A question about a pregnant patient who stopped her antidepressant may ask for a number, not a treatment decision. The board-relevant contrast is about 68% relapse after discontinuation versus 26% with continuation in a classic pregnancy cohort. The common mistake is to treat those figures as this patient’s personal odds—or as proof that stopping was necessarily wrong.
Identify the population behind the percentages
The figures came from a prospective, nonrandomized study of 201 pregnant women with a history of major depression who were taking, or had recently taken, antidepressants. Participants were euthymic for at least three months before their last menstrual period. In the discontinuation group, 44 of 65 women relapsed; in the continuation group, 21 of 82 did.
That makes the answer to a direct recall question straightforward. It also gives you the denominator you need to interpret the answer responsibly.
| Figure or detail | What it tells you | What it does not tell you |
|---|---|---|
| 68% after discontinuation | The observed proportion who relapsed in that study group | The probability that any particular patient will relapse |
| 26% with continuation | The observed proportion who relapsed in that study group | That continued medication eliminates relapse |
| 42-percentage-point difference | The arithmetic contrast between those two group proportions | The amount of relapse caused by stopping medication |
| If the stem says the patient was stable for more than a year before conception | A relevant part of the patient’s course | Whether her depression was single-episode, recurrent, severe, or likely to recur off treatment |
The study was not randomized: participants’ medication decisions were not assigned by chance. Its groups may differ in ways that affect both treatment decisions and relapse risk, so the figures support an association, not a precise causal estimate for an individual. The sample also had substantial recurrence: 44% reported five or more prior episodes. If the stem says a patient was stable for a year, that is relevant, but it does not reveal how many episodes she has had, how severe they were, or whether symptoms returned during earlier medication changes.
Don’t let a recall question become a verdict
A useful board response is to give the two numbers, name their direction, and stop before adding a conclusion the stem does not support. “Discontinuers had higher observed relapse in the cohort” is defensible. “This patient has a 68% chance of relapse” is not.
A later meta-analysis also helps explain why population and study design matter. Across the studies it pooled, discontinuation was not associated with a statistically significant increase in relapse overall; the confidence interval included no difference. In a subgroup of study populations suggestive of severe or recurrent depression, the estimated risk was higher. That does not erase the classic cohort’s result or change what a question explicitly asking for those figures expects. It does warn against treating one pair of percentages as a universal estimate for every patient with a depression history.
If the patient asks whether stopping was the right choice, the numbers alone cannot answer. First establish her baseline and current state: prior episode count and severity, functional impact, any history of suicidality or psychosis, previous relapse after reducing or stopping treatment, response and tolerability, current symptoms, and what support or psychotherapy is available. Do not infer a mild single episode—or a severe recurrent illness—from a year of stability alone.
Pregnancy status by itself is not a reason to withhold or discontinue mental-health medication. That principle does not mean every patient should take every medication unchanged. It means the discussion should weigh her illness course, treatment response, medication-related concerns, preferences, and the risks of untreated or returning depression. If she has already stopped, respond without blame, reassess symptoms and safety, and coordinate an individualized discussion and follow-up with her treatment and obstetric team.
Try a 12-minute retrieval exercise
Retrieval practice means attempting to produce an answer from memory before checking it. For this question format, practise not just recalling the percentages but also retrieving their population and limits. That tests whether you can interpret the figure rather than recognize it after rereading.
- Minutes 1–2, closed book: Write down which group had the higher relapse proportion and the approximate percentages. Don’t look at the explanation.
- Minutes 3–5: Add the study denominator in one sentence: who was studied, when they were euthymic, and whether treatment choice was randomized.
- Minutes 6–9: Answer two different prompts in separate sentences: “What figures does the MCQ ask for?” and “Can I use them as this patient’s individual probability?” Include one reason for your second answer.
- Minutes 10–12: Check your notes. Correct the exact gap—number reversal, missing population, or causal overstatement—then close the notes and repeat the answer once.
If you can recall 68% and 26% but cannot state why they are not a personal forecast, the review is not finished. Write that distinction beside the missed item; it is the reasoning step most likely to travel to the next perinatal question.
Practical takeaways
- For the classic cohort question, recall about 68% after discontinuation versus 26% with continuation.
- Treat these as observed group proportions, not an individual prediction or proof of causation.
- A year of stability does not substitute for the history of episode count, severity, functioning, and prior relapse.
- Use the figures to inform a collaborative, individualized discussion—not to judge a patient’s past decision or automatically direct a medication change.
