October 16, 2026 interva.net Editorial

After the Interview: Bangladesh's Decade of Death Reviews

A version-1 F1000Research review of Bangladesh's national MPDSR programme, still awaiting peer review, reports 2,927 maternal and 14,024 neonatal death notifications between 2019 and 2024 — with 83% of maternal deaths reviewed and 55% of neonatal deaths, and a cause profile that keeps shifting under the interviews.

maternal mortalityverbal autopsysurveillancedata quality

Much of the verbal autopsy literature stops at the moment of assignment: did the instrument capture the right symptoms, and did the algorithm or the reviewing physician land on the right cause? The harder question sits one step further along — what a health system actually does with the cause once it has it. Bangladesh has been answering that question in public since 2010, and a review of the programme’s first decade has now been posted as a version-1 article in F1000Research, where it is awaiting peer review.

The paper is a desk-based review by Animesh Biswas and colleagues, built on the government’s own MPDSR annual reports, national guidelines, published literature and meeting minutes, and it reads mostly as an implementation history. Its value for anyone working on cause-of-death data lies in the operational detail: thresholds, time windows, and the years in which the system’s own numbers did not keep up with its ambitions. The authors declare no competing interests and no specific funding; publication charges were covered by the Swedish International Development Cooperation Agency through UNFPA Bangladesh.

A pilot district, then a national programme

Maternal and perinatal death review began in 2010 as a pilot in a single district, Thakurgaon, modelled on the World Health Organization’s “Beyond the Numbers” approach, and was folded into government operational plans as a national programme. The paper describes a deliberately two-track design that treats community and facility deaths with equal attention — unusual at the time, since most maternal deaths in Bangladesh occurred at home.

At community level the sequence is specified in the guidelines: a death is notified within 72 hours, a verbal autopsy is conducted between seven and fifteen days afterwards, and a social autopsy follows immediately after the interview and should be complete within 21 days. Facility deaths are reviewed within seven days. The stated coverage rules are 100% verbal autopsy or facility review for maternal deaths and a 10% sample for perinatal deaths, with causes assigned by medical doctors — physician review of the interview, not an automated model assignment.

The article’s own timeline does not quite settle. Its abstract says the programme reached all 64 districts by 2024; the results text says nationwide coverage was in place by 2022. A table of annual figures carries a “districts” column that never rises above 21, which the paper does not explain.

What ten years of notification and review produced

The quantitative core comes from national MPDSR reports covering 2019 to 2024 — a span the paper’s methods table calls four years and its results text calls five, while the table itself prints six annual rows. Those internal inconsistencies are worth naming, because the totals they carry are the numbers most likely to be quoted onward.

Across the period, 2,927 community maternal deaths were notified and 2,420 reviewed, a review rate of 83%. Neonatal deaths were notified in far greater numbers — 14,024 — but reviewed at 7,250, or 55%. The maternal review rate stayed between 71% and 93% each year. Neonatal review was far more volatile: 54% in 2019, falling to 45% in 2020 and 30% in 2023, then recovering to 62% in 2024. In 2023 the notification count rose while reviews dropped, which is the kind of divergence a system’s own monitoring is supposed to catch.

The causes the interviews return

Among the 2,420 reviewed community maternal deaths, postpartum haemorrhage remained the leading assigned cause, followed by eclampsia; obstructed or prolonged labour accounted for 6.1% to 11.7% of deaths annually, and anaemia, sepsis and antepartum haemorrhage together for 15% to 20%. The paper reports the haemorrhage trend as “39% vs 44.9%” across the period, phrasing that leaves the direction of change ambiguous.

For the 7,250 reviewed neonatal deaths, birth asphyxia rose from 35% of assigned causes in 2019 to 47.7% in 2024. Low-birth-weight-related disorders edged down from 24% to 21.9%, and neonatal sepsis fell from 21% in 2020 to 8.2% by 2024. Those are physician-assigned causes from community interviews, and the paper is candid about the limits underneath them: recall bias, logistical difficulty, and interview quality that was “sometimes found not effective to assign causes of deaths”. Ill-defined causes are listed among the remaining constraints, particularly for neonates.

The stillbirth gap

Stillbirth review tools existed from the start, then disappeared. The first national guideline, finalised in 2016, excluded stillbirth review at both community and facility level, cited by the authors as a response to heavy workloads and thin staffing. The tools returned with the guideline revision of 2022 and 2023, and stillbirth review is now back in the community and facility instruments. The paper calls the intervening years a missed opportunity — a pointed admission in a country where many perinatal deaths are never registered anywhere at all.

Response, and the limits of a desk review

The “response” in MPDSR is the part Bangladesh treats as its signature. Social autopsy sessions are held near the household and bring together neighbours, teachers, religious leaders, Union Parishad members, health workers, volunteers, adolescents and pregnant women, working through delays in care-seeking, birth preparedness, recognition of danger signs and referral gaps. Agreed actions are recorded on a semi-structured form and routed through the upazila mechanism into DHIS2, the district health information system that now holds community notifications via a platform called Tracker Capture and facility data via Event Capture. The paper reports social autopsy findings feeding into local transport arrangements and emergency savings schemes; it also reports the friction — stigma around stillbirth and maternal death, time-intensive community meetings, incomplete facility records, and the residue of a blame culture that the non-blaming design was meant to dissolve.

During COVID-19 the routine loop stalled, and the ministry answered with video conferencing: seven districts joined in December 2020, and three conferences reached 14 districts in May 2021. Eight video sessions were later produced from the seven chapters of the national guideline.

What this design cannot show is the size of the effect. There is no counterfactual, no comparison area, and no independent field verification — the review rests on the government’s own reporting. So the paper’s concluding claim that the system has “demonstrated substantial progress in reducing maternal mortality” is a statement of programme history, not of measured impact. For the global frame, the WHO’s own decade report, published in December 2024, reviews MPDSR implementation across countries through surveys and literature and lands on the same bottlenecks: data quality, reporting, review processes, community involvement and the legal environment. The WHO’s latest estimates put roughly 260,000 maternal deaths in 2023, with a global maternal mortality ratio of 197 per 100,000 live births.

Bangladesh’s contribution to that picture is a decade of documented practice — and a set of annual reports in which the gap between notified and reviewed deaths is as informative as the cause distribution itself. A previous entry in this digest traced how one Tanzanian surveillance site measured maternal deaths over three decades of instrument change; this one describes what a national system set out to do once the interview was over.

interva.net is an independent observer: it is not affiliated with the InterVA project, Umeå University or the World Health Organization, and nothing here is advice to any ministry about how to design a death-review system.

Sources

← Back to the news digest