The recent publication by Puljic et al on the management of intrahepatic cholestasis of pregnancy (ICP), while stimulating, has various major issues that invalidate the results and conclusions.
The risk measure that the authors introduce has an inbuilt bias: mortality during 1 week for the intervention group is compared to mortality during 2 weeks for the expectant management group. The latter will naturally tend to be higher. While it is true that infants can no longer suffer fetal death, (too) early delivery could still result in higher infant mortality. Yet, although mortality during the first full year postnatal was available in the data, it was not involved in the calculations. As a result of this bias, inducing labor in women with uncomplicated pregnancies would be recommended from week ≥38 according to the authors’ calculations (Table 3, notice that the bottom row seems incorrect, because risk of expectant management does not exceed stillbirth risk). This would be an interesting proposal, but cannot be made based on these calculations.
The results are also driven by seemingly odd absolute numbers. At 36 weeks, precisely when the authors recommend intervention, mortality of infants born to ICP mothers is an order of magnitude lower than for infants born to healthy mothers (Table 2). If true this would be a very interesting insight, but perhaps it is better taken as an indication of potential data limitations. Regardless of correction for potential confounders, such numbers need some explanation.
The evidence on the correlation between bile acid levels and adverse fetal outcomes such as stillbirth is too substantive to be dismissed as “hypothesis” (pg. 667.e1). The authors further insist on significant delay when obtaining bile acid levels, necessitating decision making without laboratory results. Yet the laboratory at my institution assures me that bile acid can be measured within the hour using standard equipment. In any event, because level-dependence of risk directly affects the calculations and conclusions, this issue should have been addressed head-on rather than sidestepped with: “However, numerous retrospective studies and case reports describe the unpredictable nature of antepartum fetal deaths in pregnancies complicated by ICP after 36 weeks’ gestation.” (pg. 667.e1-667.e2). However tragic, stillbirth tends to be fraught with uncertainty. The question is how the risks of stillbirth compare for ICP vs no ICP.
Altogether, the analysis and argumentation are an insufficient basis for the recommendations made.