A screening strategy for latent tuberculosis in healthcare workers: Cost-effectiveness and budget impact of universal versus targeted screening.

OBJECTIVE

To evaluate the clinical, cost-efficiency, and budgetary implications of universal versus targeted latent tuberculosis infection (LTBI) screening strategies among healthcare workers (HCWs) in an intermediate tuberculosis (TB)-burden country.

DESIGN

Pragmatic cost-effectiveness and budget impact analysis using decision-analytic modeling.

SETTING

A tertiary-care hospital in Singapore.

METHODS

We compared 7 potentially implementable LTBI screening programs including universal and targeted strategies with different screening frequencies. Feasible targeting methods included stratification by country of origin (a proxy for risk of prior TB exposure) and by high-risk occupation. The clinical and financial consequences of each strategy were estimated relative to "no screening" (current practice) and compared to locally appropriate cost-effectiveness thresholds. All analyses were conducted from the hospital's perspective over a 3-year time horizon, based on the typical hospital planning period. Parameter uncertainties were accounted for using sensitivity analyses.

RESULTS

In our model, relative to current practice, screening new international hires and triennial screening of existing high-risk workers is most cost-effective (US$58 per quality adjusted life year [QALY]) and decreases active TB cases from 19 to 14. Screening all new hires combined with triennial universal screening, with or without annual high-risk screening or annual universal screening, reduced active TB to a range of 19 to 6 cases, but these strategies are less cost-effective and require substantially higher expenditures.

CONCLUSIONS

Targeted LTBI screening for HCWs can be highly cost-effective for hospitals in settings similar to Singapore. More inclusive screening strategies (including regular universal screening) can yield better outcomes but are less efficient and may even be unaffordable.