915
Community Health PROSPECTIVE STUDY OF SOCIAL INFLUENCES ON MORTALITY The
Study of Men Born in
1913 and 1923
L. WELIN G. TIBBLIN K. SVÄRDSUDD B. TIBBLIN S. ANDER-PECIVA B. LARSSON L. WILHELMSEN
University of Göteborg, Section of Preventive Medicine, Department of Medicine, Östra Hospital; Department of Medicine, Sahlgrenska Hospital; and Department of Family Medicine, University of Uppsala, Sweden In a study of 989 middle-aged men followed up for 9 years social influences, measured as persons per household unit, home activities, outside-home. activities, and social activities at the baseline examination, were found to be significantly associated with mortality. The association between these variables (except for home activities) and mortality was significant even when age, risk factors for coronary heart disease, and health status measured at the baseline examination were taken into consideration. Social activities may have a modifying effect on life stresses and risk factors associated with mortality.
Summary
INTRODUCTION
Cassel’ and Cobb2hypothesised that social support may be effective buffer or modifier of life stress. They suggested that social activities and psychological coping styles reduce the deleterious effects of stress and hence the risk of disease in
afternoon session, about 1 week apart. 787 (83%) of the 60-year-old men took part in the morning session, and 769 (81%) in the afternoon examination. Out of the 292 50-year-old men (drawn in 1973) 226 (77%) were examined in the morning, and 220 (75%) took part in the afternoon session. In addition to a routine physical examination blood pressure and serum cholesterol concentration was measured at the morning examination, and a questionnaire about smoking habits and alcohol consumption was completed and then checked by the examining
physician. Smoking habits were coded as: never smoked, ex-smoker, smokes 1-4 g/day, smokes 5-14 g/day, smokes 15-24 g/day, and smokes 25 g/day or more (where 1 cigarette 1 g tobacco, 1 cigarillo 2 g, and 1 cigar 5 g). Alcohol consumption was calculated from the answers to the question about the frequency of drinking and the quantity and type of alcohol consumed at any given time.6 The afternoon examination included a detailed questionnaire about social variables. The subjects completed this with the help of a psychologist. They were asked about the number of persons in the household (including the participant), marital status, and health status. For computing purposes answers to the question on health status were graded from 1 (excellent) to 7 (very poor). Activities (over the previous year) were divided into home activities (10 items); outside-home activities (14 items); and social activities (8 items) (table I). The responses, no, occasionally, or often/regular, were given a score of 0, 1, and 2 respectively. The ratings for each question were totalled to give a theoretical range for home activities of 0-20 points, for outside-home activities of 0-28 points, and for social activities of 0-16 points. =
=
=
Statistical Methods Standard correlation coefficients were computed for the social activity variables. Univariate analyses were done with Pitman’s non-parametric permutation test,and for multivariate analyses logistic regression was used. Because of the small number of 50-year-old men the data in the figures were smoothed with logistic regression. All tests of statistical significance were done on crude data.
an
RESULTS
non-specific way. A prospective population study of men born in 1913 was started in 1963 to assess possible associations between hypertension, hypercholesterolaemia, smoking, and alcohol consumption and the development of coronary heart disease (CHD). Some but not all mortality during the first decade of the study was explained by these risk factors.3,4 At an examination in 1973-74 the subjects were asked for
collected from 1973 until the end of 134 (17%) participants in the and 17 60-year-old age group (7%) in the’50-year-old age The with died. follow-up respect to mortality was group 100%.
detailed information about their social connections and activities. Traditional CHD risk factors were also measured. We present an analysis of these data and have examined the relation between social network variables and mortality during 9 years of follow-up.
follow-up, compared
a
METHODS
Study Population A random sample of 855 50-year-old men (all born in 1913) drawn from the population register in Gothenburg, Sweden, entered a prospective study in 1963.5 10 years later we drew a new sample of subjects born in 1913 from the register, including all men from the 1963 sample still alive and living in the city and men who had moved into the area in the meantime. In addition a random sample of 292 50-year-old men (born in 1923) was drawn from the register. All subjects were invited to an examination in 1973-74, and the 2 samples form the study population of this report.
Examinations For each subject the examination was divided into 1
morning and 1
Mortality
1982.
data
were
During this 9-year period
Univariate Analyses 22% of men who
TABLE I-HOME
divorced or widowers died during with 14% of the married men and
were
ACTIVITES, ACTIVITIES OUTSIDE HOME, AND SOCIAL ACTIVITIES
916
Fig 1-%mortality in men born according to number of persons
in 1913 and 1923 and grouped in the household at the baseline
Fig 2-% mortality in men born in according to home-activity score.
1913 and 1923 and
grouped
examination.
15’8% of single
men.
This trend was of borderline statistical
significance (p = 0’ 052). The number of persons per household unit was significantly inversely related to mortality (p<0-001) (fig 1). 25% of the 60-year-old men living alone died, compared with 7% of those living in households of 4 persons or more. The corresponding proportions in the sample of 50-year-old men were 1307o and 3%, respectively. In both groups of men those with the lowest degree of home activities had the highest mortality (p<0-005) (fig 2). The same was true for outside-home activities (p<0’001) (figure 3) and social activities (p<0’0001) (fig 4), for which the gradient of mortality was most pronounced: 27 - 907o in 60-year-old men with the lowest social activity score and 6 - 307o in those with the highest score. The corresponding rates for the 50-year-old men were 13-7% and 2-7%, respectively. For both age-groups combined a six-fold increase in mortality rate was found over the range of social
activity score. 8 social-activity items were analysed separately (table 11). No single variable contributed significantly to mortality in the 50-year-old men, although for most variables those who were never active tended to have a higher mortality than those who were active regularly or often. In the 60-year-old men parties at home, organised sport, visiting friends, and tradeunion meetings all contributed significantly to the social activity/mortality relationship.
Fig 3-% mortality in men born in 1913 according to outside-home-activity score.
and 1923 and
grouped
Neither frequency of drinking nor type or amount of alcohol consumed was related to mortality. Total alcohol consumption, expressed as g alcohol per person per year, was not significantly related to mortality. Intercorrelations There
was a
significant correlation between several of the
social-activity items, and this indicates that people active in field tend to be active in other areas also. The 4 measurements of social network, home activities, outside-home activities, social activities, and number of persons per household unit were also significantly intercorrelated (r= 0-09-0-53). Serum cholesterol and smoking habits were not significantly correlated to these, but systolic blood pressure was inversely correlated to social activities (r=0’07) and number of persons per household unit
one
(r=0- 14). Health Status
as a
Confounder
A poor social network appears to be an important predictor of mortality, but it may be that those who were already ill at the baseline examination tended to be more isolated, due to the illness, and more prone to death. In this study perceived health-ie, the men’s own rating of their health at the baseline examination in 1973, was used as an index of health status. As shown in fig 5 this index was significantly related to mortality
0001) and inversely related to all the network measurements (r= -0’ 10-0. 22)-the the health the lower the activity score. Therefore,
during follow-up (p<0 social worse
Fig 4-% mortality in men born in according to social-activity score.
1913 and 1923 and
grouped
917 TABLE II-INDIVIDUAL SOCIAL ACTIVITIES IN RELATION TO AGE AND MORTALITY DURING
7 YEARS OF FOLLOW-UP
also taken into account outside-home activities, social activities, and number of persons per household, but not home activities, were significantly related to mortality. The risk of dying increased 3-4 times between the low-risk and high-risk groups for social activities and 2’5 times for the other significant network-variable groups (table III). DISCUSSION
In this prospective study mortality was significantly related outside-home activities, social activities (the higher the activity the lower the mortality), and number of persons per household (the more persons the lower the mortality). No significant association was found between marital status and mortality, probably because this variable does not reflect the number of persons in the household unit very well, at least not in Scandinavia. A significant association between home activities and mortality was found with univariate analysis but not with multivariate analysis. The association between social activities and mortality was strong, even when all other factors were taken into account, and this indicates that an to
TABLE III-LEVELS OF SIGNIFICANCE FOR CORRELATIONS BETWEEN
EACH OF
4 MEASUREMENTS
OF SOCIAL NETWORK AND MORTALITY*
*In the first analysis the confounding effects of age and risk factors for CHD blood pressure, serum cholesterol, and smoking habits) were taken into account. In the second analysis the variable perceived health was added to the analysis. tp values given. :f;In highest-risk group compared with lowestrisk groups after adjustment for confounding variables. The highest-risk group was compared with the lowest-risk group in figs 1-4.
(systolic
health status is an important confounder and into account in the analysis.
must
be taken
Multivariate Analyses Home activities, outside-home activities, social activities and number of persons per household unit remained significant predictors of mortality even after taking age and the risk factors for CHD into account. When perceived health was 7 MORTALITY
40-.——————————————————————————————————————————&mdash.;,
Fig 5-% mortality in men born according to perceived health.
in 1913 and 1923 and
grouped
active social life may protect against premature death. We do not know whether the association between social network and mortality applies to women. Data from the Tecumseh Community Health Study8 indicate that the social network may be of less importance for women than for men. The association between social network and mortality may be explained by the "drift hypothesis"9 or the "vulnerability hypothesis".1 Those affected by disability drift downward in socioeconomic status and also lose contact with relatives and friends. Thus, mortality may be more likely, not because of a poor social network but because of poor health. In our study, poor health (as rated by the participant) was related to a poor social network, but it does not fully explain the raised mortality among those with a poor social network since social activities and number of persons per household unit were associated with mortality even after allowing for poor health. Cassel’s vulnerability hypothesis’ indicates 2 types of social processes of importance in disease aetiology. The first is dominated by deleterious or stress factors which enhance vulnerability to disease. The second consists of protective factors which buffer the organism from the effects of noxious stimuli. This study concerns the second type, which we have called the social network. A strong social network seems to protect middle-aged men from premature death and our data support Cassel’s hypothesis. Similar findings have been obtained from other studies. 10-18 Berkman and Symelo reported that people who lacked social and community ties were more likely to die than those with more extensive contacts. The most important protective factors among men were married status and frequent contacts with close friends and relatives (a high "index of contacts"). These factors acted independently of self reported health and habits such as smoking and alcohol consumption. These results are supported by the findings of the Tecumseh Community Health Study.8 We took several potential confounding factors into consideration. Of these health status at the start of the study was the most important. It is hard to measure general health with only a limited number of variables. Perceived health was the best general measure, but it does not take all the differences in health status into account. Alcohol is another possible confounder. There was no significant association between alcohol consumption (as obtained by questionnaire) and mortality possibly because of underreporting by high consumers-those most likely to die.
’
918
However, a strong association between an indirect measure of alcohol consumption (registration by the Temperance Board) and mortality has been reported.3’4 The associations found in this study may be explained by other confounding factors have
which
we
This
study
was
not
measured.
and the
supported by grants from the Swedish Medical Research
of Medicine,
J. The contribution ofthe social environment to host resistance. Am
Cassel 1976; 104: 107-23.
J Epidemiol
2. Cobb S. Social support as moderator of life stress. Psychosom Med 1976; 38: 300-14. 3. Wilhelmsen L, Wedel H, Tibblin G. Multivariate analysis of risk factors for coronary heart disease. Circulation 1973, 48: 950-58. 4. Tibblin G, Wilhelmsen L, Werkö L. Risk factors for myocardial infarction and death due to ischaemic heart disease and other causes. Am J Cardiol 1975; 35: 514-22. 5. Tibblin G. High blood pressure in men aged 50-a population study of men born in 1913. Acta Med Scand 1967; suppl 470.
Dogma Disputed
SIMON CHAPMAN National Health and Medical Research Council (Australia) HEALTH
agencies seeking to maximise the number of people stopping smoking must consider two broad questions. The first concerns effectiveness: is one approach more effective than another in helping smokers stop; do different methods work better with different smokers? Questions of effectiveness are asked by academics and therapists with professional interest in the intrinsic merits of various approaches. The second question asks whether interventions can be applied throughout a community: has an approach the potential to meet the mass demand its successful operation would imply; what staffing levels are necessary to penetrate a target group; are these levels realistic in times of economic recession? If these questions are asked at all, they are asked by health administrators, funding agencies, politicians, and above all by the tobacco industry. The industry’s opposition can be thought of as the litmus test of a method’s likely impact. The industry has gone on record as "fully supporting sensible and effective public education" about smoking and only the naive could see such a statement as anything less than encouragement for orthodox
smoking control activities. The industry has never opposed stop-smoking groups and therapies, but it has actively fought mass health promotion programmes2 and legislative
proposals.3 In 1982, 38% of
men and 33% of women smoked
UK4-a formidable number of people. All cessation
approaches need to be evaluated in relation to this second goal of mass application before they are assessed in terms of their effectiveness. A 5% success rate among 10 000 people is over 333 times
more
efficient than the 30%
success rate
achieved
by group work involving only 50 subjects. It is this latter sort of activity, however, that occupies nearly everyone involved in smoking cessation. *
Present address: Health Education Council, 78 New Oxford
12.
13. Dressler WW. Hypertension and culture change. New York: Redgrave Publishing Co, 1982. 14. Earp JAL. The effects of social support and health professional home visits on patient adherence to hypertension regimens. Prev Med 1979; 8: 155. 15. Kasl SV, Cobb S. Blood pressure changes in men undergoing lob loss; a preliminary report. Psychosom Med 1970; 302: 19-38. 16. Marmot MG, Syme LS, Kagan A, et al. Epidemiologic studies of coronary heart disease in Japanese men living in Japan, Hawaii and California; prevalence of coronary and hypertensive heart disease and associated risk factors. Am J Epidermiol 1975; 102: 514-25. 17. Mueller OP. Social networks: A promising direction for research on the relationship of the social environment to phychiatric disorder. Soc Sci Med 1980; 14A: 147-61. 18. Ruberman W, Weinblatt E, Goldberg JD, Chaudhary BS. Psychosocial influences on mortality after myocardial infarction. N Engl JMed 1984; 311: 552-59.
THE RESEARCH LITERATURE ON SMOKING CESSATION
STOP-SMOKING CLINICS: A CASE FOR THEIR ABANDONMENT
in the
11.
Alcohol consumption in relation to factors associated with ischemic heart disease. A co-twin control study. Acta Med Scand 1974; suppl 567. Bradley JV. Distribution-free statistical tests. London: Prentice-Hall, 1968. House JS, Robbins C, Mentzner HL. The association of social relationships and activities with mortality: Prospective evidence from the Tecumseh Community Health Study. Am J Epidemiol 1982; 116: 123-40. Lawrence PS. Chronic illness and socio-economic status. In: Jaco EG, ed. Patients, physicians and illness. New York: Free Press, 1958. Berkman LF, Syme SL. Social networks, host resistance, and mortality: A nine-year follow-up study of Alameda county residents. AmJ Epidemiol 1979; 109: 186-204. Medalie JH, Goldbourt U. Angina pectoris among 10 000 men. II Psychosocial and other risk factors as evidenced by a multivariate analysis ofa five year incidence study. Am J Med 1976; 60: 910-21. Harburg E, Blakelock EH Jr, Roeper PR. Authority and blood pressure. Psychosom
Myrhed M.
Med 1979; 41: 189-202.
REFERENCES
*
10.
University of Goteborg.
Correspondence should be addressed to K. S., Department Ostra Hospital, S-416 85 Gothenburg, Sweden.
1.
7. 8.
9.
Council, the Swedish National Association against Heart and Lung Diseases, the King Gustav V and Queen Victoria’s Foundation, the Medical Society of
Göteborg,
6.
Street, London WC1N 1AH
Most of the people working in the field are psychologists and the literature is dominated by therapeutic approaches based on individual and small group techniques. A randomly chosen year of the annual bibliography5 from the US Office on Smoking and Health which abstracts all research and commentary connected with the subject, was surveyed to establish this domination. The 1981 volume contained 2055 abstracts. 128 concerned research and descriptions of cessation techniques, but only 4 of these described massreach programmes and a further 4 described strategies that might be adopted by general practitioners and other health workers. Of the 128 abstracts, 33 were doctoral dissertations that included such esoteric procedures as "subliminal stimulation of symbiotic fantasies", "contingency contracting", and "griefwork". The study groups of the reports seldom exceeded 100 self-selected subjects. The literature on smoking cessation is biased towards therapeutic interventions modelled on psychological concepts of smoking as a behavioural or dependency problem. SMOKING CESSATION GROUPS AND CLINICS: THEIR CONTRIBUTION TO REDUCING SMOKING PREVALENCE
Raw and Heller reviewed6 the functioning of all 55 stopsmoking clinics in the UK in 1983. There are considerable difficulties in estimating the precise contribution of these clinics to a national reduction in smoking prevalence. Only 19 (36%) of the clinics had usable data and none had kept records of follow-up essential for determining the permanency of cessation. These 19 had a total annual throughput of 2390 smokers and a total abstinent population after one year of 560 (mean 23% and a range of 14-43%). If this figure were trebled to account for success rates from the remaining 36 clinics (probably a very liberal assumption), a total emerges of approximately 1680 people who appear to have stopped smoking following attendance at cessation clinics and groups in one year. How does this figure compare with the potential number of smokers who might be described as ripe to stop? When 1711 British smokers were asked "How determined are you to try and give up smoking?", 42% answered positively. If this figure were at all indicative of the proportion of smokers actively thinking of stopping, then 8 -3