Mortality and Causes of Death in a 10-Year Follow-Up of Patients Treated for Self-Poisonings in Oslo

Several follow-up studies after parasiticide show that these patients are at increased risk of suicide and other causes of death. The suicide rates vary from 1% (Rosen, 1970) to 13% (Ettlinger, 1964). The different suicide rates may be due to the different selection procedures. Selections are based on the seriousness of the suicidal intention (Rosen, 1970); different methods of parasiticides (Hawton & Fagg, 1988); self-poisonings only (Paerregaard, 1975; Rygnestad, 1988); and patients from general hospitals (Ettlinger, 1975; Hawton & Fagg, 1988); psychiatric units (Boehme, 1984; Bratfos, 1971; Cullberg, Wasserman & Stefanson, 1988; Retterstøl, 1970; Rosen, 1970), or poison units (Greer & Bagley, 1971; Paerregaard, 1975). The follow-up periods vary from 1 year (Rosen, 1970) to 31–42 years (Dahlgren, 1977). Most long-term follow-up studies seem to imply a cumulative mortality of 5–13% from suicide (Boehme, 1984; Cullberg, et al., 1988; Dahlgren, 1977; Ettlinger, 1975; Nordentoft, Breum Munck, Nordestgaard, Hunding, & Bjældager, 1993; Nordentoft & Rubin, 1993; Retterstøl, 1970; Rosen, 1976; Rygnestad, 1988).

In a previous study we presented the causes of death in an unselected 5-year follow-up of all patients treated for self-poisoning in the medical departments in Oslo during one year (Ekeberg, Ellingsen, & Jacobsen, 1991). The study of all self-poisonings in Oslo in 1980 was one of the first studies of unselected 1-year material from an entire large city (Jacobsen, Frederichsen, Knutsen, Sørum, Talseth, & Ødegaard, 1984a, 1984b). The mortality rate was 10% for females and 16% for males, and 4% of both sexes had committed suicide after 5 years (Ekeberg, Ellingsen, & Jacobsen, 1991). The excess mortality rate was highest for the female opiate abusers (63 times greater than expected). In the present study we have conducted a follow-up of the same population after 10 years. The aims of the study were to determine: (1) the 10-year mortality rate for all patients admitted to medical departments for self-poisoning in Oslo in 1980, (2) the causes of death, (3) factors that might discriminate those who died in the follow-up period from those who were alive, and (4) factors that might discriminate between suicide and other causes of death.

METHODS AND SUBJECTS

All patients discharged following treatment for self-poisonings in 1980 in the medical departments in Oslo (454,000 inhabitants) were included in the study (Jacobsen et al., 1984a, 1984b). Patients leaving the emergency room after treatment were also included. The sample was thus unselected and included all self-poisonings from drugs, narcotics, alcohol, or combinations regardless of suicidal intention. There were 952 subjects who had a total of 1125 admissions (incidence 2.8/1000). As 26 were foreigners or emigrants who could not be traced, 926 were included in the follow-up study. The incidence was highest in age group 20–29 years for both sexes, and 44% of the females and 50% of the males were younger than 30 years. As seen from Table 1, 63% of the males and 34% of the females were drug abusers, and for both sexes mostly of alcohol. Classification of abuse was based upon patient interviews and records. The basic criteria were information of daily or regular use of the respective compound, including withdrawal symptoms if the compounds were not administered. For the classification of suicidal attempt, emphasis was particularly laid on measures taken by the patient to ensure a lethal outcome, such as ingesting a supposedly lethal dose without making contact with others shortly after or the writing of a farewell letter. If the patient made contact with others after the self-poisoning or otherwise behaved in a way that made a lethal outcome unlikely, the act was considered a “cry for help.” The rate of suicidal attempt was highest among the nonabusers, 25% (Table 2).

At follow-up, all subjects were checked according to the official register by the Norwegian Central Bureau of Statistics (CBS). The causes of death were classified according to the death certificates provided by the CBS. These were based on forensic autopsy records in all cases that were not due to natural causes. The quality of this statistic in the classification of suicides has been evaluated previously by our group and was found satisfactory (Ekeberg, Jacobsen, Enger, Frederichsen, & Holan, 1985).

STATISTICAL ANALYSIS

The associations of the outcome variables death and suicide, and the explanatory variables age, gender, social group, suicidal intent, and substance abuse were determined by logistic regression modeling. We used a stepwise backwards procedure, in which variables of explanatory significance above 0.10 were excluded from the models. Age was classified in 3 levels: 15–29 years, 30–49 years, and 50 years or above. Cases with unclassified or missing data were excluded where applicable. Estimates of odds for death or suicide, which closely approximate relative risk for the event, were considered significantly different from the reference group when the 95% confidence intervals (CI — presented in parentheses) did not include 1.0. Goodness of fit for the final models was acceptable (p > .10) for death and suicide. The number of cases in the final modeling step was 869 for death and 896 for suicide in the 0- to 10-year period and 751 and 781 in the 6- to 10-year period, respectively. Since the models for years 0–10 and 6–10 were different from those describing deaths and suicides in the 0–5 year period (Ekeberg et al., 1991), we reanalyzed the data of the latter period after removal of cases that were excluded from analysis in the two former models because of incomplete information in years 6–10. The reanalyzed models of death and suicide for years 1–5 were the same as previously reported, and the odds estimates were not significantly different and are therefore not reported. Confidence intervals for mortality and suicide rates were based on Poisson distribution: CI 95% = n ± 1.96 x√n. For comparisons between proportions, Chi-square with the Yates correction was applied, and the level of significance was set at p < .05.

RESULTS

Mortality

As seen from Table 3, 22% died during the follow-up period. The mortality rate was higher among the males than among the females (P < .001). The mortality rate was particularly high among the substance abusers. Five percent of both sexes had committed suicide.

Table 3. Mortality Rate (MR) According to Substance Abuse in the 10-Year Follow-up Period

	Females		Males		Total
	n	MR (%)	n	MR (%)	N	MR (%)
No abuse	306	13	170	20	476	15
Alcohol abuse	74	26	179	31	253	28
Opiate abuse	42	12	76	31	118	25
Drug abuse	30	32	21	33	51	33
Other abuse	14	21	14	29	28	25
Average	466	17	460	28	926	22

The most common cause of death was suicide, comprising 29% of the female and 16% of the male deaths (Table 4). Of the females who committed suicide, 65% did so by another overdose, 17% by hanging, and 17% by other methods. The corresponding figures for males were 48%, 24%, and 29%, respectively. For the females, 63% of the deaths were related to suicide, abuse, or accidents, compared with 60% for the males. The suicide rate was highest during the first 2 years after discharge, and 39% of the female and 48% of the male suicides occurred during this period (Fig. 1). For deaths from nonsuicides, there were no significant differences according to the time from self-poisoning to death.

Table 4. Main Causes of Death in 10-Year Follow-up According to Sex (%)

	Females (n = 78)	Males (n = 129)	Total (N = 207)
Suicide	29	16	21
Opiate abuse	9	19	15
Heart disease	12	20	17
Accidents/wounds	12	13	13
Alcohol related	13	12	12
Lung disease	13	2	6
Cancer	9	3	5
Others	4	14	10
Total	101	99	99

• Note. The mortality rate is higher among males (p < .001). Suicide is a more common cause of death among females (p < .001).

The mortality rate was four times greater than expected for both sexes, and particularly high for the opiate abusers (Table 5). In the first 5-year period, 10% of the females and 16% of the males died. Of those who were alive after 5 years, 7% of the females and 15% of the males died in the second 5-year period.

Table 5. Excess Mortality According to Substance Abuse, Suicidal Attempts, and Sex

	Females	Males
Opiate abusers	35 (7–64)	24 (14–33)
Drug abusers	7(3–11)	4 (2–7)
Alcohol abusers	8 (5–12)	4 (3–5)
Suicidal attempts	3 (2–4)	4 (2–6)
Nonsuicidal nonabusers	2(1–3)	2 (1–3)
Average	4 (3–5)	4 (4–5)

• Note. Observed; expected numbers, 95% confidence interval, are given in parentheses.

The females had an excess suicide rate of 182 (36–327 95%CI) in the first year after the self-poisoning and 61 (36–87, 95%CI) in the total period. The corresponding figures for males were 70 (19–122) and 21 (12–30).

DISCUSSION

The main finding of the present study was the high mortality rate of 22% (17% among females and 28% among males) in the 10-year follow-up period in the unselected sample of patients treated for self-poisoning in the medical departments in Oslo. The excess mortality rate in the first 5 years of the follow-up period was 8 for both sexes (Ekeberg et al., 1991), which is twice the figure for the total period. This is partly because the expected mortality rate increases as people get older. This is, however, not the total explanation. The mortality rate for the females was lower in the second period, whereas it was the same for the males. For both sexes, however, there was a highly increased mortality rate in the second 5-year period, too. However, for both sexes, and particularly for females, the prognosis was somewhat better in the second period.

The suicide rate was particularly high in the first 2 years of the follow-up, comprising almost half of the suicides. This is a further confirmation that the suicide risk is particularly high shortly after self-poisonings (Nordentoft et al., 1993; Soukas & Lønnqvist, 1991). This calls for improved suicide prevention measures after self-poisonings.

The proportion of deaths from suicide or substance abuse or accidents was 66% in the first 5 years of follow-up and 61% after 10 years. This reflects that the likelihood of dying from natural causes increases with age, and that the majority of opiate abusers admitted for self-poisoning are at a very high risk of dying. Most of these deaths are from accidental overdose with rather low concentrations of opiates (Filseth, Fossen, Halvorsen, Hjelle, Østheim, Sortebogen, Teige, & Ekeberg, 1991), indicating that the deaths were mainly unintended. In the first 5-year period, the female opiate abusers had the highest excess mortality rate (Ekeberg et al., 1991). In the second period, there were no statistically significant differences between male and female opiate abusers. This finding, as well as the lower total female mortality rate in the second period, may indicate that the females more than the males have been able to change their way of living in a more life-preserving direction.

Our finding that death rate was particularly high during the first years after discharge confirms other studies (Hawton & Fagg, 1988; Nordentoft et al., 1993; Soukas & Lønnqvist, 1991). We have previously (Ekeberg et al., 1991) suggested that this may imply that the subjects are at particular risk of suicide shortly after self-poisonings. We also suggested that, as a consequence of the deaths, the proportion of those who are at particularly high risk in the remaining population decreases as time passes. If so, a further decline might have been expected during the follow-up period. The suicide rate was rather stable during the last eight years of the follow-up period. This indicates that the proportion of patients who are at risk of committing suicide has reached a steady state that is considerably higher than in the general population.

Both the mortality rate (31%) and suicide rate (11%) were much higher than in our study in a 10-year follow-up study from Copenhagen from the same time period (Nordentoft et al., 1993). It might be suggested that this is associated with the fact that the suicide rate in Denmark is about twice as high as in Norway (Retterstøl, 1993). The suicide rate in Finland, however, is about the same as in Denmark, and a 5-year follow-up study from Helsinki (Soukas & Lønnqvist, 1991) had a much lower mortality rate (9%) and suicide rate (3%) than in the study from Copenhagen. Therefore, the prognosis after self-poisoning is not necessarily related to the total suicide rate in the country. This raises the question whether patient characteristics or aftercare procedures may differ between Nordic capitals.

The high mortality rate among the opiate abusers reflects a major health and social problem. In an 11-year follow-up study of 300 opiate abusers from Copenhagen, 26% were dead (Haastrup & Jepsen, 1988), which is comparable to our 25% mortality among the opiate abusers. This is a further confirmation of the finding of very low psychosocial functioning in opiate abusers that are admitted to medical departments in Oslo (Rostrup, Ekeberg & Enger, 1984).

In conclusion, our main finding of the high mortality rate of 22% in a 10-year follow-up of a group of rather young patients admitted for self-poisoning calls for improved aftercare.