The nancial burden of tuberculosis-affected households in China, 2017

Background Drug-susceptible pulmonary tuberculosis patients notied and treated under national tuberculosis program, China Methods We conducted a cross-sectional study in six provinces in 2017 to assess the burden and distribution of health expenditure costs among tuberculosis-affected households and analyze related risk factors. The data was collected through face-to-face questionnaire survey. Intensity of nancial burden was the median positive overshoot beyond a threshold that dened as annual total (direct and indirect) costs exceeding 20% of the pre-TB annual household income.


Abstract
Background Drug-susceptible pulmonary tuberculosis patients noti ed and treated under national tuberculosis program, China

Methods
We conducted a cross-sectional study in six provinces in 2017 to assess the burden and distribution of health expenditure costs among tuberculosis-affected households and analyze related risk factors. The data was collected through face-to-face questionnaire survey. Intensity of nancial burden was the median positive overshoot beyond a threshold that de ned as annual total (direct and indirect) costs exceeding 20% of the pre-TB annual household income.

Results
Of 1147 patients, median (interquartile range) total costs due to tuberculosis care were US$965. 5(461.8, 2059.3). Costs due to pre-treatment care accounted for 46.3% and direct costs accounted for 82.5% for total costs. Median (interquartile range) intensity of catastrophic costs was 32%(11%, 88%). Living below poverty line, household size less than four, employment in informal sector, receiving treatment from tuberculosis dispensary and hospitalization were independent predictors of associated with higher nancial burden.

Conclusion
Despite free tuberculosis care services, the health expenditure due to tuberculosis were high among tuberculosis-affected households.

Trial registration
The Ethics Committee of Chinese Center of Disease Control and Prevention approved the study

Background
The End TB strategy targets that, by 2020, no TB-affected household should face catastrophic costs due to TB. 1 This is in line with policy to move health systems closer to universal health coverage (UHC). 2 The World Health Organization (WHO) issued TUBERCULOSIS PATIENT COST SURVEYS: A HAND BOOK 3 to guide patient cost survey in TB high burden counties. Following the manual , Viet Nam, Ghana and Myanmar carried out their national-wide survey and reported that six in ten households affected by TB or multidrug-resistant TB incurred catastrophic costs due to TB care. [4][5][6] Indonesia and Philippines reported that more than three in ten households incurred catastrophic costs. 6,7 China accounts for 8.6% of the global TB burden and 50% of the burden in Western Paci c region. 6 Despite free TB care services, and the availability of public-funded insurance systems, TB patients in China bear heavy nancial burden. 8 The information in China about the extent and determinants of catastrophic costs due to TB in TB-affected households was limited. [8][9][10][11] This information is vital to monitor the progress towards achieving End TB targets.
We report the intensity of nancial burden due to TB care (pre-treatment and treatment phases) in TB affected households and related factors in China in 2017.

Study design
This was a cross-sectional study involving primary data collection.

Health nancing in China
China is a developing country with a per capita gross national product of 7 941 US$ in 2016. The total expenditure on health in 2016 was six percent of the gross national product. 12 The health care delivery system is "mixed" with a dominant role for public sector institutions. 13 Public funded health insurance schemes (urban employee basic medical insurance (UEBMI), urban resident basic medical insurance (URBMI), and new rural cooperative medical scheme (NCMS) cover more than 95% of the population. 14 There is limited cost coverage for outpatient care. 15 National TB Programme (NTP) The prevalence of TB in the western region is 1.7 times and 3.2 times that of the middle and eastern region, respectively. 16 The National center for tuberculosis control and prevention (NCTB), which belongs to China center for disease control (CDC), manages the NTP. TB management units are established at provincial, prefecture and county levels (basic management units (BMU) at county level). One authorized TB designated medical facility (TB designated hospital, TB dispensary or CDC) is responsible for TB diagnosis and treatment at each BMU level.
TB patients are provided free chest radiography, sputum smear test and rst-line drugs in TB designated medical facilities. TB care services at referral hospitals and other general hospitals may be charged.

Patient population
The TB patient cost survey was conducted between March and June 2017, in line with the WHO recommended methodology. 3 Drug-susceptible pulmonary TB patients who had received at least two weeks of intensive phase therapy under NTP were included.

Sample size
Assuming 30% TB patients' total annual direct and indirect costs exceeding 20% of the household's annual pre-TB income , 17 relative precision as 0.2 and α error as 0.05, average cluster (de ned at county level) size of 50, between-cluster variation of 0.4, design effect of 4.36 and anticipating a non-response rate of 10%, the nal sample size was 1086, to be sampled from 22 clusters. (see Suppl Annex I, Figure I) Sampling methodology We adopted multi-stage strati ed cluster sampling. The stratifying factors were region and residence (see Suppl Annex II for the per capita GNP of the six provinces sampled). The steps followed in sampling have been summarized in Suppl Annex III.

Data collection
The face-to-face interview (at BMU in county) was done by trained investigators using a structured questionnaire (see Suppl Annex IV). Patients presented health insurance card and treatment fee documents if available. Baseline characteristics were collected at diagnosis. Costs related information was collected from symptom onset up to the day of interview.

Data management and analysis
Data were double entered and validated using EpiData (version 3.1 EpiData Association, Odense, Denmark) during July to December 2017. The analysis was conducted using STATA (version 12.1, copyright 1985-2011 StataCorp LP USA).
The calculated average monthly direct medical cost, direct non-medical costs and indirect costs during treatment were used to impute treatment costs of patients within the county for the remainder of treatment (six months for new and eight months for previously treated patients).
The analysis was weighted for multi-stage design and weighted results are presented. 18 The analysis (described below) was done separately for the pre-TB treatment phase (from symptom onset to treatment start), treatment phase (from treatment start to completion) and TB care overall (pre-TB treatment phase and treatment phase combined). Costs were described using the median and interquartile range (IQR).
Intensity of catastrophic costs was measured as the median positive overshoot beyond the 20% threshold (subtracting 20% from the total costs expressed as a proportion of annual pre-TB household income). 19 Operational de nitions used in this study have been summarized in Box. 3 Generalized linear model (Poisson regression) was built using forward stepwise method for factors associated with intensity of catastrophic costs. Age, gender and variables with unadjusted p value <0.20 were added. Decision to retain a variable in the model at each step was taken based on LR test (yes if p<0.05). In the nal model, adjusted prevalence ratios (0.95 CI) were used to summarize (infer) the association.

Patient pro le
Of 1147 patients, 811(70.7%) were male, their mean age was 51 years (range 12-89 years) and 364(31.7%) patients were interviewed during the intensive phase. New TB patients accounted for 91.6% of all the respondents and 414(36.1%) reported at least one episode of hospitalization. The median (IQR) monthly income per capita was US$190(46,243). The incomes of 223(19.4%) households were below the poverty line. The mean(standard deviation) family size was 3.5(1.9). The patient was the prime income earner in 684(59.7%) households. NCMS covered 864(75.3%) of the patients (Table I).

TB care costs
The median (IQR) direct, indirect and total costs due to TB care were US$812.1 (398.3, 1691.1), US$70.4(24.6, 296.2) and US$965.5(461.8, 2059.3) respectively (Table II). The direct costs accounted for 82.5% of the total cost, while the direct medical costs accounted for 64.6% of total costs (Table III). Of the total costs, 46.3% was incurred during the pre-treatment stage.
Family size less than four and living below the poverty line were associated with high catastrophic costs due to pre-treatment, treatment and TB care overall. Hospitalization as well as working in informal sector was associated with catastrophic costs due to treatment and TB care overall. Registration at a TB dispensary was an independent predictor for catastrophic costs due to pre-treatment care and TB care overall. Registration at CDC and age more than 65 years were independent predictors for catastrophic costs due to pre-treatment care only. Patients from middle region were less likely to incur catastrophic costs due to pre-treatment care when compared to east and west. (Table IV) Discussion This was a TB patient cost survey from China based on the WHO recommended methodology. 3 Data quality was ensured through double data entry and validation, standard data cleaning and management procedures at various levels. The data analysis was robust accounting for sampling weight and poststrati cation adjustment weight.
Our key ndings were that patients incurred high costs for both diagnosis and treatment despite the free TB care policy. Direct medical costs accounted for more than three-fths of the total costs. The intensity of catastrophic costs for TB-affected households during diagnosis and treatment were high. Risk factors for catastrophic costs were also identi ed.

Limitations
The data was collected through face-to-face questionnaire survey. Some patients may not accurately remember the exact costs incurred. We attempted to minimize recall limitation by surveying patients still on treatment and imputing costs to the entire episode assuming that all patients complete treatment.
This might overestimate the costs considering some patients may fail treatment or be lost to follow up. On the other hand, as we did not include multidrug resistant tuberculosis patients, our results could be an underestimate.
Most patients could not provide the breakdown of direct medical costs (>60% of total costs). Therefore, detailed information on components of direct medical costs is not presented. in 2012. 9 However, it is hard to directly compare our ndings with previous studies in China considering the different costs de nitions adopted.

Interpretation of key ndings
The most signi cant driver of costs was direct medical costs(65% of total costs) which was much higher than Viet Nam(44%) and Ghana(18.2%). 4,5 High direct medical costs pointed towards prescription of high-end investigations (besides sputum examination and radiography which are free) and unnecessary treatment (besides free TB drugs). These costs are paid out-of-pocket by the patients. The high direct medical costs also indicated that the TB service package and reimbursement rates of insurance schemes did not signi cantly reduce the nancial burden of TB patients. 8,11 Nearly half of the costs was spent before treatment initiation, which was higher than in Ghana(7%) and Indonesia(11%), 4,7 but was consistent with ndings from the systematic review by Tanimura et al 21 and Nigeria 22 . This suggests that TB patients incurred substantial costs before they reach TB designated medical facilities. The high pre-treatment costs may be due to poor TB awareness among patients as well as general hospitals (that are not authorized to diagnose and treat TB). This might had delayed TB care seeking and transfer out to TB designated medical facilities. 23 Many patients, especially migrants, were noti ed and probably managed for a signi cant period of time at a referral hospital before possible transfer out to TB designated medical facilities. 24 This might had contributed to the high direct costs.

High intensity of catastrophic costs
Our study calculated median intensity of catastrophic costs based on WHO's TB-speci c catastrophic cost de nition. Intensity of catastrophic costs based on different de nitions (catastrophic cost de ned as direct cost exceeding 10% of household income) has been reported elsewhere, such as China(40.8%),Nigeria(8.3%) and Benin(14.8%). 9,22,27 Our nding showed higher intensity of catastrophic costs than Nigeria and Benin, apart from the different de nition, there usually was some international or domestic special funding support for TB care in other TB high burden countries. 7 Risk factors for catastrophic costs Family size less than four and poverty were predictors for catastrophic costs(pre-treatment, treatment and overall) which was consistent with other studies in China and Indonesia. 7,9 Households with four or more members were less likely to incur catastrophic costs because in larger families, the total household income might be higher than smaller size families, thus reducing the impact of costs incurred towards TB care of one person on the household. In our study, the proportion of TB patients living below the poverty line was greater than in the general population (19.4% vs. 3.1%), 12 and was also a signi cant factor associated with catastrophic costs in all stages. Poverty suggests low capacity to pay, even lesser costs might be catastrophic.
About 36% of the TB cases reported at least one episode of hospitalization for TB care. Hospitalization was a signi cant predictor of catastrophic costs during treatment and care overall. The higher reimbursement rate of inpatients in insurance schemes might lead to the high rate of hospitalization, even as high as 55%, 9 thus leading to high costs and catastrophic costs.
When compared to TB designated hospitals, receiving treatment at CDC and TB dispensary were risk factors for catastrophic costs due to pre-treatment care. TB designated hospitals are speci c general hospitals that are authorized to manage TB. Majority of people reach CDC and TB dispensary after seeking diagnostic care in other general hospitals (not designated for TB management).
As found in other studies, economic situation is related to catastrophic costs. 8,9 Our study also found that not working in formal sector (unemployed or working in informal sector) was associated with catastrophic costs due to TB treatment. High indirect costs due to loss of wages are common in informal sector. 28

Policy implications
The high proportion of direct medical costs among total costs hints that moving towards universal health coverage is much likely to reduce the number of households incurring catastrophic costs in China. Expanding the bene t package for patients with TB and increasing reimbursement rate for outpatient care among insurance schemes may also help in reducing catastrophic costs during TB treatment. 15 Indirect medical costs and indirect costs also account for considerable amount of the total costs. Some provinces in China have implemented nutritious breakfast and travel allowance for TB patients, which can be expanded nationwide. Countries like India have implemented a TB-speci c cash transfer scheme, 29,30 China may consider the same, at least among those living below poverty line.
Reducing the high pre-treatment costs requires strengthening of TB health promotion to improve TB awareness, training the health staff of general hospitals to identify and refer presumptive TB patients to BMU and issue regulations to general hospitals to transfer TB patients to TB designated medical facilities at BMU.

Conclusion
Despite TB diagnosis and treatment being provided free of cost in China, TB patients still incur substantial costs, which hints that the current TB care policy and package are not su cient. High direct medical costs and high prevalence of catastrophic costs, both during diagnosis and treatment, show that UHC and social protection need to be reinforced urgently, if China is to meet the End TB targets of zero catastrophic costs due to TB care by 2020. 1

Declarations Ethical Approval and Consent to participate
The she Ethics Committee of Chinese Center for Disease Control and Prevention approved the study.

Consent for publication
Written informed consent for publication was obtained from all participants.

Availability of supporting data
The dataset and codebook used in this study are available on request from the corresponding author (zhanghui@chinacdc.cn, huizhang1974@126.com)

Competing interests
No competing interests was reported by the authors.

Funding
The author(s) received no speci c funding for this work. The survey was funded by WHO, the training programme were funded by the Department for International Development (DFID), UK. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.