TB and poverty: the effect of rifampicin-resistant TB on household income

SUMMARY SETTING The Republic of Moldova, one of Europe's poorest countries, also bears one of the highest burdens of rifampicin-resistant TB (RR-TB). OBJECTIVES To trace the patients’ journey through TB in terms of the relationship with poverty and assess its determinants. DESIGN This cross-sectional study used secondary data from a survey assessing catastrophic costs in RR-TB-affected households. RESULTS Data were obtained from 430 RR-TB patients. The percentage of poor TB-affected households rose from 65% prior to TB to 86% after TB treatment completion (P < 0.001). Social factors leading to poverty were identified for each stage: diagnostic period (history of incarceration: cOR 2.3, 95% CI 1.1–5.2); treatment period (being unemployed or unofficially employed: cOR 6.7, 95% CI 4.3–10.0); and post-treatment (being married or cohabiting: cOR 5.7, 95% CI 2.9–11.0). Participants who had ≥3 members in their households were more likely to be poor at all TB stages: diagnostic period (cOR 5.7, 95% CI 3.7–8.8), treatment period (cOR 3.8, 95% CI 2.5–5.6) and post-treatment (cOR 7.2, 95% CI 3.6–14.3). CONCLUSION The study identified risk factors associated with poverty at each stage of TB. These findings outline that innovative social protection policies are required to protect TB patients against poverty.

Equity in access to healthcare is today's greatest health challenge, and it will be achieved by accelerating the progress in healthcare provision to poor and socially vulnerable groups. 1 There is a strong link between poverty and health, 2 and poverty and TB have a symbiotic relationship.The WHO's Global TB report provides substantial evidence that poverty is an enabler of TB, 3 and there is a strong association between gross domestic product (GDP) per capita and TB incidence. 4Reducing poverty, advocating improved equity of access to health care and universal health care are key components of WHO's End TB Strategy 2016-2035. 5he Republic of Moldova is a lower-middle-income country where the poverty headcount ratio at national poverty lines is 23% of the population. 6The gross annual national income in the country is US$3,930 per capita. 7The Republic of Moldova is one of the 18 high-priority countries for TB of the WHO European Region, 8 and one of the 30 countries with the highest burdens of rifampicin-resistant TB (RR-TB) cases in the world, despite managing, since 2007, to reverse the rising TB incidence rate.In recent years, the TB incidence rate has continued to decline by an average of 3.2% per year. 3,8The successful treatment of RR-TB cases was 53% in 2016. 8irect TB investigations and all TB treatment are free of charge in the country. 9However, despite this, patients can face catastrophic health expenditures because of the high costs of diagnostic period care and the high indirect costs incurred during treatment.The survey conducted in 2016 showed that catastrophic costs were experienced by 26% of households affected by RR-TB, and also found that patients with RR-TB lost 30% of their income in inpatient care and 70% in outpatient care. 10nderstanding the bidirectional relationship between TB and poverty would be a powerful step towards breaking this vicious cycle and reducing the TB burden.This study aimed to determine the proportion of households affected by RR-TB who experienced poverty during the diagnostic period, during treatment and after the completion of TB treatment, and to assess the risk factors associated with poverty at each TB stage.The study findings will allow decision-makers to identify solutions and interventions for both financial and social protection to reduce poverty, and to improve TB control in the country.

Study design
This is cross-sectional study using secondary data from patients with RR-TB identified in 2016 survey on catastrophic costs carried out in the Republic of Moldova.

General setting
The Republic of Moldova is situated in south-eastern Europe and has a population at about 3 million inhabitants (including about 0.5 million inhabitants in the breakaway region of the Left Bank of the River Dniester).

National tuberculosis control
The Ministry of Health is the body primarily responsible for TB control in the country.The National Tuberculosis Control Programme (NTP) was initiated in 2001 and has been approved every 5 years since its inception.TB patients receive treatment in both inpatient and outpatient care; in 2016, approximately 90% of patients with RR-TB were hospitalised for an average of 5 months.Direct TB investigations and all TB medicines are free of charge in the country.Health insurance coverage in the general population is 70-75%, 11 but in TB patients, this is 30-35%. 12The lack of health insurance can mean that it is difficult for TB patients to access medical services before their TB diagnosis has been established and to receive medications for side-effects in special outpatient care.Temporary disability allowances and adherence incentives for the entire period of treatment are available for TB patients as welfare benefits.Temporary disability allowances are granted for those who were officially employed before being diagnosed with TB and cover 100% of wages during the entire period of treatment, and incentives are given monthly for taking 90% of doses given in outpatient care.Two types of incentive amounts were paid during the study period: US$20 per month in the period before 1 July 2015 and US$56 per month in the period after 1 July 2015.There is a national TB case-based database -Information System for Monitoring and Evaluation TB patients (SIME TB) -for registering TB patients and follow-up. 12

Study population
The study population in the baseline study 10 included 430 RR-TB patients with the following inclusion criteria: new or relapsed RR-TB cases; adult (�18 years); receiving treatment for the intensive phase of TB treatment in hospital for at least 2 months, or receiving treatment for the continuation phase of treatment as outpatients for at least 2 months, or having completed treatment for at least 2 months before interview.Patients from prisons and the Left Bank of the River Dniester were excluded.In this study, the term 'household' is defined as a group of relatives or non-relatives dwelling under one roof and sharing one budget, or a person living alone and selfmanaging who does not belong to any other household.

Data collection, sources and variables
The stages of TB were defined as the diagnostic period (2 months before TB diagnosis), treatment period (from the beginning of treatment to the end of treatment) and the post-treatment period (2 months after treatment completion).The procedures for data collection from the questionnaires are described in the original study. 10Data extracted from the original study were social and demographic variables (sex, age, residence, education, marital status, employment, household size, health insurance, history of labour migration and incarceration, smoking status, alcohol consumption, drug use), clinical variables (history of TB treatment, presence of HIV coinfection, presence of diabetes mellitus, presence of comorbidities and presence of side effects during TB treatment), sources of finance used to pay for out-of-pocket (OOP) expenditure (savings, sale of possessions, uptake of loans and income) and household income.Each respondent was interviewed just once, reporting on TB-related costs for the specific phase he/she was interviewed for.To measure the income for each household, we established the income earned by patients and other family members from wages, welfare benefits, adherence incentives and other income sources.To estimate the entire household income, we extrapolated the monthly reported income for the diagnostic period, the treatment period and the posttreatment period for all study participants.Using the World Bank's poverty line standard of US$1.90 per person per day, 13 we grouped households into poor (earning �US$1.9 per capita per day) and non-poor (earning .US$1.9 per capita per day) households.

Data analysis
Extracted data (from the original study) were imported into SPSS ® Statistics v20.0 (IBM, Armonk, NY, USA) for analyses.The data were described using frequency, proportions, mean (standard deviation [SD]), and median (interquartile range [IQR]).The v 2 test was used to assess patients' profiles and to compare proportions, and t-tests were used for continuous variables.If the expected cell frequency was less than 5, Fisher's Exact test was used instead of the v 2 test.The Kruskal-Wallis test was used to compare medians of incomes between groups.The crude odds ratio (cOR) was selected as a measure of association in the analysis of poverty determinants.We used the RawGraphs web-application to construct an alluvial plot to visualise the change in household poverty at the diagnostic period, treatment period and posttreatment stages of TB. 14 Cochran's Q test was selected to determine the differences of poverty levels between TB stages.For all analyses, levels of significance were set at 5%.

Ethical approval
The study protocol was approved by the National Committee for Ethical Expertise of Clinical Trial, Chisinau, Republic of Moldova.Written informed consent was obtained from all participants of the baseline study.

RESULTS
Demographic and clinical characteristics of study participants are shown in the Table .Of the 430 participants, 336 (78%) were male and 94 (22%) were female with an average age of 42 years (SD �12.4); 150 participants were inpatients in the intensive phase of treatment, 137 patients were outpatients in the continuation phase of treatment and 143 participants had completed treatment.Most participants were from rural areas (63%), were unofficially employed or unemployed (62%) and had no health insurance (63%) (Table ).
We measured household poverty at each stage of TB.In this study, 65% (280/430) of households were poor prior to TB and 86% were poor (368/430) after completion of TB treatment.More than a third (55/ 150, 37%) of those who were non-poor before TB became poor during TB treatment.After treatment completion, 73% (140/192) of those who were nonpoor during TB treatment became poor, and only 4% (10/238) of those who were poor during TB treatment became non-poor after treatment completion.Our analysis showed a strong relationship between poverty and TB pathways (P , 0.001) (Figure 1).
The median monthly income of TB-affected households who were classified as poor before TB diagnosis was US$100 (IQR 37-155), and this was US$94 (IQR 94-94) after treatment completion.There was a 1.4-fold decrease in median monthly income between pre-TB diagnosis and during TB treatment among those who were classified as poor at the TB treatment stage.TB-affected households who became poor after treatment completion had a median monthly income of US$94 (IQR 94-100), i.e., their income was 1.6-fold lower than that before TB (Figure 2).
Unadjusted analyses revealed several risks factors associated with poverty at each TB stage.TB patients with three or more people in their households were more likely to be poor before diagnosis (cOR 5.7, 95% CI 3.7-8.8),during treatment (cOR 3.8, 95% CI 2.5-5.6) and after treatment completion (cOR 7.2, 95% CI 3.6-14.3).Having a history of incarceration was a risk factor for poverty (cOR 2.3, 95% CI 1.1-5.2);however, being a labour migrant reduced the risk of poverty prior to TB (cOR 0.4, 95% CI 0.2-0.7).Those who were unofficially employed or unemployed were more likely to become poor during treatment (cOR 6.7, 95% CI 4.3-10.0).Being married or living with a partner was another risk factor for poverty after treatment completion (cOR 5.7, 95% CI 2.9-11.0)(Table ).
Approximately half (48%, 207/430) of TB patients experienced OOP expenditures between TB diagnosis and the end of TB treatment.Of these patients, 62% (174/280) were from poor households and 49% (74/ 150) were from non-poor households (P , 0.010).Both poor and non-poor TB patients had been obliged to use household savings, their income, to sell their possessions and to take out loans to cover OOP payments.Non-poor TB patients used savings and income to pay for OOP payments (P , 0.001; Figure 3).

DISCUSSION
This study traces the patient's journey through TB treatment and the relationship between TB treatment and poverty among patients in the Republic of Moldova in the WHO European Region.The results of this study confirm the relationship between TB and household poverty and present a few topics for discussion.
TB thrives on conditions of poverty and having TB can worsen poverty; 1 however, poverty is difficult to measure as it is multifaceted.In this study, we measured economic poverty based on the World Bank criterion of poverty as people living on US$1.90 per person per day or less. 13The study showed that a high proportion of TB patients were poor before having TB, and economically poor people are at greater risk of TB infection than the general population. 1 When we examined poverty during the different stages of TB (diagnostic, treatment and post-treatment), the proportion of participants who were poor during TB treatment decreased but increased after treatment completion.One explanation for the decrease in the proportion of poor patients during treatment is that patients with TB received welfare transfers (temporary allowances and adherence incentives) during treatment, which may have increased their household incomes.However, these temporary allowances are only given to patients with TB if they were officially employed before TB.During TB treatment, people who were not officially employed or who were not employed before TB were more likely to be living in poverty.
It has previously been shown that households can be driven into poverty once affected by TB, irrespective of the demographic or clinical characteristics, 15,16 and this was also observed in this study.The main risk factors associated with poverty at each TB stage were social factors; for example, households with three or more members were more likely to be poor during all TB stages.This may be because low household incomes cannot cover the costs of the needs of each family member, especially if the household contains children and/or dependants.The average monthly income for poor households affected by TB was about three times lower than the average monthly income in the Republic of Moldova. 17Patients with TB who were married or cohabiting were likely to become poor after treatment completion due to an imbalance between the household income and the high expenses incurred during TB treatment that was observed among non-poor households.In addition, the original study 10 showed that households experienced TB-related costs at the same time that their incomes were shrinking, thus creating a 'medical poverty trap'. 18,19aving a history of incarceration was a risk factor for poverty before being infected with TB; both poverty and TB are common experiences for ex-prisoners.Previous incarceration is a known risk factors for latent TB infection (LTBI) and the development of active TB. 20Ex-prisoners often face barriers to employment and are more vulnerable to financial instability.
Being a labour migrant was found to be a protective factor for poverty before TB.In Moldova, government wages, social pensions and other unemployment benefits are small and, consequently, there are many people who are looking for informal work or who go to work abroad; 10% of the working age population is working or looking for a job abroad. 21he limitations of this study include the use of selfreported data for incomes of the respondents and their households, the use of estimation for the entire household income and the extrapolation of monthly

Figure 1 .
Figure 1.Change in household income over time according to TB treatment stage for patients with RR-TB.Alluvial plot representing economic changes of poor (orange) and non-poor (blue) patients with RR-TB during TB stages from before being diagnosed with TB to after TB treatment completion.The three bars represent the TB stages: diagnostic period, treatment period and post-treatment period.Comparison between three groups (diagnostic period, treatment period, post-treatment) P , 0.001 using the Cochran test.RR-TB ¼ rifampicin-resistant TB.

Figure 3 .
Figure 3. Sources of finance used to cover out-of-pocket payments in (defined as household income �US$1.90person/day) and non-poor (defined as household income .US$1.90 person/day) households.Error bars represent 95% confidence intervals.v 2 tests used for proportions.*Based on household income prior to TB.

Table . A
) Sociodemographic and clinical characteristics of poor and non-poor patients with RR-TB during the TB diagnostic period and other factors contributing to poverty.*
* Poverty defined according to World Bank criterion and based on household income prior to TB diagnosis.For statistical analysis, v 2 tests were used unless expected cell frequency was less than 5 in which case Fisher's exact test was used.Missing data were excluded during hypothesis testing.†Data available from 427 patients.‡ Data available from 419 patients.RR-TB ¼ rifampicin-resistant TB; OR ¼ odds ratio; CI ¼ confidence interval; ref ¼ reference.