Stock records in Ethiopian health facilities: what the studies measured

Ask why a clinic ran out of amoxicillin and the answer usually offered is supply: the wholesaler was late, the agency was short, the budget ran out. Sometimes that is true. But the Ethiopian literature keeps surfacing a less comfortable answer — that the facility's own record of what it holds was wrong, so the reorder was placed against a number that did not exist.

The measured gap between the card and the shelf

Ethiopian studies use a clean, hard metric: inventory accuracy, defined as the absence of any discrepancy between the bin card and a physical count of the same medicine 1. It is a check anyone can run in an afternoon, and the results cluster tightly:

Two independent studies, different zones, different years — and effectively the same overall number.

Source: Alemu et al. 2023 [1]; Bekele et al. 2025 [2]

Show data table

	Inventory accuracy: bin card matched against a physical count
Primary hospitals, Gamo Zone	93.5%
Health centres, Gamo Zone	79.7%
Overall, Gamo Zone primary health care units	78.5%
Public facilities, Jimma Zone	78.3%
Health posts, Gamo Zone	62.2%

Two independent studies, different zones, different years, effectively the same number: roughly one in five stock records does not match the shelf. The Jimma Zone study also found bin cards were present for 85–100% of the essential medicines it tracked 2 — so this is not a tooling gap. The cards exist. They are simply out of date, because updating them is a second manual act that happens after the medicine has already moved.

What inaccurate records cost

Stock-outs. In Shegaw Motta general hospital, 60% of tracked essential medicines were out of stock at least once in six months, with an average stock-out duration of 38.8 days and a longest case of 157 days; the nearby health centre fared better at 20% and 11.2 days 3. In Jimma Zone, 14 of 20 facilities (70%) experienced a stock-out, averaging 33.7 days, and all 20 facilities had placed an emergency order within six months 2. An emergency order is the visible cost of an invisible record error.

Expiry. The other side of a bad record is stock nobody knew they had. Across nine Jimma Zone hospitals over two fiscal years, medicine wastage ran at 4.87% — against Ethiopia's 2% national target — for a loss of $32,453.30, plus a further $2,711.49 to dispose of it 4. Across the western Ethiopian public supply chain, 239,801 unit packs worth 20,538,198.93 ETB expired, and the strongest associated factor was poor store management (AOR 9.72, 95% CI 1.47–64.08) 5 — not procurement volume, not budget, not demand forecasting.

Wastage as a share of stock value. Both studies exceed the target; the western Ethiopia figure covers the public supply chain including EPSA.

Source: Getahun et al. 2024 [4]; Diriba et al. 2023 [5]

Show data table

	Medicine expiry against the 2% national target
Public supply chain, western Ethiopia	5%
Nine hospitals, Jimma Zone (2 fiscal years)	4.87%
National target	2%

The medicines lost are not marginal. Western Ethiopia's expiry was led by vaccines (28%) and anaesthesia drugs (19%) 5; Jimma's by central nervous system drugs (21.6%) and anti-infectives (19.6%) 4.

Share of expired stock value by therapeutic class; total 20,538,199 ETB across 239,801 unit packs.

Source: Diriba et al. 2023 [5]

Show data table

	What expired in the western Ethiopia supply chain
Vaccines	28%
Anaesthesia drugs	19%
Anti-infectives	18%
Central nervous system drugs	17%
Electrolytes	8%

Shocks amplify whatever the baseline was

The COVID-19 period gives a natural experiment in what happens when a system with poor stock visibility meets a supply disruption. Across selected Ethiopian public pharmaceutical supply agencies, mean availability of essential generic medicines fell from 67.4% to 43.3%, and mean days out of stock per month rose from 11.7 to 15.7 6.

Selected Ethiopian public pharmaceutical supply agencies. Both differences were significant at p<0.001.

Before COVID-19During COVID-19

Source: Melaku et al. 2024, BMJ Open [6]

Show data table

	Before COVID-19	During COVID-19
Mean medicine availability	67.4%	43.3%
Line-item fill rate	78%	49%

A system already short of visibility had no buffer when supply tightened.

Source: Melaku et al. 2024, BMJ Open [6]

Show data table

	Mean days out of stock per month
During COVID-19	15.7 days
Before COVID-19	11.7 days

A facility that already cannot see its own stock accurately has no capacity to absorb a shock like that. Visibility is not a nice-to-have that pays off in calm periods; it is the buffer.

What would actually change the number

The studies are unusually consistent about mechanism, which makes the implications concrete:

1. Deduct at the point of consumption. Accuracy fails because recording is a separate action from dispensing. If the same action that gives the medicine to the patient decrements the count, the discrepancy has nowhere to originate.
2. Make expiry a scheduled prompt, not a discovery. Expiry losses concentrate in slow-moving, high-value items — exactly the stock a human does not walk past daily.
3. Treat emergency orders as a defect signal. All 20 facilities in Jimma placed one within six months 2. Counted and reviewed, that number is a leading indicator of record failure.
4. Keep the physical count. No system removes the need to verify against the shelf; it only reduces how far the two drift between counts.

What the evidence does not show

Almost all of this research was conducted in public facilities — health posts, health centres, and government hospitals — often on a defined list of tracer or essential medicines. We found no equivalent published measurement of stock-record accuracy in Ethiopian private clinics, which face different incentives: stock is bought with the clinic's own money, so shrinkage and expiry hit the owner directly, while the range of items held is usually narrower.

It is reasonable to expect private clinics to do somewhat better on the items they sell frequently, and no better at all on the ones they don't. But that is an inference, not a finding, and we flag it as such. The measurement — bin card versus physical count, on a sample of items, on an ordinary Tuesday — is cheap enough that any clinic can establish its own baseline rather than borrow ours.