August 5, 2026 · 7 min read

In Pregnancy, Stopping the Antidepressant Is the Riskier Decision — Here’s What Every PMHNP Student Needs to Know

An open weekly pill organizer with only the first two compartments filled, laid out on linen beside a glass of water and a eucalyptus sprig

A patient you have been seeing for two years calls the clinic. She is eight weeks pregnant. She has been stable on sertraline 100 mg since her second depressive episode, the one that ended with a week of not getting out of bed and a partner who was scared enough to call her sister. And the first thing she says, before hello, before anything, is: “I already stopped it. I flushed them Tuesday.”

She is not being reckless. She is doing exactly what almost everyone around her — her mother, a message board, sometimes even a clinician — has implied is the responsible thing. And she has just made the single most dangerous decision of her pregnancy.

If you are training to be a PMHNP, this call is coming for you. Perinatal patients do not show up only in specialty reproductive psychiatry clinics. They show up in community mental health, in telepsychiatry panels, in integrated primary care, in the ED. And the reflex you bring to that phone call — the instinct that fires before you have time to reason — is going to determine whether that patient stays well or spends her third trimester in crisis.

New data presented in February should permanently reset that reflex.

The number that should change your instinct

At the Society for Maternal-Fetal Medicine’s 2026 Pregnancy Meeting, researchers from the University of Pennsylvania presented an analysis of 3,983 patients who gave birth in 2023 and 2024. Every patient in the cohort had a diagnosis of depression or anxiety before conception and an active SSRI or SNRI prescription three months before pregnancy. In other words: these were people who were already being treated, and whose treatment was working well enough that they were on it going in.

Here is what happened. Only 36.7% entered pregnancy still filling that prescription. Nearly 18% had no medication fills at all during pregnancy, and 64.6% had a gap of more than 60 days between fills. Discontinuation rates climbed across trimesters — 29.7%, then 31.6%, then 38.6%.

And the patients who stopped had almost twice the risk of a mental health emergency during pregnancy. Not a bad week. Emergencies: suicidality, substance overdose, psychosis. The risk peaked in the first month and again in the ninth.

The design detail that makes this convincing is the one people skip. Before pregnancy, the continuers and the discontinuers looked the same — no difference in outpatient or emergency visits for mental health reasons. These were not two different populations, one sicker than the other. The divergence started when the medication stopped.

Kelly Zafman, MD, MSCR, the maternal-fetal medicine fellow who led the work, put it plainly in comments to Healio: discontinuation “is not a safe option for many patients with an underlying mental health disorder.” She also said the magnitude of the effect surprised her, and she is someone who sees these patients every week.

An empty prenatal exam room at dawn with a fetal doppler on the wall and blinds casting stripes of light

Why “just stop the medication” feels safe when it isn’t

There is a cognitive bias hiding underneath this, and naming it will make you a better clinician for the rest of your career.

Prescribing feels like an action. Stopping feels like a non-action. If a baby is born with a complication and the mother was on sertraline, the medication is the story. If a mother relapses into severe depression, stops eating, stops attending prenatal visits, and delivers early — that is just what happened. Nobody writes the headline “Antidepressant Withdrawal Causes Preterm Birth,” even when the causal chain runs exactly that way.

Clinicians feel this asymmetry too. It is far more comfortable to be blamed for something you didn’t do than for something you did. But the patient does not experience it as symmetrical, and the outcome data certainly don’t.

This is not a new insight. Twenty years ago, Lee Cohen, MD, and colleagues published a prospective study in JAMA following 201 women with a history of major depression through pregnancy. Among those who maintained their antidepressant, 26% relapsed. Among those who discontinued, 68% relapsed. The hazard ratio was 5.0.

Sixty-eight percent. That is not a side effect. That is a coin flip weighted heavily against your patient, during the exact window when she needs functioning executive capacity, appetite, sleep, and the ability to show up for prenatal care.

What the fetal-risk evidence actually shows

The counterweight has to be real, so let’s look at it honestly.

Antidepressants are, as reproductive psychiatrists like to point out, among the most heavily studied drug classes in pregnancy. In The Carlat Hospital Psychiatry Report, Karen Horst, MD — medical director of The Women’s Place at Texas Children’s Hospital — summarized the picture this way: taken as a whole, studies have not found an increased risk of major malformations with first-trimester antidepressant exposure. She points specifically to work led by Krista Huybrechts’s group, including a study of more than 145,000 children published in JAMA Internal Medicine in 2022, which found no increase in autism spectrum disorder, ADHD, learning disorders, or developmental and behavioral disorders after in-utero antidepressant exposure once genetic factors and maternal alcohol and tobacco use were accounted for.

That last clause is the whole ballgame. The older studies that generated the scary headlines frequently failed to control for the illness being treated. Depression itself is associated with preterm birth, low birth weight, smaller head circumference, and lower Apgar scores. If you don’t separate the drug from the disease, you will hand the disease’s damage to the drug. Learn to spot this in a study abstract — confounding by indication is one of the most common reasons a psychiatric drug looks worse than it is.

Dr. Horst is candid about the limits: most of this literature is observational and database-driven, not randomized. But she is unambiguous about where the weight of evidence lands, and she describes something that should trouble you. She routinely receives referrals from patients whose own psychiatrists told them to stop their medication when they learned they were pregnant — patients who have usually already relapsed by the time she meets them.

A macro close-up of a single unbroken strand of spider silk strung with backlit golden dew droplets

The fight you are walking into

You should know that this is contested territory right now, because your patients will have read about it.

In July 2025, an FDA expert panel convened to discuss SSRI use in pregnancy and leaned heavily toward risk, with discussion of stronger warnings and even a boxed warning. The professional societies pushed back hard. The American College of Obstetricians and Gynecologists called the panel “alarmingly unbalanced,” noting that of ten panelists, only one spoke to the role of SSRIs in preventing the consequences of untreated perinatal illness. The American Psychiatric Association wrote to the agency that recent meta-analyses have found no association between prenatal SSRI exposure and overall risk of birth defects. As of this writing, no boxed warning has been finalized.

You do not have to take a political position on this to practice well. What you have to do is understand both sides well enough to hold a real conversation with a frightened patient who has seen a headline. “The FDA is looking at this and experts disagree” is a truthful answer. So is “the largest well-controlled studies we have are reassuring, and we know what happens when people stop.”

And here is the part that gets lost: mental health conditions are the leading contributor to maternal mortality in the United States. Untreated perinatal depression carries risk of suicide, preterm birth, preeclampsia, and low birth weight. The alternative to treatment is not zero risk. It has never been zero risk.

How to actually handle it

A few things worth carrying into clinical:

Screen everyone, and screen correctly. The Edinburgh Postnatal Depression Scale works in both the antenatal and postpartum periods. If you use the PHQ-9, Dr. Horst advises ignoring the sleep and appetite items — pregnancy alters both, and you will generate false positives. Pair it with the GAD-7, because anxiety is a huge share of perinatal presentations.

Have the conversation before conception when you can. The best version of this discussion happens at a routine visit with a patient of reproductive age who is thinking about pregnancy someday — not in a panicked phone call at eight weeks.

Know your agents. Sertraline has the strongest combined profile across pregnancy and breastfeeding. Fluoxetine carries the most extensive reproductive safety literature but is a less ideal choice while nursing given its long half-life. Sertraline and citalopram are more highly protein-bound with shorter half-lives, which limits accumulation in breast milk.

Do not reflexively taper before delivery. Poor neonatal adaptation syndrome — jitteriness, altered tone, feeding and respiratory difficulty — is generally mild and transient, and there is no good evidence that tapering before delivery prevents it. Meanwhile, the risk of depressive recurrence is highest in the early postpartum period, which is precisely when you would be stripping her of coverage.

Document the shared decision. Note that you reviewed the risks of no treatment to both mother and infant, the known risks of medication, and that no psychotropic is FDA-approved for use in pregnancy. Note that the patient understood and chose her course.

Use the free experts. MotherToBaby (mothertobaby.org), run by the Organization of Teratology Information Specialists, gives patients a free real-time consult with a teratology specialist. It is a second opinion that costs nothing and errs toward caution — which paradoxically makes it more persuasive to anxious patients.

Why this belongs in your board prep

Perinatal psychiatry shows up on PMHNP boards, and it shows up in a predictable shape: a stable patient, a positive pregnancy test, four answer choices. The distractor is almost always “discontinue the antidepressant.” The tempting distractor is “taper before delivery.” The correct answer usually involves continuing effective treatment, choosing an agent with good reproductive data, and documenting shared decision-making.

But do not learn it as a test item. Learn it because Dr. Horst is right that there is still a wide gap between general psychiatric knowledge and perinatal psychiatry, and you are about to enter the field with the chance to close a little of it. Somebody’s patient is going to call in a panic at eight weeks. Be the person who knows what to say.

You are building a clinical instinct right now, one topic at a time. That instinct is what separates a provider who reaches for the safest-feeling option from one who reaches for the safest option. Keep going.

Ready to build the kind of clinical reasoning that holds up on boards and in practice? PMHNPReview.org gives you practice questions, pharmacology review, and study tools designed for exactly this — turning scattered facts into the judgment you will actually use. Start today.