Intended for healthcare professionals


Follow up study of longstanding depression as predictor of mortality in elderly people living in the community

BMJ 1999; 318 doi: (Published 13 February 1999) Cite this as: BMJ 1999;318:432
  1. Tuula Pulska, research fellow (Tuula.Pulska{at}Oulu.Fi)a,
  2. Kimmo Pahkala, counsellorb,
  3. Pekka Laippala, professor in biometryc,
  4. Sirkka-Liisa Kivelä, professor in public health science and general practicea
  1. aUniversity of Oulu, Department of Public Health Science and General Practice, Aapistie 1, FIN-90220 Oulu, Finland
  2. bHealth and Social Services, Provincial State Office of Western Finland, Regional Service Unit, Vaasa, Finland
  3. cUniversity of Tampere, Tampere School of Public Health and Tampere University Hospital, Tampere, Finland
  1. Correspondence to: Dr Pulska
  • Accepted 16 July 1998

Longstanding and recurrent depressive disorders are quite common in elderly people. 1 2 Not much is known, however, of their clinical course and prognosis, including mortality. The need for treatment of longstanding, less severe depressive disorders is a matter of discussion.

We studied the relation between longstanding or recurrent depressive disorders and mortality and that between recovery from depressive disorders and mortality in elderly people.

Subjects, methods, and results

This study is based on the Ähtäri longitudinal epidemiological research project concerning depression in elderly people.13 The initial series consisted of people born in 1923or earlier and living in the municipality of Ähtäri, Finland, on 1January 1984(n=1529). In the first study in 1984-5the participation rate was 91%. The follow up study was performed in 1989-90with a participation rate of 94%. Depression was determined after semistructured interviews by the criteria of the Diagnostic and Statistical Manual of Mental Disorders, third edition (DSM-III).13

Predictors of mortality according to Cox's model (forced model)

View this table:

We examined mortality in subjects with a longstanding or recurrent course of depression and those who had recovered. Three groups were formed from those people without dementia who were alive in both 1984-5and 1989-90: people depressed in both assessments (n=78), people depressed in 1984-5but not depressed in 1989-90 (n=101), and people not depressed in both assessments (n=634). The mean (SD) age of those participating in the follow up study was 74.3(6.1) years on 1January 1989.The mortality data from the official statistics were collected for a period from the individual examination days in 1989-90to 31December 1995.

The causes of death did not differ between the groups, cardiovascular and cerebrovascular diseases and malignant neoplasms being the most common. According to Kaplan-Meier survival analysis, 48% of the people with depression at both time points had died compared with 26% in the group without depression at both times (P<0.001). In the group with depression in 1984-5but not in 1989-90, 31% had died, so the survival in this group did not differ from that in the group without depression at either time (P=0.286). The role of depression as a predictor of mortality was analysed with Cox's proportional hazards model, with age, sex, smoking, physical health, and functional abilities taken into account. Longstanding depression predicted mortality even when these factors were controlled for, while recovery from depression did not (table).


Longstanding depression seems to be a predictor for mortality in elderly people. In this study the groups of depressed people were formed on the basis of two measurements at interval of 5 years, and there were no data on the course of depression between the measurements. We assumed, however, that subjects with depression at both time points were suffering from longstanding or recurrent depression, and that this group and the group with depression at the first time point but without depression at the second differed from each other as to the course of their depression.

The results showed longstanding depression predicted mortality, whereas recovery from depression did not. More people in the group with depression at both time points than in the group who no longer had depression at the second time had had serious diseases or operations during follow up from 1984-5to 1989-90.They had hence experienced more physical and psychosocial stress, which may be an underlying factor affecting mortality.4 Apart from the variables concerning physical health and functional abilities inserted into the model, previous stress may also be a factor contributing to the higher mortality.

Clinical depression was determined over a 5year interval, which may explain why the results differ from those of a previous study. According to the results of Thomas et al, depressive symptoms (emergent symptoms, remission, or persistent symptoms) measured over a 2year interval were not associated with mortality.5

Our results support the proposal that more attention should be given to the treatment of longstanding and recurrent depressive disorders in elderly people. The effects of treatment should also be studied.


Contributors: TP participated in designing, analysing, and interpreting the results of the mortality study and wrote the initial version of this paper, which was discussed, revised, and accepted by all authors. S-LK supervised the community study on depression in old age and produced the initial ideas of the study design, methods, analyses, and interpretation of results. KP participated in designing, analysing, interpreting, and collecting data. PL was responsible for data analyses both in this mortality study and in the whole epidemiological study. TP and S-LK are guarantors of the study.

Funding: Academy of Finland, the Päivikki and Sakari Sohlberg Foundation, the YrjöJahnsson Foundation, the Juho Vainio Foundation, the Signe and Ane Gyllenberg Foundation, Tampere University Hospital Medical Research Fund, and Oulu University Hospital Medical Research Fund.

Conflict of interest: None.