ABSTRACT.
We assessed the effectiveness of food vouchers as a social protection strategy to enhance the adherence to tuberculosis treatment in health-care facilities in Brazil between 2014 and 2017. A cluster-randomized controlled trial was performed in four Brazilian capital cities. A total of 774 adults with newly diagnosed pulmonary tuberculosis were included in this study. Eligible participants initiated standard tuberculosis treatment per National Tuberculosis Program guidelines. Health clinics were assigned randomly to intervention groups (food voucher or standard treatment). The food voucher was provided by researchers, which could be used by subjects only for buying food. Most people with tuberculosis were poor, did receive benefits of the Bolsa Familia Program, and were unemployed. By Poisson regression analysis, with the total number of subjects included in the study, we found that individuals with tuberculosis who received food vouchers had a 1.13 greater risk of cure (95% CI, 1.03–1.21) compared with those who did not receive food vouchers. The provision of food vouchers improved outcomes of tuberculosis treatment and it should be enhanced even further as social protection for people with tuberculosis.
INTRODUCTION
The 2020 End Tuberculosis (TB) Strategy milestones aimed to achieve a 35% reduction in TB deaths and a 20% reduction in incidence rates.1 The WHO 2019 Global TB report1 showed that, between 2015 and 2018, the global cumulative reduction in the number of TB deaths and incidence rates was only 11% and 6.3%, respectively. In Brazil, the state of TB control is of greater concern. During the same period, the number of TB deaths in Brazil declined by only 2.6% and the incidence rate, which was 34.1 per 10,000 inhabitants in 2015, increased to 34.8 per 10,000 inhabitants in 2018.2 Part of this failure to achieve milestones may be a result of unaddressed social and economic determinants of TB.
In this regard, integrating social, economic, and health goals into TB control becomes even more important today. A recent study using data from 192 countries suggested that expanding social protection schemes to all eligible people could decrease TB incidence by 76%.3 Brazilian evidence also indicates that national social protection programs, such as the Bolsa Família Program (BFP), although not specific to TB, have an important role in both the reduction of incidence and improvement of treatment outcomes.4,5
In Brazil, one important obstacle to improving TB cure rates is nonadherence.2 Accessing diagnosis and completing TB treatment successfully is often challenging for individuals who are living in poor and food-insecure environments.6 Considering this, and accounting for the vulnerability conditions of people with TB in Brazil, several municipalities in the country have developed policies to provide food baskets during treatment.
When examining the relationship between TB, poverty, and food insecurity, the evidence of the role of food provision in TB treatment is controversial. A Cochrane systematic review7 to assess the effectiveness of different food and nutritional supplements in helping people to gain weight and recover from TB found no evidence. However, other studies in Kenya,8 Afghanistan,9 and Angola,10 designed to improve adherence found that the provision of food improved outcomes of TB treatment. Between 2008 and 2012, under the support of the Global Fund, in Brazil, 10 metropolitan regions and Manaus implemented projects to benefit people with this social program. These initiatives, however, lacked a robust assessment of their effectiveness.
Our study assessed the effectiveness of food vouchers as a social protection strategy to enhance the adherence to TB treatment in health-care facilities in four Brazilian capital cities between 2014 and 2017.
METHODS
Study design, setting, and participants.
We investigated the effect of food vouchers on the outcomes of TB treatment in people from four Brazilian capital cities using a cluster randomized trial. Brazilian capital cities were selected according to TB incidence in 2012, the representativeness of administrative regions, and the absence of a governmental policy that provides food vouchers to people with TB. The selected cities included Manaus from the northern region, Fortaleza and Recife from the northeastern region, and Porto Alegre from the southern region. The main characteristics of the study settings are presented in Figure 1. Data collection was carried out from March 2014 to April 2017, and the investigations were conducted nested within a prospective cohort study.5
Figure 1.
Demographic characteristics of Brazilian capital cities selected in the study. FHS = family health strategy; HDI = Human Development Index; MHDI = municipal human development index; TB = tuberculosis.
We include in the cluster the eligible health-care services that offered TB treatment according to the National TB Program of the Ministry of Health recommendations.11 Services that had TB treatment success rates less than the WHO target of 85% and had diagnosed more than 20 people with TB in the year before the start of data collection were selected randomly to be included in the study. The only criterion for exclusion was the presence of an ongoing study, but no such centers existed and hence no centers were excluded.
People 18 years or older with a confirmed episode of TB, bacteriologically confirmed (bacilloscopy, Xpert Mycobacterium tuberculosis/rifampin assay, and/or culture) or based on an clinical epidemiological evaluation, according to the National TB Program, were included in this study.11
Intervention.
People diagnosed with TB who started treatment in the selected facilities were invited to participate in the study by the health center staff. The intervention was implemented pragmatically in the context of a routine TB program. Eligible individuals, who consented to participate, seen at the intervention units, received monthly, nontransferable food vouchers to purchase specified products during the entire period of TB treatment. The vouchers, serialized and printed centrally, were distributed to subjects by the local research focal point.
The food vouchers were valued, according to the regional market, at roughly 26 U.S. dollars (USD). The vouchers provided access to a food basket that must be redeemed within 30 days of initial receipt of the voucher. The food baskets were based on a list of preapproved foods at central supermarkets in each city, established according to the Brazilian staple of foods, accounting for regional specificities. The list of approved foods was composed of nonperishable foods, and it included rice (4 kg), beans (2 kg), regional flours (3 kg), sugar (2 kg), noodles (500 g), powdered milk (200 g), coffee (250 g), vegetable oil (900 mL), tomato sauce (350 g), salt (1 kg), canned fish (125 g), and crackers (400 g). The food baskets were stored and distributed by local partners nearest the health facilities.
Routine TB care includes a monthly appointment for the evaluation of the individual’s clinical status, laboratory and/or image examinations, and medications given.
The intervention group was called the “food voucher group” and the control group was called the “traditional treatment group.”
Sample size and sampling.
A cluster was defined as a health-care facility. The four selected cities had 25 health-care facilities that met the inclusion criteria; we excluded six facilities because they did not agree to participate or did not follow the study recommendations (Figure 2). The sample size was calculated to determine the average number of observations required per cluster for a two-sample comparison of proportions without continuity correction. Therefore, a sample size of 19 health facilities would provide 90% power to detect an increase in the cure rate of TB treatment from 60% to 80%, with an α level of 0.05, an intracluster correlation coefficient of 0.05, and a coefficient of variation of cluster sizes of 0.7. We increased the required sample size by 20% to account for potential participant withdrawal from the study or loss to follow-up.
Figure 2.
Number of facilities and subjects selected by type of cluster in intervention and traditional health-care units.
The intervention was assigned randomly at the cluster level in each study capital by an investigator independent of the data collection team. The sampling was performed by simple randomization, according to the number of people diagnosed with TB treated in the previous year. The health facility teams were informed of the allocation of their unit during the study training. It was not possible to mask the intervention because of the nature of the study design. All people diagnosed with TB during the intervention period at intervention sites were offered the intervention.
Measurements.
Previously trained health workers enrolled people diagnosed with TB into the study at the time of their diagnosis at the health-care centers. Individuals were interviewed at the time of treatment initiation, and information about the clinical assessments was gathered. In months 2 and 6 of treatment, we interviewed the subjects again. The collected information included sociodemographic, housing, behavioral, health history, and TB treatment information.
From the baseline assessment, the evaluated sociodemographic characteristics included gender (female, male), race/skin color (black/brown/other, white), schooling (0–3, 4–7, 8–10, 11–14, > 14 years), per-capita average income (in USDs), own household goods (number), employment (no, yes), the status of BFP beneficiary (no, yes), other social protection benefits (no, yes), and health insurance ownership (no, yes). The housing conditions included type of housing (masonry, wood, other); number of rooms; number of persons in the household; and presence of private bathroom, sewerage system, waste collection, and piped water (no, yes).
Behavioral characteristics included consumption of alcoholic beverages (no, yes) and tobacco smoking (never, former, current). Regarding the clinical variables, we assessed the presence of diabetes and HIV infection (no, yes). History of TB (no, yes), the pulmonary focus of current TB (no, yes), and assignment to directly observed therapy (DOT; no, yes) were also evaluated.
During months 2 and 6, we obtained additional information, including income, employment, and status of BFP beneficiaries.
The monetary poverty variable was created through the definition of the World Bank of $5.5/day, which corresponded to 387.07 Brazilian reais in 2016.12 Thus, those people with a per-capita monthly income less than $165 were classified as poor. In addition, poverty was also assessed by the dimensions of schooling (people with schooling of less than 8 years were classified as having school deprivation) and housing (people living in a household without a private bathroom, built with nondurable material, or with more than three people per room).12
Outcomes.
The primary outcomes were the proportion of individuals who initiated TB treatment at health facilities who were cured and the proportion of default at the end of treatment. These outcomes were based on definitions adopted by the Brazilian Ministry of Health.11 Thus, the outcomes were classified as “cure” for individuals with pulmonary TB who completed their treatment and showed two negative test results with sputum smear test or, for those who did not perform sputum smear testing, demonstrated clinical improvement with no changes in physical examination. Individuals with extrapulmonary TB were considered cured when they completed their treatment and had evidence of improvement in clinical radiological and/or other complementary examinations. Default was defined as the interruption of treatment of 30 days or more. People who did not meet these criteria were considered not cured or not defaulted and were not included in the primary outcome.
Statistical analyses.
Missing data were treated with an algorithm based on random forest multiple imputations by chained equations: the missForest.13,14 The random forest test uses bootstrap aggregation of multiple regression trees to reduce the risk of overfitting, combining predictions from many trees to produce less-biased parameter estimates and confidence intervals.13 We used multiple imputations by chained equation random forest with 1,000 trees. This technique allowed us to predict individual missing values accurately, imputing continuous and/or categorical data and accounting for complex interactions and nonlinear relations.14 The imputation analysis was performed with the program R Project (version 3.3.3; R Foundation for Statistical Computing, Vienna, Austria).
Data were analyzed using Stata 14 (Stata Corp., College Station, TX). The distribution of categorical variables and the median with interquartile range (IQR) of numerical variables were calculated. A Venn diagram depicting three poverty dimensions among the study population was constructed.
Poisson regression models with the vce (cluster) command, which specify that the standard errors allow for intragroup correlation, were used to examine the association of the intervention on cure and default rates. Therefore, the observations are independent across health facilities (clusters), but not necessarily within them. The dependent variables were cure or default; the independent variable was the intervention. The adjusted models included the propensity score for each individual.15 The propensity score was estimated by a logit model with a significance level of 0.01, including the variables race/skin color, schooling, per-capita average income, employment, the status of BFP beneficiary, number of persons per dormitory, index of goods, HIV infection, other comorbidities, history of TB, and pulmonary focus of current TB. The estimates were presented as risk ratios (RRs) and 95% CIs. Furthermore, we replicated the analyses according to the assignment to DOT to assess different performances of the intervention between groups.
The study was approved by the ethics committee of the Health Sciences Center of the Federal University of Espírito Santo (No. 733541), July 30, 2014. All individuals in the study provided written informed consent. The protocol was registered at Registro Brasileiro de Ensaios Clínicos under No. UTN U1111-1221-9218.16
RESULTS
The required sample size per arm was 228 individuals. A total of 774 people diagnosed with TB were enrolled at 19 health-care facilities between 2014 and 2017. According to capital cities, 269 people (35%) lived in Manaus, 303 (39%) in Porto Alegre, 109 (14%) in Recife, and 93 (12%) in Fortaleza. Of the total, 231 individuals (30%) received food vouchers during treatment and 543 individuals (70%) received standard treatment. During months 2 and 6 of treatment, the withdraw rate was 2% (758 participants completed the stage) and 21% (614 participants completed the stage), respectively.
Characteristics of the study population at baseline are shown in Table 1. The food voucher group was composed of 133 women (58%), 127 of whom (55%) had with fewer than 8 years of schooling, and 33 (14%) and 18 (8%) persons living with diabetes and HIV, respectively. Thirty-seven (16%) had a history of TB and 112 (48%) were assigned to DOT. The standard treatment group included 330 women (61%), 253 people (46%) with fewer than 8 years of schooling, and 47 (9%) and 77 (14%) persons living with diabetes and HIV, respectively. Of these people, 105 (19%) had a history of TB and 52 (10%) were assigned to DOT.
Table 1.
Study population characteristics
| Characteristic | Traditional treatment, n (%) | Food voucher, n (%) |
|---|---|---|
| Gender | ||
| Female | 330 (61) | 133 (58) |
| Male | 213 (39) | 98 (42) |
| Skin color | ||
| Black/brown/other | 274 (50) | 196 (85) |
| White | 269 (50) | 35 (15) |
| Schooling, years | ||
| 0–3 | 111 (20) | 67 (29) |
| 4–7 | 142 (26) | 60 (26) |
| 8–10 | 85 (16) | 42 (18) |
| 11–14 | 146 (27) | 52 (23) |
| > 14 | 59 (11) | 10 (4) |
| Health insurance | ||
| No | 414 (76) | 204 (88) |
| Yes | 129 (24) | 27 (12) |
| Tobacco smoking | ||
| Never | 290 (53) | 95 (41) |
| Former | 139 (26) | 102 (44) |
| Current | 114 (21) | 34 (15) |
| Drinking of alcoholic beverages | ||
| No | 374 (69) | 179 (77) |
| Yes | 169 (31) | 52 (23) |
| Diabetes | ||
| No | 496 (91) | 198 (86) |
| Yes | 47 (9) | 33 (14) |
| HIV | ||
| No | 466 (86) | 213 (92) |
| Yes | 77 (14) | 18 (8) |
| History of tuberculosis | ||
| No | 438 (81) | 194 (84) |
| Yes | 105 (19) | 37 (16) |
| Pulmonary tuberculosis | ||
| No | 155 (29) | 29 (13) |
| Yes | 388 (71) | 202 (87) |
| Directly observed therapy | ||
| No | 491 (90) | 119 (52) |
| Yes | 52 (10) | 112 (48) |
Figure 3 depicts the overlap of the poverty dimensions among the entire study population. As shown in the Venn diagram, 49%, 64%, and 14% of the sample belonged to each of the three poverty dimensions of schooling, income, and housing, respectively. Sixty-one people (8%) were included among all three poverty dimensions. The criteria to be classified as schooling, income, or housing poverty alone was met by 75 individuals (10%), 181 individuals (23%), and 12 individuals (2%), respectively. Most participants belonged to at least two poverty dimensions, and only 180 participants (23%) were classified under the dimension of no poverty.
Figure 3.
Venn diagram of poverty dimensions.
The median monthly per-capita income (in USDs) was less among the food voucher group during treatment (Table 2). It was $67.02 (IQR, $29.49–$134.05) at baseline, $71.16 (IQR, $39.95–$117.96) in month 2 month, and $71.39 (IQR, $38.20–$134.05) in month 6 of treatment. However, the median monthly per-capita income among the traditional treatment group was also less than the threshold of $5.5 daily. Unemployment reached nearly 50% among the intervention group and 30% among the control group during all treatments. Despite this, beneficiaries of BFP at baseline included 45 people (19%) and 84 people (15%) for the food voucher and traditional treatment groups, respectively, without significant variation over time.
Table 2.
Distribution of social characteristics during tuberculosis treatment
| Characteristic | Baseline (n = 774) | Month 2 (n = 758) | Month 6 (n = 614) | |||
|---|---|---|---|---|---|---|
| Traditional treatment | Food voucher | Traditional treatment | Food voucher | Traditional treatment | Food voucher | |
| Income, median (IQR) | 118.34 (67.02–234.58) | 67.02 (29.49–134.05) | 117.96 (67.49–238.61) | 71.16 (39.95–117.96) | 133.88 (74.32–235.92) | 71.39 (38.20–134.05) |
| Employment, n (%) | ||||||
| No | 180 (33) | 117 (51) | 187 (35) | 133 (58) | 140 (32) | 89 (52) |
| Yes | 363 (67) | 114 (49) | 343 (65) | 95 (42) | 304 (68) | 81 (48) |
| Bolsa Família Program, n (%) | ||||||
| No | 459 (85) | 186 (81) | 470 (89) | 182 (80) | 383 (86) | 137 (81) |
| Yes | 84 (15) | 45 (19) | 60 (11) | 46 (20) | 61 (14) | 33 (19) |
| Poor, n (%) | ||||||
| No | 315 (58) | 84 (35) | 306 (58) | 76 (33) | 283 (63) | 63 (37) |
| Yes | 228 (42) | 147 (65) | 224 (42) | 152 (67) | 161 (36) | 107 (63) |
IQR = interquartile range.
Table 3 shows the distribution of the study population, according to the outcomes of TB treatment. Among the food voucher group, 214 individuals (93%) were cured; 466 participants (86%) in the traditional treatment group were cured. Default occurred in 14 people (6%) and 60 people (11%) in the food voucher and traditional treatment groups, respectively.
Table 3.
Distribution and estimates of the effect of food vouchers on tuberculosis treatment outcomes
| Intervention | Not cured, n (%) | Cured, n (%) | Not default, n (%) | Default, n (%) |
|---|---|---|---|---|
| Food voucher | ||||
| No | 77 (11) | 466 (86) | 483 (89) | 60 (11) |
| Yes | 17 (7) | 214 (93) | 217 (94) | 14 (6) |
| Poisson regression models* RR (95% CI) | ||||
|---|---|---|---|---|
| Participants | Cure | Default | ||
| Total population† | ||||
| Unadjusted | 1.08 (1.02–1.14) | 0.55 (0.33–0.91) | ||
| Adjusted | 1.13 (1.03–1.21) | 0.39 (0.23–0.66) | ||
| Not under directly observed therapy‡ | ||||
| Unadjusted | 1.04 (0.98–1.11) | 0.76 (0.44–1.31) | ||
| Adjusted | 1.10 (1.02–1.18) | 0.58 (0.33–1.02) | ||
| Under directly observed therapy§ | ||||
| Unadjusted | 1.15 (0.95–1.40) | 0.30 (0.09–1.00) | ||
| Adjusted | 1.18 (1.00–1.42) | 0.18 (0.06–0.54) | ||
Models adjusted by propensity score estimated by a logit model with significance level of 0.01, including the variables race/skin color, schooling, per-capita average income, employment, status of Bolsa Família Program beneficiary, number of persons per dormitory, index of goods, HIV infection, other comorbidities, history of tuberculosis (TB), and pulmonary focus of current TB.
Propensity score: mean, 0.298; SD, 0.205.
Propensity score: mean, 0.269; SD, 0.195.
Propensity score: mean, 0.409; SD, 0.203.
The cure was 13% greater (RR, 1.13; 95% CI, 1.03–1.21) among the intervention group compared with the traditional treatment group. The stratification by DOT showed an adjusted RR of 1.10 (95% CI, 1.02–1.18) for those not assigned and an RR of 1.18 (95% CI, 1.00–1.42) for those assigned also to DOT. Also, even after controlling for confounding, people with TB who received the food voucher intervention had a risk of default that was 61% less than those given traditional treatment. Although this effect has been somewhat attenuated by the absence of DOT therapy (RR, 0.58; 95% CI, 0.33–1.02), it was even greater among persons under food voucher intervention and DOT (RR, 0.18; 95% CI, 0.06–0.54).
DISCUSSION
In this cluster randomized controlled trial, we found that food vouchers indeed contributed significantly to improving adherence to treatment or successful completion of treatment. The food vouchers are not implemented in Brazil as a regular program in any of the cities that had individuals assigned for this study. A qualitative study performed in São Paulo, Brazil,17 showed that incentives such as the basic food basket and transportation stipends are relevant to patients’ adherence to treatment. Another study in Angola10 showed that during the period when basic food baskets were distributed to people with TB, the success rate of treatment was greater than during the period when there were no food baskets provided. Similar results were found in Moldova and Russia.18,19
Default from TB treatment remains a public health problem faced by many national TB programs, especially in low- and middle-income countries such as Brazil.1,20 The association between poverty and default from TB treatment is well demonstrated in the literature.4–7 Social protection interventions are promoted under pillar 2 of the WHO End TB Strategy. Previous studies have reported that poverty and TB are strongly associated.4,5,21 In our study, carried out in five Brazilian state capital cities with high TB incidences and social inequalities, 40% of participants had less than 8 years of schooling, 38% were unemployed, and 30% did not own a house. In addition, more than three fourths of the subjects belonged to a poverty stratum, and we observed that the group that received the food voucher remained among the poorest during all phases of TB treatment. The effects of housing status, low education, and being unemployed have a common feature of low socioeconomic status that increases the risk of TB and decreases the prioritization for treatment, as shown in other studies.21,22
Schooling as a measure of poverty has also been shown in other studies of individuals with TB. The frequency of illiteracy and a fewer number of years of education are common in this group.4,5,10 The close relationship between schooling and socioeconomic status reinforces the strong social determinant of TB and the need for innovative interventions in association with traditional treatment to mitigate this global public health concern.
Recent results from East Timor,23 Kenya,8 Afghanistan,9 and Angola10 using different study designs to assess improvement in adherence found that, in East Timor, provision of food did not improve outcomes with TB treatment, whereas the studies in Kenya, Afghanistan, and Angola found that nutritional support in their Food Supplementary Program was associated with a reduction in the risk of loss to follow-up. It should be noted that, apart from Brazil, these countries have a high prevalence of malnutrition and micronutrient deficiencies, and most of them have a high burden of TB/HIV comorbidities and lack of social protection.
Multimorbidity and other chronic diseases have been associated with worse TB outcomes. A previous study24 with the Brazilian population found that TB among persons with diabetes increased from 380/100,000/year in 2001 to 6,150/100,000/year in 2011. Individuals who had TB and HIV compared with those with TB only had a greater risk of default and death from TB.25 The participants of these studies are remarkably like those included in our study, suggesting that people with TB and HIV, and those with diabetes, are more likely to default. In addition, in other analyses of the Brazilian national surveillance system data, subjects who did not receive DOT were more likely to default from anti-TB treatment and die as a result of the course of the diseases; the adjusted preventable fraction of DOT in the reduction of unfavorable outcomes was 25%.26
Unfortunately, even with these previous results, our data show an impressively low rate of DOT in persons with TB of only 10%. Because of this, we also controlled the analysis by the administration of DOT. In Brazil, a country with a recognized universal health system called Sistema Único de Saúde, DOT is available to people with TB, as already demonstrated on TB treatment outcomes.26 A longitudinal database study has raised a hypothesis that this strategy associated with cash transfer could have a positive synergistic effect.27 These findings show the potential of two strategies in the national programs for TB control: food vouchers and DOT.
Although the results we report here are important to enhance TB treatment adherence, it is crucial to acknowledge the limitations of our study. We carried out an open trial, in which professionals and people with TB were not blinded to the intervention. Because we randomized the health units and people diagnosed with TB, which may vary in the geographic areas, important socioeconomic indicators such as income, unemployment, and the provision of DOT differ among them. In Brazil, most people seeking access to primary health care are women,28 and this may have affected the sample with participation bias. Because we did not collect data on the refusal rate to participate in the study, it was not possible to assess this bias accurately. With DOT, health services are very different and depend on the professionals available to perform them. In a previous study,26 only 10% of people being treated for TB in Brazil were monitored with DOT and, in the sample studied, only 58 people (7.5%) were allocated to DOTs; the proportion of participants with a voucher versus without a voucher was 5:1. To overcome this, we used propensity score matching analysis to create more homogeneous groups to make comparisons, but this difference needs to be pointed out.
This is the first prospective study conducted in Brazil that enrolled people diagnosed with TB in different regions of the country. In a previous retrospective study29 in Duque de Caxias, Rio de Janeiro State, participants were divided into two groups: those with food baskets provided monthly during TB treatment and those without any incentive. The study revealed that the cure rate was greater in the group receiving the food baskets (87.1% versus 69.7%), and the rate of noncompliance was markedly less (12.9% versus 30.3%). The WHO, based on the links between poverty/food insecurity and TB treatment outcome, has endorsed this strategy as a key element in the global response to TB control.30 Other schemes were evaluated in Peru in a randomized controlled study evaluating the effect of conditional cash transfers (≤$230 per household) aimed at enhancing TB prevention and treatment. They showed that cash transfers improved treatment outcomes.31 A study carried out in Singapore32 demonstrated the salutary effect of a nongovernmental organization–funded grocery voucher incentive scheme for low-income persons with TB on DOT. They had a greater treatment completion rate (90.0%) than those not under this scheme (86.4%).
Despite these observations, it is important to point out that the success of strategies based on food benefits, vouchers, or baskets is not explained biologically. The effect of these interventions is not only a result of an improvement in the nutritional status of people undergoing TB treatment, but also is a result of alleviation of the social vulnerability burden that affects their entire family, as highlighted by the observed large reduction in default rates. Therefore, food voucher programs are a matter of inclusion and human rights for people with TB in social development policies.
In conclusion, our study strongly suggests that food vouchers contribute to better TB treatment outcomes. We thus recommend the integration of such social protection policies into TB control program strategies to reduce TB defaults.
ACKNOWLEDGMENTS
The authors would like to honor University of California School of Public Health professor Lee W. Riley, a world-renowned leader in the field of infectious diseases and vaccinology, a friend and mentor to many, and one of the authors of this article, who passed away on October 19, 2022. During his career he made a remarkable contribution to science, and education and training of health professionals, especially in Brazil. His teachings will always remain with us. We confirm that all ongoing and related trials for this drug/intervention are registered at Registro Brasileiro de Ensaios Clínicos (U1111-1221-9218).
REFERENCES
- 1. World Health Organization , 2019. Global Tuberculosis Report 2019. Geneva, Switzerland: WHO. [Google Scholar]
- 2.Ministério da Saúde, Sistema de Informação de Agravos de Notificação [homepage on the Internet]. Tuberculose - Casos Confirmados Notificados no Sistema de Informação de Agravos de Notificação - Brasil. Available at: http://tabnet.datasus.gov.br/cgi/tabcgi.exe?sinannet/cnv/tubercbr.def. Accessed October 26, 2022.
- 3. Carter DJ et al. 2018. The impact of social protection and poverty elimination on global tuberculosis incidence: a statistical modelling analysis of Sustainable Development Goal. Lancet Glob Health 6: e514–e522. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Nery JS et al. 2017. Effect of Brazil’s conditional cash transfer programme on tuberculosis incidence. Int J Tuberc Lung Dis 21: 790–796. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Oliosi JG et al. 2019. Effect of the Bolsa Familia Programme on the outcome of tuberculosis treatment: a prospective cohort study. Lancet Glob Health 7: e219–e226. [DOI] [PubMed] [Google Scholar]
- 6. Rajeswari R Balasubramanian R Muniyandi M Geetharamani S Thresa X Venkatesan P , 1999. Socio-economic impact of tuberculosis on patients and family in India. Int J Tuberc Lung Dis 3: 869–877. [PubMed] [Google Scholar]
- 7. Grobler L Nagpal S Sudarsanam TD Sinclair D , 2016. Nutritional supplements for people being treated for active tuberculosis. Cochrane Database Syst Rev 6: CD006086. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Mansour O Masini EO Kim BJ Kamene M Githiomi MM Hanson CL , 2018. Impact of a national nutritional support programme on loss to follow-up after tuberculosis diagnosis in Kenya. Int J Tuberc Lung Dis 22: 649–654. [DOI] [PubMed] [Google Scholar]
- 9. Pedrazzoli D Houben RM Grede N de Pee S Boccia D , 2016. Food assistance to tuberculosis patients: lessons from Afghanistan. Public Health Action 6: 147–153. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Santos E Felgueiras Ó Oliveira O Duarte R , 2018. The effect of a basic basket on tuberculosis treatment outcome in the Huambo Province, Angola. Arch Bronconeumol 54: 167–168. [DOI] [PubMed] [Google Scholar]
- 11.Medeiros, A et al., 2021. Guia de Vigilância em Saúde. Brasília, DF: Ministério da Saúde, Secretaria de Vigilância em Saúde.
- 12.Banco Central do Brasil, 2016 [homepage on the Internet]. Consulta de Cotações e Boletins. Available at: https://www.bcb.gov.br/estabilidadefinanceira/historicocotacoes. Accessed October 26, 2022.
- 13.Crespo, C et al., 2017. Síntese de Indicadores Sociais: Uma Análise das Condições de Vida da População Brasileira: 2017. Rio de Janeiro, Brazil: Instituto Brasileiro de Geografia e Estatística. [Google Scholar]
- 14. Stekhoven DJ Bühlmann P , 2012. MissForest: non-parametric missing value imputation for mixed-type data. Bioinformatics 28: 112–118. [DOI] [PubMed] [Google Scholar]
- 15. Shah AD Bartlett JW Carpenter J Nicholas O Hemingway H , 2014. Comparison of random forest and parametric imputation models for imputing missing data using MICE: a CALIBER study. Am J Epidemiol 179: 764–774. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Ministério da Saúde, Organização Panamericana de Saúde, Fundação Oswaldo Cruz [homepage on the Internet]. Registro Brasileiro de Ensaios Clínicos Available at: https://ensaiosclinicos.gov.br/. Accessed October 26, 2022.
- 17. Orlandi GM Pereira ÉG Biagolini REM França FOS Bertolozzi MR , 2019. Social incentives for adherence to tuberculosis treatment. Rev Bras Enferm 72: 1182–1188. [DOI] [PubMed] [Google Scholar]
- 18. Ciobanu A et al. 2014. Do incentives improve tuberculosis treatment outcomes in the Republic of Moldova? Public Health Action 4: S59–S63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Garden B et al. 2013. Food incentives improve adherence to tuberculosis drug treatment among homeless patients in Russia. Scand J Caring Sci 27: 117–122. [DOI] [PubMed] [Google Scholar]
- 20.Sanchez, D et al., 2019. Manual de Recomendações para o Controle da Tuberculose no Brasil. Brasília, DF: Ministério da Saúde, Secretaria de Vigilância em Saúde, Departamento de Vigilância das Doenças Transmissíveis.
- 21. Shete PB Reid M Goosby E , 2018. Message to world leaders: we cannot end tuberculosis without addressing the social and economic burden of the disease. Lancet Glob Health 6: e1272–e1273. [DOI] [PubMed] [Google Scholar]
- 22. de Castro DB de Seixas Maciel EMG Sadahiro M Pinto RC de Albuquerque BC Braga JU , 2018. Tuberculosis incidence inequalities and its social determinants in Manaus from 2007 to 2016. Int J Equity Health 17: 187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Martins N Morris P Kelly PM , 2009. Food incentives to improve completion of tuberculosis treatment: randomised controlled trial in Dili, Timor-Leste. BMJ 339: b4248. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Reis-Santos B Gomes T Locatelli R de Oliveira ER Sanchez MN Horta BL Riley LW Maciel EL , 2014. Treatment outcomes in tuberculosis patients with diabetes: a polytomous analysis using Brazilian surveillance system. PLoS One 9: e100082. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. do Prado TN Miranda AE de Souza FM Dias Edos S Sousa LK Arakaki-Sanchez D Sanchez MN Golub JE Maciel EL , 2014. Factors associated with tuberculosis by HIV status in the Brazilian national surveillance system: a cross sectional study. BMC Infect Dis 14: 415. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Reis-Santos B et al. 2015. Directly observed therapy of tuberculosis in Brazil: associated determinants and impact on treatment outcome. Int J Tuberc Lung Dis 19: 1188–1193. [DOI] [PubMed] [Google Scholar]
- 27. Reis-Santos B et al. 2019. Tuberculosis in Brazil and cash transfer programs: a longitudinal database study of the effect of cash transfer on cure rates. PLoS One 14: e0212617. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Guibu IA, Álvares J. 2017. Main characteristics of patients of primary healthcare services in Brazil. Rev Saúde Pública 51 (Suppl 2): 1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Cantalice Filho João Paulo , 2009. Food baskets given to tuberculosis patients at a primary health care clinic in the city of Duque de Caxias, Brazil: effect on treatment outcomes. J Bras Pneumol 35: 992–997. [DOI] [PubMed] [Google Scholar]
- 30. World Health Organization , 2015. The End TB Strategy 2015. Geneva, Switzerland: WHO. [Google Scholar]
- 31. Wingfield T et al. 2017. Randomized controlled study of socioeconomic support to enhance tuberculosis prevention and treatment, Peru. Bull World Health Organ 95: 270–280. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Chua AP, Lim LK, Ng H, Chee CB, Wang YT, 2015. Outcome of a grocery voucher incentive scheme for low-income tuberculosis patients on directly observed therapy in Singapore. Singapore Med J 56: 274–279. [DOI] [PMC free article] [PubMed] [Google Scholar]