Designing family-centered male circumcision services: A conjoint analysis approach

Background: Male circumcision (MC) has become an important weapon in the fight against HIV/AIDS in many Sub-Saharan African countries. The successful implementation of a national MC program requires the design of circumcision services that are attractive to young men of various ages. For many potential clients, mothers and/or fathers will play an important role in the decision to be circumcised, and hence services will need to be designed with the preferences of mothers, fathers, and sons in mind. Objective: Our objective was to value multiple design characteristics of potential community-based MC services from the perspectives of mothers, fathers, and sons in Johannesburg, South Africa, and to test for concordance between their values for the design characteristics. Methods: Potential design characteristics ofMCservices were identified through open-ended interviews with key informants (n= 25). Preferences were estimated using conjoint analysis implemented as part of a cluster randomized household survey. Each participant was randomized to receive one of two possible blocks of conjoint analysis, each consisting of six forced-choice tasks comparing two possible MC services varying on 11 design characteristics. With only two levels for each attribute, our experimental design utilized a main effects orthogonal array. Data were analyzed using linear probability models, with tests of concordance of values using Wald tests generated from stratified estimates calculated using restricted least square estimation. Results: A racially and geographically diverse sample consisting of 204 fathers, 204 mothers, and 237 sons completed the survey. In aggregate, requiring a follow-up visit was the most valued design factor (p < 0.001), followed by having a lower infection rate (p < 0.001), having less pain (p = 0.001), and a private waiting room (p = 0.001). Based on stratified analysis, sons also valued having the risks and benefits of MC explained (p = 0.01) and mothers valued requiring an HIV test as part of the procedure. Requiring an HIV test was the most significant difference between the respondents (p = 0.03), with sons finding it somewhat repulsive (p = 0.30). Conclusion: Our findings suggest that valuation of aspects ofMCclinic design can diverge by decision maker. To better ensure utilization of services, these variations should be taken into account to prior to implementation of a national strategy in South Africa. Adis