electricsheepafrica/vaccine-cold-chain

Vaccine Cold Chain Management Dataset

Abstract

This dataset provides 30,000 simulated facility-level observations (10,000 per scenario) of vaccine cold chain performance across three tiers of the immunisation supply chain in sub-Saharan Africa. Each record represents a single EPI vaccine product assessed at a health facility during one monthly reporting period. The dataset captures 45+ variables spanning cold chain equipment status, continuous temperature monitoring, freeze and heat excursion events, vaccine vial monitor (VVM) status, wastage rates and causes, stock availability, outreach session delivery, and downstream immunisation impact. Three facility-level scenarios are modelled: regional vaccine store (87% temperature compliance), district hospital (65%), and rural health post (40%) — reflecting the steep gradient documented in multi-country cold chain assessments.

This dataset is entirely simulated. It must not be used for immunisation programme decisions, procurement, or regulatory purposes.

1. Introduction

1.1 The Vaccine Cold Chain Challenge

The Expanded Programme on Immunization (EPI) depends on an unbroken cold chain to deliver potent vaccines from manufacturer to child. The World Health Organization mandates that vaccines be stored continuously between +2°C and +8°C, with both freezing and heat exposure capable of permanently damaging vaccine potency — deterioration that leaves no visible signs (WHO, 2023). In sub-Saharan Africa, where immunisation infrastructure faces chronic underinvestment, maintaining this narrow temperature window is a formidable operational challenge.

1.2 Temperature Excursions: The Scale of the Problem

Automated temperature monitoring studies have revealed alarming excursion rates. A study in Cameroon using continuous electronic loggers found that freeze exposures were highest at the facility level (51%), followed by the district level (31%), with minimum temperatures reaching as low as −23.8°C. Critically, 96% of all vaccine shipments were exposed to temperatures above +8°C at some point during transport and storage (Walsh Medical Media, Cameroon). In south-west Ethiopia, automated recording devices documented frequent excursions outside the +2°C to +8°C range at routine EPI facilities, with many excursions going undetected by twice-daily manual monitoring (ScienceDirect, 2018).

A 2023 study of vaccine cold chain management in Ethiopian public health facilities found widespread gaps in knowledge and practice, with shortages caused by poor forecasting, delays in supply requests, and inadequate temperature monitoring infrastructure (PMC, 2023; doi:10.2147/JMDH.S385466). A 2025 assessment in Tanzania documented poor handling practices and infrastructure challenges, particularly in remote rural areas (PMC, 2025; PMC11951318).

1.3 Vaccine Wastage

Vaccine wastage is an inevitable consequence of multi-dose vial policies and cold chain failures. WHO-acceptable wastage rates vary by vaccine and vial size: 5% for single-dose and liquid multi-dose vials (e.g., Pentavalent, PCV), 10% for OPV, 15% for large multi-dose liquid vials, and up to 50% for reconstituted lyophilised vaccines like BCG (20-dose vial) (WHO EPI guidance; Open University HEAT module).

A multi-country study across Ghana, Mozambique, and Pakistan found that open-vial wastage for vaccines in single-dose vials or multi-dose vials kept refrigerated up to four weeks after opening ranged from 0.08% to 3%. However, for multi-dose vials where remaining doses must be discarded within six hours of opening, wastage rates were substantially higher (ScienceDirect, 2023). In The Gambia, BCG exhibited the highest wastage rates due to its 20-dose vial presentation (PMC, 2018; PMC6042329). A 2025 study in Uganda documented wastage patterns across BCG, OPV, IPV, Rotavirus, DTP-HepB-Hib, and Measles-Rubella vaccines (PMC, 2025; PMC12151339).

1.4 Equipment and Infrastructure

The cold chain equipment landscape in SSA is undergoing a transition from kerosene/gas-powered refrigerators to solar direct-drive (SDD) units, supported by Gavi and PATH investments. However, equipment functionality rates remain low — particularly at peripheral health facilities where maintenance capacity is absent. Many rural health posts rely solely on vaccine carriers with conditioned ice packs for outreach sessions, with no permanent refrigeration on site (PATH, 2022; Clean Cooling, 2025).

1.5 Rationale for This Dataset

Despite extensive survey evidence, no open, standardised, multi-variable dataset integrates cold chain equipment status, temperature monitoring, excursion events, wastage metrics, stock availability, and immunisation impact into a single analytical resource. This simulated dataset addresses that gap, enabling researchers and programme managers to develop predictive models for cold chain failure, benchmark wastage reduction strategies, and prototype equipment investment decision tools.

2. Methodology

2.1 Epidemiological Parameterization

Parameter	Regional Store	District Hospital	Rural Post	Source
Temp in range (% time)	87%	65%	40%	[2] Cameroon: 51% freeze at facility
Freeze excursion rate	9%	27%	41%	[2] 51% at facility level
Heat excursion rate	18%	40%	66%	[2] 96% of shipments exposed
Equipment functional	81%	51%	6%	[9] PATH assessment
Mean wastage rate	14%	20%	32%	[4][5] Multi-country studies
Vaccine stockout rate	8%	20%	45%	EPI programme data
VVM checking rate	80%	50%	15%	[7] WHO EVM assessment
Monthly report submitted	92%	65%	25%	EPI programme reports

2.2 Vaccine Product Selection

The dataset includes 12 EPI vaccines reflecting the standard SSA immunisation schedule:

Vaccine	Doses/Vial	Reconstitution	Freeze-Sensitive	Acceptable Wastage
BCG	20	Yes	No	50%
OPV	20	No	No	15%
IPV	5	No	Yes	5%
Pentavalent (DPT-HepB-Hib)	10	No	Yes	5%
PCV13	1	No	Yes	5%
Rotavirus	1	No	Yes	5%
Measles-Rubella	10	Yes	No	25%
Yellow Fever	5	Yes	No	25%
HPV	1	No	Yes	5%
Tetanus Toxoid	10	No	Yes	10%
COVID-19 mRNA	6	No	No	10%
Typhoid TCV	1	No	Yes	5%

2.3 Scenario Design

Scenario A — Regional Vaccine Store: Walk-in cold room, continuous electronic temperature monitoring (30-DTR or Fridge-tag), trained EPI officer, solar + generator backup, monthly distribution to districts. Analogous to regional stores in Tanzania, Ethiopia, Ghana.

Scenario B — District Hospital EPI Unit: Domestic or ice-lined refrigerator, twice-daily manual temperature monitoring, EPI nurse, irregular power supply, no backup generator. Analogous to district facilities in Uganda, Malawi, Cameroon.

Scenario C — Rural Health Post: Vaccine carrier with conditioned ice packs only, no permanent refrigeration, no temperature monitoring device, monthly outreach-based vaccination, CHW-led. Analogous to health posts in Niger, DRC, South Sudan.

3. Schema

3.1 Facility & Staffing

Column	Type	Description
facility_level	categorical	regional_store / district_hospital / rural_health_post
facility_id	string	Anonymised facility identifier
region_type	categorical	urban / peri_urban / rural
has_epi_officer	binary	Dedicated EPI officer on staff
cold_chain_staff_trained	binary	Staff trained in cold chain management

3.2 Equipment

Column	Type	Description
equipment_type	categorical	solar_direct_drive / electric_compression / gas_kerosene / ice_lined_refrigerator / vaccine_carrier_only / none
equipment_age_years	int	Age of cold chain equipment
equipment_functional	binary	Equipment operational at observation
backup_power_available	binary	Generator or solar backup present
power_outage_hours_last_month	int	Total hours of power outage

3.3 Temperature Monitoring

Column	Type	Description
monitoring_type	categorical	electronic_continuous / manual_twice_daily / none
monitoring_device_present	binary	Temperature monitoring device available
temp_log_complete	binary	Temperature log fully completed
temp_in_range_pct	float	Percentage of time temperature was +2°C to +8°C
freeze_excursion_detected	binary	Temperature dropped below +2°C
heat_excursion_detected	binary	Temperature exceeded +8°C
min_temp_recorded_C	float	Minimum temperature recorded (°C)
max_temp_recorded_C	float	Maximum temperature recorded (°C)
excursion_cause	categorical	10 root cause categories

3.4 Vaccine Product

Column	Type	Description
vaccine_name	categorical	12 EPI vaccines
doses_per_vial	int	Doses per vial (1–20)
requires_reconstitution	binary	Lyophilised vaccine requiring reconstitution
freeze_sensitive	binary	Damaged by freezing
heat_sensitive	binary	Damaged by heat exposure
acceptable_wastage_pct	int	WHO-acceptable wastage threshold

3.5 Quality & VVM

Column	Type	Description
vvm_checked	binary	Vaccine vial monitor inspected
vvm_stage	categorical	stage_1 / stage_2 / stage_3 / stage_4_discard / not_checked
shake_test_done	binary	Shake test performed for freeze-sensitive vaccines
vaccines_discarded_freeze	binary	Vaccines discarded due to freeze damage

3.6 Stock & Wastage

Column	Type	Description
vaccine_in_stock	binary	Vaccine available at observation
stockout_days_last_month	int	Days of stockout in preceding month
doses_received_last_quarter	int	Doses received from supply chain
doses_administered_last_quarter	int	Doses administered to patients
wastage_rate_pct	float	Actual wastage rate (%)
wastage_within_acceptable	binary	Wastage within WHO threshold
doses_wasted_last_quarter	int	Absolute doses wasted
wastage_cause_primary	categorical	open_vial_expiry / cold_chain_break / expired_stock / damaged_vial / contamination / VVM_stage4

3.7 Immunisation Impact

Column	Type	Description
outreach_sessions_planned	int	Planned outreach/vaccination sessions
outreach_sessions_conducted	int	Actually conducted sessions
children_missed_due_to_stockout	int	Children who missed vaccination
monthly_report_submitted	binary	Monthly EPI report submitted
report_timely	binary	Report submitted within deadline

4. Validation

4.1 Key Validation Results

Metric	Regional Store	District Hospital	Rural Post	Literature Target
Temp in range	87%	65%	40%	51% freeze at facility [2]
Freeze excursion	9%	27%	41%	51% facility level [2]
Heat excursion	18%	40%	66%	96% of shipments [2]
Equipment functional	81%	51%	6%	PATH data [9]
Vaccine in stock	92%	81%	55%	EPI targets
Mean wastage	14%	20%	32%	Multi-country [4][5]
Report submitted	92%	65%	25%	EPI programme data

4.2 Internal Consistency Checks

• Freeze-sensitive vaccines (Pentavalent, IPV, PCV, Rotavirus) show higher discard rates when freeze excursions are detected.
• Wastage rates for BCG (20-dose vial, reconstituted) are highest across all scenarios, consistent with literature.
• Equipment functionality is near-zero at rural health posts where vaccine carriers are the primary equipment.
• Power outage hours inversely correlate with temperature compliance.
• VVM stage 3/4 rates increase with heat excursion frequency.

5. Usage

from datasets import load_dataset

dataset = load_dataset(
    "electricsheepafrica/vaccine-cold-chain",
    "district_hospital"
)
df = dataset["train"].to_pandas()

# Analyse excursion rates by equipment type
excursion = df.groupby('equipment_type').agg({
    'freeze_excursion_detected': 'mean',
    'heat_excursion_detected': 'mean',
    'wastage_rate_pct': 'mean'
}).round(3)
print(excursion)

5.1 Suggested Analyses

• Cold chain failure prediction: Classify facilities likely to experience temperature excursions based on equipment, staffing, and infrastructure features.
• Wastage optimisation: Model the relationship between vial size, session frequency, and wastage to inform vial presentation decisions.
• Equipment investment prioritisation: Identify which facility types would benefit most from solar direct-drive refrigerator deployment.
• Outreach effectiveness: Analyse the relationship between cold chain reliability and outreach session completion rates.

6. Limitations

• Simulated data: Generated from statistical models parameterised by literature, not from real EPI programme data.
• No spatial modelling: Facilities are not geolocated; distance-to-supply-hub effects are approximated through scenario design.
• Simplified excursion physics: Temperature excursions are modelled as binary events rather than continuous time-temperature profiles.
• No seasonal variation: Ambient temperature effects on cold chain performance are not explicitly modelled.
• Single product per record: Cross-vaccine stockout correlations within a facility are not captured.

7. References

1. WHO (2023). Vaccine Management Handbook. EVM Standard Operating Procedures. +2°C to +8°C continuous monitoring.
2. Walsh Medical Media (Cameroon). Temperature monitoring in the vaccine cold chain in Cameroon. Freeze at facility 51%, heat 96%.
3. ScienceDirect (2018). Challenges of cold chain quality for routine EPI in south-west Ethiopia. doi:10.1016/j.vaccine.2018.05.073
4. ScienceDirect (2023). Vaccine wastage in Ghana, Mozambique, Pakistan. Open-vial wastage 0.08–3%. doi:10.1016/j.vaccine.2023.05.020
5. PMC (2018). Vaccine wastage in The Gambia: a prospective observational study. PMC6042329.
6. PMC (2025). Vaccine wastage rates and attributed factors in Uganda. PMC12151339.
7. PMC (2023). Vaccine cold chain management and associated factors, Ethiopia. doi:10.2147/JMDH.S385466
8. PMC (2025). Vaccine handling practices and conformity to cold chain systems, Tanzania. PMC11951318.
9. PATH (2022). Vaccine supply and cold chain systems in sub-Saharan Africa: current state and future directions.
10. Clean Cooling (2025). Optimising the vaccine cold chain for resilient immunisation in Africa.
11. Open University HEAT Module. Immunization: Vaccine supply and cold chain. Acceptable wastage rates by vaccine type.

Citation

@dataset{esa_vaccine_cold_chain_2025,
  title   = {Vaccine Cold Chain Management Dataset: Temperature Excursions,
             Equipment Status, Wastage, and Immunisation Impact Across Three
             Tiers of the EPI Supply Chain in Sub-Saharan Africa},
  author  = {{Electric Sheep Africa}},
  year    = {2025},
  publisher = {Hugging Face},
  url     = {https://huggingface.co/datasets/electricsheepafrica/vaccine-cold-chain},
  note    = {Simulated dataset. Not for immunisation programme decisions.}
}

License

CC-BY-4.0