The Royal College of Obstetrics and Gynecology does not endorse routine active management of intrahepatic cholestasis of pregnancy (ICP)–affected pregnancies. In contrast, the American College of Obstetricians and Gynecologists supports active management protocols for ICP. To investigate this controversy, we evaluated the evidence supporting ICP as a medical indication for early term delivery and the evolution of active management protocols for ICP. Sixteen articles published between 1986 and 2011 were identified. We created 2 groups based on whether obstetric care included active management. Group 1 comprised 6 uncontrolled reports without active management that were published between 1967 and 1983 that described high perinatal mortality rates that primarily were related to prematurity sequel. This group became the fundamental ‘core’ evidence for ICP-associated stillbirths and by extrapolation justification for active management. Group 2 was comprised of 10 reports in which the authors credited empirically adopted active management with the observed low stillbirth rates in ICP-affected pregnancies. Although the group 1 articles routinely are cited as evidence of ICP-associated stillbirth risk, the 1.2% stillbirth rate (4/331) in this group is similar to the background stillbirth rates of 1.1% (11/1000) and 0.6% (6/1000) in 1967 and 2011, respectively ( P = .062 and P = .0614, respectively). Likewise, the stillbirth rates for articles in group 2 were similar to their respective national stillbirth rate. Nevertheless, group 2 articles have become the evidence-based support for active management. We found no evidence to support the practice of active management for ICP.
Intrahepatic cholestasis of pregnancy (ICP) is a pregnancy-specific liver disorder that is characterized by generalized pruritus with elevated serum bile acids and/or liver aminotransferases. ICP affects up to 2% of pregnancies in Scandinavian countries, 0.7% in the United Kingdom, and <1% of pregnancies in Asia and Australia. In the United States, the prevalence varies from 0.32% in Connecticut to 5.6% in a primarily Latina southern California population. The parturient generally reports itching beginning in the third trimester that resolves spontaneously within months of delivery. This relatively benign maternal condition had been reported to be associated with an increased rate of unexplained term stillbirth that occurs between 37 and 39 weeks of gestation. To avoid these late-term stillbirths, there has been international acceptance of active management of ICP-affected pregnancies with the goal of delivering the infant at <39 weeks’ gestation. However, not all obstetric professionals accept this reported association between ICP and stillbirth or recommend active management for ICP-affected pregnancies.
In 2006, the Royal College of Obstetricians and Gynecologists (RCOG) published Green-top Guideline no. 43 in which they reported finding no evidence to support or refute the practice of active management for pregnancies that are affected by obstetric cholestasis. In 2011, the RGOG issued a follow-up obstetric cholestasis report that indicated that active management of ICP should be replaced by individual management decisions developed during discussions to provide the parturient with evidence concerning the known perinatal risk of early term birth vs the small but unknown risk of ICP-associated term stillbirth. Despite the publication of these RCOG evidence-based reports that questioned the association between ICP and stillbirth, active management remains the standard of care for most pregnancies affected by ICP. The American College of Obstetricians and Gynecologists Committee on Obstetrics Practice and the Society for Maternal Fetal Medicine endorsed active management of ICP-affected pregnancies in 2013. To investigate this North American obstetric controversy, we performed a systematic review of unexplained term stillbirths that were associated with ICP-affected pregnancies. Our objective was to evaluate the evidence that supported ICP as a medical indication for delivery at <39 weeks’ gestation.
Our hypothesis was that there is no difference in unexplained term stillbirth rates for ICP-affected and -unaffected pregnancies.
Methods
No human investigation approval was obtained because this is a review of published literature.
(1) Term stillbirth is defined as intrauterine death at >37 completed weeks’ gestation that occurs before the onset of labor and (2) conduct a systematic literature search of term stillbirths that were identified in ICP-affected pregnancies.
Our search strategy included a search of Medline, PubMed, Google Scholar, and Embase for articles that were published between the years of 1960 to 2013. We used the following search terms: cholestasis, intrahepatic cholestasis of pregnancy or obstetric cholestasis and stillbirth, fetal demise, intrauterine demise, or fetal death. The search was limited to articles that were published in the English language that involved human subjects. All identified articles were reviewed by the 6 authors. An article was excluded if it did not include any perinatal outcome data, was a single case report, or was a review article that added no new clinical data. Bibliographies of all articles that were determined by the authors to be relevant were manually searched for additional related publications.
Statistical analysis
The main outcome of the search was unexplained stillbirth at >37 weeks’ gestation in an ICP-affected pregnancy. We calculated the odds ratio (OR) as an estimate of the ICP-associated risk for unexplained term stillbirth.
Results
Fifty-three articles were retrieved with the use of our search criteria. Thirty-seven articles were eliminated: 1 article was a single case report; 2 articles reviewed only the genetics and epidemiology of ICP; 4 articles reviewed only placental aspects of ICP; 7 articles did not address the question of ICP and stillbirth, and 23 articles were review articles that provided no new data. The remaining relevant 16 articles used delivery data from 1967-2011 ( Table 1 ). Because many published studies had no control groups, we reviewed published national and international health statistic reports to determine the background stillbirth rate for each article reviewed ( Table 2 ).
| Article | Cases | Control subjects | Study bias | ||
|---|---|---|---|---|---|
| n | Unexplained stillbirth at ≥37 wk gestation, n | n | Unexplained stillbirth at ≥37 wk gestation, n | ||
| Roszkowski and Pisarek-Miedzińska, 1968 | 49 | 0 | 0 | Selection | |
| Johnston and Baskett, 1979 | 42 | 0 | 42 | 0 | Selection |
| Friedlaender and Osler, 1967 | 103 | 0 | 0 | Selection | |
| Laatikainen and Ikonen, 1975 | 116 | 1 | 0 | Selection | |
| Reid et al, 1976 | 56 | 1 | 0 | Selection | |
| Qiu et al, 1983 | 22 | 1 | 0 | Selection | |
| Laatikainen and Tulenheimo, 1984 | 117 | 1 | 0 | Selection | |
| Berg et al, 1986 | 100 | 0 | 0 | 0 | Selection |
| Fisk and Storey, 1988 | 83 | 0 | 0 | Selection | |
| Rioseco et al, 1992 | 328 | 4 | 319 | 3 | Selection unmatched control subjects |
| Alsulyman et al, 1996 | 79 | 0 | 79 | 0 | Selection unmatched control subjects |
| Roncaglia et al, 2002 | 206 | 0 | 0 | Selection | |
| Glantz et al, 2004 | 690 | 3 | 0 | Selection | |
| Turunen et al, 2012 | 687 | 8 | 1374 | 10 | Selection, recall unmatched control subjects |
| Pata et al, 2011 | 32 | 0 | 0 | Selection | |
| Rook et al, 2012 | 101 | 0 | 0 | 0 | Selection medical surveillance |
| Article | Study | Control subjects | P value | Review | ||
|---|---|---|---|---|---|---|
| n | Unexplained stillbirth at ≥37 wk gestation, n | n | Unexplained stillbirth at ≥37 wk gestation, n | |||
| Friedlaender and Osler, 1967 | 103 | 3 (1 “mature” stillbirth; no gestational age given) | 0 | Stillbirth rate: Denmark 1950-1962: 18/100 -14/1000 | NS | 2 stillbirths with comorbidities: 1 hydrocephalus; 1 maternal diabetes mellitus |
| Roszkowski and Pisarek-Miedzińska, 1968 | 49 | 3 | 0 | N/A | N/A | 2-yr unmatched retrospective case series |
| Johnston and Baskett, 1979 | 42 | 1 | 42 | 0 | NS | 14-yr matched retrospective case series |
| Laatikainen and Iokonen, 1975 | 116 | 0 (2 explained fetal deaths: at 38 wk gestation, 1 listeria-associated death; at 37 wk gestation, 1 twin-twin transfusion) | 0 | N/A | N/A | Uncontrolled retrospective case series |
| Reid et al, 1976 | 56 | 1 | 0 | Australian stillbirth rate 1976: 107/1000 | N/A | Retrospective case series 10-yr period |
| Qiu et al, 1983 | 22 | 1 | 0 | N/A | N/A | Uncontrolled retrospective case series |
| Laatikainen and Tulenheimo, 1984 | 117 | 1 | 0 | N/A | Uncontrolled retrospective case series | |
| Fisk and Storey, 1988 | 83 | 0 | 0 | N/A | Uncontrolled retrospective case series | |
| Berg et al, 1986 | 100 | 1 (38 wk gestation positive ABO blood type immunization) | 100 | 0 | Retrospective unmatched controlled case series | |
The study articles that were divided into expectant vs active management of ICP-affected pregnancies. The expectant management group was composed of 6 descriptive investigations that were completed before active management became common practice. These articles documented 4 unexplained term stillbirths in 331 ICP-affected pregnancies and became the core reports that linked ICP to unexplained term stillbirths. The active management group was formed by 10 studies of perinatal deaths after the empiric adoption of active management for ICP. This second group became the evidence-based support for active management of ICP as a means of avoiding unexplained term stillbirths. However, we found that the 1.12% (4/331) unexplained term stillbirths in expectantly managed cases not to be statically different from their respective national background stillbirth rate of 1.1% (11/1000) and 0.6% (6/1000) in 1967 and 2011 respectively ( P = .062 and = .061). Using the stillbirth rate in the observed group as the ICP-associated stillbirth rate, the lower stillbirth rate in the second group was attributed to active management. However, we found the stillbirth rates in both experimental groups (expectant and active) to be similar to stillbirth rates in their respective countries. We found no evidence that ICP increases the risk for unexplained term stillbirth.
We then sought to evaluate how active management became standard management for ICP-affected pregnancies.
Early reports found no reported association between ICP and fetal death
In 1968, Roszkowski and Pisarek-Miedzińska published an observational study of 49 pregnant women with jaundice. A subgroup of 3 women who had vomiting, epigastric pain, and fever were admitted with diagnosis of obstructive jaundice caused by acute or chronic cholestasis. This subgroup had 3 stillbirths: 1 stillbirth at 38 weeks’ gestation was caused by a placenta abruption. The remaining stillbirths were preterm that occurred at 25 and 27 weeks’ gestation. These authors did not conclude that obstructive jaundice increases the risk of term fetal death. Nevertheless, this article has been cited as evidence to support active management in >20 publications, most recently in 2000. In 1979, Johnston and Baskett reviewed delivery records between the years of 1963-1976 from Winnipeg, Manitoba, Canada, to perform a case series of 42 deliveries that had been complicated by ICP that were matched to 42 deliveries not affected by ICP. Of the 84 deliveries that were reviewed, the authors identified only 1 fetal death. This 1 stillbirth occurred in an ICP-affected pregnancy that was also complicated by severe preeclampsia. Johnson and Baskett did not conclude that an ICP-affected pregnancy was at increased risk for term stillbirth.
Four articles published from 1960-1970 described perinatal deaths in pregnancies that were complicated by hepatic disease. Although none of these articles contained data to support unexplained term stillbirth that was associated with ICP, they are often referenced incorrectly as evidence for this perinatal complication. Consequently, these articles have become unsubstantiated evidence to justify active management.
In 1967, Friedlaender and Osler reviewed Danish records from 1950-1962 to evaluate perinatal outcome for 103 women with icterus gravidarum. Serum bile acid concentrations were not evaluated for any of the women in this Danish study. In addition, although 99% of the parturients (102/103) had clinical jaundice that was documented by elevated bilirubin levels, only 51% of the women (53/103) had a complaint of pruritus. We suggest that this may indicate that the primary diagnosis in this cohort of pregnant women was hepatitis rather than ICP.
The authors do not provide objective evaluation of fetal maturity, such as birthweight or gestational age. The stillbirths are described only as 5 mature fetal deaths. One of the 5 mature stillbirths had hydrocephalus; another stillbirth was born to a woman who had diabetes mellitus. Therefore, in this retrospective review of ICP-affected pregnancies, there were only 3 unexplained stillbirths. The stillbirth rate of 2.9% (3/103) was similar to the European stillbirth rates in 1950 and 1960 that were reported as 1.5% (15.3/1000) and 1.5% (14.5/1000; P = .224 and .199, respectively; Fisher exact test). In 1975, Laatikainen and Ikonen searched the Helsinki Finland hospital records between 1971 and 1972 to identify 116 pregnancies that were complicated by obstetric hepatosis and matched them with 116 unaffected pregnancies. In the control group, no gestational age was given for the 1 stillbirth (1/116). One neonatal death in the ICP-affected group occurred after a cesarean delivery at 38 weeks’ gestation for late decelerations on the fetal heart tracing. In the ICP-affected group, there was a death of both twins in a 37-week gestation with the comorbid complication of intrauterine transfusion syndrome. The other perinatal deaths in the ICP-affected group occurred at 31 and 32 weeks’ gestation. It is notable that the 32-week stillbirth weighed only 800 g, which suggests that intrauterine growth restriction may have been the cause for this death. We conclude that this 1975 report by Laatikainen and Ikonen is not evidence that ICP is a risk for associated unexplained term stillbirth. Therefore, it is inappropriate to cite this article as supportive evidence for active management for ICP-affected pregnancy.
In 1976, Reid et al published a case series that used 1965-1974 data from the Royal Prince Alfred Hospital in Camperdown, Australia. Fifty-six infants were delivered from ICP-affected pregnancies. One near-term stillbirth occurred at 37 weeks’ gestation. The other intrauterine deaths in this report were preterm at <34 weeks’ gestation and clearly would not have been prevented by active management. The near-term stillbirth rate for this cohort was of 1.8% (1/56) and not significantly different from the Australian stillbirth rate of 1.0%, (7-10/1000; P = .45; Fisher exact test). In 1983, Qui et al performed a retrospective review of the perinatal outcome for 22 gravidas with ICP who delivered in China from 1970-1981. Without a control group, the author reports only 1 stillbirth (1/22) among the ICP-affected pregnancies.
Empiric use of active management
In the 1980s without regard to the background stillbirth rate, empiric use of active management was credited with a decrease in ICP-related stillbirths ( Table 3 ).
| Study | Case | Control subjects | P value | Discussion | ||
|---|---|---|---|---|---|---|
| n | Unexplained stillbirth at >37 wk gestation, n | n | Unexplained stillbirth at >37 wk gestation, n | |||
| Alsulyman et al, 1996 | 79 | 0 | 79 (Stillbirth rate: 6.1/1000 a ) | 0 | NS | Retrospective case control of actively managed intrahepatic cholestasis of pregnancy–affected pregnancies; concluded that the stillbirth rate is low because of active management |
| Rioseco et al, 1994 | 320 | 4 (At 33-38 wk gestation) | 319 | 3 (At 33-38 wk gestation) | NS | Retrospective case control of actively managed intrahepatic cholestasis of pregnancy–affected pregnancies; concluded that the stillbirth rate is low because of active management |
| Boncaglia et al, 2002 | 206 | 0 | N/A (Stillbirth rate during study period decreased from 5.5/1000 to 3.6/1000 a ) | N/A | Prospective case series of actively managed intrahepatic cholestasis of pregnancy–affected pregnancies; concluded that the stillbirth rate is low because of active management | |
| Glantz et al, 2004 | 690 | 3 | N/A (Stillbirth rate: 3.7/1000 a ) | N/A | Retrospective case series of actively managed intrahepatic cholestasis of pregnancy–affected pregnancies; concluded that the stillbirth rate is low because of active management | |
| Turunen et al, 2012 | 687 | 8 | 1374 | 20 | NS | Retrospective case control; concluded that the stillbirth rate is low because of active management |
| Rook et al, 2012 | 101 | 0 | N/A (Stillbirth rate: 6/1000 a ) | 0 | Retrospective case series of actively managed intrahepatic cholestasis of pregnancy–affected pregnancies; concluded that the stillbirth rate is low because of active management | |
a Background population stillbirth rate from public health records.
In 1984, after active management was adopted as standard care, Laatikainen and Tulenheimo published a second case series from the Helsinki University Central Hospital, Helsinki, Finland, using records from 117 deliveries that were complicated by ICP. The only fetal death in this uncontrolled case series occurred at 39 weeks’ gestation. The authors concluded that, when compared with earlier reports of ICP-associated fetal death, their 1984 observation of a 0.85% (1/117) fetal mortality rate supported active management to avoid ICP-related stillbirths. However, the earlier Helsinki ICP reports are unsuitable comparison groups. The 1975 report by Laatikainen and Ikonen did not include any unexplained term stillbirths. The 1964 report by Ikonen was a series of jaundice in late pregnancy without the characteristic pruritus of ICP.
In 1986, Berg et al prospectively collected outcomes for deliveries that occurred between 1971 and 1974 at the University of Lund, Sweden. One hundred ICP-affected deliveries were retrospectively matched with 100 unaffected deliveries. In this controlled study, the only stillbirth, which occurred in the ICP-affected group, was affected by the comorbid condition of ABO blood type immunization.
In 1988, Fisk and Storey, without any controls, performed a retrospective review of 86 ICP-affected pregnancies that were delivered at the Royal Prince Albert Hospital between the years 1975 and 1984. Although the authors documented no term stillbirths, they did identify 2 preterm stillbirths that occurred between 35 and 36 weeks’ gestation.
Although, Fisk and Storey did not report any ICP-associated unexplained term stillbirths, this article continues to be cited as evidence to support ICP-associated stillbirth risk and by extrapolation the practice of active management for ICP-affected pregnancies.
The 1990s worldwide adoption of active management for ICP-affected pregnancies
In support of active management for ICP, the American College of Obstetrics and Gynecology practice bulletin references a 1992 publication by Rioseco et al. Published in 1992, this retrospective case control study of ICP-affected and -unaffected deliveries in Santiago Chilean hospitals between 1988 and 1990. Four stillbirths occurred in the study group; there were 3 stillbirths in the control group. No gestational age is provided for any of the 7 stillbirths. Rather, the maturity of the stillbirths is documented as between 33 and 38 weeks’ gestation. There was no difference in the stillbirth rates, 0.04% (4/1000) and 0.03% (3/1000) for the study and control group, respectively. The authors attribute this lack of difference between the 2 study groups to the empiric adoption of active management. However, it is important to note that 0.03% control and 0.04 study stillbirth rates did not differ from the 1990 Chilean general population stillbirth rate of 0.06% ( P = .50 and = .75, respectively; Fisher exact test). Despite each group having stillbirth rates that were similar to rates in the general population, the authors attributed the similar stillbirth rate for the study groups to the adoption of active management. Rioseco et al justify the adoption of active management by citing several articles. As we have indicated, these citations by Rioseco et al do not provide evidence that ICP is associated with unexplained term stillbirths and therefore should not be used to support the practice of active management for ICP.
In 1996, Alsulyman at al performed a retrospective review to evaluate perinatal outcomes that were associated with active management of ICP-affected pregnancies. Antepartum surveillance records from 1988-1995 were reviewed to identify 79 ICP-affected pregnancies. The study group was matched with a control group of 79 gravidas who had antepartum testing because of increased risk for having a stillbirth because of a previous unexplained fetal death. Given the a priori assumption that ICP is associated with an increased stillbirth risk, the chosen control group was appropriate because a previous stillbirth confers a 2- to 10-fold increased risk of having another intrauterine death.
Alsulyman et al reported 2 preterm stillbirths (2/79) at 36 and 37 weeks’ gestation in the study group, with none observed in the control group (0/79). During the study period, the stillbirth rate in southern California was 0.5% (5/1000). The 0.2% (2/1000) stillbirth rate in the study group is not statistically different from the control group or the regional background stillbirth rate of 0.5% ( P = .09; Fisher exact test). Alsulyman et al acknowledged that this study was not powered to determine the association between ICP and stillbirth risk. Nevertheless, they suggested that their results supported the use of active management for ICP-affected pregnancies.
2000
In 2002, Roncaglia et al reported no stillbirths in 206 ICP-affected pregnancies that delivered between 1989 and 1997 in Monza, Italy. During the study period, the Italian stillbirth rate decreased from 0.5% (5.5/1000) to 0.3% (3.6/1000). Given the background stillbirth risk, the sample size that was used in this study did not provide the authors with adequate power to attribute the lack of stillbirths in their cohort to active management. However, the authors did use the stillbirth rates from Johnston and Baskett, Berg et al, and Fisk and Storey as background controls to support their assertion that no stillbirths occurred in this study because active management was the standard of care for ICP-affected pregnancies.
2004
Glantz et al reviewed outcomes in 690 Swedish women whose pregnancies were complicated by ICP and documented a 0.4% (3/690) stillbirth rate that was not significantly different from the 0.37% background stillbirth rate (3.7/1000; P = .693) in the general population. Without considering the population background stillbirth rate, the authors used as a control an inaccurate stillbirth risk attributed to articles by Laatikainen and Ikonen, Fisk and Storey, Rioseco et al, and Alsulyman et al.
2010-2012
Turunen et al performed a retrospective case control study to evaluate the health of women who had experienced an ICP-affected pregnancy. The authors searched the 1969-1988 discharge and labor records from the Finish Temper University Hospital to identify 687 deliveries from ICP-affected pregnancies. The identified cases were matched to 2 gravidas without ICP, the paturient admitted before and after the index case. The 1.1% (8/687) and 1.5% (20/1374) stillbirth rates for the study and control groups, respectively, were not significantly different ( P = .315; OR, 1.61; 95% confidence interval [CI], 0.063–4.09). However, the study group had a higher frequency of hospital stays that were >10 days in length (OR, 8.41; 95% CI, 6.83–10.36), higher rates of cesarean delivery (OR, 1.47; 95% CI, 1.12–1.92), induced labor (OR, 3.26; 95% CI, 2.68–3.97), preterm delivery <37 weeks’ gestation (OR, 7.02; 95% CI, 5.59–8.82), and neonatal birthweight <2500 g (OR, 1.86; 95% CI, 1.31–264). Because this was a review of women’s health outcomes after delivery, no information about prenatal management was provided. Without providing a description of appropriate care, the authors concluded that, with appropriate obstetrics care, ICP presents only a minor risk for adverse maternal or fetal outcome.
Pata et al reviewed records from 3710 deliveries in Istanbul, Turkey, to identify 32 pregnancies that were complicated by ICP. There was no fetal death in this Turkish cohort. Pata et al suggested that the official policy of early delivery at approximately 38 weeks’ gestation may have avoided any ICP-associated stillbirths. As justification of an active management policy, the authors inappropriately cited reports by Bacq et al, Alsulyman et al, and Heinonen and Kirkinen. Bacq et al characterized biochemical bile acid assessment in gravidas who were affected by ICP. Alsulyman et al, as previously described, is not evidence in support of active management. Heinonen and Kirkinen presented a review of maternal characteristics in ICP-affected pregnancies.
2012
Rook et al reviewed records from a San Francisco, California, cohort of 101 deliveries that resulted from actively managed ICP-affected pregnancies. There were no stillbirths in the study group. This study obviously has limited power to evaluate the increased stillbirth risk that was attributed to ICP because the background stillbirth rate for this northern California population was 0.6% (6/1000). Nevertheless, without objective evidence and in the absence of a control group, Rook et al credited the lack of stillbirth in ICP-affected pregnancies to the implementation of active management protocols. In contrast, the authors gave little attention to the perinatal risk that was associated with active management. This report provides objective evidence that active management is associated with adverse clinical outcomes. Specifically, the authors reported that labor was induced in 87% of affected pregnancies (88/101); the average neonatal gestational age was 37.1 ± 1.2 weeks, and 17% of these neonates (17/101) had respiratory distress. Despite the perinatal complications, Rook et al advocates for active management based on support ascribed to MacDorman et al, Glantz et al, Rioseco et al, Alsulyman et al, and Laatikainen and Tulenheimo. As we have described, these articles do not provide evidence that stillbirths are more likely to occur in ICP-affected pregnancies. It is therefore inappropriate to cite them as support for active management.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
