Abstract
OBJECTIVE
To evaluate the individual and contextual determinants of the use of health care services in the metropolitan region of Sao Paulo.
METHODS
Data from the Sao Paulo Megacity study – the Brazilian version of the World Mental Health Survey multicenter study – were used. A total of 3,588 adults living in 69 neighborhoods in the metropolitan region of Sao Paulo, SP, Southeastern Brazil, including 38 municipalities and 31 neighboring districts, were selected using multistratified sampling of the non-institutionalized population. Multilevel Bayesian logistic models were adjusted to identify the individual and contextual determinants of the use of health care services in the past 12 months and presence of a regular physician for routine care.
RESULTS
The contextual characteristics of the place of residence (income inequality, violence, and median income) showed no significant correlation (p > 0.05) with the use of health care services or with the presence of a regular physician for routine care. The only exception was the negative correlation between living in areas with high income inequality and presence of a regular physician (OR: 0.77; 95%CI 0.60;0.99) after controlling for individual characteristics. The study revealed a strong and consistent correlation between individual characteristics (mainly education and possession of health insurance), use of health care services, and presence of a regular physician. Presence of chronic and mental illnesses was strongly correlated with the use of health care services in the past year (regardless of the individual characteristics) but not with the presence of a regular physician.
CONCLUSIONS
Individual characteristics including higher education and possession of health insurance were important determinants of the use of health care services in the metropolitan area of Sao Paulo. A better understanding of these determinants is essential for the development of public policies that promote equitable use of health care services.
Keywords: Health Services, utilization; Health Services Accessibility; Health Inequalities; Social Conditions; Social Inequity; Metropolitan Zones; Multilevel Analysis
INTRODUCTION
The Brazilian Unified Health System (SUS), established in 1988, was designed to be a decentralized system with an emphasis on community services, including the Estratégia Saúde da Família (ESF – Family Health Strategy). At present, SUS suffers from underfunding and lack of trained personnel, particularly in disadvantaged areas, and this has been limiting the equitable access of poor population groups to health care units. 5 On the other hand, increased use of health care services is observed in regions with better socioeconomic conditions, particularly for services related to specialized medical care provided by private health plans. 2
The unequal use of health care services can affect the society as a whole. The systematic exclusion of population groups from health care services can lead to the emergence and dissemination of new diseases, as observed in cases of limited access to immunization schemes. a Therefore, a study of the determinants of the use of health care services is essential for identification of population groups with no or limited access to these services and can help develop public health policies.
Previous studies have analyzed the individual determinants of the use of health care services and have identified historically excluded groups, including low-income, poorly educated, and immigrant groups. 10 However, the determinants of health and access to health care services are not restricted to individual factors. Multilevel analyses that evaluate the contextual determinants of the use of health care services are also necessary to better understand the complex network of access to these services. However, few Brazilian studies have addressed these topics. 21
Other Brazilian studies have identified some individual determinants of access to health care services. An analysis of the use of dental care services indicated that wealthier individuals consulted dentists 2.8 times more often than poorer individuals in the past 12 months. 4 A study involving older people living in the municipality of Sao Paulo concluded that possession of health insurance was a determining factor for the use of health care services. 16 Another study analyzed data from the Pesquisa Nacional de Amostragem de Domicilios (PNAD – National Household Survey) and reported that the use of health care services was 1.8 times higher among patients with chronic diseases. 1
Individual income (ability to pay for health care services) is often cited as the main factor involved in access to health care services in the international literature, particularly in the United States. b Studies conducted in countries with universal access to health services have helped to identify contextual factors that influence the use of health care services using a multilevel methodology. A study conducted in Canada revealed that the use of mental health services was higher among individuals with mental illnesses living in districts with better socioeconomic conditions. 18 Furthermore, a French study showed that residents of the richest areas use health care services more often, even after adjustments for individual characteristics. 6
The place of residence can affect the use of health care services for several reasons. The geographical distribution and local availability of health care units can create barriers to the use of health services, and a short walking distance to health units is a good predictor of their use. 17 In addition, other studies have shown the importance of social capital (community participation and social cohesion). More equalitarian neighborhoods and those with less violence have greater social capital, which is associated with increased use of health services, 13 possibly because of the better information network on health care systems available for the local population. Another hypothesis is that individuals who live in neighborhoods with worse socioeconomic conditions have more health problems because of pollution and violence and therefore are more likely to use health services. 23
The presumed universality of access and the known social inequality allow for interesting opportunities for analysis of the contextual determinants of the use of health services in Brazil, particularly with regard to income inequality and violence, in which the Brazilian indicators are consistently worse compared with other developed countries. c Previous studies have indicated the contextual importance of income inequality and violence on the health of residents of Sao Paulo, 7 , 8 and this should stimulate an analysis of the use of health care services as a possible mediator of the correlation between contextual factors and health.
The aim of this study was to analyze the prevalence and factors associated with the use of health services by adults.
METHODS
We analyzed the results of the Sao Paulo Megacity study (Andrade et al 3 ). d The research was based on questionnaires given to a representative sample of the population aged ≥ 18 years living in the metropolitan region of Sao Paulo (MRSP), composed of the municipality of Sao Paulo (11,104,715 residents in 2007) and its 38 neighboring districts (8,844,543 residents).
Respondents were selected using multistratified sampling of the non-institutionalized population in the MRSP, comprising six distinct selection stages. The initial objective was to identify 5,000 households. 22 The questionnaires were administered between May 2005 and May 2007 by trained interviewers. The initial study population consisted of 5,037 individuals (response rate: 81.3%). Of these, 37 (0.7%) were excluded because of difficulties to identify the place of residence. Only subjects who lived in the same location for more than 5 years were included to avoid the misconception that the place of origin has immediate effect on the health and behavior of individuals. 11 A total of 3,588 individuals were included in the final analysis.
The questionnaires were administered during home visits with the support and supervision of the responsible academic staff. The collection instrument used was the Composite International Diagnostic Interview (WMH-CIDI) translated and adapted to Portuguese. 15 Use of health care services was analyzed by calculating the number of consultations with health professionals in the past 12 months, dichotomized between 0 and ≥ 1 owing to the distribution asymmetry, i.e., high frequency of zero values. Presence of a regular physician for routine care was analyzed by answering the following question: “Do you have a physician who you usually consult when you need routine care?”. The following individual variables were included in the multilevel models: age, gender, individual income (by tertiles because of non-linearity), and education (primary, high school, and higher education). Presence of chronic diseases in the past 12 months was evaluated, and each disease was assessed on an individual basis. The most prevalent diseases considered for analysis were as follows: cardiovascular diseases (heart attack, heart disease, hypertension, stroke), respiratory diseases (allergies, asthma, chronic obstructive pulmonary disease, and emphysema), and migraine. Major depression and anxiety disorders (panic disorder, agoraphobia, simple phobia, social phobia, generalized anxiety disorder, obsessive compulsive disorder, post-traumatic stress disorder, and separation anxiety) were the most prevalent mental illnesses in the initial study, identified using the WMH-CIDI questionnaire according to the operational criteria established by the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) 15 in the last 12 months.
The second level of the multilevel analysis included 38 adjacent districts (average of 232,751 residents in 2007) in the MRSP in addition to the municipality of Sao Paulo (MSP), divided according to its 31 neighborhoods (average of 355,467 residents), totaling 69 areas. Median income, level of income inequality (measured by the Gini index), and level of violence (using the homicide rate adjusted for age as proxy) in each of the 69 neighborhoods were calculated using data from the Brazilian Institute of Geography and Statistics (IBGE) Census of 2010 e and from Departamento de Informática do SUS (DATASUS –Department of Informatics of SUS). f
Multilevel Bayesian logistic models were adjusted using individuals as the first level and place of residence (municipality or districts) as the second level. Bayesian inference was adopted using the Markov Chain Monte Carlo (MCMC) method because of its effect on the decrease of bias in the multilevel logistic models. 20 It also allowed calculation of the deviance information criterion (DIC) to test the model fit (goodness-of-fit), in which lower values indicate better fit.
The multilevel models were calculated separately for the two dependent variables: use of health care services in the past 12 months and presence of a regular physician for routine care. The models were initially adjusted without inclusion of the independent variables (null model) to test the initial variance attributable to the place of residence. Subsequently, the models were adjusted using the individual variables and contextual variables. After the model with the best fit (lower DIC) was identified, the most prevalent chronic diseases (cardiovascular diseases, respiratory diseases, and migraine) and mental diseases (anxiety disorder and major depression) were included.
MLwiN 2.25 software, specialized in multilevel analyses, was used for statistical analysis. The models were adjusted using least squares generalized for the distribution values. The first 500 simulations were discarded as burn-ins, followed by 10,000 new iterations. Median odds ratios (MOR) were calculated to assess the percentage of variance attributable to the place of residence (second level of the model).
RESULTS
The final study group consisted of 3,588 individuals. Of these, 56.3% were women, 84.0% were under 60 years, 14.9% had complete or incomplete higher education, and 41.4% reported having health insurance (Table 1).
Table 1. Profile of the study population that used health care services in the past 12 months and the presence of a regular physician for routine care. Metropolitan region of Sao Paulo, Southeastern Brazil, 2007.
Variable | Total | Use of health care services | Regular physician | |||
---|---|---|---|---|---|---|
| ||||||
n | % | n | % | n | % | |
Total | 3,588 | 100 | 2,995 | 83.5 | 1,702 | 47.4 |
Gender | ||||||
Male | 1,569 | 43.7 | 1,182 | 75.3 | 647 | 41.2 |
Female | 2,019 | 56.3 | 1,813 | 89.8* | 1,055 | 52.2* |
Age (years) | ||||||
18 to 39 | 1,541 | 42.9 | 1,257 | 81.6 | 622 | 40.4 |
40 to 59 | 1,474 | 41.1 | 1,232 | 83.6 | 749 | 50.8* |
≥ 60 | 573 | 16.0 | 506 | 88.3* | 331 | 57.8* |
Education | ||||||
Primary education | 1,918 | 53.5 | 1,573 | 82.0 | 853 | 44.5 |
High school education | 1,136 | 31.7 | 950 | 83.6 | 534 | 47.0 |
College/University education | 534 | 14.9 | 472 | 88.4* | 315 | 59.0* |
Health insurance | ||||||
Yes | 1,486 | 41.4 | 1,334 | 89.8 | 1,037 | 69.8 |
No | 2,102 | 58.6 | 1,661 | 79.0* | 665 | 31.6* |
Income | ||||||
Low | 1,175 | 32.7 | 947 | 80.6 | 469 | 39.9 |
Average | 1,211 | 33.7 | 1,011 | 83.5 | 572 | 47.2* |
High | 1,202 | 33.5 | 1,037 | 86.3* | 661 | 55.0* |
Income inequality | ||||||
Low | 1,219 | 34.0 | 1,024 | 84.0 | 576 | 47.2 |
Average | 1,216 | 33.9 | 995 | 81.8 | 579 | 47.6 |
High | 1,153 | 32.1 | 976 | 84.6 | 547 | 47.4 |
Median income | ||||||
Low | 1,223 | 34.1 | 1,022 | 83.6 | 562 | 45.9 |
Average | 1,198 | 33.4 | 984 | 82.1 | 549 | 45.8 |
High | 1,167 | 32.5 | 989 | 84.7 | 591 | 50.6* |
Violence | ||||||
Low | 1,184 | 33.0 | 1,013 | 85.6 | 618 | 52.2 |
Average | 1,205 | 33.6 | 1,006 | 83.5 | 548 | 45.5* |
High | 1,199 | 43.7 | 976 | 81.4* | 536 | 44.7* |
* Significant difference (p < 0.05) using simple logistic regression analysis in relation to the first category of the variable.
More than 80.0% of the individuals reported at least one medical visit in the previous year. Simple logistic regression analysis indicated that women, individuals above 60 years of age, those with health insurance, those having higher education, and those with higher income were more likely (p < 0.05) to have consulted a physician in the past year. On the other hand, individuals living in regions with high violence were less likely to have consulted in the past year.
In addition, 47.4% of the individuals reported having a regular physician for routine care. Simple logistic regression analysis also indicated that the correlation was statistically significant for women, individuals belonging to older age groups, those with higher education, those with health insurance, those with average individual income, and those living in areas with high median income and low violence. Presence of a physician for routine care was strongly correlated with possession of health insurance. In addition, 69.8% individuals with health insurance reported having a regular physician in contrast with 31.6% individuals without health insurance.
Table 2 presents the multilevel models for the individual and contextual determinants of the use of health care services in the past 12 months. In the null model, the variance attributable to the place of residence (second level) was not statistically significant (p > 0.05) and the MOR was 1.24, i.e., area heterogeneity increased the chance of using a health care service in the past year by 24.0% for an individual chosen at random. The DIC was 3211.24. Inclusion of individual characteristics (model 1) decreased the DIC to 3043.99, indicating better model fit. Women, individuals with higher education, those with higher income, and those belonging to older age groups were more likely to have consulted a health professional in the past year (p < 0.05). Health insurance was included in model 2 with a statistically significant presence. Model 2 showed the best fit (DIC = 2990.11) (Table 2). Inclusion of contextual variables (models 3, 4, and 5) did not produce statistically significant results. A marginally significant result (OR: 0.78, 95%CI 0.60;1.02) was observed only for average income inequality (compared with low income inequality). The variance in the second level (place of residence) was not significant (p > 0.05) for the models with contextual variables, and the DIC value indicated worse fit compared with model 2.
Table 2. Multilevel models for the individual and contextual determinants of the use of health care services in the past 12 months. Metropolitan region of Sao Paulo, Southeastern Brazil, 2007.
Variable | Null model | Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||
OR | 95%CI | OR | 95%CI | OR | 95%CI | OR | 95%CI | OR | 95%CI | |||
Gender | ||||||||||||
Male | 0.33 | 0.28;0.40 | 0.33 | 0.27;0.40 | 0.33 | 0.27;0.40 | 0.33 | 0.27;0.40 | 0.33 | 0.27;0.40 | ||
Age | 1.02 | 1.01;1.02 | 1.01 | 1.01;1.02 | 1.01 | 1.01;1.02 | 1.02 | 1.01;1.02 | 1.02 | 1.01;1.02 | ||
Education | ||||||||||||
Primary education | ||||||||||||
High school education | 1.43 | 1.15;1.79 | 0.23 | 1.00;1.60 | 1.25 | 0.99;1.57 | 1.27 | 1.01;1.60 | 1.27 | 1.01;1.60 | ||
College/University education | 1.77 | 1.29;2.45 | 1.37 | 0.97;1.93 | 1.37 | 0.97;1.93 | 1.38 | 1.00;1.91 | 1.36 | 0.98;1.89 | ||
Individual income | ||||||||||||
Low | ||||||||||||
Average | 1.20 | 0.97;1.49 | 1.08 | 0.86;1.36 | 1.09 | 0.88;1.36 | 1.10 | 0.88;1.38 | 1.09 | 0.87;1.37 | ||
High | 1.30 | 1.02;1.65 | 1.09 | 0.84;1.40 | 1.10 | 0.86;1.42 | 1.11 | 0.86;1.43 | 1.09 | 0.85;1.41 | ||
Health insurance | 2.21 | 1.79;2.73 | 2.24 | 1.80;2.78 | 2.21 | 1.79;2.74 | 2.23 | 1.69;2.94 | ||||
Median income | ||||||||||||
Low | ||||||||||||
Average | 0.81 | 0.61;1.07 | ||||||||||
High | 0.86 | 0.64;1.15 | ||||||||||
Income inequality | ||||||||||||
Low | ||||||||||||
Average | 0.78 | 0.60;1.02 | ||||||||||
High | 0.91 | 0.70;1.19 | ||||||||||
Violence | ||||||||||||
Low | ||||||||||||
Average | 1.03 | 0.78;1.37 | ||||||||||
High | 0.97 | 0.73;1.27 | ||||||||||
Contextual variance (SD) | 0.050 | (0.035) | 0.041 | (0.031) | 0.053 | (0.033) | 0.055 | (0.034) | 0.038 | (0.034) | 0.053 | (0.033) |
Mean odds ratio | 1.24 | 1.09;1.48 | 1.21 | 1.03;1.39 | 1.25 | 1.11;1.42 | 1.26 | 1.09;1.42 | 1.20 | 1.03;1.40 | 1.25 | 1.06;1.41 |
DIC | 3.211.24 | 3.043.99 | 2.990.11 | 2.990.93 | 2.991.70 | 2.992.91 |
DIC: deviance information criterion
Table 3 presents the multilevel models for the use of health care services in the past year with the inclusion of the most prevalent chronic diseases and mental disorders. Model 2 (Table 2) was included as the starting point in all cases because it presented the best fit. Chronic and mental illnesses were analyzed individually and showed statistically significant results in models 6, 7, 8, 9, and 10. As in previous models, the variance in the second level was not significant (p > 0.05) in all cases. The model with the best fit was the one that included cardiovascular diseases (DIC = 2919.60). In model 11, which included all chronic and mental diseases evaluated, only depression was not statistically significant (OR: 1.21, 95%CI 0.82;1.78). Furthermore, model 11 presented the best fit among the models analyzed (DIC = 2882.44).
Table 3. Multilevel models for the correlation between the presence of chronic and mental illnesses and the use of health care services in the past 12 months. Metropolitan region of Sao Paulo, Southeastern Brazil, 2007.
Variable | Model 6 | Model 7 | Model 8 | Model 9 | Model 10 | Model 11 | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||
OR | 95%CI | OR | 95%CI | OR | 95%CI | OR | 95%CI | OR | 95%CI | OR | 95%CI | |
Gender | ||||||||||||
Male | 0.34 | 0.28;0.42 | 0.35 | 0.29;0.43 | 0.37 | 0.30;0.44 | 0.34 | 0.30;0.38 | 0.34 | 0.29;0.41 | 0.40 | 0.33;0.48 |
Age | 1.00 | 1.00;1.01 | 1.01 | 1.01;1.02 | 1.02 | 1.01;1.02 | 1.01 | 1.01;1.02 | 1.01 | 1.01;1.02 | 1.01 | 1.00;1.01 |
Education | ||||||||||||
Primary education | ||||||||||||
High school education | 1.27 | 1.00;1.61 | 1.23 | 0.96;1.57 | 1.32 | 1.05;1.66 | 1.27 | 1.01;1.60 | 1.29 | 1.01;1.64 | 1.29 | 1.02;1.62 |
College/University education | 1.43 | 1.02;1.99 | 1.30 | 0.92;1.83 | 1.43 | 1.02;2.00 | 1.35 | 0.98;1.87 | 1.35 | 0.96;1.90 | 1.42 | 1.01;1.99 |
Individual income | ||||||||||||
Low | ||||||||||||
Average | 1.14 | 0.91;1.44 | 1.09 | 0.88;1.37 | 1.10 | 0.87;1.38 | 1.09 | 0.87;1.37 | 1.08 | 0.86;1.36 | 1.15 | 0.90;1.45 |
High | 1.16 | 0.90;1.50 | 1.09 | 0.85;1.41 | 1.12 | 0.87;1.44 | 1.10 | 0.85;1.42 | 1.10 | 0.85;1.42 | 1.16 | 0.90;1.51 |
Health insurance | 2.32 | 1.87;2.86 | 2.21 | 1.77;2.77 | 2.22 | 1.78;2.76 | 2.23 | 1.86;2.67 | 2.24 | 1.81;2.78 | 2.29 | 1.83;2.86 |
Cardiovascular disease | 3.08 | 2.32;4.08 | 2.74 | 2.09;3.61 | ||||||||
Respiratory disease | 1.99 | 1.53;2.59 | 1.77 | 1.36;2.31 | ||||||||
Migraine | 1.70 | 1.35;2.16 | 1.41 | 1.11;1.78 | ||||||||
Depression | 1.63 | 1.14;2.33 | 1.21 | 0.82;1.78 | ||||||||
Anxiety | 1.85 | 1.40;2.45 | 1.54 | 1.14;2.07 | ||||||||
Contextual variance (SD) | 0.050 | (0.040) | 0.052 | (0.035) | 0.45 | (0.037) | 0.053 | (0.033) | 0.047 | (0.033) | 0.051 | (0.036) |
Mean odds ratio | 1.24 | 1.04;1.44 | 1.24 | 1.08;1.42 | 1.22 | 1.03;1.41 | 1.25 | 1.03;1.40 | 1.23 | 1.03;1.41 | 1.24 | 1.00;1.42 |
DIC | 2.919.6 | 2.965.09 | 2.972.92 | 2.991.70 | 2.973.26 | 2.882.44 |
DIC: deviance information criterion
Tables 4 and 5 followed the same methodology to evaluate the presence of a physician for routine care. As in the other variables, the variance attributable to the place of residence was not statistically significant (p > 0.05) in any model evaluated. Women, individuals belonging to older age groups, and those with more education had a greater correlation with the presence of a regular physician in all models evaluated. The variable of having health insurance was most strongly correlated with the presence of a regular physician in all models evaluated. The contextual characteristics of the place of residence showed no significant correlation with the presence of a regular physician. The exception was income inequality: living in regions with high income inequality was correlated with a lower probability of having a regular physician after controlling for individual factors (Table 4, model 4). In the models that included only one chronic or mental illness (Table 5, models 6-10), presence of cardiovascular or respiratory diseases was correlated with having a regular physician. Only cardiovascular diseases were significantly correlated with having a regular physician when all diseases were included in the model.
Table 4. Multilevel models for the individual and contextual determinants of the presence of a regular physician for routine care. Metropolitan region of Sao Paulo, Southeastern Brazil, 2007.
Variable | Null model | Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||
OR | 95%CI | OR | 95%CI | OR | 95%CI | OR | 95%CI | OR | 95%CI | |||
Gender | ||||||||||||
Male | 0.61 | 0.53;0.70 | 0.58 | 0.50;0.68 | 0.58 | 0.50;0.67 | 0.58 | 0.50;0.67 | 0.58 | 0.50;0.67 | ||
Age | 1.03 | 1.02;1.03 | 1.03 | 1.02;1.03 | 1.03 | 1.02;1.03 | 1.03 | 1.02;1.03 | 1.03 | 1.02;1.03 | ||
Education | ||||||||||||
Primary education | ||||||||||||
High school education | 1.56 | 1.32;1.84 | 1.24 | 1.03;1.49 | 1.23 | 1.02;1.49 | 1.24 | 1.03;1.49 | 1.23 | 1.02;1.49 | ||
College/University education | 2.02 | 1.61;2.52 | 1.25 | 0.98;1.59 | 1.27 | 1.00;1.61 | 1.27 | 1.00;1.63 | 1.26 | 0.97;1.63 | ||
Individual income | ||||||||||||
Low | ||||||||||||
Average | 1.24 | 1.05;1.46 | 0.95 | 0.79;1.15 | 0.96 | 0.80;1.16 | 0.95 | 0.79;1.15 | 0.95 | 0.79;1.14 | ||
High | 1.40 | 1.17;1.67 | 0.88 | 0.71;1.08 | 0.90 | 0.74;1.09 | 0.88 | 0.72;1.09 | 0.88 | 0.72;1.08 | ||
Health insurance | 5.31 | 4.49;6.27 | 5.32 | 4.52;6.28 | 5.33 | 4.54;6.26 | 5.32 | 4.54;6.23 | ||||
Median income | ||||||||||||
Low | ||||||||||||
Average | 0.87 | 0.68;1.12 | ||||||||||
High | 0.83 | 0.65;1.05 | ||||||||||
Income inequality | ||||||||||||
Low | ||||||||||||
Average | 0.89 | 0.70;1.12 | ||||||||||
High | 0.77 | 0.60;0.99 | ||||||||||
Violence | ||||||||||||
Low | ||||||||||||
Average | 1.01 | 0.78;1.30 | ||||||||||
High | 1.02 | 0.79;1.33 | ||||||||||
Contextual variance (SD) | 0.043 | (0.030) | 0.031 | (0.021) | 0.058 | (0.032) | 0.059 | (0.033) | 0.056 | (0.031) | 0.066 | (0.036) |
Mean odds ratio | 1.22 | 1.03;1.38 | 1.18 | 1.04;1.09 | 1.26 | 1.11;1.42 | 1.26 | 1.11;1.43 | 1.26 | 1.10;1.41 | 1.28 | 1.12;1.45 |
DIC | 4.955.52 | 4.758.92 | 4.310.18 | 4.310.62 | 4.309.70 | 4.312.08 |
DIC: deviance information criterion
Table 5. Multilevel models for the correlation between the presence of chronic and mental illnesses and the presence of a regular physician for routine care. Metropolitan region of Sao Paulo, Southeastern Brazil, 2007.
Variable | Model 6 | Model 7 | Model 8 | Model 9 | Model 10 | Model 11 | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||
OR | 95%CI | OR | 95%CI | OR | 95%CI | OR | 95%CI | OR | 95%CI | OR | 95%CI | |
Gender | ||||||||||||
Male | 0.59 | 0.51;0.69 | 0.59 | 0.51;0.69 | 0.60 | 0.51;0.69 | 0.59 | 0.50;0.68 | 0.59 | 0.51;0.68 | 0.61 | 0.52;0.72 |
Age | 1.02 | 1.01;1.03 | 1.02 | 1.02;1.03 | 1.03 | 1.02;1.03 | 1.03 | 1.02;1.03 | 1.03 | 1.02;1.03 | 1.02 | 1.01;1.03 |
Education | ||||||||||||
Primary education | ||||||||||||
High school education | 1.25 | 1.03;1.51 | 1.21 | 1.01;1.45 | 1.24 | 1.03;1.49 | 1.24 | 1.03;1.49 | 1.23 | 1.03;1.47 | 1.26 | 1.03;1.52 |
College/University education | 1.30 | 1.01;1.66 | 1.22 | 0.96;1.56 | 1.25 | 0.98;1.61 | 1.26 | 0.98;1.61 | 1.24 | 0.97;1.59 | 1.30 | 1.00;1.68 |
Individual income | ||||||||||||
Low | ||||||||||||
Average | 0.97 | 0.80;1.17 | 0.95 | 0.79;1.14 | 0.96 | 0.80;1.16 | 0.94 | 0.79;1.13 | 0.95 | 0.79;1.14 | 0.97 | 0.81;1.16 |
High | 0.91 | 0.73;1.12 | 0.88 | 0.72;1.08 | 0.89 | 0.73;1.10 | 0.87 | 0.72;1.06 | 0.88 | 0.72;1.08 | 0.90 | 0.73;1.09 |
Health insurance | 5.45 | 4.61;6.45 | 5.32 | 4.52;6.26 | 5.32 | 4.54;6.23 | 5.33 | 4.52;6.28 | 5.32 | 4.54;6.24 | 5.45 | 4.66;6.38 |
Cardiovascular disease | 1.62 | 1.36;1.94 | 1.58 | 1.31;1.90 | ||||||||
Respiratory disease | 1.19 | 1.00;1.41 | 1.14 | 0.95;1.36 | ||||||||
Migraine | 1.12 | 0.95;1.32 | 1.04 | 0.88;1.24 | ||||||||
Depression | 1.05 | 0.83;1.34 | 0.97 | 0.75;1.24 | ||||||||
Anxiety | 1.15 | 0.95;1.40 | 1.10 | 0.91;1.34 | ||||||||
Contextual variance (SD) | 0.066 | (0.034) | 0.066 | (0.034) | 0.057 | (0.032) | 0.060 | (0.032) | 0.060 | (0.032) | 0.063 | (0.034) |
Mean odds ratio | 1.28 | 1.11;1.44 | 1.28 | 1.13;1.44 | 1.26 | 1.10;1.41 | 1.26 | 1.12;1.42 | 1.26 | 1.11;1.42 | 1.27 | 1.09;1.43 |
DIC | 4.283.38 | 4.307.45 | 4.310.46 | 4.311.49 | 4.309.20 | 4.288.01 |
DIC: deviance information criterion
DISCUSSION
The results indicate that the contextual characteristics of the place of residence are not a statistically significant barrier to the use of health care services in the MRSP, contrary to the situation observed in other municipalities. 18 , 19 In contrast, individual factors, including education and possession of health insurance, were the determinants of the use of health care services in the past 12 months as well as the determinants of having a regular physician.
Recent studies have indicated a correlation between the place of residence in the MRSP and presence of chronic and mental diseases. 7 , 8 However, in the present study, use of health care services appeared to be unaffected by the place of residence. In the null model, without the inclusion of individual or contextual independent variables, the variance attributable to the place of residence was not statistically significant for any of the dependent variables studied. Furthermore, contextual variables (median income, income inequality, and violence) were not statistically significant for most cases, except for the correlation between living in a region with high inequality and lower probability of having a regular physician.
These results have implications for planning the distribution of health care services in the MRSP. Living in poorer, more unequal, or more violent areas does not appear to significantly affect the use of health care services. This may be the result of the planning strategies for health care services, such as the expansion of the ESF, which has benefited areas with worse socioeconomic conditions and reduced inequalities in access to health care in the past decade. 12
On the other hand, possession of health insurance had a strong and consistent importance in the use of health care services. Possession of health insurance was significantly correlated with the use of health care services in the past year and with the presence of a regular physician in all models. In addition, presence of chronic and mental illnesses was strongly correlated with the use of health care services in the past year. On the other hand, having a regular physician showed a weak correlation with the presence of chronic and mental illnesses, except for cardiovascular diseases.
Possession of private health insurance was correlated with an increased likelihood of using health care services and of having a regular physician even when individual variables such as education and income were included in the model. This indicates an independent correlation between possession of private health insurance and access to health care services and goes beyond the correlation with socioeconomic factors. This may be because individuals with more severe health problems are more likely to purchase health insurance (this phenomenon is known as adverse selection in economic theory). 24 Another possibility is that possession of health insurance unquestionably increases the use of health care services, particularly in relation to preventive and routine care. 14
Education was an important variable correlated with the use of health care services and its effect was greater than that found for individual income. This is in line with theories that argue that the effect of income on health would be, to some extent, a proxy for the effect of education on health, in which people with more education would be more willing to invest in their own health. 9 Therefore, regardless of the presence or absence of serious health problems (such as chronic and mental illnesses), individuals with higher education would use health care services more frequently, particularly for preventive and routine care.
The present study has some limitations. Use of health care services in the past year was analyzed retrospectively, i.e., the analysis depended on the memory of individuals. Despite the short study period (12 months), this analysis cannot overlook the occurrence of recall bias. Moreover, it was not possible to establish a temporal relationship between the onset of chronic and mental illnesses and the use of health care services. One possibility is that individuals living in areas with worse socioeconomic conditions or in areas with social inequalities would have access to health care services but would take more time to consult a health care professional after the onset of the disease. Another limitation is that it was not possible to assess the quality of health care services on an individual basis.
Use of and spending on health care services has increased worldwide, and the trend is that this growth will continue for decades. b Identification of determinants that preclude individuals from using health care services is an important issue in public health. The present study indicates that the contextual characteristics of the place of residence are not a significant barrier to the use of health care services in the MRSP. At the same time, education and possession of health insurance are important determinants of the use of these services. At present, when access to health care services for the Brazilian population is being extensively discussed in the media and in the political sphere, it is important to understand these determinants with the view of implementing policies that promote equitable use of these services.
Funding Statement
Research supported by the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP – Process 03/00204-3). The translation and development of the instrument were financed by the Fundação para a Ciência e Tecnologia de Vitória, Espírito Santo, Brazil (FACITEC – under Process 002/2003). The Sao Paulo Megacity Mental Health Survey was conducted in association with the World Mental Health Survey (WMHS) initiative. The main activities of the coordination center at Harvard University were funded by the United States National Institutes of Mental Health (R01MH070884), John D. and Catherine T. MacArthur Foundation, Pfizer Foundation, U.S. Public Health Service (R13-MH066849, R01-MH069864, and R01-DA016558), and the Fogarty International Center (Firca R03-TW006481).
Footnotes
Research supported by the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP – Process 03/00204-3). The translation and development of the instrument were financed by the Fundação para a Ciência e Tecnologia de Vitória, Espí rito Santo, Brazil (FACITEC – under Process 002/2003). The Sao Paulo Megacity Mental Health Survey was conducted in association with the World Mental Health Survey (WMHS) initiative. The main activities of the coordination center at Harvard University were funded by the United States National Institutes of Mental Health (R01MH070884), John D. and Catherine T. MacArthur Foundation, Pfizer Foundation, U.S. Public Health Service (R13-MH066849, R01-MH069864, and R01-DA016558), and the Fogarty International Center (Firca R03-TW006481).
United States. Department of Health and Human Services. National healthcare disparities report 2011. Rockville (MD): AHRQ Publications; 2012.
Clements B, Coady D, Gupta S. The economics of public health care reform in advanced and emerging economies. Washington (DC): International Monetary Fund Publications; 2012.
The World Bank. Data by country. Washington (DC); 2013 [cited 2013 Nov 1]. Available from: http://data.worldbank.org/country
This study is the Brazilian part of the World Mental Health Survey multicenter initiative, which began in 2001, involves 28 countries, and is coordinated by the World Health Organization (WHO). It was conducted in Brazil by the Psychiatric Epidemiology Center of the Instituto de Psiquiatria of the Hospital das Clínicas of the Faculdade de Medicina of the Universidade de São Paulo.
Instituto Brasileiro de Geografia e Estatística. Censo Demográfico 2010. Brasília (DF); 2013 [cited 2014 Nov 19]. Available from: http://www.sidra.ibge.gov.br/cd/cd2010RgaAdAgsn.asp
Departamento de Informática do SUS. Informações de saúde. Brasília (DF); 2014 [cited 2014 Nov 19]. Available from: http://www2.datasus.gov.br/DATASUS/index.php?area=02
REFERENCES
- 1.Almeida MF, Barata RB, Montero CV, Pereira Z. Prevalência de doenças crônicas auto-referidas e utilização de serviços de saúde, PNAD/1998, Brasil. 10.1590/S1413-81232002000400011Cienc Saude Coletiva. 2002;7(4):743–756. [Google Scholar]
- 2.Andrade LH, Viana MC, Tófoli LF, Wang YP. Influence of psychiatric morbidity and sociodemographic determinants on use of service in a catchment area in the city of São Paulo, Brazil. 10.1007/s00127-007-0263-3Soc Psychiatry Psychiatr Epidemiol. 2008;43(1):45–53. doi: 10.1007/s00127-007-0263-3. [DOI] [PubMed] [Google Scholar]
- 3.Andrade LH, Wang YP, Andreoni S, Silveira CM, Alexandrino-Silva C, Siu ER, et al. Mental disorders in megacities: findings from the Sao Paulo megacity mental health survey, Brazil. 10.1371/journal.pone.0031879PLoS ONE. 2012;7(2):15. doi: 10.1371/journal.pone.0031879. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Barros AJD, Bertoldi AD. Desigualdades na utilização e no acesso a serviços odontológicos: uma avaliação em nível nacional. 10.1590/S1413-81232002000400008Cienc Saude Coletiva. 2002;7(4):709–717. [Google Scholar]
- 5.Bastos GA, Duca GF, Hallal PC, Santos IS. Utilization of medical services in the public health system in the Southern Brazil. 10.1590/S0034-89102011005000024Rev Saude Publica. 2011;45(3):475–454. doi: 10.1590/s0034-89102011005000024. [DOI] [PubMed] [Google Scholar]
- 6.Chaix B, Merlo J, Chauvin P. Comparison of a spatial approach with the multilevel approach for investigating place effects on health: the example of healthcare utilisation in France. 10.1136/jech.2004.025478J Epidemiol Community Health. 2005;59(6):517–526. doi: 10.1136/jech.2004.025478. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Chiavegatto AD, Filho, Gotlieb SL, Kawachi I. Cause-specific mortality and income inequality in São Paulo, Brazil. 10.1590/S0034-89102012005000039Rev Saude Publica. 2012;46(4):712–718. doi: 10.1590/s0034-89102012005000039. [DOI] [PubMed] [Google Scholar]
- 8.Chiavegatto AD, Filho, Lebrão ML, Kawachi I. Income inequality and elderly self-rated health in São Paulo, Brazil. 10.1016/j.annepidem.2012.09.009Ann Epidemiol. 2012;22(12):863–867. doi: 10.1016/j.annepidem.2012.09.009. [DOI] [PubMed] [Google Scholar]
- 9.Cutler DM, Lleras-Muney A. Understanding differences in health behaviors by education. 10.1016/j.jhealeco.2009.10.003J Health Econ. 2010;29(1):1–28. doi: 10.1016/j.jhealeco.2009.10.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Derose KP, Escarce JJ, Lurie N. Immigrants and health care: Sources of vulnerability. 10.1377/hlthaff.26.5.1258Health Aff. 2007;26(5):1258–1268. doi: 10.1377/hlthaff.26.5.1258. [DOI] [PubMed] [Google Scholar]
- 11.Fone D, Dunstan F, Williams G, Lloyd K, Palmer S. Places, people and mental health: a multilevel analysis of economic inactivity. 10.1016/j.healthplace.2005.02.002Soc Sci Med. 2007;64(3):633–645. doi: 10.1016/j.socscimed.2006.09.020. [DOI] [PubMed] [Google Scholar]
- 12.Goldbaum M, Gianini RJ, Novaes HM, César CL. Utilização de serviços de saúde em áreas cobertas pelo programa saúde da família (Qualis) no Município de São Paulo. 10.1590/S0034-89102005000100012Rev Saude Publica. 2005;39(1):90–99. doi: 10.1590/s0034-89102005000100012. [DOI] [PubMed] [Google Scholar]
- 13.Hendryx MS, Ahern MM, Lovrich NP, McCurdy AH. Access to health care and community social capital. 10.1111/1475-6773.00111Health Serv Res. 2002;37(1):85–101. [PubMed] [Google Scholar]
- 14.Jeon B, Kwon S. Effect of private health insurance on health care utilization in a universal public insurance system: a case of South Korea. 10.1016/j.healthpol.2013.05.007Health Policy. 2013;113(1):69–76. doi: 10.1016/j.healthpol.2013.05.007. [DOI] [PubMed] [Google Scholar]
- 15.Kessler RC, Üstün TB. The world mental health (WMH) survey initiative version of the world health organization (WHO) composite international diagnostic interview (CIDI) Int J Methods Psychiatr Res. 2004;13:93–121. doi: 10.1002/mpr.168. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Louvison MC, Lebrão ML, Duarte YA, Santos JL, Malik AM, Almeida ES. Desigualdades no uso e acesso aos serviços de saúde entre idosos do município de São Paulo. 10.1590/S0034-89102008000400021Rev Saude Publica. 2008;42(4):733–740. doi: 10.1590/s0034-89102008000400021. [DOI] [PubMed] [Google Scholar]
- 17.Nemet GF, Bailey AJ. Distance and health care utilization among the rural elderly. 10.1016/S0277-9536(99)00365-2Soc Sci Med. 2000;50(9):1197–1208. doi: 10.1016/s0277-9536(99)00365-2. [DOI] [PubMed] [Google Scholar]
- 18.Ngamini Ngui A, Perreault M, Fleury MJ, Caron J. A multi-level study of the determinants of mental health service utilization. 10.1016/j.respe.2011.09.007Rev Epidemiol Sante Publique. 2012;60(2):85–93. doi: 10.1016/j.respe.2011.09.007. [DOI] [PubMed] [Google Scholar]
- 19.Prentice JC. Neighborhood effects on primary care access in Los Angeles. 10.1016/j.socscimed.2005.07.029Soc Sci Med. 2006;62(5):1291–1303. doi: 10.1016/j.socscimed.2005.07.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Rodriguez G, Goldman N. An assessment of estimation procedures for multilevel models with binary responses. 10.2307/2983404J R Statist Soc A. 1995;158(1):73–89. [Google Scholar]
- 21.Santos SM, Chor D, Werneck GL, Coutinho ESF. Associação entre fatores contextuais e auto-avaliação de saúde: uma revisão sistemática de estudos multinível. 10.1590/S0102-311X2007001100002Cad Saude Publica. 2007;23(11):2533–2554. doi: 10.1590/s0102-311x2007001100002. [DOI] [PubMed] [Google Scholar]
- 22.Viana MC, Teixeira MG, Beraldi F, Bassani Ide S, Andrade LH. São Paulo Megacity Mental Health Survey - a population-based epidemiological study of psychiatric morbidity in the São Paulo metropolitan area: aims, design and field implementation. 10.1590/S1516-44462009000400016Rev Bras Psiquiatr. 2009;31(4):375–386. doi: 10.1590/s1516-44462009000400016. [DOI] [PubMed] [Google Scholar]
- 23.Yip AM, Kephart G, Veugelers PJ. Individual and neighbourhood determinants of health care utilization. Implications for health policy and resource allocation. Can J Public Health. 2002;93(4):303–307. doi: 10.1007/BF03405022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Zhang L, Wang H. Dynamic process of adverse selection: evidence from a subsidized community-based health insurance in rural China. 10.1016/j.socscimed.2008.06.024Soc Sci Med. 2008;67(7):1173–1182. doi: 10.1016/j.socscimed.2008.06.024. [DOI] [PubMed] [Google Scholar]