How many hours of power does a hospital need? DRC evidence says 22
Health facilities in three provinces of the Democratic Republic of the Congo with power sufficient to run everything they needed, available for at least 22 hours a day, had inpatient mortality about 41% lower than those without, according to a study of 303 facilities in PNAS. Its authors want reliability targets tied to patient outcomes.
Hospital electricity reliability has been linked to inpatient mortality in a study of 303 hospitals and reference health centres in the Democratic Republic of the Congo, which found mortality about 41% lower where power was sufficient to run everything a facility needed and available for at least 22 hours a day.
Published in the Proceedings of the National Academy of Sciences and reported by ESI Africa on 20 October, the research covers facilities in North and South Kivu and in Haut-Katanga, 18 of them fitted with power sensors. Its authors include Ryan McCord, a postdoctoral researcher in energy economics at Duke University, and Laura Kwong, an assistant professor of environmental health sciences at the University of California, Berkeley. Across the sample that meant inpatient mortality of about 2.6% against roughly 4.4%, or, as McCord puts it, “about 18 fewer deaths for every 1,000 patients admitted”.
“A connected hospital may still lose power for hours at a time, or it may have some electricity without enough capacity to run all of its wards and equipment,” says McCord, for whom knowing whether a hospital has an electricity connection “is not enough”. “Those differences appear to matter a great deal for patients.”
A small solar system might reliably power “lights and a few basic loads” and still be “insufficient to run oxygen concentrators or other clinical equipment that is critical for patient care”, in McCord’s example, while “a large generator may provide plenty of power when it is running, but fuel shortages or breakdowns can leave the facility without electricity for long periods of time.” “It really seems to be the combination of these two things that makes the difference,” he says.
Threshold tests put the change at about 21 hours a day across the full sample and about 22 hours in the facilities with sensors, and requiring only one of the two conditions reduced the effect and removed its statistical significance. “We did not know beforehand that 22 hours would turn out to be where the mortality relationship appeared,” McCord says, and he wants “reliability targets tied to the outcomes we actually care about, in this case patient outcomes.”
“I would not take our estimate and say that installing a particular solar system at any hospital will automatically reduce mortality by 41% overnight,” says Kwong, who points out that many other things might need to change, from the equipment a facility buys to how its staff are trained. “So even in the places that maybe have made all of those other investments, there are consequences to poor reliability, even in the short-term.”
The largest differences in mortality were in internal medicine and paediatric wards, where McCord says patients “may depend on things like oxygen that cannot be disrupted when the power goes out without severe consequences to patients”, and facilities with good power also saw 17% more outpatients and 24% more nurse working days.
