© 2015 Shah et al. Background: We designed a pragmatic stepped-wedge cluster randomized controlled trial in order to evaluate provider-initiated evaluation of household contacts (HCs) of smear positive tuberculosis (TB) cases within a routine TB program in Lima, Peru. Methods/Design: National TB program (NTP) officers of San Juan de Lurigancho District (Lima, Peru) and university-based researchers jointly designed a pragmatic stepped-wedge cluster randomized trial design in order to evaluate a planned active case finding (ACF) program for all HCs of smear-positive TB cases in 34 district healthcare centres. Randomization of time to intervention initiation was stratified by health centre TB case rate. The ACF intervention included provider-initiated home visits of all new sputum smear positive TB patients in order to evaluate household contacts for active TB. Active TB was diagnosed using symptom screening, sputum screening, chest x-ray and clinical evaluation. Once initiated, ACF was provided by NTP staff and integrated into the routine DOTS TB program activities. Discussion: This study protocol describes the pragmatic stepped-wedge cluster randomized trial of active household contact evaluations within an NTP. The stepped-wedge design met overlapping needs of local TB programmers and researchers to adequately evaluate the large-scale roll out of a new control program in a TB endemic setting. Multiple planning meetings were required to develop the necessary networks and in order to understand the operations, needs and goals of the NTP staff and researchers collaborating on this project. The advantages and challenges of using this study design in practice and within existing routine TB programs in a middle-income country context are discussed. Trial registration: ClinicalTrials.gov NCT02174380. Registered 24 Jun 2014

We compared the cost-effectiveness (CE) of an active case-finding (ACF) programme for household contacts of tuberculosis (TB) cases enrolled in first-line treatment to routine passive case-finding (PCF) within an established national TB programme in Peru. Decision analysis was used to model detection of TB in household contacts through: (1) self-report of symptomatic cases for evaluation (PCF), (2) a provider-initiated ACF programme, (3) addition of an Xpert MTB/RIF diagnostic test for a single sputum sample from household contacts, and (4) all strategies combined. CE was calculated as the incremental cost-effectiveness ratio (ICER) in terms of US dollars per disability-adjusted life years (DALYs) averted. Compared to PCF alone, ACF for household contacts resulted in an ICER of $2155 per DALY averted. The addition of the Xpert MTB/RIF diagnostic test resulted in an ICER of $3275 per DALY averted within a PCF programme and $3399 per DALY averted when an ACF programme was included. Provider-initiated ACF of household contacts in an urban setting of Lima, Peru can be highly cost-effective, even including costs to seek out contacts and perform an Xpert/MTB RIF test. ACF including Xpert MTB/RIF was not cost-effective if TB cases detected had high rates of default from treatment or poor outcomes.

Setting

Peru reports among the highest multidrug-resistant tuberculosis (MDR-TB) rates in the Americas, with a growing proportion in previously untreated tuberculosis (TB) cases. The identification of clusters of primary MDR-TB compared with drug-susceptible TB (DS-TB) could help prioritize interventions.

Objective

To examine the clustering of primary MDR-TB case residences and their proximity to high-risk locations in San Juan de Lurigancho District, Lima, Peru.

Design

Enrolled primary MDR-TB and primary DS-TB cases were interviewed and their primary residence was recorded using handheld Global Positioning System devices. Kuldorff's spatial scan statistic was used for cluster detection (SaTScan(TM), v. 9.1.1). Identified clusters were visualized in Quantum Geographic Information Systems software (v1.8.0). The following cluster centers were tested: a health centre with the highest TB and MDR-TB rates (Clinic X), a hospital and two prisons. Using regression analyses, we examined predictors of primary MDR-TB cases.

Results

A statistically significant cluster of primary MDR-TB cases was identified within a 2.29 km radius around Clinic X. Proximity to Clinic X remained a significant predictor of primary MDR-TB in adjusted regression analyses.

Conclusion

We identified a hotspot of primary MDR-TB cases around Clinic X in a TB-endemic area. Causes of this clustering require investigation; targeted interventions for this high-risk area should be considered.