Conflicting data in literature and guidelines continue to feed the discussion on when to intervene with induction of labor to reduce the stillbirth risk in cases of intrahepatic cholestasis of pregnancy (ICP).

The major problem is the small numbers of events (stillbirths in an ICP cohort). To date, even cohort studies do not meet the criteria of sufficient numbers, let alone that randomized controlled trials will ever meet this requirement.

Let’s assume we want to document the gain by induction of labor to prevent 1 stillbirth in an ICP cohort of women at different gestational ages: based on incidence numbers of ICP in literature, we can calculate among pregnant women with ICP the number needed to induce (NNI) to prevent 1 stillbirth, taking into account the relative numbers and their confidence intervals (CIs).

Based on incidence figures reported in literature, at 36 weeks the risk of stillbirth in an ICP cohort is 0.068% (95% CI, 0–0.138). The stillbirth risk in a control population of the same gestational age is 0.021% (95% CI, 0.019–0.023).

Considering the wide CI of events in the ICP cohort, we calculate risk reductions for mean, worst, and best scenarios.

In the best scenario, that is the one that induction of labor will prevent most stillbirths, the control event rate (CER) of stillbirth is lowest (in the control group) and the experimental event rate (EER) is highest (in the ICP group):

• 
  

  CER = 0.019

  
  
• 
  

  EER = 0.138

  
  
• 
  

  Absolute risk increase (ARI) = 0.12

  
  
• 
  

  Relative risk increase (RRI) = 636

  
  
• 
  

  NNI = 840

In worst scenario to prove the effect of induction of labor, the number of stillbirths in the control group is highest and the number of stillbirths in the ICP group is lowest:

• 
  

  CER = 0.023

  
  
• 
  

  EER = 0

  
  
• 
  

  ARI = –0.02

  
  
• 
  

  RRI = –100

  
  
• 
  

  Number needed to “expect” instead of NNI = 4347

Mean scenario, based on the mean point estimate:

• 
  

  CER = 0.021

  
  
• 
  

  EER = 0.068

  
  
• 
  

  ARI = 0.05

  
  
• 
  

  RRI = 223

  
  
• 
  

  NNI = 2127

For each gestational week we can calculate this impact of induction on the stillbirth rate.

At 38 weeks, best scenario:

• 
  

  CER = 0.029

  
  
• 
  

  EER = 0.119

  
  
• 
  

  ARI = 0.09

  
  
• 
  

  RRI = 310

  
  
• 
  

  NNI = 1111

  
  
• 
  

  NNI is higher in comparison to 36 weeks because the baseline risk of stillbirth in the control group is rising as well

Worst scenario:

• 
  

  CER = 0.035

  
  
• 
  

  EER = 0

  
  
• 
  

  ARI = –0.04

  
  
• 
  

  RRI = –100

  
  
• 
  

  2857 expecting situations will yield 1 less stillbirth than induction

Mean scenario:

• 
  

  CER = 0.032

  
  
• 
  

  EER = 0.047

  
  
• 
  

  ARI = 0.01

  
  
• 
  

  RRI = 46.8

  
  
• 
  

  NNI = 6666

Since prospective randomized controlled trials that are able to demonstrate a real reduction of stillbirth in the ICP group need patient numbers of 35,737 in each group (control and experimental group), we can assume this ultimate trial of proof of evidence will never be conducted.

Our figures could be used to plan intervention or not and to inform patients about the risks of this intervention or expectation in cases of ICP.

The debate will continue, and only clinical consideration (risk of induction and of prematurity) and women’s choices can guide our future management.