Assessing Fidelity to HIV Index Case Testing Protocols in Malawian Clinical Settings: Results From a Cluster Randomized Controlled Trial.

INTRODUCTION

HIV index case testing is an evidence-based practice in which health care workers (HCWs) counsel people living with HIV (index clients) to offer HIV testing services to their sexual partners and biological children (contacts). For index case testing to be effective, it is important that index- and contact-client counselling protocols are implemented as intended (fidelity). We evaluated whether a digital training implementation strategy improved HCWs' fidelity to these counselling protocols.

METHODS

We conducted a cluster-randomized controlled trial at health facilities (clusters) in Southeastern Malawi. Primary study follow-up was from October 2022 to September 2023. Thirty-three clusters were randomized 1:2 to the Enhanced (n = 11) and Standard (n = 22) arms. HCWs in both arms received brief centralized in-person training on processes and protocols for index case testing. HCWs in the Enhanced arm also received decentralized tablet-guided digital training through 20 synchronous and asynchronous sessions. In all 33 clusters, a selection of index and contact client encounters were audio-recorded in the year after the digital training. We applied a 15-item fidelity checklist to assess counselling quality and adherence to index case testing protocols. Audio recordings were coded independently by two assessors blinded to the study arm. Fidelity scores ranged from 0% to 100%; higher scores indicate better protocol compliance. To compare fidelity scores between the Enhanced and Standard arms, we estimated a mean difference and 95% confidence interval (CI), using generalized estimating equations to account for clustered data.

RESULTS

We audio-recorded 341 index client-HCW and 194 contact client-HCW encounters. The mean fidelity score for index client-HCW encounters was 66.3% (CI: 62.8%-69.7%) in the Enhanced arm and 45.7% (CI: 42.7%-48.6%) in the Standard arm (Mean Difference 20.6%; CI: 16.1%, 25.2%, p<0.001). The mean fidelity score for contact client-HCW encounters was 69.0% (CI: 63.8%-74.1%) in the Enhanced arm and 57.4% (CI: 55.5%-59.3%) in the Standard arm (Mean Difference 11.5%; CI: 6.1%, 17.1%, p<0.001).

CONCLUSIONS

The digital training substantially improved HCW fidelity to index and contact client protocols. A decentralized digital training can be used to improve HCW counselling behaviour. Future research should explore the methods of assessing fidelity in trials of complex digital training interventions in other settings.