The finding:
Using Swedish national registries, researchers identified 854 361 Swedish women born between 1973-1990; this sample included all live births from this period with deceased, emigrated, and those with early endometriosis (<14 years of age) diagnosis excluded. To account for genetic factors, they identified a subsample of sisters with the same mother. They reviewed the National Patient Register (NPR) for both endometriosis and psychiatric diagnoses from 1987-2016. Researchers found a statistically significant bi-directional association for endometriosis and a variety of psychiatric disorders (affective psychotic disorders, depressive disorders, anxiety and stress-related disorders, eating disorders, substance dependence, personality disorders, and attention-deficit hyperactivity disorders).  The strongest relationship for a diagnosis of endometriosis followed by a psychiatric diagnosis was with substance dependence, which had an adjusted hazard ratio of 1.93 (CI 1.71-2.18). When women with endometriosis were compared to their unaffected female siblings, the adjusted hazard ratio (HR) ranged from 1.56 (CI 1.29-1.88) for depressive disorders to 1.98 (CI 1.34-2.93) for ADHD. When considering those who had psychiatric diagnoses and were later diagnosed with endometriosis, the adjusted HR ranged from 1.26 (CI 1.11-1.44) for substance dependence to 1.94 (CI 1.84-2.06) for anxiety and stress-related disorders. Interestingly, when women diagnosed with substance dependence or ASD were compared to their sisters without psychiatric disorders, there was no elevated risk for later endometriosis diagnosis.

Strength and weaknesses: 
Strengths include the length of study (29 years following participants longitudinally) and nationwide sample size. Female sibling comparisons helped assess the possibility of confounding genetic variables. Weaknesses include hospital record limitations such as unknown time of symptom/diagnosis onset (as records pertained more to treatment) and the potential to miss subthreshold disease burden. Additionally, the NPR features secondary care rather than primary care (where most of the psychiatric disorders are treated), meaning that only a fraction of those with psychiatric illnesses were captured in this study. The data used also does not include private insurances, though those make up a small percentage of overall care in Sweden.

Relevance:
It has been estimated that up to 10% of the female population has endometriosis, though this number varies depending on study population. This study demonstrates a bi-directional association of endometriosis and multiple psychiatric disorders. Given the frequent comorbidity of the illnesses, it is prudent that CL psychiatrists and obstetricians collaborate and screen women with known endometriosis for psychiatric illness. It is also prudent for the CL psychiatrist to consider the potential for endometriosis treatments (which are often hormonal) to cause or exacerbate psychiatric illness.  

Type of study:
(http://ebm.bmj.com/content/early/2016/06/23/ebmed-2016-110401): Cohort study

The Findings:
The authors performed a random-effects meta-analysis of studies reporting on safety (occurrence of any congenital anomaly) and/or efficacy (occurrence of relapse) of lithium exposure during pregnancy or the postpartum period for women with bipolar disorder.  Studies were required to include a control group (comparing against “general population” or “women with affective disorder”) to allow for an effect size computation.  Secondary measures related to safety included rates of cardiac anomalies, spontaneous abortion, preterm labor, and low birth weight.  The authors analyzed results from a total of 13 comparisons from 8 studies. Results of the meta-analysis revealed that lithium exposure during pregnancy (any trimester) was associated with an increased risk of any congenital anomaly when compared to women with affective disorder (OR 1.75, CI 1.2-2.5) as well as a general population (OR 2.0, CI 1.0-4.0), similar to results found when analysis was restricted to first-trimester exposure.  The findings for cardiac anomalies were less cohesive. Though risk of cardiac anomaly was higher in the lithium-exposed group when exposure occurred in the first trimester (both in comparison to a general and affective disorder specific population, the latter OR being 1.75 with a CI of 1.08-2.8), this significance disappeared when comparing women with any trimester exposure to unexposed women with affective disorder.  A significantly increased risk of spontaneous abortion in lithium-exposed women also disappeared when comparison was made to an affective disorder population. Lithium exposure was not associated with an increased risk of preterm birth or low birth weight.  With regard to prevention of mood episode relapse, lithium was associated with a significantly decreased risk (OR 0.16, CI 0.03-0.9) in two studies.  The authors also included a qualitative analysis of an additional 16 studies that did not fit the meta-analysis requirements.  In this section they noted findings from a study showing that the risk of cardiac malformations seems to triple with dosages > 900mg/day and risks generally appear to be increased with median lithium serum level > 0.64 mEq/L.

Strengths and Weaknesses:
Strengths of this study include the large sample size with resultant robust statistical power and the ability to compare findings against both a general population and an affective disorder population.  The study findings add to a sparse landscape of research that addresses important questions relevant to treatment planning in reproductive age women with bipolar disorder.  While studies included in the safety analysis had “good” quality using standardized scales assessing study quality, those included in the efficacy outcomes analysis were assessed to have a high risk of bias (as noted above this analysis included only two studies).  The follow-up time-frame examining relapse ranged from 4 weeks to 2 years, and the I²calculated at 52.7% demonstrates notable heterogeneity.  This was also an issue for the spontaneous abortion findings, which demonstrated significant study heterogeneity that limits the confidence of any conclusions.  The authors thus reflected on the need for larger controlled and nationwide database studies to confirm the findings.  Additionally, most studies did not include information about potential confounders such as exposure to other teratogenic medications, presence of substance use, or a family history of congenital malformations.  It was not possible to stratify findings to take into account use of alternative mood-stabilizing agents or type of bipolar disorder. 

Relevance:
C-L psychiatrists in inpatient and outpatient settings may be called to weigh in on lithium use during pregnancy or postpartum.  Clinical questions about whether to stop, continue, or resume lithium during this period require a synthesized understanding of the risks and benefits placed in the context of each patient’s unique psychosocial and historical factors.  The results of the study showed that lithium exposure at any time during pregnancy was associated with an increased risk of any congenital anomaly (NNH 33-38) as well as specifically with cardiac anomaly (NNH 71). However, this should be balanced with an estimated NNT of 3 for mood episode relapse prevention.  These findings generally support our understanding that lithium is a medication with substantial potential benefit in bipolar disorder counterbalanced by an increased risk of congenital malformations that, on an absolute level, is still small; the baseline rate of such is considered to be 1-2%, and this study found rates of 4% in women exposed to lithium. The authors urge close attention to other modifiable risk factors, historical course of illness as well as response to lithium, and careful monitoring of dosing and blood levels.  

Type of study:
(http://ebm.bmj.com/content/early/2016/06/23/ebmed-2016-110401): Systematic Review and Meta-Analysis