Interpretation
The maternal mortality rate in women with liver cirrhosis was 0.89% which was an 80-fold increased risk compared to pregnant controls without cirrhosis. The most common cause of maternal mortality was variceal hemorrhage during vaginal delivery, although there were large differences in the reported rates of incidence of variceal hemorrhage.34 The risk of maternal death as result of variceal bleeding may have been overestimated in our review given the decreases of maternal mortality and variceal hemorrhage in recent studies. The decrease of variceal hemorrhage is probably the result of the inclusion in (inter)national clinical guidelines to screen pregnant women with liver cirrhosis in the second trimester for early detection of esophageal varices. 35 In addition, the availability of treatment options like endoscopic variceal band ligation have become widely available during the past decades, which were already used in more recent studies included in this systematic review.4, 18, 20 This treatment is considered safe in pregnancy 36, 37 and lowering the risk of (recurrent) variceal hemorrhage and resulting in lower maternal morbidity and mortality.
The risk of mortality in both pregnant and non-pregnant patients with cirrhosis depends on etiology, severity and presence of complications, as well as the presence of comorbid conditions. 38 The reported overall 30-day mortality rate after an episode of variceal bleeding is 15-20% 8, which is similar with the mortality of 12% after a variceal bleed in pregnant women in our study. The one year mortality of non-pregnant patients with compensated liver cirrhosis is 1.0% and that of patients with compensated liver cirrhosis as well as oesofageal varices 3.4%. These data apply to all cirrhosis patients, but 67% of patients with liver cirrhosis are males older than 50 years. 39 In comparison the maternal mortality rates (<1%) among pregnant cirrhosis patients may be slightly lower, likely owing to the comparatively young age of pregnant women. Due to our study design, we were not able to investigate whether pregnancy is associated with more rapid progression of cirrhosis than can be expected according to age.
The meta-analysis showed significant heterogeneity, mostly attributed to two studies (Flemming et al. and Rasheed et al.) as demonstrated by our sensitivity analysis. The study of Rasheed et al. differs in various aspects from the other included studies. It is, in contrast to most other included studies, a prospective cohort study of patients from a middle-income country, possibly reflecting a lower level of care. This study included all pregnancies including miscarriages, whereas the other studies only described deliveries. Moreover, the etiology of liver cirrhosis, in this study solely viral hepatitis, differs from those in the other studies which included various causes. The study of Flemming et al. did not have major differences on study design, study population or etiology of liver cirrhosis. However, their case definition differed because their analysis evaluated only primiparous women.