Long-term add-on antidepressant use for bipolar depression remains controversial. This study aimed to investigate primarily the association between psychopharmacological treatments and hospitalisation (ie, hospital admission) for bipolar depression, and secondarily the association between psychopharmacological treatments and hospitalisation for bipolar mania and somatic reasons in a registry-based national Swedish cohort. In this within-subject analysis, people diagnosed with bipolar disorder were identified from Swedish nationwide registers of inpatient and specialised outpatient care, sickness absence, and disability pension between Jan 1, 2006, and Dec 31, 2021. Data for hospitalisations, and antidepressant, antipsychotic, and mood stabiliser medication use were also retrieved from national databases. Treatment periods were modelled using the PRE2DUP method. Data were analysed with a within-individual design with stratified Cox Regression models, to eliminate selection bias when calculating adjusted hazard ratios (aHRs) and 95% CIs. The main outcome was hospitalisation due to depression and secondary outcomes were mania-related and somatic hospitalisations to address the risk-benefit ratio of antidepressant treatment. The reference was non-use of antidepressant, antipsychotic, and mood stabiliser medications. We also did head-to-head comparisons (ie, comparing different drug use periods within the same individual against each other) between medications to obtain results on comparative effectiveness while minimising confounding by indication. Ethnicity data were not available. People with related lived experience were involved in the research and writing process. The study cohort included 105 495 individuals (mean age 44·2 years, SD 18·8; 65 607 [62·2%] women and 39 888 [37·8%] men). In medication class-based analyses, a higher risk of depression-related hospitalisation was associated with the use of antidepressant only (aHR 1·25, 95% CI 1·16-1·34), antipsychotic only (1·39, 1·24-1·55), antidepressant-antipsychotic combination (1·28, 1·18-1·39), and antipsychotic-mood stabiliser combination treatment (1·13, 1·03-1·24). By contrast, use of mood stabilisers only (0·89, 0·81-0·98) was associated with lower risk. For specific monotherapies, only lithium was associated with lower depression-related hospitalisation risk (0·75, 0·67-0·85). No specific antidepressant monotherapy was associated with reduced depression-related hospitalisation, while several antidepressants and antipsychotics were related to an increased risk. In head-to-head comparisons, lithium monotherapy was associated with a superior outcome compared with antidepressant monotherapy (0·59, 0·51-0·68), antipsychotic monotherapy (0·54, 0·44-0·66), lamotrigine monotherapy (0·69, 0·53-0·91), and quetiapine monotherapy (0·54, 0·41-0·71). Lithium was associated with the lowest risk of somatic hospitalisation (0·86, 0·80-0·93) when compared with non-use of antidepressants, antipsychotics, and mood stabilisers. Finally, antidepressant-only treatment (1·22, 1·03-1·44) was associated with increased risk of mania-related hospitalisation and other monotherapies and combinations were associated with a lower risk. Since medications are typically started when depressive symptoms re-emerge, all treatments might appear less effective than they actually are when the reference is non-use of medication. Lithium was the only specific monotherapy with significantly reduced risk of depression-related hospitalisations when compared with non-use of antidepressants, antipsychotics, and mood stabilisers, and with more than 30% lower risk than any antidepressant, any antipsychotic, quetiapine, or lamotrigine monotherapy in the head-to-head analysis. Lithium was also associated with the lowest risk of somatic hospitalisation. Our findings supported the use of lithium as the mainstay of treatment in bipolar disorder. The Swedish Research Council for Health, Working Life and Welfare, FORTE (grant number 2021-01079).
Read full abstract