Inequities in access to health care in different health systems: a study in municipalities of central Colombia and north-eastern Brazil

A cross-sectional study was conducted by means of a population survey in central Colombia and the northeast of Brazil. The study areas were two municipalities in each country: in Colombia, Kennedy (a district of Bogotá) and Soacha; and in Brazil, two micro-regions (3.2 and 3.3) of District 3 in Recife, Pernambuco’s capital, and Caruaru, in the interior of Pernambuco state. These four areas are the areas of the Equity-LA project (http://​www2.​equity-la.​eu/​), the broader Project in which this study is framed[52]. The areas were selected for being densely populated urban spaces with a high proportion of the population belonging to the low or medium-low socioeconomic strata and with varying geographical access to specialist care. Kennedy has approximately 1,000,000 residents and the other three study areas have about 300,000 residents. Kennedy and the two micro-regions of District 3 in Recife share the characteristic of being some of the more deprived areas in their respective cities.

The study population was made up of residents in the study areas who had had at least one health problem or had used to the health services in the three months prior to the survey.

The sample size was calculated for each study area based on the population size and an estimated proportion of 50% (maximum uncertainty principle); degree of confidence: 90% (alpha error of 0.1); precision: 2.5. The sample size was 2,163 in Colombia (1,083 in Kennedy, 1,080 in Soacha) and 2,155 in Brazil (1,076 in district 3 of Recife, 1,079 in Caruaru).

A questionnaire was designed to analyze access to health care. Based on the Behavioral Model of Health Services Use[53] and previous qualitative research[54‐58], dimensions and variables were identified. In addition, systematic reviews were conducted of studies on access and equity of access to health care and of available tools in order to identify additional variables. The literature search was carried out using the most relevant electronic databases (Pubmed, CINAHL, Social Science Citation Index, PsycInfo, Lilacs, IBECS, The Cochrane Library, System for Information on Grey Literature in Europe, Information system of the WHO Library, Panamerican Health Organization Library) to minimize the likelihood of excluding relevant studies. The search strategy included the combination of descriptors and keywords relating to equity in access to health care, delivery of health care and measurement instruments, utilizing the Boolean operator ‘AND’. Given the large number of studies identified, only studies from Colombia, Brazil, United States and Spain were included. The U.S. and Spain were included because their health systems are comparable to the Colombian and Brazilian system respectively. A total of 41 different articles were identified up to 2010. Based on these results, the first version of the questionnaire was built and validated by means of discussions with experts, after which it was adapted to both context and languages. A pre-test and a pilot test were conducted in each country to evaluate the rhythm of the interview, interviewer burden and comprehensibility. Both the rhythm of the interview and the interviewer burden proved adequate. With respect to comprehensibility, some questions were removed and others modified because the terms used were unfamiliar and people had problems understanding them. Finally, since many changes had been made, another pilot was deemed necessary, in which the questionnaire was finally considered to be adequate. The final questionnaire is divided into nine sections. The first collects information on perceived health needs and related behavior in the three months prior to the survey, in other words, whether or not the individual had used the health services and the level of care at which they had been attended. The next four sections refer to their most recent experience - within the three months prior to the survey - using the different levels of health care (primary, outpatient secondary, emergency, and inpatient care) of the SGSSS and SUS services. The last three sections include a Likert scale to measure care continuity, knowledge of the healthcare system and sociodemographic data. The Colombian questionnaire has an additional section related to insurance enrollment.

Data were collected from February to June 2011 by means of face-to-face interviews conducted by specifically trained interviewers in both countries.

Strategies to ensure the quality and consistency of data included close supervision of interviewers in the field, the review of all questionnaires and 20% of re-interviews (selected randomly). Inconsistencies during data entry were controlled through the double-entry method.

Ethics approval was obtained from the ethics committees in the participating countries: the National Committee of Research Ethics in Brazil; the Research Ethics Committee of the Health Sciences School of Universidad del Rosario in Colombia; the Institutional Review Board of the Institute of Tropical Medicine in Belgium; and the Clinical Research Ethics Committee of Parc de Salut Mar in Spain. All interviewees participated on a voluntary basis, after signing an informed consent. The right to refuse to participate or withdraw from the survey, anonymity, confidentiality and protection of data were all guaranteed.

The outcome variables are three dichotomous variables on utilization of healthcare services in the three months prior to the survey: a) consultation of a general practitioner or pediatrician, b) consultation of a specialist, c) consultation of emergency care services; and three variables on the use of preventive services by adults in the last year: d) glycemic control, e) caries prevention and f) mammography in women.

Need for care was measured by self-rated health (dichotomous: i) good - very good and good, and ii) poor - fair, poor and very poor) and having at least one chronic condition[59]. The main explanatory variables were per capita income and holding a private health plan in both countries as a proxy of socio-economic status, and type of SGSSS insurance scheme in Colombia. Per capita income (less than half minimum wage (MW), ½ - 1 MW, 2 or more MWs) was estimated by dividing the household income by the family size. Sociodemographic variables - sex, age (0–17, 18–40, 41–65, 66 or over) acolor/color- were used to adjust the models.

A univariate analysis was performed to describe the distribution of the outcome and explanatory variables for each country (Table 1). To establish the relative association between each healthcare utilization variable and socioeconomic status, prevalence ratios (PR) and their corresponding 95% confidence intervals were computed by means of Poisson regression models with robust variance (Tables 2 and3). PR was considered better suited to our study than other measures of inequality such as the Relative Index of Inequality (RII) because the socio-economic variables included in the models are not strictly hierarchical[60]. Absolute differences in utilization prevalence at both extremes of the three socioeconomic variables were also calculated. Both absolute and relative measures were used to assess socioeconomic inequalities in the utilization of healthcare services.

In addition, stratified analyses were carried out: firstly, regression models were generated stratifying by income to evaluate the effect of the insurance scheme in Colombia and additional private health insurance in Brazil on health services utilization for each socioeconomic stratum (Tables 4 and5); and secondly, regression models were made stratifying by type of insurance in both countries (contributory and subsidized in Colombia, and only SUS or SUS and private health plan in Brazil) (Table 6). Analyses were carried out with STATA statistical package version 12[61].

In both samples, the majority of participants are women, and while in the Colombia sample there is a predominance of the central age groups (18 to 40 and 41 to 65 years), in that of Brazil there is a more even distribution of the sample across all age groups. With regard to SGSSS enrollment, 56.0% are enrolled in the contributory scheme, 28.0% in the subsidized scheme, and 11.3% are uninsured. The socioeconomic status of the sample is not very high in either country: most people have less than half of the minimum wage (MW) per month, but distribution is less equal in Brazil than in Colombia. The socioeconomic level in Colombia is closely related to enrollment; the proportion of people with less than half MW in the previous month is higher in the subsidized scheme and the uninsured group than in the contributory scheme: 72.0%, 63.2% and 36.4% respectively. In Colombia 2.0% and in Brazil 20.1% hold a private health plan (PHP) (Table 1). Self-rated health status is better in Colombia: 62.2% report good or very good health, compared to 44.7% in Brazil; 30.0% claim to have at least one chronic disease, as opposed to 41.9% in Brazil (Table 1).

In the Colombia sample more than half of the individuals used the first level of care (52.2%), a higher proportion than in that of Brazil (44.5%). However, for outpatient secondary and emergency care, utilization in Brazil was significantly higher than in Colombia (Figure 1). Preventive activities in the last year differ between countries according to type of activity: while the proportions of people who used glycemic control and mammography services were similar, significant differences were found in caries prevention in adults (37.7% in Colombia vs. 19.9% in Brazil) (Figure 1).

At primary care level, per capita income does not figure as a source of inequality in Colombia but it does in Brazil, where individuals with a higher per capita income make less use of this care level (prevalence difference with respect to the poorest is -17.9, PR = 0.71, CI 95%: 0.59-0.85). In Colombia, it is the insurance scheme which is associated with the utilization of services at this level. The utilization prevalence of individuals in the contributory scheme is 58.0%, in the subsidized scheme 52.1%, and 24.7% for the uninsured (Table 2). In outpatient secondary care, per capita income is a source of inequity in both countries: higher-income individuals are more likely to use these services. In Colombia, the PR of individuals with more than one minimum wage per capita to those with the lowest income is 1.26 (CI 95%: 1.02-1.56) and in Brazil, 1.42 (CI 95%: 1.14-1.77), which is the opposite pattern of use to that detected at primary care level. Holding a private health plan is also associated with a higher prevalence of utilization in both countries. Furthermore, in Colombia, insurance enrollment reveals the same pattern as that found in primary care. Individuals in the contributory scheme display the highest use of this care level: the PR of subsidized to contributory is 0.72 (CI 95%: 0.57-0.90) and of uninsured to contributory is 0.40 (CI 95%; 0.24-0.64) (Table 2). With regard to emergency care services, the higher the per capita income is, the lower the prevalence of utilization in both countries, although once this is adjusted for need there are no perceivable inequities associated with income and private insurance in either country. However, in Colombia individuals not insured under the SGSSS are unfavorably positioned with respect to those insured under the contributory scheme (PR: 0.63, CI 95% 0.45-0.88) (Table 2).

There are noticeable inequities in preventive activities in both countries. The absolute differences in utilization prevalence (of all three socioeconomic variables) are always in favor of those who are most socioeconomically favored. In Colombia, uninsured individuals are less likely to use the three preventive services than those of the contributory scheme. Moreover, individuals in the subsidized scheme are less likely to have glycemic controls and mammographies than individuals in the contributory scheme. The inequities are accentuated in breast cancer prevention according to per capita income: the prevalence of utilization increases as the level of income rises (21.9%, 30.0% and 37.2%). In Brazil, people with a lower income and without private insurance are less likely to use the three preventive services, and the difference in prevalence of using caries prevention services is particularly marked between those without private insurance (16.1%) and those with (36.2%) (Table 3).

This second section looks closely at the effects in each income stratum of insurance scheme in Colombia and holding a private health plan in Brazil on the utilization of healthcare services, after having observed an association between higher socioeconomic status and greater use of the health services. In Colombia, in the two lower-income strata, individuals of the contributory scheme are more likely to use the primary and outpatient secondary care services than the uninsured; and they are also more likely to use outpatient secondary care than individuals in the subsidized scheme (PR = 0.70, CI 95% 0.53-0.94 in the lower per capita income stratum and PR = 0.59, CI 95%: 0.38-0.91 in the middle stratum). Inequalities between schemes were not found in the case of emergency care (Table 4). The effect of the insurance scheme on the use of preventive services is particularly noticeable in the lowest per capita income stratum, where belonging to the subsidized scheme or being uninsured significantly reduces the probability of using glycemic control services (prevalence difference between individuals in the contributory scheme and in the subsidized scheme is -11.5, and between individuals in the contributory scheme and the uninsured is -29.4) and mammography services (prevalence difference is -18.5 for the subsidized enrollees and -20.8 for the uninsured) (Table 4).

In the case of Brazil, the effect of private insurance varies depending on care level and income. It is most influential in the middle-income stratum, where individuals with a private health plan are less likely to use primary and emergency care services and more likely to use outpatient secondary care: the PR of those with insurance to those without is 1.57 (CI 95% 1.23-2.00) (Table 5). Having a private health plan implies a greater likelihood of using the three preventive services in all income strata (Table 5).

Lastly, the presence of inequities was evaluated according to insurance scheme in Colombia and type of coverage in Brazil (Table 6). In Colombia, the prevalence values for services utilization are consistently higher in the contributory scheme than in the subsidized scheme in all income strata, with the exception of emergency care and caries prevention. In the contributory scheme, level of income does not reveal inequity in health services utilization, except in the cases of outpatient secondary care and breast cancer prevention, which are used more frequently among higher income individuals. In the subsidized scheme, no inequity was detected according to level of income in use of the curative services, but it was observed in the preventive services such as caries prevention (PR comparing those in the middle income stratum with those in the lower is 1.45, CI 95%: 1.12 - 1.87) and breast cancer prevention (PR = 2.00, CI 95%: 1.28 - 3.11) (Table 6). This analysis could not be performed in the uninsured group due to its low level of use of the health services.

In Brazil, higher-income individuals are less likely to use the primary care level whether they hold a private health plan or not. In outpatient secondary care there is an income gradient in individuals with double coverage: the higher the income, the higher the probability of utilization (PR = 1.94, CI 95%: 1.22- 3.10 and PR = 1.98, CI 95%; 1.23 - 3.20) (Table 6). The prevalence of use of preventive services are higher in the group with double coverage (SUS and PHP), and the relationship with income is significant in the case of caries prevention (in both groups), as well as in mammography in the group without private insurance (Table 6).

One of the successes attributed to the Colombian reforms is an increase in the population enrolled in the SGSSS, as this was considered to facilitate access and use of the health services[7, 27, 38]. Although the uninsured population fell from 41.5% in 2000 to 12.1% in 2010[15], there is a lack of consensus in the literature on the impact this has had on equity in utilization[39]. Our results coincide with those studies in which authors show that enrollment in the SGSSS improves access to the health services[24, 27, 42], to the extent that uninsured individuals have practically no access. However, this study reveals that inequities also exist between the insurance schemes. Individuals of the subsidized scheme, with equal need, make the least use of primary, secondary and preventive care. This result is in contrast with those of other authors who conclude that enrollment guarantees access[37, 38], basing this conclusion on a utilization comparison between individuals of the subsidized scheme and the uninsured, without mentioning the differences between the two schemes. Furthermore these inequities are more pronounced in secondary care services, results which agree with the only study available that differentiates between the two care levels[7].

In emergency care, inequities between the two insurance schemes were not found, but they were found between the insured and uninsured, which appears to indicate that the legal obligation of all public and private entities to offer this type of care to everyone is not fulfilled[14], because it depends on the individual’s insurance status.

It is interesting to note that the role of the insurance scheme is more evident among lower-income individuals, as in these strata it acts as a source of inequity in the utilization of preventive and curative services (with the exception of emergency care), whilst in the higher-income population no differences according to insurance scheme were found in the utilization of services. Moreover, no inequities according to income were found in individuals belonging to the same scheme. In other words, health services utilization by the poorest population – the one which has the greatest need for it – is determined by insurance scheme and not by need.

This study corroborates the argument that the design of the SGSSS displays a structural inequity[1, 39], i.e. individuals under the subsidized scheme use the health services less than those under the contributory scheme because the benefits package offered is smaller. Thus the principle of horizontal equity is violated by the design of the health system itself, as it determines that the use of health services depends on the type of SGSSS scheme the individual is enrolled in instead of on the individual’s health needs. It will be necessary to analyze what has occurred since the new law came into effect to make the benefits package equal in the two schemes[63].

In the case of Brazil, this study is consistent with previous research in terms of the existence of inequities in access to the health services in spite of the fact that it has a national health system[25, 26, 64], and it also contributes new results: inequity in utilization varies according to care level and is more pronounced in outpatient secondary care and preventive services. There are authors who point out that inequity in access has been reduced in recent years[5, 6], although this statement should be treated with caution as they do not distinguish between care levels.

At the primary care level there is inequality in favor of the poorest (i.e. it is pro-poor), meaning that lower-income individuals make greater use of this level, which is the direct opposite of the situation in secondary care. This phenomenon has been described in previous research conducted in other countries with a national health system[8, 32, 33, 65]. Inequity in health care utilization is concentrated at the secondary level, which might be due to the fact that higher-income individuals access private secondary care services directly (via a private health plan or out-of-pocket payment) instead of accessing secondary care through the SUS[66], in order to avoid the significant access barrier of long waiting times in the SUS[62, 67]. The lack of specific studies differentiating between these two levels of care in Brazil hinders the comparison of results.

In emergency care no inequities were observed, although the utilization prevalence shows that use falls as per capita income rises and may signal a higher level of use among the poorest, which could be the consequence of the presence of access barriers at the primary care levels, such as the lack of doctors or low levels of health problem resolution reported in other studies[62].

Furthermore, the results show that people with a private health plan have more access to all levels of care than those without, thus the private health plan facilitates access to services[65], but nonetheless income inequalities persist within this group. This may be a result of the existing inequalities in relation to the coverage offered in private health plans, since higher-income individuals pay for more expensive plans, which offer better coverage in terms of complementary tests and medical exams[64].

In terms of preventive activities, our study once again reveals inequities in both countries that benefit individuals with a higher socioeconomic status. Only a few studies in Colombia on access to preventive activities analyze specific activities, and those assess inequalities in access to mammography[68] or cervical cytology[69], which hinders their comparison with our results. De Charry et al. pointed out that women in the contributory scheme reported higher levels of mammography services utilization[68]. In Brazil, some studies identify lower odds of having had a mammography among women with a low educational level, low family income and no private health plan[70, 71].

Moreover, as our results in Brazil corroborated, inequalities have been described in the preventive use of dental services in the south of the country[72] and other authors reveal inequalities in the use of dental services without differentiating between curative and preventive visits[5]. Furthermore, studies in both countries that do not distinguish between the types of activity identify that the higher the individual’s socioeconomic level, the higher the probability that the medical visit was for preventive rather than curative reasons[7, 28, 35]. This cycle, in which the poor (lower-income, subsidized scheme and uninsured individuals in Colombia, and those without a private health plan in Brazil) have worse health and worse access to curative health care and preventive activities, makes them particularly vulnerable and may be related to worse health outcomes.

Finally, we should bear in mind that the evaluation of equity in access based on the use of services has been widely criticized for ambiguities in its interpretation: it is difficult to differentiate between necessary use and excessive or insufficient use, both in terms of quantity and quality of care received[73]. In this study, quality of care has not been assessed, although in both countries some results suggest that care quality varies according to socioeconomic level. In Colombia, fewer subsidized scheme enrollees reported that their health problem was solved than contributory scheme enrollees. In Brazil, higher-income individuals reported higher levels of positive outcomes[62]. As there is insufficient data to serve as evidence, further evaluation of equity in access to health services is required, focusing not only on utilization of services but also on quality of care.