The power of health in the Democratic Republic of the Congo: Electricity quality and reliability in medical facilities of North Kivu Province

High-quality reliable electricity is essential for health service delivery, yet current World Health Organization (WHO) and World Bank Multi-Tier Framework (MTF) indicators rely on binary, recall-based survey metrics that fail to capture the incidence, duration, voltage or frequency of electricity in low- and middle-income country (LMIC) health facilities. Between 2022 and 2024, we deployed real-time, cellular-connected sensors in 25 health facilities across North Kivu Province, Democratic Republic of Congo (DRC), to continuously monitor outages, “uptime,” and the percentage of time electricity supply fell within ±10 % of nominal voltage and ± 5 % of nominal frequency. Facilities experienced an average of 0.67 outages per day (min–max: 0–2.04), with a median duration of 2.01 h (0.03–10.9). Voltage compliance averaged 73.9 % (10.0–99.6 %) and frequency compliance 82.1 % (9.3–99.7 %). Sixteen facilities achieved ≥99 % uptime, the upper threshold mapping to the WHO's current definition of reliable electricity. Uptime varied widely (46.7–100 %), underscoring the limitations of a reliable/not-reliable approach. Compared to those in Nigeria and Kenya, North Kivu facilities experienced fewer outages but substantially longer durations, with better voltage quality than Nigeria but worse frequency quality. By demonstrating the feasibility of continuous, auditable monitoring in a conflict-affected LMIC, this study shows how sensor-based metrics provide more sensitive and clinically relevant indicators of electricity performance than WHO and MTF frameworks. These findings support updating global electrification metrics to incorporate real-time uptime, voltage, and frequency indicators that better capture risks to biomedical equipment and patient outcomes. • We collected power quality and reliability metrics at 25 facilities in North Kivu, eastern DRC over 29 months. • Mean outage incidence across the sample was 0.67/day (min–max: 0–2.04), median outage duration: 2.01 h (0.03–10.9). • 73.9 % of recorded voltage values fell within ±10 % of nominal 230 V (min–max: 10.0–99.6 %) and 82.1 % of frequency values fell within ±5 % of nominal 50 Hz (9.3–99.7 %). • 16 out of the 25 facilities achieved “two nines” (99 %) of uptime. Facility uptime ranged from 46.7 % to 100 % (mean 93.2 %, median: 99.5 %, SD: 14.5). • Outage incidence was overall lower in North Kivu than in comparable samples in Nigeria and Kenya, but outage durations were longer. Frequency quality was overall lower in North Kivu than in Nigeria and Kenya, but voltage quality was better than in Nigeria. • Sensors offer significant improvements over survey recall in assessing the impact of energy poverty on healthcare outcomes.