The recent study of postpartum psychiatric illness [Citation1] is unfortunately flawed by the decision by Munk-Olsen et al. to exclude effects of prior pregnancy losses.
Numerous studies (including one by Munk-Olsen) have shown that prior induced abortion or miscarriage increases the risk of postpartum psychiatric disorders [Citation2,Citation3]. Indeed, multiple losses increase risk [Citation3].
Record linkage has also shown elevated rates of primary care following abortion [Citation4]. Prior abortions may therefore explain higher recourse to primary care before delivery for women more prone to postpartum disorders [Citation1].
Properly analyzed, Munk-Olsen’s data would likely confirm that screening for prior pregnancy losses offers a means to identify women who may require more postpartum care. That would be a very actionable finding.
Unfortunately, Munk-Olsen's studies are disturbingly inconsistent in their methodologies. For example, in their primary care study [Citation1], they (a) exclude all women with any history of treatment for mental health disorders prior to childbirth and (b) also show the consultation rate ratio for two years before childbirth through to one year after.
These excellent study design choices reduce confounding issues by limiting subjects to the most psychologically healthy while providing a good objective metric for pre-pregnancy, during pregnancy, and post-pregnancy health.
In their fetal death study [Citation2], Munk-Olsen also (a) excluded women with prior psychiatric contact and (b) used a 12-month period prior to fetal death for their baseline. But then an oddity occurs. Unlike other researchers [Citation3], they decide to define fetal death to exclude abortions. Then, even more oddly, they modify their results by controlling for exposure to induced abortions while at the same time omitting information about how abortions affected mental health.
This erratic treatment of abortion-associated effects is further highlighted by the methodological choices employed in Munk-Olsen’s only two abortion studies [Citation5,Citation6]. In these studies, analyses for women without a prior history of mental disorders are omitted, which is exactly the opposite of what she does elsewhere [Citation1,Citation2].
Also, rather than showing treatment rates for psychological conditions for two years prior to pregnancy outcome [Citation1], or even one year [Citation2], an inconsistent baseline is employed. She uses a nine-month period, covering the entire time women delivering were pregnant but a mix of pre-conception time and pregnant time for those who had abortions. In her response to comments, Munk-Olsen admits this baseline “may not be directly compatible” [Citation6]. But is it not the whole point to find methodological choices that make groups “directly comparable”?
Further confounding is introduced by mixing women who had one or more abortions into the comparison group of women giving birth [Citation5,Citation6], an approach contrary to prior record linkage studies. This cross-adulteration makes it impossible to compare women who abort their first pregnancies with those who deliver their first pregnancies.
Such methodological choices consistently tend to obscure rather than elucidate the associations between abortion and mental health in Danish records.
Concerns over obfuscation are heightened by Munk-Olsen’s refusals to provide any additional data. For example, when a request was made to show the rate of contact for mental health treatments before conceptions and after pregnancies, Munk-Olsen’s asserted contact rates had no informative value [Citation6], a response inconsistent with her primary care methodology [Citation1]. Similarly, when a colleague requested a simple count of the number of women included in her study who had had both abortions and deliveries and the percentage who had had psychiatric contact [Citation5], Munk-Olsen emailed that it would take too much time and effort to calculate.
Despite this pattern of non-responsiveness, I again request Munk-Olsen to re-analyze the data presented here [Citation1] to show segregated effects of miscarriage and abortion history on consultation rates in general, and on mental health consultation rates in particular, both before and after subsequent deliveries.
References
- Munk-Olsen T, Pedersen HS, Laursen TM, Fenger-Grøn M, Vedsted P, Vestergaard M. Use of primary health care prior to a postpartum psychiatric episode. Scand J Prim Health Care. 2015;33(2):127–33.
- Munk-Olsen T, Bech BH, Vestergaard M, Li J, Olsen J, Laursen TM. Psychiatric disorders following fetal death: a population-based cohort study. BMJ Open. 2014. doi:10.1136/bmjopen-2014-005187.
- Giannandrea S a M, Cerulli C, Anson E, Chaudron LH. Increased risk for postpartum psychiatric disorders among women with past pregnancy loss. J Womens Health (Larchmt). 2013;22(9):760–768.
- Østbye T, Wenghofer EF, Woodward CA, Gold G, Craighead J. Health services utilization after induced abortions in Ontario: a comparison between community clinics and hospitals. Am J Med Qual. 16(3):99–106.
- Munk-Olsen T, Laursen TM, Pedersen CB, Lidegaard Ø, Mortensen PB. Induced first-trimester abortion and risk of mental disorder. N Engl J Med. 2011;364(4):332–9.
- Munk-Olsen T, Laursen TM, Pedersen CB, Lidegaard O, Mortensen PB. First-time first-trimester induced abortion and risk of readmission to a psychiatric hospital in women with a history of treated mental disorder. Arch Gen Psychiatry. 2012;69(2):159–65.