Stopping the SSRI Is Not Removing the Risk

The clinical scene below is a fictional composite created for education. It does not describe an actual patient.

The message arrives on a Monday morning: "I just found out I am pregnant. Should I stop the sertraline today?"

The patient has been well for more than a year. Her record contains a diagnosis of recurrent major depression, but the medication discussion is scattered across several notes. One note says she previously became ill after stopping treatment. Another says only that pregnancy risks were reviewed. There is no summary of how severe the prior episode was, how quickly symptoms returned, or what treatment would replace the medication if she stopped it.

Her obstetric clinician tells her to ask psychiatry. Psychiatry tells her to confirm the plan with obstetrics. Before either appointment occurs, she cuts the dose in half. Three days later she feels dizzy, sleeps poorly, and becomes frightened that the depression is returning.

Those symptoms do not answer the question. They could reflect medication discontinuation, recurrent anxiety or depression, pregnancy, sleep loss, or more than one process. The immediate failure is not that the "wrong" choice was made. It is that a high-stakes treatment change began before anyone defined the competing risks or took ownership of the plan.

Pregnancy makes medication counseling more important. It does not make the decision one-sided.

Continuing an SSRI involves questions about the specific drug, dose, timing, maternal effects, fetal and neonatal outcomes, and the limits of available evidence. Changing or stopping it involves a different set of questions: the patient's likelihood of recurrence, discontinuation symptoms, loss of functioning, access to psychotherapy, speed of psychiatric follow-up, and what happens if symptoms return.

The comparison is therefore not medication risk versus no risk. It is one clinical pathway versus another.

That distinction matters because the phrase "avoid exposure" sounds complete. It is not. A patient remains exposed to the course of the underlying illness, to the consequences of a treatment change, and to the quality or fragmentation of the care system around her.

Much of the evidence on psychiatric medication in pregnancy is observational. That is often necessary, but it creates a recurring interpretation problem.

People who continue medication may differ from people who do not. They may have more severe or recurrent illness, different medical conditions, different substance exposure, or different access to care. If an outcome is more common in the medicated group, the prescription may be one contributor, no contributor, or part of a more complicated causal pathway. The diagnosis and the circumstances around it can travel with the medication in the data.

A 2026 JAMA Psychiatry Special Communication argues that this separation is essential. Its abstract concludes that accumulated evidence suggests little or no SSRI-attributable risk for the most serious outcomes once major depressive disorder and related factors are considered.[1] That is an important synthesis, but it is not a universal safety certificate. It is a Special Communication rather than a randomized trial, and its conclusion is a class-level synthesis rather than a drug-specific guarantee.

Good counseling moves from class-level reassurance or alarm to a more exact question: what is known about this medication, at this dose and stage of pregnancy, for the outcome the patient is worried about?

The American College of Obstetricians and Gynecologists recommends against withholding or discontinuing medication for a mental health condition because of pregnancy or lactation status alone.[2] It labels this a strong recommendation based on low-quality evidence.

The last word matters.

"Status alone" rejects a shortcut. It does not require every patient to remain on every medication. It means pregnancy by itself is not a complete clinical rationale. The decision still has to include the treated disorder, current stability, prior course, treatment response, alternatives, medication-specific evidence, and the patient's priorities.

The evidence grade is useful context. The recommendation rejects a status-only rule, but it does not erase uncertainty or turn a guideline into a drug-specific guarantee.

This is why "continue" and "stop" are both incomplete plans. Either choice needs a reason, a monitoring strategy, and a contingency if the expected course changes.

Stopping a maintenance medication is sometimes described as the absence of treatment. Clinically, it is an intervention. It changes exposure, physiology, symptom protection, and the monitoring burden.

The risk is also not uniform. A 2020 systematic review and meta-analysis did not find a statistically significant pooled increase in relapse across all included populations. In a subgroup analysis limited to populations suggestive of severe or recurrent depression, however, discontinuation was associated with higher relapse risk.[4] Only one study contributed to that subgroup, so the result is better treated as a signal for individual risk formulation than as a transportable estimate. Illness history changes the meaning of the same medication decision.

A frequently cited 2006 prospective cohort found substantially more relapse among selected pregnant patients who discontinued antidepressants than among those who maintained them.[5] The study is relevant, but it was not randomized and its sample had highly recurrent depression. A later JAMA correction supplied financial disclosures omitted from the original publication; it did not correct the reported data or analysis. The study should inform counseling, not be converted into a promise about what will happen to one patient.

MotherToBaby's 2026 depression fact sheet emphasizes the variables that belong in the conversation: symptom severity, previous hospitalizations, how quickly symptoms returned after earlier changes, and how quickly treatment worked when restarted. It also advises patients to make medication changes with their clinicians and notes that gradual dose reduction may be used when discontinuation is chosen.[3]

The practical lesson is not "never stop." It is that stopping without a relapse formulation, replacement plan, or follow-up interval is not lower-risk care. It is undocumented risk transfer.

The diagnosis label is only the beginning. The record should show episode number, severity, suicidality or psychosis when relevant, hospitalization history, functional impairment, recurrence pattern, comorbidity, and any history suggestive of bipolar disorder. A patient with one mild episode and a patient with recurrent severe depression are not entering the same decision.

Name the medication, dose, duration, gestational timing, treatment benefit, adverse effects, and the exact outcomes under discussion. Absolute and background risk are more useful than an isolated relative-risk headline. Uncertainty should be stated, not hidden inside the phrase "risks and benefits discussed."

Past missed doses, tapers, discontinuation symptoms, relapses, and response after restarting treatment can be more clinically relevant than a generic class statement. The timeline matters. So does distinguishing recurrence from short-term discontinuation effects.

Psychotherapy is an evidence-based treatment, but writing "consider therapy" is not access. ACOG recommends psychotherapy as first-line treatment for mild to moderate perinatal depression while also recognizing that access can be limited.[2] Is a clinician available? How soon? At what frequency? What happens while the patient waits? If medication is reduced, the alternative plan and symptom monitoring should begin before protection is removed, not after deterioration.

MotherToBaby recommends coordination between obstetric and mental-health clinicians across pregnancy, delivery, and postpartum.[3] Coordination must be operational. The note should identify the prescriber, the clinician monitoring symptoms, the next contact, the threshold for urgent reassessment, and the postpartum plan. "Follow up as needed" leaves the most important part unspecified.

A defensible note does more than record consent language. It allows another clinician to reconstruct the reasoning.

It should state:

This does not make the evidence perfect. It makes the decision visible.

Continuing an SSRI during pregnancy is not automatically the correct answer. Stopping it is not automatically the safer answer.

The clinically honest question is not, "How do we remove risk?" It is, "Which risks are present on each path, which ones matter most for this patient, and who will monitor what happens next?"

Pregnancy should trigger better counseling, not a status-only rule. A medication change should trigger a treatment plan, not a handoff gap.

1. Wisner KL, Oberlander TF, Osborne LM, Huybrechts KF. Depression and SSRI Treatment During Pregnancy: Prioritizing Maternal Mental Health. JAMA Psychiatry. Published online August 12, 2026. doi:10.1001/jamapsychiatry.2026.2405.

2. American College of Obstetricians and Gynecologists. Treatment and Management of Mental Health Conditions During Pregnancy and Postpartum: ACOG Clinical Practice Guideline No. 5. Obstetrics & Gynecology. 2023;141(6):1262-1288. doi:10.1097/AOG.0000000000005202.

3. MotherToBaby. Depression. Updated April 1, 2026.

4. Bayrampour H, Kapoor A, Bunka M, Ryan D. The Risk of Relapse of Depression During Pregnancy After Discontinuation of Antidepressants: A Systematic Review and Meta-Analysis. Journal of Clinical Psychiatry. 2020;81(4):19r13134. doi:10.4088/JCP.19r13134.

5. Cohen LS, et al. Relapse of Major Depression During Pregnancy in Women Who Maintain or Discontinue Antidepressant Treatment. JAMA. 2006;295(5):499-507. doi:10.1001/jama.295.5.499. A subsequent correction supplied financial disclosures omitted from the original publication.

This article is for educational purposes and is not medical advice. Medication decisions during pregnancy should be individualized with the patient's treating clinicians.