This is a good example of the fact that the best available evidence changes as new research data become available. The contrasting results of the three trials4-6
conducted in women with previous episodes of postpartum depression may be explained in part by differences in trial methodology and/or the preventive efficacy of different drug classes. The positive result of the earliest trial (of nortriptyline or other antidepressant against placebo) may be an artefact produced because of its inferior methods (non-randomised and non-blind, which allow selection bias; the women feeling themselves to be at greatest risk opting for treatment, for example, and response bias; those using antidepressants perhaps being more greatly influenced by a placebo effect). The more recent trial of nortriptyline is less subject to these biases and found no benefit.
The pilot trial evaluating sertraline, however, reported a positive effect.6
Differing modes of action of the two classes of drugs may explain the difference in preventive efficacy, but this was a very small trial (only 17 patients), and the very high recurrence rate in the placebo group (4/8 = 50%) is hard to reconcile with earlier experience.
While adverse events occurred more frequently among women receiving sertraline, again because of the small numbers in the trial, this could be a chance occurrence. And of course, trial data are not the best source of evidence for rare adverse events, for which huge patient numbers are required. There are minimal data on which to base decisions about the safety of breastfeeding while taking sertraline. Expert Concensus Guidelines7
based on multiple case series by several investigators suggest that sertraline may be used with minimal risk in breastfeeding mothers; however, there are no published controlled long-term evaluations of infants exposed to selective serotonin reuptake inhibitors through breast milk.8
The updated search also located two Cochrane reviews9,10
and a review protocol11
that seemed relevant, although the trials included in these reviews did not necessarily enrol women with previous episodes of postpartum depression. One of the reviews (including only two trials) found that synthetic progestogen increased the risk of postpartum depression and reported that the preventive effect of oestrogens is unknown.9
In another review, the effect of psychosocial and psychological interventions for preventing postpartum depression was evaluated.10
In a subgroup analysis (specified a priori
), the effect of a variety of non-pharmaceutical interventions was evaluated in women defined as ‘high risk’ and the general population: trials selecting participants based on ‘at risk’ criteria had more apparent success in preventing postpartum depression (relative risk [RR] = 0.67, 95% CI = 0.51 to 0.89) than those enrolling women from the general population (RR = 0.87, 95% CI = 0.66 to 1.16). Given that a previous episode of depression (of any type) is one of the strongest risk factors for a future episode,12
these data (which should be interpreted cautiously) suggest a possible alternative to antidepressant treatment for this woman. A review evaluating the effectiveness of antidepressant drugs in the prevention of postpartum depression is currently only in protocol form.11