Innovative timing strategies for tuberculosis household contact investigation: cost-effectiveness analysis from a randomized trial in rural and urban South Africa (Kharituwe Study)

Abstract

Background: Household contact investigation (HCI) for tuberculosis (TB) is recommended but often limited by resource constraints, particularly for individuals unavailable during business hours.

Methods: We conducted an economic evaluation from January 1, 2022, through December 31, 2022, nested within a randomized trial in South Africa ("Kharituwe") comparing standard HCI for TB and two novel strategies: HCI during holiday periods in a rural setting and off-peak HCI during weekends and evenings in an urban setting. Costs were derived from 2022 expenditures, and secondary TB cases were defined by positive sputum cultures. As a secondary outcome of the Kharituwe Study, we assessed the incremental cost-effectiveness ratio (ICER) of each strategy against a hypothetical no-HCI scenario from the health system perspective in 2022 US dollars. Cost-effectiveness was assessed using a country-specific willingness-to-pay threshold of US$3015 per disability-adjusted life year (DALY) averted. The trial is registered with clincaltrials.gov (NCT04520113).

Findings: Relative to a hypothetical no-HCI approach, standard HCI was estimated to cost US$1400 [95% uncertainty interval (UI): $1000-$2100] per DALY averted in the urban setting and US$3600 [95% UI: $2500-$5400] in the rural setting. Corresponding cost-effectiveness ratios were US$1900 [95% UI: $1300-$2800] for off-peak (urban) and US$6400 [$3900-$10,000] for holiday-based (rural) HCI. Personnel costs, travel costs (in the rural setting), and TB prevalence among contact persons were primary drivers of cost-effectiveness.

Interpretation: HCI for TB is likely cost-effective in urban South Africa and may be cost-effective in rural settings, which face barriers including long travel times and lower TB prevalence. Holiday-based HCI was not found to be cost-effective. Integrating HCI for TB into broader home-based interventions may improve cost-effectiveness.

Funding: Funding was provided by the United States National Institute of Allergy and Infectious Diseases (Grant # 5R01AI147681).

Keywords: Economic evaluation; Household contact investigation; South Africa; Tuberculosis.

Fig. 1

One-way sensitivity analysis: cost-effectiveness of contact investigation for tuberculosis in rural and urban South Africa. Panels A–D present results of our one-way sensitivity analyses evaluating four strategies for TB household contact investigation, each relative to a hypothetical no contact investigation approach. Parameters are ordered according to their influence on the estimated cost-effectiveness estimate along the y-axis, while the x-axis indicates the estimated cost per disability-adjusted life year averted. The length of each bar indicates the difference between the primary estimate of cost-effectiveness (shown as a vertical black line and labeled to the right of the line) and the corresponding estimate when the parameter is at the upper (blue) or lower (red) bound of its range of variation.

Fig. 2

Incremental cost-effectiveness scatter plots of household contact investigation strategies for tuberculosis relative to a hypothetical no household contact investigation approach. These cost-effectiveness scatterplots were generated by running 1000 simulations in a probabilistic sensitivity analysis. Each point on the scatter plot represents an iteration of the simulation, with the corresponding incremental cost-effectiveness ratio (ICER) estimate comparing the strategy of interest to a hypothetical no contact investigation approach. The base-case ICER estimate for each strategy compared to a hypothetical no contact investigation approach is also included as a dark circle toward the center of each ellipse. The estimated country-specific cost-effectiveness threshold for South Africa (US$3015 per disability-adjusted life year averted) is shown as a diagonal line. Points falling below the cost-effectiveness threshold are considered cost-effective. Costs are reported in 2022 US dollars.

Fig. 3

Cost-effectiveness acceptability curves. These cost-effectiveness acceptability curves present the probability of each of four contact investigation strategies being cost-effective (y-axis) across a spectrum of different cost-effectiveness thresholds (shown on the x-axis), compared to a hypothetical no contact investigation approach. Results are based on probabilistic sensitivity analysis as described in the manuscript text. All costs are given in 2022 US dollars.