The Cost of Community Outreach HIV Interventions in Thailand: A Cohort Study

Despite having an estimation of 440000 people living with HIV in 2018, the HIV epidemic in Thailand has become mature and new cases reported have been rapidly declining thanks to its successful prevention programs and scaling up of anti-retroviral therapy (ART). Thailand aimed to achieve its commitment to end the HIV epidemic by 2030 and implemented a cascade of HIV interventions through Reach-Recruit-Test-Treat-Retain (RRTTR) program.


Background
It was estimated that 480000 people were living with HIV in Thailand 2018 and about 15000 people died from AIDS-related illnesses.(1) Nevertheless, the HIV epidemic in Thailand has become mature and new cases reported have been rapidly declining since 1992 thanks to its successful prevention programs along with the scaling up of anti-retroviral therapy (ART).(1) (2) With the aim to achieve its commitment to end the AIDS epidemic by 2030, Thailand has implemented a cascade of HIV interventions through Reach-Recruit-Test-Treat-Retain (RRTTR) program to address the gaps between HIV prevention and life-long treatment system.(2) Domestic resources account for more than 85% of HIV nancing and TB responses of Thailand.Although the international funding for HIV was merely a small fraction, it mainly contributed to HIV interventions for non-Thai key-affected populations (KAPs) while majority of domestic funding focused on treatment and care.(3) (4) In 2015, Thailand launched a series of HIV preventive interventions through Reach-Recruit-Test-Treat-Retain (RRTTR) program with the support of the Global Fund which acted as a strategic short-term plan to facilitate the transition of HIV nancing to domestic funded disease response by taking into consideration that if the Global Fund investments were to cease.Under this program civil society organizations (CSOs) and non-governmental organizations (NGOs) were contracted as implementing agencies (IAs) for community-based preventive interventions in an active case nding approach to reach out KAPs, created demand for HIV counselling and testing (HCT) with adequate linkages to treatment initiation and retention through collaboration with government health service providers.(5) A total of 38 tailored HIV intervention packages covering the Reach and Recruit components of RRTTR for different KAPs were implemented in 27 provinces with high burden of HIV and TB.HIV prevention services were also integrated with TB services provision for migrant population at the community level.
(6) Since the interventions implemented under RRTTR program were new, yet there are limited empirical evidence regarding the cost, e ciency, and impact of these interventions.It is believed that e ciency gains or cost reductions could be achieved through economies of scale and scope.The former is achieved by cost savings resulting from an increase in scale of operations while the latter by cost reduction through providing jointly services rather than separately.(7)(8) This study aimed at nding evidence for Economies of Scope and Scale.
We calculated unit costs per person of tailored HIV intervention packages designed for men who have sex with men (MSM) and transgender people (TG), male sex workers (MSW), female sex workers (FSW), people who inject drugs (PWID), and migrant workers (MW) at the community level which were implemented under this program.We used these estimates to evaluate the economies of scale and scope.The results from this study aimed to provide strategic information for policy makers on setting priorities and optimizing e ciency in resources used for HIV programs within a sustainability nancing mechanism when scaling-up.

Methodology
Research Setting: This study focused on community-based HIV interventions implemented by Non-Governmental Organizations (NGOs) as Implementing agencies (IAs) under RRTTR program in the provinces highly affected by HIV epidemic in Thailand (as shown in Figure 1).This was a cohort study, and the data were collected covering a period of 21 months (from January 2015 to September 2016).Peer-led community outreach activities were organized by outreach workers and peer educators who were recruited, trained, and supervised by IAs.Outreach sessions were often provided at a meeting venue in the community closed to where KAPs live or work i.e., beauty salons, parks, stadium, universities, local markets, department stores, entertainment venue, public transportation stations, campaign events, beauty contests, and drop-in-centers (DiCs) etc. KAPs were encouraged for HIV testing at the outreach session and those who were willing to get tested will be referred to receive service for Voluntary Con dential Counselling and Testing (VCCT) in collaboration with local government health service providers.VCCT services were often provided at the local public hospital/ health service center, mobile clinic, and community-based HIV testing centers (or DiCs).Outreach workers of SRs provided pre-test and post-test counselling, while HIV testing was done by professional health staffs (nurses and medical technicians) from local government hospitals using pin-prick blood testing with Same Day Result (SDR).It is worth noted that HIV testing and provision of ARV were delivered at the government health facilities, and the cost of which were incurred under the public health system.

Interventions and Program
Program Outputs: The numbers of KAPs reached out by community-based HIV intervention in the provincial level were de ned as the program outputs of in this study.This number included KAPs reached by outreach activities (through individual or group session) and social media (clients registered with unique identi cation code (UIC) using on-line applications) who received a tailored package of HIV service, participated in behavioral change program, received IEC materials regarding HIV and STI prevention, commodities (such as condom, lubricants, or clean needles & syringes), counselling services and were referred for HIV/STI testing (for those who were willing to undertake the test).The data were collected from IAs routine reports covering reporting period of January 2015 to September 2016.

Costing Methodology
This study used top-down costing approach and focuses on health care providers perspective whereas interventions were implemented by IAs at the community level.Patients' costs were not collected and included in this study.We collected data from expenditures records and progress update annual reports of IAs covering expenditures from 1 st January 2015 to 30 th September 2016.The data were keyed into excel sheets, pivotal tables Indirect Program Cost (Overhead cost): At above site level, overhead costs (include program management, health system strengthening, monitoring and evaluation etc.) were calculated and allocated accordingly to all 15 IAs who received funding through primary recipient (PR) agency who received funding through the Global Fund program.
Unit Cost: We calculated unit cost per targeted KAP reached by a community-based HIV intervention or unit cost per person to receive service from a community-based HIV intervention.The unit cost was de ned as total cost divided by total program outputs.The unit cost at IAs level was rst calculated by allocating all direct program cost (operational cost) at site level while excluding overhead cost.The unit costs including both direct program cost and overhead cost from above IAs level were then calculated.The overhead costs were allocated proportionally to IAs' individual intervention by using allocation strategies based on each intervention program outputs.USD per person reached for PWID (Table 3).The lowest unit cost per person reached was 10.8 USD for Migrant in Samut Sakhon province while the highest was 324.5 per person reached for PWID in Samut Prakan (Table 2).
After inclusion of overhead cost (Indirect Program Cost) at above site level into calculation, average unit cost was 29.7 USD per person reached for MSM/TG, 33.1 USD per person reached for MW, 31.9USD per person reached for FSW, 36.3USD per person reached for MSW, and 179.1 USD per person reached for PWID (Table 3).Migrant intervention in Samut Sakhon remained as the intervention with the lowest unit cost (21.6USD per person reached) despite about 50% increased after adding up overhead cost.PWID intervention in Samut Prakan remained the one with the highest unit cost (340.5USD per person reached) with merely 5% increase after adding up the overhead cost (Table 2).
The unit cost per person to received tailored HIV intervention package were put in scattered plots with both polynomial and linear trend lines to identify potential evidence of economies of scales.Interestingly, the result shows the presence of economies of scales.This indicates that the higher number of KAPs reached by community-based HIV interventions, the lower they will cost.

Discussion
Thailand has achieved the rst and third 90s of UNAIDS' 90-90-90 targets in 2018 whereas more than 90% of people living with HIV in Thailand were aware of their status and of those 80% were on treatment thanks to its successful HIV prevention, testing and treatment programs.(1) As the country HIV epidemic is mature and concentrated in KAPs, its approach of targeting population who were at the highest risk in highly priority geographical sites appeared to be both feasible and effective.We found e ciency increased across full range of scales examined among HIV interventions led by community based IAs in Thailand.The cost per key outputs (cost per person reached) and scale are found to be correlated negatively (Fig. 2).Despite the presence of up-turn point for cost per person reached (Fig. 2), it is merely due to one data point.And it is noteworthy that regression trends were downward sloping in both cases with simple liner function.Therefore, even though we only observed unit costs of interventions for a period of 21 months, this study suggested that community outreach HIV interventions will only become less costly overtime after scaling up.(8) (17) In addition, we calculated unit cost by allocating overhead costs using different cost allocation ratios (i.e., according to the size of targeted KAPs in the implementing province and the amount of grants/funding that IAs received to implement interventions) as sensitivity analysis and the trends are shown in Fig. 3.
Nevertheless, this research study is subject to some potential limitations that could be addressed in future studies.All interventions led by CSOs under this program was initially intended to be lasted for two years.However, it was later extended for another two years.At the time of data collection for this study, we were only able to collect data covered the rst 21 months periods after its initial implementation.This study applies only the cost of HIV services provided at the community level which does not include the cost incurred from patients' perspective.Due to limitations of the secondary data, this study does not investigate potential contextual factors that might act as cost drivers for interventions and discuss whether the same results, or better results might have been achieved with greater e ciency.The cost and e ciency of TB prevention services integrated with HIV services to MW should be evaluated and discussed along with HIV services in further studies.In addition, other components of RRTTR (i.e., testing and provision of treatment under conventional public health system etc.) should also be evaluated in future studies.

Conclusion
This study suggested that community outreach HIV interventions led by CSOs in Thailand will only become less costly as they are scaled up overtime.The integration of HIV prevention services with TB prevention services may also reduce costs.Further studies are required to follow up with this ongoing intervention programs in depth to determine how to improve e ciency of HIV prevention services in Thailand.

Figures
Page 16/18 Activities: Depends on type of targeted KAPs, each intervention package included a combination of 1) behavioral change interventions through outreach activities, harm reduction intervention (for PWID only), distribution of commodities i.e., condoms, lubricants, needles, syringes and behavior change communication (BCC) & IEC materials etc., 2) HIV testing and counselling, referral to health facilities for HIV testing and/or STI testing, TB screening and testing (Migrant population only), 3) referral to treatment, providing care & support for those who tested positive, and following up on those who are on ART for treatment adherence, and 4) community health system strengthening.
were constructed to summarize and calculate unit cost.Shared costs between two or more interventions incurred by the same IAs were allocated proportionally by using different allocation strategies based on type of activities as they were described on records or by looking into program outputs yielded by interventions.Direct Program Cost (IAs Operational Cost): All cost incurred at site (provincial) level or IAs level were de ned as Direct Program Cost.Costs were disaggregated into different level, rstly by program areas then by interventions, activities, and cost inputs.Shared cost across different KAPs or program area were allocated accordingly using allocation criterion based on their program outputs.The cost included both recurrent and capital costs utilized by IAs which were reported into 13 cost inputs; Human Resources (HR), Travel related costs (TRC), External Professional services (EPS), Health Products -Pharmaceutical Products (HPPP), Health Products -Non-Pharmaceuticals (HPNP), Health Products -Equipment (HPE) including HIV test kit, Procurement and Supply-Chain Management costs (PSM), Infrastructure (INF), Nonhealth equipment (NHE), Communication Material and Publications (CMP), Indirect and Overhead Costs, and Living support to client/ target population (LSCTP).

Table 1 :
Characteristics of community outreach HIV interventions in this study tested through HIV counselling and testing services.The highest number of HIV tests referred by a single intervention was achieved in Chiang Mai whereas 4179 MSM/TG were tested followed by the intervention in Bangkok of 4076 MSM/TG tested.The lowest number of tested was recorded in Samut Prakan intervention for PWID which has only 20 of PWID tested for HIV.In case of TB interventions, a total of 56099 migrants were reached and 878 TB suspects were identi ed and referred for TB con rmation to local hospitals for further arrangement.CostThe total cost of all 45 community-based outreach interventions was 9473316 USD from January 2015 to September 2016 whereas 82.7% (7837234 USD) were spent for HIV interventions and the rest of 12.3% (1636082 USD) for TB interventions.It comprised Direct Program Cost (Operational Cost) 6686157 USD (70.6%) by Implementing Agencies (IAs) at the site level and Indirect Program Cost or Overhead Cost 2787160 USD (29.4%) incurred at above IAs level (Table 3).HIV interventions cost could be further divided into direct program cost 5576092 USD (71.1%) at IAs level and overhead cost or Indirect Program Cost 2261142 USD (28.9%) above IAs level.Without including overhead costs, the cost for 14 MSM/TG interventions was 1775546 USD, 4 MSW interventions 668854 USD, 1 FSW intervention 294562 USD, 14 PWID interventions 1958613 USD, and 6 MW interventions 878518 USD, respectively.After adding up overhead costs, the total cost for MSM interventions was 2929664 USD, MSW 918065 USD, FSW 424937 USD, PWID 2186047 USD, and MW 1378521 USD, respectively (Table3).
878ResultsThis study identi ed 45 community-based HIV/TB interventions(38 HIV interventions and 8 TB interventions) delivered by IAs to KAPs in 27 provinces of Thailand.Among 38 HIV interventions, there were 14 interventions targeting MSM/TG population, 12 harm reduction interventions for PWID and their partners, 7 interventions for migrant workers, and 5 interventions for sex workers (4 MSW and 1 FSW) and their clients.All TB interventions program were focused on migrant workers (Table 1).otherinterventions focus on MSW/FSW (7714), MSM/TG (7040) and MW (5948).The prevention program for MSM/TG in Bangkok alone contributed to the highest number in terms of targets reached by a single intervention program having 25438 MSM/TG received services, while prevention program for PWID in Satun province achieved the lowest targets by 161 PWID received services (Table 2).Average number of targets reached by any intervention was 5025 KAPs.Among all KAPs reached out by interventions, 32309 werereferred to Note: KAPs (Key Affected Population), IAs (Implementing Agencies), IPC (Indirect Program Cost), Excl (Exclude), Incl (Include), FSW (Female sex workers), MSM/TG (Men who have sex with men/ Transgender), MSW (Male Sex Worker), PWID (People who inject drug).The average unit cost (with the inclusion of program cost while excluding overhead cost or indirect program cost) for a KAP to received HIV services delivered by IAs was 18 USD per person reached for MSM/TG, 21.1 USD per person reached for MW, 22.1 USD per person reached for FSW, 26.5 USD per person reached for MSW, and 161

Table 3
Cost, inputs, outputs, and average unit cost of HIV interventions among different key affective populations (7) results in this study show that communitybased outreached interventions led by NGOs and CSOs were critical in Thailand's new approach for ghting against the epidemic.Community-based HIV interventions has reached out a signi cant number of KAPs in high priority geographical sites and brought them into HIV prevention and life-long treatment system under RRTTR program.We found that unit cost per person reached by HIV intervention package varies across targeted populations and geographical areas in Thailand.Average unit cost per person of an intervention package for MSM/TG was the lowest followed by interventions for MW, MSW, FSW, and PWID, respectively.In Thailand, HIV prevention services led by NGOs in partnership with public health facilities for MSM/TG and other KAPs were available countrywide prior to RRTTR program especially in large cities and tourist destinations where MSM/TGs were concentrated.(2)(5)(9)(10) Some IAs were, therefore, able to adapt new HIV intervention programs into their existing programs rather than creating new implementing structures and plans.As a result, this might have reduced start-up cost and increased the e ciency of interventions.(7)Average unit cost of an intervention for PWID was found ve folds higher compare to other HIV interventions.In addition, this study found very low uptake among HIV intervention services for PWID which suggested that intervention for PWID were less viable compared to interventions focusing on other KAPs.There were studies indicated that policy and legal constraints has displaced many PWID access to lifesaving healthcare services in Thailand.Fear of disclosure, Stigma and discrimination in healthcare settings and concerns over con dentiality were among main factors which causes low demand in HIV services for PWID.(11)(12) (13) Therefore, the uptake of services was often low and more intensive service delivery with comprehensive package was needed in reaching out these PWID population.Unit costs for HIV intervention for MW in Samut Prakan and Samut Sakhon were the lowest among all interventions.It is noteworthy that HIV and TB prevention services were integrated and implemented simultaneously to MW by the same IAs under RRTTR program.There is a possible cause of economy of scope observed here whereas IAs were able to distribute xed cost for reaching out more cases from both HIV and TB interventions.In addition, sharing of services could reduce input prices and task shifting could lead to greater e ciency through integrated services.(7)(14) However, there are some contextual factors should be considered when we interpret the trend of unit costs of HIV interventions which include IAs' program maturity, geographic structural factors (i.e., how di cult or ease to physically access the location), quality of services provided and e ciency of service delivery etc.For instance, program immaturity is likely associated with high start-up cost for new service or location and insu cient service delivery process.And higher quality services with relatively more comprehensive care or package is likely to result in higher unit cost and low targeted reached.(15)(16)