Outcomes

The outcome was VMMC AE yield relative to the expected (maximum) rate of VMMC AEs in this population. The expected maximum rate of VMMC AEs in this population was assumed to be 2%, the commonly accepted safety standard [20–22]. Consequently, an AE yield less than 2% would be considered an underestimate of AEs while an AE yield of over 2% would be considered an overestimate of AEs. The percentage change in AE yield was calculated relative to 2% (i.e., percentage yield in SoC and 2wT divided by 2%) and expressed as a percentage. Data for effectiveness were obtained from the RCT, with reported AE rates as the base case estimates, and 95% confidence intervals were used in sensitivity analysis [25] Parameters for the outcome estimation are shown in Table 1.

*2wT RCT clinical outcome results published previously [26].

**Assumed no uncertainty.

***Cost of VMMC minus costs of circumcision kits [26].

We used a micro-costing approach for cost estimation. The parameter estimates, ranges used for sensitivity analysis, and data sources are presented in Table 1. We divided costs into the following categories: text messaging (2wT arm only), follow-up phone calls (as follow-up to 2wT), in-clinic post-VMMC follow-up, outreach post-VMMC follow-up (for missed day 2 appointments in SoC, including phone calls), and management of AEs.

The cost of text messaging for 2wT was calculated by summing the program costs of the text message service and the personnel cost of sending and responding to text messages. The cost of the text message service was calculated by multiplying the unit cost of a single text message by the mean number of texts sent per client in the trial. Because text messages are sold in bundles of 250, the cost of a single text message was estimated as a fraction of the bundle cost. The cost of 2wT-specific personnel was calculated using research logs to estimate the full time equivalents (FTE) spent by the nurse responsible for managing the 2wT system (including sending and responding to texts), and multiplying the FTE estimate by wage. Reports from research staff showed that the FTE required by the nurse to manage the texting was estimated at five to 15 minutes per hour, assuming an 8-hour work day and five working days a week, over the nine-month time period of the trial. The cost of follow-up phone calls made to 2wT clients was calculated by summing the costs of the phone service and the personnel costs of making calls. The cost of the phone service was calculated by multiplying the unit cost of a call by the mean number of calls per client in the trial as shown in call logs.

The cost of in-clinic follow-up was calculated by summing the personnel cost of follow-up and the costs of clinical supplies provided by the MoHCC VMMC program. The personnel cost of in-clinic follow-up was estimated by multiplying the time spent per client (by one clinic clerk and one VMMC nurse) by their wages. The use of one clinical clerk and one VMMC nurse reflects the pattern of routine post-VMMC follow-up in the MoHCC VMMC program. The cost of supplies included the components of a MoHCC wound dressing bundle of gloves, antiseptic, alcohol rub, gauze, bed liner, and distilled water.

The cost of follow-up for missed day two visits in SoC was calculated by summing the personnel costs of outreach, transportation costs, costs of phone calls (made during attempts to reach client’s homes), and costs of clinical supplies obtained from both RCT records and VMMC program logs. The costs of personnel were calculated based on the common practice of using one driver and one nurse for home visits and multiplying their wages by the time spent in this outreach activity. The nurse and driver also received a lunch allowance per outreach which was also included. The costs of phone calls during outreach assumed an average of one phone call per client home visit. The cost of clinical supplies included the components of a wound dressing bundle as described above. The costs of outreach transportation were calculated by multiplying the cost of fuel per unit distance by the mean distance of follow-up, both obtained from program data.

The costs of AE management were calculated by multiplying the probability of occurrence of the AE type from the trial (bleeding, infection, and swelling) [25] by the costs of managing the different AE types (minor surgery, antibiotics, and analgesics/antipyretics). The cost of minor surgery was assumed to equal to the cost of dorsal slit VMMC minus the cost of circumcision kits. All costs were estimated in 2018 US dollars ($US).