Non-medical financial burden in tuberculosis care: a cross-sectional survey in rural China

We designed and conducted a cross-sectional survey in TB patients. In China, the administrative
demarcations move downward from country to provinces to prefectures/cities to districts/counties
and to towns. The study was undertaken in Zhenjiang City, Jiangsu Province in eastern
China; Yichang City, Hubei Province in central China; and Hanzhong City, Shaanxi Province
in western China. Sample size calculations indicated that a minimum of 792 TB cases
(264 in each city) were necessary as the assumed sample proportion of catastrophic
expenditure on non-medical costs were 20 %, with 5 % as half the width of the confidence
interval, and ??=?0.05. In each city one county/district was randomly selected by
a random number from each category of those with high, middle and low GDP per capita.
Three townships/streets were then selected at random in each selected county/district
and thirty TB cases were randomly selected from each township/street using a list
of registered cases. Patients who completed normal treatment or stopped treatment
during 2012 were included in the study. We excluded patients with communication barriers,
such as those with hearing impairments. We also excluded patients with serious diseases
and migrant workers who did not join the survey within the study period. Patients
with mental illnesses were excluded as well. In total, 797 TB patients were recruited
and informed consent of each participant was obtained.

Interviews were conducted by trained enumerators using a structured questionnaire
to collect the medical and non-medical costs (transportation, accommodation, and nutritional
supplementation) of TB treatment. Information regarding personal demographic and socio-economic
status (age, sex, education, family income/expenditure, etc.; Table 1), reimbursements from health insurance, and financial assistance from government
agencies was also collected. The field survey was conducted between April 2013 and
May 2013.

Table 1. Basic characteristics of TB patients according to study site
^a

Only patients with a ‘confirmed’ TB diagnoses were included. Most had at least one
sputum smear test and one chest x-ray. Indirect expenditures on transport and accommodation
incurred by patients, their families and others related to seeking and accessing TB
treatment during pre-diagnostic, diagnostic and post-diagnostic periods, as well as
during hospitalization where applicable, were recalled by patients and their caregivers,.
The cost for nutritional supplementation during TB treatment was estimated by extracting
the cost of extra food expenditure (such as meat, milk, vitamins, etc.). We attempted
to minimize the recall bias via in-depth interviews with the patient.

Ethical approval was sought and granted for this research by the Ethical Committee
of China CDC. It was recognized that the right and the welfare of the subject were
adequately protected; the potential risks were outweighed by the potential benefits.
The ethical approval number was 201307.

Statistical analysis

We quantified non-medical costs by aggregating the transport, accommodation, and nutritional
supplementation expenditures related to TB health care. Overall, 752 patients reported
complete non-medical costs for transport, accommodation, and nutritional supplementation,
while others missed some portion of the above. Cases with missing data were deleted
when analyzing the corresponding costs. Mean and median non-medical expenditures were
calculated and compared across subgroups using Mann–Whitney U and Kruskal-Wallis tests
and a 5 % significance level. Linear regression was then used to model the relationships
between non-medical costs and the explanatory variables available from the survey
data. We also separated the transport plus accommodation costs and the additional
nutrition cost for the multi-variate analysis. All the statistical analysis was done
using the SAS version 9.3 statistical software package (SAS Institute Inc., Cary,
North Carolina).

We considered the following patient variables to be potentially correlated with non-medical
costs as they were major risk factors for the non-medical cost: residence location
(the three study cities), gender, age (65 years or??=65 years), residence type (urban
or rural), inpatient care (with or without), health insurance (covered or uncovered),
education level (never attended school, primary school, junior high school, high school
or higher), family income (as a proportion of the median in each city), and patient
category (new or relapse patient).