This study provides the first empirical data on the prevalence of mental disorders and associated severity levels in the adult community population living in households within the Brazil's largest metropolitan area, which may serve as a model of what might be seen in other megacities of the LAC region specifically, and in the developing world generally. The results reveal that mental disorders are notably prevalent and the estimated 10% prevalence of ‘severe’ cases indicates that in this megacity there are more than one million adults with impairment levels indicating special need for mental health care. Comorbidity is quite a common phenomenon, with most of the morbidity concentrated in around 40% of the active cases that present two or more disorders. In addition, this study offers (1) evidence on the burden of mental health in a developing country where prior epidemiological data are scarce; (2) a comparison of the results with estimates from other WMH surveys, since the same methods were applied in this consortium initiative; and (3) an examination of the relationships between psychiatric morbidities and facets of urban life, such as exposure to violence, neighborhood social deprivation, and migration status.
Compared to corresponding prevalence estimates of WMH-CIDI-diagnosed DSM-IV mental disorders from the other 23 participating countries of the WMH Survey 
, our estimate of 29.6% is larger than the corresponding value in the United States (26.2%) and about two times the estimates seen for the other upper-middle income participating countries 
. Also, by comparison with results from the other countries, the SPMA seems to have the largest proportion of severely affected cases (10%), well above the US estimate (5.7%), the New Zealand (4.7%) 
, and those from the 14 countries reported elsewhere 
In our megacity, the anxiety disorders qualify as the most frequently observed condition 
and major depression emerged as one of the most prevalent disorders, with higher estimated prevalence than has been seen elsewhere in other participant countries 
. The estimate of SUD prevalence in São Paulo (3.6%) is higher than Colombia's and Mexico's, the other two LAC countries in the WMH surveys, which reported estimates of 2.8% and 2.5%, respectively 
. With respect to impulse-control disorders, the SPMHS' prevalence of intermittent explosive disorders exceeds the estimate of US (3.1% vs. 2.6%) and stands as the highest IED prevalence estimate among the WMH sites that assessed this disorder 
The characteristics of our sample reflect the pattern of population growth of this megacity over the last decade: about one-half of the adult SPMA residents are in-migrants coming from other small cities and rural areas, most of them now living in suburban and peripheral deprived neighborhoods of the SPMA 
. Concurrently, there has been a widespread scaling-up of urban violence 
, increasing the feeling of insecurity among people living in the megacity. As it happens, the SPMHS-estimated level of exposure to violence rivals to what has been experienced in armed conflict countries such as Lebanon 
In Brazilian health statistics of recent years, violence and injuries have been found to be one of the main sources of morbidity and mortality by external causes 
. In our survey, crime-related events were found to be associated with all classes of mental morbidities and disorder severity, confirming previous reports that SUD, PTSD, and depression are frequent among individuals exposed to traumatic events 
. SUD may increase the risk of violence victimization, over and above any purported effects of SUD on crime or violent behavior 
A country's internal migration of workers due to economic reasons often has been described in relation to a ‘healthy migrant effect’ 
. This favorable pattern was observed in the SPMA, with migrants less prone to present mood and ICD, as compared to non-migrants in the same area. Nevertheless, the potential effects of migration were observed unevenly in some subgroups of our sample, with several determinants possibly interacting with each other, e.g., being male or female, urban/rural origin, and neighborhood context, and some vulnerable subgroups were disclosed in our exploration of product-terms. For example, migration-related stressors in combination with high NSD might work to increase the likelihood of being an active case of anxiety disorder in men. In this context, our observed patterns of male-female variations with respect to migration status, urbanicity, and NSD deserve more detailed future analysis, possibly with probing into issues that will clarify the forces that brought the in-migrants to the megacity, clarifying the temporal sequence of events and processes at play during the causal pathways that lead toward increasing risk, severity, or non-persistence of mental disorders.
In these SPMHS estimates, previous exposure to an urban environment is associated with increased odds of presenting an ICD, and to a lesser extent, mood disorders and more severe disorders. These findings may be consistent with earlier reports that psychiatric disorders are more common among the inhabitants of urbanized areas 
. A “breeder hypothesis” has been used to link the detrimental consequences of exposure to urbanicity to poor mental health status 
. Other WMH sites that have surveyed urban areas, such as Colombia and European locations 
, also found higher prevalence of mental disorders in more urbanized areas than less ones.
In contrast to this general pattern, non-migrant women raised in less urbanized areas of the SPMA seem to have been more vulnerable to mood disorders than women raised in more urbanized regions, or perhaps have more persisting mood disorder once it starts. Also, non-migrant women living in high NSD areas were also more likely to present an ICD than those from no/low NSD conditions. This might be due to the fact that in most peripheral deprived areas of the SPMA there is a predominance of woman-headed households with low education 
.Poverty among urban women may account for perpetuation of mechanisms of poor mental health 
The lack of male-female difference in ICD and drug dependence is in contrast with findings from other WMH countries 
, wherein for most externalizing disorders the estimates for men exceed those for women. Our data suggest a male-female convergence in externalizing disorders in the megacity, which might imply a growing burden of mental disorders in women 
. The findings of greater male-female differences in migrants from rural areas in mood disorders and migrants living in no/low NSD in anxiety disorders is consistent with previous reports that migration places women in a more vulnerable position in relation to men. How gender interact with other social contexts to shape health of migrant population is still an open matter 
With respect to age, most mental disorders, particularly the moderate/severe cases, were more common in early adulthood and midlife, suggesting impact on role-functioning during the important years of employment in the labor force 
. With respect to marital status, our finding that previously married residents were more likely to present an anxiety, mood, or ICD suggests the lack of social support of those divorced as one of complex pathways to mental disorders, as described by Kendler and colleagues in their research on depression in women 
. The association linking loneliness and poor social relationships (including separation/divorce/widow status) with ill-health outcomes and mortality was recently clarified in a meta-analysis 
. Changing marriage patterns, with increased social isolation, is considered both a predictor and a putative cause for poor mental health in urban areas 
, and of course, becoming separated or divorced may be a consequence of an active mental disorder as well.
Use of services
As expected, disorder severity was found to be related to treatment seeking and receipt of services, which we surmise to be linked with the distress and impairment that accompany mental disorders 
. Nonetheless, the majority of SPMA adults with active mental disorders did not receive treatment services. The finding of one third of those with serious disorders in the previous year receiving treatment is similar to findings in upper-middle income countries in the WMH survey consortium 
, but the SPMA estimate is not quite one-half the mean value observed in higher income countries.
Among severe/moderate cases, treatment in the specialty mental health sector was more common than general medical treatment, indicating an incipient mental service provision, contrasting with mental health care deregulation described in relation to our previous findings from a more circumscribed survey of neighborhoods in central São Paulo. Nevertheless, inequality and lack of integration also were observed within the SPMA 
. Possibly, the gap in mental health treatment in other regions of the country is even worse.
Of course, this study has some limitations, and a few of the more salient ones should be mentioned. First, data are not representative of Brazil nor of the world's megacities in general; however, the detailed assessment of the population needs in this area is important for further tailoring policies and strategies to improve the mental health of the population to be served 
. Second, the target population was restricted to people living in a large metropolitan area; generalization to rural or small city life is not warranted, even though an estimated 85% of the Brazilian population lives in urban areas 
. To the extent that these two limitations exist, they are likely to increase the prevalence rates.
Third, the migrant group is heterogeneous, coming from diverse settings. Different ages at migration, socio-economic condition, and lengths of residence in the SPMA could interfere in the adaptation and acculturation process. Future analyses will be carried on using survival models to account for time-varying and time-invariant characteristics.
Fourth, only residents in households were surveyed, whereas the homeless and those institutionalized were not assessed. Fifth, household surveys relying on self-report assessments may induce unwillingness to participate and of non-disclosure; for instance, for alcohol or other drug use and problems. To the extent that these two biases exist, it will make our estimate conservative.
Sixth, this report does not include some clinically important disorders - notably, non-affective psychosis and dementia. Although the WMH-CIDI inquired about psychotic symptoms, this information does not allow the diagnosis of non-affective psychosis. Previous studies have shown that these symptoms are overestimated in lay-administrated interviews 
. However, non-affective psychotic subjects might be captured as cases, as many are comorbid with anxiety, depression, and substance use disorders 
. Therefore, if severity is underestimated in the WMH-CIDI results will be conservative. The exclusion of elderly with cognitive impairments that was unable to answer the questionnaire did not allow detecting dementia, what can have lowered the rate of cases in this age group.
Finally, the cross-sectional nature of our data does not allow determining the direction of association of sociodemographic variables with disorders assessed herein.
This epidemiological survey of mental disorders experienced by adults living in a large and heterogeneous urban area has produced findings that may be a basis for current and future concern – not only in Brazil, but also in the LAC region, and perhaps in other megacities of the developing world. The observed estimates for the prevalence of mental disorders are among the largest ever seen in corresponding epidemiological surveys that have been conducted in other countries, with comparable field survey methods. A large proportion, one-third, of the active mental disorder cases qualify as ‘severe’ cases and most of these active and severe cases remain untreated. The heavy burden experienced by those with two or more disorders, as indicated by the association with severity, must be taken into account when planning services and prevention strategies.
These results call attention for the public health impact of mental disorders and offer an important foresight to stakeholders and health care providers 
. If the world human agglomeration will be settled mostly in large urban centers and megacities during the rest of this new century, the case of SPMA deserves attention as a potential forewarning of what might be occurring elsewhere.
The low rate of treatment suggests that the incipient integration of mental health promotion and care into the rapidly expanding Brazilian primary health system 
should be strengthened, reaching disadvantaged individuals without access to mental health services. Also, it is important to work with mental health promotion and early recognition of cases, particularly among young and males, those who are the group with less access to services in our survey.
Given the substantial burden of these mental health problems, it is important enhance the role of non-specialist health workers and other professionals, such as teacher and community leaders, in the recognition, detection and, eventually treatment of mental disorders. One potentially useful approach in poorly resourced countries is known as task-shifting or task-sharing 
. Under this approach, there is up-regulation of capacities of primary medical care providers and non-medical professionals for effective treatment of mental disorders; core packages of mental health services are integrated into routine primary care. When accompanied by careful supervision and mentoring by mental health specialists, this approach can be used to scale up the mental health workforce in highly populated developing countries, particularly in the context of disadvantaged or especially vulnerable groups living in more deprived areas that otherwise might be outside the reach of mental health specialists