Maternal and Perinatal Death Review in a Hospital

The first requirement is knowing which deaths to review, and the answer is broader than the obvious. Maternal deaths are identifiable but not always flagged as such, particularly where death occurs after transfer to intensive care under another speciality, some weeks after delivery, or following a pregnancy that did not reach term. Deaths from indirect causes in a pregnant or recently pregnant woman are the ones most often missed entirely.

Build identification into routine processes rather than relying on the maternity unit to notice. A check on the pregnancy status of any death in a woman of reproductive age, applied at the point of certification, catches cases that would otherwise never reach the review. It is a small addition to an existing step and it is the difference between reviewing the deaths you knew about and reviewing all of them.

Define the perinatal scope explicitly using standard definitions, including stillbirths and early neonatal deaths, and state the thresholds you apply. Ambiguity here produces inconsistent inclusion between periods, which makes any trend meaningless, and it is invariably the stillbirths that fall out of scope when the definition is loose. Review near misses too, where a woman survived severe complications, because they carry the same system lessons in far greater numbers.

The most common failure in the meeting itself is that discussion begins before the facts are established, and the group spends its time reconstructing a chronology from memory. Someone should assemble a factual case summary beforehand: a timeline from the records with times, observations, decisions, who was present and what was requested and when, with gaps in the record identified as gaps rather than filled by inference.

That assembly is genuinely difficult and it is where the review's quality is determined. Times matter enormously — the interval between a deterioration being observed and being escalated, between escalation and a senior attending, between a decision and its execution — and those intervals are what the records were never designed to make legible. Extracting them is work, and doing it properly is worth more than the discussion that follows.

Circulate the chronology in advance so the meeting examines it rather than constructs it. Where the record does not support a time, say so; a review that fills a documentation gap with a plausible assumption has manufactured a fact, and the gap itself is frequently the most important finding in the case.

Use a consistent framework rather than open discussion, because open discussion goes wherever the most senior person in the room takes it. A workable structure examines factors across several domains in turn: the woman's own circumstances and any delay in reaching or accessing care, care at the facility itself including assessment, escalation and treatment, and organisational factors such as staffing, availability of blood and equipment, and the fit of protocols to actual conditions.

Ask at each point what would have had to be different, and whether that difference was achievable with what was available at the time. This distinguishes findings that can be acted on from those that cannot, and it protects the review from the twin failures of concluding that nothing could have been done and of concluding that everything should have been.

Classify preventability using a defined scale and record the reasoning, not just the category. The scale matters less than that it is applied consistently and that the reasoning is written down, because the value over time is in comparing cases and seeing patterns. A series of reviews each concluding with an unexplained category is a set of opinions rather than a body of evidence.

Review proceedings need protection if people are to speak honestly, and that protection has to be real and understood. Establish who attends, that discussion is confidential to the process, how records are kept and who may access them, and what is shared outward and in what form. Ambiguity produces guarded participation and guarded participation produces useless reviews.

Support the staff involved, deliberately and separately from the review. Clinicians and midwives involved in a maternal or perinatal death carry it, and being called to a review can compound that considerably. Tell them in advance what the review is and is not, offer support that exists independently of the process, and do not let the first they hear of it be a meeting invitation.

Handle the family's perspective with equal deliberation. What the family was told, when, and by whom is often a contributing factor in its own right, and their account frequently contains information the records do not. Whether and how to involve them requires judgement and local policy, but their experience should not be absent from a review of what happened to them.

Reviews generate recommendations and hospitals accumulate them. The recurring pattern is a body of well-reasoned findings, each assigned to someone, none with a date, and none subsequently checked, so that the same contributing factor appears in a review two years later and nobody notices it is the same one.

Give every action an owner, a date and a definition of what completion looks like, and review outstanding actions at the start of every meeting before any new case is discussed. That single ordering change does more for implementation than any amount of emphasis, because it makes the backlog visible to the same group that created it.

Weight actions by how reliably they work. Changing a system so an error becomes difficult, or making a resource available that was not, is far more effective than issuing a reminder or arranging training, which are the actions most often chosen because they are the easiest to agree. A review whose recommendations are consistently at the weak end is producing activity rather than change, and it is worth auditing your own action list for that pattern specifically.

Individual reviews find individual causes. The larger value comes from reading across them: contributing factors that recur, times of day or days of week that appear disproportionately, a particular transfer or escalation route implicated repeatedly, a resource whose unavailability shows up in unrelated cases. None of that is visible case by case and all of it is actionable.

This requires the reviews to be recorded in a structured, comparable form rather than as narrative minutes. Coding contributing factors against a consistent framework makes aggregation possible; free-text minutes make it a research project nobody undertakes. Keeping that structured record alongside the hospital's other quality and incident data, as a platform such as HealUDoc allows, is what turns a series of meetings into a source of evidence.

Report the aggregate to clinical governance annually alongside your maternal and perinatal indicators, including what was found, what was changed and what remains outstanding. It is uncomfortable reading and it is the strongest evidence a maternity service can offer that it examines its own outcomes seriously — which is, in the end, what the whole apparatus is for.