Twelve maternal deaths, a small budget, and a provincial demand for measurable targets
You are asked to chair the district maternal death review committee. Last year the district recorded 12 maternal deaths for 18,000 live births. Five of the 12 were residents of the estate sector, which forms 22% of the district population. Verbal autopsy suggests that three of the estate deaths occurred at home or on the way to hospital; two followed delays of more than two hours in arranging transport at night; one woman had four antenatal visits at the estate dispensary but no haemoglobin estimation. Two of the 12 deaths were initially coded as cardiac deaths and were reclassified as maternal only after the committee reviewed them. Postpartum haemorrhage accounts for four deaths, sepsis three, hypertensive disorders two, and the remainder other causes. The committee has one meeting and a small budget, and the provincial director wants an action plan with measurable targets.
1. Interpret these figures. How valid is the maternal mortality ratio, and what does a delays-based reading of the estate deaths tell you?
Read the worked answer
The ratio. 12/18,000 × 100,000 = 66.7 per 100,000 live births. Two caveats before using it: the numerator is small, so the ratio is unstable (a single additional death moves it by 5.6), and the numerator is probably an undercount - the two cardiac deaths were captured only because the committee re-reviewed them, which implies other deaths, including early-pregnancy and indirect causes, are being missed. Denominators from estate populations are also uncertain. So 66.7 is a minimum estimate, and year-to-year comparison is unreliable; a three-year rolling figure is better.
Equity signal. Five of 12 deaths (41.7%) came from 22% of the population. That is roughly 1.9 times the share expected on population alone, and it is the strongest single finding in the dataset.
Delays reading. Three deaths at home or en route, and two with transport delays over two hours at night, point to the second delay - reaching care - as the dominant, addressable failure in the estate sector. But delay is usually layered: the woman or family may also have delayed the decision to seek care (first delay), and the facility may have delayed adequate treatment (third delay) after arrival. The woman with four antenatal visits and no haemoglobin is a third-delay/quality-of-care failure in the antenatal period, not a transport failure. Haemorrhage and sepsis, the leading causes here, are precisely the causes that respond to prompt uterotonics, antibiotics and blood - so the delays are causally connected to the cause distribution.
2. Prioritise interventions for the estate sector and attach measurable targets to each.
Read the worked answer
Prioritisation should follow the delays that account for the most avoidable deaths, weighted by feasibility within one budget cycle.
Second delay - transport and access (highest priority, 5 of 12 deaths).
- A named 24-hour transport scheme per estate, with a phone tree and a pre-agreed first-contact person; a standby vehicle arrangement with the estate management.
- Maternity waiting accommodation near the referral hospital for women at high risk or far from care.
- Targets: proportion of estates with a documented and tested night transport plan (0% → 100% in 12 months); median call-to-arrival time (audit of a sample of calls, target under 60 minutes); proportion of estate births attended by a skilled attendant (baseline and annual).
Third delay - quality of care (4 haemorrhage deaths, 3 sepsis).
- Active management of the third stage of labour and a PPH drill at every facility; sepsis protocol with early antibiotics; 24-hour availability of blood or a documented referral arrangement.
- Targets: proportion of PPH cases receiving a uterotonic within one hour (target ≥95%); case-fatality among PPH admissions (target zero for the year); proportion of facilities passing a quarterly drill assessment.
First delay and antenatal quality. Haemoglobin at first visit and at 28 weeks for every estate pregnancy; iron and folate supply; danger-sign counselling in the woman's own language.
- Targets: proportion of estate antenatal records with a documented haemoglobin (0% for the index case → ≥90%); proportion of pregnant women with a documented danger-sign discussion.
A single meeting and a small budget mean two or three interventions done properly, not ten done nominally.
3. How would you strengthen the data system so that next year's committee is working with trustworthy numbers?
Read the worked answer
Find the missing deaths. Link birth and death registers, hospital admission and ICU records, and estate dispensary records, and search for deaths of women of reproductive age rather than relying on cause-of-death coding. Review all deaths in women aged 15–49, not just those already labelled maternal - that is how the two cardiac deaths surfaced, and it is the single most useful change.
Standardise the review. A structured tool for every death: dates and times of onset, decision to seek care, arrival at first facility, referral, definitive treatment, and outcome, mapped explicitly onto the three delays. Verbal autopsy for home deaths, with a trained interviewer and a clinician review panel. Classify each death as avoidable, possibly avoidable or not avoidable, with the reason recorded.
Close the loop. A recommendation without an owner and a date is not a recommendation. Each action gets a named responsible officer, a deadline and an indicator. The committee should review the previous year's actions first, and report to the provincial director with the actions completed, not only the deaths counted. Findings should be anonymised and returned to the estates and facilities concerned.
Targets for the system itself: 100% of maternal deaths reviewed within 30 days; 100% of deaths in women aged 15–49 screened for possible maternal causation; proportion of previous-year recommendations with evidence of implementation. A ratio built on a complete numerator and correct denominator is worth more than a small budget spent on another campaign.