All research
Financing & billing5 min read6 sources

Who pays: out-of-pocket, insurance, and why billing accuracy is a supply problem

The most-quoted figure for Ethiopian out-of-pocket spending is 31%. The WHO-sourced series puts it at 46.3% for 2023. Both are cited here, because the gap between them is the whole point about dated statistics.

Published

Share

Start with a disagreement, because it demonstrates the problem with undated health statistics better than any argument could.

The figure quoted almost everywhere for Ethiopian out-of-pocket health spending is 31% — it derives from Ethiopia Health Accounts and is repeated in the peer-reviewed literature as recently as 2025 2. The WHO Global Health Expenditure Database series, distributed through World Bank Open Data, puts out-of-pocket payments at 46.3% of current health expenditure in 2023, up from 33.1% in 2020 and 37.0% in 2021 1.

Out-of-pocket payments as a share of current health expenditure

WHO Global Health Expenditure Database series for Ethiopia. The widely quoted 31% predates this window.

Source: WHO GHED via World Bank Open Data [1]

Show data table
Out-of-pocket payments as a share of current health expenditure
201835.5%
201937.8%
202033.1%
202137%
202244.6%
202346.3%

Those are different measurement bases and different years, and neither is wrong. But a clinic planning around "about a third" when the most recent internationally comparable series says "about a half" is planning around the wrong decade. Every statistic needs a year attached to it.

Households pay a large share directly, and it hurts a minority badly

Using the 2018/19 Ethiopia Socioeconomic Survey (6,770 households), 1.49% of households faced catastrophic health expenditure and 0.89% faced impoverishing expenditure at the 10% threshold 2. Households with a member making more facility visits had higher odds of catastrophic spending 2.

Those national percentages sound small, and they are frequently misreported as far larger figures. What they describe is a concentrated harm: most households spend little, and a small share are pushed under by a single episode of care.

For the facility, billing accuracy is a drug-supply input

The clearest reason to care about billing in an Ethiopian clinic is not accounting hygiene. It is that retained revenue pays for medicines.

Ethiopia's primary health care costing study found internal revenue collected through user fees and insurance payments funded almost 12% of drug and medical-supply costs in health centres and nearly 28% in primary hospitals 3.

{
  "type": "bars",
  "title": "Share of drug and medical-supply costs funded by facility-retained revenue",
  "subtitle": "Money lost in billing does not surface as a smaller bank balance — it surfaces as an empty shelf.",
  "unit": "%",
  "max": 40,
  "source": "Alebachew et al. 2023 [3]",
  "data": [
    { "label": "Primary hospitals", "value": 28 },
    { "label": "Health centres", "value": 12 }
  ]
}
``` Money lost in the billing process does not show up as a smaller balance at year end. It shows up as an empty shelf — and, per the outpatient satisfaction evidence, a patient who does not get their prescription is far less likely to be satisfied [6](#ref-6).

Ethiopia's provider payment assessment confirms fee-for-service remains the backbone of retained facility revenue, while noting it drives higher claim volumes — and therefore more surface area for error [5](#ref-5).

## Claims fail on arithmetic, not on clinical judgment

For clinics contracted with community-based health insurance, the second leak is the claim itself. A mixed-methods study of CBHI schemes in two districts of northeast Ethiopia documented claims **partially rejected after medical audit** for a specific and mundane set of reasons [4](#ref-4):

- requests for services outside the covered list
- **mathematical errors**
- billing for patients belonging to other districts
- **duplicate medicine orders within a single treatment period**

None of these are clinical disputes. All are what happens when a claim is assembled by hand from a paper chart.

Rejection is not the only risk. The same study found facilities left unreimbursed for an entire fiscal year, and a referral hospital declining to renew its contract because the scheme could not settle the previous year's charges [4](#ref-4). A clinic carrying insurance receivables is extending credit whether it planned to or not.

## Two payment workflows, one desk

Insured and cash patients are different processes. Cash is collected at the point of service and reconciles daily. Insurance is claimed after the fact, audited, partially paid, and settled months later. Running both through one undifferentiated register is where reconciliation becomes impossible.

The patient experience differs too: in Hadiya Zone, satisfaction was **65.8% among insured outpatients versus 44.95% among uninsured ones** at the same facilities [6](#ref-6).

Concretely, the evidence supports four controls:

1. **One price list applied by the system**, not recalled by whoever is at the desk.
2. **Charges generated from the clinical record**, so services performed cannot fall off the invoice.
3. **Claim totals computed, not summed by hand** — mathematical errors are a named rejection cause [4](#ref-4).
4. **Duplicate detection within a treatment episode** — also a named rejection cause [4](#ref-4).

## What this note does not claim

No peer-reviewed Ethiopian study we could find quantifies under-billing or charge leakage as a share of revenue in private clinics. The percentages that circulate on this topic ("3–5% of charges are never captured") come from vendor and consultancy marketing in the United States, and are not cited here.

The two out-of-pocket figures are not directly comparable: the 31% figure comes from Ethiopia Health Accounts for an earlier period [2](#ref-2), while 46.3% is the WHO-sourced share of current health expenditure for 2023 [1](#ref-1). We present both rather than choosing, and neither is disaggregated for private clinics specifically.

National CBHI enrolment figures vary widely between sources and reporting years; we have not quoted a single national coverage percentage because we could not verify one from a primary source.

References

Every figure above links to one of these. DOIs resolve to the publisher of record.

  1. [1]World Health Organization Global Health Expenditure Database (via World Bank Open Data) (2025). Out-of-pocket expenditure (% of current health expenditure) — Ethiopia, indicator SH.XPD.OOPC.CH.ZS. World Bank Open Data, series retrieved July 2026. data.worldbank.org/indicator/SH.XPD.OOPC.CH.ZS?locations=ET
  2. [2]Tadiwos YB, Kassahun MM, Mebratie AD (2025). Catastrophic and impoverishing out-of-pocket health expenditure in Ethiopia: evidence from the Ethiopia socioeconomic survey. Health Economics Review 15:15. doi.org/10.1186/s13561-025-00602-1
  3. [3]Alebachew A, Abdella E, Abera S, et al. (2023). Costs and resource needs for primary health care in Ethiopia: evidence to inform planning and budgeting for universal health coverage. Frontiers in Public Health 11:1242314. doi.org/10.3389/fpubh.2023.1242314
  4. [4]Hussien M, Azage M, Bayou NB (2022). Financial viability of a community-based health insurance scheme in two districts of northeast Ethiopia: a mixed methods study. BMC Health Services Research 22:1072. doi.org/10.1186/s12913-022-08439-8
  5. [5]Koricho M, Zerayacob T, Abebe F, et al. (2024). An Assessment of Provider Payment Mechanisms (PPMs) in Ethiopia: Implications for Redesign of PPMs and Progress Toward Universal Health Coverage. Health Systems & Reform 10(1):2377620. doi.org/10.1080/23288604.2024.2377620
  6. [6]Rad MH, Ejajo T, Elilo LT, et al. (2024). Disparities in satisfaction among insured and uninsured adult outpatient department service users in Southern Ethiopia, 2022: a comparative cross-sectional study. BMC Health Services Research 24:807. doi.org/10.1186/s12913-024-11176-9

Measure this in your own clinic

National averages are a starting point, not a diagnosis. Bloom Medicine records what happens in the consultation room once, then reuses it for the invoice, the stock count and the chart — so your own numbers become countable.

More research

View all