October 2, 2026 interva.net Editorial

Pakistan Counted 115,600 Households to Learn What Kills Its Children

A national verbal and social autopsy survey reconstructed causes of death for thousands of stillbirths, maternal deaths and child deaths across Pakistan — running four different coding methods in parallel.

verbal autopsyInterVAchild mortalityfield studies

Pakistan registers the fact of a death far more reliably than its cause. Routine civil registration “does not capture causes or social determinants for most of these deaths,” the authors of a new national study write — a blind spot that leaves health planners guessing about what is killing the country’s mothers, newborns and children, and where money might actually change the outcome.

To fill part of that blind spot, a research team led by Aga Khan University, with collaborators at the Institute for International Programs at Johns Hopkins University, carried out verbal and social autopsy interviews across 115,600 households in urban and rural districts of all four Pakistani provinces and three administrative regions. The work is not a single paper but a four-part series in the Journal of Global Health, all published on 11 September 2026: a design-and-methods paper, an analysis of neonatal and under-five deaths, one on stillbirths and maternal deaths, and one on mortality among children and adolescents aged 5–19.

A survey built to be coded more than once

Verbal autopsy is the standard substitute for medical certification where most deaths happen outside hospitals — a structured interview with a family about the symptoms, circumstances and care sought before a death, from which a probable cause is assigned. The WHO has published an instrument for exactly this; software implementations such as InterVA-5, the model developed at Umeå University, and InSilicoVA, distributed through the openVA toolset, turn the questionnaires into coded causes.

What sets the Pakistan survey apart is how many coding routes it ran at once. Causes were assigned by three computerised methods — InterVA-5, InSilicoVA and an Expert Algorithm — and independently by physician-certified verbal autopsy, in which trained paediatric and obstetric physicians reviewed each case. Because every method carries its own assumptions about how symptoms map to causes, running them side by side makes the choice of coder a deliberate part of the study design rather than an afterthought. The series reports the survey’s findings and its analytical framework; it is, as the authors describe it, an effort to document the determinants of mortality, not only its distribution.

Where child deaths concentrate

The under-five analysis drew on 3,880 neonatal and post-neonatal deaths recorded in the three years before the survey. Neonatal mortality came out at 39.8 deaths per 1,000 live births (95% confidence interval 39.3–40.3), rising to 40.8 in rural areas against 37.3 in urban ones. Post-neonatal under-five mortality was 35.9 per 1,000 (35.4–36.4), again higher in the countryside (36.8) than in cities (33.6).

The reported causes are familiar from VA work elsewhere and stubbornly preventable. Preterm birth complications, birth asphyxia and infections dominated among newborns; infectious diseases, diarrhoeal disorders and pneumonia led beyond the neonatal period. These are population-level estimates derived from interviews, not individual clinical diagnoses, and they carry the uncertainty that VA cause assignment always involves.

Two thousand stillbirths, and a pattern that repeats

A second paper examined 2,076 stillbirths and 848 maternal deaths captured in the 2017–2018 study window. It put the national stillbirth rate at 30.5 per 1,000 total births and the maternal mortality ratio at 176.2 per 100,000 live births, both higher in rural districts. Strikingly, the leading causes were broadly the same for stillbirths and for mothers: pregnancy-induced hypertension, obstetric haemorrhage, sepsis and obstructed labour.

The social-autopsy arm tried to explain the routes to those deaths, and traced many of them through delays that will be familiar from maternal mortality research — difficulty recognising that an illness was severe, delaying the decision to seek care, problems with transport, and further delay once a provider was reached. The authors conclude that the rates they observed exceed both Pakistan’s national targets and the Sustainable Development Goal targets for 2030, with the sharpest inequities in rural Balochistan and Khyber Pakhtunkhwa.

The burden shifts with age

Among the 1,322 deaths recorded in children and adolescents aged 5–19, the age distribution tilted toward the oldest group: 45.8% were 15–19, 24.4% were 10–14 and 29.8% were 5–9. Boys accounted for 59% of the deaths. The cause mix changed with age in a way that maps onto the global picture — infectious diseases were the largest contributors in younger children, while non-communicable diseases and accidents or injuries grew more prominent among adolescents. Obstetric complications stood out as a contributor to death among female adolescents, a reminder that the boundary between “child health” and “maternal health” is not clean.

What the numbers can and cannot settle

A VASA survey is not a death certificate, and its outputs are estimates with confidence intervals rather than verdicts on individual lives. But a nationally sampled, multi-method survey of this scale produces something the routine system cannot: a cause-of-death profile disaggregated by age, sex and province, and a companion account of the care-seeking failures around each death. The authors argue that the design points toward integrating verbal and social autopsy into national surveys and building the capacity of community health workers to conduct it — a proposal, reported here as the study’s own conclusion, that raises the same questions of cost, training and data quality that follow VA wherever it is scaled up.

interva.net is an independent digest of verbal autopsy and cause-of-death research. It is not affiliated with, endorsed by, or speaking for the InterVA project, Umeå University, or the World Health Organization.

Sources

← Back to the news digest