Decline in suicide mortality after psychiatric hospitalization for depression in Finland between 1991 and 2014

Depression is the most important mental disorder in terms of suicide mortality. Numerous studies over time have estimated the lifetime risk of suicide in depression, including a recent Danish national study1. Organization of services and treatment practices for depression have undergone major changes over the past decades, including remarkable growth in the use of antidepressants, emphasis on community‐based services, and deinstitutionalization. Temporal trends in suicide mortality among psychiatric patients with depression can be expected, but have not been investigated.

We followed a Finnish population‐based cohort of depressive patients (N=56,826), with a first lifetime hospitalization due to depression between 1991 and 2011, up to the end of the year 2014 (maximum follow‐up: 24 years). Here we report both cumulative risk of suicide and temporal trends in suicide mortality.

This study stems from the MERTTU research project2. Complete data at individual level via the Finnish identity codes were linked from the Finnish Hospital Discharge Register, containing data on all inpatient treatments, and Statistics Finland's register on causes of death.

We identified from the Finnish Hospital Discharge Register all ≥18‐year‐old people with a psychiatric diagnosis admitted to a psychiatric hospital or a psychiatric ward of a general hospital between 1987 and 2011. We next obtained data on patients’ hospitalizations with psychiatric diagnoses between 1980 and 2011. Baseline hospitalizations for depressive disorder (principal diagnosis) in 1991‐2011 were identified through the Finnish ICD‐9³ codes 2961A‐G and 2968A (used in 1987‐1995) and ICD‐10 codes F32‐33. Since 1987, the national guidelines have stipulated applying operationalized criteria for clinical psychiatric diagnoses (the Finnish ICD‐9³ was based on the DSM‐III‐R criteria and the ICD‐10 on the Diagnostic Criteria for Research). The Finnish Hospital Discharge Register displays complete coverage and good accuracy of mental health diagnoses4.

We excluded patients with previous psychiatric hospitalizations since 1980, with a principal diagnosis of psychotic disorder at baseline, or who died by suicide during the baseline hospitalization.

We retrieved the dates and causes of death for all cases, and then identified suicides (ICD‐9 codes E950A‐K, E951A‐E957A, E959A‐C, E959X; ICD‐10 codes X60‐X76, X78, X80‐X84, Z91.5, Y87.0). Finland has high medico‐legal autopsy rates (performed by a forensic pathologist who identifies all suicide and unnatural deaths). Overall, the death investigation process leaves few undetermined deaths5.

Patients were followed up from the day of discharge to death by suicide or other cause, or until December 31, 2014, whichever occurred first. Events other than suicides were treated as censored. Diagnostic conversions to a principal psychotic or bipolar disorder were ignored because of risk of inducing survival bias.

The survival function and cumulative risk of suicide were estimated with the Kaplan‐Meier product limit estimator. For time‐trend analyses, we formed consecutive 5‐year cohorts by admission year. We estimated for each cohort (years 1991‐1995 as reference) the age‐ and gender‐adjusted proportional hazards for suicide over 3 years (equal length of individual follow‐ups) and maximum 24‐year follow‐up (varying length of individual follow‐ups). We used software packages R and Survo.

A national cohort of 56,826 patients (25,188 men and 31,638 women) with first lifetime hospitalization for depression was followed for 628,514 person‐years (follow‐up: mean 11.1 years, median 10.7 years, maximum 24 years). Of 15,063 patients who died during the follow‐up, 2,567 (17.0%) died by suicide (1,598 men, 969 women). The cumulative risk of suicide was 6.13% overall (95% CI: 5.80‐6.46%), 8.64% in men (95% CI: 8.00‐9.27%), and 4.14% in women (95% CI: 3.83‐4.45%). The suicide incidence rate in men was 23.05 per 1,000 person‐years (95% CI: 21.20‐25.02) for the first 12 months; 8.84 per 1,000 person‐years (95% CI: 7.69‐10.10) for 12‐24 months; and 6.12 per 1,000 person‐years (95% CI: 5.17‐7.20) for 24‐36 months. The corresponding rates in women were 9.73 per 1,000 person‐years (95% CI: 8.68‐10.87), 3.82 per 1,000 person‐years (95% CI: 3.17‐4.57), and 3.19 per 1,000 person‐years (95% CI: 2.60‐3.88).

Relative to baseline years 1991‐1995, the age‐ and gender‐adjusted hazard ratio for suicide within 3 years post‐discharge was 0.69 (95% CI: 0.61‐0.79, p<0.0001) in 1996‐2000, 0.54 (95% CI: 0.47‐0.63, p<0.0001) in 2001‐2005, and 0.48 (95% CI: 0.42‐0.56, p<0.0001) in 2006‐2011. The corresponding hazard ratio for maximum 24‐year follow‐up was 0.70 (95% CI: 0.63‐0.77, p<0.0001) in 1996‐2000, 0.57 (95% CI: 0.51‐0.64, p<0.0001) in 2001‐2005, and 0.49 (95% CI: 0.43‐0.55, p<0.0001) in 2006‐2011.

These Finnish cumulative risks of 8.6% in men and 4.1% in women are slightly higher than the cumulative Danish incidences of 6.7% in male and 3.8% in female depressive in‐ and outpatients1. The difference likely reflects higher overall suicide mortality in Finland, and our inclusion of only inpatients. Overall, hospital samples may overestimate the suicide mortality in depression by about 30‐50%6.

Our findings show a considerable and consistent decline in long‐term suicide mortality since 1991, in contrast with the results of a recent meta‐analysis of discharged general psychiatric inpatients7. In Finland, the suicide rates peaked in 1990 and were approximately halved by 2014. Our data are thus consistent with this overall pattern.

Following worldwide trends, numerous changes have occurred in Finland since 1990. First, the national Suicide Prevention Project was implemented in the early 1990s. Second, the consumption of antidepressants has eight folded during the study period8. Third, per capita alcohol consumption rose in 1990‐2005, showing a downtrend since 2007. Fourth, the deinstitutionalization process has resulted in a reduction of about 60% in the number of psychiatric beds (equalling in 2011 the average in countries which are part of the Organisation for Economic Co‐operation and Development (OECD): 71 per 100,000 vs. 70 per 100,000)⁸. From 1994 to 2011, the number of inpatient days have halved, and the number of treated patients reduced by 10%9. Fifth, the availability of psychiatric outpatient care has improved and outpatient‐oriented services are associated with lower suicide mortality2.

We conclude that the cumulative suicide risk in depression depends on time period and context. The large downtrend in suicide mortality of psychiatric inpatients in Finland over current treatment era is encouraging for ongoing efforts to prevent suicides in depression.

Kari I. Aaltonen1,2, Erkki Isometsä1, Reijo Sund3,4, Sami Pirkola2,5
1Department of Psychiatry, University of Helsinki and Helsinki University Central Hospital, Helsinki, Finland; 2Department of Public Health Solutions, Mental Health Unit, National Institute for Health and Welfare, Helsinki, Finland; 3Centre for Research Methods, Faculty of Social Sciences, University of Helsinki, Helsinki, Finland; 4Institute of Clinical Medicine, University of Eastern Finland, Kuopio, Finland; 5Faculty of Social Sciences, University of Tampere and Pirkanmaa Hospital District, Tampere, Finland