NAOMAR ALMEIDA-FILHO?
BECOMING MODERN AFTER ALL THESE YEARS: SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
ABSTRACT. This paper takes a critical standpoint, both theoretical and methodological, to revisit Inkeles and Smith’s hypothesis on the association between modernization and mental health. First it is proposed a critical evaluation of the premises of the conceptual treatment of the relationships between social change and mental health prevailing during the past two decades. Secondly, results from epidemiologic research on the psychological outcomes of social development in Latin America are reviewed, emphasizing the methodological improvements which occurred during the past two decades. Selected findings of an epidemiological survey recently conducted in urban Brazil are then presented, focusing on a case-control analysis of the socio-economic correlates of individual mental health. Finally, some of the implications of the new evidence concerning the social change and mental health hypothesis are discussed, as an attempt to interpret these findings in the light of recent developments of theories on social change and health in the contemporary world.
INTRODUCTION About twenty-five years ago, as part of the Harvard Project on the Social and Cultural Aspects of Development, Alex Inkeles and David Smith coordinated an extensive cross-national survey on the social and psychological consequences of the (as called at that time) modernization process in developing countries, reported in a book entitled Becoming Modern (Inkeles and Smith 1974). Concerning mental health, the basic hypothesis being tested was that individual “modernizing experiences” were causally related to psychological maladjustment. As a whole, Inkeles and Smith (1970, 1974) concluded that their results gave virtually no support to the assertion that education, urban living experience, factory work, mass media contact, and individual modernity were regularly associated with increased emotional problems in developing countries. Comparisons between psychosomatic symptom ? M.D., Ph.D., Dean, Instituto de Sa´ude Coletiva, Universidade Federal da Bahia, Brazil. Adjunct Professor, Department of Epidemiology, School of Public Health, University of North Carolina at Chapel Hill, USA. Visiting Professor, Department of Anthropology, Universit´e de Montr´eal, Qu´ebec, Canada. Senior Researcher (I-A) of the Brazilian National Research Council – CNPq.
Culture, Medicine and Psychiatry 22: 285–316, 1998. © 1998 Kluwer Academic Publishers. Printed in the Netherlands.
286
NAOMAR ALMEIDA-FILHO
scores of newly settled migrants and of rural dwellers showed no significant differences. Among factory workers, rural origin was not related to higher psychological stress levels, even after matched comparisons by education and length of urban living experience. Their major finding, however, was related to the economic placement of the migrant worker: “Indeed, taking all the measures into account, moving in itself seems to be neither here nor there with regard to psychic health. Perhaps the critical factor is whether or not your post-migratory status permits you to become integrated into a stable, meaningful, and rewarding [economic] role in your new environment” (Inkeles and Smith 1970: 106). Despite the presence of methodological limitations,1 such as a sampling design which did not include urban unemployed or underemployed subjects, the psychosocial section of the Harvard Study opened a new avenue of research on the social determinants of psychic health. The proposition that, regarding mental health, the economic dimensions of the modernization process are more important than modernizing experiences and individual modernity will be herein designated as “Inkeles and Smith’s hypothesis”. Given the limited scientific production in Latin American psychiatric epidemiology at that time, there was no evidence available to test such a fertile hypothesis on more specific grounds. Indeed, as far as scientific evidence is concerned, Latin American psychiatric epidemiology has recently provided sufficient data to assess this hypothesis more effectively. However, rather than seeing the association of modernization with psychological well-being as a mere set of hypotheses that can now be empirically tested, we shall take this opportunity to challenge the conceptualization behind Inkeles and Smith’s hypothesis, to the extent that their approach was in many ways naive to major issues of political economy. The argument developed in this paper is that, taking a critical theoretical standpoint to revisit the premises of the modernization-mental health hypothesis, we need to place Inkeles and Smith’s contribution in historical perspective, considering overall that their interpretation of the social world was largely biased by the political and ideological context of the applied social sciences of their times. With this aim, I will first present a critical evaluation of the conceptual background of the relationships between social change and mental health prevailing at the time of Inkeles and Smith’s study. Secondly, I intend to briefly review some of the epidemiological research on psychological outcomes of social development in Latin America, emphasizing the methodological improvements which occurred during the past two decades. Then, I will present selected findings of an epidemiological survey recently conducted in urban Brazil, focusing on a case-control
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
287
analysis of the evidence produced on the socio-economic correlates of individual mental health. Finally, I will discuss some of the implications of the new evidence concerning the social change and mental health hypothesis, as an attempt to contextualize these findings into new conceptual models.
SOCIAL CHANGE AND MENTAL HEALTH: CONCEPTUAL BACKGROUND To better appreciate the landmark nature of Inkeles and Smith’s study, we need to consider that at that time the dominant theoretical framing of that question was based largely on two competitive hypotheses: acculturation pressures versus urban stressors, respectively supported by the pioneering findings of the prevalence surveys of Stirling County and Midtown Manhattan.2 What characterized such a context was the scarcity of good quality epidemiological data – understandable by the still incipient stage of methodological development – in line with speculative approaches based on development theories as applied to Third World countries. As a result, one could easily find dramatic depictions of the psychosocial effects of urbanization and industrialization more rhetorical than scientific. Exemplary of such a trend, whose premises have had a long-standing impact on the ways scholars have thought about urban experiences in Latin America, is the following quotation from Argandoña and Kiev’s Mental Health in the Developing World (1972): As in almost all the countries of the world at present, in Latin America man is going through startling changes. Never before has so much been required of the human being. Migration to the city means homelessness for thousands who live in close physical proximity, separated from village ways, in a climate of alienation and insecurity. . . . In the overcrowded cities, the collective atmosphere makes it impossible for the human being to feel at home: he becomes an anonymous element. . . . We still do not know the ultimate effects of these transformations on the human being, but it can be said that present living conditons are harmful physically and psychologically. . . . Deprived of emotional and intellectual satisfactions arising from his belonging to a community, normal adaptation is disrupted and mental and emotional disorder can result. (Argandoña and Kiev 1972: 51–52).
By the time Inkeles and Smith concluded their work, researchers represented the process of social change as a series of factors integrated into a structure of determination thought to converge onto mental illness outcomes. Some of these factors were taken as correlates of modernization, and specific combinations of them were in some cases confounded with the change process itself. Among such factors, the notions of urban stress,
288
NAOMAR ALMEIDA-FILHO
goal-striving stress, life change, and social support could be understood as part of a “sociologism” given their origins in social relations theories. As explanations oriented by a “culturalism” approach, I would consider the notions of nostalgia (loss of culture), marginal man, cultural shock, and acculturation which can all be regarded as consequences or components of the change process itself. Problems in conceptualization Despite the efforts of researchers and thinkers, there has never been a satisfactory agreement about the scope and definitions of the different notions employed to describe the historical changes of human populations and their consequences as social and cultural processes, such as acculturation, assimilation, integration, adjustment, adaptation, modernization, displacement, and the like. Propositions to systematize this conceptual mess have fallen into two broad patterns: (a) taxonomies and classification systems can be made more and more complex, yet using the original single terms; or (b) the authors can start playing what I would call “the game of adjectives.” Some authors contended that “assimilation and acculturation are separate, distinct processes,” and that “acculturation is a necessary, though not a sufficient, condition for assimilation to occur” (Teske and Nelson 1974: 365). For Price (1968), the basic process is the “adaptation” of human beings, which can be analysed in terms of “assimilation” (incorporation into the social fabric), “acculturation” (change of cultural characteristics), and “adjustment” (successful outcome for individual social and psychological health). On the contrary, others proposed that assimilation, integration, separation and “marginality” should be considered as sub-categories and possible outcomes of “acculturation” processes, defined by the cross combination of two dimensions: cultural identity and cultural communication (Berry 1980). In this scheme, assimilation and integration would be examples of successful acculturation. For DeVos (1976), adjustment refers exclusively to the individual personality, while adaptation is defined in terms of the individual’s relationship to society. The debate on such a terminological confusion would become even more complicated if we considered the (now demodé) notion of Westernization. Examples of this kind can be found in the Latin American literature, as in Mariátegui and Samañez’s (1968) description of a “transculturation” process, with the dychotomous consequences of “adjustment” or “marginality” at the cultural level, and of “security” or “uncertainty” at the psychological level. The “game of adjectives” had been played by many authors. Borrie (1959), discussing consequences of immigration, used the expressions
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
289
“economic absorption” and “cultural integration.” Brody (1969) added “psychological adaptation,” forming with the other two a triad of conditions supposed to be protective for the migrant’s mental health. Weppner (1972) extended the number of levels and adjectives, proposing “economic adjustment,” “cultural assimilation,” “linguistic assimilation,” “social integration,” and “psychological adaptation.” Graves (1970) once mentioned a certain “psychological acculturation.” Two-term adjectives were also frequently employed by some authors, and the favorite ones seemed to be “psychosocial” and “psychocultural,” which could be used interchangeably in combination with several of the above mentioned key terms. Examples are “psychocultural stress” (Wittkower and Dubreuil 1973), “psychosocial adjustment” (Barger 1977), “psychocultural adaptation” (Spindler 1968), among others. The problem of naming the notions employed in this area of research is undoubtedly a profitable topic for a semantic analysis which obviously cannot be pursued here. Nevertheless, I would acknowledge that both processes described above – the “complex(class)ification” and the “adjectivation” – are indicative of a more general crisis in the paradigms of the disciplines concerned with the research question of mental health consequences of social change. When the basic notions of a given scientific discourse have problems in becoming concepts (i.e. in “meaning” concrete processes or phenomena), correlative notions may be created to aid them in approaching regions of reality previously uncovered by the notions alone. As these additions will merely mend incomplete representations, and as they also have the fate of suffering from the same original lack of meaning, the additions themselves need further complementation. Often, the result of such a process is a multiplication of definitions and a proliferation of correlative notions. An alternative way to face such a symptom of that paradigmatic crisis can be the use of a “semantic band-aid,” which is made relatively effective through attaching an adjective to a noun. Following this rationale, two-term adjectives thus would be suggestive of a double failure in producing meaning for concrete processes or phenomena through the process of concept formation. Given this background, the ambitious project of Inkeles and Smith had the merit of re-conceptualizing the problem in the light of scientific methodology. They were successful in avoiding such “language traps” by defining research hypotheses and empirical goals with a partial awareness of the limits of the kind of objectivity peculiar to sociocultural research.
290
NAOMAR ALMEIDA-FILHO
A critique of culture change models Acculturation, defined as the forced outcome of geographical moves from a rural traditional context to a modern industrial society, was employed by Cassel, Patrick and Jenkins (1960) in a theoretical system of the health implications of social change. Regarding the basic underlying processes in the model, Cassel et al. assumed that a traditional culture provides the rural migrant with “a design for living quite appropriate to the social situation of the folk community, but a culture adapted to rural life may increase rather than decrease the stresses of the rural migrant to an urban situation” (Cassel, Patrick and Jenkins 1960). The basic coherence of this model was given by the ideas of function (intra structures) and congruency (inter structures but within the social systems), whose deviations would correspond to the notions of acculturation and adaptation. For these authors, Leighton’s social disorganization hypothesis (Leighton 1959) is not at all applicable to the study of health consequences of social change because the modern, industrial context “is not simply ‘disorganized’ and ‘secularized’ [. . . but] it is highly organized on different principles” (Cassel, Patrick and Jenkins 1960: 946). For this model, social change may lead to cultural incongruency, depending upon the “goodness of fit” between the migrant’s culture and his/her actual social situation. Such incongruities, in turn, tend to arouse excessive stress on the individual’s affiliative network and on his/her psychic apparatus that may or may not be absorbed by either system. Further developments of the original scheme proposed “social support” as a mediating variable related to the capability of stress absorption by the individual’s social network (Kaplan, Cassel and Gore 1977). The basic hypothesis therefore was that the non-absorption of “sociocultural stresses” aroused in the social change process is associated with the onset of psychiatric and psychosomatic symptoms. Implicit in these conceptual models were hypotheses of distinct processes related to social change considered as risk factors for mental ill/health, for example, the uprooting process, migration (moving and re-insertion), cultural marginality, the hostile urban environment as a social stressor, overcrowding, anomie and social isolation. The idea of a traditional-modern cultural continuum, at the basis of such models, can be subjected to several general criticisms. Overall, such a notion represents a highly ideologized picture of traditional social contexts, as if they were “societies,” “cultures,” or “subcultures,” as homogeneous as the culture concept would expect, insofar as those contexts are thought to be formed by people who share common “cultural traits” and common social goals. To accept this perspective means to deny the existence of complex relationships of exchange,
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
291
exploitation and domination, as well as the heterogeneity, diversity and conflict inherent in any social context. Let us call this an “illusion of homogeneity.” How did such a conceptual bias operate in the social change and mental health theoretical frameworks? Initially, the status of non-modern, rural, migrant, poor, unemployed was taken as an attribute of an individual or a group, which would define a higher or lower risk of being mentally ill. In that body of research, it did not matter whether or not the unemployed migrant had been a member of the exploited classes of his/her place of origin who took another subordinate position in the social context of destination. To say that in one sentence: they tended to consider that “all rural, migrant, non-modern people are alike.” But such an interpretation is obviously fallacious, so absurd that it is amazing how influential it has become in this area of research. A more plausible approach, the second illusion of homogeneity, was sometimes proposed: one could admit internal inequalities among the underprivileged study subjects, but inequalities that were not related to the larger economic, social and political context. These inequalities were then turned into heterogeneities, that could be either naturally or normally distributed, implying a homogeneous distribution of chances to be different. For example, the researcher might have been concerned with the question of whether a migrant is of rural origin (maybe from another “culture”) or had moved from another urban area. It is natural (i.e. inevitable) to be born in a certain location, or belong to a given culture. Or else, the subjects could have been classified according to their availability of social support, coping styles, levels of need-achievement, degree of preparedness for change, and the like. Support, goal-striving stress, coping, preparedness, were all personal issues that were randomly or naturally distributed within the human collective studied. Besides, these were all likely to be dealt with as internal to the micro-social group, to the family, to the individual personality, and therefore the expectation of meaningful changes could be successfully restricted to levels within the reach of social policy or educational programs. Furthermore, the traditional-modern dichotomy implies a primary contradiction between countryside and city which, to say the least, may be misleading, in that it disguises the network of economic and political domination established by the expansion of the global economic system. Thus, the basic contradiction is not between two competing cultural systems, and not even between different modes of production, even though such oppositions possibly play a key role in the historical development of these societies. Rather, the opposition to be explored is between conflicting
292
NAOMAR ALMEIDA-FILHO
practices of social groups within either a cultural or an economic system. The basic contradiction should not be tradition versus modernity (concretely expressed by the polarities urbanites-ruralites, Indians-Westerners, Occident-Orient), but indeed domination versus subordination, either in the economic or the political sphere. Finally, we must not confound the acculturation of societies that have been subject to some kind of “cultural invasion” with the acculturation process experienced by individuals or groups who move from one cultural setting to another. Even if they were essentially similar processes, the predicted effects on individual mental health might have been quite distinct, depending upon the adopted theoretical model. Furthermore, in this connection, structural explanations based on a political economy perspective could be well integrated into such a model through the notion that the process of globalization of the economy is at the basis of cultural (ideological) invasions and that macro-economic processes produce massive human dislocations and the subsequent acculturation pressures.
EPIDEMIOLOGICAL RESEARCH ON SOCIAL CHANGE AND MENTAL HEALTH The relationship between social change and mental health was precisely the basic scientific question which, back in the early sixties, gave a start to psychiatric epidemiological research in Latin America. Following upon Seguín’s clinical description of a “psychosomatic disadaptation syndrome” among Indians recently migrated to metropolitan areas (Seguín 1956), a series of community surveys on the relationships between migration, cultural change and mental health was conducted in Peru (Caravedo, Rotondo and Mariátegui 1963). In all samples, urban groups had higher levels of depression, agressiveness, and alcoholism than the rural groups; epilepsy, depressive reactions, and psychosomatic disorders were found to be higher among migrants, and psychoses and anxiety reactions to be more prevalent among nonmigrants. These authors proposed to explain their findings through the conventional urban stress model, as follows: The worsening emotional symptomatology among peasants newly arrived to Lima . . . suggests a probable influence of negative factors which perhaps are related with the new living conditions and with the demands of an urban environment, where major stresses and uncertainties are more present than in the simple and rural environment they come from (Rotondo, Mariátegui and Bambarén 1963: 56).
Since then, psychiatric epidemiology in Latin America has evolved based primarily on large-scale community surveys. This is probably due
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
293
to the poor condition of the health information systems in the continent which follows the even more precarious situation of the health care systems in those countries. The few studies conducted in clinical settings which attempted to derive “treated” morbidity indicators have not been successful, either because case-register data are unreliable and/or because the way health care has been organized in Latin America prevents the establishment of population-based estimates. A review of epidemiological investigations completed in 13 Latin American countries showed that 249 research reports were produced between 1970 and 1988 (Almeida-Filho 1991). According to this evaluation, the decade of 1980 was characterized by a growing scientific production in the field, with an average number of publications on psychiatric epidemiology in Latin America around 25 per year. The plateau corresponding to this recent period seems to have resulted from the establishment of institutionalized research groups with a more professional attitude toward psychiatric epidemiologic investigation, particularly in Puerto Rico, Mexico, Argentina and Brazil. Undoubtedly, psychiatric epidemiology is booming in Latin America. Although the methodology of the majority of investigations reviewed was still extremely deficient, with inadequate or simplistic techniques for data analysis, there has been a major push for methodological improvement of study designs in Latin American psychiatric epidemiology. One can also observe a recent trend towards the use of standardized diagnostic procedures and systems in the context of surveys covering a wide range of psychiatric conditions. Next, I will present a summary of methodological features and findings of the most important of such research efforts, outlining their contributions to the exploration of the mental health consequences of social change in the subcontinent. Results from previous studies Many surveys carried out in Latin America aimed at generating preliminary data to be used to plan community mental health programs (Grimson et al. 1972; Tarnopolski et al. 1977; González et al. 1978; Di Marco 1982; Gallegos and Miguez 1984; Penayo, Caldera and Jacobsson 1992). Even though providing valuable information on the overall unequal distribution of mental problems in society, these studies did not directly address the issue of social change and mental health. Using the Cornell Medical Index as the basic data collection device, Ponce (1970) conducted a community survey to study the housing, health and mental conditions of in-migrants to the metropolitan area of Lima, Peru, with a stratified sample of 779 individuals. The author reported that
294
NAOMAR ALMEIDA-FILHO
migrants had slightly worse mental health conditions than nonmigrants, and that those who had come from the highlands and rural areas showed higher rates of anxiety, depression and stress. With regard to length of urban residence, Ponce’s findings point to an increase in depression and anxiety, and a decrease in stress levels, along the time scale. For none of these results did the author present any significance testing nor any controlled analysis for confounding factors. However, stratified analysis by a socioeconomic ecological classification of neighborhoods revealed an excess of cases in the lower strata for both migrant and nonmigrant groups. In his discussion, Ponce emphasized that “there was a clear relationship between economic level and reactions of maladaptation, which were more intense in the lower economic levels,” pointing to an interaction effect of migration and socioeconomic status (Ponce 1970: 459). Coutinho (1976) carried out a cross-sectional prevalence study in a ruined slum area in Bahia, Brazil. The study was based on a census of 1,196 adults, consisting mostly of unskilled workers, craftsmen, prostitutes and petty criminals. A two-stage design was employed for the data collection: firstly, probable cases were screened with a non-validated standardized questionnaire and secondly, they were interviewed in a health center by medical residents. The overall prevalence of mental disorders was very high (49%), while neuroses and alcoholism were the most frequent diagnoses, both approximately 23%. There was an excess of psychiatric morbidity among women, but the only significant difference reported was for alcoholism (more prevalent among men). Higher prevalences of mental illness were found also among single, illiterate, unemployed and lower income subgroups. A well-conducted large-scale study of the association between migration and psychological disturbance in the Colombian city of Cali was reported by Micklin and León (1978). A Spanish translation of Langner’s 22-item scale and a questionnaire on occupational and residential history were applied to a multistage sample of 681 respondents, all employed or seeking work. The independent variables received a detailed treatment, with the distinction of “migrant status,” classified according to six life-cycle intervals, from “migrant type,” classifed by size of places of residence. The findings of the Cali study clearly indicate that gender, education and social mobility were associated with levels of psychiatric disturbance: females, illiterate and lower class people had the highest scores. There were no clear-cut differences in scores by migrational status, once controlled for life-cycle interval of migration. With regard to type of migration, the authors found that, for definitive moves to urban areas, individuals of rural origin consistently had higher scores in the symptom
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
295
scale. Migrants from other large urban areas showed the lowest score levels, regardless the life-cycle of migration. Analyses of variance of these findings revealed that “when all variables were considered simultaneously, however, the effect of migration on psychiatric disturbance virtually disappears” (Micklin and León 1978: 103). Even though not explicitly acknowledged by its authors, the Cali study was a careful, elegant test of Inkeles and Smith’s hypothesis, still limited by the same sampling bias of the Harvard survey, namely not including unemployed and underemployed workers (the informal sector of the economy). In Brazil, Almeida-Filho and Bastos (1982) conducted a pioneering case-control study to assess the effect of migration experience on depressive states in women. The group of cases was formed by 30 women who had been admitted to a university psychiatric ward with the diagnosis of depression (either reactive or endogenous). Sixty age-matched controls were taken from a sample of participants in a community survey who had no psychiatric symptoms. Three-fourths of the cases were migrants as compared to only one-third of controls. Migrant women showed a risk of receiving a diagnosis of depression nearly six times higher than non-migrants, even after controlling for occupation, educational level and marital status. Improving research designs Following the development of a multiple stage research design adapted for the local sociocultural conditions, a series of population studies of psychiatric morbidity has been carried out in the State of Bahia, Brazil. In that context, an adult screening instrument named QMPA – Questionário de Morbidade Psiquiátrica de Adultos (Santana 1982) was developed and tested for validity,3 being since then largely employed for epidemiologic studies of mental disorders in Brazil. In Bahia, the QMPA was applied in three community surveys: the first one in a poor-income district of Salvador (Santana 1982), the second in the nearby industrial town of Camaçari (Almeida-Filho et al. 1983), and the third in a rural area of the São Francisco Valley (Barbosa and Almeida-Filho 1986). The Salvador study was based on a cluster surface sample of 1,549 individuals older than 14 years. The data collection employed a two-stage design, with the administration of the QMPA followed by psychiatric interviews with the subgroup screened and a proportional subsample of negatives in the test. As reported by Santana (1982), the overall prevalence of psychiatric morbidity for the adult population was 20%, consisting primarily of neurotic and psychosomatic disorders (17%). Women had significantly higher overall prevalence of psychiatric disorders (24%) than
296
NAOMAR ALMEIDA-FILHO
men (16%), while men had 11 times more alcoholism than women. A non-significant association of migration status and low income was found with prevalence of psychiatric morbidity, especially for neurotic and psychosomatic disturbances. The study of Camaçari (Almeida-Filho et al. 1983), was based on a random sample of 1,067 adults, stratified according to the census districts of the town. It also employed a two-stage strategy for data collection, with a first phase of QMPA administration, but only screened individuals were interviewed by psychiatrists for the second phase. The annual overall prevalence of emotional disorders reached 24%, but less than 7% of the cases were considered to be have social impairment of any kind. In that investigation, the consumption of psychotropic drugs (mostly minor tranquilizers) was also investigated: almost 18% of the sample were using that type of medication by the time of the data collection, the majority of them without a medical prescription. Barbosa and Almeida-Filho (1986) reported results of a census on the mental health status of a rural population (n = 1,002) living in four neighboring villages in the region of São Francisco, in the backlands of Bahia, which employed a similar methodology, based on an adapted version of the QMPA. The overall prevalence of anxiety disorders was 14.5%, with women yielding a prevalence 2.2 times greater than men. Prevalences increased with age but were negatively associated with education (19% for illiterates; 12% for those completing elementary school; 9% for highschool graduates). Finally, the prevalence of anxiety disorders among those who legally possessed the land was slightly lower than among sharecroppers (14% as compared to 16%). These rates were consistently below those found in urban areas, with comparable methods. To approach the effect of social factors on mental health, thus testing Inkeles and Smith’s hypothesis, I used the dataset of the Salvador study presented above (Almeida-Filho 1982). The traditional hypothesis of an association between migration and mental health was tested using a multivariate technique, controlling for the confounding of gender, age, marital status, education, position in the labor market, income and family size. The hypothesized association was statistically significant only when controlling confounders one-by-one. The simultaneous adjustment for gender, education and marital status practically erased the migration effect. A similar approach was used to test the alternative hypothesis of an association between displacement from the labor market (expressed by unemployment and under-employment) and poor mental health. In this case, the association was statistically significant, even after adjusting for the whole set of confounders. Interpreting these findings, my conclusion
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
297
was then fully in support of Inkeles and Smith’s prediction, in that “geographical mobility as a life-change or as a cultural change may not be the basic process of interest for the study of social factors and mental illness in Third World social formations” (Almeida-Filho 1982: 115). In spite of better study-designs and screening instruments more adequately developed and tested, the investigations of this series were based on less structured diagnostic procedures, probably of low reliability. As a consequence, the comparability of their findings was limited to local research contexts. In addition, the test of Inkeles and Smith’s hypothesis had been performed with stress levels as a broadly defined outcome variable, without a clear clinical definition of psychopathology. A new generation of research Indeed, data collection designs developed by Latin American researchers have been cost-efficient and suitable for application in underdeveloped areas. However, these investigations have generated data regarding highly specific population groups, which prevents the generalization of their results to the general population. Despite being somehow too expensive for underdeveloped countries with other health priorities, case-registers in well-defined catchment areas would be needed to produce morbidity indicators, particularly for low-prevalence disorders. Anyway, community morbidity studies with improved and more adequate methodology are still needed in many regions of the continent, for certain population groups, and for selected psychopathological conditions, targeted at specific research hypotheses. Currently, a new generation of psychiatric morbidity studies has been developed in Latin America, following the guidelines established by the wave of ECA (Epidemiological Catchment Area) surveys conducted in the US (Robins and Regier 1991). The results of such studies, carried out in Puerto Rico (Canino et al. 1987) and in Brazil (Mari et al. 1993; Almeida-Filho et al. 1997), with a similar design and classification system, have shown that an important proportion of the urban adult population of these countries (ranging from 19% to 34%), suffer from some kind of psychiatric disturbance. In general, distribution patterns and prevalence rates specific for diagnostic categories are similar to those found in the original ECA studies, with a predominance of phobias, alcohol abuse, anxiety and depressive disorders. In addition to producing good-quality data on the epidemiology of mental disorders using more advanced and standardized diagnostic systems, studies of this new generation may become an important source of evidence for the exploration of psychosocial correlates of the occurrence
298
NAOMAR ALMEIDA-FILHO
of emotional disturbances in these societies. More specifically, they may reinforce the clinical basis of such research and enhance the comparability of research findings, allowing the approach of the modernization-stress hypothesis in a cross-cultural fashion. Considering that the influence of sociocultural processes on mental health traditionally has been a major research agenda in the continent (Almeida-Filho 1987), the natural (social) history of mental illness ought to be studied in its particularities. With this aim, analytical investigations such as case-control studies are highly recommended in order to provide a better knowledge about social risk factors active in the Latin American populations. In the next section, I will analyse some preliminary data of a case-control study nested in a large-scale survey recently conducted in Brazil (Almeida-Filho et al. 1997), as an attempt to test whether Inkeles and Smith’s hypothesis still holds true in the current historical context of Brazilian society.
PSYCHOSOCIAL CONSEQUENCES OF MODERNIZATION: NEW EPIDEMIOLOGICAL EVIDENCE FROM URBAN BRAZIL The Brazilian Multicentric Study of Psychiatric Morbidity was conducted in three major urban areas of different regions of Brazil (Brasília, São Paulo and Porto Alegre), to estimate the overall prevalence of psychiatric disorders in the community.4 A two-stage cross-sectional design was applied to a representative sample of adults in the three research sites. All sampled subjects were screened for the presence of psychopathology and a subsample was selected for a confirmatory psychiatric interview. A more detailed account of the survey methods can be found elsewhere (AlmeidaFilho et al. 1997; Mari et al. 1993). In this section, I will first introduce a few methodological features of that study, followed by a case-control analysis of the evidence produced with regard to socio-economic determinants of selected non-psychotic conditions (anxiety disorders, phobias, somatization, and depression). Research methods The Brasília site was chosen for a pilot study, where the fieldwork was completed in the first semester of 1991. The São Paulo survey was conducted in September–December 1991 and in Porto Alegre it was done in December 1991–March 1992. In the first phase of the fieldwork,5 the following instruments were applied: (a) a Family Chart, covering demographic and socio-economic information (including migrational and
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
299
occupational history); and (b) the QMPA, a 44-item screening instrument developed by Santana (1982) for the detection of psychiatric morbidity in epidemiological studies adequate for the Brazilian urban environment, as described above. The study used a cluster sampling strategy, based on a socio-economic stratification of all census units within metropolitan limits.6 This strategy was successfully employed in Brasília and Porto Alegre, but in São Paulo, the biggest Brazilian metropolis, a sampling stratification was performed only in three of the 48 districts which compose the metropolitan area. The final sample sizes were the following: Brasília – 2,345 sample, 285 subsample; São Paulo – 1,742 sample, 236 subsample; Porto Alegre – 2,384 sample, 315 subsample; total – 6,470 sample, 836 subsample. For the second phase, a subsample of individuals considered as probable cases (QMPA scores above 7) were interviewed by a team of psychiatrists specially trained for this type of investigation, with diagnostic purposes. A subsample of the non-suspected was also interviewed by the second phase team, in order to reduce the false positiveness of the psychiatric examination. Examiners and examinees did not have any previous information on scores or on screening status. Diagnostic interviews included basically the application of a Brazilian version of the DSM-III Symptom Checklist, translated and adapted by Miranda et al. (1987). For the case-control study, all subjects positively diagnosed by the second phase interviewers as suffering from non-psychotic disorders, and whose symptoms had begun within the past 12 months, were considered as cases. Controls were randomly selected among those who, examined by the diagnostic team, did not fulfill any of the diagnostic criteria of DSM-III and were therefore considered as “disease-free”. Matched by research site, more than one control per case was taken whenever possible, aiming at increasing the power of statistical testing (Rothman 1986). After the application of the selection procedures, the study groups were constituted as follows: anxiety disorders – 124 cases and 124 controls; phobic disorders – 137 cases and 137 controls; somatization disorders – 60 cases and 120 controls; affective disorders – 58 cases and 174 controls. Data analysis The basic dependent variable of the present analysis is the occurrence of selected non-psychotic disorders, covering the following categories: (a) affective disorders (depression, dysthymia); (b) anxiety disorders (generalized anxiety states, post-traumatic acute or chronic disturbances, atypical anxiety states); (c) phobic disorders (agoraphobia, social and simple phobias); and (d) somatization disorders (conversive and dissociative
300
NAOMAR ALMEIDA-FILHO
disorders, somatization syndrome, psychogenic pain). Independent variables are migrational status, with subjects classified as migrants (from inside and outside the region) and non-migrants (city-born subjects), and position in the formal labor market, with the categories of displaced (unemployed, under-employed) and placed (salaried, retired, employer). Due to the peculiarities of domestic labor, “housewife” was taken as a separate analytical category. Age, gender, education, marital status, household density (resident/room ratio), and religious affiliation were all taken as potential confounding variables in the analysis. As a first step, the adequacy of pooling the case subsamples was evaluated by assessing the patterns of association between outcome and independent variables in each research site. A discrepancy was found only for affective disorders in the Brasília subsample, determining its exclusion from some analysis of this diagnostic group. The case-control design is perhaps the most cost-efficient tool for epidemiologic hypothesis testing. It compares in a retrospective fashion the differential exposure to a putative risk-factor between currently diseased (case) and non-diseased (control) groups (Rothman 1986). Despite its incapacity of producing measures of disease occurrence (such as incidence or prevalence), the case-control study can estimate the relative risk, i.e. it can measure the magnitude of a hypothesized association, as well as assess its statistical significance. Appropriate epidemiological and statistical analyses were performed with the EPIINFO (Dean et al. 1990), while logistic regression modelling used the MULTLR software (Campos-Filho and Franco 1989). Below, I will analyse a series of odds-ratios, which is the association measure of choice for case-control studies (Kahn and Sempos 1989). The odds-ratio (OR) varies from 0 to ∞, with a neutral point at 1.0. Therefore, an estimated odds-ratio above 1.0 suggests a risk factor, while a value below that mark may be interpreted as indicating a protective factor. Confidence intervals express the range of variation (upper and lower limits) within which the odds ratio can be estimated within significance levels (usually p < 0.05).7 Selected results on migration and position in the labor force Let us first take a look at the psychopathological profile of the samples studied. The overall lifetime prevalence of psychiatric symptoms of any kind, standardized by age, was high in the three areas surveyed: the metropolitan region of Brasília showed the highest rates, around 51%, followed by Porto Alegre, with 43%, and São Paulo, around 30%. Ageadjusted prevalences of more severe disorders, recognized as those in need
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
301
of care, ranged from 19% (São Paulo) to 34% (Brasília and Porto Alegre). Anxiety disorders comprised the highest prevalence rates, from around 10% (Porto Alegre and São Paulo) up to 18% (Brasília), followed by phobic states, from 8% (São Paulo) to 14% and 17% (Porto Alegre and Brasília, respectively). Alcohol abuse/addiction yielded the most consistent prevalence levels, with all rates varying from 8% to 9%. Somatization and depression prevalences showed much variation inter-areas: somatization – 3% in São Paulo to 5% in Porto Alegre and 8% in Brasília; depression – from 2% and 3% for respectively São Paulo and Brasília, to 10% in Porto Alegre. Table 1 presents the analysis of the association between migration experience and the occurrence of non-psychotic disorders. Crude overall measures of association revealed that migration has no effect over the outcomes anxiety and phobias, because the respective odds ratios were not statistically different from 1.0. For somatization and depression, the estimated odds ratios (almost 2.0) reached borderline significance levels. Stratified analyses by gender yielded no significant effects of migration over the mental health indicators for any of the subgroups analysed. Heterogeneity tests suggested presence of interaction between gender and migration, for anxiety and phobias, at a borderline (0.10) level of statistical significance. Logistic regression analysis did not change these patterns and, furthermore, failed to detect any relevant interaction between migration and the other independent variables. This can be interpreted as an overall rejection of the hypothesis of a direct effect of migration status on the studied psychological outcomes. Table 2 allows the testing of the hypothesis that regular, stable, formal employment would be protective for the individual’s mental health. Overall crude measures of association were consistent with this prediction, with the odds ratios for anxiety disorders, phobias, somatization and depression located below the unity. Moreover, with the exception of anxiety, all measures reached levels of statistical significance. Stratification by gender suggests that, for all conditions under analysis (especially phobias and depression), employment seems to be a source of protection for women but not for men. However, such indication of interaction was not supported by the heterogeneity test, even at a borderline level of significance. Logistic regression adjustments resulted in erasing out the expected protective effect of employment on the mental health indicators, indicating that the joint effect of the other independent variables analysed indeed was confounding the hypothesized relationship. As shown in Table 3, there seemed to be an effect of displacement from the formal labor market on the occurrence of selected non-psychotic
302
NAOMAR ALMEIDA-FILHO
Table 1. Odds ratios and 95% confidence intervals for the association between migration status and non-psychotic disorders Diagnosis (N of cases)
Overall or (ci)
Females or (ci)
Males or (ci)
Adjusted# or (ci)
Anxiety (124)
1.32 (0.73–2.39)
1.74 (0.81–3.77)
0.61 (0.21–1.76)
0.96 (0.50–1.85)
Phobias (137)
1.08 (0.61–1.90)
1.26 (0.60–2.63)
0.56 (0.20–1.62)
0.65 (0.34–1.24)
Somatization (60)
1.99a (0.87–4.60)
1.89 (0.71–5.06)
2.71 (∗ –∗ )
1.32 (0.52–3.33)
Depression (58)
1.87a (0.90–3.94)
1.79 (0.70–4.64)
2.00 (∗ –∗ )
1.71 (0.70–4.14)
# Logistic regression adjustment for Age, Gender, Education, Marital Status, Religious Affiliation, Household Density, and Position in the Labor Market. ∗ Calculations made impossible due to small numbers. a p < 0.10
Table 2. Odds ratios and 95% confidence intervals for the association between employment and non-psychotic disorders Diagnosis (N of cases)
Overall or (ci)
Females or (ci)
Males or (ci)
Adjusted# or (ci)
Anxiety (124)
0.68 (0.39–1.20)
0.84 (0.40–1.76)
2.33 (0.52–∗ )
1.35 (0.75–2.43)
Phobias (137)
0.45c (0.26–0.78)
0.52b (0.25–1.06)
1.53 (0.39–∗ )
0.93 (0.53–1.64)
Somatization (60)
1.41b (0.20–0.84)
0.67 (0.27–1.65)
1.55 (∗ –∗ )
0.98 (0.44–2.25)
Depression (58)
0.51b (0.25–1.01)
0.53a (0.23–1.21)
1.40 (0.23–∗ )
0.88 (0.38–2.04)
# Logistic regression adjustment for Age, Gender, Education, Marital Status, Reli-
gious Affiliation, Household Density, and Migration Status. Calculations made impossible due to small numbers. 0.05 p < 0.10 0.005 < p < 0.05 p < 0.005
∗ a b c
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
303
Table 3. Odds ratios and 95% confidence intervals for the association between position in the labor force and non-psychotic disorders Diagnosis (N of cases)
Categories
Overall or (ci)
Adjusted# or (ci)
Anxiety (124)
Placed+ Displaced Housewife
1.00 1.55 (0.73–3.29) 1.67a (0.85–3.14)
1.00 1.28 (0.54–3.04) 1.57 (0.66–3.77)
Phobias (137)
Placed+ Displaced Housewife
1.00 1.56 (0.75–3.24) 2.91c (1.53–5.53)
1.00 1.20 (0.57–2.53) 0.81 (0.39–1.68)
Somatization (60)
Placed+ Displaced Housewife
1.00 2.11a (0.86–5.21) 3.02c (1.32–6.95)
1.00 1.34 (0.64–2.80) 1.44 (0.71–2.92)
Depression (58)
Placed+ Displaced Housewife
1.00 1.31 (0.54–3.15) 2.79c (1.27–6.17)
1.00 1.10 (0.41–2.95) 0.78 (0.29–2.03)
# Logistic regression adjustment for Age, Gender, Education, Marital Status,
Religious Affiliation, Household Density, and Migration Status. Baseline category. 0.05 < p < 0.10 0.005 < p < 0.05 p < 0.005
+ a b c
disorders, considering the crude overall odds ratios above 1.0. However, with the exception of somatization disorders, none of them reached statistically significant levels, which means that these weak associations could be attributed to random sampling variation. As expected, adjustment through logistic analysis resulted in the reduction of all these measures of association to the unity, again revealing the effect of confounding variables on the hypothesized relationship. On the other hand, the housewife role was a consistent, significant risk factor for all disorders considered, with the odds ratios corresponding to phobias, somatization and depression approaching 3.0. According to Table 4, gender-stratified estimates showed that the pattern of economic displacement as a risk factor holds only for women, with significant odds ratios above 2.0 for all conditions but depression. However, with the exception of phobic states, none of the non-psychotic disorders were significantly associated with household labor. Despite not reaching levels of statistical significance, the odds ratios for males were
304
NAOMAR ALMEIDA-FILHO
Table 4. Odds ratios and 95% confidence intervals for the association between position in the labor force and non-psychotic disorders, stratified by sex Diagnosis (N of cases)
Categories
Females or (ci)
Males or (ci)
Anxiety (124)
Placed+ Displaced Housewife
1.00 2.84a (0.97–8.53) 1.31 (0.62–2.77)
1.00 0.41 (∗ –1.82) –
Phobias (137)
Placed+ Displaced Housewife
1.00 2.65b (0.92–7.30) 2.09b (1.01–4.32)
1.00 0.65 (∗ –2.49) –
Somatization (60)
Placed+ Displaced Housewife
1.00 2.25a (0.90–7.30) 1.56 (0.61–4.00)
1.00 0.67 (∗ –6.82) –
Depression (58)
Placed+ Displaced Housewife
1.00 1.50 (0.50–4.46) 1.90 (0.79–4.56)
1.00 0.84 (∗ –5.02) –
# Logistic regression adjustment for Age, Gender, Education, Marital Status,
Religious Affiliation, Household Density, and Migration Status. + Baseline category. ∗ Calculations made impossible due to small numbers. a 0.05 < p < 0.10 b 0.005 < p < 0.05 c p < 0.005
consistently below the unity, suggesting perhaps a “protective” effect of participating in the informal sector of the economy, only for this gender group. Accordingly, heterogeneity tests suggested the presence of interaction of gender with displacement on the occurrence of anxiety disorders (p = 0.02) and phobias (p = 0.07).
TOWARD NEW THEORETICAL FRAMEWORKS Overall prevalence estimates found in this survey were remarkably higher than those established by previous investigations with comparable methodology (Santana 1982; Almeida-Filho et al. 1983). Assuming that these studies measured similar nosological entities, then the mental health status of the urban Brazilian population might have worsened during the 1980 decade. On the other hand, morbidity profiles according to diagnostic
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
305
groups identified in the present study are consistently close to the findings of the Bahian research series mentioned above. Again, there seems to be an increment in the prevalence levels of disorders considered as adaptive, reactive or, in another perspective, produced through a process of social determination, such as the majority of non-psychotic disorders. How to explain these findings? Have they resulted from methodological differences between two generations of research, such as diverse sampling strategies or classification systems? Probably not, particularly if we consider the consistency and magnitude of the differences observed. The most probable explanation is that a constellation of risk factors has been active during the time frame considered, as part of a powerful process of change in Brazilian society. But what are these risk factors? Are they individually based, such as modernizing personal experiences or exposure to mass media? Or related to population dynamics, such as demographic density or geographical dislocations? Or are they socially determined, as for example social class mobility or the formation of an urban reserve army of labor, which has produced the exclusion of people from the labor market? The case-control analysis produced no evidence in support of the hypothesis of a direct, positive association between migration experience and the occurrence of any of the non-psychotic disturbances considered. Surprisingly enough, placement into the formal labor market (having a regular, stable job) did not seem to effectively protect against suffering such disorders. However, stratified analyses showed a consistent pattern of interaction between labor-related variables and gender, in the following direction: being regularly employed seemed to be somehow a risk-factor for men but not for women. Among women, housewife activity was associated with a higher risk for all disorders studied. As far as the gender interaction effect is concerned, the fact that statistical significance criteria were not met has to be taken in relative terms given the overwhelming consistency observed for all disorders analysed. In summary, results from this study pointed to a rejection of Inkeles and Smith’s hypothesis, at least as a general prediction of the kind tested by previous research (Micklin and Leon 1978; Almeida-Filho 1982). Indeed, a formal service or industrial job nowadays does not seem anymore to be a protective factor for mental health, but even worse, as for males, it appears as a source of psychological unrest, a potential risk-factor for nonpsychotic disorders. Furthermore, from these results, gender (and correlate social processes and roles) has come to be considered as a major fundamental dimension for the ethno-epidemiological study of mental health (Paykel 1991).
306
NAOMAR ALMEIDA-FILHO
As discussed above, traditional approaches would advocate theoretical models based on the notion of modernization, through correlate processes of migration, urbanization, acculturation and Westernization. As I have noted elsewhere (Almeida-Filho 1987), the Latin American epidemiological literature has been profuse in such models, sometimes barely implicit in the research objectives and methods but many times clearly exposed. Even ancient psycho-anthropological constructs such as nostalgia, “marginal man” or modal personality traits were called for, as in the Peruvian first wave of studies which mentioned a cultural shock mechanism of this kind to account for the higher rates of psychopathology among Indian migrants to the cities (Rotondo, Caravedo and Mariátegui 1963; Ponce 1970). During the eighties, such culturalist models have been considered outdated, and replaced by economistic explanations, based largely on the dynamics of the labor force in societies recently integrated to the economic world-system. A theoretical move like this has been justified by the failure to find a clean positive association between psychopathology and migration after controlled analysis in improved research designs, as in Micklin and Leon’s (1978) and in my own early work (Almeida-Filho 1982). Indeed, Micklin and León (1978), who initially considered migration as a life crisis causally related to psychological stress, eventually found evidence to support the hypothesis that the urban social mobility system would be the main factor involved in being emotionally disturbed under such circumstances. I myself pointed out the class formation process at the root of the migration-cultural change-modernization problem of Third World societies, as part of the contemporary history of an all-expanding capitalist mode of production, incorporating and collapsing old modes of production, generating more complex economic and social formations (Almeida-Filho 1982). This alternative theoretical model was clearly based on the identification of modernization processes with the expansion of the capitalist mode of production, in which the so-called development process resulted in placing into a market economy increasing numbers of the population of formerly underdeveloped areas (Oliveira 1977). This implies that social change means the process of constant transformation of the social class structure, determined by a contradiction between different modes of production at the economic basis of the social formation (Castells 1977; Srour 1978). Following upon such transformations, some social classes may lose their position in the social structure, and their members would be selected for absorption by the emerging classes or to be displaced from the new social order. Of course, the dynamics of this selection-displacement
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
307
process could be understood in parallel to the processes of rural-to-urban migration, or modernization, or urbanization (Singer 1975; Castells 1977). In the metropolis, territorial domain of modernity where production, circulation and consumption are more tightly controlled, the social selection process would eventually lead those who have been successively displaced to form the so-called “reserve army of labor” and the “lumpen-proletariat” (Souza and Faria 1980). Within this framework, one could hypothesize – as I did fifteen years ago (Almeida-Filho 1982) and a few pages above – that the condition of being displaced implies, for the individual, extreme uncertainty and social stress, with permanent or temporary repercussions at a psychic level. However, in the present work, no evidence was produced to support such a hypothesis. Now I think that we do need further theoretical moves, in order to account for the deep changes in the social reality and to understand the new failures in testing key hypotheses such as Inkeles and Smith’s. Indeed, the economic world-system and Latin American social formations have changed profoundly in recent years. According to Castells (1992: 6), one of the most influential theorists of the historico-structural perspective summarized above, the economic world-system is not anymore expanding geographically (as an imperialism) because its material basis has been radically transformed by “one of the most important technological revolutions of human history,” forming a new type of economy based more on information technology and on the phenomenon of “hypertrading,” i.e. the interconnection of production, distribution and consumption flows across national borders. One of the most striking features of such economic globalization has been pervasive structural unemployment mostly when and where work for all was once promised (Freeman and Soete 1994). The political changes of Eastern Europe and the unequal economic evolution of peripheral countries may justify the proposition that the Third World has disappeared (Harris 1986). In addition, a Fourth World is under formation as whole countries and particular regions of given countries get more and more economically and socially underdeveloped. After all these years, modernization has been more and more implicated with the production of social inequality, particularly in societies that, once upon a time, compounded the so-called “developing world.” The metaphorical nature of development theories is no recent discovery, as seen in more realistic early descriptions of Third World urban environments, such as for example Brody’s The Lost Ones, written in the early seventies: [Latin American big cities] are characterized by little opportunity for self-determination, inadequate employment opportunities, racial and social class-based discrimination, a mobility blocking educational system, inadequate welfare agencies, little opportunity for realizing latent potential, poor information dissemination, lack of adequate housing
308
NAOMAR ALMEIDA-FILHO
and persistent exposure to severe physical (often occupational) stresses with inadequate protective and reparative facilities (Brody 1973: 580).
In Latin America, the majority of the population now live in urban areas and, in average, have more access to education, electricity, and mass media communication than ever. In several of these countries, the long term trend has been one of economic growth, despite the recession of the 1980 decade, resulting in overall higher per capita income. However, “in average” and “per capita” are abstractions rather than concrete reality, and the social reality of these countries is that the rich are fewer but richer and the poor are in larger number and have become more and more miserable. Indeed, the distribution of wealth in these countries has been characterized by an increasing income concentration in a few sectors of the population (CEPAL 1991), producing what has been called “the new poverty”. Currently, it is estimated that 46% of the Latin Anerican population are below the poverty line and that such new poverty has grown at a rate of 2.5% per annum since 1986 (PNUD 1993). Brazil, for example, has undergone very rapid and deep economic, social and political changes. During the seventies, time-frame of my former investigations, the country’s GNP grew at an average of 9% per annum, while remaining one of the most unequal societies in Western history, with less than 10% of the population accounting for almost 50% of total income. During the past decade, Brazil’s economy has been dominated by a deep economic crisis, with recession, rising inflation, high public fiscal debts and the need to service a huge external debt (SELA 1988). The country’s economic growth slowed down to less than 1% per annum but, in parallel, urbanization, fertility declines and increasing political participation have maintained steady improvement trends (World Bank 1989). Such a crisis background, which has opened the 1990’s with a deep recession in the formal economy and therefore has brought unemployment rates to a historical peak (CEPAL 1991), may help us to figure out how, to some point, having a stable job may represent lower earnings and opportunities as well as higher social control, responsibility and pressures over the worker as compared to being part of the more dynamic informal sector of the economy (Medeiros and Salm 1994). One could then understand why, at the lowest levels of subsistence, classical indicators of social and economic position are not helpful anymore as predictors of poor mental health outcomes. Let me conclude by going back to the subtitle of the present paper: social change and mental health. The very expression ‘social change’ means simultaneously the existence of a concrete social reality and its constant historical transformation. However, one should not forget that the
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
309
proper meaning of social change, and modernization as its fate, has also been subject to change itself. Interpretation models of society ought to be transformed in order to account for such a changing object of inquiry, the social reality under historical transformation. Surely modernization theory implied a highly optimistic approach to the issue of social change. As a matter of fact, Inkeles and Smith’s enterprise was set out to demonstrate that becoming modern would not be too bad for one’s (mental) health. They succeeded in showing the adaptive power of the human being to overcome the general stresses and uncertainties of sociocultural change, as acute and profound as they be. They also found that some structural consequences of modernization were more important for the mental health of people than its suprastructural correlates, but they expected, as candidly as any development theorist could, that such deleterious consequences were transient but necessary by-products of the social change process, to be solved when modernity (and jobs, and education, and housing, and communication) would have become available for all. Any theoretical interpretation of the present findings should consider the fundamental issue of meaning. Conventional epidemiologic reasoning of risk-factors, confounding variables and interaction terms, is useless in this case. For instance, to understand how employment can be protective only for women but not for men one needs to admit that the same exposure factor (such as a regular job, or the lack of it) may have distinct meanings for different genders (Lennon 1987; Rosenfield 1989). This might also explain the historical changes in the effect of some of these social factors, to the extent that displacement from the labor market was tested as a riskfactor for poor mental health in the past and now it appears not associated with the occurrence of non-psychotic disorders. Again, the social and individual meaning of being unemployed or underemployed has certainly changed in parallel with the overall transformation of the social context during the decade. And now? Now that development has turned into its opposite – recession, for many formerly “developing” countries? Now that modernization is not anymore equivalent to Westernization but has become more “Orientalization”? Now that talking about Third World countries does not make any sense, also because there is not anymore a Second World? Now that the evidence from bigger and better epidemiologic research is not so clearcut in support of such grand theories? Now that, as we saw above, being either employed or displaced from the labor market does not seem to affect the individual’s psychological outcome in a context of deep economic crisis? Now that gender, along with all related economic and power corre-
310
NAOMAR ALMEIDA-FILHO
lates, emerges as the fundamental heterogeneity as far as mental health is concerned? At last, does it still make sense to talk about modernity (and its psychosocial consequences) in the contemporary economic, social and cultural context? Maybe not. Social scientists do not seem interested anymore in modernization or equivalent constructs, simply because they are already debating how close is modernity to its end, and we would thus be in late modernity (Giddens 1990, 1992), or whether it is over or not, and here come post-modern times (Lyotard 1985; Jameson 1992, 1994). Is this sort of discussion inopportune, or at least dislocated, in a society such as Brazil (or any other Latin American country) in the 90’s under continued economic and political crisis? Maybe so. Nonetheless, while typical fourth-world epidemics such as famine, dengue and cholera are breaking out afresh, a country like Brazil has been able to develop a sophisticated telecommunication infrastructure, with satellite-based broadcasting networks and massive informatization of its national banking system. Indeed, by our peripheral condition, this perception of everyday misery simultaneous with the disclosure of openended potential for growth, everywhere, daily provokes a bitter critical sense that we can apply to all scientific, technological and political discourses to which we are subjected. The challenge, therefore, may be to develop a scientific approach to the mental health repercussions of social change capable of accounting for the perverse coexistence of underdevelopment with over-exploitation, producing a “new poverty” afflicted by the “new morbidity,” which allows the simultaneity of a mass of pre-modern subjects with an elite of post-modern rulers.
ACKNOWLEDGMENTS I would like to thank the CNPq – Conselho Nacional de Desenvolvimento Científico e Tecnológico of Brazil for supporting the writing and revision of this paper through a Research Fellowship Award. Special thanks are due to Prof. Jair Mari, from the UNIFESP – São Paulo, and to Prof. Evandro Coutinho, from the ENSP/FIOCRUZ, for their participation as co-PI in the Brazilian Multicentric Study of Psychiatric Morbidity. Duncan Pedersen, Gilles Bibeau and Byron Good read earlier drafts of the manuscript and contributed much to its final form.
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
311
NOTES 1. The study methodology was based on the application of a battery of questionnaires, including a Psychosomatic Symptoms Test, to selected samples of rural cultivators, urban migrants, urban service workers, and industrial workers, involving almost 6,000 subjects. Two Latin-American countries, Argentina (n = 817) and Chile (n = 931), were included, along with East Pakistan, India, Nigeria and Israel. 2. In the Stirling County study, Leighton et al. (1963) had reported that the occurrence of migration prior to age 20 was “a factor comparable in importance to low education and low occupational position in increasing the risk of psychiatric disorder”. These authors interpreted their results “as indicating a deleterious effect of migration during childhood and adolescence on an individual’s subsequent mental health” (Leighton et al. 1963: 301). Srole et al. (1962), in Midtown Manhattan, had found higher frequency of psychiatric symptoms among urban migrants to New York City than among migrants from rural areas, while both migrant groups performed better than nonmigrants. In both studies, migration was conceived as a correlate (either cause or consequence) of the exposure of individuals to social and cultural change. 3. In epidemiologic investigation, researchers have been more concerned with “concurrent validity,” which implies analyses of the performance of a given diagnostic tool vis-à-vis some previously accepted standard, in terms of sensitivity (capacity of identifying cases) and specificity (capacity of recognizing healthy subjects). The issue of “face validity,” even though quite relevant for cross-cultural studies, has been indeed neglected by Western-centered epidemiologic research on mental health. 4. This research was commissioned by the Mental Health Division of the Ministry of Health of Brazil and received financial support of the Ministry of Health, PanAmerican Health Organization, FINEP and FAPERGS. Myself, Jair Mari and Evandro Coutinho worked as PI of the investigation. Drs. Josimar Franca, Sergio Andreoli and Jefferson Fernandes were respectively field-directors of the Brasilia, Sao Paulo and Porto Alegre surveys. Dr. Vilma Sousa Santana collaborated with the elaboration of the research proposal and in planning the fieldwork of the pilot-study, while Dr. Ellis Busnello acted as consultant for diagnosis issues. The data analysis presented herewith was largely due to Dr. Coutinho’s invaluable assistance. The utilization of survey material for the purposes of the present paper has been kindly authorized by the commissioning agency and the research coordination team. 5. The research team for the first phase was composed of a total of 51 interviewers (16 in Brasília, 15 in São Paulo and 20 in Porto Alegre). They were either medical students or had a university degree in a health profession (nursing, psychology), and received a 72-hour training program developed by the University of Bahia for interviewing and questionnaire administration, basically through techniques of role playing, and participated in several reliability tests. They were also trained in the field, in order to get them acquainted with each research setting, by practicing application of questionnaires to families not included in the sample. In the fieldwork, each interviewer was in charge of approximately 60 families and they were instructed to do the first visit at meal times, weekends and in the evenings, up to three times before considering that unit missing. Family composition information was provided preferably by the housewife (or someone performing this role), while the screening questionnaire was answered by every family member older than 14 years. The survey received major coverage of the local media, helping to reduce the refusal rate to a minimum. Field supervisores monitored closely the data collection, re-visiting participating families at random to ensure
312
NAOMAR ALMEIDA-FILHO
quality control of the fieldwork. The duration of the first phase of fieldwork varied from one (São Paulo) to three months (Porto Alegre). The second phase of the survey employed a total of 25 psychiatrists and psychologists with clinical experience (6 in Brasília, 8 in São Paulo and 11 in Porto Alegre), trained in the use of the Checklist by the staff of the Department of Psychiatry of Escola Paulista de Medicina. The training course comprised 40 hours, including clinical supervision and a reliability check of the interviewers. Results of the diagnostic interview were also registered in terms of the diagnostician’s degree of confidence in presence of psychopathology, duration and level of severity of symptoms. The interval between the interview of the first phase and the diagnostic examination varied between 1 and 4 weeks. Due to problems related to the type of research setting and the population covered (out-migration and irregular working schedules), it was impossible to locate 42 subjects (15 in Brasília, 16 in São Paulo, 11 in Porto Alegre) assigned for the second phase interviewing. 6. In each site, census units were randomly selected and, with the help of updated surface maps, cluster blocks and housing units were identified and also randomly selected for inclusion in the study. Sample size calculations were based on modules of 900 individuals, within a 2% precision interval for overall prevalence estimates of up to 2 strata, which resulted in a baseline sample size of 1,800 subjects. Considering an expected proportion of losses and refusals of 10%, the number of eligible interviewees for the first screening phase was established at 2,000 subjects for each research site. For the second phase of diagnostic confirmation, a random selection subsample (n = 300) was drawn for each base sample, including 30% of screened probable cases and 10% of probable non-cases. 7. The odds ratios and confidence intervals were calculated with Cornfield approximations. In a second step, the presence of effect modifiers and potential confounders was explored through stratified analysis (Rothman 1986). Mantel-Haenszel summary odds ratios and the respective test-based confidence intervals were then calculated. In order to control for the simultaneous confounding effect of the whole set of study variables, logistic regression analyses were performed following Hosmer and Lemeshow (1989) guidelines. Stepwise procedures were not employed due to their dependence upon criteria of statistical rather than epidemiological significance for inclusion of variables in the model. The presence of interaction between the independent variables and the potential effect-modifiers was assessed with Woolf’s homogeneity test (Kahn and Sempos 1989). Only interaction terms which had reached a “likelihood ratio test” (Hosmer and Lemeshow 1989) significant to the 0.90 level were maintained in the logistic models.
REFERENCES Almeida-Filho, N. 1982 The Psychosocial Costs of Development: Labor, Migration and Stress in Bahia, Brazil. Latin American Research Review 17: 407–432. 1987 Social Epidemiology of Mental Disorders: A Review of Latin American Studies. Acta Psychiatrica Scandinavica 75(1): 1–10. 1991 Epidemiolog´ıa Psiqui´atrica en America Latina: Temas Metodol´ogicos y Perspectivas de la Investigaci´on. In R. Alarc´on, ed., Desde Nuestra Propria Entraña – Homenage al Profesor Javier Mari´ategui (pp. 184–200). Lima: Universidade Peruana Cayetano Heredia.
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
313
Almeida-Filho, N., and S.B. Bastos 1982 Estudo caso-controle da associação entre migração e desordens depressivas em mulheres. Jornal Brasileiro de Psiquiatria 31(1): 25–30. Almeida-Filho, N., V. Santana, D.M. Coutinho, V. Aguiar, and D. Antonissen ´ 1983 Prevalência de Desordens Mentais em uma Area Industrial da Região Metropolitana de Salvador. Universitas 32: 59–72. Almeida-Filho, N., J.J. Mari, E. Coutinho, S. Andreoli, J. Fernandes, and J. França 1997 Brazilian Multicentric Study of Psychiatric Morbidity: Methodological Features and Prevalence Estimates. British Journal of Psychiatry 171: 524–529. Argandoña, M., and A. Kiev 1972 Mental Health in the Developing World. A Case Study in Latin America. New York: The Free Press. Barbosa, J., and N. Almeida-Filho 1986 Prevalence of Emotional Disorders in a Rural Area of Bahia, Brazil. Rio de Janeiro: 11th World Congress of Social Psychiatry, Annals. Barger, W. 1977 Culture Change and Psychosocial Adjustment. American Ethnologist 4: 471–495. Berry, J. 1980 Acculturation as Varieties of Adaptation. In A. Padilla, ed., Acculturation: Theory, Models and Some New Findings (pp. 9–26). Boulder: Westview. Borrie, W. 1959 The Cultural Integration of Immigrants. Paris: UNESCO. Brody, E. 1969 Migration and Adaptation: The Nature of the Problem. In E. Brody, ed., Behavior in New Environments: Adaptation of Migrant Populations (pp. 13–21). Beverly Hills: Sage. 1973 The Lost Ones. Social Forces and Mental Illness in Rio de Janeiro. New York: International University Press. Campos-Filho, N., and E. Franco 1989 A Microcomputer Program for Multiple Regression by Unconditional and Conditional Maximum Likelihood Methods. American Journal of Epidemiology 129: 439–444. Canino, G., H. Bird, M. Rubio-Stipec, M. Bnaro, and P. Shrout 1987 The Prevalence of Specific Psychiatric Disorders in Puerto Rico. Archives of General Psychiatry 44: 727–733. Caravedo B., H. Rotondo, and J. Mariátegui 1963 Estudios de Psiquiatria Social en el Peru. Lima: Ediciones del Sol. Cassel, J., R. Patrick, and D. Jenkins 1960 Epidemiological Analysis of the Health Implications of Culture Change: A Conceptual Model. Annals of the NY Academy of Sciences 84: 938–949. Castells, M. 1977 Imperialismo y Urbanizacion en America Latina. Barcelona: Editorial Gustavo Gilli. 1992 A Economia Informacional: A Nova Divisão Internacional do Trabalho e o Projeto Socialista. Caderno CRH 17: 5–34. CEPAL 1991 Panorama Social de America Latina. Santiago de Chile: Naciones Unidas – Comisi´on Economica para America Latina y Caribe.
314
NAOMAR ALMEIDA-FILHO
1992 El Perfil de la Pobreza en America Latina a Comienzos de los Años 90. Santiago de Chile: Naciones Unidas – Comisi´on Economica para America Latina y Caribe. Coutinho, D. 1976 Prevalência de Doenças Mentais em uma Cpmunidade Marginal: Um Estudo do Maciel. Tese de Mestrado em Medicina, Universidade Federal da Bahia, Salvador, 86 p. Dean, A.G., J.A. Dean, A.H. Burton, and R.C. Dicker 1990 EPIINFO version 5.01b. Stone Mountain, Georgia: USD Inc. DeVos, G. 1976 Responses to Change. New York: Van Nostrand. DiMarco, G. 1982 Prevalencia de desordenes mentales en el area metropolitana de la Republica Argentina. Acta Psiquiatrica y Psicologica de America Latina 28(2): 93–102. Freeman, G., and L. Soete 1994 Work for All or Mass Unemployment? London: Pinter Publishers. Gallegos, C., and H.A. Miguez 1984 Prevalencia de Alteraciones Psiquiatricas: Estudio Piloto. Revista Costarricense de Ciencias Medicas 5(1), 37–43. Giddens, A. 1990 The Consequences of Modernity. Cambridge: Polity Press. 1992 Modernity and Self-Identity. Cambridge: Polity Press. Gonzalez, V., A. Yamhure, and R. Garcia 1978 Epidemiologia de los Trastornos Mentales en Bogota. Acta Psiquiatrica y Psicologica de America Latina 24(2): 93–99. Graves, T. 1970 The Personal Adjustment of Navajo Indian Migrants to Denver, Colorado. The American Anthropologist 72: 35–54. Grimson, W.R., A.C. Blanco, and M. Estrugamou 1972 Investigacion Epidemiologica de Entidades Psiquiatricas. Boletin de la Oficina Sanitaria Panamericana 73(6): 572–585. Harris, N. 1986 The End of the Third World. London: Penguin Books. Hosmer, D.W., and S. Lemeshow 1989 Applied Logistic Regression. New York: John Wiley & Sons. Inkeles, A., and D. Smith 1970 The Fate of Personal Adjustment in the Pprocess of Modernization. International Journal of Comparative Sociology 2(1): 81–113. 1974 Becoming Modern: Individual Change in Six Developing Countries. Boston: Harvard University Press. Jameson, F. 1992 Postmodernism, or the Cultural Logic of Late Capitalism. Durham: Duke University Press. 1994 The Constraints of Postmodernism. In: The Seeds of Time. New York: Columbia University Press. Kahn, H.A., and C.T. Sempos 1989 Statistical Methods in Epidemiology. New York: Oxford University Press. Kaplan, B., J. Cassel, and S. Gore 1977 Social Support and Health. Medical Care 15(5): 47–58.
SOCIAL CHANGE AND MENTAL HEALTH IN LATIN AMERICA
315
Leighton, A. 1959 My Name is Legion. New York: Basic Books. Leighton, D., J. Harding, D. Macklin, A. MacMillan, and A. Leighton 1963 The Character of Danger. Psychiatric Symptoms in Selected Communities. New York: Basic Books. Lennon, M.C. 1987 Sex Differences in Distress: The Impact of Gender and Work Roles. Journal of Health and Social Behavior 28: 290–305. Lyotard, J.F. 1985 The Post-Modern Condition. Minneapolis: University of Minnesota Press. Mari, J.J., N. Almeida-Filho, E. Coutinho, S.B. Andreoli, C.T. Miranda, and D. Streiner 1993 The Epidemiology of Psychotropic Use in the City of Sao Paulo. Psychological Medicine 23: 467–474. Mariátegui, J., and F. Samañez 1968 Sociocultural Change and Mental Health in the Peru of Today. Social Psychiatry 3(1): 35–40. Medeiros, C.A., and C. Salm 1994 O Mercado de Trabalho em Debate. Novos Estudos CEBRAP 39: 49–65. Micklin, M., and C. León 1978 Life Change and Psychiatric Disturbance in a South American City: Effects of Geographic and Social Mobility. Journal of Health and Social Behavior 19(1): 92–107. Miranda, C., J.J. Mari, M.R. Jorge, and J.A. Del Porto 1987 Versão Brasileira do Invent´ario de Sintomas do DSM-III. São Paulo: Escola Paulista de Medicina (mimeo). Oliveira, F. 1977 A Economia Brasileira: Cr´ıtica a` Rrazão Dualista. Cadernos CEBRAP 2: 3–110. Paykel, E.S. 1991 Depression in Women. British Journal of Psychiatry 158(suppl.): 22–29. Penayo, U., A. Caldera, and P. Jacobsson 1992 Prevalencia de Trastornos Mentales en Adultos de Subtiava, Le´on, Nicar´agua. Bolet´ın Of Sanit Panam 113(2): 137–149. PNUD 1993 Reforma Social y Pobreza – Hacia una Agenda Integrada de Desarollo. Washington: Programa de las Naciones Unidas para el Desarollo. Ponce, O.V. 1970 Migraci´on Interna a la Metr´opolis. Lima: Imprenta de la Universidad de San Marcos. Price, J. 1968 The Migration and Adaptation of American Indians to Los Angeles. Human Organization 27(2): 168–175. Robins, L., and D. Regier, eds. 1991 Psychiatric Disorders in America. New York: The Free Press. Rothman, K. 1986 Modern Epidemiology. Boston: Little Brown Co. Rotondo, H., J. Mariátegui, G. Bambaren, B. Caravedo, and M. Ofono 1963 Personalidad B´asica, Migracio´ n y Salud Mental. Boletin de la OPAS 52(2): 140– 142.
316
NAOMAR ALMEIDA-FILHO
Rosenfield, S. 1989 The Effect of Women’s Employment: Personal Control and Sex Differences in Mental Health. Journal of Health and Social Behavior 30(3): 77–91. Santana, V. 1982 Estudo Epidemiol´ogico das Doenças Mentais em um Bairro de Salvador. S´erie de Estudos em Sa´ude (Secretaria de Sa´ude da Bahia) 3: 122 p. Seguin, C. 1956 Migration and Psychosomatic Adaptation. Psychoanalytic Medicine 18: 404–410. SELA 1988 Las Consecuencias Sociales del Endeudamiento Externo de America Latina. Caracas: Sistema Econ´omico Latino Americano. Singer, P. 1975 Economia Pol´ıtica da Urbanização. São Paulo: Brasiliense. Souza, G.A., and V. Faria 1980 Bahia de Todos os Pobres. Petr´opolis: Vozes. Spindler, G. 1968 Psychocultural Adaptation. In E. Norbeck, ed., The Study of Personality: An Interdisciplinary Appraisal. New York: Holt, Rinehart & Winston. Srole, L., R. Langner, S. Michael, M.K. Opler, and T. Rennie 1962 Mental Health in the Metropolis. New York: McGraw-Hill. Srour, R. 1978 Modos de Produção: Elementos da Problem´atica. Rio de Janeiro: Graal. Tarnopolsky, A., R. Caetano, I. Levav, C. Campillo, J. Del Olmo, and L.U. Pinheiro 1977 Prevalence of Psychiatric Morbidity in an Industrial Suburb of Buenos Aires. Social Psychiatry 12(1): 75–88. Teske, R., and B. Nelson 1974 Acculturation and Assimilation: A Clarification. American Ethnologist 1(2): 351– 367. The World Bank 1989 Adult Health in Brazil: Adjusting to New Challenges. Washington DC: The World Bank, Brazil Department. Weppner, R. 1972 An empirical Test of the Assimilation of a Mirant Group into an Urban Milieu. Anthropological Quarterly 45(4): 262–273. Wittkower, E., and G. Dubreuil 1973 Psychocultural Stress in Relation to Mental Illness. Social Science and Medicine 7: 691–704.
Instituto de Saúde Coletiva Universidade Federal da Bahia Rua Padre Feijó, 29 – 4◦ andar Campus Canela – Salvador – Bahia Brazil – 40110-270