Parent project and intervention description
This microcosting analysis was conducted within in the aPS scale-up project (NIAID R01AI134130), a collaborative implementation science study between the Ministry of Health (MOH) National AIDS and STI Control Program (NASCOP), PATH-Kenya and University of Washington, conducted in 31 health facilities in Kisumu and Homa Bay counties in western Kenya [14]. Details of the aPS scale-up project and its implementation procedures have been published [14]. The overall goal of the project was to implement and evaluate the effectiveness of aPS when integrated within routine HTS, and assess implementation outcomes including acceptability, demand, integration, implementation fidelity, and costs.
The aPS scale-up project focused on evaluating aPS as a strategy for increasing HIV testing among men [14]. Though aPS is offered to both male and female clients receiving HTS in Kenya, this project deliberately targeted adolescent girls and women and their partners to address challenges in finding men who are difficult to mobilize through other interventions and may increase the burden of HIV through risky sexual behavior if not tested and treated Briefly, female clients testing HIV-positive at participating facilities (female index clients) received information on aPS and were screened for eligibility by healthcare workers. Those eligible were ≥15 years of age - those between 15 and 18 years were emancipated minors as per Kenya guidelines [15], newly diagnosed HIV-positive, at low risk of intimate partner violence (IPV), not pregnant, and had at least one sexual partner within the last 3 years. Participants were classified to have either low, moderate, or high IPV risk using the IPV screening tool adopted from the national APS guidelines [16]. Women at low risk of IPV did not fear IPV from their partner or had never experienced any form of IPV (emotional, physical, sexual). Pregnant women were excluded as they were considered a vulnerable population and were instead offered home-based couple counseling and testing.
Consenting female index clients were asked to provide names and contact information for all MSPs in the last three years, a process called partner elicitation. HTS providers contacted MSPs via phone and/or physical (in-person) tracing to notify them of their potential HIV exposure and offer HIV testing. MSPs testing HIV-positive were asked to enroll in aPS and provide contact information for their female sex partners, who were also followed up, notified of their exposure, and offered HTS. Female index clients and sex partners testing HIV-positive were encouraged to link to care and followed up at 6 weeks, 6 months, and 12 months to assess linkage to care, ART initiation, and HIV viral load suppression.
Cost data collection
We estimated the incremental financial and economic costs of integrating aPS into the HTS program using a payer perspective [17] following principles outlined in the Global Health Cost Consortium Reference Case [18]. Financial costs represent actual expenditure on goods and services, while economic costs reflect the value of resources used to produce output. We used activity-based ingredients approach to identify key aPS activities, inputs, resource use, and associated prices and values of goods and services. We identified key activity cost centers and used microcosting methods to quantify and value inputs from each activity across facilities.
Cost data were collected during three field visits. In August 2018, we collected start-up costs from MOH NASCOP and PATH offices associated with one-time planning, training and awareness activities that took place before the project started. We then collected recurrent costs one year after study initiation from 14 facilities in Kisumu (n = 8) in August 2019, and then in Homa Bay (n = 6) in January 2020. Facilities were purposefully sampled based on location (county, urban/peri-urban/rural) and client volume (based on patient volumes receiving HTS at the facilities) after consultation with the site team.
We extracted data from project expense reports and MOH budgets, and obtained supplementary information on all key activities and resource use for the aPS integration. We also conducted semi-structured interviews with key health personnel at the MOH and PATH, as well as facility administrators to obtain information on time use and shared program costs (rent, personnel salaries, and supply prices from MOH sources) that were not available from expense and budget reports. We disaggregated costs by facility to evaluate variations in incremental costs and average unit cost per MSP.
We included health system costs incurred during provider elicitation of MSPs from female index clients, as well as for phone and physical tracing for MSPs. We excluded costs of eliciting and tracing female sex partners of HIV-positive MSPs as this was not the primary focus of the costing analysis. We included the costs of HIV testing and linking MSPs testing HIV-positive to care, but did not include the costs of ART since these costs are incurred under the national HIV care and treatment program, which is separate from the HTS/aPS program. We also excluded the cost of research activities not part of routine aPS delivery.
Cost analysis
We differentiated between new aPS costs, and shared program costs to support the integration of aPS services into the existing HTS program. New costs were those related to aPS inputs and activities not conducted prior to aPS scale-up e.g., aPS microplanning meetings, initial trainings, sensitization, transport costs for physical tracing, communication costs for phone tracing, personnel (service delivery including partner elicitation, phone and physical tracing, exposure notification, HIV testing, and linkage to care) and aPS supervision. Shared program costs from the current HTS program were allocated based on the share of the activity or input used in aPS. These included the share of program costs for vehicles, equipment, overheads, HTS supplies, health facility administration, and refresher training.
We distinguished between fixed and variable costs. Fixed costs included overheads (e.g. building costs, water, and electricity), capital (vehicles, equipment), and non-service delivery personnel costs (i.e. health facility administration and aPS supervision). We allocated building space based on the proportion of time in the HTS visit taken up by aPS activities at each facility. Rental costs were estimated from MOH rates for government facilities or rental rates from nearby commercial properties. Capital costs were annualized over the expected useful life (assumed to be five years) using a 3% annual discount rate [18]. Similarly, start-up costs (microplanning, sensitization and training), which occurred once during the project, were treated as a type of fixed costs and annualized over five years using a 3% discount rate.
We estimated variable costs by measuring resource use across the 14 facilities. Personnel time was captured as a proportion of full-time work allocated to aPS. Salaries were converted into hourly wages based on the assumption that full-time employment was equivalent to 2080 hours/year. We estimated personnel time using time-and-motion observation for partner elicitation at the clinic, and MSP outreach by phone and physical tracing. To estimate personnel time cost, we multiplied the cost per minute (including both salary and benefits) by the median time spent on aPS activities including: 1) partner elicitation by the number of female index clients seen, 2) phone tracing by the number of MSPs traced on phone accounting for approximately 40% repeat calls, and 3) physical tracing multiplied by the number of MSPs traced physically accounting for 10% repeat physical tracing attempts. Based on facility data reports, we assumed that of the MSPs who were successfully traced, 70% were traced on phone and 30% by physical tracing. Estimates for the phone calls and physical tracing attempts were based on facility reports and staff opinion.
Phone call costs were estimated as a percentage of airtime assigned to the facility per year used to call MSPs elicited through aPS. Transport costs were estimated by multiplying the number of expected commutes per year, mainly through public transport, by the average cost of each commute. For supplies and commodity costs, we observed resource use during HIV testing and multiplied the relevant quantities by input costs obtained from program budgets or centralized price lists.
Cost data were collected and analyzed in templates designed in Microsoft Excel (Microsoft, Redmond, USA). We adjusted costs to 2019 currency and converted to US dollars (USD) using the 2019 average exchange rate (1 USD = KSh 101) [19]. Additional details about the costing methodology, including the Excel file used for the analysis, are available in the Supporting Information.
Program volume
We used data collected by the implementation project staff to obtain the number of MSPs traced, tested, tested HIV-positive, and on ART over a one-year period as recorded in the MOH HTS facility registers. These data were compiled from April 1, 2019 to March 31, 2020 to capture costs at least one year prior to the COVID-19 pandemic.
Cost metrics
For each facility, we first estimated incremental costs by summing the start-up and recurrent costs. We then calculated the average unit cost per MSP traced, tested, testing HIV-positive, and on ART by dividing the incremental costs by the number of MSPs traced, tested, testing HIV-positive, and on ART, respectively. Lastly, we estimated the weighted average incremental facility cost per year by weighting the annual incremental costs in our sample of 14 facilities by the number of MSPs in each facility. We also explored all cost metrics by facility to assess how client volume and location affect total incremental and average unit costs per MSP. We estimated cost shares by activity and input to explore how resources and activities were utilized within aPS.
Scenario analysis
We estimated the costs of integrating aPS under two scenarios: 1) as-implemented, which replicates the current national HTS program where financial support is received from both government and external funding sources, and 2) MOH-only, in which, based on expert opinion from MOH and site staff, we excluded costs associated with international non-governmental organizations (NGO) i.e., we assumed that all HTS providers transitioned into the MOH human resource system, that only MOH staff would supervise aPS delivery, and that no international NGO overhead costs were incurred.
Ethical approval
This study received ethical approval from the Kenyatta National Hospital Ethical and Scientific Review Committee (P465/052017) and the University of Washington Institutional Review Board (STUDY00002420). This study was conducted in accordance with the Declaration of Helsinki, and all study participants gave informed consent for enrolment and follow-up prior to study participation.