Psychological Dysfunction
community's mental health. We include a discussion of social indicator research because the focus on disease states that characterizes psychiatric inquiries into the community is. unduly restrictive. Attainment of quality of life is an equally compelling goal for communities and individuals as the more delimited objective of the eradication of disease, and is no more elusive. Research evidence shows newer measures of subjective well-being are not full of caprice, as once was feared, although they certainly are influenced by contextual variables at least as much as measures of mental distress and disorder. We shall expand on these points.
The chapter is organized to first provide a brief review of the current state of our knowl- edge from psychiatric epidemiology, and in going so we note some of the major methodologi- cal problems that confront that research effort. We then tum to a discussion of a subset of social indicators referred to as measures of subjective well-being, and review the promise those measures hold for enriching our knowledge base within community psychology.
PSYCHIATRIC EPIDEMIOLOGY Basic Concepts
Kleinbaum, Kupper, and Morgenstern (1982) have identified four major goals of epide- miology:
1. To describe the health status of populations.
2. To examine the etiology of disease by identifying causative factors.
3. To predict disease occurrence and its distribution in the population.
4. To control the distribution or spread of disease by preventing new cases and eradicat- ing existing cases.
The concern for the welfare of populations, and the emphasis placed on preventive methods, has made this approach among the most widely adopted perspectives among re- searchers within community psychology.
The most common measures of health and illness relied on in epidemiology are estimates of point prevalence, the number of cases of disorder existing within the community at a given point in time, and incidence, the number of new cases over a designated time period. Relative risk for contracting a disease is another common indicator; relative risk is a comparison of the probabilities of becoming ill between groups that differ in some important characteristic. Odds ratios, in particular, have been used to communicate levels of risk within a readily understood conceptual framework; the odds of having a disease is the probability of contracting a disease over a specified time period divided by the probability of not contracting the disease. The relative odds of contracting lung cancer, for example, between smokers and non-smokers is about 10 to 1.
The application of these public health methods to the study of mental health problems is the focus of psychiatric epidemiology. Psychiatric disorders define a special set of disease entities within epidemiology; they pose complex problems in case identification and differen- tiation. Most mental health problems (with the possible exception of manic-depressive disorder and some forms of dementia) lack specific biochemical markers associated with disordered behavior. The distinction between "sick" and "well" is not always clear-cut in functional terms either. The establishment of a set of operational rules to guide the classifica- tion of mental disorders (Spitzer, Endicott, & Robins, 1978) has made psychiatric diagnosis among clinicians more reliable. Even in those circumstances, however, only manifestations of
the disorder are in evidence-manifestations that may arise from a multitude of underlying conditions and may be treated by a number of different methods. Nevertheless, the operational definitions of caseness have greatly improved the reliability of judgments among studies of mental disorder among researchers who have adopted these common definitions.
Of great significance to public health is the heterogeneity in long-term prognosis among those who develop mental disorders. Harding, Zubin, and Strauss (1987) reviewed studies of chronicity in schizophrenia and found evidence of full recovery among 20-34% of those initially labeled chronically ill. Evidence of such variability in recovery has led to an increas- ing number of investigations into the course of the disorder. Special attention has been given to the prediction and prevention of relapse through studies of the patients' families (Falloon, 1988; Vaugh & Leff, 1976), as well as the care and treatment provided by mental health care providers. Early detection of disordered behavior of children has been associated with poor outcomes in some studies (Gersten, Langner, Eisenberg, & Simcha-Fagan, 1975), due perhaps to labeling of the person and the social stigma attached to that label (Horwitz, 1982; Scheff, 1974). Others (Gove, 1980) have argued that when effective treatments are available, early detection should result in beneficial results. It is estimated that one-third of the homeless suffer from some major mental disorder (Levine & Rog, 1990). The treatment of these people and their impact on the communities in which they live has become a major public health issue in recent years due to the growing numbers and visibility of these indigent and often disturbed people.
The counterpoint to these studies of the poor is a recent report from Vaillant and Schnurr (1988) who followed an initially healthy group of 188 college freshman for 45 years. He found that a quarter of them eventually warranted a DSM-III diagnosis during the follow-up years;
nearly half evidenced significant psychiatric impairment at some point, but most recovered.
Risk for severe psychological disturbance appears high even for the healthiest groups; this state is not irreversible provided the person has adequate resources with which to cope with his or her difficulties.
Epidemiological investigations in general, and especially those that pursue etiological evidence concerning mental disorders, cannot isolate a single cause of the disease. Instead, these studies rely on the careful examination of risk factors that may increase the odds of the person developing a disorder, but which by themselves are neither necessary nor sufficient causes of the disorder. Such factors are studied to see if they covary with the disease state, precede the occurrence of the disease in time, and survive critical tests of spuriousness through the examination of other potential explanatory mechanisms.
Research Designs
The most common study design in psychiatric epidemiology is a cross-sectional field survey of a representative sample of a community population. Such a design can accomplish much in the way of describing the population and providing clues to possible risk factors through an examination of the covariates of disordered states. Cohort studies examine groups of people over time who differ in one or more risk factors to determine if there are differences in incidence rates due to initial differences in the risk factors under study. This design is optimal for the testing of causal hypotheses, but is rarely accomplished in psychiatric epidemi- ology, as it is necessary to track very large cohorts in order to detect differences in incidence for disorders that have low base rates. A useful compromise is the retrospective case-control design, where the history of known cases are examined to identify features that distinguish
them from similar controls. Three problems plague this design. First, it is very difficult to distinguish antecedent events from consequences, post hoc. Recall may be differentially affected by the presence of a disorder, for example. Second, often the researcher himself cannot avoid searching for significance in the data set, increasing the probability of chance findings. Third, it is impossible to match cases and controls perfectly, which allows for the possibility that other (unmeasured) differences between cases and controls are responsible for both the disorder and the identified antecedent risk factor. The success of this design depends, therefore, on the development and testing of a limited set of a priori hypotheses concerning etiology, in the veridicality of historical information gathered, and the adequacy of sampling methods for identifying and testing carefully matched controls.
Defining a Case
The most nagging issue in the psychiatric epidemiologic literature has been defining caseness. The issue of caseness centers around three questions: (1) What is the unique subset of affects, cognitions, and behaviors associated with a specific psychiatric disorder; (2) How much symptomatology needs to be reported or observed to warrant the classification of psychologically impaired?, and (3) How long should the symptoms persist in order to judge them to be associated with an underlying pathological state of the person and not due to a transitory reaction to a stressful life circumstance? When there is no agreement as to how these questions should be answered, the consequence is the prevalence rates of mental disorders range widely-from less than 1% to over 60% in studies conducted around the world (Dohrenwend & Dohrenwend, 1969; Schwab & Schwab, 1978). These large variations in rates raised questions regarding both the reliability of measurement and the validity of the criterion employed to define mental disorders.
The problem of reliability has been dealt with by using structured and standardized interview schedules to decrease error variance due to administration and interviewer differ- ences. Also, multiple "expert" raters have been employed to enhance and check reliability.
The issue of validity is much stickier because there is no single agreed upon "gold standard"
for judging mental disorder. Most researchers have little trouble differentiating severely disturbed individuals from those who show no sighs of psychopathology. The problem occurs towards the middle of the mental health continuum, where there are no clear-cut distinctions between normal and abnormal behavior.
Most of the early studies, from the 1950s until the mid-1970s, had two common charac- teristics: (1) mental disorder was defined on a continuum, and (2) specific diagnoses were rarely made. In place of a differential diagnosis, level of impairment in functioning was assessed as a common denominator of illness severity. In recent years the zeitgeist has changed, resulting in a move away from global ratings of psychological distress/impairment to an attempt to identify prevalence rates of specific disorders. There has been a concerted effort to reestablish medicine within psychiatry including the use of specific psychiatric labels. The reasons for this are many, including political and economic pressures to demonstrate the scientific basis of psychiatry (Klerman, 1986). The publication of the 1980 Diagnostic and Statistical Manual of Mental Disorders (DSM) (American Psychiatric Association, 1980) provided more objective criteria for making diagnoses, which allowed for reliable classifica- tion of cases and non-cases by diagnosticians. This was a significant methodological advance for investigations of mental disorders as specific disease entities because it provided a standard set of criteria for research, permitting comparison of findings across studies. Whether the
classification of disordered behavior into highly differentiated diagnostic entities is really more than convenient fiction is still a subject of much debate (for example, Mirowsky & Ross, 1989). We shall provide some examples of these difficulties as we review current efforts.
Measuring Psychological Distress: Instrument Development
The type of research instruments employed in psychiatric epidemiologic studies have varied depending on the needs of the investigators. In general, however, there has been a move away from the reliance on clinicians and human decision-making to define caseness and toward the use of trained administrators and computerized scoring. From the MidTown Manhattan study came the Langner 22-item screening inventory (Langner, 1962), one of the most widely used "first generation" measures of psychiatric condition. The measure provided scores on global impairment in functioning (for more recent advances in this approach see National Institute of Mental Health, 1985). Since that time a "second generation" of invento- ries have been developed that have numerous symptom dimensions, such as somatization, depression, and psychoticism. Another refinement is their inclusion of intensity measures, which allows the respondent to report his or her subjective level of symptomatic distress. The questions on these newer inventories tend to be more specific and allow for less interpretation on the part of the respondent. The SCL-90-R (revised) (Derogatis, 1977) represents one of these second-generation screening inventories. One of its most attractive features is that it was developed on psychiatric outpatients and focuses on measuring neurotic symptomatology. The nine symptom dimensions of the SCL-90-R are somatization, obsessive-compulsive, interper- sonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psy- choticism. Each of the 90 items can be rated on a fi ve-point scale ranging from "not at all" to
"extremely. "
Questions have been raised by researchers as to whether these screening inventories actually measure distinct forms of psychological upset. Dohrenwend and his colleagues (1980) have argued that most screening inventories assess demoralization, a non-specific form of dis- tress (see Frank, 1973). Dohrenwend et al. (1980) contend that demoralization is the psycho- logical equivalent to temperature in assessing one's physical condition. An elevated score tells you something is wrong, but not what it is.
THE EPIDEMIOLOGIC CATCHMENT AREA PROGRAM
Development of the Diagnostic Interview Schedule
This dissatisfaction with general impairment measures, and the desire to advance knowl- edge of rates of specific mental health problems using present psychiatric nomenclature led to the development of a new and comprehensive research instrument: The NIMH Diagnostic Interview Schedule (DIS). This set of measures was developed by Robins, Helzer, Croughan, and Ratcliff (1981) to collect data for a large multicommunity survey project entitled the Epidemiologic Catchment Area (ECA) program. The ECA program was an outgrowth of questions and recommendations made by the 1978 Report of the President's Commission on Mental Health (Regier, Myers, Kramer, Robins, Blazer, Hough, Eaton, & Locke, 1985). The primary objective of the ECA program was to obtain prevalence rates of specific mental
disorders, rather than prevalence rates of global impainnent (Eaton, Regier, Locke, & Tanke, 1981).
Four respected instruments developed during the 1970s were considered for incorpora- tion into the DIS: The Present State Examination (PSE) (Wing, Cooper, & Sartorius, 1974), the Psychiatric Epidemiological Research Interview (PERI) (Dohrenwend et al., 1980), the Sched- ule for Affective Disorders and Schizophrenia (SADS) (Spitzer & Endicott, 1977), and the Renard Diagnostic Interview (RDI) (Helzer, Robins, & Crougham, 1978). The DIS needed to employ DSM-III criteria, make psychiatric diagnoses in the absence of key infonnants or medical records, and be reliably administered by trained lay interviewers in a reasonable length of time. Only the RDI met all of these criteria, so it was developed to meet the needs of the ECA program. The DIS was designed to make lifetime diagnoses using the Feighner criteria (Feighner, Robins, Guze, Woodruff, Winokur, & Munoz, 1972), and make distinctions between current and past diagnoses. The Folstein-McHugh Mini Mental State Examination (Folstein, Folstein, & McHugh, 1975) was used to assess cognitive impainnent.
The ECA Communities
The five sites selected for the ECA program included New Haven, Connecticut (Yale University); Baltimore, Maryland (Johns Hopkins University); St. Louis, Missouri (Washing- ton University); five contiguous counties in the Piedmont area of North Carolina (Duke University); and Los Angeles, California (University of California at Los Angeles). Multiple sites were employed to assess special populations of interest (black, Hispanic, aged, rural, and urban) and to assess the replicability of findings across sites. At each site approximately 3000 community residents over age 18 were interviewed, along with 500 residents of institutions.
Detailed infonnation about the sampling and methodology has been described by Eaton and Kessler (1985).
ECA Estimates of Magnitude of Mental Health Problems
Roughly one in five persons in the previous six months or one in three persons during their lifetime experienced psychological problems significant enough to warrant a diagnosis of a DIS/DSM-III disorder or severe cognitive impainnent. Rates were surprisingly consistent across sites, with the exception of a high rate of phobias in Baltimore. If phobias are excluded, 6-month and lifetime prevalence rates averaged 15% and 25%, respectively (Burnam, Hough, Kamo, Escoban, & Telles, 1987; Kamo et aI., 1987).
These rates are quite similar to the well-known Stirling County and Midtown Manhattan studies conducted during the 1950s. The Midtown Manhattan study's estimated point preva- lence impainnent rate was 23%, while the Stirling County's lifetime impainnent rate was 31.1% (28.7% mild, 2.3% moderate, and 0.1%, severe). In the Stirling County study, 57% of the sample were estimated to be psychiatric cases sometime during their life, but only 24% of the sample were judged to be both significantly impaired and psychiatric cases.
The leading lifetime diagnoses in the ECA sample were alcohol abuse or dependence (15%), phobias (12% with Baltimore site excluded), drug abuse or dependence (6.5%), major depressive episode (6%), dysthymia (3.5%), and antisocial personality (3%). The percentages in parentheses are gross estimates based on averages across ECA sites (Kamo et al., 1987;
Robins et al., 1984). The leading 6-month diagnoses were phobias (8% with Baltimore site excluded), alcohol abuse or dependence (5%), major depressive episode (3%), and drug abuse
or dependence (2%). Dysthymia could not be estimated for 6-month prevalence rates since time of symptom onset was unknown.
Demographic Correlates of Lifetime Disorders Gender
Men had substantially high rates for substance (particularly alcohol) abuse/dependence and antisocial personality, while women had higher rates for phobia and major depressive episode. Trends were found suggesting that women have higher rates than men for dysthymia and panic disorder. Few sex differences were found among the young-those in the 18 to 24 age group (Robins et aI., 1984).
Age
People less than 45 years of age had rates of mental disorder two times higher than persons 45 years of age and older. Substance abuse/dependence and antisocial personality were found to be primarily young person's disorders, while cognitive impairment was most prevalent in those 65 years of age and older (Robins et aI., 1984).
Ethnicity
Ethnic differences were only found at the Los Angeles site, where a large number of Mexican-Americans (n
=
1243) were compared with non-Hispanic whites. Non-Hispanic whites had higher rates of drug abuse/dependence and major depressive episode than Mexican-Americans. Certain age and sex differences were found by ethnic group. Drug abuse/dependence was uncommon among Mexican-American women of any age and for men and women over 40 years of age of either ethnic group. Mexican-American males abused alcohol5lt2 times more frequently than drugs, whereas non-Hispanic whites abused drugs and alcohol in nearly equal proportions. However, recent data suggest an increase in use of drugs among Mexican-American males. Substantial higher prevalence rates among native-born Mexican-Americans in comparison to immigrants have also been found, which suggests that exposure to American majority culture may lead to higher substance abuse (Burnam et al., 1987).Education
College graduates had fewer psychiatric disorders than those with less education, but differences for specific disorders rarely were found across all sites. The only consistent finding across sites was for cognitive impairment, suggesting that the rate of decline in intellectual abilities leading to dementia is slower for those of greater educational attainment.
Urbanization
Overall rates of psychiatric disorder were found to be higher in inner-city areas (Robins et aI., 1984). This finding is consistent with Faris and Dunham's (1939) early work based on hospital admission rates. Rates of substance use disorders, antisocial personality, and cogni- tive impairment were found also to be higher in urban, rather than rural, areas. Only panic disorder was more prevalent in rural/small town areas, according to Robins et al. (1984).