Mortality in affective disorders

Mortality in affective disorders

Journal of Affective Disorders 65 (2001) 263–274 www.elsevier.com / locate / jad Research report Mortality in affective disorders ¨ b , Michael Phil...

92KB Sizes 1 Downloads 47 Views

Journal of Affective Disorders 65 (2001) 263–274 www.elsevier.com / locate / jad

Research report

Mortality in affective disorders ¨ b , Michael Philipp c Barbara Schneider a , *, Matthias J. Muller a

Centre of Psychiatry, Department of Psychiatry and Psychotherapy I, University of Frankfurt /Main, Heinrich-Hoffmann-Str. 10, D-60528 Frankfurt /Main, Germany b Department of Psychiatry, University of Mainz, Mainz, Germany c Psychiatric Clinic Landshut, Landshut, Germany Received 6 November 1999; received in revised form 5 July 2000; accepted 4 August 2000

Abstract Background: To investigate the mortality rates in affective disorders due to unnatural and natural causes with respect to illness subtype and social–demographic features. Methods: Mortality data were determined from a prospective study of 354 outpatients with affective disorders during a follow-up period of 5 years. Death from natural and unnatural causes was compared to sex- and age-specific expectations in the general population. Standardized mortality rates (SMR) in diagnostic subgroups and the influence of social–demographic features were investigated. Results: The observed 30 deaths represented nearly three times (SMR, 2.9) the number expected on the basis of age- and sex-standardized reference population rates. Death from natural causes occurred with the same rate as expected (SMR, 1.0), death from unnatural causes was 28.8 times higher than expected. Women with affective disorders had a considerable high risk to die from unnatural causes (SMR, 47.1). A significant excess of unnatural death was found in all subtypes of affective disorders, particularly in recurrent major depressive episodes (SMR, 46.7). Limitations: The sample was restricted in size. Therefore subgroup differences and multiple relationships of risk factors could not be analyzed with high statistical power. Conclusions: The results corroborate earlier findings of excess mortality in major affective disorders and strengthen the view that suffering from recurrent major depression confers per se an important biological risk for suicide. Natural causes of death in affective disorders are comparable to expectations from reference populations. Social–demographic characteristics may contribute to an additional risk of premature death by suicide, particularly in women.  2001 Elsevier Science B.V. All rights reserved. Keywords: Over-mortality; Unnatural death; Natural death; Subtypes of affective disorders; Social–demographic features

1. Introduction It is well-known for decades that the mortality risk of patients with affective disorders is elevated *Corresponding author. Tel.: 1 49-69-630-14784; fax: 1 4969-630-15920.

(Odegard, 1952; Malzberg, 1953; Bratfos and Haug, 1968; Tsuang and Woolson, 1978; Angst and Stassen, 1986; Black et al., 1987b; Zilber et al., 1989; Newman and Bland, 1991; Surtees and Barkley, 1994; Kouzis et al., 1995; Hansen et al., 1997; Harris and Barraclough, 1998; Angst et al., 1999; Hoyer et al., 2000), with much of the excess mortality ac-

0165-0327 / 01 / $ – see front matter  2001 Elsevier Science B.V. All rights reserved. PII: S0165-0327( 00 )00290-1

264

B. Schneider et al. / Journal of Affective Disorders 65 (2001) 263 – 274

counted for by suicide. Several studies have found that lifetime risk of suicide in major mood disorders is about 30 times higher than in the general population (Pokorny, 1964; Guze and Robins, 1970; Diekstra, 1989). Psychological autopsies have shown that about 30–70% of suicides in a general population are due to affective disorders (Robins et al., 1959; Dorpat and Ripley, 1960; Barraclough et al., 1974; Beskow, 1979; Rich et al., 1986). Historically, mortality studies of psychiatric patients have focused on diagnostically heterogeneous groups of patients. Research interest has recently moved towards studying mortality in particular diagnostic groups and different categories of psychiatric disorders, motivated by the finding that mortality risks may be specific to diagnosis, and subtype (Black et al., 1987b). In bipolar affective disorder, mortality was 2.1–2.6 times higher than in age- and sex-matched control groups (Dalgard, 1966; Newman and Bland, 1991), in unipolar depressed patients excess mortality was 1.9–6.2 (Malzberg, 1937, 1953; Newman and Bland, 1991; Surtees and Barkley, 1994; Kouzis et al., 1995). Black et al. (1987b) have found that the suicide risk in affective disorders is determined by the extent of depressive symptoms. In patients with mixed or manic bipolar disorder, the severity of concurrent depressive symptoms in mania, rather than the presence of a depressive syndrome per se, was associated with suicidality (Strakowksi et al., 1996). Excess mortality from unnatural death has been found in all diagnostic subgroups of affective disorder (Tsuang and Woolson, 1978; Black et al., 1987b) with higher rates of suicide in unipolar than in bipolar patients in most of the studies (Angst et al., 1979; Black et al., 1987b; Dingman and McGlashan, 1986; Perris and D’Elia, 1966; Newman and Bland, 1991). Beside the difficult differential lifetime diagnosis between unipolar and bipolar depression, there are, however, also conflicting results. In a study assessing mortality rates over 25 years in about 200 patients (Brodaty et al., 1997), the subtype of depressive illness did not differentially predict mortality. Morrison (1982) reported data from a general practice found patients with bipolar subtypes to have a 7.5-fold greater risk of suicide than patients in unipolar disorders. This differences could be partly explained by additional risk factors of suicide which

should be taken into account (e.g. comorbidity with alcohol dependence and anxiety disorders). Different authors have found ambiguous results about the suicide risk in male and in female patients in the different diagnostic subgroups: Tsuang and Woolson (1978), Newman and Bland (1991) and Brodaty et al. (1997) showed a higher risk attributable to female gender with respect to affective illness and to unnatural death, whereas other authors (Barraclough and Pallis, 1975; Modestin and Kopp, 1988; Henriksson et al., 1993; Isometsa¨ et al., 1994a,b) have found male gender to be a risk factor for unnatural death, especially in depressed patients. Furthermore, suicidality seems to be influenced by ethnic, social–demographic and psychological features independently from diagnosis and diagnostic subtype, but data on that issue are scarce. Studies within different contexts show that higher mortality rates were found in all unmarried, i.e. separated, divorced, widowed and never married subjects (Robinson, 1962; Dublin, 1963; Hartelius, 1967; ´ 1980; Davidson and Philippe, 1983) with Fuse, higher risk in separated, divorced and widowed than in unmarried subjects (Tuckman and Youngman, 1963; Cohen et al., 1966; Murphy, 1962; Adam, 1985). In previous studies (Barraclough and Pallis, 1975; Roy, 1984; Pokorny, 1983; Johansson et al., 1997) living alone has been found to be a very strong risk factor for suicide in psychiatric illness (SMR, 2.15). Only in a Swedish population unmarried subjects show minimum death rates (Jacobsson and Renberg, 1986). But most of these studies are not focused on patients with affective disorders. In a recent study, comparing patients with a first suicidal attempt, patients with adjustment reactions were significantly less educated, had a lower social status and in most cases were unmarried when compared to patients with major depression attempting suicide (Polyakova et al., 1998). This finding could imply that in patients with major affective disorders social–demographic features may contribute to a lesser degree to suicidality than in patients with acute adjustment reactions. Furthermore a few studies (Durkheim, 1897; Hoyer and Lund, 1993) have raised the crucial question whether marriages with multiple children or children per se offer protection against suicide. With respect to natural deaths, a significant excess

B. Schneider et al. / Journal of Affective Disorders 65 (2001) 263 – 274

in affective disorders has been found in several studies (Malzberg, 1937; Zilber et al., 1989; Weeke et al., 1987, Newman and Bland, 1991). However, other investigators have only observed either a slight insignificant or no over-mortality from natural causes (Martin et al., 1985b; Murphy et al., 1987; Hewer ¨ and Rossler, 1997). But only in a few former studies, mortality in affective disorders was investigated taking into account the specific illness subtype (Black et al., 1987a). Our study aims to describe the association between mortality from natural and from unnatural causes in outpatients with different subtypes of affective disorders with special regard to social– demographic characteristics in comparison to the general population.

2. Methods

2.1. Sample characteristics Originally, a sample of 559 consecutively admitted inpatients of a psychiatric hospital aged 18–80 without organic brain disorders were assessed (64.6% females, mean age at index evaluation 42.2614.6 years; mean age at onset 36.8614.0 years). A total of 354 of the 559 patients with affective disorders (mean age at index evaluation 43.2614.2 years, range 18–80; mean age at onset 38.4613.7 years, range 11–74, 242 females) were interviewed with the Polydiagnostic Interview (Philipp and Maier, 1986) which includes quantitative measures of all psychopathological symptoms and social–demographic features (marital status, existence of children, highest educational and professional level) between July 1, 1983 and July 1, 1988. Cases, in which precise information about a social– demographic feature could not be obtained, were dropped from mortality analyses. The psychiatric diagnoses were generated with the aid of a computer program (Philipp and Delmo, 1987). Five years later, all index patients were contacted again for a second interview. In the follow-up study, 352 of the 354 patients (99.4%) could be located. Information about the deceased patients were obtained from their relatives, friends, general practitioners, psychiatrists and from medical records.

265

Deaths were first categorized as due to ‘natural’ or ‘unnatural’ causes. Unnatural deaths refer to external causes and included suicides, accidents and homicides (ICD-9 No. 800–999). Natural deaths included those in which no ‘external’ cause could be identified; they were categorized as due to circulatory system diseases (including cardiovascular and cerebral-vascular disorders), neoplasm, infection and other causes of natural death.

2.2. Mortality analysis The observed mortality in the total sample, as well as in various subgroups, was compared to sex- and age-specific expectations derived from 1988 death rates for Germany. The technique was similar to the methods described in detail by Woolson et al. (1978) and used in a series of mortality studies (Tsuang, 1978; Tsuang and Woolson, 1978; Tsuang et al., 1980; Martin et al., 1985a). Sex- and age-specific mortality rates of the German population were calculated from vital statistics according to age range by intervals of five years. These age- and sexspecific deaths rates were then multiplied by the number of persons of a particular cohort exposed to risk during each 5-year interval of age. Patients were considered to be at risk of death from all causes up to the point of death or the follow-up interview. The expected number of deaths for the cohort was obtained by summing these products. The ratio of observed to expected number of deaths is termed the standardized mortality ratio (SMR). The Poisson distribution was used to test the significance of any excess in observed over expected mortality. Cases in which precise information about a social–demographic feature could not be obtained were dropped from mortality analyses. Additionally, for SMR, of diagnostic subgroups, approximated 95% confidence intervals (CI95) were calculated. In choosing the mortality rates of Germany in 1988 as the reference, it was assumed that death rates were constant in the study years surrounding 1988 (1983– 1993) and that any urban–rural differences between the study and reference population did not affect the overall results. The patients for whom the mortality status could not be ascertained were dropped from the statistical analyses. Social–demographic characteristics of the suicide-group, the group of the still

B. Schneider et al. / Journal of Affective Disorders 65 (2001) 263 – 274

266

alive, and the group of patients who died from other causes were compared non-parametrically by x 2 tests, Mann–Whitney U-tests (MWU-tests) and Kruskal–Wallis analysis of variance (KW-ANOVA). The two-tailed level of statistical significance was set at a 5 0.05; due to the explorative nature of the study, no adjustment for multiple testing was performed. SMR values $ 3 were detectable with a power of approximately 80% (with a 5 0.05) in the smallest subgroup analyzed.

3. Results The age at index evaluation and the age at onset of illness in the different diagnostic subgroups are shown in Table 1. Significant differences between different diagnostic subgroups were found with respect to the existence of children in the total group (P 5 0.029, x 2 -test) and in the female (P 5 0.0046, x 2 -test), but not in the male subgroup. No further

differences between diagnostic subgroups emerged in any other social–demographic feature. Data from 352 (99.4%) of the 354 index patients could be evaluated 5 years after the index assessment; 322 patients were still alive and 30 (8.5%) were dead. Mortality status could not be ascertained for a 56-year-old woman (at index evaluation) who died at age 60. Two patients had moved; their new addresses could not be found out. Ten deaths were attributed to natural causes (five dying from cancer, four dying from diseases of the cardiovascular system and one dying from peritonitis caused by diverticulitis), 19 deaths to unnatural causes, 18 of them to suicide and one to car-accident as a passenger. Thirty patients (8.3% of the 352 patients traced) died, 10.4 deaths were expected (SMR, 2.9; CI95, 1.8–4.0). The observed mortality was significantly higher than expected (P , 0.0001) and was significantly increased among women (24 deaths observed, 5.6 expected; SMR, 4.3, P , 0.0001; CI95, 2.5–6.1), but not among men (six deaths observed, 4.8 expected; SMR, 1.3, P . 0.10). Mortality from

Table 1 Sample characteristics DSM-III-R ICD-9

Mania

Total sample Age at onset of illness Mean6S.D. Range Age at index evaluation Mean6S.D. Range

n 5 26

Bipolar disorder

MDE

Between group differences a

Recurrent

Single episode

n 5 48

n 5 137

n 5 143

30.8612.9 [16–66]

36.5611.1 [16–57]

41.3613.9 [11–74]

37.6614.0 [15–69]

P , 0.0001

37.7614.2 [19–68]

42.0612.3 [19–70]

47.9613.6 [22–76]

40.7614.9 [18–80]

P 5 0.0006

Men Age at onset of illness Mean6S.D. Range Age at index evaluation Mean6S.D. Range

n 5 12

n 5 12

n 5 36

n 5 52

29.0611.3 [16–50]

40.967.5 [29–52]

40.0612.9 [20–65]

37.6614.9 [15–66]

n.s.

34.9614.2 [20–66]

46.267.6 [29–58]

45.4614.5 [22–76]

40.0615.8 [19–80]

P 5 0.048

Women Age at onset of illness Mean6S.D. Range Age at index evaluation Mean6S.D. Range

n 5 14

n 5 36

n 5 101

n 5 91

32.4614.4 [16–66]

34.9611.9 [16–57]

42.0614.2 [11–74]

37.5613.5 [16–69]

P 5 0.0035

40.0614.2 [19–68]

40.6613.3 [19–70]

48.7613.2 [22–76]

41.0614.4 [18–72]

P 5 0.0006

a

Results from Kruskal–Wallis analysis of variance; n.s., not significant.

B. Schneider et al. / Journal of Affective Disorders 65 (2001) 263 – 274

all causes was not significantly different between men and women as evidenced by x 2 -tests (P . 0.10). When comparing the three groups of patients (still alive, died from natural causes, and died from unnatural causes) with respect to several features, the following results were obtained. There were significant global differences between the three groups of patients (KW-ANOVA) in the age at index evaluation (P 5 0.0001), but not in the age at onset (P 5 0.086). Furthermore, significant differences were yielded in the distribution of educational levels (P 5 0.042), but neither in the distribution of professional levels (P . 0.10), marital status, nor in the existence of children ( x 2 -tests, P . 0.10, respectively). Particular testing between group differences (MWU-tests) revealed, that patients who died from natural causes were significantly older at index evaluation (62.269.4, range 43–76 years) than the still alive patients (42.6614.1, 18–80 years, P , 0.001) and the patients who died from unnatural causes (48.4614.6, 22–76 years, P 5 0.005); patients who died from natural death were significantly older at age at onset (57.1610.0, 38–71 years) than patients who were still alive (37.9613.3, 11–58, P 5 0.026), but not than patients who died from unnatural death (39.3617.1, 15–74 years, P . 0.10). The mean age at death was 55.8614.3 years (range 27–78 years). The patients who died from natural causes (66.068.8; range 48–78 years) were significantly older than the patients who died from unnatural causes (50.1614.1; 27–77 years; P , 0.01). At age of death, women who died from natural deaths (67.465.9; 61–76 years) were also significantly older than women who died from unnatural deaths (47.9614.1; 27–77 years; P , 0.01). This result could not be found in male patients (natural deaths: 62.7615.0; 48–78 years; unnatural deaths: 62.066.2; 55–67 years). No differences in the mean age at death from all causes were found between men and women (male patients: 62.3610.3; 48–78 years; female patients: 54.2614.9; 27–77 years). Subjects who committed suicide (mean age at onset: 36.8619.4 years; mean age at death: 49.8614.4; 27–77 years) had not a significantly earlier onset, but died earlier (P , 0.001) than patients who died from other reasons (mean age at onset 41.9626.7 years; mean age at death: 64.868.6, 48–78 years). Thirteen

267

of the 18 patients (72.2%) who committed suicide, died within 2 years after index evaluation.

3.1. Death from unnatural causes In the total sample of patients, the observed mortality of unnatural causes (n 5 19) was in significant excess of that of the expected deaths (0.7 deaths expected; SMR, 28.8; CI95, 15.6–42.0). The female subgroup experienced significantly higher mortality due to unnatural causes than the general population (Table 2) and also the male subgroup (Table 2). Mortality from unnatural causes was not significantly different between men and women as evidenced by x 2 -tests (P . 0.10). Table 2 shows the mortality rates in the diagnostic subtypes for the total group of patients, and separately for male and female patients. As Table 2 indicates, the observed mortality from all causes as well as from unnatural causes was significantly higher than expected in all diagnostic groups except mania. Particularly high was the mortality from unnatural causes in patients with recurrent major depressive episodes (MDE). About 60% of all deaths were caused by suicide (50% in the bipolar, 72.2% in the group with MDE, recurrent, and 62.5% in the group with MDE, single episode). Thus, suicide was the single largest cause of death in patients with affective disorders. In the male subgroup, the observed mortality from unnatural causes was significantly higher than expected only in patients with recurrent MDE (Table 2); in the female subgroup the results were similar to those of the total sample (Table 2). Generally, women had a considerable higher risk to die from unnatural causes (SMR, 47.1) than men (SMR, 7.1, Table 2). Table 3 shows the mortality rates in different age groups. In all age groups, patients died significantly more often than expected from all causes as well as from unnatural causes; this result was also found in women, but not in men (Table 3). Especially high was the over-mortality from unnatural death in the total sample of patients older than 49 years. Patients with recurrent MDE died significantly more often than expected in the highest age group from all causes (SMR, 3.2; P , 0.0001; CI95, 1.7–5.3) and in the middle and highest age groups from unnatural causes (30–49 years: SMR,

B. Schneider et al. / Journal of Affective Disorders 65 (2001) 263 – 274

268

Table 2 Index diagnosis (DSM-III-R) and unnatural death during follow-up a All causes of death

Total sample (n 5 354) Mania (n 5 26) Bipolar, depressed (n 5 48) MDE, recurrent (n 5 137) MDE, single (n 5 143) Men (n 5 112) Mania (n 5 12) Bipolar, depressed (n 5 12) MDE, recurrent (n 5 36) MDE, single (n 5 52) Women (n 5 242) Mania (n 5 14) Bipolar, depressed (n 5 36) MDE, recurrent (n 5 101) MDE, single (n 5 91)

Unnatural causes of death

Obs

Exp

SMR

CI95

P-value

Obs

Exp

SMR

CI95

P-value

30 0 4 18 8 6 0 1 4 1 24 0 3 14 7

10.37 0.50 0.90 5.10 3.87 4.78 0.28 0.36 1.91 2.23 5.59 0.22 0.54 3.19 1.64

2.9 0 4.5 3.5 2.1 1.3 0 2.8 2.1 0.5 4.3 0 5.6 4.4 4.3

1.8–3.9 – 1.2–11.4 2.1–5.6 – – – – – – 2.8–6.4 – 1.2–16.2 2.4–5.4 1.7–8.8

P , 0.0001 n.s. P 5 0.024 P , 0.0001 n.s. n.s. n.s. n.s. n.s. n.s. P , 0.0001 n.s. P 5 0.032 P , 0.0001 P 5 0.0022

19 0 2 14 3 3 0 0 3 0 16 0 2 11 3

0.76 0.06 0.09 0.30 0.31 0.42 0.04 0.04 0.14 0.19 0.34 0.02 0.05 0.16 0.12

25.0 0 22.2 46.7 9.7 7.1 0 0 21.4 0 47.1 0 40 68.8 25.0

15.1–39.0 – 4.0–80.3 21.7–71.6 2.0–28.3 1.5–20.9 – – 4.4–62.6 – 26.9–76.4 – 4.8–144.5 34.3–123.0 5.2–73.1

P , 0.0001 n.s. P 5 0.0074 P , 0.0001 P 5 0.0073 P 5 0.016 n.s. n.s. P 5 0.0008 n.s. P , 0.0001 n.s. P 5 0.0024 P , 0.0001 P 5 0.0005

a Obs, number of observed deaths; Exp, number of expected deaths; SMR, standard mortality rate; MDE, major depressive episode; CI95, 95% confidence interval; P-values, two-tailed Poisson test; n.s., not significant.

Table 3 Age at index evaluation and unnatural death during follow-up a Age (years)

Index sample

All causes of death Obs.

Exp.

SMR

76 157 121

2 7 21

0.25 1.64 8.48

Men , 29 30–49 . 49

29 47 36

0 1 5

Women , 29 30–49 . 49

47 110 85

2 6 16

Total sample , 29 30–49 . 49

Unnatural causes of death CI95

P-value

Obs

Exp

SMR

CI 95

P-value

7.9 4.3 2.5

1.0–28.6 1.7–8.8 1.5–3.8

P 5 0.0122 P , 0.0001 P , 0.0001

2 6 11

0.15 0.29 0.33

13.8 20.9 33.3

1.7–49.8 7.7–45.5 16.6–59.6

P 5 0.0003 P , 0.0001 P , 0.0001

0.15 0.68 3.94

0 1.5 1.3

– – –

n.s. n.s. n.s.

0 0 3

0.10 0.16 0.16

0 0 18.3

– – 3.8–53.5

n.s. n.s. P , 0.0001

0.10 0.96 4.54

20 6.3 3.5

2.4–72.2 2.3–13.7 2.0–5.7

P , 0.0001 P , 0.0001 P , 0.0001

2 6 8

0.05 0.13 0.17

43.5 46.5 48.2

5.3–157.1 17.1–101.2 20.8–95.0

P , 0.0001 P , 0.0001 P , 0.0001

a Obs, number of observed deaths; Exp, number of expected deaths; SMR, standard mortality rate; P-values, two-tailed Poisson test; * P , 0.05; ** P , 0.01, *** P , 0.001; n.s., not significant.

27.8; P , 0.0001; CI95, 5.7–81.2; . 49 years: SMR, 60.2; P , 0.0001; CI95, 28.9–110.8). In patients with single MDE, patients aged 30–49 years showed a significant over-mortality from unnatural death (SMR, 27.0, P , 0.001, CI95, 5.6–79.0). Additional explorative analyses revealed that the over-mortality from all as well as from unnatural causes could also be found in the subgroups of

marital status and of patients with and without children (Table 4). Married women (SMR, 56.5, P , 0.0001, CI95, 26.9–103.9) and not married women (SMR, 46.3, P , 0.0001, CI95, 15.0–108.0), women with children (SMR, 40, P , 0.0001, CI95, 17.3–78.8) and childless women (SMR, 72.7, P , 0.0001, CI95, 13.2–143.3) as well as unmarried men (SMR, 8.7, P 5 0.022, CI95, 1.1–31.4) and men

B. Schneider et al. / Journal of Affective Disorders 65 (2001) 263 – 274

269

Table 4 Marital status, existence of children and unnatural death during follow-up a Index sample

All causes of death Obs.

Exp.

SMR

CI95

P-value

Unnatural causes of death Obs

Exp

SMR

CI95

P-value

Total sample Married Not married With children Without children

198 143 210 144

19 9 17 18

6.96 3.13 7.82 2.49

2.7 2.9 2.2 7.2

1.6–4.3 1.3–5.5 1.3–3.5 4.3–11.4

P , 0.0001 P 5 0.0020 P 5 0.0048 P , 0.0001

11 7 11 8

0.38 0.26 0.41 0.23

29.1 26.7 41.9 35.1

14.5–52.1 10.7–55.0 13.5–48.5 15.1–69.1

P , 0.0001 P , 0.0001 P , 0.0001 P , 0.0001

MDE, recurrent Married Not married With children Without children

84 48 93 44

11 6 12 6

3.28 1.71 3.77 1.36

3.4 3.5 3.2 4.4

1.7–6.0 1.3–7.6 1.6–5.6 1.6–9.6

P 5 0.0008 P 5 0.0031 P 5 0.0008 P 5 0.0046

8 5 9 5

0.16 0.09 0.18 0.08

50.3 53.8 50.3 60.2

21.7–99.1 17.5–125.5 23.0–95.4 19.6–140.6

P , 0.0001 P , 0.0001 P , 0.0001 P , 0.0001

MDE, single episode Married Not married With children Without children

77 63 83 60

6 2 4 4

2.74 1.08 3.11 0.77

2.2 1.9 1.3 5.2

– – – 1.4–13.3

n.s. n.s. n.s. P 5 0.014

2 1 2 1

0.15 0.08 0.16 0.11

13.2 11.9 12.2 9.2

1.6–47.8 – 1.5–44.1 –

P 5 0.020 n.s. P 5 0.024 n.s.

a

Obs, number of observed deaths; Exp, number of expected deaths; SMR, standard mortality rate; MDE, major depressive episode; P-values, two-tailed Poisson test; n.s., not significant.

with children (SMR, 14.3, P 5 0.003, CI95, 2.9– 41.7) died significantly more often from unnatural causes than expected. In patients suffering from recurrent MDE, death rates from all causes and from unnatural causes were similarly increased independently of the marital status and being a parent (Table 4). In patients with single MDE, patients without children died significantly more often from all causes, married patients and patients with children from unnatural causes (Table 4). Patients with bipolar affective illness (n 5 48) and mania (n 5 26) were not further analyzed due to the rather low number of observations. Patients with a middle educational level (n 5 183; SMR, 3.2, P , 0.0001; CI95, 1.9–4.9), housewives or skilled workers (n 5 94; SMR, 3.6, P , 0.001; CI95, 1.8–6.5) as well as patients with apprenticeship (n 5 176; SMR, 2.4, P , 0.01; CI95, 1.8–6.5) died significantly more often than expected from all causes of death and also from unnatural causes (SMR, 22.5–40.1, P , 0.001 or , 0.0001 in each subgroup). In patients suffering from recurrent MDE, the death rates from all causes and from unnatural causes were similarly increased in the subgroups of educational and professional levels as in the whole sample. In patients with single MDE, only patients

with a middle educational level (SMR, 24.1, P 5 0.0006, CI95, 5.0–70.7) and people with apprenticeship (SMR, 14.0, P , 0.05, CI95, 5.0–70.7) died significantly more often from unnatural causes. Due to the low number of observations for patients with lowest educational level (n 5 4) and unskilled workers (n 5 12), mortality was not further analyzed in these subgroups.

3.2. Death from natural causes Observed mortality from natural causes (total sample, n 5 10; male, n 5 3; female, n 5 7) was not significantly increased when compared to the expected values from general population (9.7, 4.4, and 5.3, respectively; SMR, 1.0, 0.7, and 1.4, respectively). In all diagnostic subtypes and in all groups of social–demographic features, patients did not die more often from natural causes than expected except childless patients with a single depressive episode (SMR, 4.6, P 5 0.02, CI95, 0.9–13.3). We did not find an association between any combination of social–demographic features and natural death, neither in the whole patient sample, nor in the groups with recurrent or single major depressive episodes.

270

B. Schneider et al. / Journal of Affective Disorders 65 (2001) 263 – 274

4. Discussion The results of the present study confirm previous findings of excess mortality in major affective disorders, especially from unnatural causes (SMR, 25.0). The results also strengthen the view that suffering from major depression is per se the most important risk factor of suicide. Women with affective disorders had a considerable higher risk to die from unnatural causes (SMR, 47.1) than men (SMR, 7.1). Social–demographic characteristics per se did not substantially increase mortality risk from unnatural causes but may confer an additional risk. The frequencies of death from natural causes were equally high as in the general population. Yet, our study has some limitations: limited study population to those patients admitted to a psychiatric hospital, the small number of patients in some subgroups which was associated with a low statistical power to detect associations between mortality and special social– demographic features, prospective study without following psychological autopsy of the died patients, lack of documentation of treatment before death, short follow-up period (which possibly leads to high suicide rate, Angst et al., 1999) and causes of death not confirmed by postmortem examination. The mortality rate in our study was nearly three times higher than expected and death from unnatural causes, especially from suicide, accounted for this excess mortality. These findings are consistent with data from other studies (Tsuang and Woolson, 1978; Murphy et al., 1987; Newman and Bland, 1991). We also found a higher mortality from all causes in women (SMR, 4.30), but not in men. The higher mortality rate in female subjects relative to the general population was attributable to the preponderance of women among subjects who died from suicide (16 women out of 18 suicides). This confirms the findings by Black et al. (1987b) and Brodaty et al. (1997), but contradicts those of other investigators (Bruce et al., 1994; Isometsa¨ et al., 1994a,b; Hoyer et al., 2000) who reported that affective disorders and depression, respectively, increased mortality more in men than in women. Like Angst et al. (1980), we found that patients with affective disorders died from other causes than suicide were significantly older at age at onset and at age at the time of death.

A major quest of this study was to determine whether subtypes of affective disorders were associated with different risks of mortality. The overall mortality rate was increased in all subtypes of affective disorder except mania. The death rate from all causes was highest in patients with bipolar affective disorder (SMR, 4.5) and lowest in patients with a single depressive episode (SMR, 2.1). The mortality rate in the bipolar subgroup is higher compared with the findings of Dalgard (1966) and Newman and Bland (1991), whereas the overall mortality in depressed patients was similar to other studies (Malzberg, 1937, 1953; Newman and Bland, 1991; Surtees and Barkley, 1994; Kouzis et al., 1995). Mortality from all causes was higher in female bipolar patients corroborating the results of Odegard (1952) and Newman and Bland (1991). Contrary to our results, Newman and Bland (1991) have found nearly the same overall mortality rates in depressed women and men. Comparable to other studies (Tsuang and Woolson, 1978; Black et al., 1987b; Newman and Bland, 1991; Hoyer et al., 2000), the present analysis revealed statistically significant excess mortality from unnatural death in all diagnostic subgroups of affective disorder, with highest mortality in unipolar depression. Yet, all previous studies which found higher rates of suicide in unipolar than in bipolar disorder (Perris and D’Elia, 1966; Angst et al., 1979; Dingman and McGlashan, 1986; Black et al., 1987b; Newman and Bland, 1991; meta-analysis by Harris and Barraclough, 1997) did not distinguish between the type of course of unipolar depression and assessed only inpatients. Our findings in depressive outpatients suggest that recurrent depressive episodes confer the highest risk of suicide. Whereas manic episodes seem not to be associated with an particular risk of suicide, depressive states are clearly linked to increased suicidality (Isometsa¨ et al., 1994a). Thus, a patient who suffers from both manic and depressive episodes during the course of illness may spend less time at an increased risk of suicide than a patient who experiences recurrent depressive episodes. Therefore, at least in a short-term follow-up, patients with bipolar affective disorder appear to have a lower risk of suicide whereas long-term follow-up studies show a similar risk for bipolar and unipolar subtype.

B. Schneider et al. / Journal of Affective Disorders 65 (2001) 263 – 274

In addition to differences emerging from diagnostic subtypes, gender seems to affect the risk of suicide independently. Women seem to have a higher risk of unnatural death regardless of the subtype of affective disorder. Contrary to the results of other studies which found male gender to be a risk factor for unnatural death (Barraclough and Pallis, 1975; Modestin and Kopp, 1988; Henriksson et al., 1993; Isometsa¨ et al., 1994b; Hoyer et al. 2000), we — like Newman and Bland (1991) — found a higher SMR, for unnatural death in bipolar and unipolar women as compared to men. With respect to age, studies on general psychiatric populations have found higher global mortality in the younger age groups similar to our results (Rorsman, 1974; Craig and Lin, 1981; Brook, 1985; Martin et al., 1985a,b; Casedebaig and Quemada, 1989; Zilber et al., 1989; Hansen et al., 1997). However, in the present study excess mortality from unnatural death was highest in patients older than 49 years. In line with the results of Kouzis et al. (1995), but contradictory to the results of previous studies (Robinson, 1962; Dublin, 1963; Hartelius, 1967; Davidson and Philippe, 1983), marriage obviously did not protect from over-mortality. The above mentioned studies, however, are mostly rather old and did not focus on patients with affective disorders. Rees and Lutkins (1967) and Jagger and Sutton (1991) have found that mortality risk is highest after loss of a spouse. Because married people have an increased likelihood to loose a spouse by separation or death, this could be a reason for the high mortality rate in people who were married at index evaluation. Unfortunately, we had no information about changes of the marital status in all suicided patients. Moreover, we took only marital status into account and did not count the increasing number of non-married couples in our society nowadays. In the whole patient sample as well as in patients with recurrent depressive episodes, mortality was increased in both, patients with children and childless patients. Previous studies (Durkheim, 1897; Hoyer and Lund, 1993) found a lower suicide risk in female patients with children. In our study, mortality from unnatural causes was only slightly higher in women with children compared to childless women. However, marital status and existence of children

271

does not tell us much about the protection offered by intimacy with others. Participation in work, social activities as well as friendship and family in a broader sense should be taken into consideration when the quality of the social network is discussed. Finally, the present study did not find an association between low educational level and a high suicide rate. Our finding is consistent with an English study showing the absence of a social class trend in suicide for the years 1951, 1961 and 1971 (Bulusu and Anderson, 1984) and of social–economical status and occupation in different psychiatric disorders (Babigian and Odoroff, 1969; Innes and Millar, 1970). More recently, Kouzis et al. (1995) observed education not as a predictor of mortality, but low household income. But prospective studies about the influence of educational and occupational level on mortality and suicidality in affective disorders do not exist. In summary, the present data are in line with previous studies indicating no substantially, but sometimes significantly, increased risk for patients with affective disorders to die from natural causes (Martin et al., 1985b; Murphy et al., 1987; Zilber et al., 1989; Newman and Bland, 1991; review by Harris and Barraclough, 1998). In the present study, mortality from natural death was insignificantly lower than expected in men and slightly increased in women. However, similarly increased SMRs in men and women were reported (Newman and Bland, 1991; Murphy et al., 1987). Previous studies (Malzberg, 1937; Kerr et al., 1969) suggested that an excess in premature natural deaths in these patients was associated with the effects of long-term institutionalization, substandard care of concurrent medical illness, prolonged periods of extreme psychiatric symptoms, lack of modern psychopharmacological treatment with agitation and malnourishment for long periods (Avery and Winokur, 1976). In the present study, no patient who died from a natural cause had experienced long-term hospitalization. Furthermore, our data indicate that current treatment of affective disorders is not associated with higher risk to die from natural causes per se, although no data regarding the specific actual treatment of the patients were available. Former studies observed a non-significant increased mortality from natural causes in all age

272

B. Schneider et al. / Journal of Affective Disorders 65 (2001) 263 – 274

groups, especially at younger age (Martin et al., 1985b), but did not find mortality from natural causes associated with marital status, nor educational level nor with professional level (Martin et al., 1985b; Babigian and Odoroff, 1969; Innes and Millar, 1970). But, as already mentioned, all these studies comprised different psychiatric disorders, not only affective disorders. Our results showed a slightly higher risk of natural death only in the subgroup of patients with single MDE: childless patients with a single MDE had increased rates of mortality from natural as well as from all causes. Patients with affective disorders did not have a higher risk to die from natural causes than the general population. But the diagnosis of affective disorder per se, especially of recurrent major depressive episodes, is the most important risk factor for suicide. However, other risk factors, e.g. age and female gender, may also be independently or illnessrelated — of relevance for committing suicide. Thus, adequate treatment and prophylaxis, e.g. with lithium in recurrent major depression (Avery and Winokur, ¨ 1976; Barraclough, 1972; Muller-Oerlinghausen et al., 1992; Ahrens et al., 1995; Isometsa¨ et al., 1994a) are necessary. As in almost all cases, the suicides occurred within the first 2 years of follow-up, aftercare including regular appointments in outpatient services and intensive cooperation between psychiatric hospitals and outpatient services, is especially important for at least that period of time.

References Adam, K.S., 1985. Attempted suicide. Psychiatr. Clin. North Am. 8, 183–201. ¨ Ahrens, B., Muller-Oerlinghausen, B., Schou, M., Wolf, T., Alda, M., Grof, E., Grof, P., Lenz, G., Simhandl, C., Thau, K., ¨ Vestergaard, P., Wolf, R., Moller, H.J., 1995. Excess cardiovascular and suicide mortality of affective disorders may be reduced by Lithium prophylaxis. J. Affect. Disord. 33, 67–75. Angst, J., Angst, F., Stassen, H.H., 1999. Suicide risk in patients with major depressive disorder. J. Clin. Psychiatry 60 (suppl 2), 57–62. Angst, J., Felder W Frey, R., 1979. The course of unipolar and ¨ bipolar affective disorders. In: Schou, M., Stromgren, E. (Eds.), Origin, Prevention and Treatment of Affective Disorders. Academic Press, London, pp. 215–226. Angst, J., Stassen, H.H., 1986. Verlaufsaspekte affektiver Psycho¨ sen: Suizide, Ruckfallrisiko im Alter. Vortrag am 7. In:

‘Weissenauer’ Schizophrenie-Symposium Bonn, 5. / 6. Dezember 1986. Angst, J., Stassen, H.H., Gross, G., Huber, G., Stone, M.H., 1980. Suicide in affective and schizoaffective disorders. In: Marneros, A., Tsuang, M.T. (Eds.), Affective and Schizoaffective Disorders. Springer, Heidelberg, pp. 168–185. Avery, D., Winokur, G., 1976. Mortality in depressed patients treated with electroconvulsive therapy and antidepressants. Arch. Gen. Psychiatry 33, 1029–1037. Babigian, H.M., Odoroff, L C., 1969. The mortality experience of a population with psychiatric illness. Am. J. Psychiatry 126, 470–480. Barraclough, B.M., 1972. Suicide prevention, recurrent affective disorder and lithium. Br. J. Psychiatry 121, 391–392. Barraclough, B.M., Bunch, J., Nelson, B., Sainsbury, P., 1974. A hundred cases of suicides: clinical aspects. Br. J. Psychiatry 125, 355–373. Barraclough, B.M., Pallis, D.J., 1975. Depression followed by suicide: a comparison of depressed suicides with living depressives. Psychol. Med. 5, 55–61. Beskow, J., 1979. Suicide and mental disorder in Swedish men. Acta Psychiatr. Scand. 56 (suppl 277), 1–138. Black, D.W., Winokur, G., Nasrallah, A., 1987a. Is death from natural causes still excessive in psychiatric patients? A followup of 1593 patients with major affective disorder. J. Nerv. Ment. Dis. 175, 674–680. Black, D.W., Winokur, G., Nasrallah, A., 1987b. Suicides in subtypes of major affective disorder. A comparison with general suicide mortality. Arch. Gen. Psychiatry 44, 878–880. Bratfos, O., Haug, J.O., 1968. The course of manic-depressive psychosis. Acta Psychiatr. Scand. 44, 89–112. Brodaty, H., MacCuspie-Moore, C.M., Tickle, L., Luscombe, G., 1997. Depression, diagnostic sub-type and death: a 25 year follow-up study. J. Affect. Disord. 46, 233–242. Brook, O.H., 1985. Mortality in the long-stay population of Dutch mental hospitals. Acta Psychiatr. Scand. 71, 626–635. Bruce, L.M., Leaf, P.J., Rozal, G.P.M., Florio, L., Hoff, R.A., 1994. Psychiatric status and 9-year mortality data in the New Haven epidemiologic catchment area study. Am. J. Psychiatry 151, 716–721. Bulusu, L., Anderson, M., 1984. Suicides 1950–82. Popul. Trends 35, 11–17. Casedebaig, I., Quemada, N., 1989. Mortality in psychiatric inpatients. Acta Psychiatr. Scand. 79, 257–264. Cohen, E., Motto, J.A., Seiden, R.H., 1966. An instrument for evaluating suicide potential: a preliminary study. Am. J. Psychiatry 122, 886–891. Craig, T.J., Lin, S.P., 1981. Mortality among psychiatric inpatients. Arch. Gen. Psychiatry 38, 935–938. Dalgard, O.S., 1966. Mortality in patients with functional psychoses. Nord. Med. 16, 680–684. Davidson, F., Philippe, A., 1983. Epidemiologie du suicide. Gazette Med. France 90, 1665–1677. Diekstra, R.F.W., 1989. Suicide and attempted suicides: an international perspective. Acta Psychiatr. Scand. 354 (Suppl), 1–24. Dingman, C.W., McGlashan, T.H., 1986. Discriminating characteristics of suicides. Acta Psychiatr. Scand. 74, 91–97.

B. Schneider et al. / Journal of Affective Disorders 65 (2001) 263 – 274 Dorpat, T.L., Ripley, H.S., 1960. A study of suicide in the Seattle area. Comp. Psychiatry 1, 349–359. Dublin, L.I., 1963. Suicide: A Sociological and Statistical Study. Ronald, New York. Durkheim, E., 1897. Le Suicide. Free Press, New York. ´ T., 1980. Some characteristics of suicide in Japan. Psychiatr. Fuse, J. Univ. Ottawa 2, 89–94. Guze, S.B., Robins, E., 1970. Suicide and primary affective disorders. Br. J. Psychiatry 117, 437–438. Hansen, V., Arnesen, E., Jacobsen, B.K., 1997. Total mortality in people admitted to a psychiatric hospital. Br. J. Psychiatry 170, 186–190. Harris, E.C., Barraclough, B., 1997. Suicide as an outcome for mental disorders: a meta-analysis. Br. J. Psychiatry 170, 205– 228. Harris, E.C., Barraclough, B., 1998. Excess mortality of mental disorder. Br. J. Psychiatry 173, 11–53. Hartelius, H., 1967. A study of suicides in Sweden 1951–63 including a comparison with 1925–1950. Acta Psychiatr. Scand. 43, 121–127. Henriksson, M.M., Aro, H.M., Marttunen, M.J., Heikkinen, M.E., ¨ E.T., Kuoppasalmi, K.I., Lonnqvist, ¨ Isometsa, J.K., 1993. Mental disorders and comorbidity in suicide. Am. J. Psychiatry 150, 935–940. ¨ ¨ von Patienten mit Hewer, W., Rossler, W., 1997. Mortalitat ¨ funktionellen psychischen Erkrankungen wahrend des ¨ Zeitraums stationarer Behandlung. Fortschr. Neurol. Psychiat. 65, 171–181. Hoyer, E.H., Mortensen, P.B., Olesen, A.V., 2000. Mortality and causes of death in a total national sample of patients with affective disorders admitted for the first time between 1973 and 1999. Br. J. Psychiatry 174, 76–82. Hoyer, G., Lund, E., 1993. Suicide among women related to number of children in marriage. Arch. Gen. Psychiatry 50, 134–137. Innes, G., Millar, W.M., 1970. Mortality among psychiatric patients. Scot. Med. J. 15, 143–148. ¨ E.T., Henriksson, M.M., Aro, H.M., Heikkinen, M.E., Isometsa, ¨ Kuoppasalmi, K.I., Lonnqvist, J.K., 1994a. Suicide in major depression. Am. J. Psychiatry 151, 530–536. ¨ E.T., Henriksson, M.M., Aro, H.M., Lonnqvist, ¨ Isometsa, J.K., 1994b. Suicide in bipolar disorder in Finland. Am. J. Psychiatry 151, 1020–1024. Jacobsson, L., Renberg, E., 1986. Epidemiology of suicide in a Swedish county Vasterbotten 1961–1980. Acta Psychiatr. Scand. 74, 459–468. Jagger, C., Sutton, C., 1991. Death after marital bereavement — is the risk increased? Stat. Med. 10, 395–404. Johansson, L.M., Sundquist, J., Johansson, S.E., Bergman, B., 1997. Ethnicity, social factors, illness and suicide: a follow-up study of a random sample of the Swedish population. Acta Psychiatr. Scand. 95, 125–131. Kerr, T.A., Schapira, K., Roth, M., 1969. The relationship between premature death and affective disorder. Br. J. Psychiatry 115, 1277–1282. Kouzis, A., Eaton, W.W., Leaf, P.J., 1995. Psychopathology and mortality in the general population. Soc. Psychiatry Psychiatr. Epidemiol. 30, 165–170.

273

Malzberg, B., 1937. Mortality among patients with involutional melancholia. Am. J. Psychiatry 93, 1231–1238. Malzberg, B., 1953. Further studies in mortality among patients with mental disease. Acta Med. Scand. 227 (suppl), 215–230. Martin, R.L., Cloninger, D.R., Guze, S.B., Clayton, P.J., 1985a. Mortality in a follow-up of 500 psychiatric outpatients. I. Total mortality. Arch. Gen. Psychiatry 42, 47–57. Martin, R.L., Cloninger, D.R., Guze, S.B., Clayton, P.J., 1985b. Mortality in a follow-up of 500 psychiatric outpatients. II. Cause-specific mortality. Arch. Gen. Psychiatry 42, 58–66. Modestin, J., Kopp, W., 1988. Study on suicide in depressed inpatients. J. Affect. Disord. 15, 157–162. Morrison, J.R., 1982. Suicide in psychiatric practice population. J. Clin. Psychiatry 43, 348–352. ¨ Muller-Oerlinghausen, B., Muser-Causemann, B., Volk, J., 1992. Suicides and parasuicides in a high-risk population on and off lithium long-term medication. J. Affect. Disord. 25, 261–270. Murphy, G.E., 1962. Clinical identification of suicidal risk. Arch. Gen. Psychiatry 27, 356–359. Murphy, J.M., Monson, R.R., Olivier, D.C., Sobol, A.M., Leighton, A.H., 1987. Affective disorders and mortality: a general population study. Arch. Gen. Psychiatry 44, 473–480. Newman, S.C., Bland, R.C., 1991. Suicide risk varies by subtype of affective disorder. Acta Psychiatr. Scand. 83, 420–426. Odegard, O., 1952. The excess mortality of the insane. Acta Psychiatr. Scand. 27, 253–267. Perris, C., D’Elia, G., 1966. A study of bipolar, manic-depressive) and unipolar recurrent depressive psychoses. Acta Psychiatr. Scand. 194 (suppl), 172–189. Philipp, M., Delmo, C.D., 1987. The Polydiagnostic Interview PODI. University of Mainz, Germany. Philipp, M., Maier, W., 1986. The polydiagnostic interview: a structured interview for the polydiagnostic classification of psychiatric patients. Psychopathology 19, 175–185. Pokorny, A.C., 1983. Prediction of suicide in psychiatric patients. Arch. Gen. Psychiatry 40, 249–257. Pokorny, A.D., 1964. Suicide rates in various psychiatric disorders. J. Nerv. Ment. Dis. 139, 499–506. Polyakova, I., Knobler, H.Y., Ambrumova, A., Lerner, V., 1998. Characteristics of suicidal attempts in major depression versus adjustment reactions. J. Affect. Disord. 47, 159–167. Rees, W.D., Lutkins, S.G., 1967. Mortality of bereavement. Br. Med. J. 4, 13–16. Rich, C.L., Young, D., Fowler, R.C., 1986. San Diego suicide study. I: young vs. old subjects. Arch. Gen. Psychiatry 43, 577–582. Robins, E., Gassner, S., Kayes, J., Wilkinson, R.H., Murphy, G.E., 1959. Some clinical considerations in the prevention of suicide based on a study of 134 successful (completed) suicides. Am. Public Health 1, 888–899. Robinson, P.I., 1962. Suicide: causes and prevention statistics and public health significance. Postgrad. Med. 32, 154–159. Rorsman, B., 1974. Mortality among psychiatric patients. Acta Psychiatr. Scand. 50, 354–375. Roy, A., 1984. Suicide in recurrent affective disorder patients. Can. J. Psychiatry 29, 319–322. Strakowksi, S.M., McElroy, S.L., Keck, Jr. P.E., West, S.A., 1996.

274

B. Schneider et al. / Journal of Affective Disorders 65 (2001) 263 – 274

Suicidality among patients with mixed and manic bipolar disorder. Am. J. Psychiatry 153, 674–676. Surtees, P.G., Barkley, C., 1994. Future imperfect: the long-term outcome of depression. Br. J. Psychiatry 164, 327–341. Tsuang, M.T., 1978. Suicide in schizophrenics, manic depressives, and surgical controls. Arch. Gen. Psychiatry 35, 153–155. Tsuang, M.T., Woolson, R.F., Fleming, J.A., 1980. Causes of death in schizophrenia and manic-depression. Br. J. Psychiatry 136, 239–242. Tsuang, M.T., Woolson, R.F., 1978. Excess mortality in schizophrenia and affective disorders. Do suicides and accidental deaths solely account for this excess? Arch. Gen. Psychiatry 35, 1181–1185.

Tuckman, J., Youngman, W., 1963. Suicide risk among persons attempting suicides. Public Health Rep. 78, 585–587. Weeke, A., Juel, K., Vaeth, M., 1987. Cardiovascular death and manic-depressive psychosis. J. Affect. Disord. 13, 287–292. Woolson, R.F., Tsuang, M.T., Fleming, J.A., 1978. A method for analyzing mortality data collected in follow-up studies. Methods Inf. Med. 17, 116–120. Zilber, N., Schufman, N., Lerner, Y., 1989. Mortality among psychiatric patients — the groups at risk. Acta Psychiatr. Scand. 79, 248–256.