checkedThe Carlat Psychiatry Podcast2026-11-16last verified 2026-11-16

do 1 in 25 women with untreated postpartum psychosis kill their children?

verdict: misattributed
a real number attached to the wrong thing
“Without it, one in 25 go on to murder their children.”

a host, a psychiatrist said on The Carlat Psychiatry Podcast, 2026-11-16, at 07:01. hear it in context →

the episode: Kraepelin on Clancy Part 3 (at 07:01). we link the show's own page; we do not re-host audio.

first, the part that holds

postpartum psychosis is a psychiatric emergency, and the risk of a mother harming or killing her baby is real. the 4% figure (about 1 in 25) is real too. it appears in the literature and has been repeated in american clinical reviews for years.

what we found when we went and looked

we traced the 1-in-25 figure to a 1985 scottish follow-up of 82 women treated for postpartum illness, which reported a "probable incidence of infanticide" of 4%. a 2017 review in archives of women's mental health, which covers more than 4,000 childbearing psychoses, says the figure comes from a misreading of that study. only 29 of the 82 women had psychotic illness, both filicides (one definite, one probable) were by depressed mothers, and none was by a mother with acute psychosis without depression. the same review puts filicide at 4.5% (1 in 22) in depressive psychoses and below 1% in early-onset postpartum psychoses (9 of 1,115). it also says the risk would be lower still once the mother is under medical care. no study we found measures the rate in untreated postpartum psychosis.

assertion by assertion

assertionverdictwhat the record says
about 1 in 25 (4%) women with postpartum psychosis kill their childrenmisattributedthe 4% comes from a 1985 follow-up of 82 women with mixed postpartum illness, followed for an average of 16 years. most had depression, only 29 had psychotic illness, and the count was one definite and one probable filicide, both by depressed mothers. a 2017 review of more than 4,000 cases calls the move from this sample to postpartum psychosis 'a misreading'. its own estimate is 4.5% for depressive psychoses and under 1% for early-onset postpartum psychoses (9 of 1,115). the review compiles published cases, which may over-represent dramatic outcomes, so its rates are rough too. [1]
the 1-in-25 rate applies specifically to women whose psychosis is untreated ('without it')unverifiablenone of the sources we read gives an infanticide rate for an untreated cohort, and such a cohort would be very hard to study. the 1985 sample that the 4% comes from were women treated for postpartum illness, as described in the 2017 review. that review says the risk 'would be even lower, once the mother had come under medical care', but it gives no number for treated versus untreated cases. a 2023 review links homicide to 'inadequately treating' postpartum psychosis, also without a number. [1][2]
the deaths are murders, meaning deliberate killingoverstatedthe 2017 review says some filicides in postpartum psychosis without overt depression 'appear to be accidental, without intent to kill'. in five of the cases it compiled, there may have been 'no mens rea'. some deaths in the literature are deliberate, especially in depressive psychoses, but not all fit the word murder. [1]
infanticide is a recognized, elevated risk in postpartum psychosis, with published estimates of 1% to 4.5%confirmeda 2021 review in brain sciences says 'infanticide is rare, occurring in between 1 and 4.5% of all cases', citing the 2017 review. a 2023 review in current psychiatry reports gives the same range and calls postpartum psychosis a psychiatric emergency because of the risk of child harm, infanticide and suicide. the 4.5% upper bound is for depressive psychoses, not postpartum psychosis as a whole. [2][3]

both directions

for the speaker: the number has a real origin: a 1985 study reported a probable infanticide incidence of 4%. american clinical reviews repeat 4% for untreated postpartum psychosis. the 2017 review confirms the risk of a mother killing her child is real and finds about 1 in 22 (4.5%) in depressive postpartum psychoses, close to the speaker's figure. other reviews treat postpartum psychosis as an emergency because of the risk of child harm and suicide, which matches the urgency of the speaker's point.

against: the 4% came from a small mixed sample: 82 women, mostly with depression, followed for 16 years, with two filicides (one only 'probable'), both by depressed mothers. a 2017 review of more than 4,000 childbearing psychoses calls applying it to postpartum psychosis a misreading. for early-onset postpartum psychosis, the review's rate is under 1% (9 of 1,115), and some of those deaths appear accidental. the review says the risk is lower once the mother is under care, but the 4% is not a rate for untreated women. the 1985 women had been treated, and no source we read reports a rate for an untreated group. the 2021 and 2023 reviews give a range of 1% to 4.5%, not a single 1 in 25 figure.

what this cannot say

we could not open the 1985 study or its abstract in this session; the publisher and pubmed pages blocked automated access. our account of its sample and findings comes from the 2017 review's description of it. we also could not open the 2009 american review credited with spreading the 4% figure, so we cannot say how it presented the number. the 2017 review's rates come from compiled published cases and the author's own series of 321. that is not a population cohort: denominators are uncertain, and published cases may lean toward severe outcomes. filicide in postpartum psychosis is rare enough that every estimate rests on small counts. no source we found can say what the rate would be with no treatment at all.

sources, each one fetched and read

  1. Brockington I. Suicide and filicide in postpartum psychosis. Archives of Women's Mental Health, 2017;20(1):63-69. doi:10.1007/s00737-016-0675-8 What we read there: 'Recent American papers have given the much higher figure of 4 %... This error can be traced to a misreading of a Scottish study (Davidson and Robertson 1985), in which 82 patients were followed for a mean of 16 years: only 29 were suffering from bipolar, paranoid or schizophrenic psychoses, the rest from depression; one definite and one probable filicide were committed by depressed mothers.' 'Twenty-four cases among 535 depressive psychoses is 1/22 (4.5 %).' 'The frequency in early onset psychoses, including accidental filicides, is 9/1,115 (less than 1 %). The risk would be even lower, once the mother had come under medical care.' Fetched 2026-11-16 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC5237439/
  2. Friedman SH, Reed E, Ross NE. Postpartum psychosis. Current Psychiatry Reports, 2023;25(2):65-72. doi:10.1007/s11920-022-01406-4 What we read there: 'infanticide is rare occurring in between 1 and 4.5% of cases'; postpartum psychosis 'is often considered a psychiatric emergency due to the elevated risk of child harm, infanticide, and suicide'; it lists 'accidents, suicide, homicide, and neglect' as risks of inadequately treating it, without giving rates by treatment status. Fetched 2026-11-16 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC9838449/
  3. Perry A, Gordon-Smith K, Jones L, Jones I. Phenomenology, epidemiology and aetiology of postpartum psychosis: a review. Brain Sciences, 2021;11(1):47. doi:10.3390/brainsci11010047 What we read there: 'infanticide is rare, occurring in between 1 and 4.5% of all cases', citing the 2017 review on suicide and filicide in postpartum psychosis. Fetched 2026-11-16 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC7824357/

speakers are described by role, not named. one sentence is quoted under fair use, with attribution and a link to the original at the timestamp; we do not re-host audio or video. nothing on this page is a reason to start, stop or change a medication. if you have the file behind anything marked unverifiable, the research board is where people can answer. adam