Type of study: Case control study

The finding: Olanzapine and quetiapine were associated with a moderate increased risk for development of gestational diabetes in women who continue use of the respective medication during pregnancy versus those who discontinue before the start of pregnancy. Olanzapine was also associated with a positive dose-response relationship between use and risk for gestational diabetes. There was no difference in risk for gestational diabetes associated with risperidone, ziprasidone, or aripiprazole when comparing continuers versus discontinuers.

Strength and weaknesses: Previous studies have had mixed results regarding the risk for developing gestational diabetes in women taking atypical antipsychotics during pregnancy. Studies often compare pregnant women prescribed atypical antipsychotics entering the perinatal period versus those who are not. This introduces potential confounding by indication as women who are not prescribed atypical antipsychotics may differ from those who are in many ways. Some of these differences (i.e., psychiatric illness, healthy diet, exercise) serve to increase risk of developing gestational diabetes. A major strength of this study is all women included were prescribed an atypical antipsychotic leading up to pregnancy, which allows the comparison of women who discontinued atypical antipsychotic during pregnancy versus those who continued, minimizing the aforementioned confounding variables. Furthermore, pre-existing diabetes was used as an exclusion factor, which was not always the case in previous studies. An additional strength includes use of the Medicaid Analytic eXtract claims-based database which allowed for obtainment of a large enough sample size to compare five of the most commonly prescribed atypical antipsychotics (aripiprazole, ziprasidone, risperidone, quetiapine, olanzapine) regarding risk of developing gestational diabetes. Medicaid finances close to 80% of all antipsychotic prescriptions and nearly 50% of all pregnancies. Lastly, the authors performed extensive exploratory analyses (co-variate, risk-stratified, bias, cumulative dose-response) to control for potential confounders such as demographic data, psychiatric diagnoses, comorbidity, other medication use, history of gestational diabetes, and duration of antipsychotic treatment.

Weaknesses: Obesity is a major risk factor for development of gestational diabetes. Obesity is incompletely captured in claims data captured by databases like the Medicaid Analytic eXtract database. The authors attempted to adjust for this by using the Massachusetts General Hospital National Pregnancy Registry for Atypical Antipsychotics to estimate the prevalence of overweight or obese women. The potential for bias over a range of obesity prevalence differences (0%-25%) between continuers and discontinuers was examined in exploratory analysis. An additional weakness is the possibility of confounding influencing the quetiapine analysis. Quetiapine is associated with increased off-label use in comparison to other atypical antipsychotics. Women prescribed quetiapine off-label (i.e., insomnia, anxiety) may be more likely to discontinue upon learning of pregnancy. The authors note the relationship between continuing quetiapine and developing gestational diabetes is maintained when restricting analysis to women with a diagnosis of a psychiatric disorder, but comment further research is needed to confirm this finding. Furthermore, the quetiapine sample was the largest sample size and was associated with the highest rate of previous gestational diabetes in comparison to the other atypical antipsychotics examined, which may have influenced results. Lastly, it was not possible to determine reasons for discontinuation based on claims-based data, which could indicate additional confounding not examined in exploratory analyses.

Relevance: This is the largest study examining the relationship between continued atypical antipsychotic use during pregnancy and development of gestational diabetes. With the large sample size, it was possible to examine individual antipsychotics to compare individual risk. This information can be used to help guide the risk/benefit conversation between clinician and patient to determine if use of an atypical antipsychotic during pregnancy is appropriate, and also provides additional information regarding which atypical antipsychotic may be most appropriate for the patient on an individual level pertaining to risk of gestational diabetes.