Some relationships in development data are contested. This one mostly isn't. Across the 154 countries with both figures, physicians per 1,000 people and infant mortality correlate at −0.73 — more doctors, fewer infant deaths, and the pattern holds through nearly the whole range.

That is about 53% of the variation in infant mortality tracked by a single workforce number. In this field, that is a strong result.

The scale of the gap

The extremes are hard to absorb. Niger has 0.04 physicians per 1,000 people — roughly one doctor for every 25,000 people. Cuba has 9.54, more than two hundred times as many per head.

Infant mortality runs from South Sudan at 71.9 deaths per 1,000 live births to San Marino at 1.2. A child born in South Sudan is about sixty times more likely to die before their first birthday than one born in San Marino.

The countries at the hard end of both:

• South Sudan — 0.04 physicians per 1,000, 71.9 infant deaths per 1,000
• Niger — 0.04 physicians, 65.6 deaths
• Zimbabwe — 0.14 physicians, 62.4 deaths
• Central African Republic — 0.07 physicians, 58.8 deaths
• Chad — 0.09 physicians, 56.3 deaths
• Sierra Leone — 0.13 physicians, 54.6 deaths

And the other end:

• San Marino — 4.63 physicians, 1.2 deaths
• Belarus — 4.72 physicians, 1.8 deaths
• Norway — 4.97 physicians, 2.0 deaths
• Andorra — 5.07 physicians, 2.4 deaths
• Portugal — 5.85 physicians, 2.7 deaths
• Austria — 5.51 physicians, 2.8 deaths

Why the correlation is strong — and what it hides

Physician density is not really measuring doctors. It is a proxy for whether a country has a functioning health system at all: trained staff, facilities to employ them in, supply chains, referral pathways, and a state capable of paying for the lot. Countries with 0.04 doctors per 1,000 are missing all of it, not just the doctors.

That matters because it changes what the number implies. Air-dropping physicians into Niger would not move infant mortality much on its own, and the interventions that actually drive the number down at the low end are mostly not doctor-delivered: skilled birth attendance, antenatal care, immunisation, clean delivery conditions, and treatment for neonatal infection. Midwives and community health workers do most of that work, and they do not appear in this indicator at all.

Cuba is the case that proves the point in the other direction. It has the highest physician density on earth at 9.54 per 1,000 — well clear of any wealthy European country — and its infant mortality, while genuinely low for its income level, does not lead the world. Doctors are necessary and not sufficient.

The ceiling effect

Look again at the low-mortality list. Portugal has 5.85 physicians per 1,000 and Norway has 4.97, yet Norway's infant mortality is lower. Belarus, at 4.72, records 1.8 — better than Portugal, Andorra and Austria despite fewer doctors per head.

Above roughly 3 physicians per 1,000, the relationship flattens out almost entirely. Everything separating a rate of 2.8 from a rate of 1.2 is happening in neonatal intensive care quality, maternal health before conception, preterm birth rates and how deaths are registered — not in workforce headcount.

This is worth stating plainly because it is where cross-country comparisons get abused most often: the correlation is carried almost entirely by the bottom of the range. It says a great deal about the difference between 0.04 and 3 doctors per 1,000, and close to nothing about the difference between 4 and 6.

A measurement caveat

Infant mortality depends on birth and death registration, and registration is weakest exactly where mortality is highest. Figures for the worst-affected countries are modelled estimates with wide uncertainty bands rather than counted events. The direction of the finding is not in doubt; the precision of individual country values at the top of the table is.

Note too that the two snapshots are a year apart — physician density from 2023, infant mortality from 2024 — which is normal for this kind of comparison and adds a little noise.

For a still stronger relationship built on infrastructure rather than staffing, see sanitation access and maternal mortality at −0.80.

Explore the full maps for physicians per 1,000 people and infant mortality, or try the daily map game. Source: World Bank Open Data, 2023 and 2024.