected output datasets and any results generated using the tool? Yes\n\nAre the conclusions about the tool and its performance adequately supported by the findings presented in the article? Yes",
"responses": [
{
"c_id": "11385",
"date": "07 May 2024",
"name": "Benjamin Haibe-Kains",
"role": "Author Response",
"response": "We thank Dr Faouzi for his feedback regarding our manuscript titled ‘Med-ImageTools: An open-source Python package for robust data processing pipelines and curating medical imaging data’. We have added our responses to their comments here in subdivided sections of comments/questions (Q), in bold, our corresponding responses/answers (A), and any relevant sections of the manuscript as an italicized bullet point, with changes emphasized in bold. Q: The authors present their Python package for data processing pipelines and curating medical imaging data in this manuscript. Overall the manuscript is well written and clearly presents the main contributions of this package to the community. Nonetheless, some points on the software side could be improved. Please find my comments below: A: We thank the reviewer for their positive understanding of the authors’ work and constructive feedback, especially on the technical aspects of the package. Q: How does the output format used compare to other formats such as other popular formats such as BIDS (Brain Imaging Data Structure, which is maybe less general since it's intended for brain imaging data)? A: We thank the reviewer for bringing the BIDS standard to the authors’ attention. The BIDS standard is very effective in standardizing brain imaging datasets by adhering to an intuitive structure describing various modalities of brain MRI. Med-ImageTools is mainly geared towards use cases that are broader and less specific than BIDS and its implementation would require fundamental reworkings of the package. However, the benefit of having a BIDS-native output cannot be understated and this feature will be added to the development team’s ongoing development goals. We have revised the relevant section from ‘Current Workflows’ as follows: While Med-ImageTools has many modular functions for image, contour, and dose input/output (IO) built on popular frameworks such as SimpleITK, TorchIO and PyDicom, such functionalities are redundant and available in other open-source packages as well. We have tailored our core features towards broader use cases and development workflows, instead of modality or disease type specific workflows, such as BIDS. Our main contribution is in the development of AutoPipeline and will mainly discuss its functionalities and implementation. Q: Figure 2b (and the corresponding text): Raw saved running time, even though useful, is irrelevant without knowing the total running time. Please provide also the saved time as a percentage of the total running time. A: We appreciate the reviewer’s point about Figure 2b. However, the reported time saved is reported in core-hours, which is a comprehensive unit of total running time. This is because the crawling occurred independently from the end-to-end processing of the datasets to ensure an accurate measure of core-hours saved. Reporting core-hours saved allows the reader to estimate their time/cost savings based on their specific hardware availabilities, either local machine, server-based, or cloud infrastructure, agnOSTIC to the hardware used by the authors. We have emphasIZED this point in our manuscRIPT as fOLLOWS: However, when scaled up to larger TCIA datasets such as OPC-Radiomics 13 (n=606) and NLST 14 (n=26254), it can save researchers 24.7 core-hours and 1072 core-hours, respectively (Figure 2b). The resources saved are reported in core-hours to allow a hardware-agnostic estimation of time and cost savings. While it may vary depending on the research infrastructure utilized, these databases can result in significant savings in billing. Q: Figures 2c and 2d: \"1 lines\" => \"1 line\" Q: The authors should consider renaming the \"master\" branch as the \"main\" branch (see https://sfconservancy.org/news/2020/jun/23/gitbranchname/). Q: Is there a reason for Python 3.8 not being tested in the continuous integration (CI) (maybe a limit on the number of jobs)? By the way, Python 3.7 is no longer supported and Python 3.11 has been released for over 9 months (https://devguide.python.org/versions/). The Python versions tested in the CI should probably be updated. A: We thank the reviewer for their comments. The figures and repository have been updated to address these points. Q: Flake8 is installed in the CI but never used. I ran the command locally (flake8 | wc -l) and obtained 2163 errors or warnings, which is not good. It would be nice to clean the source code a bit. A: We thank the reviewer for raising the concern about the source code’s sanitation. The authors have reviewed the flake8 output and incorporated it into our codebase. You can find all changes in the following commit: https://github.com/bhklab/med-imagetools/commit/a8c0c3be53b4c9b43da12603b222d261d417a18c. Q: I don't understand the point of using the \"-s\" option with pytest. I don't find the output (when using this option) to be clear or useful to identify tests that would fail for instance. In particular, I don't think that printing stuff when running the tests in the CI is relevant. A: We thank the reviewer’s attention to detail in Med-ImageTools’ CI/CD framework. The -s option is enabled for verbose outputs from Med-ImageTools to be carried through to the standard output of the tests in CI. In the past we have had OS-specific issues that were elucidated through the -s flag, but have removed it for future releases. Q: There is no code coverage associated to the tests, which is an important limitation. Please add code coverage to your project. There are \"only\" 16 tests for the moment, so I suspect that a substantial part of the source code is not tested. A: We thank the reviewer for highlighting the lack of code coverage association of these tests. We agree that this is a clear limitation of the Med-ImageTools’ development pipeline, and have significantly increased the number of unit tests and integrated them into our CI/CD process. Future development efforts will continue to add more tests to our pipeline, including a badge detailing code coverage on our README."
}
]
}
] | 1
|
https://f1000research.com/articles/12-118
|
https://f1000research.com/articles/13-1238/v1
|
16 Oct 24
|
{
"type": "Research Article",
"title": "Do more pregnancies increase the risk of periodontal disease?",
"authors": [
"M. Helmi",
"Eman AlJoghaiman",
"M. Helmi"
],
"abstract": "Background Hormonal changes in pregnancy and their induced effect on periodontal health are well documented. The present study is aimed at the potential repercussions of multiple pregnancies on periodontal health.\n\nMaterials and methods Our study utilized data from key sections of the NHANES. All the pertaining and relevant data for the study is collected. Our exposure variable was the number of pregnancies, and the outcome variable was periodontal disease. The number of pregnancies is classified as one, two, three, four, or more. Age, gender, race/ethnicity, education, poverty/income ratio, marital status, and other variables. Multiple logistic regression models were employed to assess the impact of multiple pregnancies on periodontal disease.\n\nResult The crude and multiple logistic regression analyses revealed that none of the variables were significantly associated with the prevalence of periodontitis. In univariate analysis, patients with one or two pregnancies had higher odds of experiencing periodontitis (OR 1.154, 95% CI 0.748-1.779), (OR 1.464, 95% CI 0.864-2.483) respectively. However, these associations did not reach statistical significance.\n\nConclusion Within the limitation of the study, there is no significant relationship between parity and the prevalence of periodontitis, the longitudinal study may be warranted to delve deeper into any potential associations.",
"keywords": [
"Parity",
"Multiple Pregnancy",
"Periodontal Disease",
"NHANES"
],
"content": "Introduction\n\nPeriodontal diseases are influenced by a variety of factors.1 This prevalent oral condition is initiated by the accumulation of dental biofilm and is further exacerbated by various local and systemic elements.2 Notably, hormonal factors play a significant role in impacting periodontal health.3 Fluctuations in progesterone and estrogen levels during different life stages, including puberty, pregnancy, and menopause, have been identified as contributors to adverse effects on periodontal health.4 The hormonal influence leads to gingival changes that worsen pre-existing dental biofilm-induced gingivitis. Moreover, the absence of estrogen, without the presence of dental biofilm, can result in desquamative changes in the gingiva, representing the other end of the hormonal spectrum.5\n\nDuring the transitional phases between puberty and menopause, the nine-month duration of pregnancy introduces alterations to periodontal health, manifesting as both localized and generalized changes in the gingiva. Localized changes are characterized by the presence of a pregnancy tumor, while generalized changes manifest as an overall enlargement of the gingiva.6 This period is associated with heightened inflammation of the gingival tissues, commonly known as pregnancy gingivitis. Symptoms include redness, tenderness, and swelling of the gingiva, accompanied by spontaneous bleeding or bleeding during routine activities such as tooth brushing or eating.7 Typically commencing in the second month of pregnancy, these changes can peak in severity during the third trimester.8\n\nThe pathogenesis of altered periodontal health during pregnancy involves a combination of factors.4 A key aspect is the rise in levels of hormones such as oestradiol, oestriol, and notably, progesterone. These hormones play a central role in modifying the host immune-inflammatory response to oral bacteria. Specifically, Prevotella intermedia, a Bacteroides species thriving on estrogen and progesterone, experiences a significant increase during pregnancy, serving as a primary bacterial factor.9\n\nFurthermore, these hormones exhibit specific receptors on gingival fibroblasts and epithelial cells, influencing gingival changes. Additionally, they act on endothelial cells, increasing vascular permeability and contributing to the overall alterations observed in the gingiva during pregnancy.10 This intricate interplay underscores the multifaceted nature of periodontal health dynamics in the context of pregnancy.\n\nProgesterone emerges as the key hormone driving these changes, yet estrogen also plays a significant role in inducing vascular changes.11 Simultaneously, deficiencies in host inflammatory cells, particularly in neutrophil chemotaxis, contribute to the adverse aspects of periodontal health. Significantly, vitamin D deficiency is a prevalent concern in the pregnant population, as highlighted in various publications.12 The role of vitamin D in periodontitis is well-documented.\n\nThe susceptibility to oxidative stress increases during pregnancy due to heightened metabolic demands and increased tissue oxygen requirements, a factor strongly implicated in the inflammatory process.13 Additionally, the presence of gestational diabetes may further disrupt host defense mechanisms, altering the delicate balance between health and disease.14 The intricate interplay of these hormonal, nutritional, and metabolic factors underscores the complexity of periodontal health dynamics during pregnancy.\n\nWhile numerous studies have delved into the connection between pregnancy and periodontal disease, there is a scarcity of research exploring the impact of multiple pregnancies on the increased risk of periodontal issues.15,16 Given the tissue changes occurring with each pregnancy and the cumulative effect of repeated exposure in subsequent pregnancies, there is a hypothetical expectation of a deteriorating impact on periodontal health.4 Therefore, understanding the potential repercussions of multiple pregnancies on periodontal health holds significant importance for both expectant mothers and healthcare professionals.\n\nBy acknowledging this potential correlation, healthcare providers can underscore the importance of maintaining good oral hygiene practices and seeking appropriate dental care during pregnancy to mitigate any possible risks. In this context, an effort has been made to investigate the relationship between multiple pregnancies and periodontal health. This exploration aims to enhance our comprehension of potential risks and preventive measures associated with periodontal disease in individuals experiencing multiple pregnancies. Such knowledge can serve as a guide for healthcare professionals, enabling them to offer pertinent advice, facilitate early detection, and provide timely interventions to support optimal periodontal health outcomes in pregnant women.\n\n\nMethods\n\nThe National Health and Nutrition Examination Survey (NHANES) is a comprehensive, cross-sectional survey conducted in the United States, aiming to provide a nationally representative overview of non-institutionalized individuals living in households. Participants in this survey undergo a series of assessments, including the completion of a questionnaire, medical and dental examinations, and various laboratory tests. The protocols for collecting oral health data in the NHANES 2011–2012 and NHANES 2013–2014 cycles were approved by the Centres for Disease Control and Prevention National Centre for Health Statistics Research Ethics Review Board. Written informed consent was secured from all survey participants.\n\nOur study utilized data from key sections of the NHANES, including demographic information, examination results, questionnaire responses, and limited access data. The focus was on individuals aged 18 years and older who underwent a dental examination, with exclusion criteria in place to remove edentulous subjects from our analysis, we ensure a thorough and targeted evaluation of oral health within a diverse and nationally representative sample of the U.S. population.\n\nExposure variable\n\nNumber of pregnancies\n\nOutcome variable\n\nPeriodontal disease\n\nFor this investigation, we used the NHANES complete periodontal examination data to calculate periodontal disease indices using Eke et al. definition of periodontal disease17). According to this definition, periodontal disease was classified as follows: Severe periodontitis: ≥2 interproximal sites with loss of attachment (LOA) ≥6 mm (not on the same tooth) and ≥1 interproximal site with probing depth (PD) ≥5 mm; Moderate periodontitis: ≥2 interproximal sites with LOA ≥4 mm (not on same tooth), or ≥2 interproximal sites with PD ≥5 mm (not on same tooth); Mild periodontitis: ≥2 interproximal sites with LOA ≥3 mm, and ≥2 interproximal sites with PD ≥4 mm (not on same tooth) or one site with PD ≥5 mm and finally, no periodontitis group whose has no evidence of mild, moderate, or severe periodontitis.18\n\nCovariate variable\n\nTo ensure a comprehensive examination and control for any factors that might influence the outcome, our analysis includes a range of covariates. These covariates serve the crucial purpose of minimizing the impact of potential confounders, allowing us to scrutinize the relationship between the exposure and outcome with greater precision. The diverse set of covariates comprises age, gender, race/ethnicity, education, poverty/income ratio, marital status, occupation, smoking habits, alcohol consumption, dental insurance coverage, dental visit frequency, and body mass index (BMI). Age is categorized into six groups: (18-30), (31-40), (41-50), (51-60), and over 60 years. Gender is identified as either female or male. Race and ethnicity are classified as non-Hispanic White, non-Hispanic Black, Mexican American and other Hispanic, and non-Hispanic Asian. Poverty indices are categorized into low, middle, and high. Marital status is delineated as yes or no. Occupation is categorized as working and non-working. Dental visits are categorized as regular and not regular. Dental insurance coverage is classified as yes or no. Smoking status is divided into never, former, and current smokers. Alcohol consumption is classified as alcohol drinker and non-alcohol drinker. The number of pregnancies is classified as one, two, three, four, or more. Lastly, education level is categorized as less than high school, high school, and college graduate or above. By meticulously examining and accounting for these covariates, we aim to obtain results that are closer to the true relationship between the exposure and outcome, free from the confounding effects of other variables.\n\nThe data were obtained by consolidating demographic, health questionnaire, clinical examination, and limited access data from NHANES (2011–2012) with corresponding files from NHANES (2013–2014). To ensure unbiased point estimates and accurate variance estimation, considering the complex sampling design of NHANES, we applied proper sampling weights and utilized a licensed version of SAS survey procedures, following the recommendations of the National Centre for Health Statistics and the Centres for Disease Control and Prevention.\n\nAn analysis of the demographics and disease status of the study population was conducted using the Rao-Scott chi-squared test. Additionally, both simple and multiple logistic regression models were employed to assess the impact of multiple pregnancies on periodontal disease. The multiple regression model included age, sex, race, income, and education level as explanatory variables. The selection of these potential confounders was based on either current literature evidence or their association with insurance and dental care utilization variables observed in bivariate analysis. The significance level was set at p ≤ 0.05, ensuring a rigorous evaluation of the relationships within the study.\n\n\nResults\n\nTable 1 presents the demographic characteristics of the study subjects, including weighted percentages. Among the 2128 subjects, more than one-quarter were aged over 60 years [insert specific age], 42.9% identified as non-Hispanic white, 34.6% had a high household level, and over half had either an associate or college degree. The majority of subjects had some form of health insurance and had visited the dentist within the 12 months preceding the survey.\n\na Weighted row percentages.\n\nThe subjects were categorized based on the severity of periodontitis, dividing them into groups of no, mild, moderate, and severe periodontitis (refer to Tables 2 and 3). The prevalence of periodontitis exhibited a significant difference primarily based on the subjects’ age. Subjects in older age brackets were consistently more likely to have some form of periodontitis compared to their younger counterparts. Additionally, the prevalence of periodontitis on average was higher among pregnant subjects compared to those who were not pregnant.\n\na Weighted row percentages.\n\n* By Rao-Scott chi-square test.\n\na Weighted row percentages.\n\nIn Table 4, approximately 9.4% of subjects with health insurance and 9.2% of subjects without health insurance exhibited some form of periodontitis. Analyzing periodontitis concerning the time elapsed since the last dental visit, we observed that 11% of subjects who had a dental visit within 12 months prior to the survey had periodontitis. For those with a dental visit more than 12 months but less than 24 months prior, 8.5% had periodontitis, and 9.8% of those with their most recent dental visit more than 2 years before the survey had periodontitis.\n\na Weighted row percentages.\n\nBoth crude and multiple logistic regression analyses revealed that none of the variables were significantly associated with the prevalence of periodontitis. In univariate analysis, patients with one and two pregnancies had higher odds of experiencing periodontitis (OR 1.464, 95% CI 0.864-2.483), (OR 1.154, 95% CI 0.748-1.779) respectively. However, these associations did not reach statistical significance (p > 0.05). Patients with dental visits in the 1-2 year range had greater odds (OR 1.129, 95% CI 0.772-1.651) of having periodontitis, but this association was not statistically significant (p > 0.05) (see Table 5).\n\n\nDiscussion\n\nThe primary aim of this study is to investigate the association between multiple pregnancies and the severity of periodontal disease/periodontitis. According to the study results, there was no discernible difference in the prevalence of periodontitis between individuals with single pregnancies and those with multiple pregnancies. However, it was noted that the prevalence of periodontitis was higher in pregnant individuals compared to non-pregnant ones. The study findings also indicated that patients with two pregnancies had higher odds of experiencing periodontitis than those with only one pregnancy, although this difference did not reach statistical significance. Both unadjusted and adjusted odds ratios for the number of pregnancies suggested higher odds of periodontitis during pregnancy.\n\nThe findings of the current study align with earlier reported research.19,20 Previous studies indicated higher gingival index and periodontal probing depth among women with prior pregnancies compared to primigravida. However, after adjusting for factors such as age, socio-economic status, education, and other associated risk factors, no correlation was identified. Nevertheless, some studies have demonstrated a significant association between increased gingival scores and periodontal probing depth in women with multiple pregnancies compared to those with a single pregnancy.21 Additionally, research has shown heightened gingival inflammation and increased periodontal probing depth during pregnancy.22–24\n\nPiscoya et al. conducted a study exploring various factors, including the number of pregnancies and the prevalence of periodontitis. Their findings, organized in a hierarchy, revealed that schooling, family income, smoking, body mass index, and bacterial plaque were associated with the prevalence of periodontitis, but not with multiple pregnancies.20 Another study concluded that, in addition to other factors, pregnant women with two or more previous births [multigravidae] exhibited significantly higher gingival index and periodontal probing depth scores compared to those with one previous birth. However, the increased gingival index and periodontal probing depth in multigravidae might be attributed to untreated gingival or periodontal disease from the first pregnancy persisting during subsequent pregnancies. Furthermore, factors such as low socio-economic status and lower educational levels could contribute to negligence of oral hygiene, leading to an increased prevalence of periodontitis.21\n\nThe variations in study findings may be attributed to several factors. Multiparous women typically tend to be older than prima gravida women, leading to a longer cumulative exposure to etiological agents of disease. Additionally, the absence of treatment during or between pregnancies results in untreated periodontal disease persisting into subsequent pregnancies.21 Multiparous women, especially those with young children, may prioritize other systemic health conditions, diverting attention, time, energy, and finances away from personal dental care. This tendency can result in neglected oral care, increased plaque accumulation, and a higher prevalence of gingival conditions.19\n\nTherefore, the observed higher gingival index and probing depth in multiparous women may be more related to sociodemographic backgrounds than to a true association with parity. It is widely recognized that existing periodontal disease is exacerbated during pregnancy, and pregnancy itself does not directly cause gingival or periodontal disease. Pregnancy-associated physiological changes can superimpose gingival inflammation on pre-existing dental plaque accumulation. If oral health is well-maintained with the absence of gingival inflammation before pregnancy, the condition of pregnancy itself may not induce gingivitis or periodontitis. Notably, existing studies have not taken into account pre-pregnancy gingival inflammation and treatment for the periodontal condition, which could influence the study outcomes.20\n\nHormonal alterations during pregnancy contribute to an increase in specific periodontal pathogens, such as Prevotella intermedia, which utilizes elevated hormone levels as a nutrient. Physiological microvascular changes observed in pregnancy, coupled with exposure to altered dental biofilm, may exacerbate pre-existing gingival conditions. The surge in estrogen levels, particularly progesterone, reaches a 20-fold increase, leading to changes in vascular permeability that cause gingival swelling and elevated crevicular fluid levels.25 The heightened production of prostaglandins, in addition to the vascular burden, may intensify gingival inflammation, result in the loss of keratinization of gingival epithelium, and foster fibroblast proliferation. Furthermore, the altered host response is characterized by decreased chemotaxis and phagocytic capacity of neutrophils, along with the down-regulation of IL-6 production. This exposes the gingival tissue to microbial attack, resulting in increased gingival inflammation.26\n\nGestational diabetes compounds the detrimental effects on host immunity, fostering the proliferation of pathogenic microflora and increasing the risk of periodontal disease.25 Repeated exposure to these events during multiple pregnancies is anticipated to lead to heightened periodontal destruction. However, it has been observed that postpartum, there is a substantial decrease in gingival inflammation, and gingival health is often restored to the pre-pregnancy state.27 Furthermore, the experiences of gingival disease during previous pregnancies may prompt individuals to undergo periodontal treatment, contributing to the restoration of gingival health.28 This could explain the absence of a significant difference in periodontitis among individuals with one pregnancy and those with multiple pregnancies in the present study.\n\nThere are several limitations to our study. Notably, the exacerbation of existing periodontal conditions during pregnancy is a well-known phenomenon. Unfortunately, our study lacked data on the status of gingival inflammation before pregnancy, which could have influenced the study outcomes. Additionally, being a retrospective cross-sectional study, our investigation relied on data from a study not specifically designed to address our hypothesis, potentially introducing clinical variations in the disease process. The utilization of dental care after the first pregnancy was not explored, and if a significant number of individuals underwent periodontal treatment after the initial pregnancy and before subsequent pregnancies, it might impact the outcomes. While a standardized protocol was followed for the diagnosis of periodontal disease, there remains a possibility of misclassification, albeit likely to be non-differential. Periodontal status was assessed only at the baseline survey, and changes over follow-up were not considered. It is conceivable that individuals initially free of periodontal disease might develop the condition later, potentially leading to an underestimation of the association for those groups. To address these limitations, future studies should aim for larger longitudinal prospective designs to validate the findings from this initial study.\n\nWhile the study did not identify a difference in the prevalence of periodontitis between single and multiple pregnancies, the findings hold significance on two fronts. Firstly, the data were obtained from a substantial sample size, highlighting increased gingival and periodontal changes during pregnancy. This underscores the need to educate all women about these findings, aiming to prevent periodontal changes during pregnancy that may impact their regular daily routines. Secondly, as periodontal disease is deemed a risk factor for pregnancy outcomes, its control assumes prime importance. Addressing and managing periodontal health becomes crucial in optimizing pregnancy outcomes.\n\n\nConclusion\n\nIn summary, this study has explored the relationship between parity and the prevalence of periodontitis, revealing no significant association between the prevalence of periodontitis and the number of pregnancies. However, a longitudinal study may be warranted to delve deeper into any potential associations.\n\n\nAuthors’ contributions\n\nBoth Authors contributed equally from the idea to the preparing the draft and both authors reviewed and prepared the final draft of the study.\n\n\nEthics and consent\n\nThe protocols for collecting oral health data in the NHANES 2011–2012 and NHANES 2013–2014 cycles received approval from the Centres for Disease Control and Prevention National Centre for Health Statistics Research Ethics Review Board. All survey participants provided written informed consent before publishing their information.\n\nhttps://www.cdc.gov/nchs/nhanes/irba98.htm",
"appendix": "Data availability\n\nFigshare: Do more pregnancies increase the risk of periodontal disease?, https://doi.org/10.6084/m9.figshare.25662546 29\n\nThe project contains the following underlying data:\n\n• Demographical Characteristics (e.g., gender, income, BMI …), Periodontal Disease status, Pregnancy status, and dental visits.\n\nData are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).\n\n\nReferences\n\nAlJehani YA: Risk factors of periodontal disease: review of the literature. Int. J. Dent. 2014; 2014: 1–9. Publisher Full Text\n\nKinane DF, Stathopoulou PG, Papapanou PN: Periodontal diseases. Nat. Rev. Dis. Primers. 2017; 3: 17038. Publisher Full Text\n\nSteinberg BJ, Hilton IV, Iida H, et al.: Oral health and dental care during pregnancy. Dent. Clin. N. Am. 2013; 57(2): 195–210. Publisher Full Text\n\nBoyapati R, Cherukuri SA, Bodduru R, et al.: Influence of Female Sex Hormones in Different Stages of Women on Periodontium. J. Midlife Health. 2021; 12(4): 263–266. PubMed Abstract | Publisher Full Text\n\nThomas KE, Chitra N: Periodontal changes pertaining to women from puberty to postmenopausal stage. Int. J. Pharm. Bio. Sci. 2013; 4(2): B766–B771.\n\nSoory M: Hormonal factors in periodontal disease. Dent. Update. 2000; 27(8): 380–383. Publisher Full Text\n\nWu M, Chen SW, Jiang SY: Relationship between gingival inflammation and pregnancy. Mediat. Inflamm. 2015; 2015: 623427. PubMed Abstract | Publisher Full Text | Free Full Text\n\nCoventry J, Griffiths G, Scully C, et al.: ABC of oral health: periodontal disease. BMJ. 2000; 321(7252): 36–39. PubMed Abstract | Publisher Full Text | Free Full Text\n\nGare J, Kanoute A, Meda N, et al.: Periodontal Conditions and Pathogens Associated with Pre-Eclampsia: A Scoping Review. Int. J. Environ. Res. Public Health. 2021; 18(13). PubMed Abstract | Publisher Full Text | Free Full Text\n\nMariotti A, Mawhinney M: Endocrinology of sex steroid hormones and cell dynamics in the periodontium. Periodontol. 2000. 2013; 61(1): 69–88. PubMed Abstract | Publisher Full Text\n\nMorelli EL, Broadbent JM, Leichter JW, et al.: Pregnancy, parity and periodontal disease. Aust. Dent. J. 2018; 63: 270–278. PubMed Abstract | Publisher Full Text\n\nSllamniku Dalipi Z, Dragidella F: Calcium and Vitamin D Supplementation as Non-Surgical Treatment for Periodontal Disease with a Focus on Female Patients: Literature Review. Dent. J (Basel). 2022; 10(7). Publisher Full Text\n\nHansson SR, Nääv Å, Erlandsson L: Oxidative stress in preeclampsia and the role of free fetal hemoglobin. Front. Physiol. 2014; 5: 516.\n\nAbariga SA, Whitcomb BW: Periodontitis and gestational diabetes mellitus: a systematic review and meta-analysis of observational studies. BMC Pregnancy Childbirth. 2016; 16(1): 344. PubMed Abstract | Publisher Full Text | Free Full Text\n\nAhnoux A, Aoussi EL, Anongba DS, et al.: Pregnancy and periodontal health. Study of 133 pregnant women. Odontostomatol. Trop. 2003; 26(102): 37–40. PubMed Abstract\n\nMoore S, Ide M, Wilson RF, et al.: Periodontal health of London women during early pregnancy. Br. Dent. J. 2001; 191(10): 570–573. PubMed Abstract | Publisher Full Text\n\nEke PI, Page RC, Wei L, et al.: Update of the case definitions for population-based surveillance of periodontitis. J. Periodontol. 2012; 83(12): 1449–1454. PubMed Abstract | Publisher Full Text | Free Full Text\n\nAljoghaiman E, Helmi H: Do more pregnancies increase the risk of periodontal disease?2023.\n\nMachuca G, Khoshfeiz O, Lacalle JR, et al.: The influence of general health and socio-cultural variables on the periodontal condition of pregnant women. J. Periodontol. 1999; 70(7): 779–785. PubMed Abstract | Publisher Full Text\n\nde Vasconcellos Piscoya MDB , de Alencar Ximenes RA , da Silva GM , et al.: Periodontitis-associated risk factors in pregnant women. Clinics. 2012; 67(1): 27–33. PubMed Abstract | Publisher Full Text | Free Full Text\n\nTaani D, Habashneh R, Hammad M, et al.: The periodontal status of pregnant women and its relationship with socio-demographic and clinical variables. J. Oral Rehabil. 2003; 30(4): 440–445. PubMed Abstract | Publisher Full Text\n\nMasoni S, Panattoni E, Rolla P, et al.: Stomatological problems related to pregnancy. A statistical study. Minerva Stomatol. 1991; 40(12): 791–796. PubMed Abstract\n\nRaber-Durlacher J, Van Steenbergen T, Van der Velden U, et al.: Experimental gingivitis during pregnancy and post-partum: clinical, endocrinological, and microbiological aspects. J. Clin. Periodontol. 1994; 21(8): 549–558. PubMed Abstract | Publisher Full Text\n\nAbraham-Inpijn L, Polsacheva O, Raber-Durlacher J: The significance of endocrine factors and microorganisms in the development of gingivitis in pregnant women. Stomatologiia. 1996; 75(3): 15–18. PubMed Abstract\n\nBendek MJ, Canedo-Marroquín G, Realini O, et al.: Periodontitis and gestational diabetes mellitus: a potential inflammatory vicious cycle. Int. J. Mol. Sci. 2021; 22(21): 11831. PubMed Abstract | Publisher Full Text | Free Full Text\n\nRaju K, Berens L: Periodontology and pregnancy: An overview of biomedical and epidemiological evidence. Periodontol. 2021; 87(1): 132–142. PubMed Abstract | Publisher Full Text\n\nGonzalez-Jaranay M, Téllez L, Roa-López A, et al.: Periodontal status during pregnancy and postpartum. PLoS One. 2017; 12(5): e0178234. PubMed Abstract | Publisher Full Text | Free Full Text\n\nVilla A, Abati S, Pileri P, et al.: Oral health and oral diseases in pregnancy: a multicentre survey of Italian postpartum women. Aust. Dent. J. 2013; 58(2): 224–229. PubMed Abstract | Publisher Full Text\n\nAljoghaiman E: The Relationship Between Mental Health and Periodontal Disease: Insights from NHANES Data. [Dataset]. figshare. 2024. Publisher Full Text"
}
|
[
{
"id": "332865",
"date": "04 Nov 2024",
"name": "Dr. Talal Shihayb",
"expertise": [
"Reviewer Expertise Periodontitis-systemic diseases",
"oral epidemiology",
"advanced methods",
"meta-research in oral health"
],
"suggestion": "Approved With Reservations",
"report": "Approved With Reservations\n\ninfo_outline\nAlongside their report, reviewers assign a status to the article:\n\nApproved The paper is scientifically sound in its current form and only minor, if any, improvements are suggested\n\nApproved with reservations\nA number of small changes, sometimes more significant revisions are required to address specific details and improve the papers academic merit.\n\nNot approved Fundamental flaws in the paper seriously undermine the findings and conclusions\n\nThe authors assessed the effects of number of pregnancies on periodontitis. Studying pregnancy outcomes in relation to oral health is crucial. However, I have some comments on the paper:\nIntroduction No comments.\nMethods Can the authors explain why they only selected the 2011-2012 and 2013-2014 cycles of NHANES?\nThe authors excluded edentulous participants from their study. Dealing with edentulous patients is tricky and not straight forward. If the reason for being edentulous was due to periodontal disease, the perhaps edentulous participants may be considered as the most severe form of periodontitis (since the authors are studying prevalent periodontitis and not incident periodontitis). Given that the authors are studying the accumulated exposure of pregnancy on periodontal disease, the results could be different had edentulous participants due to periodontal disease been included in the analysis, I suggest the authors add edentulous as a separate category or at least justify/discuss excluding edentulous participants from the study in the methods/discussion section.\nIf by edentulous participants, the authors meant without any teeth, then any person with 1 tooth is automatically classified as mild periodontitis or no periodontitis and cannot be classified as moderate or severe. However, all included participants should have the chance to be in group of periodontitis, which is unfortunately prevented including participants with just 1 tooth. The authors perhaps may have already included those with at least 2 permanent teeth and if so, should make this clear. If not, then inclusion of participants with at least 2 teeth (if authors are still excluding edentulous participants) in the study or at least discussion of the impact of this aspect on the result is important.\nThe authors defined the exposure as the number of pregnancies. Did the authors include all pregnancies whether completed or not? Providing more details on this aspect would make the paper more clear. Furthermore, the authors used it as a categorical variable instead of a continuous exposure. Unfortunately, this results in loss of information. I suggest the authors add this to the discussion section.\nThe authors have defined the outcome of chronic periodontitis according to CDC/AAP. I suggest replacing the word periodontal disease with chronic periodontitis in the manuscript as the term periodontal disease is an umbrella term that includes multiple conditions.\nI suggest adding more details on how smoking and alcohol were classified or at least cite their webpages if the authors classified them as NHANES originally did.\nThe authors wrote the following under statistical analysis: “To ensure unbiased point estimates and accurate variance estimation, considering the complex sampling design of NHANES, we applied proper sampling weights and utilized a licensed version of SAS survey procedures, following the recommendations of the National Centre for Health Statistics and the Centres for Disease Control and Prevention.” I think using clustering variables to correctly estimate standard errors should be added as both sampling weights and clustering are need to correctly estimate point estimates and standard errors, respectively.\nThe authors have pointed out the covariates that they included in their study and mentioned the following under statistical analysis: “The multiple regression model included age, sex, race, income, and education level as explanatory variables. The selection of these potential confounders was based on either current literature evidence or their association with insurance and dental care utilization variables observed in bivariate analysis. The significance level was set at p ≤ 0.05, ensuring a rigorous evaluation of the relationships within the study.” Determining how the confounding variables were selected is very crucial in order to estimate the causal effect of multiple pregnancies on chronic periodontitis. In this study, the authors rightly determined age, sex, race, income, and education level as confounding variables based on previous knowledge and literature. However, the following on insurance and dental visit was not clear: “their association with insurance and dental care utilization variables observed in bivariate analysis.” Furthermore, assessing associations or confounding variables based on p-values should be avoided (please check [1],[2])\nAs a point related to the one above and based on the criteria the authors went with for determining confounding variables, unfortunately, smoking and diabetes (well-known confounding variables) were left out of the multivariable logistic regression model. Therefore, residual confounding exists in the result. I suggest adding them to the model or at least discuss how the residual confounding of these would impact the odds ratio of number of pregnancies on periodontitis.\nResults The authors did not elaborate on any missing data. Authors should show the frequency and % of missing data for each variables. In addition, authors should describe how they dealt with any missing data in their analysis and discuss its implications on the results.\nThe authors mentioned that they have included the following covariates: “age, gender, race/ethnicity, education, poverty/income ratio, marital status, occupation, smoking habits, alcohol consumption, dental insurance coverage, dental visit frequency, and body mass index (BMI)” Although marital, status and occupation were mentioned in table 4, they along with smoking habits were not mentioned in table 1. Smoking habits was not even mentioned in any table. Authors should add these.\nIn table 5, patients wrote: “Patients with dental visits in the 1-2 year range had greater odds (OR 1.129, 95% CI 0.772-1.651) of having periodontitis, but this association was not statistically significant (p > 0.05) (see Table 5).” The odds ratios of the confounding variables are of no interest to the authors in the study as they are do not correctly estimate their causal effect. This phenomena is well-known as Table 2 fallacy (3). Kindly just report the odds ratios of main variable of interest (number of pregnancies) and remove the other from text or table.\nPlease round the odds of pregnancy on periodontitis to 2 digits as 3 digits adds nothing and just complicates reading the results.\nI suggest removing displaying/discussing the results in text as significant or non-significant and instead focus on the point estimates as well as their precision (please check [1]and[2])\nDiscussion and conclusion The authors wrote in the limitations: “Additionally, being a retrospective cross-sectional study, our investigation relied on data from a study not specifically designed to address our hypothesis, potentially introducing clinical variations in the disease process.” The word retrospective should be removed as this was a cross-sectional study.\nAuthors should further discuss the impact of no-temporality, pregnancy misclassification likelihood and effect, missing data, and leaving out important confounding variables.\n\nIs the work clearly and accurately presented and does it cite the current literature? Yes\n\nIs the study design appropriate and is the work technically sound? Partly\n\nAre sufficient details of methods and analysis provided to allow replication by others? Partly\n\nIf applicable, is the statistical analysis and its interpretation appropriate?\nPartly\n\nAre all the source data underlying the results available to ensure full reproducibility? Yes\n\nAre the conclusions drawn adequately supported by the results? Partly",
"responses": [
{
"c_id": "13245",
"date": "03 Feb 2025",
"name": "Eman AlJoghaiman",
"role": "Author Response",
"response": "Dear Reviewer, Thank you for your insightful review and valuable suggestions to enhance our manuscript. We have carefully revised the manuscript and highlighted the changes accordingly. We sincerely appreciate your time and thoughtful feedback. Best regards,"
}
]
},
{
"id": "343703",
"date": "26 Dec 2024",
"name": "Nancy Ajwa",
"expertise": [
"Reviewer Expertise Public Health",
"Periodontology",
"Orthodontology",
"Dentistry"
],
"suggestion": "Approved",
"report": "Approved\n\ninfo_outline\nAlongside their report, reviewers assign a status to the article:\n\nApproved The paper is scientifically sound in its current form and only minor, if any, improvements are suggested\n\nApproved with reservations\nA number of small changes, sometimes more significant revisions are required to address specific details and improve the papers academic merit.\n\nNot approved Fundamental flaws in the paper seriously undermine the findings and conclusions\n\nThe current structure provides a good foundation for the research paper. I have a few minor comments on the paper, which I am splitting in the headings so that authors can easily identify and improve the respective sections. Introduction: -Consider moving some of the detailed descriptions of the hormonal mechanism to later paragraphs for better flow. -Add a brief sentence early in the introduction highlighting why studying multiple pregnancies (versus single pregnancy) is particularly important -Consider mentioning any geographical or population-specific gaps in current knowledge that this study helps address Methods: -Clarify the definition of pregnancy exposure (whether it includes completed pregnancies only or all pregnancies) -Add information about how missing data was handled in the analysis Results: -Round the odds ratios to 2 decimal places instead of 3 for better readability -Include smoking status data in Table 1, as it's an important variable that was mentioned in the methods but not shown in the results Discussion: -Add a brief discussion of the limitations of using pregnancy as a categorical rather than continuous variable -Include a short paragraph acknowledging the potential impact of excluding edentulous participants on the study findings\nThese modifications would improve the manuscript while maintaining its overall scientific merit and indexing. The core findings and conclusions remain sound, and these suggested changes are minor in nature.\n\nIs the work clearly and accurately presented and does it cite the current literature? Yes\n\nIs the study design appropriate and is the work technically sound? Yes\n\nAre sufficient details of methods and analysis provided to allow replication by others? Yes\n\nIf applicable, is the statistical analysis and its interpretation appropriate?\nYes\n\nAre all the source data underlying the results available to ensure full reproducibility? Yes\n\nAre the conclusions drawn adequately supported by the results? Yes",
"responses": [
{
"c_id": "13244",
"date": "06 Feb 2025",
"name": "Eman AlJoghaiman",
"role": "Author Response",
"response": "Dear Reviewer, Thank you for your insightful review and valuable suggestions to enhance our manuscript. We have carefully revised the manuscript and highlighted the changes accordingly. We sincerely appreciate your time and thoughtful feedback. Best regards,"
}
]
}
] | 1
|
https://f1000research.com/articles/13-1238
|
https://f1000research.com/articles/13-951/v1
|
22 Aug 24
|
{
"type": "Study Protocol",
"title": "The effect of surface treatments on the bond strength of polyetheretherketone posts: a systematic review protocol",
"authors": [
"Hanen Boukhris",
"Aymen Ben Hadj Khalifa",
"Hayet Hajjami",
"Souha Boudegga Ben Youssef",
"Aymen Ben Hadj Khalifa",
"Hayet Hajjami",
"Souha Boudegga Ben Youssef"
],
"abstract": "Abstract*\nBackground Polyetheretherketone (PEEK) is widely used in the biomedical field due to its outstanding biological and mechanical properties. Originally employed as a temporary abutment in implantology, recent research has expanded its indications for more definitive applications, such as frameworks and dental post and core. This shift requires a thorough assessment of PEEK’s adhesion and mechanical characteristics. However, PEEK’s inert properties and intricate chemistry create difficulties in surface treatment, resulting in reduced surface energy and inadequate adhesion. Inducing specific physical and chemical changes aims to overcome these challenges and enhance adhesion for PEEK. Despite its numerous clinical trials, standardized protocols remain lacking. This systematic review aims to assess the impact of surface treatments on the bonding performance of PEEK posts.\n\nMethods A detailed search of the literature will be conducted across several databases including PubMed, Scopus and clinical trial registries. Additional databases such as Cochrane Central, EMBASE, Web of Science and EBSCO will also be included. The search strategy will target controlled randomized studies and non-randomized clinical trials evaluating the impact of surface treatments on PEEK post adhesion strength. The Newcastle-Ottawa Scale (NOS) will be used to assess bias in non-randomized studies, while the Cochrane Risk of Bias (ROB II) tool will be employed for evaluating randomized controlled trials. Data extraction will focus on study design, treatment methods, outcomes and results. This systematic review protocol will adhere to the guidelines for systematic reviews outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA).\n\nDiscussion The discussion will explore the implications of findings on clinical practice, highlighting the importance of enhancing PEEK’s bioactivity and surface energy to improve bonding efficacy in dental procedures. Moreover, it will suggest areas for future research to advance dental materials science, aiming to optimize the utilization of PEEK in dental applications\n\nSystematic review registration PROSPERO: CRD42024529783 (Registered on 08/04/2024).",
"keywords": [
"Polyetheretherketone (PEEK)",
"Posts",
"Surface treatments",
"Bond strength",
"Systematic review"
],
"content": "Introduction\n\nEndodontically treated teeth (ETT) with over 50% loss of coronal structure are prone to shear forces during chewing and often necessitate post-and-core placement. Posts are employed to secure the core material, enhancing the stability and retention of the final restoration.1,2\n\nThe growing demand for aesthetic improvements in dental treatments has led to the widespread use of prefabricated fiber posts. These posts offer advantages like uniform stress distribution, biocompatibility, and ease of handling, making them preferable over metal posts for restoring endodontically treated teeth. However, they can cause mechanical stress at the restoration margin and fail to strengthen the tooth structure.\n\nDespite their lower elasticity modulus compared to metal posts, fiber posts still exhibit significantly greater stiffness than dentin. Fiber post and core buildup materials can fail due to several mechanisms, such as cracking of the resin matrix, fracture of the fibers, and detachment at the interface.\n\nA novel material with both low Young’s modulus and satisfactory aesthetics has emerged. Thermoplastic polymer Polyetheretherketone (PEEK) has excellent properties. It has a Young’s modulus (3-4 GPa) lower than that of dentin which helps reduce stress on both restorations and teeth. These qualities make PEEK suitable for various dental applications including post and core systems.3–6\n\nHowever, the achievement of adhesion between PEEK and resin materials is challenging due to its low surface energy and resistant surface modification. Various treatments including chemical and micromechanical methods are recommended to improve the bonding of composite resin to PEEK posts.7–14\n\nThe goal of this systematic review is to assess how effectively PEEK posts perform with various surface treatments. This is crucial for dental professionals as it provides valuable information for developing dependable bonding protocols for PEEK posts and cores.\n\n\nProtocol\n\nThis protocol outlines the process for conducting a systematic review in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.15,16 The methodology from the F1000 journal will be followed to ensure accuracy and consistency at every step. The review protocol is registered in PROSPERO (CRD42024531175) since April 8, 2024.\n\nPrimary objective: To evaluate the effectiveness of various surface treatments in the improvement of bonding strength for Polyetheretherketone (PEEK) posts.\n\nSecondary objectives:\n\n- To provide evidence-based recommendations for clinicians in the choice of surface treatments for PEEK posts during bonding procedures.\n\n- To identify gaps in the literature and to propose possible suggestions for further research.\n\nThe inclusion criteria are structured according to the PICOS model. This model is designed to specify the key components of the research question.\n\nThe research question for this systematic review is: What is the effect of various surface treatments on retention and bond strength of polyetheretherketone (PEEK) posts used in dental restoration compared to untreated PEEK posts?\n\n• Types of participants: This systematic review will focus on patients requiring dental restoration with polyetheretherketone (PEEK) posts. Eligible participants must be free from medical conditions affecting bone healing, avoid parafunctional habits such as bruxism and have no occlusal problems.\n\n• Intervention types: This systematic review will explore PEEK (Polyetheretherketone) use in dental posts. Patients will receive interventions involving surface treatments like etching with 98% sulfuric acid and sandblasting with 50 μm alumina oxide (Al2O3).\n\n• Types of outcomes: This systematic review will focus on evaluating PEEK posts in dental restorations with different surface treatments before bonding. Key outcomes include retention rate, fracture resistance compared to traditional posts, and marginal adaptation quality. It will also assess bond strength between PEEK posts and dental materials, post-operative sensitivity, restoration longevity and clinical success rate overall.\n\n• Measures of effect: This systematic review will measure effects using quantitative and qualitative assessments. Quantitative measures include statistical analysis of retention rates, fracture resistance, bond strength and restoration longevity.\n\nQualitative measures involve evaluating marginal adaptation quality, post-operative sensitivity and clinical success rates through observational data and patient-reported outcomes. Meta-analytical techniques may be used to synthesize findings across studies for comparison.\n\n• Study types: Included articles will mainly include randomized controlled trials and prospective or retrospective cohort studies. These studies specifically investigate the use of polyetheretherketone (PEEK) posts in dental restorations, focusing on surface treatments. They are chosen for their substantial data and rigorous methodology to meet the research objectives effectively.\n\nExcluded articles will cover case reports, case series, abstracts, discussions, interviews, editorials, and opinion pieces, along with research that does not center on PEEK posts or surface treatments. Additionally, studies lacking adequate data or methodology will be omitted to ensure the review’s reliability and relevance.\n\nA combination of keywords and precise subject headings relevant to the topic will be employed in the refined search strategy, ensuring a comprehensive exploration of pertinent literature. In addition to MEDLINE, databases such as Web of Science, EBSCO, Scopus, Cochrane Central and EMBASE will be meticulously searched to reduce the likelihood of missing relevant studies.\n\nTo achieve comprehensiveness, attempts will be made to locate grey literature and active clinical trials through sources such as dissertations, conference proceedings and clinical trial registries. The expert panel will offer guidance in identifying grey literature sources and evaluating their pertinence to the review.17\n\nFurthermore, reference lists of included studies will be systematically examined as part of the search strategy to identify supplementary articles not retrieved solely through electronic databases.\n\nThis approach aims to reduce publication bias and ensure a thorough review of the available evidence.\n\nFor this systematic review, the study selection process will involve a comprehensive search across all identified databases. Two independent reviewers will screen titles and abstracts to exclude irrelevant studies. Full-text articles of potentially eligible studies will be assessed using predefined inclusion criteria. These criteria will focus on studies that evaluate surface treatments of PEEK in dental post applications.\n\nReviewers will check article reference lists for additional relevant studies. Articles meeting inclusion criteria will proceed to data extraction.\n\nDuring screening, any discrepancies among reviewers will be resolved through discussion. If needed, an additional reviewer (HH) will be consulted to ensure accuracy and consensus.\n\nTo guarantee the reliability of the findings, the methodological quality and risk of bias of the included studies will be evaluated using standardized tools.18\n\nFor randomized controlled trials, the Cochrane Risk of Bias (ROB II) tool will be utilized to assess methodological rigor across key domains: randomization procedures, adherence to intended interventions, completeness of outcome data, outcome measurement and accuracy in reporting results. Each domain will be categorized as having low, high or unclear risk of bias.19\n\nThe quality assessment of non-randomized studies will be conducted using the Newcastle-Ottawa Scale (NOS) which evaluates them based on three main criteria: the selection of study participants, group comparability and outcome ascertainment. Each study will be scored on these criteria with higher scores indicating better methodological quality.20\n\nTwo independent reviewers will conduct the assessments to minimize bias and enhance reliability. Any discrepancies in the assessments will be resolved through discussion or by consulting an additional reviewer (HH) to reach a consensus.\n\nMethodological rigor and assessment of bias will ensure that the review’s conclusions are based on high-quality evidence.\n\nData items: The following details will be extracted from the selected studies: participant demographics, specifics of the interventions, outcome measures, study characteristics (publication year, author, study design …) and results pertaining to bond strength and surface characteristics.\n\nExtraction method: A standardized data extraction form will be created in a Microsoft Excel sheet to systematically capture relevant data from each included study.\n\nData extraction will be conducted by two reviewers working autonomously to ensure consistent and precise handling of the information.\n\nDiscussion will be initiated to resolve any discrepancies, and if needed, input will be sought from a third reviewer (HH).\n\nThis structured approach will guarantee thorough and dependable data extraction for subsequent analysis.\n\nThe extracted data will be rigorously analyzed and synthesized to assess the effectiveness of surface treatments for PEEK in dental post materials. Initial descriptive analysis will summarize study details, participant characteristics, intervention particulars and outcome measures such as bond strength and surface characteristics. Quantitative synthesis, including meta-analysis where feasible, will calculate effect sizes with 95% confidence intervals and assess heterogeneity across studies using statistical tests such as Tau-squared, Cochran’s Q test and I-squared, systematically categorized to understand the range of variability.\n\nSubgroup analyses will explore variations in treatment methods and material types. Sensitivity analyses will test result robustness, and qualitative synthesis will offer a narrative summary where quantitative synthesis is not possible. Findings will be interpreted in the context of clinical relevance, discussing methodological strengths and limitations while proposing directions for future research in optimizing PEEK’s performance in dental applications. Forest plot will be used to depict the results, providing a concise visualization of aggregated study effects.21\n\n\nDiscussion\n\nThe outcomes of this systematic review will be highly relevant for practitioners focused on aesthetic and digital dentistry, especially in the management of damaged teeth. By evaluating the efficacy of surface treatments for polyetheretherketone (PEEK) in dental post materials, this review aims to provide evidence-based guidance on enhancing bond strength and surface characteristics critical for durable dental restorations. The findings are anticipated to inform clinical decision-making, facilitating the selection of optimal surface treatment strategies to improve the longevity and aesthetic outcomes of PEEK-based restorations. Moreover, this review will identify areas where current research is lacking and propose avenues for future investigation, aiming to advance the field of dental materials science and enhance patient care in aesthetic and functional dental rehabilitation.\n\nNo ethical approval is needed for this systematic survey. The authors intend to present the findings at target conferences and publish the research findings in a peer-reviewed journal adopting open science practices.\n\nThis systematic review is currently in the data analysis process. The protocol of this systematic review was submitted to PROSPERO registry on 8th April, 2024 (CRD42024529783).",
"appendix": "Data availability\n\nNo data are associated with this article.\n\nFigshare: PRISMA-P checklist for The Effect of Surface Treatments on the Bond Strength of Polyetheretherketone Posts: A Systematic Review protocol, https://doi.org/10.6084/m9.figshare.25771182.v1. 22\n\nData are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).\n\n\nReferences\n\nMezied MS, Alhazmi AK, Alhamad GM, et al.: Endocrowns Versus Post-core Retained Crowns as a Restoration of Root Canal Treated Molars - A Review Article. J. Pharm. Bioallied Sci. 2022 Jul; 14(Suppl 1): S39–S42. PubMed Abstract | Publisher Full Text\n\nÖzyürek T, et al.: Fracture strength of endodontically treated teeth restored with different fiber post and core systems. Odontology. 2020; 108: 588–595. PubMed Abstract | Publisher Full Text\n\nArcila L, de Carvalho RN , Bottino M, et al.: Indications, materials and properties of 3D printing in dentistry: a literature overview. ResSocDev. 2020; 9(11). Publisher Full Text\n\nKessler A, Hickel R, Reymus M: 3D printing in dentistry-state of the art. Oper. Dent. 2020; 45(01): 30–40. PubMed Abstract | Publisher Full Text\n\nAlsadon O, Wood D, Patrick D, et al.: Fatigue behavior and damage modes of high performance poly-ether-ketone-ketone PEKK bilayered crowns. J. Mech. Behav. Biomed. Mater. 2020; 110: 103957. PubMed Abstract | Publisher Full Text\n\nBae SY, Park JY, Jeong ID, et al.: Three-dimensional analysis of marginal and internal fit of copings fabricated with polyetherketoneketone (PEKK) and zirconia. J. Prosthodont. Res. 2017; 61(02): 106–112. PubMed Abstract | Publisher Full Text\n\nLee KS, Shin MS, Lee JY, et al.: Shear bond strength of composite resin to high performance polymer PEKK according to surface treatments and bonding materials. J. Adv. Prosthodont. 2017; 9(05): 350–357. PubMed Abstract | Publisher Full Text | Free Full Text\n\nAglar I, Ates SM, Yesil Duymus Z: An in vitro Evaluation of the Effect of Various Adhesives and Surface Treatments on Bond Strength of Resin Cement to Polyetheretherketone: Bond Strength of Resin Cement to PEEK. J. Prosthodont. 2019; 28: e342–e349. Publisher Full Text\n\nMonteiro LC, Pecorari VGA, Gontijo IG, et al.: PEEK and Fiberglass Intra-Radicular Posts: Influence of Resin Cement and Mechanical Cycling on Push-out Bond Strength. Clin. Oral Investig. 2022; 26: 6907–6916. PubMed Abstract | Publisher Full Text\n\nSong CH, Choi JW, Jeon YC, et al.: Comparison of the microtensile bond strength of a polyetherketoneketone (PEKK) tooth post cemented with various surface treatments and various resin cements. Materials (Basel). 2018; 11: 916. PubMed Abstract | Publisher Full Text | Free Full Text\n\nBenli M, Eker GB, Kahraman Y, et al.: Surface characterization and bonding properties of milled polyetheretherketone dental posts. Odontology. 2020; 108: 596–606. PubMed Abstract | Publisher Full Text\n\nÇulhaoğlu A, Özkır S, Şahin V, et al.: Effect of various treatment modalities on surface characteristics and shear bond strengths of polyetheretherketone-based core materials. J. Prosthodont. 2020; 29: 136–141. PubMed Abstract | Publisher Full Text\n\nChaijareenont P, Prakhamsai S, Silthampitag P, et al.: Effects of different sulfuric acid etching concentrations on PEEK surface bonding to resin composite. Dent. Mater. J. 2018; 37: 385–392. PubMed Abstract | Publisher Full Text\n\nMathew S, Raju IR, Sreedev CP, et al.: Evaluation of push out bond strength of fiber post after treating the intra radicular post space with different post space treatment techniques: a randomized controlled in vitro trial. J. Pharm. Bioallied Sci. 2017; 9: S197–S200. PubMed Abstract | Publisher Full Text\n\nMoher D, Liberati A, Tetzlaff J, et al.: Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009; 6(7): e1000097. PubMed Abstract | Publisher Full Text | Free Full Text\n\nMoher D, Shamseer L, Clarke M, et al.: Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Syst. Rev. 2015; 4(1): 1. PubMed Abstract | Publisher Full Text | Free Full Text\n\nMigliavaca CB, Stein C, Colpani V, et al.: How are systematic reviews of prevalence conducted? A methodological study. BMC Med. Res. Methodol. 2020 [cited 2020 Dec 12]; 20(1): 96. PubMed Abstract | Publisher Full Text | Free Full Text\n\nChecklist for Randomized Controlled Trials Critical Appraisal tools for use in JBI Systematic Reviews.[cited 2020 Dec 10].\n\nSterne JAC, Savović J, Page MJ, et al.: RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ (Clinical Research ed.). 2019; l4898. Publisher Full Text\n\nWells GA, Shea B, O’Connell D, et al.: The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. Ottawa Hospital Research Institute; Accessed January 20, 2024.\n\nBoukhris H, Hajjami H, Ben youssef S.: Clinical Outcomes of Polyetheretherketone (PEEK) Hybrid Prosthesis in All-on-Four Rehabilitation: A Systematic Review Protocol [version 1; peer review: awaiting peer review]. F1000Res. 2024; 13(507). Publisher Full Text\n\nBoukhris H: PRISMA-P checklist for The Effect of Surface Treatments on the Bond Strength of Polyetheretherketone Posts: A Systematic Review protocol. Dataset. figshare. 2024. Publisher Full Text"
}
|
[
{
"id": "316835",
"date": "18 Sep 2024",
"name": "Marwa Emam",
"expertise": [
"Reviewer Expertise Fixed prosthodontics",
"implant",
"occlusion",
"ceramics",
"dental materials",
"esthetics",
"digital dentistry."
],
"suggestion": "Approved With Reservations",
"report": "Approved With Reservations\n\ninfo_outline\nAlongside their report, reviewers assign a status to the article:\n\nApproved The paper is scientifically sound in its current form and only minor, if any, improvements are suggested\n\nApproved with reservations\nA number of small changes, sometimes more significant revisions are required to address specific details and improve the papers academic merit.\n\nNot approved Fundamental flaws in the paper seriously undermine the findings and conclusions\n\nReview report for the submitted manuscript “The effect of surface treatments on the bond strength of polyetheretherketone posts: a systematic review protocol”\nI would like to thank the authors for submitting their protocol for review. The use of PEEK posts in dental applications is a developing subject, so a systematic assessment of surface treatment procedures is extremely important. PEEK's biocompatibility and mechanical strength make it excellent for dental posts, but the inert surface makes bonding challenging. This review could fill a crucial gap in clinical practice. Overall, the language is clear, but minor editing of English language might be required in the final submitted review. Author might want to explore options for refining their text to improve its flow and readability. This can make their article more engaging and clear. 1. Title: - Approved as is. 2. Abstract: Kindly give a hint on the different surface treatment techniques of PEEK for the reader.\n\n3. Introduction: -I think the introduction is a bit shallow. I would prefer the introduction to begin with PEEK material (history, chemical composition and properties) as it is the main objective of the study. Then going through the different applications quickly and concentrating on the endodontically treated teeth and post indications and PEEK post (ready & custom made) as one of the options available for clinicians. Finally, stating PEEK adhesion properties as a weakness that different surface treatments aim to improve. -Kindly connect short paragraphs into paragraphs of 6-7 lines paragraphs. -Kindly add references to the second & third paragraph. - The authors could elaborate on the therapeutic relevance of PEEK's bond strength as compared to conventional materials like metal, zirconia or fiber posts. 4. Methods: -The protocol is well-structured and follows known principles for systematic reviews (PRISMA). The use of numerous databases (e.g., PubMed, Scopus, Cochrane Central) will ensure a thorough literature review. The intended use of risk-of-bias tools (ROB II and NOS) and data extraction procedures demonstrates a commitment to evidence of excellent quality synthesis. Intervention types: There is no mention of the surface treatments of PEEK aimed for exploration. Only 2 examples were mentioned so kindly give more details regarding treatments to be studied. Outcomes:\nMore precise definitions would be useful in the outcomes section. While bond strength and retention are indicated, it is unclear how they would be tested. Providing additional details on the anticipated outcomes and their clinical significance could enhance the protocol. Evaluation of methodological quality and risk of bias: -The protocol’s description of bias assessment tools (ROB II and NOS) is good, but how will studies with unclear risks of bias be handled in the analysis? Also kindly give more details regarding the “standardized tools” you have mentioned as means of evaluation. Meta analysis: While the authors highlight the possibility of meta-analysis, it would be interesting to describe how they intend to deal with possible heterogeneity among studies (e.g., differences in surface treatments, study designs). 5. Conclusions: Although this protocol presents a well-designed systematic review that is expected to yield useful information about the surface treatments of PEEK posts in dental applications, it would be strengthened by adding the details addressed in the comments above for further enhancement.\n\nIs the rationale for, and objectives of, the study clearly described? Yes\n\nIs the study design appropriate for the research question? Yes\n\nAre sufficient details of the methods provided to allow replication by others? Yes\n\nAre the datasets clearly presented in a useable and accessible format? Yes",
"responses": [
{
"c_id": "13237",
"date": "06 Feb 2025",
"name": "Hanen Boukhris",
"role": "Author Response",
"response": "We appreciate the reviewers’ valuable feedback, which has helped improve the clarity, depth, and rigor of our manuscript. Below, we provide detailed responses to each comment and indicate the revisions made accordingly. 1.Introduction Comment: \"The introduction should start with PEEK’s material properties, history, and applications before discussing its role in post-and-core restorations.\" Response: We have reorganized the introduction to first present PEEK’s chemical composition, mechanical properties, and biomedical applications before focusing on its role in post-and-core restorations. The revised introduction now follows a logical progression, improving the contextual framework of the study. Comment: \"Please provide more details on the different surface treatment techniques of PEEK.\" Response: We have expanded the discussion on surface treatments, including plasma treatment, acid etching (98% sulfuric acid), sandblasting (50 μm alumina oxide), laser surface modification, silanization, and air abrasion. These modifications and their rationale are now clearly described. Comment: \"There is limited discussion on the comparative performance of PEEK posts against conventional materials like metal, zirconia, and fiber posts.\" Response: The revised manuscript now includes a comparative analysis highlighting the strengths and limitations of PEEK versus conventional post materials, emphasizing its clinical relevance. Comment: \"How will studies with unclear risk of bias be handled in the analysis?\" Response: We have specified that studies with an unclear risk of bias will undergo sensitivity analysis, and their influence on the final conclusions will be carefully considered. Comment: \"Clarify how heterogeneity among studies will be addressed in the meta-analysis.\" Response: A heterogeneity assessment will be conducted using I² statistics, and subgroup analysis or meta-regression will be applied if significant heterogeneity is detected. Comment: \"Provide more precise definitions of key outcomes.\" Response: We have revised the outcomes section to specify assessment methods for bond strength (e.g., push-out tests), fracture resistance, and marginal adaptation quality. These responses, along with the corresponding revisions, have strengthened our manuscript and improved its clarity. Thank you for your constructive feedback."
}
]
}
] | 1
|
https://f1000research.com/articles/13-951
|
https://f1000research.com/articles/13-825/v1
|
23 Jul 24
| {"type":"Systematic Review","title":"The use of medicinal marijuana for treating Cerebral Palsy: a l(...TRUNCATED)
| [{"id":"330978","date":"28 Oct 2024","name":"Natalia Szejko","expertise":["Reviewer Expertise Cannab(...TRUNCATED)
| 1
|
https://f1000research.com/articles/13-825
|
https://f1000research.com/articles/13-1367/v1
|
15 Nov 24
| {"type":"Clinical Practice Article","title":"“Unmasking the Uncommon”: A case series of multi-dr(...TRUNCATED)
| [{"id":"341910","date":"25 Nov 2024","name":"Godwin Justus Wilson","expertise":["Reviewer Expertise (...TRUNCATED)
| 1
|
https://f1000research.com/articles/13-1367
|
https://f1000research.com/articles/12-199/v1
|
20 Feb 23
| {"type":"Research Article","title":"Social distancing between personal belongings during the COVID-1(...TRUNCATED)
| [{"id":"186686","date":"24 Jul 2023","name":"Leon O. H Kroczek","expertise":["Reviewer Expertise exp(...TRUNCATED)
| 1
|
https://f1000research.com/articles/12-199
|
https://f1000research.com/articles/12-1441/v1
|
07 Nov 23
| {"type":"Research Article","title":"Epidemiological profiles and causes of sudden deaths of various (...TRUNCATED)
| [{"id":"244085","date":"22 Feb 2024","name":"Lorraine Du Toit-Prinsloo","expertise":["Reviewer Exper(...TRUNCATED)
| 1
|
https://f1000research.com/articles/12-1441
|
https://f1000research.com/articles/12-1094/v1
|
01 Sep 23
| {"type":"Research Article","title":"Food consumption and adherence to dietary guidelines among Jorda(...TRUNCATED)
| [{"id":"204311","date":"12 Sep 2023","name":"Lana Agraib","expertise":["Reviewer Expertise My areas (...TRUNCATED)
| 1
|
https://f1000research.com/articles/12-1094
|
End of preview. Expand
in Data Studio
RottenReviews: Benchmarking Review Quality with Human and LLM-Based Judgments
Quick links: 📃 Paper | ⚙️ Code
RottenReviews is a benchmark dataset designed to facilitate research on peer review quality assessment using multiple types of evaluation signals, including human expert annotations, structured metrics derived from textual features, and large language model (LLM)-based judgments.
Note: This HF repo only contains the raw files and the human annotation data records. Some dataset components are available only in our Google Drive. Follow repository documentation for downloading the processed files.
🧠 Dataset Summary
Peer review quality is central to the scientific publishing process, but systematic evaluation at scale is challenging. The RottenReviews dataset addresses this gap by providing a large corpus of academic peer reviews enriched with reviewer metadata and multiple quality indicators:
- Raw peer reviews from multiple academic venues (e.g., F1000Research, Semantic Web Journal, ICLR, NeurIPS) spanning diverse research areas
- Reviewer profiles (when available) linked via external scholarly metadata
- Quantifiable metrics capturing interpretable aspects of review text and reviewer behavior (e.g., lexical diversity, topical alignment, hedging)
- Human expert annotations over a subset of reviews across multiple quality dimensions (e.g., clarity, fairness, comprehensiveness)
- LLM-based judgments generated using structured prompts for automated quality assessment
The dataset was introduced to support research on benchmarking and modeling peer review quality at scale. It contains thousands of submissions and reviewer profiles, making it one of the most comprehensive resources for peer review quality analysis.
📌 Usage Example
from datasets import load_dataset
dataset = load_dataset("Reviewerly/RottenReviews", "ICLR2024") # Select partiotion from ['ICLR2024', 'NIPS2023', 'F1000Journal', 'SemanticWebJournal', 'human_annotated_data']
# Access processed reviews
processed_reviews = dataset["data"]
print(processed_reviews[0])
🎯 Tasks & Applications
RottenReviews supports a wide range of research tasks, including:
- Peer Review Quality Prediction
- Benchmarking LLM-Based Review Evaluation Methods
- Correlation Analysis Between Metrics and Human Judgments
- Reviewer Behavior and Metadata Modeling
- Interpretability Studies for Review Quality Signals
🧾 License & Citation
The dataset and accompanying code are released under the license specified in the RottenReviews repository. If you use this dataset in academic work, please cite the accompanying RottenReviews paper.
@inproceedings{ebrahimi2025rottenreviews,
title={RottenReviews: Benchmarking Review Quality with Human and LLM-Based Judgments},
author={Ebrahimi, Sajad and Sadeghian, Soroush and Ghorbanpour, Ali and Arabzadeh, Negar and Salamat, Sara and Li, Muhan and Le, Hai Son and Bashari, Mahdi and Bagheri, Ebrahim},
booktitle={Proceedings of the 34th ACM International Conference on Information and Knowledge Management},
series = {CIKM '25},
pages={5642--5649},
year={2025},
url = {https://doi.org/10.1145/3746252.3761506},
doi = {10.1145/3746252.3761506}
}
- Downloads last month
- 17