https://orcid.org/0000-0002-9114-6157
Abstract
HIV status awareness is critical for ending the HIV epidemic but remains low in high-HIV-risk and hard-to-reach sub-populations. Targeted, efficient interventions are needed to improve HIV test-uptake. We examined the incremental cost-effectiveness of offering the choice of self-administered oral HIV-testing (HIVST-Choice) compared with provider-administered testing only [standard-of-care (SOC)] among long-distance truck drivers. Effectiveness data came from a randomized-controlled trial conducted at two roadside wellness clinics in Kenya (HIVST-Choice arm, n = 150; SOC arm, n = 155). Economic cost data came from the literature, reflected a societal perspective and were reported in 2020 international dollars (I$), a hypothetical currency with equivalent purchasing power as the US dollar. Generalized Poisson and linear gamma regression models were used to estimate effectiveness and incremental costs, respectively; incremental effectiveness was reported as the number of long-distance truck drivers needing to receive HIVST-Choice for an additional HIV test-uptake. We calculated the incremental cost-effectiveness ratio (ICER) of HIVST-Choice compared with SOC and estimated 95% confidence intervals (CIs) using non-parametric bootstrapping. Uncertainty was assessed using deterministic sensitivity analysis and the cost-effectiveness acceptability curve. HIV test-uptake was 23% more likely for HIVST-Choice, with six individuals needing to be offered HIVST-Choice for an additional HIV test-uptake. The mean per-patient cost was nearly 4-fold higher in HIVST-Choice (I$39.28) versus SOC (I$10.80), with an ICER of I$174.51, 95% CI [165.72, 194.59] for each additional test-uptake. HIV self-test kit and cell phone service costs were the main drivers of the ICER, although findings were robust even at highest possible costs. The probability of cost-effectiveness approached 1 at a willingness-to-pay of I$200 for each additional HIV test-uptake. HIVST-Choice improves HIV-test-uptake among truck drivers at low willingness-to-pay thresholds, suggesting that HIV self-testing is an efficient use of resources. Policies supporting HIV self-testing in similar high risk, hard-to-reach sub-populations may expedite achievement of international targets.
Having a choice to self-administer oral HIV testing improves HIV test uptake at a low willingness to pay value compared with only the provider-administered HIV testing
Introduction
The HIV epidemic in the sub-Saharan Africa region remains a major global public health challenge, with over 1 million people living with HIV unaware of their status (Joint United Nations Programme on HIV/AIDS, 2021). Lack of awareness of HIV status has downstream implications along the HIV care continuum. Early diagnosis promotes timely linkage to care, antiretroviral therapy (ART) initiation and viral suppression, as well as achievement of the UNAIDS 90–90–90 targets (Joint United Nations Programme on HIV/AIDS, 2021). However, uptake of HIV testing services remains low in some sub-populations, such as long-distance truck drivers (LDTDs) (Botão et al., 2016; Lalla-Edward et al., 2017), that are disproportionately impacted by HIV.
LDTDs are particularly difficult to reach in order to promote uptake of health services. An extremely mobile population, LDTDs have irregular work schedules and hours that are discordant with healthcare facility operating hours, limiting the accessibility and utilization of healthcare services, including routine HIV testing (Lalla-Edward et al., 2018). This is relevant given that they travel for many days away from their families, which increases the likelihood of engaging with multiple partners along truck routes, which increases risk of acquiring and/or transmitting HIV (Morris and Ferguson, 2006).
Emerging evidence suggests that patient-centred care delivery, such as HIV self-testing, improves HIV test uptake (Kelvin and Akasreku, 2020; Jamil et al., 2021) due to its acceptability, scheduling flexibility and user privacy (Figueroa et al., 2015). The introduction of self-administered oral HIV testing to complement the provider-administered HIV testing has improved uptake of HIV testing among high-risk sub-populations (Ortblad et al., 2017), including LDTDs, by nearly 3-fold (Kelvin et al., 2018; 2019). But evidence on cost (George et al., 2018; Matsimela et al., 2021) and cost-effectiveness (Okoboi et al., 2021) of self-administered oral HIV testing among hard-to-reach sub-populations, including LDTDs, remains limited.
With a nearly 30% shortage in global funding for HIV response programs in 2019 (UNAIDS, 2021) and emerging financing challenges due to the COVID-19 pandemic (The Global Fund, 2021), there is need to prioritize available limited resources. In this context, the current study leveraged individual-level health data from a recent randomized controlled trial (Kelvin et al., 2018) to examine the economic costs and willingness to pay for the choice of self-administered oral HIV testing (HIVST-Choice) compared with provider-administered HIV testing (SOC), among LDTDs presenting for care at a roadside wellness clinic in Kenya. This study was designated as exempt by the Virginia Commonwealth University Institutional Review Board.
Methods
Effectiveness data
Effectiveness data came from a randomized controlled trial, which was conducted in 2015 at two Kenyan roadside wellness clinics to which LDTDs were already presenting for care. The trial compared HIV testing uptake among LDTDs in two arms: HIVST-Choice and SOC (Kelvin et al., 2018). In HIVST-Choice, LDTDs (n = 150) were offered the choice of using (1) provider-administered rapid finger-prick or (2) self-administered rapid oral HIV testing at the clinic. If the two options were declined, they were offered a third option: (3) self-administered rapid oral HIV testing for home use. In SOC, LDTDs (n = 155) were offered provider-administered finger-prick HIV testing only. LDTDs in HIVST-Choice had 2.8, 95% CI [1.5, 5.4], higher odds of HIV testing uptake compared with those in SOC (Kelvin et al., 2018). This statistically significant difference emerged only after including offering the option to self-test at home in the analysis.
Cost data
Economic cost data came from the literature, including from a related study on SMS text messaging to promote HIVST among truck drivers in Kenya (George et al., 2018) and reflected a societal perspective. We used economic costs from the literature, since programmatic and budgetary data were available for a related study (George et al., 2018), but not for the current study. For the literature review, we identified candidate cost sources based on geographical setting, year of the study and relevancy to the current study. Studies conducted in Kenya (setting of the trial) were prioritized, supplemented by contextually relevant studies from lower-middle-income sub-Saharan African countries when local data were unavailable. We restricted the search to studies with cost data that were collected within 10 years from the trial implementation to capture more recent healthcare delivery and utilization patterns.
Costing approach
Micro- and gross costing approaches were used to assign per-patient costs. We used micro-costing for costs such as medical costs (consumables and HIV test kits), labour (medical staff salary) and cost of patient time spent conducting the HIV test. These costs required more precise estimation for resources utilized by multiplying the quantity of resources and the unit cost (Xu et al., 2014). Gross-costing aggregates costs for an intervention to estimate the per-patient cost for resources that cannot be explicitly allocated at the patient level based on individual utilization (Clement et al., 2008). Gross costing was used for costs such as capital, cell phone, cell phone service and overheads by dividing the aggregated cost by the number of patients. Capital costs were annualized and then converted into monthly values. This approach was also used in a prior related study in a similar setting (George et al., 2018). Costs were estimated based on the HIV testing procedure (provider-administered or self-administered oral) performed and setting (clinic only or clinic and home). Medical costs included the cost of HIV testing kits [OraQuick for self-administered, I$16.97 (Kelvin et al., 2017); Colloidal Gold test for provider-administered, I$1.56 (Iyer et al., 2013)] as well as medical supplies (self-administered, provider-administered) used in the HIV testing process at the clinic (Cherutich et al., 2018), which varied based on the HIV testing procedure.
Labour costs included salaries for the HIV testing counsellor per-patient (self-administered, provider-administered) (Kelvin et al., 2017), non-clinical healthcare facility staff (cost per-patient tested at the clinic; cost per-patient visiting the clinic but testing at home) (George et al., 2018) and one-time training for the HIV testing counsellor on how to use the HIV self-administered oral kit (Kelvin et al., 2017).
Other costs included: test administration costs (testing at the clinic only, visiting the clinic but testing from home) (George et al., 2018), facility overhead (e.g. utility bills) (George et al., 2018) and cell phone (cost of purchasing smartphones) and cell phone service (monthly cost to facilitate phone-based interviews and post-test counselling) (Kelvin et al., 2017). The cell phone and cell phone service costs were calculated by dividing the total cost of purchasing four phones and cell phone service for 3 months with the number of LDTDs in the intervention. These costs were allocated using gross-costing approaches (Clement et al., 2008).
Patient time spent at the healthcare facility or conducting home testing for HIV, including pre- and post-test counselling, was considered time lost that could have been alternatively used to economically benefit the patient. Patient time (for self-administered tests, provider-administered tests) was calculated as the product of the time spent testing for HIV and the hourly wage of a truck driver, estimated based on income data of participants in the trial. The HIV testing process took 40 and 50 min for participants that tested using SOC and oral self-administered test, respectively (Kelvin et al., 2017).
Costs were reported in 2020 international dollars (I$) (The World Bank, 2022), which is a hypothetical currency that captures differences in local currency purchasing power and thus allows cross-country cost comparisons. We used the international dollar, versus the US dollar, given its stability over time and the potential for substantial variation in market exchange rates, even in the short term (Turner et al., 2019), when using other currencies. Where applicable, costs were adjusted for inflation using the World Bank GDP deflator (The World Bank, 2023) to account for data collected at different points in time.
Statistical analysis
We conducted analysis using the time horizon (3 months) over which the original trial was conducted in order to provide decision makers with information that could inform policy decisions relevant to the short term. To do so, we used two statistical models. A generalized linear Poisson regression model with robust variance was used to estimate the effectiveness and incremental effectiveness of HIVST-Choice and a generalized linear gamma regression model with a log-link function, which accommodates skewness in the data, to estimate the incremental cost. Equation 1 shows the generalized linear model for estimating effectiveness, incremental effectiveness and incremental costs of HIVST-Choice.
where |$Outcom{e_i}$| represents HIV testing uptake or cost assigned to the|${\ }{i^{th}}$| patient; |$Choic{e_i}$| is a binary variable equal to 1 if a patient is assigned to HIVST-Choice and equal to 0 if assigned to SOC; |${\beta _1}$| is the coefficient of interest—effect of HIVST-Choice on the outcome. |${X_{\bf{i}}}$| and |${\rm B}$| are vectors and coefficients, respectively, of the control variables. We controlled for four variables that have been found to impact HIV testing uptake (Staveteig et al., 2013) and/or are contextually applicable to the study. These include the healthcare facility, participant age, the purpose of the participant’s clinic visit (HIV testing vs not) and whether the participant had paid for sex in the last 6 months, thereby increasing risk of HIV acquisition.
Effectiveness and incremental effectiveness
We estimated the effectiveness of HIVST-Choice as the relative risk of HIV test uptake for the HIVST-Choice compared with SOC, which was used in estimating the incremental effectiveness, the Number Needed to Treat (NNT) (Cook and Sackett, 1995). NNT is interpreted as the number of LDTDs who need to be offered HIVST-Choice for an additional driver to be tested for HIV. The NNT approach was selected as an alternative to more traditional measures of incremental effectiveness such as disability adjusted life years (DALYs) averted because the primary outcome (HIV test uptake) in the trial was an intermediate outcome, and the trial time period (3 months) was too short to meaningfully estimate or apply a final outcome, such as DALYs, for a long-term, complex chronic disease such as HIV.
Economic costs and incremental costs
Per patient economic costs, by trial arm, were calculated by multiplying resources utilized at the individual level with unit costs. We summarized per patient economic costs by trial arm using the mean with 95% CIs, since decisions are made based on expected costs (Glick et al., 2014). Incremental costs reflect the difference in mean per-patient costs between HIVST-Choice and SOC estimated using a generalized linear gamma model. Future costs and benefits were not discounted given the short analytic time horizon (3 months).
Incremental cost-effectiveness
The ICER for HIVST-Choice compared with SOC was estimated by dividing the difference in costs by the difference in effectiveness. Non-parametric bootstrapping was used to estimate 95% CIs since the data (cost and effectiveness variables) were not normally distributed (Glick et al., 2014). Although there is no specific number of resamples required for the bootstrapping method, it is recommended to resample more than 1000 bootstrap samples to generate unbiased estimates (Glick et al., 2014); We created 1500 bootstrap samples. We used the bias-corrected method to generate CIs because this approach corrects for bias and skewness in the distribution of bootstrap estimates (Tibbe and Montoya, 2022). The ICER represents the cost per additional HIV test uptake for LDTDs offered the choice of a self-administered oral test cost compared with offering only the provider-administered HIV test. We used this interpretation given availability of intermediate, versus final, outcomes as our effectiveness measure.
Missing data
In the analytic sample, nine (3%) participants were missing at least one data point. Using Little’s test, we examined the randomness assumption about the missing data and found that the data were missing completely at random (MCR) across trial arms (Little, 1988) implying that the missing data were not systematically correlated with other variables across trial arms. This was confirmed by creating dummy variables and conducting t-tests comparing the outcomes and other co-variates between LDTD with missing and those without missing each variable. Participants (9) with data MCR were excluded from the analysis since they had no significant impact on the study outcomes.
Uncertainty
We assessed the uncertainty of costs using a tornado diagram and the cost-effectiveness acceptability curve (Löthgren and Zethraeus, 2000; Maiwenn, 2013). Further, we examined the best- (all lower bound costs values) and worst-case scenarios (all upper bound values). Although these scenarios may be unrealistic in practice, they can provide insight into the policy impact of the most optimistic and pessimistic cases in cost variation. The acceptability curve summarized the probability of HIVST-Choice being cost-effective compared with SOC at different willingness-to-pay thresholds. We generated the acceptability curve from a joint distribution of incremental costs and incremental effects, which was estimated using non-parametric bootstrapping (Löthgren and Zethraeus, 2000).
Results
Baseline
As in the original study (Kelvin et al., 2018), HIVST-Choice significantly increased HIV testing uptake among LDTDs. Approximately 87% (130/149) of LDTDs tested for HIV in HIVST-Choice compared with 73% (114/156) of LDTDs in SOC. LDTDs in HIVST-Choice were 23% more likely to test for HIV relative to those in SOC. The incremental effectiveness, measured as the NNT, was 6.25, 95% CI [4.51, 8.02], meaning that for every six LDTDs offered the choice of a self-administered oral HIV test, one additional driver tested for HIV.
The mean per-patient cost (Table 1) was nearly four times higher in HIVST-Choice (I$39.28) compared with SOC (I$10.80). The majority (>70%) of the mean per-patient cost in the HIVST-Choice was attributed to cell phone service use (I$13.16), HIV test kit (I$11.42) and overheads (I$3.93), while in the SOC, it was attributed to overheads (I$3.35), patient time (I$1.98) and HIV testing counsellor salary (I$1.79). The incremental cost was I$27.92, 95% CI [I$24.71, I$31.13], representing the adjusted difference in the mean per-patient costs between the HIVST-Choice and SOC.
Mean costs per patient, by cost component and trial arm, reported in 2020 I$
| HIVST-Choice arma | SOC armb | ||||
|---|---|---|---|---|---|
| Cost component | Mean | 95% CI | Mean | 95% CI | P-valuec |
| HIV test kit | 11.42 | [10.12–12.72] | 1.13 | [1.02–1.24] | <0.001 |
| Medical supplies | 0.27 | [0.24–0.29] | 0.33 | [0.29–0.36] | <0.001 |
| Labour | |||||
| HIV testing counsellor | 2.59 | [2.43–2.76] | 1.79 | [1.61–1.97] | <0.001 |
| Health facility staff | 1.01 | [0.94–1.08] | 0.87 | [0.78–0.96] | 0.015 |
| Training | 0.09 | [0.08–0.09] | 0.00 | [0.00–0.00] | <0.001 |
| Capital costs | |||||
| Health facility | 1.57 | [1.47–1.68] | 1.36 | [1.21–1.50] | 0.014 |
| Equipment | 2.38 | [2.23–2.53] | 0.00 | [0.00–0.00] | <0.001 |
| Overhead | 3.93 | [3.66–4.19] | 3.35 | [3.01–3.69] | 0.009 |
| Cell phone service | 13.16 | [12.34–13.98] | 0.00 | [0.00–0.00] | <0.001 |
| Patient time | 2.86 | [2.68–3.05] | 1.98 | [1.78–2.18] | <0.001 |
| Cost per patient | 39.28 | [36.57–42.00] | 10.80 | [9.72–11.88] | <0.001 |
| HIVST-Choice arma | SOC armb | ||||
|---|---|---|---|---|---|
| Cost component | Mean | 95% CI | Mean | 95% CI | P-valuec |
| HIV test kit | 11.42 | [10.12–12.72] | 1.13 | [1.02–1.24] | <0.001 |
| Medical supplies | 0.27 | [0.24–0.29] | 0.33 | [0.29–0.36] | <0.001 |
| Labour | |||||
| HIV testing counsellor | 2.59 | [2.43–2.76] | 1.79 | [1.61–1.97] | <0.001 |
| Health facility staff | 1.01 | [0.94–1.08] | 0.87 | [0.78–0.96] | 0.015 |
| Training | 0.09 | [0.08–0.09] | 0.00 | [0.00–0.00] | <0.001 |
| Capital costs | |||||
| Health facility | 1.57 | [1.47–1.68] | 1.36 | [1.21–1.50] | 0.014 |
| Equipment | 2.38 | [2.23–2.53] | 0.00 | [0.00–0.00] | <0.001 |
| Overhead | 3.93 | [3.66–4.19] | 3.35 | [3.01–3.69] | 0.009 |
| Cell phone service | 13.16 | [12.34–13.98] | 0.00 | [0.00–0.00] | <0.001 |
| Patient time | 2.86 | [2.68–3.05] | 1.98 | [1.78–2.18] | <0.001 |
| Cost per patient | 39.28 | [36.57–42.00] | 10.80 | [9.72–11.88] | <0.001 |
Abbreviations: SOC = Standard of care.
Participants were offered the choice to test for HIV using (1) the provider-administered HIV testing or (2) self-administered oral HIV-testing under the supervision of a provider. If the truck driver declined the two options, they were offered a third option; (3) self-administered oral HIV-testing outside the clinic (at home) without supervision of a provider.
Participants were offered on-site provider-administered HIV testing.
The P-values are from the Wilcoxon rank sum test for differences in median costs by trial arm.
Mean costs per patient, by cost component and trial arm, reported in 2020 I$
| HIVST-Choice arma | SOC armb | ||||
|---|---|---|---|---|---|
| Cost component | Mean | 95% CI | Mean | 95% CI | P-valuec |
| HIV test kit | 11.42 | [10.12–12.72] | 1.13 | [1.02–1.24] | <0.001 |
| Medical supplies | 0.27 | [0.24–0.29] | 0.33 | [0.29–0.36] | <0.001 |
| Labour | |||||
| HIV testing counsellor | 2.59 | [2.43–2.76] | 1.79 | [1.61–1.97] | <0.001 |
| Health facility staff | 1.01 | [0.94–1.08] | 0.87 | [0.78–0.96] | 0.015 |
| Training | 0.09 | [0.08–0.09] | 0.00 | [0.00–0.00] | <0.001 |
| Capital costs | |||||
| Health facility | 1.57 | [1.47–1.68] | 1.36 | [1.21–1.50] | 0.014 |
| Equipment | 2.38 | [2.23–2.53] | 0.00 | [0.00–0.00] | <0.001 |
| Overhead | 3.93 | [3.66–4.19] | 3.35 | [3.01–3.69] | 0.009 |
| Cell phone service | 13.16 | [12.34–13.98] | 0.00 | [0.00–0.00] | <0.001 |
| Patient time | 2.86 | [2.68–3.05] | 1.98 | [1.78–2.18] | <0.001 |
| Cost per patient | 39.28 | [36.57–42.00] | 10.80 | [9.72–11.88] | <0.001 |
| HIVST-Choice arma | SOC armb | ||||
|---|---|---|---|---|---|
| Cost component | Mean | 95% CI | Mean | 95% CI | P-valuec |
| HIV test kit | 11.42 | [10.12–12.72] | 1.13 | [1.02–1.24] | <0.001 |
| Medical supplies | 0.27 | [0.24–0.29] | 0.33 | [0.29–0.36] | <0.001 |
| Labour | |||||
| HIV testing counsellor | 2.59 | [2.43–2.76] | 1.79 | [1.61–1.97] | <0.001 |
| Health facility staff | 1.01 | [0.94–1.08] | 0.87 | [0.78–0.96] | 0.015 |
| Training | 0.09 | [0.08–0.09] | 0.00 | [0.00–0.00] | <0.001 |
| Capital costs | |||||
| Health facility | 1.57 | [1.47–1.68] | 1.36 | [1.21–1.50] | 0.014 |
| Equipment | 2.38 | [2.23–2.53] | 0.00 | [0.00–0.00] | <0.001 |
| Overhead | 3.93 | [3.66–4.19] | 3.35 | [3.01–3.69] | 0.009 |
| Cell phone service | 13.16 | [12.34–13.98] | 0.00 | [0.00–0.00] | <0.001 |
| Patient time | 2.86 | [2.68–3.05] | 1.98 | [1.78–2.18] | <0.001 |
| Cost per patient | 39.28 | [36.57–42.00] | 10.80 | [9.72–11.88] | <0.001 |
Abbreviations: SOC = Standard of care.
Participants were offered the choice to test for HIV using (1) the provider-administered HIV testing or (2) self-administered oral HIV-testing under the supervision of a provider. If the truck driver declined the two options, they were offered a third option; (3) self-administered oral HIV-testing outside the clinic (at home) without supervision of a provider.
Participants were offered on-site provider-administered HIV testing.
The P-values are from the Wilcoxon rank sum test for differences in median costs by trial arm.
In the base case, the ICER was I$174.51, 95% CI [I$161.54, I$186.80]. That is, offering LDTDs the choice of a self-administered oral test cost approximately I$175 per additional HIV test uptake compared with offering only the provider-administered HIV test.
Uncertainty
Our findings were robust to variations in economic costs. The costs of cell phone service and HIV self-testing kit were the key cost drivers and had the largest impact on the ICER (Figure 1), with the upper bound of the ICERs for both sensitivity analyses falling well below traditional willingness-to-pay thresholds. We also examined the potential impact of reducing the HIV self-testing kit cost to US$2, based on the Bill and Melinda Gates Foundation agreement with manufacturers of the self-testing kits and low-income countries (OraSure Technologies, 2017). With the reduced cost, the ICER was I$124 for each additional HIV test uptake. There was limited variation in the ratio for both the best- and worst-case scenarios (see supplementary material).
One-way sensitivity analysis of unit costs
When we used non-parametric bootstrapping to examine the joint distribution of incremental costs and incremental effectiveness, all the data points on the cost-effectiveness plane remained in the northeast quadrant (Figure 2, Panel A). This indicates that while HIVST-Choice increased both costs and HIV testing uptake, there was less uncertainty in the cost per additional HIV test uptake compared with SOC since all the data points clustered in the same quadrant. Using these data points to calculate ICERs, the probability that the ICER fell below a given willingness-to-pay threshold approached 1 when willingness-to-pay exceeded I$200 (Figure 2, Panel B).
Joint distribution of cost and effect of HIV testing uptake, and the cost-effectiveness acceptability curve
Discussion
HIV status awareness remains low, particularly in high-risk sub-populations in sub-Saharan Africa (Joint United Nations Programme on HIV/AIDS, 2021), and to sustainably improve the uptake of HIV testing in these populations requires efficient, innovative and targeted strategies. Leveraging individual-level data from a randomized controlled trial (Kelvin et al., 2018), we examined the incremental cost-effectiveness of the choice of provider-administered or self-administered oral HIV testing compared with offering provider-administered HIV testing only among LDTDs seeking care at two roadside clinics in Kenya. We found the intervention effective—increasing the probability of HIV testing uptake by 23%—and cost-effective when decision maker willing-to-pay is approximately I$175 per additional HIV test uptake in this population, with this finding robust across multiple sensitivity analyses.
We interpret our findings in terms of decision maker willingness-to-pay for an additional HIV test uptake when offering HIVST-Choice to LDTDs. We acknowledge that cost-effectiveness analysis studies typically assume use of a final outcome (such as DALYs), versus intermediate outcomes as in the current analysis, and thus our findings may not be directly comparable to other studies. Future research on the cost-effectiveness of this intervention using final, longer-term outcomes is warranted to provide estimates of the value for money of the HIVST-Choice intervention that are comparable with emerging literature on this topic.
Our findings contribute to an emerging literature on the value for money of HIV self-testing in hard-to-reach sub-populations, such as LDTDs. The current study complements a cost analysis by George et al. of an SMS-based intervention among LDTDs and female sex workers that found self-administered oral HIV testing per test costs more (double) than routine facility-based testing (SOC),(George et al., 2018) which is consistent with the current study. However, the work found the cost for each additional LDTDs tested was <$1 (George et al., 2018). This differs substantially from the current analysis for several reasons, including analysis of a different intervention (SMS-based messaging), the higher costs of cell phones and cell service in the current analysis (Kelvin et al., 2017), the inclusion of patient time costs and methodologic differences in estimating additional costs for each additional HIV test uptake (George et al., 2018). Other work on female sex workers, another high-risk and hard-to-reach population, found HIV self-testing was cost-effective in this population and in settings with high prevalence of undiagnosed HIV, with findings forecast over a 50-year time horizon comparable to findings in the current study (Cambiano et al., 2019).
Our work complements other economic evaluations of HIV self-testing in the literature, although these prior studies have largely focused on the overall population of people living with HIV and forecast economic costs and final health outcomes over longer time horizons. In Zimbabwe and Malawi, e.g. provider-administered facility-based HIV testing plus HIV self-testing was cost-effective compared with only the provider administered HIV testing at $253 per quality-adjusted life year gained (Maheswaran et al., 2018). Similar to findings in this study, the cost of an HIV self-testing kit was higher compared with the provider-administered HIV testing kit and was one of the key variables impacting the cost-effectiveness of HIV self-testing. In Zimbabwe, HIV self-testing was cost-effective when efficacy was at least 20% (Cambiano et al., 2015), which is comparable to the effectiveness (23%) of HIVST-Choice.
The ICER (I$175) in the current study is comparable to prior studies on cost-effectiveness of self-testing using intermediate outcomes in sub-Saharan Africa. In Uganda, among men who have sex with men, peer-delivered HIV self-test kits were $147.30 per additional HIV-positive person identified compared with the standard facility-based and hotspot provider-administered HIV testing (Okoboi et al., 2021). Similarly, in Malawi, findings were comparable, with facility-based HIV self-testing costing $103 per additional newly diagnosed individual compared with standard provider-administered HIV testing (Nichols et al., 2020).
Findings indicate that self-care interventions (Narasimhan et al., 2020; World Health Organization, 2022)—in this case, choice of self-administered oral HIV testing for high-risk and hard-to-reach sub-populations—are an effective strategy to improve HIV test uptake, with our estimates of incremental cost-effectiveness, despite differences in time horizon and outcome, falling below multiple thresholds for cost-effectiveness and aligning with the current literature. In Eastern and Southern Africa, HIV testing uptake remains low despite healthcare services being geographically and temporally convenient (North Star Alliance, 2018). For example, roadside wellness centres, such as those run by North Star Alliance, offer a broad menu of healthcare services, including HIV testing and treatment, close to truck stops where LDTDs, sex workers and roadside community residents congregate and interact and are open after hours when these groups are more likely to have time to seek services, but despite the geographic proximity to health care services, uptake of HIV testing remains low (North Star Alliance, 2018). HIVST-Choice provides an efficient potential solution to some of the limitations (e.g. lack of flexibility and privacy) of SOC HIV testing offered at the roadside wellness clinics and conventional primary health clinics (Lalla-Edward et al., 2019).
Our results contribute to a substantial gap in knowledge on efficient strategies to improve HIV status awareness among LDTDs and other hard-to-reach sub-populations in the sub-Saharan Africa, a region with more than half of the world’s HIV population. In Kenya and Uganda, along the trans-African highway, sexual interaction between transport workers and communities at truck stops was estimated to contribute up to 4148 new HIV infections per year (Morris and Ferguson, 2006). International and local policy makers could implement strategies such as HIVST-Choice that improve HIV testing uptake as one strategy to reduce onward HIV transmission by diagnosing people early and enrolling them on treatment. Further, people living with HIV who are aware of their HIV status are likely to have fewer sexual partners, use condoms, seek treatment and have reduced viral load compared with those unaware of their status (Delavande and Kohler, 2012), thus reducing onward HIV transmission, even if they have unprotected sex. Findings from this study thus may provide supporting evidence to guide policy makers in their decision-making and implementation of HIV self-testing, particularly in high-risk and hard-to-reach sub-populations.
This study has limitations. First, we estimated economic costs and willingness to pay using the trial’s shorter 3-month time horizon, rather than a longer lifetime or similar time horizon, which limits comparability of our findings with existing literature. We selected this time horizon to make our findings more relevant to the shorter-term time horizons over which policy decisions are made. Future research can complement these findings by examining the efficiency of the intervention over a longer time horizon. Second, our estimates of effectiveness and incremental effectiveness were based on an intermediate outcome (HIV test uptake), which limits comparability with cost-effectiveness studies in literature that use final outcomes (e.g. disability adjusted life years). However, our findings align with the existing literature, albeit limited, that use final outcomes when examining the cost-effectiveness of HIV self-testing in hard-to-reach populations. Third, economic cost data were derived from literature since programmatic and budgetary data from the trial were not available. Additionally, we conducted sensitivity analyses to account for uncertainty in cost estimates, with our findings remaining robust. Further research that considers how sources for economic costs (e.g. programmatic budgets versus the literature) impact estimates of incremental cost-effectiveness and policy conclusions is warranted.
Conclusions
As countries aim to achieve UNAIDS targets with limited resources available, innovative, tailored and efficient strategies are imperative, particularly for sub-populations that are hard-to-reach and at high risk of acquiring and transmitting HIV. This study finds offering self-administered oral HIV testing as a testing choice at roadside wellness clinics in Kenya to be an efficient use of resources compared with SOC of offering provider-administered blood-based HIV testing only. Future studies should examine the cost-effectiveness of self-administered HIV testing outside the clinic setting among high-risk and hard-to-reach sub-populations and consider the long-term costs and health benefits of HIV testing.
Supplementary data
Supplementary data is available at HEAPOL Journal online.
Data availability
The data that support the findings of this study cannot be shared publicly due to partipant privacy considerations but are available on reasonable request to the corresponding author.
Funding
This study was supported by the International Initiative for Impact Evaluation (3IE # TW2.2.06). Additional support came from the National Institutes of Health’s National Institute of Allergy and Infectious Diseases (NIAID), the Eunice Kennedy Shriver National Institute of Child Health & Human Development (NICHD), the National Cancer Institute (NCI), the National Institute on Drug Abuse (NIDA), the National Heart, Lung, and Blood Institute (NHLBI), the National Institute on Alcohol Abuse and Alcoholism (NIAAA), the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), the Fogarty International Center (FIC), the National Library of Medicine (NLM) and the Office of the Director (OD) under Award Number U01AI096299 (Central Africa-IeDEA). Support was also received from the Virginia Commonwealth University Blick Scholar Award and the Einstein-Rockefeller-CUNY Center for AIDS Research [P30-AI124414] which is supported by the following National Institutes of Health (NIH) Co-Funding and Participating Institutes and Centers: NIAID, NCI, NICHD, NHBL, NIDA, NIMH, NIA, FIC and OAR.
Author contribution statement
Conception or design: Deo Mujwara, April Kimmel.
Data collection: Deo Mujwara, Elizabeth Kelvin, Gavin George, Eva Mwai.
Data analysis and interpretation: Deo Mujwara, Bassam Dahman, Daniel Nixon, Tilahun Adera, April Kimmel.
Drafting the article: Deo Mujwara, April Kimmel.
Critical revision of the article: Deo Mujwara, Elizabeth Kelvin, Bassam Dahman, Gavin George, Daniel Nixon, Tilahun Adera, Eva Mwai, April Kimmel.
Reflexivity statement
Gender and seniority
Authors in this study included three women and five men with various levels of seniority: Deo Mujwara (recent graduate), Elizabeth Kelvin (associate professor), Bassam Dahman (associate professor), Gavin George (professor), Daniel Nixon (physician), Tilahun Adera (professor), Eva Mwai (researcher), April Kimmel (associate professor).
Regional location and experience
Authors are from different regions and have expertise in HIV, resource allocation, biostatistics and epidemiology. The first author, Deo Mujwara, has experience in health policy and HIV resource allocation in sub-Saharan Africa. Two authors, Gavin George and Eva Mwai, are based in South Africa and Kenya, respectively, and their work has largely focused on HIV in sub-Saharan Africa. April Kimmel, Elizabeth Kelvin, Bassam Dahman, Tilahun Adera and Daniel Nixon are based in the USA. April Kimmel has extensive research experience in HIV policy and resource allocation in sub-Saharan Africa. Elizabeth Kelvin has worked on various research projects in East and South Africa, including the randomized, controlled trial which serves as the basis for this study. Bassam Dahman is a biostatistician with expertise in statistical methodologies, HIV and other chronic conditions. Daniel Nixon is an HIV clinician and researcher, with experience in sub-Saharan Africa. Tilahun Adera has extensive research experience in epidemiology in various settings including countries in Africa.
Ethical approval.
Ethical approval for this research was waived by the authors institute IRB.
Conflict of interest statement:
None declared.
