Pharmacological treatments and risk of readmission to hospital for unipolar depression in Finland: a nationwide cohort study

Background: Little is known about the comparative effectiveness of long-term pharmacological treatments for severe unipolar depression. We aimed to study the effectiveness of pharmacological treatments in relapse prevention in a nationwide cohort of patients who had been admitted to hospital at least once as a result of unipolar depression.

Methods: Our nationwide cohort study investigated the risk of readmission to hospital in 1996-2012 in all patients in Finland who had been admitted to hospital at least once for unipolar depression (without a diagnosis of schizophrenia or bipolar disorder) in Finland between Jan 1, 1987, and Dec 31, 2012. We used nationwide databases to obtain data for hospital admission, mortality, and dispensed medications. Exposure and non-exposure periods for medications were established using the PRE2DUP method. The primary analysis was within-individual analysis of readmission to hospital in the total cohort, in which each individual was used as his or her own control to eliminate selection bias. Putative survival and protopathic biases were controlled in sensitivity analyses. Since 33 independent statistical comparisons were done for specific medications, the level of statistical significance was set at p<0·0015.

Findings: Data from 123 712 patients were included in the total cohort, with a mean follow-up time of 7·9 years (SD 5·3). Lithium use was associated with a lower risk of re-admission to hospital for mental illness than was no lithium use (hazard ratio [HR] 0·47 [95% CI 0·40-0·55]; p<0·0001), whereas the groups of antidepressants (HR 1·10 [1·06-1·13]; p<0·0001) and antipsychotics (HR 1·16 [1·12-1·20]; p<0·0001) were not associated with a reduced risk of readmission to hospital. Risk of hospital readmission was lower during lithium therapy alone (HR 0·31 [0·21-0·47]; p<0·0001) than during use of lithium with antidepressants (HR 0·50 [0·43-0·59]; p<0·0001). After lithium, clozapine (HR 0·65 [0·46-0·90]; p=0·010) and amitriptyline (HR 0·75 [0·70-0·81]; p<0·0001) were the specific agents associated with the next lowest risk of readmission. In the sensitivity analyses controlling for survival and protopathic biases, all drugs were associated with lower rates of readmission to hospital than they were in the primary analysis, showing the same rank order in comparative effectiveness. The lowest mortality was observed during antidepressant use (HR 0·56 [0·54-0·58]; p<0·0001).

Interpretation: Our results indicate that lithium, especially without concomitant antidepressant use, is the pharmacological treatment associated with the lowest risk of hospital readmission for mental illness in patients with severe unipolar depression, and the outcomes for this measure related to antidepressants and antipsychotics are poorer than lithium. Lithium treatment should be considered for a wider population of severely depressed patients than those currently considered, taking into account its potential risks and side-effects.

Funding: The Finnish Ministry of Health.