 |
 |
|
|
|
 Many studies of rural–urban differences in mental disorders have been conducted to identify risk factors and social causes of mental illness. In the past, these studies often involved multidisciplinary teams who took into account psychiatric as well as social and cultural variables. Today, such studies are usually instigated by mental health service planners concerned about possible disparity in service provision between rural and urban areas, as well as about the regional organization of care, which is a particular concern at the European level. Most studies have reported a higher prevalence of mental disorders, particularly depression, in urban areas. This has been attributed to several factors: first, the decline in community relationships and social isolation in cities (1–3); second, greater stress relating to housing, work, marriage, child rearing, and security, combined with inadequate resources to cope with the stresses of urban life (4,5) and high urban levels of hostility (6); third, higher concentrations of poverty in city centres (7); and fourth, poor social integration and social withdrawal (1), coupled with sociocultural disintegration (including family and marital disintegration), which limit social networks (8). In addition, migration from rural to urban areas, which involves stress factors and coping resources as well as changes in culture, may also play a part (9,10). Several extensive mental health surveys of adult populations in the US and Canada have evaluated rural–urban differences in mental health. The milestone US Epidemiologic Catchment Area Study reported that individuals living in urban settings had a significantly higher risk of major depression (2.4%), as measured by the DIS (11), than did those living in rural areas (1.1%) (12). In a study conducted in Quebec, using DSM-III criteria, a significantly different point prevalence of major depression was noted for rural and urban areas (3.7%, compared with 1.1%) (13). This study suggested that the difference was mainly attributable to unemployed men and women without partners. The researchers were also able to differentiate 3 levels of urban–rural diversity in the prevalence of depression: a metropolis (Montreal); a small provincial city (Rimouski), where depression was the lowest; and a rural area (around Rimouski). This finding underlines the importance of how the notions of rural and urban are defined. The Stirling County Study supported the view that the proportion of psychiatric disorders is lower in rural and more integrated societies (6). However, a mental health survey conducted in the Edmonton metropolitan area found a lower 6-month prevalence of major depression (3.2%), compared with that found in the Stirling County Study, as measured by the DIS (14). Moreover, the National Comorbidity Survey found no difference between metropolitan areas, smaller cities, and rural areas when the current (30-day) and 12-month prevalence of major depression were compared (15,16). Similarly, no significant difference in the prevalence of major depression between rural and urban areas was seen in studies conducted in Taipei, Taiwan (17), and Seoul, Korea (18), nor was it seen in the Ontario Health Supplement Study (19). Several published reports on rural–urban differences and mental health issues describe studies from the UK. Over 2 decades ago, researchers collected data on depression in women from 2 highly differentiated samples: an urban group in a south London suburb and 2 other groups living on 2 Scottish islands, one of which included a small town (20,21). The results showed a significant decrease in depression associated with rural living. In addition, several environmental factors, specific for each sample (referred to as “provoking agents” and “major difficulties”), were shown to predict part of the variance. More recently, 2 of 3 nationwide surveys conducted in the UK compared psychiatric morbidity in urban and rural areas (22,23). In the Health and Lifestyle Survey, individual interviewers categorized psychiatric morbidity into 3 classes according to subjects’ place of residence: urban without open space, urban with open space, and rural (22). The ORs for psychiatric morbidity adjusted for sociodemographic variables supported the idea of rural living as a protective factor for mental health. The more recent National Morbidity Survey found a similar result, although in this study, rural living did not explain any significant amount of variance in the prevalence of either drug or alcohol dependence (23). Another recent large-scale survey was conducted in the Netherlands, where rural areas were defined according to national population density criteria (bottom 80% of counties); the results again supported an advantage for rural living (24). The ORs (adjusted for age and sex) for mood and substance use disorders as well as for comorbidity (2 or more disorders) were significantly lower in rural areas. Other European studies comparing depression in rural and urban areas have produced diverse results (21,24–29). Although most studies showed a higher prevalence of depression in large cities, compared with rural environments, the findings were by no means concordant, and the studies are difficult to compare because of methodological diversity. Inconsistencies among studies may be attributed to differences in several areas: the mental health measurement instruments used, definitions of depression and of urban and rural settings, methods of calculating prevalence, and sociodemographic characteristics across the rural study areas. In particular, depression was defined quite diversely in terms of the nosological entity studied, classification system, and method of diagnostic assignment. The distinction between depressive symptoms and depressive disorders seems important, since, for some authors, it is the former category that differs between rural and urban areas (30). The definition of urban and rural settings also has been a concern, given the variability in conceptualization across studies. The most widely used discriminants have been population density, interviewer judgment, and size of the communities where people live, defined by different cut-off thresholds (for example, 5000 or 10 000 inhabitants). Finally, the sociodemographic composition of urban and rural populations differs, for example, in age, marital status, and employment status. This needs to be controlled for carefully, since many of these variables have been shown to be independently associated with variance in the prevalence of mental health disorders. Attempts were made to eliminate this diversity in the European multicentre ODIN study, which interviewed a randomized sample of people residing in specified urban and rural areas in 4 European countries (Finland, Ireland, Norway, and the UK) in a 2-step procedure (screening plus standardized clinical interview) (31). The ODIN investigators found large urban–rural differences in the prevalence of depressive disorders in the British Isles (that is, in the UK and Ireland) but not in the 2 participating Nordic countries (Finland and Norway). They also observed notable differences between the urban sites studied, which did not appear between the rural sites. Compared with the corresponding rural site, a marked urban preponderance in the prevalence of depressive disorders was seen in women in the UK and Ireland, whereas, in men and in the total sample, this difference did not reach statistical significance. Logistic regression analysis including selected risk factors for depression showed an even higher risk of depressive disorders in both Dublin and Liverpool, compared with the Finnish urban site (Turku), which had the lowest urban prevalence. In addition, the ODIN investigators found that factors such as lack of confidence and difficulty in getting practical help from neighbours were important predictors of depressive disorders (32). Most of the previous studies, however, failed to consider disability or impairment levels when comparing the prevalence of major depression and depressive symptoms in rural and urban areas. Therefore, the findings provide limited information for mental health service planning and for understanding the etiology of depressive disorders (33). To address this issue, Wang analyzed data from the 1998–1999 Canadian NPHS (which collected data on mental health as well as on impairment) and reported that NPHS participants in rural areas showed a lower prevalence of MDE than those in urban areas, after controlling for the effects of race, immigration status, work, and marital status (34). Nonimmigrants and white participants in rural areas had a lower prevalence of MDE than did those in urban areas, and such differences depended on age and geographic region. Rural and urban participants did not differ in 2-week disability and interference in daily life due to depressive symptoms. The author concluded that the reasons for the rural–urban differences in the prevalence of MDE were complex and might depend on the individuals’ age, immigration status, race, work, marital status, and province of residence. A European study, the ESEMeD, has offered a unique opportunity to compare rural–urban differences in mental disorders in diverse countries, along with other relevant variables; it uses a common methodology and a similar definition of rural and urban areas across participating countries. This paper presents results of the ESEMeD study. It has 2 principal objectives: to determine whether rural–urban differences in mental health exist in European Union countries after sociodemographic variables are controlled for and to assess whether the characteristics of individuals with mental health disorders are similar in rural and urban areas, particularly concerning self-reported impairment. Material and MethodsThe ESEMeD 2000 study was a transversal survey carried out between 2001 and 2003 in the general population of Germany, Belgium, Spain, France, the Netherlands, and Italy. Subjects The target population consisted of noninstitutionalized individuals, aged 18 years and over, residing in private households in the 6 countries studied. A stratified, multistage, random sample (without replacement) was drawn in each country from the most representative national database available (that is, the electoral roll in Italy, the postal directory in the Netherlands, a randomly generated list of telephone numbers in France, and household registries in Belgium, Germany, and Spain.) Professional interviewers interviewed subjects at home. The weighted overall response rate was 61.2%. Data Collection The CIDI, a comprehensive, fully structured, diagnostic interview, was used to assess mental disorders (16,35). This allowed retrospective assignment of psychiatric diagnoses according to the ICD-10 (36) and DSM-IV (37) criteria. Using standard computerized algorithms, lifetime (12-month) and current (1 month before interview) prevalence rates were estimated. Two further questions were included to capture information on the DSM-IV criteria for impairment (Q1: “How severe was your emotional stress during those times: very severe, severe, mild, moderate?” Q2: “How often during those times was your emotional stress so severe that you could not carry out your daily activities: often, sometimes, rarely, never?”). In addition, psychological distress was assessed with the Mental Health subscale of the SF12, a short version of the SF36 that contains only 12 items. This instrument generates 2 summary scores: the PCS score, indicating the physical quality of life, and the MCS score, indicating the mental quality of life. Each scale uses all 12 items but with different weights. In this study, we used the MCS-50 version of the mental summary scale, which considers a score of 50 as neutral and generates positive and negative scores by subtracting 50 from the total score (a higher score indicates a better quality of life and therefore less psychological distress). The population size of the community where participants resided was recorded. Communities were classified as rural or urban according to the number of inhabitants determined from national census data. A bimodal dichotomization of above and below 10 000 inhabitants was used. Statistical Analysis We used STATA/SE V8.1 (38) for the statistical analysis. We calculated results with svy commands, which enabled calculation of the CI within the sampling design. A probability threshold of P < 0.05 was considered statistically significant. We compared categorical variables with the chi-square test. We used univariate and multivariate logistic regression analyses to compare diagnostic groups and to adjust for country and sociodemographic variables. ResultsRural–Urban Differences in Mental Health Across Countries Table 1 presents the sociodemographic distribution of the rural and urban populations in the different countries studied. In France, Germany, and Spain, the distribution of subjects by age group and marital status varied significantly between rural and urban areas. We therefore applied weighting to take into account these sociodemographic differences when calculating the prevalence of psychiatric disorders. For the unemployed category, we calculated rates from the workforce population only (excluding retirees, students, housewives, and others who had not worked for at least 6 months).
We compared 12-month prevalence rates of psychiatric disorders in men and women according to DSM-IV criteria (Table 2). Overall, we observed a higher risk of mental health disorders for subjects living in an urban environment. This association was particularly strong for depressive disorders and less so for anxiety disorders. In contrast, we observed no such association for alcohol-related disorders. However, these rural–urban differences were not identical in all countries. In France and Germany, higher rates of all types of disorder appeared in the rural sample, whereas in Belgium, rates were higher in the urban sample. We saw no rural–urban differences in the remaining countries. Looking at individual psychiatric disorders, we found rural–urban differences for mood disorders in France and Germany, and for anxiety disorders in France, but no differences for alcohol-related disorders. We also noted a sex interaction, finding significant rural–urban differences only in men, except for those in Belgium, where the urban preponderance of psychiatric disorders was restricted to women.
Sensitivity to the Definition of Rural and Urban We reanalyzed the data using a cut-off threshold of 5000 inhabitants to define rural and urban communities (France was the only country where this was possible). This yielded qualitatively similar differences in prevalence rates of psychiatric disorders between rural and urban areas as in the principal analysis (data not shown). We performed a multivariate regression analysis to take into account the potential influence of sociodemographic variables. This yielded an OR for having any mental disorder over 12 months of 1.29 (95%CI, 0.98 to 1.69) in the urban population, compared with the rural population, and a corresponding OR for a mood disorder of 1.08 (95%CI, 0.73 to 1.59). Neither OR was statistically significant; hence, using a more restrictive definition did not change the results. In a second step, we reanalyzed the data for the 3 countries in which we observed rural–urban differences (Belgium, France, and Germany), assigning communities to 1 of 3 classes: metropolis (>100 000 inhabitants), medium-sized city (10 000 to 100 000 inhabitants), or rural (<10 000). The resulting prevalence values were somewhat different from those obtained from a bimodal definition of rural and urban (Table 3). For example, the lower risk of psychiatric disorder associated with urban living among women in Belgium no longer appeared for those living in a metropolis, remaining only for those in medium-sized cities. In addition, this lower risk associated with women in medium-sized Belgian cities also emerged for anxiety disorders. Similarly, in France, it was only in medium-sized cities that we observed an altered risk of psychiatric disorders, compared with rural areas. Living in a medium-sized city in Germany was associated with an elevated risk of mood disorders, and living in a metropolis was associated with an elevated risk of alcohol-related disorders.
Sociodemographic Determinants of Rural–Urban Differences For those countries with demographic differences between rural and urban populations (France, Germany, and Spain), we used multivariate logistic regression to control for these differences (Table 4). This approach confirmed the rural– urban differences in prevalence of any mental disorder observed for France and Belgium. In contrast, the rural–urban difference seen in Germany seemed to be due to differences in marital status. An independent effect of marital status also appeared in France (with higher prevalence in previously married subjects), whereas in Spain, an age effect occurred (with higher prevalence in younger subjects).
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||