© International Epidemiological Association 1999
International Journal of Epidemiology 1999;28:456–460
Printed in Great Britain
Mortality and causes of death among Danish medical doctors 1973–1992 Knud Juel,a Johannes Mosbechb and Eva Støttrup Hansenc
Background To examine the mortality pattern of Danish doctors for the period 1973–1992. Methods
A historical prospective cohort study based on the membership register of the Danish Medical Association. The study population consisted of 21 943 medical doctors, 6012 of whom were women. The doctors’ cause-specific mortality was compared with that of the general population.
Results
The study covered about 277 000 person-years. A total of 2387 deaths occurred from 1 January 1973 to 31 December 1992. The doctors’ mortality was lower than that of the general population. Both sexes showed a standardized mortality ratio (SMR) below one for cancer, circulatory diseases and other natural causes. Mortality due to lung cancer was particularly low. The SMR for suicide was significantly increased, 1.6 for males (95% CI : 1.4–1.9) and 1.7 for females (95% CI : 1.1–2.5). The suicide rate was increased, in particular because of an increased number of suicides by poisoning. In addition female doctors displayed a relatively high mortality due to accidents and other types of violent death.
Conclusions Compared with the general population the doctors’ mortality was low, but the mortality from external causes was increased, mainly due to an excess number of suicides. Keywords
Doctors, suicide, mortality, causes of death
Accepted
21 December 1998
Previous studies on the mortality of Danish doctors have indicated an increased risk of suicide and of violent death.1 Previous studies on Danish doctors indicate that in the period 1935–1975, the doctors’ mortality has decreased relative to that of the general population: standardized mortality ratio (SMR) 1.18 in 1935–1939,2 0.99 in 1955–1959,2 and 0.88 in 1970– 1975.1 The reason for this pattern remains unclear, but an excess mortality from heart disease was noted within the period 1935–1944.2 Studies from other Scandinavian countries covering doctors’ mortality in the 1970s have concurrently shown a total mortality well below that of the general population, and a raised suicide rate.3–5 Studies from countries outside Scandinavia have shown similar results.6–8 In the previous studies, female doctors have been few in number, and possible gender-specific differences in the doctors’ mortality pattern have remained concealed. Since the late 1960s, women have constituted an increasing percentage of Danish
a The Danish Institute for Clinical Epidemiology, Copenhagen, Denmark. b The Danish National Board of Health, Copenhagen, Denmark. c University of Copenhagen, Panum Institute, Department of Occupational
and Environmental Health, Denmark. Reprint requests to: Knud Juel, Danish Institute for Clinical Epidemiology, 25, Svanemollevej, DK-2100 Copenhagen, Denmark. E-mail:
[email protected]
medical students, and today the number of female doctors is high enough to allow for gender-specific analyses of the mortality among Danish doctors. The gender-specificity of the doctors’ health problems is of growing concern because of the rapidly increasing number of women in this previously male dominated profession. The present study was conducted to investigate the mortality pattern among Danish doctors with regard to possible genderspecific differences and possible time trends/cohort effects. Particular concern was devoted to suicide and other types of violent death.
Material and Methods The study comprised members of the Danish Medical Association (DMA) identified from membership files available from 1 January 1973 onwards. Each record contains information about the unique personal identification number, the date of qualification, and the date of joining the DMA. The DMA register can be assumed to be virtually complete, as membership was almost universal throughout the study period; it is now about 94% of actively employed doctors. Almost every doctor joins the DMA immediately after graduation. A total of 86 doctors who joined the DMA after having emigrated were excluded from the study, as were 29 doctors for whom only incomplete data were available.
456
MORTALITY OF DANISH DOCTORS
457
Table 1 Mortality of members of the Danish Medical Association, 1973–1992 compared with that of the general population, by gender and subdivision Males
Females
Observed
SMRa
95% CI
Observed
SMR
95% CI
AJHDb
199
0.76
0.66–0.88
87
1.02
0.81–1.25
AMSc
906
0.73
0.68–0.78
124
0.86
0.71–1.02
OGPd
885
0.85
0.79–0.90
85
0.77
0.62–0.95
1990
0.78
0.75–0.82
296
0.87
0.77–0.97
84
0.83
0.66–1.03
17
1.00
0.58–1.60
2074
0.78
0.75–0.82
313
0.88
0.78–0.98
Subdivision
All subdivisions No subdivision Total a Standardized mortality ratio. b Association of Junior Hospital Doctors. c Association of Medical Specialists. d Organization of General Practitioners.
After the above exclusions, the study population consisted of 21 943 individuals (15 931 [73%] men and 6012 [27%] women). The population comprised 9565 doctors (including 1669 women) alive on 1 January 1973 who were or had been a member of the DMA, and 12 378 doctors (including 4343 women) who had joined the DMA between 1 January 1973 and 31 December 1992. The date of entry into the study was 1 January 1973 for the first group and the date of joining the DMA for the second. The date of leaving the study was the date of death, the date of emigration, or 31 December 1992, whichever came first. All the records were checked in the Danish Central Population Register to obtain information about the vital status of each individual on 31 December 1992: as alive and living in Denmark, dead, or emigrated. Information on cause of death was subsequently obtained from the Cause of Death Register of the National Board of Health. For the study period 1 January 1973 to 31 December 1992 person-years were calculated on the basis of individual entry and exit dates. The Danish general population was used as reference group, and the comparisons were based on gender-, age-, period- and cause-specific mortality rates. The statistical analysis of observed differences was based on the assumption that the observed number of deaths follows a Poisson distribution.9 Internal comparisons were made to evaluate possible differences by medical subdivision, and age at graduation. For these analyses a log-linear model was used in which the observed number of deaths was assumed to follow a Poisson distribution.10 The cause of death was coded according to the 8th Revision of WHO’s International Classification of Diseases (ICD-8). In the analyses, the following diagnostic groups were considered: all causes; cancer (140–209); lung cancer, including malignant neoplasm of the larynx, trachea, bronchi and lung (161–162);
breast cancer (174); circulatory diseases (390–458); other natural causes (000–136, 210–389, 460–796); respiratory diseases (460–519); cirrhosis of the liver (571); symptoms and ill-defined conditions (780–796); violent death (E800–E999); suicide by poisoning (E950); other suicides (E951–E959) and violent death due to injury undetermined whether accidentally or purposely inflicted (E980–E989).
Results The study covered about 277 000 person-years, 24% derived from women and 76% from men. The maximum follow-up time was 20 years, and the average follow-up time 13.4 years. At the end of 1992, 2387 doctors had died (2074 men and 313 women), 1337 had emigrated (1000 men and 337 women), and 18 219 (12 857 men and 5362 women) were still alive and living in Denmark. Data on cause of death were missing for 125 of the 2387 deaths (5.2%). Most of these deaths had occurred outside Denmark. Table 1 shows the total mortality of the study population, distributed by medical subdivision and gender. A significant difference (P , 0.01) was seen between the mortality of male members of the Association of Medical Specialists (AMS) and male members of the Organisation of General Practitioners (OGP). The mortality of the small group of doctors not associated with subdivision did not differ significantly from that of the other groups. As for a large proportion of the deaths in this group information on cause of death was missing, the whole group was excluded from subsequent analyses, which thus comprises 2286 deaths. Among doctors who were members of the DMA before 1 January 1973, the SMR for men was highest for the oldest cohorts, whereas the pattern was less clear for women (Table 2).
Table 2 Mortality 1973–1992 of members of the Danish Medical Association who joined the association before 1973, by gender and year of graduation Males
Females
Observed
SMRa
95% CI
Observed
SMR
95% CI
1017
0.85
0.80–0.90
121
0.93
0.77–1.12
1940–1959
683
0.74
0.69–0.80
92
0.74
0.59–0.90
>1960
176
0.66
0.57–0.77
42
0.96
0.69–1.29
1876
0.79
0.75–0.82
255
0.85
0.75–0.97
Graduation year 1904–1939
Total a Standardized mortality ratio.
458
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
Table 3 Mortality 1973–1992 of members of the Danish Medical Association who joined the association after 1 January 1973, by gender and age at entry Age at entry 35 years Total
Gender
Observed
SMRa
95% CI
M
79
0.57
0.45–0.71
F
32
0.99
0.68–1.40
M
35
1.33
0.92–1.85
F
9
0.92
0.42–1.74
M
114
0.69
0.57–0.83
F
41
0.98
0.70–1.32
Table 4 Mortality 1973–1992 of members of the Danish Medical Assocation, by gender and cause of death Cause of death
Gender
Observed
SMRa
95% CI
M
470
0.73
0.66–0.80
Cancer
F
96
0.86
0.70–1.06
Lung cancer
M
94
0.48
0.38–0.58
F
8
0.53
0.23–1.05
Breast cancer
M
0
0.0
–
F
30
1.29
0.87–1.84
M
376
0.84
0.76–0.93
F
58
0.80
0.61–1.03
M
874
0.76
0.71–0.81
F
86
0.68
0.54–0.84
M
320
0.61
0.54–0.68
F
53
0.75
0.56–0.98
Respiratory diseases
M
81
0.45
0.35–0.56
F
13
0.66
0.35–1.13
Cirrhosis of the liver
M
28
0.60
0.40–0.87
F
8
1.56
0.67–3.07
Symptoms, etc.
M
55
0.60
0.46–0.78
All other cancers
a Standardized mortality ratio.
Circulatory diseases Other natural causes
High age at the time of membership of the DMA (usually by the time of graduation) was related to higher mortality in men but not in women (Table 3). The SMR for female doctors was close to one. Thus, a clear difference is seen between men and women. Compared with the general population, both male and female doctors had a lower mortality for cancer, circulatory diseases and other natural causes. In particular, the mortality due to lung cancer was low (Table 4). The mortality from external causes was significantly increased among both men and women. Among the men, the increase could be ascribed to suicides by poisoning; whereas for the women, both suicides by poisoning, fatal accidents, and other types of violent death occurred in excess. Table 5 shows the mortality from violent deaths by gender and medical subdivision. The overall suicide mortality was significantly increased. An increased frequency of suicide by poisoning was a common feature in all subgroups particularly among male members of the AMS and OGP. Female junior doctors showed increased numbers of accidents and of suicides by poisoning as well as by other methods. We found that poisoning was involved in 17 of the 21 cases in which the death certificate did not state whether the cause of death was accident, suicide or homicide. Some of these deaths were probably due to suicide, since accidental poisoning among doctors ought to be rare. This conclusion is supported by the finding of only three registered poisoning accidents (E850– E859) in the total of 2286 deaths.
F
7
0.56
0.23–1.16
Violent deaths
M
270
1.23
1.08–1.38
F
54
1.71
1.28–2.23
Suicide (total)
M
168
1.64
1.40–1.91
F
26
1.68
1.10–2.46
Suicide from poisoning
M
103
4.35
3.55–5.28
F
17
2.44
1.42–3.90
M
65
0.83
0.64–1.05
F
9
1.05
0.48–2.00
Uncertain if accident, suicide or homicide
M
17
1.30
0.76–2.08
F
6
2.21
0.81–4.81
All other violent deaths
M
85
0.81
0.65–1.00
1.65
1.03–2.49
Other suicides
F
22
No data (abroad)
M
43
F
6
No data (Denmark)
M
13
F
1
Total
M
1990
0.78
0.75–0.82
F
296
0.87
0.77–0.97
a Standardized mortality ratio.
Discussion This study demonstrated a reduced mortality and an increased suicide rate among Danish doctors as compared with the general population. The findings are consistent with those obtained in other studies on doctors’ mortality.6,7,11 Doctors constitute a stable, homogeneous group into which most members enter at around the age of 25–30 and almost nobody leaves the profession before retirement. The ideal control group would be another highly educated stable professional group assumed to be similar to doctors in all aspects, except those directly related to being a doctor. In previous studies, doctors have been compared with the general population, with all occupationally active individuals, with other university graduates, and with certain professional groups such as lawyers and clergymen.1,3,8,12,13 We have chosen the general Danish population
as the reference because of the availability of cause-specific mortality rates. When the ‘expected’ mortality is calculated from the mortality rates of the general population, the SMR of a particular occupational group is likely to be biased in a negative direction as far as total mortality is concerned (the so-called ‘healthy worker effect’). The same is the case for chronically disabling diseases, whereas the SMR estimates for sudden death are hardly affected. For Danish males, the mortality rate generally decreases with increasing social class and educational level, whereas the same is not the case for the women, for whom a high educational level seems to indicate a death risk close to that of the general population.14
MORTALITY OF DANISH DOCTORS
459
Table 5 Violent death 1973–1992 of members of the Danish Medical Association, by cause of death, gender and medical subdivision AJHDa
AMSb
OGPc
SMRd Cause of death Suicide from poisoning
Other suicides
All other violent deaths
Violent deaths, total
SMRd
SMRd (95% CI)
Gender
Observed
(95% CI)
Observed
(95% CI)
Observed
M
29
2.99 (2.00–4.29)
43
5.89 (4.26–7.93)
31
4.65 (3.16–6.60)
F
11
2.61 (1.25–4.80)
3
1.52 (0.31–4.43)
3
2.58 (0.53–7.55)
M
16
0.54 (0.31–0.87)
22
0.85 (0.53–1.29)
27
1.17 (0.77–1.70)
F
8
1.71 (0.74–3.37)
1
0.41 (0.01–2.28)
0
0.00 –
M
27
0.64 (0.42–0.94)
36
0.92 (0.65–1.28)
39
1.06 (0.75–1.45)
F
13
1.80 (0.96–3.07)
11
2.32 (1.16–4.15)
4
0.97 (0.26–2.49)
M
72
0.88 (0.69–1.11)
101
1.40 (1.14–1.70)
97
1.46 (1.18–1.78)
F
32
2.03 (1.39–2.87)
15
1.64 (0.92–2.70)
7
1.04 (0.42–2.15)
a Association of Junior Hospital Doctors. b Association of Medical Specialists. c Organization of General Practitioners. d Standardized mortality ratio.
Danish doctors generally smoke less than the general population,15 a difference that will bias the SMR estimates for smoking-related diseases in a negative direction. As expected, the doctors’ mortality was lower than that of the general population. The lowest mortality (SMR = 0.57) was observed for male doctors who graduated 1973 or later and who were under 35 years old at the time of graduation, while older cohorts of male doctors tended to approach the mortality of the general population. For female doctors the mortality generally did not differ much from that of the general population. The gender differences may be due to differences in the selection of males and females into the medical profession and/or may reflect socially related differences in stress levels. The data do not allow us to conclude whether the gender difference in mortality is a new phenomenon or not. Few women graduated in the earlier period of the century and there are certainly significant differences in the selection of the women who entered the profession at the beginning of the century and those who have entered within more recent years. For males the study demonstrated that members of the OGP had a higher mortality than members of the AMS, a difference that persisted when the data were analysed by period of graduation and cause of death. Similar findings have been reported from Finland and the UK.7,11 In particular, this study showed a higher mortality from lung cancer among the general practitioners than among the members of the AMS, a finding that most likely reflects the higher proportion of daily smokers in the former group compared with the group of hospital specialists.15 In general, the Danish doctors displayed a very low mortality from lung cancer and non-malignant respiratory diseases, a finding that corresponds well with the low frequency of smokers among doctors.15
We found no significantly increased mortality from breast cancer. A raised breast cancer risk is a well-known phenomenon among highly educated women, who typically have a reproductive career characterized by few and late births. The main finding in this study was a consistently increased mortality from suicide with an overall SMR around 1.65, mainly due to an excess number of suicides by poisoning. These findings are consistent with those previously reported.6,8,13,16–18 In addition, our findings document that the female doctors experience a suicide risk that seems to be comparable to that of the male doctors. One earlier study has demonstrated increased mortality due to suicide among Danish female nurses (SMR = 1.67),14 a finding that indicates easy access to drugs and knowledge about their effects and lethal doses to be of importance.18 When dealing with their own illnesses, knowledge about diagnosis and prognosis may also be important for chronically ill doctors. In addition, the combination of strenuous working conditions and a possible self-selection of individuals with particular personality traits may contribute to the doctors’ high risk of suicide. The consistent finding of a high rate of suicide among doctors warrants further investigation in order to identify possible preventive measures. A future study should focus on cases that have committed or attempted to commit suicide.
Conclusion Our study on Danish doctors’ mortality demonstrated a low overall mortality, a very low mortality from a number of smoking-related diseases, and a high rate of suicide mainly due to poisoning. These findings are consistent with those previously reported. Besides, our results demonstrated that the high risk of suicide affects female as well as male doctors.
460
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
The reasons for the doctors’ high suicide rate should be further investigated.
8 Lindeman S, Läärä E, Hakko H, Lönnqvist J. A systematic review on
gender-specific suicide mortality in medical doctors. Br J Psychiatry 1996;168:274–79. 9 Bailar JC, Ederer F. Significance factors for the ratio of a Poisson
variable to its expectation. Biometrics 1964;20:639–43.
References 1 Jeune B. Mortality among practicing physicians in Denmark 1971–
1975. Ugeskr Læger 1980;142:2577–82. 2 Lindhardt M, Frandsen E, Hamtoft H, Mosbech J. Causes of death
among the medical profession in Denmark. Dan Med Bull 1963; 10:59–64. 3 Arnetz BB, Hörte LG, Hedberg A, Theorell T, Allander E, Malker H.
Suicide patterns among physicians related to other academics as well as to the generel population. Acta Psychiatr Scand 1987;75:139–43.
10 Berry G. The analysis of mortality by the subject-years method.
Biometrics 1983;39:173–84. 11 Asp S, Hernberg S, Collan Y. Mortality among Finnish doctors. Scand
J Soc Med 1979;7:55–62. 12 Curnen MGM, Turgeon LR, Valanis B, Skovron ML, Varma AAO,
Fleiss JL. Cancer in Connecticut physicians and lawyers: a study by birth cohorts. Conn Med 1985;49:729–37. 13 King H. Health in the medical and other learned professions. J Chronic
Dis 1970;23:257–81.
4 Borgan JK, Kristofersen LB. Mortality by Occupation and Socio-economic
14 Andersen O. Mortality and Occupation 1970–80. Copenhagen: Central
Group in Norway 1970–1980. Oslo: Central Bureau of Statistics of Norway, 1986, pp.1–84
15 Madsen M, Pedersen JT, Nielsen PE. Smoking habits among Danish
5 Rimpelä AH, Nurminen MM, Pulkkinen PO, Rimpelä MK, Valkonen
T. Mortality of doctors: do doctors benefit from their medical knowledge? Lancet 1987;i:84–86. 6 Carpenter LM, Swerdlow AJ, Fear NT. Mortality of doctors in different
Bureau of Statistics, 1985. physicians, nurses and midwives in 1989. Ugeskr Læger 1990;152: 3336–40. 16 Harrington JM, Oakes D. Mortality study of British pathologists
1974–80. Br J Ind Med 1984;41:188–91.
specialties: findings from a cohort of 20 000 NHS hospital consultants. Occup Environ Med 1997;54:388–95.
17 Boxer PA, Burnett C, Swanson N. Suicide and occupation: a review of
7 Doll R, Peto R. Mortality among doctors in different occupations.
18 Nordentoft M. Suicide among physicians. Ugeskr Læger 1988;150:
Br Med J 1977;i:1433–36.
the literature. J Occup Environ Med 1995;37:442–52. 2440–43.