RX+DX= PPH Risk: What prescription dispensing data tells us about antidepressants and risk of postpartum hemorrhage.

 Regular contributor (and brand new PhD!) Walker Karraa shares a new study examining the relationship between antidepressant medication and postpartum hemorrhage.  Walker questions the lead researchers on other factors present during labor and birth that may have as much or more impact on the likelihood of PPH, as the influence of antidepressant medication and inquires if those factors were examined.  Read Walker’s assessment and interview and share in the comments section your thoughts on this research.  How might you respond to students, patients and clients who ask about this potential increased risk of hemorrhage?  – Sharon Muza, Community Manager, Science & Sensibility


© www.revolutionpharmd.com

A recent study, Use of antidepressants near delivery and risk of postpartum hemorrhage: Cohort study of low income women in the United States (Palmsten, Hernándéz-Diaz, Huybrechts, Williams, Michels, Achtyes, Mogun, & Setogouchi, 2013) is described as:

“This study is the first to report an association between exposure to antidepressants at the time of delivery and risk of postpartum hemorrhage in a US population and in a population with a diagnosis of depression” (p. 6). Further inquiry into the study provides ample opportunity to consider the intersection of method, measurement and maternal mental health with regards to the use of antidepressant medication and potential risks.

RX: Prescription-dispensary records

The objective of this epidemiological cohort study was to “determine whether use of serotonin or non-serotonin reuptake inhibitors near to delivery is associated with postpartum hemorrhage” (Palmsten et al., 2013, p. 1). The methods involved analyzing pharmacy dispensing records of 106,000 women, ages 12-55, previously identified through Medicaid Analytic eXtract (MAX) who had live births between the years 2000 and 2007, and had been given a medically coded diagnosis of mood or anxiety disorder as defined by the ICD-9 codes (296.x, 300.x, 309.x, or 311.x). Outcome was identified as women who had received an ICD-9 code for postpartum hemorrhage (666.x), atonic postpartum hemorrhage (666.1x), and only inpatient postpartum hemorrhage (Palmsten, et al., 2013).

DX: Connecting diagnosis to depression

Citing a 2000 Canadian epidemiological study (West, Richter, Melfi, McNutt, Nennstiel & Mauskopf), the authors determined that being given the medical code with one of the aforementioned diagnoses was a reasonable predictive measurement of maternal depression. Palmsten et al., (2013) stated “The positive predictive value for depression with these codes was 77%, indicating that most women in this subcohort likely had depression” (Palmsten, et al., 2013, p. 2).

Measuring exposure to antidepressants was addressed by dividing the women of this large cohort into four groups based on their pharmacy dispensing data: (a) current, or antidepressant dispensing supply that overlapped with the delivery date, (b) recent, or antidepressant dispensing supply on at least 1 day in the 1-30 days before delivery date, (c) past, or antidepressant dispensing supply ending between 1-5 months before delivery date, and (d) a reference group with “no exposure”, or no record of antidepressant dispensing supply in the five months before delivery.

I am very grateful to the study’s lead author, Dr. Kristin Palmsten, and senior author, Dr. Soko Setoguchi for taking the time to unpack the pharmacoepidemiological methodology used in this study, and offer suggestions for how childbirth professionals can address findings in practice.

WK: Can you explain the use of prescription dispensing data with regards to estimating exposure at the time of delivery in lay terms?

KP and SS: In our study, we had information on the date a woman was dispensed an antidepressant prescription, the type of antidepressant received, and the number of days for which the prescription was intended to cover. Using this information, we estimated whether a woman had antidepressants available near the time of delivery. Because women may not have taken antidepressants on the days we assumed, there will be some error in our measurement of the exposure. However, this is the best available measurement for drug exposure in studies with large numbers of women like ours.

WK: How does a prescription dispensing data collection measure blood serum platelet levels of exposure? Given that prescription dispensing data is an epidemiological estimate, what would you suggest is the best language to use when describing estimated, relative, or actual risk of postpartum hemorrhage if a woman is taking an SSRI or a non-SSRI prior to delivery?

KP and SS: The risk for postpartum hemorrhage was 2.8% for women not using antidepressants and it was 3.9% for women who were exposed to antidepressants near the time of delivery. We did not have biologic samples available to assess antidepressant exposure.

Given the breadth of the study and potential implications of assumed causality, I then asked the authors for feedback on their following concluding statements (Palmsten, et al., 2013):

• “Although we cannot rule out residual confounding, our study indicates that there might be
about one excess case of postpartum hemorrhage for every 80 to 100 women using antidepressants near the time of delivery, if we assume causality” (p. 6);
• “Our study suggests that all classes of antidepressants are associated with an increased risk for abnormal bleeding” (p. 6);
• “The absolute increase in risk associated with antidepressant exposure in the month before delivery is small, but women and their physicians should be aware of the potential risks when making treatment decisions near the end of pregnancy” (p. 6).

WK: Based on these statements, how would you recommend childbirth educators respond to women’s concerns regarding the use of SSRI and non SSRI in pregnancy?

KP and SS: Our study found that women who use SSRI or non-SSRI antidepressants near the time of delivery had an increased risk for postpartum hemorrhage. We could not exclude the possibility that other factors associated with antidepressant use might actually have caused postpartum hemorrhage, and it is important to remember that the increase in risk of postpartum hemorrhage among antidepressant users is small. In our study, the risk for postpartum hemorrhage was 2.8% for women not using antidepressants and it was 3.9% for women using antidepressants near the time of delivery. These findings as well as the harmful effects of untreated depression should be considered in decisions regarding antidepressant use during pregnancy.

WK: How do you see the risk of exposure to “all classes of antidepressants” in consideration of the literature demonstrating the adverse effects of untreated perinatal mood and anxiety disorders on fetal development, birth, and postpartum health and wellness of mother and baby?

KP and SS: Practitioners and pregnant women should consider and balance the potential risks of antidepressants and the harmful effects of untreated depression and depressive relapse on maternal and offspring health.

WK: How would you address the impact of the underlying disorder in the assessment of risk? For example, depressed women are more likely to be overweight/obese, which is also associated with hemorrhage. (Blomberg, 2011).

KP and SS: We cannot rule out the possibility that obesity, alcohol use, drug use, or other factors related to maternal depression or the severity of the depression; contribute to the higher risk of postpartum hemorrhage among women who use antidepressants during pregnancy. This uncertainty of our results should be a part of the antidepressant treatment decision by practitioners and pregnant women.

WK: Regarding potential mediators, your study included delivery characteristics of short labor, long labor, forceps or vacuum extraction and induced labors. For induction, was protocol considered? For example, use of Cervadil or not, or the length of time and levels of Pitocin given prior to delivery? How might you look forward to including data like this in future analyses?

KP and SS: We did not have information on the type or duration of induction. Further studies are needed to confirm our results and these would be important factors to consider in future studies.

WK: How do you perceive the relationship between these findings and pain management in labor and delivery?

KP and SS: We did not assess the role of pain management in this study, but pain management and epidural use are important factors to consider in future studies.

WK: Childbirth educators are often interested in the relationship between outcome measures and hospital labor and delivery protocol. Many hospitals have protocols regarding external fetal monitoring (EFM) that requires being in bed, and not eating or drinking in labor. For a woman who is also on an SSRI or non SSRI, how might either or both of these practice protocols confound exposure and risk of postpartum hemorrhage stated in this study? (Particularly because serotonin receptors in the gut involved in metabolizing SSRIs?)

KP and SS: Many factors influence bioavailability of antidepressants and birth outcomes. We did not have information on EFM in our study and we cannot speculate how EFM interact with antidepressants and postpartum hemorrhage.

WK: How might APGAR scores of infants be considered within this discussion?

KP and SS: While we did not have APGAR scores in our database, the impact of maternal mood and anxiety disorders and maternal antidepressant use on infant outcomes is another critical piece to be considered in the balance of antidepressant treatment decisions around the time of pregnancy.

WK: Unfortunately, many medical care providers do not screen for perinatal mood and anxiety disorders in pregnancy, despite validated and available short tools available (such as PH-Q-9 or PHQ-2). In assessing exposure to antidepressant medication and increased risk for postpartum hemorrhage, how do you see your data potentially bridging that gap?

KP and SS: We hope this study and others on antidepressant safety during pregnancy underscore the importance of maternal mood and anxiety disorders on pregnancy outcomes, the complex treatment decisions that women with mood and anxiety disorders face, and the importance of discussing treatment options before, during, and after pregnancy with patients.


The opportunity to create cross disciplinary dialogue connecting reader with research, researcher with reader creates the causes for future collaboration, increased understanding, and growth in the field. Given the findings posited in this study, the scope and limitations of the prescription dispensing epidemiological methods—there is much to learn regarding the issue of antidepressants and postpartum hemorrhage. Pharmacy dispensing records cannot measure the exposure perfectly, as having a prescription does not insure consuming the prescription. As noted by the authors, bioavailability of blood serum was not a resource. Controlling for timing, dosage, frequency, missed doses, or titration cannot be measured through prescription records, yet the authors concluded the records and analyses of the records estimate a likelihood of exposure and conclude risk of increased chance of postpartum hemorrhage.

As increased awareness of maternal mortality brings our understanding of the significance of further research into preventing PPH, critical analysis of the relationship, or lack of relationship, between perinatal mood and anxiety disorders and psychopharmacological treatment must continue to develop. I look forward to seeing the next phase of research that emerges from the work of this team, and thank them for their contribution to the discussion.

Correspondence regarding this research paper may be directed to the lead author, Dr. Kristin Palmsten.


Blomberg, M. (2011). Maternal obesity and risk of postpartum hemorrhage. Obstet Gynecol,118 (3):561-8. doi: 10.1097/AOG.0b013e31822a6c59.

Palmsten, K., Hernándéz-Diaz, S., Huybrechts, K. F., Williams, P. L., Michels, K. B., Achtyes, E. D., Mogun, H. & Setogouchi, S. (2013). Use of antidepressants near delivery and risk of postpartum hemorrhage: Cohort study of low income women in the United States. BMJ, 347:f4877 doi:10.1136/bmj.f4877.

Salkeld, E., Ferris, L. E., & Juulink, D. N. (2008). The risk of postpartum hemorrhage with selective serotonin reuptake inhibitors and other antidepressants. Journal of Clinical Psychopharmacology, 28, 230-234.

West, S.L., Richter, A., Melfi, C.A., McNutt, M., Nennstiel, M.E., & Mauskopf, J. A. (2000). Assessing the Saskathchewan database for outcomes research studies of depression and its treatment. Journal of Clinical Epidemiology, 53, 823-831.

Childbirth Education, Guest Posts, Maternal Mental Health, New Research, Postpartum Depression , , , , , , ,

  1. | #1

    While an interesting and provocative study, one major limitation acknowledged by the authors is that they didn’t have the ability to determine the severity of hemorrhage. The conventional definition of hemorrhage at vaginal birth is 500 mL, the same as a blood donation, and at cesarean delivery, it is 1000 mL, still not a severity that produces symptoms or requires transfusion. The researchers conclude that 1 more woman per 100 (absolute excess risk) currently taking antidepressants experiences hemorrhage compared with unexposed women. Since only a small proportion of women with the ICD-9 diagnosis of hemorrhage would experience severe hemorrhage, the clinical importance of the effect of antidepressants is questionable.

  2. avatar
    | #2

    @Henci, 1 per 100 is HUGE in clinical terms. How many people do you see in a day? Can you imagine if every 100th one just started bleeding out? You would spend a lot of time covered in blood.

    And a minimal blood loss doesn’t mean that the mother wasn’t facing a life-threatening situation, just that someone was able to treat it before the mother lost more blood. Presumably we are not letting someone continue to bleed out just to see whether they are going to lose a lot of blood or a little, right?

  3. | #3

    Hi Henci and Liz,
    Thank you for taking the time to read and comment on this article. I too had questions. It feels like we are all wondering about the validity of measure. I was curious about how data on whether a woman had a prescription dispensed for an antidepressant could measure her risk of having PPH.And as the researchers shared, it is an estimate. Actual exposure of medication cannot be measured by these methods. And confounding co-variables are so difficult to access (given the legal fear surrounding research with pregnant populations) much less control. It is hard for everyone, researchers included. I am so grateful for the attempt by these researchers to try to use the best methods they have to examine this topic. And I know future research will stem from this work.

    Do either of you have any thoughts on how a national birth registry, such as those in Sweden, etc., might benefit measurements of blood loss?

  4. avatar
    | #4

    Almost all research depends on estimates, and all estimates have some level of uncertainty. The fact that there is unlikely to be a perfect correlation between a prescription and blood serum levels is mostly irrelevant (and why epidemiologists use intention-to-treat models). To the extent that the error is very bad, it will reduce the statistical significance of the coefficients, but we still expect the coefficient to be a reasonable estimate of the true relationship. We will only get WRONG coefficients if the error is correlated with the independent variable — if, for example, women who hemorrhage are more likely to fill prescriptions they don’t use. Unless you have such an explanation, concerns about the measure are not a particularly strong critique of the work.

  5. | #5

    Hi Liz,
    I hear your passion! I trust Henci, myself and many if not all of the readers here understand statistics as they apply here. Thanks for the refresher, though!

    I hope that the conversation, the dialogue, the discussion can continue.

  6. | #6

    I’m just saying that the undoubtedly extremely low odds of a woman having severe postpartum bleeding, i.e., bleeding exceeding 1000 mL or requiring blood products, has to be weighed against the risks of a depressed woman discontinuing her medication because her care provider or she is alarmed about the possibility of a postpartum “hemorrhage” that amounts to no more than excessive bleeding that never posed a danger.

  7. avatar
    | #7

    I absolutely agree. We know that untreated depression is a major risk factor for poor perinatal outcomes. What this study cannot answer is whether a statistically-significant difference is actually a clinically-significant difference — i.e., whether the increased risk of PPH actually outweighs the clinical benefit of antidepressants. If you lose 500mL at a vaginal birth, you’re in the same category as someone who needs five units of PRBCs, but clearly you’re not in the same clinical situation. Definitions matter. Without good definitions, the utility of the data are limited.

  1. No trackbacks yet.