The Mobile WAChX trial (NCT02400671) evaluates one-way versus two-way communication versions of a semiautomated SMS text messaging intervention to improve ART adherence and retention in care in peripartum women at 6 facilities in Nairobi and Western Kenya. One-way SMS text messaging consisted of weekly automated motivational and educational SMS text messages and clinic visit reminders. Participants randomized to the two-way SMS text messaging arm additionally had the capability of communicating with a nurse through the SMS text messaging system. Participants randomized to the control arm received no SMS text messages (standard of care).

The study procedures for the randomized controlled trial are described in detail in a previous paper [15]. For costing analysis, convenience sampling was used to collect the cost data from 2 of the 6 sites: an urban health center in Nairobi (facility A) and a rural subcounty hospital in Western Kenya (facility B). We used a provider perspective to analyze the incremental financial and economic costs of implementing Mobile WAChX alongside existing maternal and child health (MCH) services from January 2017 to December 2017, as well as the average incremental cost per beneficiary and average incremental cost per contact. Beneficiaries are service users who accessed Mobile WAChX SMS text messaging communication, including automated messages and personalized messaging with nurses, through the Mobile WAChX system. The number of contacts is equal to the number of total messages sent successfully to the beneficiaries. Data on the number of beneficiaries and contacts at each facility were extracted from ongoing project monitoring and evaluation reports and electronic databases of SMS text messages.

We used an activity-based ingredients approach to identify all activities undertaken to deliver the Mobile WAChX project. Activities included intervention planning, project partner sensitization, staff training, system development, and delivery of SMS text messages. After identifying all intervention-related activities, we quantified the resources used and valued these by using the best available data on salaries and commodity prices. We used a combination of data collection methods to collect primary resource-use and cost data, including obtaining prices from project expense reports, administrative records, and published government salaries, and conducting semistructured interviews with facility-based health workers and project administrators. Time-motion studies were conducted in both facilities to record staff time spent on intervention activities (eg, recruiting, screening, and registering participants, sending SMS text messages to users). Data collection was conducted between October 2017 and January 2018.

We organized cost data into one-time fixed costs and variable costs (Table 1). Fixed costs were categorized according to the following activities: intervention planning, preparation of intervention sites, development of the Mobile WAChX SMS text messaging management system, initial training, and sensitization meetings with facility staff and Kenyan MOH officials. All fixed costs were used for one-time start-up activities, where we assumed a 5-year useful life. Variable costs included recurrent costs, which were required to sustain the intervention. They were divided into mutually exclusive input categories for personnel, communication, equipment, and overhead. Personnel costs included salaries, benefits, and allowances of facility-based health workers as well as of staff in charge of personnel supervision and coordination. A study nurse and a retention officer delivered Mobile WAChX-related activities. Shared facility costs included a study coordinator and a data manager who were responsible for supervising all the 6 sites and some shared communications costs such as the costs of developing a Mobile WAChX system and cost of the system-hosting platform. We allocated these shared facility costs based on the annual share of clients served at the facility, as a percentage of the total annual number of clients served for all 6 facilities. In addition to the cost of a system-hosting platform, other communication costs included cost of internet data bundles for using the internet-based Mobile WAChX system, airtime cost for making patient follow-up phone calls, costs of sending SMS text messages to users, and short code toll-free numbers used by participants to deliver messages to the nurse. Equipment costs (mobile phones, laptops, and furniture) were annuitized over the useful life of 10 years by using a discount rate of 3%. All costs in this evaluation were expressed in USD, using the official exchange rate of 1 USD to 103.25 Kenyan Shillings (2017 exchange rate). In addition, research time and other research costs were removed from the costing analysis. International staff time was also excluded to better reflect the costs of the program when implemented and scaled locally.

We developed an Excel-based model (Microsoft Excel version 15.28, Redmond) to estimate total incremental costs and incremental unit costs. The sum of all the activity cost categories reflects all the resources required to deliver the Mobile WAChX intervention. All activities were mutually exclusive, thereby avoiding double counting. We used project output data on the number of beneficiaries per month per facility and number of messages sent by the system and nurses per month. These data were collected as part of the Mobile WAChX monitoring and evaluation strategy. We first estimated the total incremental cost for each facility to deliver the Mobile WAChX intervention and divided this by the number of women receiving the intervention to determine the cost per beneficiary. We also estimated the cost per contact, defined as the total cost divided by the total number of messages sent to participants during 2017, for each facility. We analyzed the costs of one-way and two-way SMS text messaging interventions separately. We also estimated the average weighted costs for both facilities by using project output data on the number of beneficiaries per facility.

In addition to estimating intervention costs, we estimated 3 scenarios. The first scenario was to estimate the cost when the two-way Mobile WAChX intervention was implemented in all 6 facilities in this project. In this scenario, the same intervention and personnel were applied to every facility where they shared most start-up costs from system development, cost of system-hosting platform, as well as personnel supervision and coordination costs. The number of participants was the total number of participants receiving intervention in all 6 sites. The second scenario was to estimate a more typical scenario where the MOH of Kenya supports these activities after the pilot phase. We applied the MOH salary scale to service delivery health workers in the current project. Finally, we calculated the incremental cost effectiveness ratios (ICERs) of comparing two-way SMS text messages to no intervention in the 2 facilities. Due to unavailable efficacy data in the Mobile WAChX trial, we used clinical outcome results from a similar randomized controlled trial, which assessed whether two-way SMS text messaging interventions improved plasma HIV-1 viral RNA load suppression at 12 months in 3 clinics in Kenya [7]. The efficacy of mobile SMS text messaging intervention for medication adherence was obtained from a meta-analysis study [16]. Other inputs were using project output data and our analyses.

Among the 152 HIV-infected women in facility A who received the Mobile WAChX intervention, 76 (50.0%) were randomized to the two-way SMS text messaging group and the other 76 (50.0%) were randomized to the one-way SMS text messaging group. A total of 80 women participated in facility B, of whom 39 were randomized to the two-way SMS text messaging intervention group and 41 were randomized to the one-way SMS text messaging intervention group (Table 2). The beneficiaries in these 2 facilities accounted for 27.7% (152/548) and 14.6% (80/548) of the total number of project beneficiaries at the time of data collection in facilities A and B, respectively. The total number of automated SMS text messages from the system sent to beneficiaries in the 2 facilities in 2017 was 6924 in the two-way SMS text messaging intervention group and 7318 in the one-way SMS text messaging intervention group. The nurses in the 2 facilities sent 1386 personalized messages in response to two-way participant messages during 2017.

Table 3 summarizes the weighted average total annual incremental costs and incremental unit costs for beneficiaries in 2017. The weighted average total cost of the Mobile WAChX intervention was estimated at US $3725 for the two-way SMS text messaging group and US $2542 for the one-way SMS text messaging group. Fixed costs were US $2936 for two-way SMS text messaging and US $1757 for one-way SMS text messaging intervention groups (Figure 1), while variable costs were similar across the 2 groups (US $789 and US $785, respectively). The weighted average cost per beneficiary for the two-way group was higher than that for the one-way SMS text messaging intervention group (US $62 and US $41, respectively) (Figure 2). In addition, the weighted average cost per contact was estimated at US $0.85 for the two-way and US $0.66 for the one-way SMS text messaging intervention group (Figure 3). The detailed annual incremental costs and cost per beneficiary by input categories for each facility and the weighted average estimates are presented in Multimedia Appendix 1, Multimedia Appendix 2, and Multimedia Appendix 3.

Figure 4 presents the cost shares by activity and input cost categories. Personnel cost accounted for the largest share of the total costs. The personnel cost in the one-way SMS text messaging group (US $825/US $2542, 32.4%) was lower than that of the two-way SMS text messaging group (US $1794/US $3725, 48.2%), as one-way messaging required no personnel time responding to messages. The second largest share of the total costs was related to software development of the SMS text messaging management system (two-way: US $573/US $3725, 15.4% vs one-way: US $569/US $2542, 22.4%) and communication (two-way: US $470/US $3725, 12.6% vs one-way: US $261/US $2542, 10.3%) such as flat rate fees for internet usage and mobile phone minutes. In the two-way SMS text messaging group, an estimated 10.2% (US $381/US $3725) of the total costs was used for purchasing equipment and mobile devices to set up the intervention site in the facility and to implement the Mobile WAChX intervention. Overhead costs accounted for 7.8% (US $291/US $3725) of the total costs in the two-way SMS text messaging group, in the form of a clinic collaboration fee to use a clinic room for study activities (US $48 per month). Two trainings were conducted at the beginning of the project to teach nurses and retention officers how to use the Mobile WAChX system and orient them to implement the intervention, which costs US $98 per facility (3% of the total costs). The remaining budget was allocated to sensitization (US $64) and microplanning (US $54), which accounted for 3.2% (US $118/US $3725) of the total costs. The main cost drivers were similar in both two-way and one-way SMS text messaging intervention groups. The variable cost shares by input categories are shown in Figure 5. The majority of the total costs were related to personnel in both two-way and one-way SMS text messaging intervention groups (US $1794/US $2936, 61.1% vs US $825/US $1757, 46.9%, respectively), followed by communication (US $470/US $2936, 16.0% vs US $261/US $1757, 14.9%, respectively), equipment (US $381/US $2936, 13.0% vs US $381/US $1757, 21.7%, respectively), and overhead (US $291/US $2936, 9.9% vs US $291/US $1757, 16.5%, respectively).

Cost estimates in the 2 scenarios compared to the baseline scenario estimates reflected in Table 3 are presented in Multimedia Appendix 4. Expanding the two-way SMS text messaging intervention to all 6 facilities would have the effect of decreasing the estimated cost per beneficiary by 31% (from US $62 to US $43; US $19/US $62) and cost per contact by 26% (from US $0.85 to US $0.63; US $0.22/US $0.85). When building on scenario 1 and replacing research project health worker salaries with the more typical salaries for MOH staff, the cost per beneficiary decreased by 50% (from US $62 to US $31, US $31/US $62) and the cost per contact decreased by 46% (from US $0.85 to US $0.46, US $0.39/US $0.85).

The ICERs of comparing two-way SMS text messaging versus no intervention in the 2 facilities and the input parameters are summarized in Multimedia Appendix 5 and Multimedia Appendix 6, respectively. According to published RCT and meta-analysis estimates, implementing two-way SMS text messaging intervention was estimated to have suppressed viral load in 11 patients and achieved medication adherence among 15 patients, with total annual incremental costs of US $7084 [7,16]. Thus, the ICERs were US $644 per viral load suppression and US $472 per medication adherence.