Professional Documents
Culture Documents
To cite this article: Christina Annette Herrera, Tracy A. Manuck, Gregory J. Stoddard, Michael
W. Varner, Sean Esplin, Erin A. S. Clark, Robert M. Silver & Alexandra G. Eller (2017): Perinatal
outcomes associated with intrahepatic cholestasis of pregnancy, The Journal of Maternal-Fetal &
Neonatal Medicine, DOI: 10.1080/14767058.2017.1332036
Article views: 3
Download by: [The UC San Diego Library] Date: 08 June 2017, At: 06:23
THE JOURNAL OF MATERNAL-FETAL & NEONATAL MEDICINE, 2017
https://doi.org/10.1080/14767058.2017.1332036
ORIGINAL ARTICLE
CONTACT Christina A. Herrera christina.herrera@hsc.utah.edu University of Utah Department of OBGYN, Division of Maternal Fetal Medicine, 30 N.
Medical Drive, Room 2B200, Salt Lake City, UT 84132, USA
Presentations: This study was presented in part in poster format at the Society of Maternal Fetal Medicine (SMFM) annual Pregnancy Meeting in
February 2016.
ß 2017 Informa UK Limited, trading as Taylor & Francis Group
2 C. A. HERRERA ET AL.
ventilation use, transient tachypnea of the newborn audited (and deemed accurate) by manual physician
(TTN), and stillbirth [8]. review (CAH).
Interventions for preventing adverse perinatal out- The primary outcome was composite neonatal mor-
comes with intrahepatic cholestasis of pregnancy are bidity which included any of the following: hypoxic
of unproven efficacy. Management schemes differ due ischemic encephalopathy (HIE), intraventricular hemor-
to a lack of high-quality evidence to guide providers. rhage (IVH) grade 3 or 4, bronchopulmonary dysplasia
Fetal non-stress tests are often performed in an effort (BPD), necrotizing enterocolitis (NEC), or perinatal
to attenuate the risk of adverse perinatal outcomes death (including stillbirth). These outcomes were
including asphyxia and stillbirth. In general, non-stress determined by ICD-9 coding. Secondary outcomes
tests have a low false-negative rate (less than 1% risk included individual neonatal morbidity and perinatal
for stillbirth within 7 d of a reactive (normal) test) [9]. outcomes according to antenatal testing frequency.
This is appealing given the association between still- Perinatal asphyxia was defined as an arterial cord
birth and cholestasis of pregnancy. However, non- blood gas with pH less than 7.00 and/or a five minute
stress tests have not been shown to reduce stillbirth in Apgar score less than 5 [14]. Delivery GA included all
this population, and fetal death has been documented births (as opposed to only live births). Of note, during
within days of a reactive non-stress test in several the study epoch, there was no institutional standard
cases [8,10–13]. Non-stress tests were introduced as a for delivery timing, but delivery at 37 weeks was cus-
means of predicting fetal asphyxia due to chronic tomary. Additionally, there was no protocol for ante-
processes such as placental insufficiency [9]. Stillbirth natal testing scheme, which was provider dependent.
in the setting of cholestasis of pregnancy is thought For the primary analysis, women were categorized
to be a sudden, unpredictable event [7], yet antenatal into four groups based on maximum serum bile acid
testing is commonly used for surveillance. level: no cholestasis (bile acids <10 lmol/L), mild cho-
We sought to examine perinatal outcomes among lestasis (bile acids 10–39 lmol/L), moderate cholestasis
women with cholestasis according to bile acid levels (bile acids 40–99 lmol/L), and severe cholestasis (bile
and antenatal fetal testing patterns. We hypothesized acids 100 lmol/L). For antenatal testing analysis,
that neonatal morbidity would be positively correlated women were grouped by the frequency of non-stress
with bile acid levels and that antenatal testing would tests, as determined by billing records: twice weekly or
not prevent adverse outcomes. more, weekly, or less than weekly. Women who did
not have testing at one of our hospitals were excluded
from the analysis as they could have had testing at an
Materials and methods
outside clinic (testing records inaccessible to us),
We performed a retrospective cohort study of women which would bias our results. If women had varying
with intrahepatic cholestasis of pregnancy over frequencies of testing during the observation period,
10 years (2005–2015) within the Intermountain they were categorized by the most frequent testing
Healthcare system, a large healthcare system that received.
includes 22 hospitals. Administrative and clinical elec- In order to appropriately account for multiple gesta-
tronic medical record data are maintained in an tions and mothers with more than one pregnancy in
Enterprise Data Warehouse that captures data across our cohort, we used mixed effects models in all analy-
multiple information systems. The Enterprise Data ses (including patient demographics). We used mixed
Warehouse was queried to identify women who deliv- effects linear regression for continuous outcomes,
ered at an Intermountain facility with a diagnosis of mixed effects gamma regression for severely skewed
intrahepatic cholestasis of pregnancy, pruritus, or itch- continuous outcomes, and mixed effects Poisson
ing using ICD9 codes (648.93, 576.8, 698.9). Our study regression for binary outcomes. We performed multi-
was approved by the Intermountain Institutional variable mixed effects regression models for primary
Review Board (IRB#1024616). outcomes to control for potential confounders, begin-
Women were included if they met initial ICD9 cod- ning with all potential confounders with p values less
ing criteria and had serum bile acid testing performed. than .20, and then eliminating them in a backwards
Women were considered to have intrahepatic cholesta- variable selection fashion. We included multiple gesta-
sis of pregnancy with bile acids 10 lmol/L. Women tions in our analysis as this group is at high risk for
with normal bile acids (<10 lmol/L) were categorized cholestasis. In order to account for possible confound-
as controls. Women were excluded if they had a major ing of neonatal outcomes by including multiple gesta-
fetal chromosomal or structural abnormality. In order tions, we controlled for these pregnancies in our
to ensure data accuracy, 5% of medical records were regression models. There was no adjustment for
THE JOURNAL OF MATERNAL-FETAL & NEONATAL MEDICINE 3
multiple comparisons and no imputation was made pregnancy remained associated with the composite
for missing data. All p values were derived from a two- neonatal morbidity (p ¼ .019) and meconium stained
sided comparison. Statistical analyses were performed amniotic fluid (p ¼ .004). However, NICU stay was only
using STATA software (v14.1, College Station, TX). related to delivery GA (p ¼ .065, Table 3c). These find-
ings persisted after controlling for multiple gestations.
Figure 1 shows the incidence of neonatal morbidity
Results
and delivery GA according to bile acid level.
Seven hundred and eight-five pregnancies from 759 Women with severe cholestasis of pregnancy were
mothers including 843 neonates (731 singletons, 50 delivered earlier in gestation (mean 36.5 weeks,
twins, and four triplets) met inclusion criteria. Two p ¼ .002, Table 4). There was no difference in the inci-
hundred and ninty-eight had no laboratory evidence dence of spontaneous preterm birth between groups
of cholestasis of pregnancy, 347 had mild cholestasis (p ¼ .729). Of the 34 neonates (from 22 pregnancies)
of pregnancy, 108 had moderate cholestasis of preg- with composite morbidity, 11 were singletons, 20 were
nancy, and 32 had severe cholestasis of pregnancy. twins, and three were triplets. Among those pregnan-
Maternal demographics and pregnancy characteristics cies with neonatal morbidity, seven had no antenatal
are shown in Table 1. Women with severe intrahepatic testing, 14 had some form of antenatal testing which
cholestasis of pregnancy were more likely to be was normal prior to delivery, and one was delivered
Hispanic (p ¼ .047) and have a higher pre-pregnancy due to a non-reactive non-stress test in the setting of
BMI (p ¼ .042). Women with moderate intrahepatic comorbidities (subject 1 of Table 6).
cholestasis of pregnancy were more likely to have mul- Perinatal outcomes according to antenatal testing
tiple gestations (p ¼ .037). Women without intrahepatic frequency are shown in Table 5. Of patients under-
cholestasis of pregnancy were more likely to have going antenatal testing, 150 women had non-stress
hypertensive disorders (p ¼ .030). tests twice weekly or more, 43 had weekly testing, and
Neonatal outcomes adjusted for GA according to 53 were tested less than once per week. Bile acid lev-
maternal bile acid level are presented in Table 2. els did not differ between groups according to testing
Women with severe cholestasis of pregnancy were frequency (p ¼ .50, Table 5). Women receiving twice
more likely to have the composite neonatal morbidity weekly or more testing were delivered earlier in gesta-
(p ¼ .004) and meconium-stained amniotic fluid tion (mean 36.6 weeks, p ¼ .016), but the difference in
(p ¼ .014), while women with moderate cholestasis of GA was 4 d. Women with antenatal testing less than
pregnancy were more likely to have longer NICU stay once per week were diagnosed at a later GA than
(p ¼ .052). After controlling for potential confounders women with higher testing frequencies (p < .001).
(Tables 3(a) and 3(b)) including Hispanic ethnicity, pre- Four women had abnormal antenatal testing
pregnancy BMI, hypertensive disorder, cesarean deliv- prompting delivery with no evidence of perinatal
ery and delivery GA, severe intrahepatic cholestasis of asphyxia in these neonates. These pregnancies are
Table 1. Maternal demographics and pregnancy characteristics stratified by maximum serum bile acid level.
No ICP, bile acids Mild ICP, bile acids Moderate ICP, bile acids Severe ICP, bile acids
<10 lmol/L 10–39 lmol/L 40–99 lmol/L 100 lmol/L
(n ¼ 298) (n ¼ 347) (n ¼ 108) (n ¼ 32) p Value
Maternal age (years), mean ± SD 27.9 ± 5.7 28.0 ± 5.1 28.1 ± 5.0 29.7 ± 6.2 .496
Nulliparous 128 (43.0) 126 (36.3) 40 (37.0) 8 (25.0) .129
White race 242 (81.8) 266 (77.3) 79 (73.8) 19 (59.4) .530
Hispanic ethnicity 41 (13.9) 57 (16.6) 22 (20.6) 11 (34.4) .047
Multiple gestations
Twins 16 (5.4) 20 (5.8) 12 (11.1) 2 (6.3) .037
Triplets 0 (0) 2 (0.6) 2 (1.9) 0 (0)
Pre-pregnancy BMI, kg/m2, mean ± SD 26.0 ± 6.0 25.2 ± 5.7 24.2 ± 5.6 26.7 ± 6.3 .042
Current smoker 6 (2.0) 11 (3.2) 1 (0.9) 0 (0) .597
Hypertensive disorder 25 (8.4) 10 (2.9) 5 (4.6) 1 (3.1) .030
Diabetes 28 (9.4) 23 (6.6) 4 (3.7) 3 (9.4) .156
Treated with UDCA NA 131 (37.8) 52 (48.1) 14 (43.8) .148
GA at cholestasis diagnosis, mean ± SD NA 34.4 ± 3.7 33.9 ± 3.8 33.2 ± 5.0 .113
Maximum bile acid result, lmol/L, median (IQR) NA 19 (14, 26) 55 (45, 76) 122 (112.5, 159) NA
Maximum AST result (U/L), median (IQR) NA 49 (33, 81) 72 (44, 148) 68 (46, 125) NA
Maximum ALT result (U/L), median (IQR) NA 63 (34, 129) 89 (250, 50) 86.5 (237, 60) NA
Cesarean delivery 89 (29.9) 73 (21.0) 33 (30.6) 7 (21.9) .109
Data are N (%) unless otherwise specified. NA: not applicable. Statistical comparison not of interest, groups differ as defined. IQR: interquartile range
(25th, 75th percentiles). ICP: intrahepatic cholestasis of pregnancy; BMI: body mass index; UDCA: ursodeoxycholic acid; GA: gestational age; AST: aspartate
aminotransferase; ALT: alanine transaminase.
4 C. A. HERRERA ET AL.
Table 3a. Multivariable mixed effects Poisson regression Table 3c. Multivariable mixed effects Poisson regression
model with neonatal composite morbiditya. model with NICU length of staya.
RR (95% CI) p Value RRb (95% CI) p Value
Delivery GA 0.56 (0.45–0.69) <.001 Delivery GA 0.74 (0.72–0.77) <.001
Bile acid levels Bile acid levels
No ICP Referent No ICP Referent
Mild ICP 1.01 (0.30–3.37) .987 Mild ICP 9.8 (8.1–11.5) .261
Moderate ICP 2.03 (0.58–7.13) .272 Moderate ICP 20.9 (15.4–26.2) .065
Severe ICP 5.59 (1.33–23.53) .019 Severe ICP 21.3 (10.9–31.7) .157
Bile acid levels – controlling for multiple gestations Bile acid levels – controlling for multiple gestations
No ICP Referent No ICP Referent
Mild ICP 0.90 (0.30–2.69) .846 Mild ICP 10.3 (9.0–11.7) .249
Moderate ICP 1.38 (0.44–4.30) .577 Moderate ICP 14.4 (11.8–17.1) .075
Severe ICP 5.21 (1.63–16.61) .005 Severe ICP 15.0 (9.8–20.1) .972
a
Potential confounders dropped from the final model (p > .20) included RR: relative risk; GA: gestational age; ICP: intrahepatic cholestasis of
Hispanic ethnicity, pre-pregnancy BMI, hypertensive disorder, and cesar- pregnancy.
ean delivery. a
Potential confounders dropped from the final model (p > .20) included
Hispanic ethnicity, pre-pregnancy BMI, hypertensive disorder, and cesar-
ean delivery.
b
Table 3b. Multivariable mixed effects Poisson regression Reported in days. Counts of days are modeled so RR represents the ratio
of mean days in NICU.
model with meconium-stained amniotic fluida.
RR (95% CI) p Value
Delivery GA 1.24 (1.03–1.50) .026
Bile acid levels
No ICP Referent One stillbirth (subject 1, Table 7) may have been
Mild ICP 1.80 (0.90–3.59) .094 related to cholestasis of pregnancy since it was other-
Moderate ICP 2.04 (0.81–5.11) .129
Severe ICP 4.82 (1.64–14.18) .004 wise unexplained. This patient had a reactive non-
Bile acid levels – controlling for multiple gestations stress test one day prior to the fetal demise. The
No ICP Referent other two had causes of stillbirth unrelated to choles-
Mild ICP 1.79 (0.89–3.58) .100
Moderate ICP 2.03 (0.81–5.09) .132 tasis of pregnancy.
Severe ICP 4.75 (1.60–14.13) .005 The observed sample sizes for the four bile acid
a
Potential confounders dropped from the final model (p > .20) included level (cholestasis) groups were no (n ¼ 314), mild
Hispanic ethnicity, pre-pregnancy BMI, hypertensive disorder, and cesar-
ean delivery. (n ¼ 317), moderate (n ¼ 124), and severe (n ¼ 34).
Relative to the observed incidence for the primary
outcome of neonatal morbidity in the no cholestasis
summarized in Table 6. Two of these cases had a con- group (1.6%), these sample sizes provided 80%
current comorbidity for which testing was already indi- power, using a two-sided alpha 0.05 comparison,
cated, which may have also impacted the decision to to detect a difference between the no cholestasis
deliver. There were three stillbirths in the cohort. group and the following incidences for the three cho-
Details are summarized in Table 7. INCODE classifica- lestasis groups: mild (5.6%), moderate (7.3%), and
tions were used to categorize the cause of death [15]. severe (11.9%).
THE JOURNAL OF MATERNAL-FETAL & NEONATAL MEDICINE 5
Figure 1. Shown are adjusted means and proportions using marginal estimation [25] following the appropriate mixed effects
regression model, controlling for gestational age in composite morbidity, and perinatal asphyxia.
Table 5. Pregnancy and neonatal outcomes stratified by testing frequency among patientsa with intrahepatic cholestasis of
pregnancy.
Twice weekly or more Once weekly Less than weekly
(n ¼ 150) (n ¼ 43) (n ¼ 53) p Value
Delivery GA, mean ± SD 36.6 ± 1.3 37.2 ± 1.4 37.1 ± 1.2 .016
<37 weeks 24 (16.0) 10 (23.3) 13 (24.5) .380
<36 weeks 23 (15.3) 3 (7.0) 4 (7.6) .107
<34 weeks 9 (6.0) 2 (4.7) 1 (1.9) .590
<32 weeks 1 (0.7) 0 (0) 0 (0) NA
<28 weeks 0 (0) 0 (0) 0 (0) NA
Delivery GA singletons, mean ± SD 36.9 ± 1.2 37.5 ± 1.1 37.2 ± 1.0 .013
Composite neonatal morbidity 9 (5.2) 4 (8.2) 1 (1.8) .537
Perinatal asphyxia 4 (2.3) 0 (0) 0 (0) NA
GA at ICP diagnosis, mean ± SD 32.4 ± 3.6 32.4 ± 3.8 34.6 ± 3.6 <.001
Maximum bile acid result (lmol/L), median (IQR) 25.5 (16, 44) 26 (19, 39) 24 (17, 45) .496
Maximum bile acids >40 42 (28.0) 10 (23.3) 15 (28.3) .682
Multiple gestations
Twins 18 (12.0) 6 (14.0) 2 (3.8) .149
Triplets 3 (2.0) 0 (0) 0 (0)
One or more BPPs performed 27 (18.0) 5 (11.6) 5 (9.4) .289
Abnormal NST prompting delivery 2 (1.5) 2 (4.7) 0 (0) >.99
Number of admissions not resulting in delivery, median (IQR) 1 (0,2) 1 (0,2) 1 (0,1) NA
Data are N (%) unless otherwise specified. NA: not applicable. Outcome too rare to fit a model without over-fitting. IQR: interquartile range (25th, 75th
percentiles). ICP: intrahepatic cholestasis of pregnancy; GA: gestational age; NST: non-stress test; BPP: biophysical profile.
a
These data include only those women who underwent testing at one of our clinical sites.
6 C. A. HERRERA ET AL.
Table 6. Summary of intrahepatic cholestasis of pregnancy cases with abnormal testing prompting delivery.
Delivery
Testing gestational Max bile Delivery Arterial cord
Subject frequency NST result agea acids type Apgars gas (pH, BE) Additional maternal comorbidities or history
1 Weekly Non-reactive 30 30 Cesarean 6, 7 7.3, 2.2 Lupus, antiphospholipid syndrome, pre-eclampsia,
history of 2 prior stillbirths
2 Twice weekly Non-reactive 34 16 Vaginal 8, 8 7.3, 3.6 Depression on Quetiapine and Clonazepam
3 Twice weekly Non-reactive 36 24 Vaginal 8, 9 NA None
4 Weekly 1 Late deceleration 39 14 Vaginal 8, 9 NA None
NA: not available; GA: gestational age; NST: non-stress test; BE: base excess.
[16] Stoll BJ, Hansen NI, Bell EF, et al. Neonatal outcomes of [21] Geenes V, Chappell LC, Seed PT, et al. Association of
extremely preterm infants from the NICHD Neonatal severe intrahepatic cholestasis of pregnancy with
Research Network. Pediatrics. 2010;126:443–456. adverse pregnancy outcomes: a prospective popula-
[17] Walsh MC, Szefler S, Davis J, et al. Summary proceed- tion-based case–control study. Hepatology.
ings from the bronchopulmonary dysplasia group. 2014;59:1482–1491.
Pediatrics. 2006;117:S52–S56. [22] Henderson CE, Shah RR, Gottimukkala S, et al. Primum
[18] Wright CJ, Kirpalani H. Targeting inflammation to non nocere: how active management became modus
prevent bronchopulmonary dysplasia: can new operandi for intrahepatic cholestasis of pregnancy.
insights be translated into therapies? Pediatrics. Am J Obstet Gynecol. 2014;211:189–196.
2011;128:111–126. [23] Jain R, Suri V, Chopra S, et al. Obstetric cholestasis:
[19] Fernandez CLA, Fajardo CA, Favareto MV, et al. outcome with active management. J Obstet Gynaecol
Oxygen dependency as equivalent to bronchopulmo- Res. 2013;39:953–959.
nary dysplasia at different altitudes in newborns [24] Glantz A, Marschall H-U, Mattsson L-A. Intrahepatic
1500 g at birth from the SIBEN network. J Perinatol. cholestasis of pregnancy: relationships between bile
2014;34:538–542. acid levels and fetal complication rates. Hepatology.
[20] Chappell LC, Gurung V, Seed PT, et al. Ursodeoxycholic 2004;40:467–474.
acid versus placebo, and early term delivery versus [25] Muller CJ, MacLehose RF. Estimating predicted proba-
expectant management, in women with intrahepatic bilities from logistic regression: different methods cor-
cholestasis of pregnancy: semifactorial randomised respond to different target populations. Int J
clinical trial. BMJ. 2012;344:e3799. Epidemiol. 2014;43:962–970.