October 22, 2026 interva.net Editorial

One Death in Four, Rated Avertable by an Operation

A verbal autopsy dataset from Butajira, Ethiopia, put 636 deaths in front of three surgeons. A quarter were judged definitely or probably avertable by surgery — and cancers, not injuries, led the list.

verbal autopsycause of deathfield studiesglobal health

On 20 October a group of emergency, critical-care and surgery clinicians published a commentary in the Journal of Health, Population and Nutrition on the World Health Assembly’s Global Strategy and Action Plan for Integrated Emergency, Critical and Operative Care. Their argument concerns sequence rather than ambition: before advanced intensive care and specialised surgery, systems should guarantee foundational care — early recognition of severe illness, basic airway and oxygen support, essential medicines, timely referral.

Six hundred and thirty-six interviews, three surgeons

Any such argument needs a numerator, and the mortality half of it usually comes from households. A study published in BMC Surgery in March supplies one from the Butajira Health and Demographic Surveillance Site in south-central Ethiopia: 833 deaths registered between 2014 and 2019, of which 680 had complete cause-of-death information, and 636 survived duplicate and irreconcilable records to form the analysis set.

Two senior surgical residents assigned probable underlying causes to ICD-10 codes from WHO verbal autopsy procedures — the questionnaire instrument the agency publishes in its current multilingual edition — and then rated each death separately: definitely avertable by surgery, probably, possibly, not avertable, or an operation already performed. Where the two disagreed, a senior surgeon arbitrated. The interview establishes what a family says killed someone; the avertability rating is a second judgment, laid on top by clinicians who know what a scalpel can change.

What a quarter of the deaths looked like

Of the 636 deaths, 167 (26.2%) were rated definitely or probably surgically avertable. A further 6% were possibly avertable, and in 0.8% an appropriate operation had already happened.

The distribution is not what a reader primed for trauma expects. Neoplasms supplied 62 of the 167 (37.1%), roughly three times their 12.3% share of all deaths. Injuries supplied 37 (22.2%), against 10.1% overall; renal disorders 29 (17.4%), against 6.6%.

Among women, cancers were nearly half of the avertable deaths (32, 49.2%), with injuries at 18.5% and perinatal causes at 9.2%. Among men, cancers were 29.4%, renal disorders 28.4% and injuries 24.5%. Every avertable renal death was male, most of them bladder outlet obstruction from benign prostatic hyperplasia — an enlarged prostate, and the operation that relieves it. Among adults, men were 1.62 times as likely as women to die of an avertable condition (95% CI 1.16–2.24).

A judgment, not a measurement

Verbal autopsy yields a cause distribution; it does not yield a verdict on preventability. Nothing in a family’s account says that a death needed an operation. Three clinicians reading cause codes supplied that verdict, and a panel with a different view of what counts as operable could return a different number from the same interviews.

The machine-coded route this site follows most often — the InterVA model and the openVA family of algorithms — has no equivalent output at all: it estimates the probability that a cause belongs to a death, and stops there. Avertability has to be added on top, by people.

The prominence of cancers is a statement about timing as much as about capacity. A tumour that was resectable when it was still a diagnosis arrives at a surveillance interview as a death. The authors list what they believe produces that outcome: no surgical care at the primary level, delays in the decision to seek surgery, thin family and social support, and hospital fees families cannot afford.

What the study does not settle

They compare their figure with a community study in Sierra Leone that reported 25% of deaths avertable, and which used neither verbal autopsy nor surgeons in the assessment. They also mark a divergence from Malawi, where avertable deaths clustered outside health facilities — a pattern that did not appear in Butajira.

The limitations are substantial. Recall and social-desirability bias among bereaved families, and the subjectivity of physicians assigning cause, both bear on the data. Of the 833 registered deaths, 153 had no complete cause information and could not be included, and there were no data on distance to the nearest facility, household income or community health insurance. It is one surveillance site, over six years that ended before the care strategy existed.

Above all, there is no reference standard: no complete autopsy, no clinical adjudication of the deaths themselves. The 26.2% records what a review panel concluded about interviews, not how often an operation would in fact have saved a life. It is a coverage signal carried in a cause-of-death table.

Read beside this week’s commentary, the discomfort is useful. Triage, oxygen, referral and essential operations at the first level where they can be done correspond to what the Butajira panel flagged: obstructed prostates, injuries, perinatal emergencies, cancers that needed an operating list. Neither document shows whether building that package would move the 26.2%. A verbal autopsy can say what a quarter of deaths looked like. It cannot say what would have prevented them.

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