Are defined as places where people are isolated from the rest of society for a period of time and come under the complete control of the officials who run the institution?

The few existing studies of the prevalence of social isolation indicate that a total absence of relationships is relatively rare for the elderly. 23 Furthermore, when social isolation is identified as a condition of older individuals, the phenomenon is generally accepted as the continuation of a lifelong pattern rather than a development of late life. However, elderly individuals, perceiving themselves to be frail and dependent, may isolate themselves to disguise their loss of autonomy. The limited empirical evidence available on the prevalence of social isolation 1 , 46 seems to corroborate a low prevalence of true social isolation. Theoretical as well as empirical work on the effect of social isolation makes use of social support as an indicator of the degree of social isolation.

Although evidence of total isolation among older individuals may not exist, there has been work suggestive of the existence of relatively low levels of social support for some elderly. In a follow-up study of research involving a community population, 35 percent of the older individuals surveyed reported that they had no confidant (a confidant being defined as someone with whom to discuss serious problems, who was easily available, and with whom there was at least monthly contact). 17

The very early, almost primitive level of definition, measurement, and hence attribution of specific risk to this factor in disease and its outcome is undoubtedly the basis for the lack of estimates of the cost of social isolation, although related areas have been studied. There have been rough estimates made of the cost to families and other care givers for specific diseases, notably Alzheimer's disease; 16 in addition, there have been efforts made to analyze the cost-effectiveness of prevention programs for older individuals. The general conclusion regarding most of this work is that the methods have not been rigorous and that this area requires the attention of carefully designed and scientifically managed research. 2

The theoretical base as well as empirical work on social isolation and disease has not been consistent in differentiating prevention of occurrence (primary prevention) and the various levels of morbidity and mortality (secondary and tertiary prevention). One seminal theory, linking social isolation and resultant stress to weakened host resistance, suggests a nonspecific effect. This tendency is confirmed in the studies reported below on all-cause mortality rates. The studies suggest effects at all stages, pre- and post- impairment, disability, and handicap.

Although specific investigations do not always fit neatly in the various stages of prevention, the work on lack of social support and its effects is a singularly apt illustration of this broader scope in prevention. There have been both empirical investigations and the building of theoretical constructs centering on the strength of social supports in preventing the occurrence of disease, that is, primary prevention. However, much attention has also been focused, particularly for the older population, on secondary or tertiary prevention when the population of interest includes a large percentage of individuals with existing disease. Often, this means a chronic condition that will be a characteristic of the individual to the end of his or her life; therefore, the focus on impairment, disability, and handicap is an appropriate model. In this way a broader, more inclusive strategy can be developed to analyze the effect of social supports (or of any risk factor). The paradigm-like proposition derived from the World Health Organization's (WHO) efforts to achieve standard definitions of disability was used by one of the working groups of the Public Health Service's Health Objectives for the Nation: Year 2000. 33 The framework suggested by the group was that of a hierarchy at the disease end of the spectrum of health status, beginning with the development of disease and leading to impairment, disability, and finally handicap, in that order. The positive end of this spectrum begins with independence, leads to productivity, and ends with life satisfaction. This framework is the background against which impairment, disability, and handicap are considered.

The mechanism by which social isolation, one strong source of stress, contributes to morbidity and mortality affords insights into specific outcomes in the course of disease. Much research on social isolation points toward a nonspecific effect of this factor on health status. Rahe, 35 for example, notes similarities between the non-specificity of recent life change risk factors and coronary heart disease risk factors, which can come from a variety of sources. Early theorizing regarding causal pathways that link stress to biological changes was pioneered by Cassel. 9 He described stress in epidemiologic terms as affecting host resistance and thereby increasing susceptibility to disease. Cassel did not specify the constituents of stress nor identify specific diseases as a result. Rather, he hypothesized a generalized weakening of the host that makes the individual susceptible to insults. Commenting on Cassel's hypothesis, Reed and coworkers 37 assert that it is unlikely that specific social processes are inherently stressful to most people in most places and that it is more likely that the individual reacts differently to the situation depending on perception of the situation, personality, prior experience, and means of coping. However, this type of approach would still have consequences for the occurrence of disease and for subsequent impairment, disability, and handicap.

Kasl 26 describes the stages of disease subsequent to stress using as a framework the epidemiologic schema that describe the spectrum of health through the natural history of disease development. These stages are as follows: (1) asymptomatic status, risk factor absent; (2) asymptomatic status, risk factor present; (3) subclinical disease susceptible to detection; (4) initial symptom experience; (5) initial event (diagnostic criteria for a disease are met); (6) course of disease; (7) institutionalization; and (8) mortality. He observes that the very broad concepts and general theoretical formulations about underlying processes make it more difficult to identify optimal points in the overall causal matrix at which to consider prevention and intervention.

The state of the art in measurement of social supports has not reached the point where there can be differentiation of the stage of disease at which a lack of social supports has the greatest impact.

Several salient questions illustrate these limitations:

Are the factors that relate to the initial onset of disease the same as those that affect the course and outcome of disease?

How are existing biological risk factors and the structure of social supports representing psychological risk factors related in the overall disease etiology?

Will social isolation or other psychosocial stressors occurring early in life affect outcome in the same manner as they would if they occurred later in life?

The preceding discussion reflects the early state of the art regarding the role of social isolation in disease. Because of this limited understanding, a precise assessment of the burden occasioned by this risk factor and the potential for prevention is, of necessity, tenuous. An examination of theoretical and empirical work is essential to begin increasing the accuracy of burden assessment. This approach will serve as well as a background for the discussion of preventability that follows.

The role in health of such phenomena as social support and social isolation has its roots in early consideration and theoretical formulation of the mind/body controversy. Despite the hazy and at times unscientific approaches to this area of human functioning, the role of psychosocial concepts in disease and health demands attention. Eastwood described the beginnings of “psychosomatic medicine,” which addressed the mind/body paradox, as an attempt to identify psychological variables that promote diseases. 15 This early “psychosomatics” movement was reflected in a 1964 WHO report, 49 which concluded that the relationship of mind and body was a dynamic one and that the human system can be affected by either psychological or physiological insult and stress. Over the years, and particularly during the 1960s and 1970s, a significant corpus of work emerged in this area. 13 , 20 , 44

This early, pioneering approach evolved into later large-scale studies to test the role of social supports in mortality and morbidity. This trend began in the mid-1970s and is typified by the studies of House and colleagues, 22 Berkman and Syme, 3 and Blazer. 4 Later research examined prevention strategies designed to delay the onset of disease, affect the early detection of disease, contain the course of functional impairment in the presence of disease, and maintain the highest quality of life in the face of impairments and disability as the result of disease. (This approach clearly addresses all of the stages of the WHO classification and the framework of the Health Objectives for the Nation working group.) The three prospective cohort studies noted above 3 , 4 , 22 showed higher rates of mortality from all causes for socially deprived older persons. These studies used different measures of social support and varying time intervals over which the effect took place. Yet the analysis in each is convincing in linking the absence of social support to higher mortality rates.

Berkman and Syme 3 reported on a nine-year follow-up study of a random sample of 6,928 adults in Alameda County, California. They found that people who lacked social and community ties were more likely to die in the follow-up period than those with more extensive contacts. Their findings also indicated that the “association between social ties and mortality was . . . independent of the self-reported physical health status at the time of the 1965 survey, year of death, socioeconomic status, and health practices such as smoking, alcoholic beverage consumption, obesity, physical activity, and utilization of prevention health services as well as a cumulative index of health practices” (p. 186). Social support in this study was defined as the presence of a spouse, contact with friends, or church and community group membership.

Seventeen-year mortality data from the Alameda County Study were used to examine the relative importance of social ties as predictors of survival at different ages. 43 Comparisons of the relative importance of four types of social ties reveal an interesting shift across the age groups. Marital status assumes primary importance for those aged less than 60 years at baseline. However, ties with close friends or relatives assume greater importance for those aged 60 and older.

A study reported by Blazer 4 of 331 persons aged 65 and older has a similar finding of a higher death rate among persons who were socially deprived. This effect on mortality occurred over a shorter period—30 months. The measurement of social isolation was based on three factors: roles and available attachments, perceived social support, and frequency of social interactions.

A similar finding of decreased mortality risk associated with higher levels of social relationships was reported by House and coworkers, 22 but the finding in this instance was limited to men. The study, the Tecumseh County Health Survey, followed a cohort of 2,754 adults for 9 to 12 years. Trends were similar for women but generally nonsignificant. The measure of social support comprised intimate social relationships, formal organizational involvement outside of work, active and relatively social leisure, and passive and relatively solitary leisure. The findings of these large studies appear to have relevance for each stage of prevention.

There have been other major studies that looked at the relationship of social networks to specific diseases. Haynes, 19 for example, examined coronary heart disease using cross-sectional data from the Framingham study. He showed that, for men and women over the age of 65, marital dissatisfaction or disagreements were significantly related to the prevalence of coronary heart disease. This association occurred only for the older age group (aged 65 and older), suggesting that risk factors may change at different ages. 27

Reed and colleagues 36 reported that prevalence rates for myocardial infarction, angina, and all coronary heart disease were associated with a lack of social network (the study controlled for 12 other known risk factors). The associations, however, were with prevalence of disease rather than mortality and thus draw attention to primary prevention. The study population comprised 4,653 men of Japanese ancestry in northern California and a cohort in Honolulu who were taking part in the Honolulu Heart Program. The measure of social support consisted of structural questions regarding the individual's social network: marital status, closeness of parents, number of living children, number of persons in the household, frequency of social activities, frequency of discussing serious personal problems with coworkers, frequency of attendance of religious services, and number of social organizations attended regularly.

Wortman and Conway 50 reviewed the literature on social support and recovery from illness, the majority of which constitutes studies of recovery from disease in the hospital. In general, Wortman and Conway found that the effect of social support in recovery from disease is less clear than in longitudinal studies over a long period of time with all-cause disease outcome measures. They note that spurious results can be obtained because of the inability to disentangle socially competent and nonneurotic individuals who may have easier access to social support than less socially able persons and who may be more effective in negotiating the health care system. In spite of these dangers, they believe the majority of interventions provide clear evidence that social support facilitates recovery.

Wallston and coworkers 47 believe research evidence supporting a direct link between social support and physical health is more modest than other reviewers have claimed. They categorized studies in terms of the effect of social support on illness onset, the use of health services, adherence to medical regimens, recovery, rehabilitation, and adaptation to illness. (This effort is the most direct parallel to the WHO stages of disability and to various stages and types of prevention.) Wallston and colleagues maintain that these studies fail to distinguish between psychosocial assets as a buffer against the negative health effects of stressful situations and social support as a buffer against illness occurrence. There have been attempts to identify a relationship between social network characteristics and patterns of utilization of health care, but these authors feel the findings from this research are equivocal. In addition, they believe there is evidence for a relationship between social support and adherence to medical regimens. They conclude that there is consistent evidence for positive effects on recovery, rehabilitation, adaptation, and mortality but that it is unclear whether this outcome is due to one type of support or to combinations of support.