---
license: cc-by-4.0
task_categories:
  - tabular-classification
language:
  - en
tags:
  - healthcare
  - health-workforce
  - burnout
  - brain-drain
  - retention
  - task-shifting
  - sub-saharan-africa
  - lmic
pretty_name: "Healthcare Worker Workforce & Retention (Burnout, Migration, Staffing)"
size_categories:
  - 10K<n<100K
configs:
  - config_name: urban_tertiary
    data_files: data/hcw_urban_tertiary.csv
  - config_name: district_hospital
    data_files: data/hcw_district_hospital.csv
    default: true
  - config_name: rural_health_centre
    data_files: data/hcw_rural_health_centre.csv
---

# Healthcare Worker Workforce & Retention Dataset

## Abstract

This dataset provides **30,000 simulated healthcare worker records** (10,000 per scenario) from sub-Saharan Africa. Each record contains 45+ variables including cadre, workload, burnout, salary, migration intentions, and workplace safety. Three settings: urban tertiary (22% high burnout), district hospital (59%), and rural health centre (83%).

## 1. Introduction

SSA faces a critical healthcare worker shortage that could be 93% larger than current estimates when accounting for disease burden (McKinsey 2024). Brain drain, low salaries, burnout, and poor working conditions drive attrition. Task shifting is widespread but insufficient. WHO's Health Workforce Support and Safeguards List highlights the severity.

**This dataset is entirely simulated. It must not be used for workforce planning decisions.**

## 2. Methodology

### 2.1 Parameterization

| Parameter | Value | Source |
| --- | --- | --- |
| HCW shortage scale | 93% larger | McKinsey 2024 |
| Brain drain | Major issue | ScienceDirect 2025 |
| Task shifting | Widespread | PMC 2025 |
| Salary dissatisfaction | ~55% | WHO AFRO |
| Vacancy rate rural | ~60% | BMJ GH 2022 |

### 2.2 Scenario Design

| Scenario | Doctor Ratio | Vacancy | Burnout High | Migration |
| --- | --- | --- | --- | --- |
| Urban tertiary | 15% | 25% | 22% | 24% |
| District hospital | 5% | 41% | 59% | 16% |
| Rural HC | 1% | 61% | 83% | 8% |

## 3. Schema

| Column | Type | Description |
| --- | --- | --- |
| id | int | Unique identifier |
| age | int | Worker age |
| sex | categorical | M / F |
| cadre | categorical | doctor / nurse / clinical_officer / lab_tech / pharmacist / chw |
| burnout_score | int | Burnout score (0-100) |
| burnout_level | categorical | low / moderate / high |
| intention_to_leave | binary | Plans to leave |
| intention_to_migrate | binary | Plans to migrate abroad |
| migration_destination | categorical | UK / USA / Middle East / SA / Europe |
| salary_satisfaction | categorical | satisfied / neutral / dissatisfied |
| vacancy_rate | derived | Staffing adequacy |
| task_shifting | binary | Performing shifted tasks |

## 4. Validation

<p align="center">
  <img src="validation_report.png" alt="Validation Report" width="100%">
</p>

Key validation checks:

- **Burnout gradient**: 22% → 59% → 83% ✓
- **Vacancy gradient**: 25% → 41% → 61% ✓
- **Salary dissatisfied**: ~55% ✓
- **Migration higher in urban** (doctors) ✓
- **Low salary = #1 reason** to leave ✓

## 5. Usage

```python
from datasets import load_dataset
dataset = load_dataset("electricsheepafrica/health-workforce-retention", "district_hospital")
df = dataset["train"].to_pandas()
```

## 6. Limitations

- **Simulated**: Not from real HR records.
- **No facility-level data**: No bed counts or catchment.
- **Simplified**: No detailed training pipeline.
- **Cross-sectional**: No longitudinal tracking.

## 7. References

1. McKinsey (2024). SSA healthcare worker shortage.
2. ScienceDirect (2025). HCW migration mitigation Africa.
3. PMC (2025). Skilled HCW shortage and surplus.
4. BMJ GH (2022). Projected HWF requirements 2023-2030.
5. PMC (2025). HWF shortage addressing health needs.

## Citation

```bibtex
@dataset{esa_hcw_workforce_2025,
  title={Healthcare Worker Workforce and Retention Dataset},
  author={Electric Sheep Africa},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/electricsheepafrica/health-workforce-retention}
}
```

## License

[CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)