população quilombola : inquérito no Sudoeste da Bahia Medication use by the “ quilombola ” population : a survey in Southwestern Bahia , Brazil

MÉTODOS: Estudo transversal de base populacional com 797 quilombolas adultos de Vitória da Conquista, BA, em 2011. Utilizou-se análise de variância para comparar as médias de medicamentos por indivíduo segundo variáveis demográficas, socioeconômicas e de comportamentos relacionados à saúde. Foram estimadas as prevalências, razões de prevalência e os respectivos intervalos de confiança de 95%. Análise múltipla foi conduzida por meio de regressão de Poisson com variância robusta.


INTRODUCTION
da Conquista: Avaliação de Condicionantes de Saúde), in Vitória da Conquista, BA, Northeastern Brazil, in 2011.d The 797 participants were selected using probabilistic two-stage sampling: i) proportional division between the five districts that contain quilombola communities, according to the population and selection of each community, chosen through simple random selection; ii) random selection of residences according to the proportional distribution per district.Residents aged 18 and over in the selected residences were invited to take part in the research.
When calculating the sample size, the following were considered: an estimated prevalence of 50.0%given the heterogeneity of the events measured, accuracy of 5%, 95% confidence interval (95%CI) and a design effect of 2. An extra 30.0%was added to the number obtained, to cover losses, producing a sample size of 884 individuals.Details of the survey's sampling process can be found in another publication.d The losses encountered (15.5%) were smaller than forecast in the beginning of the study, but significantly higher among males and younger individuals (18 to 34 years old).The main reasons for losses were not being in the residence and refusals.
The data were obtained using a semi-structured questionnaire adapted from the National Health Survey.e A pilot study was carried out to verify the recruitment dynamic, test the data collection instruments and confirm the viability of the investigation.Individual interviews were carried out at the homes and applied using laptop computers (HP Pocket Rx5710) between September and October 2011.
The dependent variable was medication use, obtained based on the following question: "In the last 15 days have you used any medications?",checked by the presentation of packaging or prescription.For those who responded yes, the name, pharmaceutical form and dosage of each were recorded, and whether or not they had been prescribed by a health care professional (doctor, dentist or nurse).The medications were classified according to the Anatomical Therapeutic Chemical (ATC) Classification System, f levels 1 (anatomical) and 2 (therapeutic).The  Feb 15].Available from: http://www.whocc.no/atc_ddd_index/medications were classified by their presence, or not, in the Relação Municipal de Medicamentos Essenciais (REMUME g -Municipal List of Essential Medicines) in force at the time.The units of analysis were the individual and the medications.The mean number of medications per interviewee was used as an indicator of intensity of use.Each medication was broken down by principle active ingredients with the help of the Pharmaceutical Specialties Dictionary h in order to calculate the mean number of active ingredients per interviewee.Independent variables were age, marital status, schooling, work status, economic level (economic classification defined by the Associação Brasileira de Empresas de Pesquisa (Brazilian Institute of Research Companies), i self-perceived state of health, number of self-reported morbidities, frequency of home visits form a community health worker or health care professional and number of medical appointments in the preceding 12 months.The number of morbidities was defined based on adding the interviewee's self-reported morbidities and included in the questionnaire (hypertension, diabetes, hypercholesterolemia, heart disease, stroke, asthma or asthmatic bronchitis, arthritis, chronic spinal problems, tuberculosis, depression, other mental disease, lung disease and osteoporosis).Gestational hypertension or diabetes were not considered.
Differences between the means of active ingredients per interviewee were compared using variance analysis.The prevalence of medication use was calculated based on the number of participants who responded having used at least one medication in the 15 days preceding the interview divided by the total number of interviewees.The Prevalence Ratio was used as an estimate of association between medication use and the explanatory variables in question.This measure and its 95%CI were estimated using Poisson regression with robust variance.Poisson multiple regression with robust variance was used to obtain estimates of the prevalence ratios for medication use, adjusted for potential confounding factors.Those variables which had an association with medication use with a level of significance < 20% in the univariate analysis were included in the initial model.A level of significance of 5% was used for the tests and to judge whether the variables would remain in the final model.Models were compared using Akaike's Information Criterion (AIC) and Bayesian Information Criterion (BIC).The fit of the model was evaluated using Chi-square test.Barreiras (CAAE 0118.0.066.000-10, on 29/10/2010) and of the Universidade Federal de Minas Gerais (CAAE 0118.0.066.203-10, on 13/7/2011), and followed the standards set by the Declaration of Helsinki and by Resolution 196/96 of Conselho Nacional de Saúde.Participants were informed of the research objectives and procedures beforehand and assured of the confidentiality of all data by reading the informed consent form, expressing their agreement to take part in the study.

RESULTS
Among the 797 adult individuals who responded to the survey, 54.3% were female; the majority were aged between 35 and 59 years old (41.5%).More than half (61.4%) cohabited and 72.4% had up to four complete years of schooling (Table 1).The majority were unemployed and 85.6% belonged to economic class D or E. The predominant self-perception of health was good or very good and 41.2% of the interviewees reported having none of the morbidities in question.With regards to health care service use, half of the participants had not seen a doctor and 50.3% of them had received a monthly visit from a community health worker or health care professional.The frequency of medication use was 41.9% (95%CI 38.5;45.4).
Participants used 714 different types of medicine, corresponding to 853 active ingredients (mean = 1.1 active ingredients/individual; standard deviation -1.7; range = 0 to 15).The majority of medications had been prescribed by a doctor, dentist or nurse (83.3%) and 70.0% of them appeared in REMUME.
The most commonly used medications belonged to Cardiovascular, Nervous and Musculo-skeletal system, and Alimentary tract and metabolism (Table 2), with means of consumption higher for these groups.The most commonly used therapeutic subgroup were diuretics (mean of 0.15/individual), followed by agents acting on the renin-angiotensin system (0.13), analgesics (0.08) and anti-inflammatory and antirheumatic products (0.07).Drugs used in diabetes predominated in Alimentary tract and metabolism.
The prevalence of medication use was significantly higher in women than in men (50.3% and 31.9%,respectively) (Table 3).Medication use was positively and significantly associated with being female, older age groups, better economic level (classes D, C and B2), worse self-reported health, a higher number of self-reported diseases and a higher number of medical appointments in the preceding 12 months (Table 3).A negative significant association was observed with having five or more complete years of schooling and being in work at the time of the interview.
The mean number of active ingredients taken by the interviewees was significantly higher among females, older ages groups, those with lower levels of schooling, those who were out of work, those with a better economic level, with worse self-evaluated health, a higher number of reported diseases and medical appointments (Table 3).
The following proved to independently associated with more frequent medication use in the quilombola population: a) being female; b) being aged 60 and over; c) higher socio-economic level, with dose-response gradient; d) self-evaluating health as bad or very bad; e) higher number of self-reported morbidities, with dose-response gradient (Table 4).The values predicted by the models were shown to be adequate to the observed values.

DISCUSSION
The frequency of medication use in this population (41.9%) was lower than that observed in studies of adults in Fortaleza, CE, Northeastern Brazil (49.7%), 1 Lorena, SP, Southeastern Brazil (51.3%), 7 Pelotas, RS, Southern (65.9%), 2 and in Brazil (49.0%). 3This may be partially explained by quilombola's minor access to medications, once they live in rural areas, where public health care services are generally limited.
For those resident in more distant locations, with no regular provision of public transport, travelling from their place of residence to acquire medications can be difficult. 6The percentage of medications present in REMUME was high as, in low income populations, more prescription of medications available in municipal health care network pharmacies is common.In studies of the older adults in Brazil, lower levels of medication use were found in small cities in the countryside (such as 69.1% in Bambuí, MG, Southeastern Brazil), 10 in areas of low socio-economic levels (the periphery of Fortaleza, 60.7%) 4 and in rural areas (Carlos Barbosa, RS, Southern Brazil, 63.5%). 6Medication use in quilombola communities do not seem to differ from that observed in rural or low income communities.
4][5][6]11 Sans et al 13 explain that medication use is higher among women because of the higher frequency of medical appointments and the subsequent greater probability of health problems being detected and diagnosed.Moreover, various health care programs (prenatal, breast cancer and cancer of the uterus prevention) are aimed at women, making them more prone to medicalization. 2 In this study, the number of medical appointments was significantly higher among women.
1][12] Other national studies on the adult population 2,3 showed similar data, although the ATC system was not used to classify the medications, which limits comparisons.
Research in Catalonia, 13 Spain, showed the same therapeutic groups, albeit in a different order of classification, with drugs used for the nervous system predominating.
There is no set prescription standard, which depends on the characteristics of the health care system and of the population assessed.However, Ribeiro et al 11 stated that it was possible that the prescriber takes on patterns of prescribing according to the age of the patients and according to ideological and market pressures, which would explain the similarities observed.
There were high levels of consumption of analgesics and anti-inflammatory and antirheumatic products.This may be explained by the greater tendency to selfmedicate, common among users of these therapeutic classes, 14 partly due to their being over-the-counter drugs and their sale being, often, irregular.In fact, 65.6% of analgesics and 46.4% of anti-inflammatory and anti-rheumatic products had not been prescribed by a health care professional.
Medication use increased with age, a trend that is consistent with the literature 2,3,5,7 and that may reflect the higher prevalence of morbidities that occur with advancing age.This effect was confirmed in the adjusted analysis, in which this variable was independently associated with medication use.Schooling negatively influenced the amount of medications used.This finding differed from those found among the elderly in Belo Horizonte, MG, 11 and Rio de Janeiro, RJ, 12 in the population of Fortaleza, CE, 1 and of Brazil, 3 but agrees with that found in the population of Pelotas, RS. 2 The effect, however, did not persist after adjusting for the other variables.The communities in question were homogenous with regards to the level of schooling (more than 70.0% had up to four complete years of schooling), which meant that the cutoff point for this variable was lower than that adopted in other studies, making comparisons difficult.Other factors investigated, such as age, may also confound the association.In fact, an association was found between age and schooling, indicating that the proportion of individuals aged 60 and over, without schooling, was 75.7% (against 33.3% and 10.8% in the 35 to 59 and 18 to 34 age groups, respectively), and at the same time, they used more medications.
Higher medication use among individuals in higher economic levels was consistent with the study by Bertoldi et al, 2 although level A and B1 were not identified in the quilombola population.Other studies that used monthly household income as an indicator of economic level 1,6,10 found similar results.This association was confirmed after adjusted analysis.
Medication use was lower among individuals who were working at the time of the interview, although this effect did not persist after adjusting for the other variables.
Rural work has characteristics that may interfere in the use of medications.Dal Pizzol 6 highlighted that rural workers, who remain laboring the whole day, may stop using one or more medications more frequently than those in urban environments if the use is made more difficult by the conditions of rural work itself.
A trend was observed for the frequency and number of medications used to increase with worsening of health.1][12][13] The effect of this variable was confirmed in the adjusted analysis.Another variable that indicates the health status of the population was the number of self-reported comorbidities, which shows individuals' chronic health problems and which was associated with medication use even after adjustment.
The variables related to health care services, frequency of home visits from community health care workers or health care professionals and the number of medical appointments behaved differently.The frequency of home visits proved not to be associated with medication use, even with the study population coming from a rural zone and considering this visit an opportunity in which medications could have been supplied.
Frequency of medication used increased as the number of medical appointments increased, even after adjustment, as had been indicated in other studies. 1,11,12Arrais et al 1 suggested that this could be related to increasing medicalization of society, in which the majority of appointments end in a prescription.The Brazilian Health Care System faces difficulties in overcoming the practices of the bio-medical model, as the population value and prefer curative practices, individualized care based on prescriptions to activities promoting health and preventing health problems.Hand in hand with this practice often goes irrational use of medications, which can lead to a variety of problems, such as adverse reactions, iatrogenic disease, resistance (antimicrobial) and unnecessary costs. 2 This study had some limitations.The use of a 15-day recall period to assess medication use was prioritized to enable comparison with other studies, as the majority used this period.However, this strategy may result in memory bias.To avoid this limitation, analysis was restricted to medications for which there was proof of use, through showing packaging or prescription.The differential loss, observed in males and in the 18 to 34 age group, may have caused an overestimate of the use of medications by the overall population and by males, as lower consumption is expected in underrepresented groups.
Assessment of individual determinants of consumption in quilombolas indicates that women, as well as the elderly, are the group with the greatest propensity to use medications; they should, therefore, be the groups given preference when developing specific strategies for ensuring rational use.Higher numbers of medical appointments also significantly increased medication use, which reinforces the need to intensify promotional strategies in the day-to-day work of the health care services.Knowing the profile of medication use in the quilombola population is the first step in understanding access and discussing rational use.Aspects such as selfmedicating and polypharmacy, as well as the lifestyle, beliefs and values of the quilombolas in seeking health care and using medication need to be investigated in greater detail.

Table 2 .
Distribution of medications by group and subgroup according to anatomical and therapeutic classification (1 st and 2 nd levels of the ATC), a and ratio of number of medications per individual.COMQUISTA Project, Vitória da Conquista, BA, Northeastern Brazil, 2011.
a It includes anatomical therapeutic groups (1 st ATC level) with frequency above 2% and the most frequent therapeutic subgroups (2 nd ATC level), totaling less than 80.0% within each level.

Table 3 .
Prevalence and prevalence ratio (PR) for medication use, mean and standard deviation (SD) for the number of active ingredients according to the variables analyzed.COMQUISTA Project, Vitória da Conquista, BA, Northeastern Brazil, 2011.(N = 797) a Significant values (p < 0.05) b p estimated by variance analysis

Table 4 .
Prevalence ratios, adjusted for medication use, of the variables included in the final regression model.COMQUISTA Project, Vitória da Conquista, BA, Northeastern Brazil, 2011.