Sources of medicines for hypertension and diabetes in Brazil: results from the National Health Survey

This study aims to analyze the sources people over 18 years of age use to obtain medication for the treatment of hypertension and diabetes, according to sociodemographic characteristics from 2013 to 2019. Data from the Brazilian National Health Survey were analyzed. Most individuals with diagnosis and prescription to pharmacological treatment reported obtaining medicines exclusively from one type of source. The percentage of people who acquired hypertension medicine exclusively from public pharmacies decreased, from 24.5% in 2013 to 16.2% in 2019; while there was an increase in those obtaining from the Popular Pharmacy program, from 23.5% to 31.4%; as well as for out-of-pocket payment, which rose from 30.9% to 35.5% The percentage of people who acquired diabetes medication exclusively from public pharmacies increased from 7.4% to 18.6% and with out-of-pocket payment increased from 21.6% to 26.8%, while the percentage of those who acquired from the Popular Pharmacy program decreased from 47.2% to 36.4%. The percentage of those who acquired medication from various sources decreased for both hypertension and diabetes. For men, white, and those with higher education, the source of medication acquisition, for both conditions, was mostly by out-of-pocket payment. The high number of medicine acquisition from public sources represents an advance in Brazil’s response to the treatment of these conditions, but reducing regional differences still represents a challenge to be overcome by the healthcare system.


Introduction
Methods This is a cross-sectional study conducted using secondary data from the 2013 and 2019 Brazilian National Health Survey (PNS).
The PNS is part of the Integrated System of Household Surveys (SIPD), of which the sample structure is known as the Master Sample. The Master Sample is a set of census sectors or aggregates of selected sectors to meet the needs of various surveys conducted by the Brazilian Institute of Geography and Statistics (IBGE). The target population is composed of people dwelling in permanent private households throughout Brazil.
The primary units of the Master Sample are stratified according to four criteria: administrative, geographic, urban or rural situation, and a statistical criterion, which subdivides the strata based on the three previous criteria into homogeneous strata, according to the total income of the households and the number of households 18  In the first stage of selection, as part of the SIPD, the primary sampling units (PSU) are obtained by simple random sampling among those previously selected for the Master Sample, respecting the stratification of the PSUs of the Master Sample 18 .
In the second stage, a fixed number of permanent households in each PSU selected in the first stage are then selected by simple random sampling. The selection of households is made based on the Brazilian National Address List for Statistical Purposes (CNEFE) in its last update before the conclusion of this stage of the sampling plan 18 .
In the third stage, in each selected household, a resident from a list of eligible residents constructed at the time of the interview is randomly invited to answer the individual part of the questionnaire 18 .
The PNS questionnaire is divided into three parts, namely: household information, information on all residents, and information about an individual selected among the household residents 19 . The PNS 2013 began data collection in August 2013 and ended in February 2014, collecting 6,069 PSU and 64,348 households with interviews, totaling 205,546 residents with a completed questionnaire. In this edition, individuals aged 18 or older were selected for individual interviews. In total, 60,202 individuals in this age group were interviewed and responded to the individual questionnaire 18 .
To allow comparisons between the 2013 and 2019, the IBGE conducted a new calibration of the expansion factors of the PNS 2013, considering the revision of the Population Projection of the Federative Units by sex and age for the period 2010-2060, released in 2018. This same population projection was used to calibrate the weights of the PNS 2019, thus ensuring comparability between the two editions of the survey 5 . In this study, data from this version of the PNS 2013 were used.
The PNS 2019 had its data collected from August 2019 to March 2020. It has a sample design similar to the 2013 edition, but with some changes regarding the sample size of the PSU and households, in addition to changes in the eligible age group for individual interview, which changed to 15 years of age or more, different from 18 years of age in 2013.
The survey was conducted in 8,036 PSU and had 94,114 households with completed interviews. In these households, questionnaires were completed for 279,382 residents, and 94,114 individuals aged 15 years and over were randomly selected to answer specific interview questions 19 .
The analyses of this article were conducted with the sample of individuals ≥ 18 years old who reported a medical diagnosis of arterial hypertension (2013: n = 10,017; 2019: n = 19,838) and/or diabetes (2013: n = 6,185; 2019: n = 6,338) and those who were using medication as a treatment. An important methodological difference appears in the PNS 2019, in which an intermediate question about the existence of medication prescriptions was included. This question enables identifying people who have the disease and a medical prescription but do not undergo the treatment, which was not possible to be evaluated in the previous survey.
The sources of medicines for hypertension and diabetes (oral medicines and/or insulin) were expressed as: (1) obtaining all from the Popular Pharmacy program; (2) obtaining all from the public pharmacies of the SUS; and (3) out-of-pocket payment of all medicines, when "none" was mentioned in the other sources and the individual paid for their medicines; and (4) various, when the medications were obtained from more than one source.
The proportions of medicine acquisition by source were estimated for both years of investigation and the prevalence ratios for obtaining medication for arterial hypertension and diabetes were estimated, according to the following variables: sex (male; female); age group, in years (18 to 39; 40 to 59; 60 or more); schooling level (no education or incomplete primary education; complete primary education and incomplete secondary education; complete secondary education and incomplete higher education; complete higher education); geographic region of residence (North; Northeast; Central-West; Southeast; South), whether or not they have health insurance, and per capita household income expressed in quartiles for each year (2013-2019). To enable the adjusted analysis, the skin color variable was grouped into whites and non-whites.
The analyses were performed using the statistical package Stata version 13.0 (https://www. stata.com), using the svy command set and considering the weight and the complex plane of the sample. Poisson regression model was used to estimate crude and adjusted prevalence ratios (PR) and 95% confidence intervals (95%CI); in the evaluation of the statistical significance of the differences between groups, a 5% significance level was considered. To define the final model, a "back-
The sociodemographic characteristics of individuals using medication for the treatment of either condition were similar in the two years investigated, being higher among women, in people aged 50 to 69 years, of white race/skin color, with incomplete primary education, without health insurance, in the upper middle-income quartile, and living in the Southeast Region (Table 1).  Most individuals reported obtaining medications for hypertension and diabetes from a single source in the two years analyzed. Obtaining medications for diabetes from a mixed source decreased from 23.8% (95%CI: 18 Regarding the treatment of hypertension, after adjustment, the obtainment exclusively from public pharmacies was lower in males and in white individuals in 2019 and, for both years of investigation, it was lower in the age group of 40 years or more, and among the most educated, being 1.6 times higher in the South Region, when compared with the North Region. Obtaining through the Popular Pharmacy program was lower (40%) among people with complete higher education in 2013 and with health insurance, for both years; whereas it was higher in the South Region compared to the North Region for both years (1.4 and 1.6 times more, respectively). Obtainment exclusively through out-ofpocket payment was higher among men in 2019, and in individuals who self-reported as white, with complete higher education for both years investigated (1.5 and 1.6 times more, respectively) ( Table 2).
Regarding the obtainment from various sources, hypertension medication was higher in the older age groups and lower in the more educated in the two years investigated, and 20% lower in the Central-West Region, when compared with the North in 2013 (Table 2).
Obtaining oral medications and/or insulin for diabetes from public pharmacies, after adjustment, was 70% lower in individuals residing in the Northeast Region, when compared with the North; 50% in the lower and upper middle-income quartile in 2013; and 50% lower in individuals with health insurance and in upper quartile of income in 2019. Obtainment exclusively from the Popular Phramacy program, for the two years investigated, was lower in individuals aged 60 years or older (30% lower in 2013 and 20% in 2019); 40% and 20% lower in individuals with a complete higher education, with health insurance, and in the residents of the South Region in relation to the North (1.5 and 1.9 times more respectively). The acquisition of diabetes medication exclusively with out-of-pocket payments was 1.3 times higher in whites in 2013, and in the two years analyzed, it was higher in males (1.2 e 1.3 times more respectively), in individuals with complete higher education (1.3 and 1.2 times), and significantly higher in individuals with health insurance (2.9 times more in 2013 and 2 times more in 2019). A lower rate of this source of acquisition (60% less) was observed in residents in the South Region of the country, when compared with the North, both in 2013 and 2019 ( Table 3).
Obtainment of diabetes medication through various sources was higher in men, among the most educated, in the elderly (aged 60 years or older) about 1.5 times more in individuals with health insurance, in the highest income quartile in 2019; and in whites in the two years investigated (Table 3).   Table 2 Crude and adjusted prevalence ratios (PR) for the sources for the acquisition of hypertension medication by adults aged 18 Table 3 Crude and adjusted prevalence ratios (PR) for the sources of obtaining medicines by adults aged 18 years or over to treat diabetes, according to

Discussion
The results of PNS, in the two years evaluated, show that most Brazilians with hypertension and diabetes use medication for their treatment, which reinforces the importance of monitoring the provision of medication responsible for controlling these conditions. However, comparisons between the results regarding the sources for obtaining these medications in the PNS 2013 and 2019 should be done cautiously. This is a consequence of the differences in the questionnaires of the two surveys, which represent an improvement in the evaluation of the use of drugs for hypertension and, especially, for diabetes. The 2019 survey included a question about prescribed medications for both hypertension and diabetes. The previous format went from questioning disease diagnosis to questioning medication use, underestimating the proportion of use. It was necessary to assume that a fraction of the people with the diagnosis did not necessarily use any type of medication. The new question allow for evaluate people who had been prescribed the medication but did not adhere to the treatment. Another important change was that in 2019 oral medicines and insulin were evaluated in separate questions, regarding the prescription, obtainment, and use of these medications, while in 2013 there was only one question on the subject, without distinction. The analysis of the Risk and Protective Factors Surveillance System for Chronic Noncomunicable Diseases Through Telephone Interview (Vigitel) in 2011, 2014, and 2017 corroborates the results from the 2013 PNS, when, about 80% of individuals aged 20 years or older with a diagnosis of hypertension, reported using medicines for their treatment 16 . Considering the 2019 data, an increase in the use of drugs for the treatment of hypertension was observed, when compared with the national survey carried out in 2013 (94.6%; 95%CI: 93.5-95.5), which also investigated the use of pharmacological treatment among individuals who had had a medical prescription 12 .
For diabetes, it is important to highlight the difference in the proportion of individuals with medical indication and use of oral hypoglycemic agents in relation to insulin, which was much lower. This difference may suggest barriers to access or adherence to insulin treatment, which can be explained by the fact that insulin dispensing is more restricted by the required storage conditions, its use depends on other supplies, in addition, to the discomfort which is inherent to its administration.
Analyzing the sources of obtainment, a large part of the population uses a single source to obtain all medication for the treatment of both chronic diseases herein investigated. For people with hypertension, the main single source of acquisition is out-of-pocket payment, with a decrease in the obtainment from public pharmacies and an increase in the obtainment from the Popular Pharmacy program from 2013 to 2019, which has also been observed in another study representative of the state's capitals and the Federal District 16 and that can be explained by the implementation of free antihypertensive drugs, through the campaign Saúde Não Tem Preço (Health is Priceless).
As for people being treated for diabetes, an increase was observed for the obtainment exclusively from public pharmacies, with a decrease from the Popular Pharmacy program, which was also observed in a study that analyzed data from sources for the obtainment for diabetes oral medications from 2012 to 2018 17 .
Despite the differences between the sources within the analyzed years, the free sources -those from the public pharmacies and from the Popular Pharmacy program -remains as the main source for acquiring medicines for the treatment of these conditions, especially for diabetes. However, it is worth noting that almost a third of Brazilians affected by these conditions pay for the treatment in full, evidencing the importance of strengthening public pharmaceutical policies and guaranteeing population's accesses to medication 15,16,17,20 .
When analyzing the sources for obtaining hypertension medication, considering the sociodemographic characteristics, we observed that male individuals and self-declared white people were less adherent to public pharmacies. Additionally, there is a decrease in the use of this source with increasing age and schooling. Regarding the Popular Pharmacy program, the obtainment of antihypertensive drugs was lower among individuals with health insurance, and there was a difference between the South and the North individuals. On the other hand, the use of one's own funds for medication was positively associated with being male, having declared white skin color, and, significantly, with having completed higher education.
Cad. Saúde Pública 2022; 38 Sup 1:e00152721 Notably, the sources of access to hypertension and diabetes medication have a social distribution. Those less educated and non-white seek free sources more, while the more educated, white people prefer the out-of-pocket payment sources.
The use of more than one source -be it SUS, Popular Pharmacy program, or private pharmacy networks -increased among older people. This was more intense in people with less education. This may be the result of a greater complexity of the treatment, resulting from aging and the consequent need for a greater number of medication.
Obtaining medication from one's own fund was higher in men than in women. This suggests a general male behavior of lower use of healthcare services.
The South Region concentrates the highest proportion of free sources of medicines (SUS and Popular Pharmacy program) for hypertension, which shows a certain contradiction, in which places with better economic conditions offer greater possibility of free access to medicines. The explanation for this significant difference can have several origins, one of which notes that the expansion of Popular Pharmacy program took place in regions with greater infrastructure and in larger pharmacies networks, due to the technological needs for implementing the sales system. The acquisition of diabetes medication from SUS is lower in this region. These regional differences are not restricted to medication, according to several authors. The study published by Viacava et al. 21 addresses the regional differences in the various aspects of the Brazilian health services. Regarding this issue, the classic text written by Hart 22 discuss the inverse care law, in which regions that need it most receive fewer medical resources. Although public policies toward medication accessibility may have reduced these differences in Brazil in recent decades, the situation still deserves attention.
It is also important to highlight the differences within the PNS questionnaires relating to the options offered as sources of medicines. In the 2013 survey, the first option was "Health plan", which did not occur in 2019. This is because the supply of medicines via health plans, except for some specific products, occurs only during hospitalization. The supply of medicines for outpatient use is restricted to very few plans and covered by reimbursement. The PNS questioning strategy is to not ask the next source when an individual answers that they acquire all their medication from one source. Thus, in 2013, those who answered that they obtained all their medication from the health insurance were not asked about obtaining them from the Popular Pharmacy program or from the public health system. This strategy is based on a logic aiming toward the coherence of the answers, since if everyone answered, there would be no logical possibility of another source of medication. However, the order in which the sources are presented may affect the answers when the first source is chosen. This may have underestimated the acquisition of medication from the SUS and the Popular Pharmacy program by presenting an option of access to hypertension and diabetes medication that, in practice, does not exist.
Among the limitations of our study, we highlight the use of self-reported morbidity, which may underestimate the prevalence of the diseases. Moreover, information bias may arise regarding the indication of drug treatment and the correct identification of the source of obtaining drugs.
The high availability of hypertension and diabetes medication from free sources, despite regional and sociodemographic differences, represents an advance in Brazil's response to the treatment of these conditions. On the other hand, simply obtaining the drugs does not guarantee their proper use. Those who seek to improve the population's health should focus on increasing adherence to treatments, promoting correct monitoring of the disease, and expanding health care to encompass changes to lifestyles. The reduction of regional differences, with the improvement of services in the most vulnerable areas still represents a challenge to be overcome by the Brazilian health system.