Adaptation of WHO’s generic tuberculosis patient cost instrument for a longitudinal study in Africa

Figures & data

Table 1. Characteristics and limitation of the WHO TB patient cost survey and the adapted version for longitudinal studies in Africa

	Generic WHO TB patient cost survey	Adapted patient cost survey – for TB Sequel
Design	Cross-sectional facility-based	Longitudinal facility-based
Geographic location	National survey	Five African study clinics located in The Gambia, Mozambique, South Africa and Tanzania.
Population	All patients (adults and children) registered for TB (drug-susceptible or drug-resistant) treatment in the National Treatment Programme (NTP).	Adult patients (≥18 years) registered for TB (drug-susceptible or drug-resistant) treatment in the National Treatment Programme (NTP) in the study sites
Sampling	Random cluster sampling or a national simple random sampling where electronic registers are available.	Consecutive sampling – as eligible participants present at the facility [10]
	Sample size based on assumed prevalence of catastrophic costs.	Sample size based on prevalence (>20%) of the outcome of interest which is severe pulmonary function impairment, measured by spirometry at least 2 years after treatment initiation
Enrolment	Consecutive patients registered for TB treatment who are attending a sampled facility and are a minimum of 14 days into either the intensive or continuation phase.	Starting (+/- 14 days) TB treatment after TB diagnosis.
Interview schedule	Each patient is interviewed only once	Patients are interviewed multiple times at defined study visits (M0, M2, M6, M12 and M24).
Resources and time requirements	The survey team is typically comprised of a principal investigator, a survey coordinator, a data manager and interviewers – these are temporary or short-term positions.	The team composition includes data capturers, quality control manager, and a statistician. Permanent or long-term positions are needed for continuity and retention of experienced staff.
	Once off interview and consent (±45 minutes), either paper-based or electronic survey.	Multiple interviews (±45 minutes), only paper-based survey with data entry into an electronic database.
	Qualitative interviews can be performed both with patients and household members, as well as health and social service staff and policy-makers [2].	No qualitative component but includes additional component to study social consequences (e.g. quality of life, pain, disability and stigma related to TB) [10].
Interviewers	Facility-based	Facility- and community-based
Calculating total costs	Requires some retrospective data collection, forward projections and imputations to calculate the total costs.	Total costs available for each phase of TB treatment (i.e., pre-treatment, intensive phase of treatment, continuation phase of treatment and post-TB treatment).
	Only patients in the intensive phase receive questions on the costs incurred prior to diagnosis.	All patients receive questions on the costs incurred prior to diagnosis at the enrolment/M0 visit.
Outcomes	Treatment outcomes are typically not available	Treatment outcomes are available
	Estimate the cost per patient treated for TB	Can estimate the cost per patient treated for TB by outcome Repeated measures of health-related quality of life, pain, depression or anxiety and disability related to TB is collected at M2, M6, M12 and M24. May contribute to better estimates of disability-adjusted life years (DALYs).
Limitations	Survey findings can only be generalized to a subset of people with TB who receive care in the NTP. Conclusions cannot be drawn about all people with TB in the country.	Survey findings can only be generalized to a subset of people with TB who receive care in the NTP in the study sites. Conclusions cannot be drawn about all people with TB in the country.
		The study site may not necessarily be where the patient goes for routine TB care or to collect medication. This may result in issues with recall as participants try to distinguish between routine treatment versus study related visits.
		Direct costs, income loss or coping costs may be underestimated as study sites reimburse patients for travel and time at each visit or may provide support (e.g. food parcels, supplements, travel vouchers) to promote retention.
	Recall bias – a major challenge for the estimation of total patient costs incurred. This mainly affects cost estimates for the pre-treatment period.	To minimize recall bias, patients receive questions on the costs incurred prior to diagnosis at the first visit (M0). During subsequent visits, patients are asked to recall since the date of their last interview. Interviewers are also trained to make it easier for the patient to recall by using local methods of time structuring.
	Costs after treatment completion are not included.	Costs after treatment completion are included. Patients are followed for a minimum of 24 months after TB treatment initiation.
	The current methodology only includes TB patients enrolled and retained in care.	A longitudinal design can only consider earlier aspects of the cascade of care, prior to when a patient becomes lost to follow up.

Figure 1. The WHO generic TB patient cost survey instrument was adapted for use at use at enrollment/start of TB treatment (Day 0) and at Month 2, Month 6, Month 12 and Month 24; time points intended to capture costs incurred for different phases of TB disease (e.g. pre-treatment, on-treatment and post-treatment)

Table 2. Approaches to estimating annual household income using the TB patient cost instrument adapted for longitudinal studies*

Catastrophic total costs^& = Episode direct costs^a ± Episode indirect cost^b > threshold value (e.g. 20% base case) Household capacity to pay^c,d
Direct medical and non-medical costs^a	Total indirect (productivity) loss^b
These are payments made directly by the patient or their household member. Direct medical payments include payments for formal medical professionals, informal traditional or alternative practitioners, clinics, health centers, pharmacies, and hospitals – for medical services and products (e.g. medicines, consultation fee, payment to DOTS provider/supporter, day charges for hospitalization, diagnostics, lab tests and procedures). This excludes prepayment for health services – for example health insurance premiums – and where relevant, net of any reimbursements to the individual who made the payments. Non-medical direct costs include travel, accommodation, food or other non-medical payments incurred by the patient, their household member or caregiver/guardian while picking up medication or during the visit/hospital stay for TB care (e.g. nutritional or food supplements, interest on loans taken out to meet the costs of TB, day charges for time in hospital etc.).	Indirect costs can be estimated using two alternative methods; Self-reported household income loss net of welfare or social assistance payments. This includes the cost that the caregivers bear by contributing their time and in-kind services. Any gain or loss of income reported during each phase or for the entire episode is considered the total indirect cost [18]. Opportunity cost of time spent away from the daily productive routine. Participants are asked to self-report time spent seeking and receiving care as well as the average number of hours they work each day, if this changed since the last interview, and if so, by how much. The time total period of absence (in hours) is multiplied by the hourly wage rate of the absent worker. The hourly wage can be estimated from directly reported data, household asset ownership, or national statistics.
The sum of the direct medical and non-medical costs for the different phases/entire episode is used as the numerator.	Indirect costs for the different phase/entire episode is used as the numerator in the catastrophic total cost equation.
Household capacity to pay^c,d
Self-reported current income (detail or prompted ranges)^c	Estimated income based on asset scoring^d
Participants are asked to self-report their monthly individual and household income, which is then used to calculate the annual income. If participants don’t know or refuse to answer they are asked to identify their individual and household income from a list of prompted ranges. Detailed questions are used to solicit information about income (e.g. employment, property income, income from household production of services or goods) and non-salary income (e.g. travel voucher, food vouchers, disability grant, in-kind or cash transfers, public assistance, donations etc.) Annual household income before the onset of symptoms is typically used for the denominator.	Annual household income can be estimated using asset ownership or dwelling characteristics (e.g. number of rooms, type of toilet facility, electricity supply, source of drinking water). A standard asset index can be used to divide households into five socio-economic quintiles. For each income quintile, mean household permanent income can be extrapolated from the National Income Dynamics Survey. This estimate usually represents income before TB. Estimating income based on assets is a suitable alternative if income is hard to report accurately or is subject to great variability over time (e.g. casual or seasonal work, large informal sector work etc.).
Alternative indicator of catastrophic costs – Coping strategies
Coping strategies can be used as an indicator of economic catastrophe. Participants are asked if the individual or their household used savings, borrowed money, sold property or took out a loan to cover the cover of TB, and if so, how much. Participants are also asked about the economic consequences experienced (e.g. pulling children out of school, household food security etc.)

*Adapted from [2] and [18].

Abbreviations: TB tuberculosis; DOTS Directly Observed Treatment Short-course & Participants are interviewed at time points intended to capture costs incurred for different phases of TB disease (e.g. pre-treatment, on-treatment and post-treatment). Catastrophic total costs can be calculated for the entire TB episode or for each phase of TB.

a = Direct medical and non-medical costs; b = Total indirect (productivity) loss; c = Self-reported current income (detail or prompted ranges); d = Estimated income based on asset scoring.

Sweeney S, Mukora R, Candfield S, et al. Measuring income for catastrophic cost estimates: limitations and policy implications of current approaches. Soc Sci Med. 2018;215:7–15.

 PubMed Web of Science ®Google Scholar

Tuberculosis patient cost surveys: a handbook. WHO/HTM/TB/2017.24. Geneva: World Health Organization; 2017 [cited 2020 Nov 26]. Available from: http://who.int/tb/publications/patient_cost_surveys/en/

 Google Scholar