INTRODUCTION
Male circumcision has been demonstrated to reduce the risk of male acquisition of HIV.1–3 4 5 6 5 7–9
Studies have examined the cost and cost-effectiveness of VMMC in sub-Saharan Africa. Estimates of the average cost per surgical VMMC range from $29 in sites integrated into regular health facility activities in Kenya10 11 10–13 13,14 13,15,16
Studies modelling the cost-effectiveness of VMMC in terms of cost per HIV infection averted9,10,13,16–26 10 26 15,20,24,27
Models have also suggested that programs can be more cost-effective by targeting men in most at-risk age groups. In Zimbabwe, the modelled cost per HIV infection averted dropped from $1035 when assuming the current age distribution (13–29 years) to $811 when only men aged 20–24 years were circumcised.17 9,22–25
These age-targeted models, however, have recognized limitations: they are not based on empirical data that account for variations in cost stemming from different demand creation and service delivery approaches, which may be necessary to increase uptake of VMMC services for those at highest risk. Previous studies have documented the barriers to client demand for VMMC.28–32 33
As far as we are aware, no articles have been published either (1) presenting primary cost data on the resources needed to design and implement locally-tailored demand creation approaches or (2) analyzing the outcomes resulting from these additional costs in terms of HIV infections and DALYs averted, which would allow policymakers to weigh the effect of additional investments against other HIV prevention strategies, and, more broadly, investments in other disease areas. This study seeks to bridge that gap in the literature by analyzing the cost and cost-effectiveness of a targeted demand creation intervention in the context of a randomized controlled trial in Tanzania evaluating VMMC uptake among men aged 20–34 years in 2 settings with different HIV-prevalence rates.34
METHODS
Setting and Trial
A randomized controlled trial was conducted in Njombe and Tabora regions of Tanzania in November–December 2014 and February–March 2015, respectively. By the start of the intervention, VMMC coverage across age groups was higher in Njombe (80%) than in Tabora (55%)35 36 https://links.lww.com/QAI/B160 37,38
The trial evaluated locally-tailored demand creation and delivery of VMMC services. In addition to all the standard demand creation components present in the control arms (which included elements of mass media engagement, community mobilization, and targeted service delivery), the following were added to the intervention arms: (1) demand creation communication stressing non-HIV benefits of VMMC and the voluntary nature of HIV testing services before VMMC; (2) additional peer promoters and the involvement of 2 circumcised men from the community as auxiliary peer promoters; (3) separate waiting and group education areas for men aged ≥20 years during service delivery; (4) engagement of female partners in community-based demand creation and education about post-circumcision healing and abstinence (see Supplemental Digital Content 2 for details on demand creation activities in both arms, https://links.lww.com/QAI/B160
Details of the trial design and outcomes have been published.34
Parent sites and spin-off sites received demand creation activities (for 2 weeks and 1 week, respectively) before surgical services were offered. VMMC services were delivered for 26–34 days in each parent site. Clinical staff remained in the cluster for the duration of the trial and did not perform any other clinical services other than the VMMCs. Sites stayed open for follow-up visits after circumcisions ended. Surgeries were conducted using a mixture of reusable and disposable kits.
Costing Methods
Costs of all activities related to surgery (which includes HIV testing services and waste management), demand creation, and monitoring and supervision for the control and intervention arms of the trial were included. Costs of start-up activities for the intervention arms, namely orientation meetings for peer promoters and auxiliary peer promoters and the development of demand creation materials (which includes formative research to determine user preferences), were included in the analysis. Start-up activities for the standard of care delivery and surgery training were excluded; these activities took place long before the start of the intervention and cost data were not available.
The study adopted a provider's perspective (ie, costs of intervention delivery). A top–down costing approach (ie, dividing overall program costs by outputs) was used to account for inefficiencies, down time, and wastage. Limitations of this approach (a lack of information on individual service provision and an over reliance on implementing organization records) are acknowledged.
Staff, supplies, and capital costs (ie, goods with a market value of ≥$100 and useful life of more than 1 year) for surgery and demand creation activities were calculated using cluster-specific data. Total monitoring and supervision costs and start-up costs (in the intervention arms) were calculated and allocated across clusters in both arms. Costs of developing demand creation materials were annuatized over 5 years. Financial records of the implementing organization and the Central Medical Stores catalog were used to calculate prices of supplies and capital goods.39
The number of days worked by clinical and demand creation staff was ascertained by attendance record review. Supervision and start-up staff time was determined through interviews with the implementing organization. Salary costs were obtained from health care providers who conducted the VMMCs and averaged by clinical category. Demand creation staff were paid per diems at the standard rate of the implementing organization. Supervision and start-up salary information was gathered from interviews and record review with the implementing organization.
Costs per cluster were estimated and unit costs (average costs per VMMC) were calculated by dividing total cluster costs by the total number of circumcisions performed per cluster. Cost drivers were explored by calculating average cluster cost components in each trial arm per region (ie, total cost components across all cluster per arm/region divided by the total number of clusters in each arm/region).
Tests of statistical significance to examine cost differences between arms were not performed because of the low number of clusters per region. Cost data were collected in Tanzanian shillings (TSH) and converted to 2015 U.S. dollars using an average rate of 2130 TSH per $1.40
Effectiveness Analysis
Data on the number, client age, and location of VMMCs performed were obtained from the trial data set.34 34 https://links.lww.com/QAI/B160
The Decision Makers' Program Planning Tool (DMPPT) version 2.1 was used to estimate the primary number of HIV infections averted over a 15-year time horizon from the total number of VMMCs performed during the trial, as well as the costs saved from ART averted.26,41 26,41 1–3,9
DALYs averted following HIV infections averted as a result of VMMC were calculated using a standard approach.42 43 44 https://links.lww.com/QAI/B160 1–3,9,26,41,43–48
Cost-Effectiveness and Sensitivity Analysis
The primary outcome of analysis is the ICER estimated by comparing cost per DALY averted using a tailored demand creation approach against cost per DALY averted using the standard of care approach. Intermediate measures are total cost per cluster and unit cost per VMMC, per HIV infection averted, and per DALY averted.
A univariate deterministic sensitivity analysis was performed to measure the effect of uncertainty on a number of different parameters (including ART coverage and cost, time horizon, and supplies costs) on the ICERs.
Ethical approval was obtained from the London School of Hygiene & Tropical Medicine Research Ethics Committee, the Medical Research Coordinating Committee of the National Institute for Medical Research of Tanzania, and the Institutional Review Board of the U.S. Centers for Disease Control and Prevention. The study was also registered at ClinicalTrials.gov
RESULTS
Average total costs per cluster were higher in the intervention arms than in the control arms both in Njombe ($46,696 and $39,105, respectively) and in Tabora ($54,408 and $40,443, respectively). The breakdown by cost category (as percentage of total cost) was similar between control and intervention arms, with recurrent staff costs accounting for the largest portion of total costs (between 68% and 80%). The majority of these staff costs were related to surgery. Demand creation–related staff costs were more than 2 times higher in the intervention arms than in the control arms ($6577 vs. $2438, respectively) (Table 1 ).
TABLE 1.: Breakdown of Average Cluster Costs by Region and Arm (USD and %)
Given the VMMC delivery model implemented, most costs across both arms of the trial can be considered cluster-level fixed costs (ie, costs that remain constant at the cluster level independent of outcome). Only consumable surgery supplies can be considered variable costs (ie, costs that vary with output), and they accounted for 4% to 6% of the total costs in Njombe and 12% to 13% in Tabora.
Average costs per VMMC per cluster varied greatly, between $40.86 and $336.66. Average costs per VMMC per arm were lower in the intervention clusters than in the control clusters: $81.65 vs. $101.31, respectively. They were also, on average across both arms, more than 2 times lower in Tabora than in Njombe: $65.01 vs. $152.02, respectively. Cluster-specific unit costs and total VMMCs can be found in Supplemental Digital Content 5, https://links.lww.com/QAI/B160 Fig. 1 ).
FIGURE 1.: Plot graph of average cost per VMMC per cluster. The graph shows the average cost per VMMC per cluster (unit cost) plotted against the number of VMMCs per cluster. The data suggest that unit costs decrease as the number of VMMCs per cluster increases.
The impact of VMMC on HIV infections averted over 15 years would be greater in the intervention arm than in the control arm across the 2 regions: 287 infections would be averted compared with 169, respectively (Table 2 ). Although the total number of VMMCs was higher in Tabora than in Njombe, the impact on HIV infections averted was projected to be higher in Njombe than in Tabora, with 266 infections averted compared with 190, respectively, because of higher projected HIV incidence in Njombe. The costs per HIV infection averted were lower in the intervention arm than in the control arm both in Njombe ($1424 vs. $1917) and in Tabora ($2212 vs. $3018).
TABLE 2.: Effectiveness: HIV Infections Averted, DALYs Averted, and Costs
Similar trends emerged in DALYs averted (Table 2 ). In total, more DALYs were averted in the intervention arms (2468) than in the control arms (1444). Again, despite fewer total VMMCs in Njombe than in Tabora, 2280 DALYs were averted in Njombe and 1632 in Tabora. The costs per DALY averted were lower in the intervention arm than in the control arm in Njombe ($166 vs. $224) and in Tabora ($257 vs. $354).
The ICERs of DALYs averted (without including cost savings from ART costs avoided from clients who did not acquire HIV) were $71 in Njombe and $143 in Tabora. However, when ART costs averted were considered, cost savings were achieved resulting in negative ICERs in both regions (Fig. 2 ).
FIGURE 2.: Incremental cost-effectiveness plane. The 2 squares denote the ICERs for both study regions. Their position in the South-East quadrant suggests that the intervention dominates the control because it is both less costly and more effective.
Univariate sensitivity analysis suggests that the ICERs were relatively robust to variations in VMMC cost inputs. The greatest impact on the ICERs was driven by the cost of ART. ICERs remained negative in all variations tested, indicating that the demand creation intervention would remain cost-saving under most scenarios. Modifying the yearly cost of ART between $175 and $1545 (one-third to 3 times the base cost) led to variations in the base–case ICERs from −233% to 78% in Njombe and from −287% to 96% in Tabora. The variables examined held the same hierarchy in both regions in terms of relative sensitivity (Fig. 3 ).
FIGURE 3.: Tornado diagrams of the percentage changes in the base–case ICER (including averted costs of ART) from a deterministic 1-way analysis of key input variables per region. (1) Light blue bars show the direction and magnitude of change of the ICER when the input is at its minimum value. Conversely, dark blue bars show the direction and magnitude of change of the ICER when the input is at its maximum value. (2) Base case assumptions of parameters: ART costs: $515; ART coverage: 70%; time horizon for infections averted: 15 years; disability weights: HIV symptomatic pre-AIDS 0.274 (0.184–0.377), HIV/AIDS receiving ART 0.078 (0.052–0.111); AIDS: not receiving ART 0.582 (0.406–0.743); per diems demand creation: peer promoter $21 USD, auxiliary peer promoter $11 USD; surgical supplies costs (consumables): Tabora $24,431 (control) and $35,164 (intervention), Njombe $8070 (control) and $14,207 (intervention); clinical staff wages: $7938 USD; TSH to US dollar exchange rate: 2130 TSH:$1 USD; working days per year: 191. (3) The dotted line indicates the point at which the ICER goes from being cost-saving (left) to not cost-saving (right).
DISCUSSION
To our knowledge, this is the first economic evaluation examining a locally-tailored demand creation strategy to improve uptake of VMMC among target populations.
Although higher costs associated with the demand creation activities were incurred in the intervention arms, the unit costs per VMMC were lower in the intervention arms because of increased uptake. Despite higher unit costs and lack of impact of the intervention on the proportion of clients in the 20–34 year age group in Njombe, costs per infection averted and per DALY averted were lower due to higher projected HIV incidence. When taking into account the costs of ART provision avoided because of HIV cases averted, providing VMMCs was cost-saving in both arms of the trial and across the 2 regions. However, the intervention arms were more cost-saving than the control arms. Staff costs made up the largest proportion of costs in both arms of the trial, followed by supplies. Although this is consistent with other studies, staff costs as a percentage of total cost were higher in our study than in others.13,49
Most other studies used ingredients-based costing, which only capture direct time spent providing VMMC services. This study used a top–down costing approach, which accounts for all staff costs, whether or not the staff were fully utilized; this may explain our relatively higher staff costs. The service delivery approach used in the trial (in both arms) required all VMMC staff to remain in the same cluster for a minimum of 4 weeks regardless of client volume and to not perform other clinical activities. In routine service delivery (ie, non-research), staff move frequently, as dictated by client volume. Therefore, staff costs in our study can be considered to be fixed costs. One other published study had a similar delivery method to ours, although in their case the mobile clinics remained in the field for fewer days (12 on average vs. approximately 30 in our study).12
Although these higher costs related to fixed clinical staff could be interpreted as an inefficiency, this approach may be an essential component of an effective strategy in delivering services to harder-to-reach populations. Whether staff-heavy modalities of delivery are essential to an increased uptake of VMMC among harder-to-reach clients needs to be assessed and piloted in routine service delivery settings.
Our findings suggest that economies of scale were achieved. Figure 1 shows that the higher the number of VMMCs per cluster, the lower the unit costs, likely because of the high proportion of cluster-level fixed costs (between 87% and 96%) across both arms and regions. Although it is beyond the scope of this study to explore economies of scale in detail (and further work in this area, such as cost function analyses, would be a worthwhile endeavor), our study suggests that once the initial costs of setting up a cluster are incurred, a lower cost per VMMC can be attained if demand creation activities succeed in attracting more uncircumcised men.
The average cost per HIV infection averted ranged from $1424 in the intervention arm in Njombe to $3018 in the control arm in Tabora. These costs were within the ranges found in other studies, although the range is extremely wide. The locally-tailored demand creation strategy in the intervention arms required additional start-up costs (orientation meetings and a demand creation design stage, including formative research), as well as additional demand creation staff (peer promoters and auxiliary peer promoters) and additional demand creation supplies (ie, more flyers and banners, etc.). However, the average cost per HIV infection averted and cost per DALY averted were lower in the intervention arms than in the control arms in both regions. This suggests that additional spending on demand creation, which in turn may lead to higher demand, is justified as it makes the intervention more cost-effective than the standard of care.
Although average unit costs per VMMC were higher in Njombe for both the control and intervention arms, the average costs per HIV infection averted and per DALY averted were lower in Njombe. Fewer people were circumcised during the trial in Njombe, but the impact per circumcision on HIV acquisition was projected to be greater in that region because of higher projected HIV incidence.
Total costs of the intervention decreased when incorporating cost savings attributable to ART averted. The ICERs for both regions were negative: the intervention dominated the control as the intervention was both less expensive and more effective. We therefore found that the intervention was not only cost-effective but was also cost-saving. Other studies have also found that VMMC is cost-saving when compared with a no-VMMC scenario once accounting for ART costs averted.15,20,24,27
It is important to note, however, that as coverage of VMMC increases in the future, greater efforts will need to be made to reach an increasingly small number of uncircumcised men. Demand creation strategies may need to be strengthened or modified, incurring additional cost. These changes could potentially make the intervention less cost-effective.
Although the intervention increased overall VMMC uptake, the proportion of men in the target age range did not increase significantly in Njombe. Despite this, we found that, even if cost savings were observed in both regions, impact and cost-effectiveness in Tabora were lower than in Njombe. This suggests that the overall projected HIV incidence may be more important than client age in determining cost-effectiveness of VMMC.
Limitations
We may have underestimated the cost-effectiveness (DALYs averted) of the intervention for several reasons: (1) although VMMC reduces the risk of acquiring other infections,50
Patient costs were not included in this study. Given the period of recommended physical inactivity post-VMMC, the costs of wages lost to the individual may be substantial.51
The intervention arms had several components. However, activity-specific costs and effects were not disaggregated. Future research could look at which specific components yield the greatest uptake of VMMC and the consequent cost implications.
Some of the costs in our study, such as monitoring and supervision costs, were collected for each region and could not be disaggregated at the cluster level. Therefore we allocated these costs equally across clusters in the same region, potentially underestimating the costs in clusters that required greater oversight.
Trial implementation took place during the dry season in Tabora and the rainy season in Njombe, when more agricultural activity takes place. This may have led to differences in implementation costs and demand for services among men of productive ages.
Three separate, self-contained models and tools were used (costs, DMPPT, and DALYs), and therefore, a combined probabilistic sensitivity could not be undertaken. Consequently our analysis does not capture full uncertainty.
REFERENCES
1. Auvert B, Taljaard D, Lagarde E, et al. Randomized, controlled intervention trial of male circumcision for reduction of HIV infection risk: the ANRS 1265 Trial. PLoS Med. 2005;2:e298.
2. Bailey RC, Moses S, Parker CB, et al. Male circumcision for HIV prevention in young men in Kisumu, Kenya: a randomised controlled trial. Lancet. 2007;369:643–656.
3. Gray RH, Kigozi G, Serwadda D, et al. Male circumcision for HIV prevention in men in Rakai, Uganda: a randomised trial. Lancet. 2007;369:657–666.
4. WHO, UNAIDS. New Data on Male Circumcision and HIV Prevention: Policy and Program Implications: Montreux, Switzerland: WHO, UNAIDS; 2007.
5. World Health Organization. Voluntary Medical Male Circumcision for HIV Prevention in 14 Priority Countries in East and Southern Africa: Progress Brief June 2016.
6. Ally M, Bukuku M, Menon V, et al. Costs and Impacts of Scaling Up Voluntary Medical Male Circumcision in Tanzania. Tanzanian Ministry of Health and Social Welfare, USAID | Health Policy Initiative Costing Task Order and Joint United Nations Program on HIV/AIDS. 2012.
7. UNAIDS, WHO, SACEMA Expert Group on Modelling the Impact Cost of Male Circumcision for HIV Prevention. Male circumcision for HIV prevention in high HIV prevalence settings: what can mathematical modelling contribute to informed decision making? PLoS Med. 2009;6:e1000109.
8. Stover J, Kripke K. Estimating the effects of targeting voluntary medical male circumcision (VMMC) programs to different age groups: the decision makers program planning toolkit (DMPPT 2.0). IAS Conference. Melbourne, Australia; 2014.
9. Kripke K, Perales N, Lija J, et al. The economic and epidemiological impact of focusing voluntary medical male circumcision for HIV prevention on specific age groups and regions in Tanzania. PLoS One. 2016;11:e0153363.
10. Marseille E, Kahn JG, Beatty S, et al. Adult male circumcision in Nyanza, Kenya at scale: the cost and efficiency of alternative service delivery modes. BMC Health Serv Res. 2014;14:31.
11. Tchuenche M, Palmer E, Hate V, et al. The cost of voluntary medical male circumcision in South Africa. PLoS One. 2016;11:e0160207.
12. Larson B, Tindikahwa A, Mwidu G, et al. How much does it cost to improve access to voluntary medical male circumcision among high-risk, low-income communities in Uganda? PLoS One. 2015;10:e0119484.
13. Menon V, Gold E, Godbole R, et al. Costs and impacts of scaling up voluntary medical male circumcision in Tanzania. PLoS One. 2014;9:e83925.
14. Bollinger L, Adesina A, Forsythe S, et al. Cost drivers for voluntary medical male circumcision using primary source data from sub-Saharan Africa. PLoS One. 2014;9:e84701.
15. Fieno JV. Costing adult male circumcision in high HIV prevalence, low circumcision rate countries. AIDS Care. 2008;20:515–520.
16. Njeuhmeli E, Forsythe S, Reed J, et al. Voluntary medical male circumcision: modeling the impact and cost of expanding male circumcision for HIV prevention in eastern and southern Africa. PLoS Med. 2011;8:e1001132.
17. Awad SF, Sgaier SK, Ncube G, et al. A reevaluation of the voluntary medical male circumcision scale-up plan in Zimbabwe. PLoS One. 2015;10:e0140818.
18. Binagwaho A, Pegurri E, Muita J, et al. Male circumcision at different ages in Rwanda: a cost-effectiveness study. PLoS Med. 2010;7:e1000211.
19. Gray RH, Li X, Kigozi G, et al. The impact of male circumcision on HIV incidence and cost per infection prevented: a stochastic simulation model from Rakai, Uganda. AIDS. 2007;21:845–850.
20. Kahn JG, Marseille E, Auvert B. Cost-effectiveness of male circumcision for HIV prevention in a South African setting. PLoS Med. 2006;3:e517.
21. Bollinger LA, Stover J, Musuka G, et al. The cost and impact of male circumcision on HIV/AIDS in Botswana. J Int AIDS Soc. 2009;12:7.
22. Kripke K, Okello V, Maziya V, et al. Voluntary medical male circumcision for HIV prevention in Swaziland: modeling the impact of age targeting. PLoS One. 2016;11:e0156776.
23. Kripke K, Chen PA, Vazzano A, et al. Cost and impact of voluntary medical male circumcision in South Africa: focusing the program on specific age groups and provinces. PLoS One. 2016;11:e0157071.
24. Kripke K, Chimbwandira F, Mwandi Z, et al. Voluntary medical male circumcision for HIV prevention in Malawi: modeling the impact and cost of focusing the program by client age and geography. PLoS One. 2016;11:e0156521.
25. Kripke K, Vazzano A, Kirungi W, et al. Modeling the impact of Uganda's safe male circumcision program: implications for age and regional targeting. PLoS One. 2016;11:e0158693.
26. Kripke K, Njeuhmeli E, Samuelson J, et al. Assessing progress, impact, and next steps in rolling out voluntary medical male circumcision for HIV prevention in 14 priority countries in Eastern and Southern Africa through 2014. PLoS One. 2016;11:e0158767.
27. Chao TE, Sharma K, Mandigo M, et al. Cost-effectiveness of surgery and its policy implications for global health: a systematic review and analysis. Lancet Glob Health. 2014;2:e334–e345.
28. Ashengo TA, Hatzold K, Mahler H, et al. Voluntary medical male circumcision (VMMC) in Tanzania and Zimbabwe: service delivery intensity and modality and their influence on the age of clients. PLoS One. 2014;9:e83642.
29. Hatzold K, Mavhu W, Jasi P, et al. Barriers and motivators to voluntary medical male circumcision uptake among different age groups of men in Zimbabwe: results from a mixed methods study. PLoS One. 2014;9:e85051.
30. Evens E, Lanham M, Hart C, et al. Identifying and addressing barriers to uptake of voluntary medical male circumcision in Nyanza, Kenya among men 18-35: a qualitative study. PLoS One. 2014;9:e98221.
31. Skolnik L, Tsui S, Ashengo TA, et al. A cross-sectional study describing motivations and barriers to voluntary medical male circumcision in Lesotho. BMC Public Health. 2014;14:1119.
32. Plotkin M, Castor D, Mziray H, et al. “Man, what took you so long?” Social and individual factors affecting adult attendance at voluntary medical male circumcision services in Tanzania. Glob Health Sci Pract. 2013;1:108–116.
33. Sgaier SK, Baer J, Rutz DC, et al. Toward a systematic approach to generating demand for voluntary medical male circumcision: insights and results from field studies. Glob Health Sci Pract. 2015;3:209–229.
34. Wambura M, Mahler H, Grund JM, et al. Increasing voluntary medical male circumcision uptake among adult men in Tanzania. AIDS. 2017;31:1025–1034.
35. Njeuhmeli E, Gorgens M, Gold E, et al. Scaling up and sustaining voluntary medical male circumcision: maintaining HIV prevention benefits. Glob Health Sci Pract. 2016;4(suppl 1):S9–S17.
36. Tanzania National Bureau of Statistics. National Census Survey. 2012.
37. Tanzania Commission for AIDS (TACAIDS), Zanzibar AIDS Commission (ZAC), National Bureau of Statistics (NBS), Office of the Chief Government Statistician (OCGS), and Macro International Inc. 2008. Tanzania HIV/AIDS and Malaria Indicator Survey 2007–08. Dar es Salaam, Tanzania: TACAIDS, ZAC, NBS, OCGS, and Macro International Inc.
38. Tanzania Commission for AIDS (TACAIDS), Zanzibar AIDS Commission (ZAC), National Bureau of Statistics (NBS), Office of the Chief Government Statistician (OCGS), and ICF International 2013. Tanzania HIV/AIDS and Malaria Indicator Survey 2011–12. Dar es Salaam, Tanzania: TACAIDS, ZAC, NBS, OCGS, and ICF International.
39. Medical Stores Department. Medical Stores Department Catalogue 2013–2014.
40. OANDA. [cited 19 December 2015]. Available from:
www.oanda.com. Accessed September 1, 2017
41. Kripke K, Opuni M, Schnure M, et al. Age targeting of voluntary medical male circumcision programs using the decision makers' program planning toolkit (DMPPT) 2.0. PLoS One. 2016;11:e0156909.
42. Fox-Rushby JA, Hanson K. Calculating and presenting disability adjusted life years (DALYs) in cost-effectiveness analysis. Health Policy Plan. 2001;16:326–331.
43. World Health Organization. Global Health Observatory Data Repository. 2017.
44. Nkingwa M. HIV care and treatment services in Tanzania. Ministry of Health and Social Welfare of the United Republic of Tanzania, National AIDS Control Program. 2014.
45. Brown T, Grassly NC, Garnett G, et al. Improving projections at the country level: the UNAIDS estimation and projection package 2005. Sex Transm Infect. 2006;82(suppl 3):iii34–40.
46. Johnson LF, Mossong J, Dorrington RE, et al. Life expectancies of South African adults starting antiretroviral treatment: collaborative analysis of cohort studies. PLoS Med. 2013;10:e1001418.
47. Salomon JA, Haagsma JA, Davis A, et al. Disability weights for the global Burden of disease 2013 study. Lancet Glob Health. 2015;3:e712–e723.
48. Stover J, Andreev K, Slaymaker E, et al. Updates to the spectrum model to estimate key HIV indicators for adults and children. AIDS. 2014;28(suppl 4):S427–S434.
49. Mutabazi V, Bitega JP, Muyenzi Ngeruka L, et al. Cost analysis of adult male circumcision with the PrePex device versus surgery in Rwanda. Urol Nurs. 2014;34:303–311.
50. Kacker S, Frick KD, Quinn TC, et al. Financial implications of male circumcision scale-up for the prevention of HIV and other sexually transmitted infections in a sub-Saharan African community. Sex Transm Dis. 2013;40:559–568.
51. Tchuenche M, Hate V, McPherson D, et al. Estimating client out-of-Pocket costs for accessing voluntary medical male circumcision in South Africa. PLoS One. 2016;11:e0164147.
