Discussion
In this study, we found that in utero exposure to BDZs in women with a mental disorder(s) was associated with a significantly lower gestational age 3.2 days, 95% CI -5.8 - -0.53, p = 0.019. We also found that the use of BDZs during pregnancy is associated with prolonged psychiatric care after delivery concerning the mother. Other parameters like birth weightthe presence of congenital malformations, and the APGAR scores, were not different compared to non-exposure of BDZs.
Concerning the lower gestational age, our results are in line with Huitfeldt et al. and Grigoriadis et al. Huitfeldt et al reported in 2020 a reduction of gestational age of 2.1 days ((95% CI -3.3 - -0.9 days) after in utero exposure of BDZs. The recent systematic review and meta-analysis by Grigoridaris et al in 2020 included studies with prospectively collected data concerning a BDZ exposed and non-exposed cohort, regardless of psychiatric disease, and excluded co-exposure with other psychotropic drugs. They found that exposure to BDZs was associated with a reduction of 6.2 in gestational age (MD – 6.2 days weeks (95% CI -11.2 days - 0.01), P = 0.02). These results, together with our results, a significant shorter gestational age, seems not to be of clinical relevance. Other studies lack information about the indication of use, which might have biased the results. Our cohort comprises BDZs exposure with exposure to other psychotropic drugs, also possible/usable in general clinical care. We found that co-administered psychotropic drugs did not contribute to the lower gestational age. An explanation from a clinical perspective could be that women using BDZs at the end of pregnancy, are more prone to inducing birth and so a shorter duration of pregnancy, possibly because of fatigue symptoms in this vulnerable population and the associated risks of post-partum depression.
We found no association of BDZs exposure with birthweight which indicates that BDZs exposure in utero does not alter the intrauterine growth of the fetus. Our results are conflicting with the results ofGrigoriadis et al. in 2020 who found that SGA differed significantly between BDZ exposed and non-exposed cohorts after correction for publication bias (OR 1.38 (95% CI 1.04, 1.85), P = 0.03), however, the indication for use was not given. Wang et al in 2010 found also an increase in the proportion of SGA among women using BDZs.
We found no significant association between BDZs exposure and the risk for minor and major congenital malformations in our study, where 26.5% of the women in our cohort were exposed to BDZs during the first trimester. Our results are in line with a meta-analysis of Eneto et al (2011). They included studies with BDZ exposure in at least the first trimester with an unexposed control group with only live births and found no association with major malformations (OR 1.06 (95% CI 0.91, 1.25)) or oral clefts (OR 1.19 (95% CI 0.34, 4.15)) in cohort studies. However, in case-control studies, they found an association between major malformations (OR 3.01 (95% CI 1.32, 6.84)) or oral clefts (OR 1.79 (95% CI 1.13, 2.82)) and BDZ exposure. The included case-control studies may be hampered by recall bias from exposed neonates and the use of concomitant co-medication. A recent meta-analysis by Grigoriadis et al (2019) included cohort studies with prospectively collected data about major congenital malformations and cardiac malformations in BDZ exposed and unexposed pregnancies with live births. They found no association with congenital (OR 1.08 (985% CI 0.93, 1.25)) and cardiac malformations (OR 1.27 (95% CI 0.98, 1.65)) and BDZ exposure alone. However, the risk of malformations increased in a sub-group of women using BDZ and antidepressants with an OR of 1.40 (95% CI 1.09, 1.80)). It is not clear whether these results were due to combined exposure to BDZ and antidepressants or antidepressants exposure alone, which is associated with cardiac malformations.
We found no differences in major malformations in the exposed group (4.1%) compared to the reference group (6.1%) after exclusion of some risk factors for the development of congenital malformations, such as teratogenic drugs. EUROCAT reported a prevalence of 2.33% of major congenital malformations in live births in the Netherlands between 2011 and 2018, which seems lower compared to our study population. An explanation could be that our cohort consists of women with a high-risk pregnancy and delivery or puerperium in the hospital, which could increase the selection and surveillance for congenital malformations. All women in our study had a mental disorder, which is also a risk factor and related to other risk factors for congenital malformations, such as smoking or drug use.
Previous studies reported that the percentage of minor malformations ranged from 14.7% to 40.7%. We found 7.3% minor malformations in our cohort, however, minor malformations were also not consequently reported, screened, and diagnosed. Minor malformations can be used to detect different syndromes or major congenital malformations, especially related to teratogenics which we did not find. However, these data must be interpreted cautiously because of methodological issues about their variation in definition, diagnosis, and reporting.
Strengths and limitations
Observational studies are, by their nature, limited in their ability to establish causation with certainty. Consequently, our results are not necessarily reflective of a causal relationship.
It could be that the effects we found, were due to the severity of illness, which we didn’t measure. That is, women with more severe illness may have stayed on the drug, and thus, the illness, and not necessarily the drug, may have affected the effects. This limitation is a result of confounding, that is, the BZD association is distorted because other factors that the study groups differ on, such as diagnosis or indication, can also be associated with adverse outcomes. Effects of confounding by indication can explain a portion or all of the associations found.
However, we tried to reduce selection and confounding bias by using a DAG and using potential confounders published in literature.
Our results are limited by relatively small sample size, and so our findings should be interpreted with caution. Unfortunately, we could not register the use of drugs by trimester of pregnancy. Therefore, any influence of drugs during the organogenesis and the prevalence of congenital malformations in our population could not be directly related to the use of drugs, including BDZs.
In the exposed group the use of BDZs was grouped, regardless of indication, dosage, or half-life of individual drugs. These factors were accounted for by introducing the severity of illness (intermittent versus continuous use of BDZs (data not shown); the number of registered mental illnesses; use of other psychotropic drugs), introducing risk factors for mental disease (educational level; presence of support system), however, we did not find differences between both groups. Although residual confounding could be possible.
Further, it could be that we underestimated the number of congenital malformations, as we only considered live births in our cohort. Therefore, miscarriages and planned abortions may have contributed to the number of malformations.
From the literature, we know that there may be an association between BDZ exposure and miscarriages, but this needs further study. and it could be that children with specific malformations were lost to follow-up because of their specific malformations and further treatment in a tertiary clinic.
Finally, the study population is a high-risk population of pregnant women with severe mental disorders who are referred to our hospital. Women using BDZ while pregnant and treated by their general practitioner were not included. This might overestimate the risks we found and limit our study’s generalizability.
One of the strengths of our study is that only women are included with a defined BDZ-exposure for a mental disorder, determined by a manual review of the EPD. We used a rich dataset, including all main possible confounders, to analyze our data, including dosage of BDZ, the indication of use, frequency and duration of use, and excluding occasional use of BDZ at the end of pregnancy during hospitalization. This may prevent misclassification.
Also, our reference group were women with mental disorders not using BDZs. We know that mental disorders themselves can also lead to different adverse neonatal and maternal outcomes. So confounding by indication might therefore be limited in our study.