68677 Cost -Effectiveness of Harm Reduction Interventions in Guangxi Zhuang Autonomous Region, China Human Development Unit East Asia and Pacific Region The World Bank May 2007 Document of the World Bank Draft report: 05/08/2012 Acknowledgements This note has been prepared by Emiko Masaki (Asian Development Bank); Jie Chen, Qiuying Zhu, Xiaomin Chen, Yi Chen, and M. Lu (Guangxi CDC); Elliot Marseille (Consultant, Health Strategies International); and J. Khan (Consultant, University of California, San Francisco);. The work was done under the supervision of Ana Revenga, Task Team Leader for the China AIDS AAA Program, and benefited from inputs from Mead Over, Shiyong Wang and David Wilson (World Bank). 2 Draft report: 05/08/2012 Table of Contents Acknowledgements ............................................................................................................. 2 Table of Contents ................................................................................................................ 3 I. Background ...................................................................................................................... 5 II. Objectives of the study ................................................................................................... 6 III. Effectiveness of Harm Reduction Programs ................................................................. 8 IV. Harm Reduction Programs in Guangxi....................................................................... 10 V. Methods ........................................................................................................................ 12 Data Collection ............................................................................................................. 13 Cost effectiveness ......................................................................................................... 14 VI. Key findings................................................................................................................ 17 Methadone Maintenance Treatment Programs ............................................................. 17 Needle Exchange Programs .......................................................................................... 18 Sex Worker programs ................................................................................................... 21 Scale and unit cost ........................................................................................................ 22 Cost composition .......................................................................................................... 23 Cost-effectiveness ......................................................................................................... 24 High Risk Prefectures ............................................................................................... 25 Low Risk Prefectures ................................................................................................ 28 VII. Discussion ................................................................................................................. 32 References ......................................................................................................................... 34 3 Draft report: 05/08/2012 Executive Summary Issues: China‟s epidemic is mostly concentrated in high-risk populations, especially injection drug users and sex workers. Implementation of harm reduction programs in China has expanded rapidly in the past few years. The government resources devoted to AIDS nearly tripled between 2004 and 2005, and have increased sixty-fold relative to their level in 2000. This level of commitment warrants a careful examination of the efficiency with which these programs operate. As China‟s AIDS program is scaled up, issues concerning the allocation and effectiveness of resource use are rapidly gaining importance. Description: The objective of the study is to present the cost and cost-effectiveness of harm reduction interventions in Guangxi Province to help policymakers and program staff to mount a well-targeted, cost-effective, evidence-based HIV/AIDS prevention response. The study consists of two parts. The first part is to compile and analyze the resources used and their costs for delivering harm reduction services. The second part is to model the epidemic impact of behavioral changes produced by these HIV prevention interventions. The study examined three types of harm reduction activities: methadone maintenance treatment (MMT); needle exchange programs (NEPs); and a program for sex workers. The data were collected from six intervention sites in Guangxi Province during the period of December 2005 to April 2006. Conclusions: Economic costs of MMT services ranges from US$29,599 to US$73,639, with the unit costs estimated at US$19.0-US$49.4 per client-month. Unit cost per needle distributed is low as US$0.1 and the unit cost per condom distributed is $0.45. There are some large variations in prevention costs, outputs and efficiency across programs and services. Understanding the determinants of these variations may yield insights that could be translated into higher levels of efficiency. Personnel and recurring services account for a large portion of the total costs, nearly 40% for NEPs and 56% for MMT programs. Thus, identifying ways of increasing productivity may be promising avenues for enhancing efficiency. The cost-effectiveness analysis, based on an epidemic model liking with behavioral parameters, found the needle exchange program to be the most cost- effective of three interventions examined in this study. 4 Draft report: 05/08/2012 I. Background HIV prevention cost and cost-effectiveness studies are particularly important in the current era of increased funding for HIV when the prospects for „scaling-up‟ transitions from rhetoric to reality. The need to spend this money well exceeds the normal obligation to spend public money competently: the lives of millions literally depend upon how efficiently available funds are allocated. Refinement and extension of the existing cost and cost-effectiveness knowledge base supports the targeting of prevention funds where they can have the greatest impact on the epidemic[1]. In China, with its large population and potential for a significant epidemic, efficient spending for HIV prevention could pay particularly high dividends. Until the early 2000s, HIV/AIDS incidence in China remained low, even among vulnerable communities. However, following major economic and social changes, HIV/AIDS prevalence rose in specific sub-populations and provinces. China now recognizes HIV/AIDS as a major development threat and has mobilized increased national and international resources to contain the epidemic. Both international and national resources to combat HIV/AIDS in China have grown rapidly in recent years. Resources are no longer the primary constraint to effective AIDS action in China. China‟s major challenge is to increase its capacity to use these resources effectively. To that end, several development partners are supporting national and provincial capacity building. China‟s epidemic is currently believed to be concentrated in high-risk populations. We believe that the provision of cost-effectiveness information early in the scale-up process when program plans are relatively fluid, can help policy makers and program staff to mount a well-targeted, cost-effective, evidence-based HIV/AIDS prevention response. 5 Draft report: 05/08/2012 II. Objectives of the study The over-arching objective of this study is to understand the relationships between spending on HIV prevention programs and desired prevention outputs such as needles exchanged, condoms distributed, and counseling sessions conducted and outcomes, i.e., cases of new HIV infections prevented. When combined with the epidemiological modeling of the impact of behavioral changes on the HIV epidemic, this cost information will provide the foundation for cost-effectiveness analyses that are needed to help guide the spending of HIV prevention funds. Because of its budgetary implications during the current „scale-up‟ stage, we are particularly interested in understanding how unit costs vary by scale. In addition to our primary objective of addressing cost-effectiveness and scale questions, we believe that this study can also achieve an important secondary goal, that of documenting the range and degree of variation in:  Program scale and cost  Cost composition, e.g., personnel, recurring goods, capital goods.  Unit program costs, i.e., cost per unit of output. While the cost-effectiveness data will assist high-level planners to allocate funds among program types, these comparative site-level data can assist program managers to assess the efficiency of their individual projects, and begin to identify ways to enhance that efficiency. Experience from other projects suggests that these monitoring data can sometimes be useful adjuncts to the budgeting process, and help managers to monitor monthly program costs and outputs. To that end, we will send back to each site the final report that includes the cost and output data compiled from all the sites. We also envision a half-day workshop 1 in which study results are presented to the relevant program managers and they in turn have an opportunity to comment on these data and, more importantly, to share their views on the obstacles and facilitators of service delivery efficiency. The study is conducted in two parts. The first part is to compile and analyze the resources used and their costs for delivering harm reduction services in Guangxi Province. It also carefully documents the level and type of services these expenditures can support in response to the needs of local communities. The second part is to model the impact of behavioral changes produced by these HIV prevention interventions on the epidemic. In a later stage, these two parts of the study are combined and analyzed to derive measures of cost-effectiveness for HIV prevention interventions in Guangxi. The recent surveillance results suggest that most new infections are concentrated among injection drug users (IDUs) and sex workers (SWs) in Guangxi [2, 3]. We therefore selected harm reduction programs for IDUs and for SWs as the focus of this study. During 2005-2006, the Guangxi CDC and the World Bank team undertook field work to cost the harm reduction 1 The tentative agenda for the proposed workshop is outlined in Annex (D) 6 Draft report: 05/08/2012 activities at three methadone maintenance programs, two needle exchange programs and one SW intervention sites in five counties and cities in Guangxi. The data reported here are derived from that costing exercise. While the number of sites for which data are now available is too small to draw firm conclusions, we are able to offer some preliminary findings. We have also established the analytic foundation on which findings that are more definitive can rest following receipt of data from additional sites in Guangxi and possibly from other provinces. 7 Draft report: 05/08/2012 III. Effectiveness of Harm Reduction Programs Approaches to reduce HIV risk among IDUs include needle exchange programs, methadone maintenance treatment, detoxification programs, voluntary counseling and testing, and community-level outreach programs to change risk behaviors. Effectiveness and cost of the interventions vary by many factors including location, service delivery modes, size of risk groups, risk behaviors, etc. Growing evidence of intervention effectiveness has become available and been documented. Measuring the cost-effectiveness presupposes estimation of the magnitude of changes in well-defined measures of effectiveness. Since making such estimation falls outside the purview of the current study, we began with a search of the available peer reviewed and „grey‟ literature on the effectiveness, costs, and cost effectiveness of methadone maintenance, needle exchange and sex worker services in preventing HIV transmission but found no studies specific to China. Thus, we turned to data from other countries. We found a meta-analysis of 47 studies of needle exchange program effectiveness in reducing HIV risk in the United States, Europe, and Australia[4]. The observation period for studies in this review ranged from 1986 to 1997. The reported mean reduction in the prevalence of needle sharing was 18%. This was derived from the reported “R� coefficients, assuming a 50%/50% distribution of participants allocated to the treatment and control conditions and observed rates of needle-sharing between 20% and 80%. Other reviews[5, 6] found a wide but generally higher range in risk reductions, 17 - 70%, so this meta-analysis result may be conservative. With regard to methadone maintenance, a Cochrane review evaluated 28 studies of 7,900 participants published between 1989 and 2003 of the relationship of such services to HIV risk [7, 8]. Because of heterogeneous methods, it was not possible to derive summary measures of effectiveness. Nevertheless, the data strongly support the conclusion that methadone substitution dramatically reduces both needle sharing and risky sexual behavior. For example, of the four studies that reported on multiple partners or the exchange of sex for drugs or money, three reported a significant reduction in these behaviors. Four of six studies reported statistically significant reductions in episodes of unprotected sex. Of the two studies with nonsignificant reductions, one found that fewer participants reported unprotected sex at follow-up and in the other, most participants reported unprotected sex (84% at baseline, 88% at follow-up). All of the four studies that reported on the effect of methadone maintenance on seroconversion found that participation reduced HIV incidence. While none of the studies reviewed were conducted in China, they provide substantial evidence that MMT can reduce HIV transmission. The eight studies of sex worker program from seven countries found relative reductions in risk from 0.17 to 0.90, with the plurality of results between 0.50 and 0.60 [9]. The highest and lowest reductions both occurred in sex contacts with no condom. HIV incidence was examined in three studies, and decreased by 0.62, 0.68, and “less than expected.� Reduction in STI incidence was 0.33 – 0.77 in two studies. The randomized control trial had a relatively low risk reduction (0.24). However, this intervention was 8 Draft report: 05/08/2012 low intensity (just distributing condoms), whereas the interventions with highest behavioral risk reduction [10, 11] had more varied, frequent, and intensive interactions with intervention participants. 9 Draft report: 05/08/2012 IV. Harm Reduction Programs in Guangxi The first pilot Methadone Maintenance Treatment (MMT) program in Guangxi was started in Nanning, the capital of Guangxi province, in November 2003. The following year in 2004, three additional pilot MMT projects were initiated in Liuzhou, Huchi, and Wuzhou. Since the beginning of the program, these four facilities have registered over 1,100 IDUs. As shown in Table 1, the number of MMT programs increased rapidly in the period 2004 - 2006, and the number are expected to grow at a much faster rate in coming years. As of May 2006, 11 new MMT sites received approval and are to begin operation later this year. An additional 8 sites have submitted application and are awaiting approval. Table 1. Expansion of harm reduction interventions (on-going) in Guangxi ~2001 2002 2003 2004 2005 2006 2007 Total* MMT 1 4 4 (15)** 25 4 NEP 4 5 11 29 31 31 31 SW 2 2 2 24 30 30 (50)** >50 30 Source: Guangxi Center for HIV/AIDS Prevention and Control * as of June 2006 ** the figures are as of June 2006, and the figures in ( ) are the number of sites that are planned by end of 2006 In 1999 the national and local CDC started the first pilot needle exchange programs in Liuzhai and Tiandong, followed by Yongning and Ningming in 2002. An additional 25 sites have been selected for NEP expansion. There are 31 sites currently providing needle exchange services. There are variations in the mix of interventions provided in each NEP site; however, most of the NEP programs in Guangxi follow the general protocols developed by Ministry of Health. A typical needle exchange program includes the harm reduction activities of: (a) collection and safe disposal of used needles and syringes; (b) social marketing of new needles and syringes, including direct distribution and redemption of pharmacy/clinic vouchers; (c) community education; and (d) support of drug use cessation. More than 50,000 IDUs are currently registered with the Guangxi public security office and the actual number of IDUs is estimated to be several times higher. As shown in Figure 1, the IDU epidemic expanded dramatically in late 1990s and leveled off in the recent years. However, HIV prevalence rates in some areas remain very high, more than 50 percent in Wuzhou and 30 percent in Pingxiang. The growth of the number of harm reduction programs, particularly SW programs and NEPs lagged the epidemic by several years but expanded rapidly starting in 2003 – 2004. 10 Draft report: 05/08/2012 Figure 1. HIV prevalence and harm reduction programs in Guangxi HIV prevalence among IDUs and harm reduction programs in Guangxi 0.6 35 0.5 30 HIV prevalence rate 25 0.4 20 0.3 15 0.2 10 0.1 5 0 0 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 NEP SW MMT Wuzhou Liuzhou Nanning Ningming Pingxiang 11 Draft report: 05/08/2012 V. Methods We administered survey instruments to gather information on program costs and outputs at three MMT, two NEP and one SW site. Our data collection methods were adapted from those developed in the NIH-funded “PANCEA� project that evaluated the cost and efficiency of 200 HIV prevention programs in five countries, Uganda, South Africa, India Mexico and Russia [12]. Data were gathered, tabulated and summarized using Excel® spreadsheets. We selected six intervention sites with three types of harm reduction activities: MMT, needle exchange programs, and a program for sex workers. During the period of December 2005 to April 2006, we collected program costs, outputs, and epidemiological data at 6 sites in five cities and counties. We selected Nanning, Liuzhou2, and Wuzhou for the MMT intervention and Ningming and Wuzhou for needle exchange programs. Pingxiang was selected for a sex workers program. These sites were selected as a convenience sample based on the availability of data and the willingness of program managers to participate. These sites may therefore not be statistically representative of all harm reduction sites in Guangxi. As we add more sites to the study sample, the likelihood and magnitude of potential bias will diminish. Locations of these intervention sites are shown in the below map (Figure 2). Brief descriptions of the harm reduction programs at the each study site are included in Annex (A). Figure 2. Map of harm reduction programs in Guangxi # Liuzhou # # # c # # # # # # # c # # Nanning # # c # # Wuzhou Intervention Site # # c MMT site # # # # NEP site # # # # Study site # # Pingxiang # c # # Ningming 2 The data from Liuzhou was not included in the analysis and results presented in this report due to the problem of the incomplete data. Thus the results presented are based on the data from 5 intervention sites. 12 Draft report: 05/08/2012 Data Collection Outputs. The study team gathered data on key service outputs for each intervention type. The “key� outputs are those with the greatest potential to reduce HIV transmission and are the project‟s primary raison d‟etre. For MMT programs, we defined a single key output: the number of person-months of methadone maintenance. This output is the main basis for assessing program efficiency and for predicting reductions in risky behaviors and HIV infections. For the NEPs, we gathered data on two key outputs. These were the number of individuals served and the number of needles or syringes exchanged. Costs categories. The cost of program resources was divided into five categories: personnel compensation including benefits, recurrent goods, capital goods, recurrent services, and facility space. We report economic costs. Financial costs are payments actually made by the program, and exclude the value of donated services and goods and of subsidies. Economic costing attaches a value to donated inputs, and is the approach more appropriate for assessing the costs of scaling up prevention. To calculate economic costs, we obtained estimates of the market value of donated or heavily subsidized inputs for personnel and other items. Cost data were collected by month for 12 months at each site. The data are reported over the period December 2005 to May 2006. Personnel compensation was computed for all personnel contributing to the work of the program, including clinicians, technicians, and support personnel. If a particular person was contributing part time to the program, only that proportion of salary was included in the salary cost. A similar approach was used with other shared resources. Recurrent goods included both medical and non-medical supplies with an expected life of under 12 months. Capital goods include such durable items as furniture, electrical fixtures, air conditioner, lab equipment, and computers. The life of the capital goods was assumed to be six years, and therefore, one-sixth of the cost was allocated to each fiscal year. Recurrent services include training of staff, cleaning and building maintenance, utilities, photocopying, printing, postage and courier, and some miscellaneous items. Building costs refers to the rent that was actually paid; or if space was donated or subsidized, to the rent for an equivalent space at that location and time. Allocating costs. Many facilities that provide services to IDUs, such as hospitals and clinics, provide other services that share personnel and other inputs. We explicitly allocated inputs to the intervention. Personnel allocations were based on the proportion of time spent on tasks relevant to the HIV prevention intervention under study or of space dedicated to the intervention. A similar approach was used for other inputs. Quality assurance. Rigorous on-site assessment and careful review process increased the likelihood that the collected data were valid. The study team first reviewed the collected data on-site during the field work, and then conducted an additional in-depth review of all data gathered from each site. We scrutinized each response to determine its “face validity� and consistency with other information gathered at the same program. Any missing or seemingly incorrect data were itemized in a structured memo and sent to the data collection sites. The data were revised as necessary and reviewed a second time by 13 Draft report: 05/08/2012 the study team. This process of review and clarification was repeated until the study team was satisfied with the accuracy and completeness of the data. Data sources. Cost and output data were drawn primarily from financial records and project reports, including those overseen by outside reviewers such as funding agencies. When such documents were available, we used them. However, available written records were often incomplete or imperfectly matched to the survey instruments. Thus, records often required interpretation by respondents to provide needed data, or respondent recall of key facts. Without inclusion of less formal data sources, many programs would have been excluded. We were willing to sacrifice some precision for a much more inclusive portrayal of the universe of HIV prevention programs. Data collection visits. The data collection team consisted of two data collectors. The visits require 1 - 2 full days of the team at each site; “full� visits required a second visit of similar duration. Often data collection was done in a greater number of partial days or spread over several weeks, to accommodate respondent schedules. Along with data cleaning and data entry performed off-site, each study site thus required 10 to 15 person- days of work. Survey Instruments. Output data were captured in an Excel spreadsheet that allows for direct by-month entry of the relevant data. This was a straightforward process for MMT and NE programs because they have standardized output types, e.g. number of client- months of MMT services. For the SW programs that typically have a more heterogeneous set of activities, (e.g., peer counseling, condom distribution, STI services) we asked respondents to define the activities of their program rather than attempting to impose a pre-determine list. Cost instruments were extensively adapted by PANCEA from templates produced by UNAIDS[13]. Survey instruments used for the study are attached in Annex (B) and original instruments are available from http://hivinsite.ucsf.edu/InSite?page=pancea. Analysis. For this report, we calculated the cost by category and the key outputs for each program separately and in aggregate. We then calculated the unit costs for each program and an average unit cost across each program type, i.e., MMT, NEP and SW. We generated scatter plots of efficiency against scale and fit regression trend lines to these data. Finally, we compared the efficiency of the Chinese NEP against the efficiency of six Russian NEPs studied by the PANCEA project. Cost effectiveness We undertook an exploratory modeling of cost effectiveness by combining using an existing epidemiological and cost-effectiveness model for HIV prevention with the new cost and output data collected as part of the current study and presented in this report. 14 Draft report: 05/08/2012 Modeling tools For the epidemiological modeling work, we adopt the Multiple-Intervention Multiple- Group Epidemic and Cost-Effectiveness Model developed by James G. Kahn and others from the PANCEA project. A detailed description of the model and its methodology are presented in Annex C. Modeling scenarios Guangxi Province with a population of 44 million is both large and epidemiologically diverse. For example, HIV prevalence varies from area to area by a factor of 10 in some risk groups. Modeling Guangxi‟s epidemic and the effect of risk reductions on that epidemic thus suggests that we portray multiple community scenarios. Since these risk profiles are linked to geographic areas we modeled three geographically-defined scenarios conceived as high, medium and low risk geographic areas. Guangxi CDC has already divided the province into three sub-regions by risk group for IDUs. The prevalence in the low risk sub-region is <10 %; the medium risk region has a prevalence of 15% - 40% and the high risk region has a prevalence of 40% - 50%. The modeling results of high and low risk scenarios were presented in the section of Key Findings. Data for model specification The effectiveness modeling, required demographic, epidemiological, and risk behavioral data by risk region and reasonable estimates of the initial distribution of risk groups in the population by sex, and the percent of various risk groups, such as IDUs, SWs, and clients. In addition, portraying the scenarios required model input parameter values reflecting the three risk levels defined by the geographic areas.  Size of the general population  Size of the high risk populations (IDUs, SWs)  HIV prevalence in IDUs; in non-IDU SWS; and in the general population  HIV risk factors including o Percent married by risk group type (e.g., SW-IDUs, SW-non-IDUs, non- SW-IDUs, both male mad female o Frequency of needle sharing and bleaching o Condom use patterns by type of partners (e.g., regular, casual souse, client of sex worker) o Frequency of sex acts by partner type When available, these data were supplied to us by the Guangxi CDC and were derived from a variety of behavioral and epidemiological surveys. However, in many cases not all of the ideal data items were available. In these instances, we imputed estimates from the data that are available. We derived estimates of the effectiveness of the interventions from the published literature. Following assembly of these data, we compared the epidemic impact and cost effectiveness of spending $10 million over five years on methadone maintenance or on needle exchange or $5 million over five years on sex worker programs (since $10 million could not easily be absorbed by sex worker programs). We projected the epidemic impact 15 Draft report: 05/08/2012 of these five-year spending scenarios to both 10 and 20 years. We also examined the consequences of spending $40 million over 20 years on needle exchange programs. Due mainly to the gaps in data available on the timeline of this study, the results as presented in the “Key Findings� section should be considered provisional. Nevertheless, they do demonstrate the types of policy-relevant cost-effectiveness determinations that can be made using the epidemiological and cost-effectiveness model and the cost and output data developed in this report. 16 Draft report: 05/08/2012 VI. Key findings Table 2 summarizes the core findings of our study. These are the annual costs, the level of key outputs produced for each of the program types, and the unit costs per output. In the following three sections, we present results that are more detailed for each of the harm reduction program types. Table 2. Summary Results USD Six HIV prevention programs in Guangxi Costs, Outputs and Efficiency Annual economic Site Annual outputs Economic cost per output costs Methadone maintenance programs1 Nanning 73,693 1,493 49.4 per client-month Wuzhou 29,599 1,558 19.0 per client-month Ave, all MMT sites 51,646 1,526 33.8 per client-month Needle exchange programs2 Ningming 9,149 90,773 0.10 per needle/syringe Wuzhou 19,271 157,720 0.12 per needle/syringe Ave, all NEP sites 14,210 124,247 0.11 per needle/syringe Interventions for commercial sex workers3 Pingxiang 30,914 2,515 12.3 per client reached 1. Client-month of services delivered 2. Needles/syringes distributed 3. Clients reached Methadone Maintenance Treatment Programs As reported in Table 2 economic costs of MMT services for the fiscal year studied ranges from US$29,599 to US$73,693. Unit costs per client-month of MMT services were estimated at US$19.0-US$49.4 with an average cost of US$33.8 per client-month. Figure 2 below displays the relative contribution of each of five expenditure categories to total costs of MMT services in Nanning. Personnel accounts for about one-third of the total. Personnel and building rent together account for about 70% of the total. Capital costs assume an average six-year time to replacement for each item and constitute less than one percent of total costs. Building cost calculations reflect an apportionment of total building costs to MMT activities based on the number of employees involved in MMT compared with all employees working in the building. 17 Draft report: 05/08/2012 Figure 2. Cost of MMT services by cost category Building 16.2% Building 33.7% Personnel 35.0% Personnel 35.0% Capital goods Capital MMT in Nanning MMT in Wuzhou 5.2% goods Recurring 0.6% Recurring goods Recurring services Recurring 21.8% goods 20.9% services 9.8% 21.9% Total costs = US$ 29,599 Total costs = US$ 73,693 As shown in Figure 3, MMT services expanded rapidly since beginning of the program in Nanning, which increased by 66% over the data collection period. The graph also indicates some frustration in output services. We presume that this is partly due to delays in obtaining sufficient supplies of methadone. Figure 3. Program outputs of MMT services 1000 6000 900 Clients (Nanning) 5000 800 Clients (Wuzhou) 700 Methadone (Nanning) 4000 MMT clients Methadone (Wuzhou) 600 Methadone 500 3000 400 2000 300 200 1000 100 0 0 Nov Nov Nov Jan Jun Jan Jun Jan Jun Jul Jul Jul Mar Mar Mar Apr Apr Apr Feb Aug Sep Feb Aug Sep Feb Aug Sep Oct Oct Oct May May May Dec Dec Dec 2004 2005 2006 Needle Exchange Programs Table 3 summarizes annual costs, outputs and unit costs per output of needle exchange programs in our study sites. Economic costs of needle exchange services range from a low of $9,149 to a high of $19,271 by the different sites with mean economic costs of $14,210. Unit cost per needle distributed is very low around US$0.1. 18 Draft report: 05/08/2012 The level of outputs over the fiscal year studied varies by the program site. The number of clients reached by the program in the fiscal year ranged from 1,674 to 6,442; and the number of condom distributed ranged from 5,590 to 22,656, a 4-fold difference (Table 3). Table 3. Summary costs, outputs, and unit costs of NEP services in Guangxi USD Ningming Wuzhou Average Annual costs (USD) 9,149 100% 19,271 100% 14,210 100% Personnel 2,423 26% 6,563 34% 4,493 32% Recurring goods 5,163 56% 10,313 54% 7,738 54% Recurring services 530 6% 1,919 10% 1,224 9% Capital goods 380 4% 238 1% 309 2% Building 653 7% 240 1% 447 3% Annual outputs IDU clients reached 1,674 6,442 4,058 Needles distributed 90,773 157,720 124,247 Condom distributed 5,590 22,656 14,123 Unit cost (USD) cost per needle distributed 0.10 0.12 0.11 cost per condom distributed 0.02 0.53 0.27 cost per client-VCT service - 1.66 1.66 cost per client reached 5.47 2.99 4.23 Figure 4 compares the cost distributions by category (i.e., personnel, recurring goods, recurring services, capital goods, and building) in our study sites. The distribution of costs across different cost categories between these two sties is very similar. In both programs, spending on personnel is high, more than a quarter of their total costs, while a small fraction of the cost are spent on capital goods and building. In contrast to the cost distribution patterns in MMT programs, the needle exchange programs are much less capital intensive. Figure 4. Cost-breakdown of needle exchange services in Ningming and Wuzhou Capital Building Building goods 1.2% Capital 7.1% 1.2% goods Recurring 4.2% Personnel services 26.5% 10.0% Personnel 34.1% Recurring services NEP in Wuzhou NEP in Ningming 5.8% Recurring Recurring goods goods 53.5% 56.4% Total costs = US$19,271 Total costs = US$9,149 19 Draft report: 05/08/2012 The number of clients seen per year at needle exchange programs is usually in the range of 100 – 200 in Ningming, but the number has been declining since the beginning of the program. In contrast, the number of clients and the number of needles distributed in Wuzhou have been increasing rapidly over the last 12 months period. The figure also indicates that there was a sudden drop in the number of needles distributed in December 2005 to January 2006 in both sites. We speculate that this drop may be partly caused by disruption in supply of needles or delay in funding from the parent organization, but this needs to be confirmed by the intervention sites. Figure 5. Program outputs of needle exchange services 1000 18000 900 Clients (Ningming) 16000 800 Clients (Wuzhou) 14000 Needles (Ningming) 700 Needles-Syringes Needles (Wuzhou) 12000 600 Clients 10000 500 8000 400 6000 300 200 4000 100 2000 0 0 Nov Nov Nov Jan Jun Jan Jun Jan Jun Jul Jul Jul Mar Apr Mar Apr Mar Apr Aug Sep Aug Sep Aug Sep Feb Feb Feb Oct Oct Oct Dec Dec Dec May May May 2004 2005 2006 Figure 5 displays the level of key outputs in the two study NEPs over time. The number of clients at the Ningming site actually declined modestly but steadily starting from November 2004. The number of needles also declined until December of 2005, after which it rose dramatically. This indicates a large increase in the number of needles exchanged per client. The reasons for both of these trends, i.e., steady decrease in clientele accompanied by a sudden increase in needles distributed is worth understanding in more detail through interviews with project managers and other qualitative data collection methods. In Wuzhou, by contrast, clientele rose dramatically following inception of the program, accompanied by a roughly similar proportional increase in the number of needles exchanged. In February of 2006, there was decline in needles exchanged from about 15,000 the previous month to fewer than 12,000. In March, the number rose again to about 15,700. We do not know whether the one-month dip was due to factors on the demand or on the supply side. Again, interviews with project managers could yield useful information about obstacles to efficient service delivery and how these obstacles might be addressed. 20 Draft report: 05/08/2012 Sex Worker programs Figure 6 shows the cost distributions by category for SW services in Pingxiang. The annual cost of the program is US$30,914 with more than a half of the cost spent on recurring services. The annual costs of building and capital goods were very small totaling slightly over $1,250. The unit cost per condom distributed was $0.45 and the unit cost per client reached was slightly over $8. The level of outputs over the fiscal year studied varies by month with a pronounced peak in September, 2005 (Figure 7). The total number of clients reached by the program in the fiscal year was 3,773; the total number of VCT provided was 217; and the number of condom distributed was 68,939. Figure 6. Cost of sex worker program by category Capital goods Building Personnel 2.2% 1.9% 21.8% Recurring Personnel services Recurring 54.9% Recurring goods goods Recurring services 19.3% Capital goods Total cost = US$30,914 Building Figure 7. Outputs from sex worker program in Pingxiang SW Program Outputs in Pingxiang 12000 100 90 10000 80 70 8000 VCT services 60 Condoms Condoms 6000 50 VCT 40 4000 30 20 2000 10 0 0 Nov Nov Jan Jun Jan Jun Jul Jul Mar Mar Apr Apr Feb Aug Sep Feb Aug Sep Oct Oct May May Dec 2005 2006 21 Draft report: 05/08/2012 Scale and unit cost We compared unit cost and scale data between the 14 PANCEA Russian sites and the two Chinese NEP sites in Ningming and Wuzhou. The mean unit cost for the Russian sites was $0.32 per needle or syringe exchanged. For the two Chinese sites the average was $0.11, or 65% lower. As can be seen in Figure 8, the average cost for the Russian sites is raised substantially by the results from just one site, „C‟. If this site is removed from the comparison, the average unit cost for the 13 remaining Russian sites is $0.22, still double the unit cost of the Chinese sites. Figure 8: Unit costs for 14 Russian and Two Chinese NEPS 1.80 1.60 US$ per needle/syringe 1.40 1.20 1.00 0.80 0.60 0.40 0.20 - Russia (Ave.) Guangxi (Ave.) Site I Site J Site C Site D Site H Site N Site L Ningming Site G Site F Site M Site A Site B Site E Site K Wuzhou Note: Unit costs of Site A to Site O are based on the PANCEA‟s data from Russia Figure 8 displays the unit cost per syringe or needle exchanged by the volume of annual exchanges for 14 Russian sites and the two Guangxi Province sites observed in the current study, Ningming and Wuzhou. The Russian sites exhibit significant scale economies as shown by the logarithmic trend line fitted to these data (y = -0.239 ln(x) +2.91). The two Chinese sites show no significant scale economies, and in fact unit costs actually rise slightly with scale (y = -0.039 ln(x) -0.3413). Two sites is an inadequate basis from which to draw any conclusions, but this figure illustrates the comparisons that are possible as the number of Chinese NEPs in our study sample increases. 22 Draft report: 05/08/2012 Figure 9. Cost per needle exchanged by annual needle volume in 14 Russian and 2 Chinese NEPs 1.80 1.60 US$ per needle/syringe 1.40 1.20 Russia 1.00 Guangxi 0.80 Log. 0.60 (Russia) Log. 0.40 (Guangxi) 0.20 - 0 50,000 100,000 150,000 200,000 Needles/syringes exchanged Cost composition As shown in Table 4, comparison of the cost structure of the Russian and Chinese NEPs revealed significant differences. Recurring goods accounted for 55% of costs in the Chinese sites and only 22% in the Russian sites. This in turn is due in part to differences in the unit costs of the needles/syringes themselves. In the sites we studied, their unit costs were about US $0.065 in Russia, and $0.049 in Guangxi, China. Table 4. Comparison of NEP cost composition between Chinese and Russian NEPs Chinese NEPs Russian NEPs (n=2) (n=14) Personnel 30% 46% Recurring goods 55% 22% Recurring services 8% 16% Capital goods 3% 12% Building 4% 4% Note: Russian data is subject to revision 23 Draft report: 05/08/2012 Cost-effectiveness We estimated cost-effectiveness for each of the three interventions using cost data from above combined with the PANCEA HIV epidemic model adapted to Guangxi. Due to time constraints, this exercise was completed quickly, with a minimum of consultation with CDC Guangxi before the modeling and none afterwards. Thus, the results should be considered suggestive rather than definitive estimates of the impact and cost- effectiveness results possible with the PANCEA model. We would welcome opportunities to refine these analyses through further discussions and data gathering. We estimated cost-effectiveness for two of the three groupings of prefectures according to HIV risk – the high risk prefectures which are defined as having an HIV prevalence among IDUs exceeding 40%, and the low risk prefectures defined as having an HIV prevalence among IDUs under 15%. These groupings illustrate how the findings vary by HIV prevalence and according to the reported risk behaviors. A complete analysis would also look at the medium risk prefectures. For each risk grouping, we set parameter values in the PANCEA model to represent population size and mix; HIV prevalence; sexual partner patterns and condom use; and needle sharing. The quality of data available for parameter value assignment varied, as reported to us by CDC Guangxi. In some instances (e.g., HIV prevalence in high risk areas), we could not reconcile the data and thus made some adjustments. For most inputs (e.g., condom use), we had better data for the high risk groups (SWs, IDUs) than for the general population, and therefore made plausible estimates from other settings. The value of key epidemic parameters – and the characterization of the quality of these estimates – are reported in the tables below. Importantly, most uncertainties in epidemic input values have less consequence for cost-effectiveness estimates than do uncertainties in intervention costs and efficacy (which directly determine cost-effectiveness). Our assumptions for intervention cost and efficacy are summarized below. These are derived from the costing exercise reported here and the literature review of effectiveness also reported here.  Methadone maintenance: $428 per person-year 3 , 80% reduction in needle sharing.  Needle exchange: $55 per person-year 4 , 30% reduction in needle sharing.  Sex worker programs: $24 per person-year 5 ; 50% reduction in episodes without condom use and 25% reduction in STIs. 3 The unit cost of MMT per person year ranges from $228 to $588. 4 The unit cost of needle exchange ranges from $36 to $65 per persion-year. 5 It assumes that average program client participates 3 months per year. 24 Draft report: 05/08/2012 High Risk Prefectures The key epidemic inputs for the high risk prefectures are described in the table below. Comments on the data and its adaptation for this purpose are in the technical notes under the table. Table 5. Summary of key epidemic inputs for the high risk areas Guangxi high risk prefectures Needle HIV STI Sex acts per Needle Risk group Size Condom use shares per prevalence prevalence year cleaning year Female 15-49 y.o. 3,653,439 0.0018 SW non-IDU 43,476 0.0300 0.27 521 0.750 "Casual" 155,637 0.0040 0 108 0.215 "Married" 2,788,560 0.0010 0 102 0.050 IDU (incl SW) 15,454 0.1000 0.1 165 0.143 31 0.5 Youth (15-19) 650,312 0.0005 0 31 0.250 Male 15-49 y.o. 3,987,299 0.0082 Client of SW 398,730 0.0080 0 24 0.713 "Casual" 311,273 0.0050 0 55 0.267 "Married" 2,463,289 0.0020 0 124 0.079 IDU 56,420 0.4000 0.042 78 0.211 87 0.5 Youth (15-19) 757,587 0.0005 0 23 0.257 bold solid local data upright estimated using data from local or similar settings underlined adjusted in model for consistency or calibration italics guessed based on plausibility & calibration Technical notes Size data are thought to be accurate at the aggregate level, but more conjectural for specific risk groups, due to difficulties in measurement (sample size, access, definition). HIV prevalence in male IDUs estimated at 47% from AEM model, representing more HIV infections than the male population at 0.63% prevalence. Thus, we lowered IDU HIV% to 40% and raised male HIV% to 0.82%, via very low (perhaps too low) prevalence in other male groups. Similar issues in females. To review & discuss. STI prevalence is for syphilis only. Sex acts per year are based on empirical data only for SWs with clients, otherwise local expert opinion and plausible values. The mix across partner types is less certain. Condom use has been estimated locally only for sex workers (in a location where there is a SW intervention) and IDUs. Values reported represent a weighted mean for all partner types. Needle shares per year reported values were 50 (based on general discussion not formal survey); values of 36 and 93 used for stable predicted prevalence in female and male IDUs, respectively. To review & discuss. As specified in the technical notes, we set the sharing values to yield stable HIV prevalence in IDUs for upcoming years. The predicted HIV prevalence for Guangxi high risk areas over 20 years, without additional HIV prevention, is represented in the figure below. The HIV prevalence is predicted to drop in non-IDU SWs, perhaps reflecting high condom use rates present currently, after implementation of risk reduction programs. This leads to lower HIV prevalence in IDUs in later years. 25 Draft report: 05/08/2012 Figure 10. Predicted HIV prevalence, high risk areas, no new HIV prevention HIV Prevalence 45.0% SW 40.0% Casual F 35.0% Spouse F 30.0% IDU-SW F 25.0% Youth F 20.0% Client 15.0% Casual M 10.0% Spouse M 5.0% IDU M 0.0% 0 2 4 6 8 10 12 14 16 18 20 Youth M Years The predicted HIV prevalence with a $10 million needle exchange program, spread over 5 years, is in figure 11. There is a sharp reduction in IDU HIV prevalence for 5 years, followed by a slowly rising HIV prevalence assuming no new prevention spending. Figure 11. Predicted HIV prevalence, high risk areas, with needle exchange, $10 million over 5 years HIV Prevalence 45.0% SW 40.0% Casual F 35.0% Spouse F 30.0% IDU-SW F 25.0% Youth F 20.0% Client 15.0% Casual M 10.0% Spouse M 5.0% IDU M 0.0% 0 2 4 6 8 10 12 14 16 18 20 Youth M Years 26 Draft report: 05/08/2012 The following figure 12 shows a steady drop in HIV prevalence if the needle exchange program is supported at $40 million spread over 20 years. As expected, such sustained prevention activities has a far greater effect on long-term prevalence than only five years of funding. The reductions are roughly linear suggesting that for the ranges examined here, incremental benefits do not decline as prevalence falls. Figure 12. Predicted HIV prevalence, high risk areas, with needle exchange, $40 million over 20 years HIV Prevalence 45.0% SW 40.0% Casual F 35.0% Spouse F 30.0% IDU-SW F 25.0% Youth F 20.0% Client 15.0% Casual M 10.0% Spouse M 5.0% IDU M 0.0% 0 2 4 6 8 10 12 14 16 18 20 Youth M Years Figure 13 summarizes the impact on expected new HIV infections over 10 years. In the first scenario “Off� signifies no new prevention funding or activities. Nearly 90,000 (88,259) new HIV infections are predicted. Scenario A is $10 million spent on methadone maintenance, spread over 5 years; 84,537 new infections are predicted. Scenario B is $10 million spent on needle exchange, spread over 5 years; 77,050 new infections are predicted. Scenario C is $5 million spent on SW programs, spread over 5 years (according to our data, it is not possible with current SW program design to spend all $10 million due to limited numbers of SWs and low unit costs.) This yields an estimated 86,608 new infections. 27 Draft report: 05/08/2012 Figure 13. Predicted HIV infections, high risk areas, for different prevention scenarios HIV infections over 10 years by Scenario 100,000 90,000 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10,000 Babies 0 "Off" A B C D Men Prevention Resource Allocation Scenario Women The number of infections averted in each prevention scenario and their associated cost- effectiveness are summarized below. The most cost-effective intervention is needle exchange, at approximately $900 per HIV infection averted. The other two interventions both cost about $3,000 per infection averted. These results, shown in Table 6, are first- order approximations only. They are somewhat sensitive to epidemic inputs, and very sensitive to estimates of unit cost and of effectiveness. Table 6. Cost and epidemic impact of interventions in high risk areas by different prevention scenarios Methadone Needle Sex worker maintenance Exchange programs (Scenario A) (Scenario B) (Scenario C) HIV infections prevented over 10 years, with 5 years of program operation 3,722 11,210 1,651 Cost per infection prevented ($) 2,687 892 3,029 Low Risk Prefectures The key epidemic inputs for low risk prefectures are summarized in the table below. HIV prevalence varies the most between areas. Of note, condom use rates differ little from those found in the high risk areas, perhaps reflecting increased awareness of HIV in both areas. STI rates are largely derived from the high risk areas, due to data limitations. Also, despite vast difference in HIV prevalence in IDUs (the definition of the risk groupings), needle sharing rates are assumed to be the same. We made this assumption because CDC 28 Draft report: 05/08/2012 Guangxi is not aware of good data on sharing rates, and especially not on differential sharing rates by risk areas. This is an important area for review and discussion. Table 7. Summary of key epidemic inputs for the low risk areas Guangxi low risk prefectures Needle HIV STI Sex acts per Needle Risk group Size Condom use shares per prevalence prevalence year cleaning year Female 15-49 y.o. 3,733,256 0.00035 SW non-IDU 38,117 0.0071 0.27 539 0.606 "Casual" 159,037 0.0010 0 108 0.213 "Married" 2,856,090 0.0002 0 102 0.050 IDU (incl SW) 15,493 0.0258 0.1 165 0.071 31 0.5 Youth (15-19) 664,520 0.0001 0 31 0.250 Male 15-49 y.o. 4,086,396 0.00190 Client of SW 408,640 0.0030 0 24 0.554 "Casual" 318,073 0.0020 0 55 0.247 "Married" 2,526,058 0.0002 0 124 0.072 IDU 57,210 0.0875 0.045 78 0.215 87 0.5 Youth (15-19) 776,415 0.0005 0 23 0.256 bold solid local data upright estimated using data from local or similar settings underlined adjusted in model for consistency or calibration italics guessed based on plausibility & calibration Technical notes Size data are thought to be accurate at the aggregate level, but more conjectural for specific risk groups, due to difficulties in measurement (sample size, access, definition). Difference in total for males represents MSMs (assumed very few for this exercise). HIV prevalence estimates are most solid for SWs and IDUs. We assumed very low values to approach the overall estimated HIV prevalence for all females and all males. To review & discuss. STI prevalence is for syphilis only. Data are largely imputed from high risk area, due to gaps in low risk area Sex acts per year are based on empirical data only for SWs with clients, otherwise local expert opinion and plausible values. The mix across partner types is less certain. Condom use has been estimated locally only for sex workers and IDUs. Values reported represent a weighted mean for all partner types. Needle shares per year reported values were 50 (based on general discussion not formal survey). Values of 36 and 93 were derived for high risk areas to achieve near stable prevalence; they lead to sharply increasing HIV prevalence in IDUs in low risk areas. To review & discuss. Predicted HIV prevalence is shown below in figure 14. Of note, HIV prevalence in IDUs (followed later by other groups) rises, due to sharing risk behaviors that are similar to those in the high risk areas. Whether this increased prevalence is likely over time (representing just a delay in some infections as compared with high risk areas), or other factors enduringly suppressing HIV prevalence, is unknown. Obtaining better data on needle sharing patterns (and sex risks) would help to assess these possibilities. 29 Draft report: 05/08/2012 Figure14. Predicted HIV prevalence, low risk areas, no new HIV prevention HIV Prevalence 35.0% SW 30.0% Casual F Spouse F 25.0% IDU-SW F 20.0% Youth F 15.0% Client 10.0% Casual M 5.0% Spouse M IDU M 0.0% 0 2 4 6 8 10 12 14 16 18 20 Youth M Years Figure 15. Predicted HIV prevalence, low risk areas, with needle exchange, $10 million over 5 years HIV Prevalence 30.0% SW Casual F 25.0% Spouse F 20.0% IDU-SW F 15.0% Youth F Client 10.0% Casual M 5.0% Spouse M IDU M 0.0% 0 2 4 6 8 10 12 14 16 18 20 Youth M Years Figure 16 summarizes the impact on expected new HIV infections over 10 years in low risk areas. The first scenario “Off� is no new prevention spending. An estimated 35,344 new HIV infections are predicted. Scenario A is $10 million spent on methadone maintenance, spread over 5 years; 33,383 new infections are predicted. Scenario B is $10 million spent on needle exchange, spread over 5 years; 29,790 new infections are 30 Draft report: 05/08/2012 predicted. Scenario C is $5 million spent on SW programs, spread over 5 years yielding an estimated 34,566 new infections. (As above, $10 million cannot be absorbed by SW programs). Figure 16. Predicted HIV infections, low risk areas, for different prevention scenarios (see text) HIV infections over 10 years by Scenario 40,000 35,000 30,000 25,000 20,000 15,000 10,000 5,000 Babies 0 "Off" A B C D Men Prevention Resource Allocation Scenario Women The number of infections averted and associated cost-effectiveness are summarized below in Table 8. As with high risk areas, needle exchange is the most cost-effective, with methadone maintenance and SW programs costing substantially more per HIV infection averted. Cost-effectiveness ratios are about twice as high (therefore less favorable) in the low risk settings as in the high risk settings. The difference in cost- effectiveness is mainly due to less severe epidemic conditions: Lower prevalence, means less HIV exposure to susceptibles, hence less HIV incidence, and finally fewer HIV infections averted by the same proportionate reduction in risk and incidence. Table 8. Cost and epidemic impact of interventions in low risk areas by different prevention scenarios Methadone Needle Sex worker maintenance Exchange programs (Scenario A) (Scenario B) (Scenario C) HIV infections prevented over 10 years, with 5 years of program operation 1,960 5,553 778 Cost per infection prevented ($) 5,101 1,801 6,429 31 Draft report: 05/08/2012 VII. Discussion Implementation of harm reduction programs in Guangxi Province has expanded rapidly, from 7 in 2001 to 34 currently and projected 65 by the end of 2006. This level of commitment warrants a careful examination of the efficiency with which these programs operate. Our study of the unit costs, cost composition, and output levels from three MMT, two NEPs and one SW site in Guangxi Province highlights key issues concerning implementation of the existing programs and for further expansion of harm reduction programs in Guangxi Province. There are some large variations in prevention costs, outputs and efficiency across programs and services. For example, the two MMT programs have a 2.6-fold variation in unit costs (cost per client-month of service), from $19.0 per client-month-in Wuzhou to $49.4 per client month in Nanning. Unit costs as measured by cost per needle/syringe exchanged varied from $0.10 in Ningming to $0.12 in Wuzhou, a more modest but still sizable variation. Some of this variation may reflect real differences in the cost and demand structure faced by the respective sites. However, it is also possible that some of this variation in efficiency reflects varying quality of program management and planning. If so, this suggests that a primary challenge for policy makers, and program managers alike, is to identify methods for raising program efficiency. The level of program outputs in the NEP and SW sites sometimes rose and fell dramatically within short periods. Understanding the determinants of these fluctuations through qualitative data may also yield insights that could be translated into higher levels of efficiency. Personnel and recurring services account for a large portion of the total costs, 66% of total costs in the Nanning MMT site, and 79% of the total at the Wuzhou site. In both cases personnel is the largest category with 35% of the total. In the NEP sites, there is a similar uniformity in cost structure, with recurring goods and services and personnel combined constituting 89% of the total in Wuzhou and 98% of total in Ningming. Recurring goods is by far the largest cost category at the NEPs. Identifying the major determinants of recurring goods and finding potential ways to economize there may be a viable path to increased efficiency. Similarly, identifying ways of increasing productivity may be one of the most promising avenues for enhancing the efficiency of the personnel- intensive MMT sites. Physical space constitutes a surprisingly large portion of total costs in the MMT sites, 34% in Nanning and 16% on Wuzhou. It is not clear whether there are opportunities to reduce these costs, but site managers should be encouraged to explore this possibility. No clear indication of economies of scale in NEPs in Guangxi, based on the limited sample sites. The two NEP sites for which we have unit cost data show essentially constant returns to scale where number of needles or syringes distributed is the output of interest. Data from additional sites are required before we can draw any conclusions. By contrast, the Russian PANCEA NEP sites showed significant economies of scale. 32 Draft report: 05/08/2012 Average unit costs in the Russian sites are double that of the Guangxi NEPs we studied. However, the largest and most efficient Russian sites had unit costs in the same range or lower than the Chinese sites. Unit cost and scale data are available from only two NEP sites. This is an insufficient sample from which to draw conclusions. However, the Russian data suggests that efficiency gains may be feasible as NEPs increase their clientele. Referrals from NEP to MMT. Establishing routine referral links from NEPs to MMT programs could be an effective and relatively easily implemented method of increasing MMT uptake and possibly cost-effectiveness. Such links may also remove the stigma that is sometimes associated with NEP programs, i.e., that they encourage drug abuse. 33 Draft report: 05/08/2012 References 1. Bertozzi, S., et al., Chapter 18: HIV/AIDS Prevention and Treatment, in Disease Control Priorities in Developing Countries, D. Jamison, et al., Editors. 2006, Oxford University Press: New York. 2. Beyrer, C., Hidden epidemic of sexually transmitted diseases in China: crisis and opportunity. Jama, 2003. 289(10): p. 1303-5. 3. Ministry of Health of China, National HIV/STI Surveillance Report. 2004. 4. Ksobiech, K., A meta-analysis of needle sharing, lending, and borrowing behaviors of needle exchange program attenders. AIDS Educ Prev, 2003. 15(3): p. 257-68. 5. Kahn, J., Chapter 14: How do NEPs affect HIV drug and sex risk behaviors? The public health impact of needle exchange programs in the United States and abroad., in Report prepared for the Centers for Disease Control and Prevention. 1993: San Francisco. 6. Lurie, P., E. Drucker, and A. Knowles. Still working after all these years: increasing evidence of needle exchange program (NEP) effectiveness in studies published since 1993. in XII International AIDS Conference. 1998. Geneva. 7. Gowing, L., et al., Substitution treatment of injecting opioid users for prevention of HIV infection., in Cochrane Database Syst Rev. 2004. 8. Gowing, L.R., et al., Brief report: Methadone treatment of injecting opioid users for prevention of HIV infection. J Gen Intern Med, 2006. 21(2): p. 193-5. 9. Marshall, N., et al., The Effectiveness of HIV Prevention Interventions in Developing and Transitional Countries: A Systematic Review. Under review, 2006. 10. Bhave, G., et al., Impact of an intervention on HIV, sexually transmitted diseases, and condom use among sex workers in Bombay, India. Aids, 1995. 9 Suppl 1: p. S21-30. 11. Egger, M., et al., Promotion of condom use in a high-risk setting in Nicaragua: a randomised controlled trial. Lancet, 2000. 355(9221): p. 2101-5. 12. Marseille, E., et al., Assessing the efficiency of HIV prevention around the world: methods of the PANCEA project. Health Serv Res, 2004. 39(6 Pt 2): p. 1993-2012. 13. UNAIDS, Costing Guidelines for HIV Prevention Strategies. In UNAIDS Best Practices Collection/00.31. 2000, UNAIDS: Geneva. 34 Draft report: 05/08/2012 ANNEXES Annex 1: Descriptions of Study Sites Liuzhou (MMT) Liuzhou is the biggest industry base in Guangxi, and is located in the middle of the Guangxi province. Governing 4 urban districts including Chengzhong, Yufeng, Liunan and Liubei, and 6 counties including Liujiang, Liucheng, Luzhai, Rong‟an, Sanjiang and Rongshui, its geographic area is 18,600sq.km, among which 658.3sq.km is urban area. Population by the end of 2003 is 3,512,600, among which 958,900 is urban population, and minority population including Zhuang, Miao, Yao and Dong is 1,687,000. The total number of registered drug users in Liuzhou is 10,000, among which 4,800 is in urban area. 83.1% of drug users are male while 16.9% are female, and. 95% of the drug users are IDU. The first needle exchange program was started in September 2003 with the support from AusAID. More than 700 drug users are covered under this program. In baseline surveys conducted in 2004, 22.5% of the drug users indicated that they “have shared needles in the last week�, 50% indicated that they “did not use condoms in their sex behavior with sex workers�. Liuzhou is one of the three MMT sites in Guangxi that started in 2004 during the second pilot phase. The facility now covers about 400 injecting drug users. 35 Draft report: 05/08/2012 Nanning (MMT) Nanning is the capital city of Guangxi with six urban districts including Qingxiu, Liangqing, Xingning, Yongning, Jiangnan and Xixiangtang and one county, Wuming. Its geographic area is 22,300sq.km, among which 1,834sq.km is urban area. Population of Nanning by the end of 2003 is 6,416,700, among which 1,457,700 is urban population. Nearly a quarter of the city‟s population lives in urban area, less than 10% of the total geographic area of Nanning city. The first pilot MMT has started in November 2003, as a part of the Red Cross Hospital. Three additional pilot MMT projects were initiated in Liuzhou, Hechi, and Wuzhou in December 2004. Since the beginning of the program, nearly 370 IDUs were registered in the facility and of those 149 patients are currently receiving methadone treatment. As for May 2006, 11 new MMT sites have got approval and are being prepared for operations in later year. Also, another 8 sites have submitted application and are waiting for the approval. The Needle Exchange program was started in September 2003 with support from AusAID, covering about 800 IDUs. According to the 200X baseline survey, 60% of the drug users indicated that they “have shared needles in the last week�, 7% indicated that they “did not use condoms in their sex behavior with sex workers�. Ningming (NEP) Ningming is located in the southwest part of Guangxi, borders Vietnam in the south; shares a 212 kilometers frontier with Vietnam. In 2000, the total population is 390,000, male 200,000 and female 190,000. Urban population is 570,000 and rural population is 330,000. There are more than 15 minority groups but 78.52% is Zhuang and 20.98% is Han. The mobile population in Ningming is about 500,000 every year, mainly from Hunan, Hubei and Sichuan province, and the other areas of Guangxi. Ningming is bordering 3 villages of Vietnam which are Tongmian, Aidian and Zhilang. About 3000-4000 people are going to Vietnam from Ningming with formal documents such as laissez-passer or passport, but the actual number could be much more because on the border there are many paths without guard connecting China and Vietnam, the residents on the frontier could pass in and out of Vietnam freely. By the end of 2003 the total number of registered IDUs is about 1800. The AusAID funded NEP was started in October 2002, covering about 600 IDUs as of May 2006. The needles sharing rate among IDUs has reduced from 94% in baseline survey to 22.5% in the 24 months follow-up survey. The rate of condom use with casual sex partners has reduced from 19% in baseline survey to 7.7% in the 24 months follow-up survey. Pingxiang (SW) Pingxiang lies in the southwest part of Guangxi, borders Vietnam and shares a 97 kilometers frontier with Vietnam. There are 2 national ports, 6 frontier trading spots and one national frontier economic cooperation district. People could pass the border by 36 Draft report: 05/08/2012 Xianggui Railway, 322 national highway and Ping‟er River. Geographic area is 650sq.km. Population by the end of 2003 is 102,400, among which 72,300 is rural population. Minority population including Zhuang, Miao and Jing is 85,400. The growth rate of population is 4.62‰. Pingxiang city, a border city of China to Vietnam, hosts a number of mobile populations, including day traders, temporary workers, and sex workers. As of November 2005, there were 352 reported HIV cases and 26 deaths. A major transmission mode is through injecting drugs followed by sexual transmission. In 1998, the first HIV prevention program targeting sex workers (named as “Women Health Center Project�) started with the funding from Netherland Development Organisation, and covering more than 300 women. The most recent survey data on sex workers in Pingxiang indicated that condom use in last sex behavior was high at 80.1% in 2006 increased from 67.1% in 2003. However the consistent use of condom still remains low at 62.2% in 2005. More than 90% of the sex workers in Pingxiang are aware of the condom distribution services in the Women Health Center. Starting from 2005，the center expanded its outreach servers and provides condoms in hotel rooms. Since the Women Health Center was set up in 1998, staffs make regular visits to entertainment places to provide health education and counseling, and trainings on the correct use of condoms, and trainings for peer educators to sex workers. In addition, STI clinic was opened in Pingxiang CDC to provide medical service for target population. In 30 hotels, IEC materials are provided in the lobby and condoms are provided in hotel rooms. In those villages with severe HIV/AIDS prevalence, interventions such as health education and condoms promotion are conducted. In 2005, another sex workers program supported by FHI has been initiated in Puzhai one of the six trading spots, at the border of Vietnam. The program mainly focuses on Vietnamese sex workers and truck drivers, and provides outreach education as well as STI services at the project funded clinic. Wuzhou (MMT + NEP) Wuzhou is located near Guangzhou, and is function as trading market between Guangzhou and Guangxi. The total population of Wuzhou city is approximately 480,000. At the end of 2005, 980 HIV cases were reported. The surveillance conducted in 2003 indicated that more than 50% of IDUs were HIV positive. In 2004, the AusAID supported the first NEP in Wuzhou, which also provides services including condom distributions and IEC. At present, three NEPs are operating in both urban and rural areas in Wuzhou. More than 1,400 IDUs have registered and nearly 700 IDUs are currently served by the three NEP sites. The first MMT in Wuzhou was established in August 2005. Wuzhou was one of the four MMT sites that are currently operating in Guangxi. Since the beginning of the program, 143 patients have registered and 130 patients are active and currently seeking methadone treatment. The second MMT facility is currently under preparation and is expected to start services by the end of 2006. 37 Draft report: 05/08/2012 Annex 2: Survey Instruments SECTION 1: Characteristics of Facility To be asked of facility manager / director Facility location: Data collection date: Data Source: 1st 2nd 3rd 1 Type of facility 5 Why did you choose to open the facility here? (up to 3 reasons) 2 Type of institution 3 What year was this facility founded? 6 Has this facility always been at this location 4 When did you begin providing services at this location? 7 Do people with HIV/AIDS work in this facility? Data Source: 1 2 3 4 5 6 7 8 For each of the following services VCT CSM STI care IEC schools MTCT SW RR Have you ever provided this service in this facility? (yes, no, dk, n/a) 8 (If no, go to next intervention type) 9 Do you still provide this service? (yes, no, dk, n/a) Do you pay for any organization outside your facility to provide these services? 10 (yes, no, dk, n/a) 11 In what year did you begin providing this service? 12 If "No" to q 9: In what year did you stop providing this service? If "No" to q 9: Why did you stop providing this service? (Skip to next intervention 13 type) 14 If "Yes" to q 9: Do you provide services here, elsewhere, or both? Data Source: 1 2 Days and hours of service Work Week Weekend 15 How many days do you regularly provide services? 16 How many hours each day do you regularly provide services? 17 How many staff who see clients are usually present during regular hours on these days? 18 How many staff who see clients are usually present AFTER regular hours on these days? 1 2 3 4 For each of the groups that you serve, 2nd important target 3rd important target Most important group 4th important target group group group 19 What is your target population? 20 Can you estimate how many people there are in this area in each group? 21 How many among this group are HIV-infected today? (percent) 22 How many among this group were HIV-infected one year ago? (percent) 23 How did you make these estimates? 38 Draft report: 05/08/2012 SECTION 2: Institutions and Governance To be asked of facility manager / director Data Source: 1 Are you part of a larger organization? (if independent, go to Q8) yes/no 2 Is your parent organization incorporated or registered? yes/no 3 What type of organization are you part of? 4 How many similar facilities does this organization operate in this country? number 5 How often are you visited by a supervisor? 6 How often do you have to submit a report to your supervisor? 7 How much time does your supervisor spend here each month? (hrs) number 8 Does the funding you receive depend on your performance? (If no, go to Q10) yes/no 9 How is your performance measured? (list up to three criteria) Data Source: Community Client groups groups 10 Do you meet with (…) to discuss facility operation or policy? (If no, skip to Q12) 11 How often do you meet with (…) to discuss facility operation or policy? Parent This facility organization 12 Is there a governing board for (…)? (If no, go to Q18) 13 Are clients or members of the local community represented in the governing board? 14 Can the board vote to remove the facility director? 15 How often does the governing board meet? 16 Is the governing board elected or appointed? 17 Who elects (appoints) the governing board? Data Source: 18 Which group or person exercises ultimate authority on the following decisions? 1 purchasing small supplies and consumables? 2 purchasing larger equipment and capital goods? 3 hiring and firing? 4 salaries and bonuses? 5 promotions and job assignments? 6 If user fees, levels and whom to charge? 7 general budget questions 8 program strategies and priorities 9 program operational policies 39 Draft report: 05/08/2012 SECTION 3a: Inputs -- Buildings, physical plant and general quality To be asked of facility manager or building manager. Data Source: 1 Number of buildings occupied by entire facility For up to three main buildings where facility services are performed Building 1 Building 2 Building 3 (or were in the most recent fiscal year): 2 Brief description / label (no address here, for confidentiality) 3 Do you occupy part or all of this building? 4 Do you share space with any other organization in this building? What is the approximate total area occupied by your facility in this 5 building? 6 What is the approximate total physical area of this building? What is the TOTAL number of people who work in, or are based at, 7 this facility in this building? 8 What year did you buy or begin using these premises? Data Source: 9 Do you own or rent? (if own, go to Q14) 10 What was your previous month's rent? Was your monthly rent the same for all of the most recent fiscal 11 year? (If "yes", go to Q19) 12 If "no", what was your monthly rent previously? 13 On what date did your monthly rent change? (mm/yy) 14 Total purchase price of land and building. 15 Monthly mortgage payment Was your mortgage payment the same for all of the most recent 16 fiscal year? (y/n) (If "yes", go to Q19) 17 If "no", what was your payment previously? 18 On what date did your payment change? (mm/yy) Data Source: 19 What year was this building built? 20 Material of walls 21 Material of floors 22 Material of roof 23 Are there toilet facilities for clients? (if "no", go to Q25) 24 Does the toilet flush? Is there soap and water for employees and patients to wash their 25 hands? 26 Do you have a waiting room or area in this building? How do you dispose of waste? (choose one, if multiple use 27 Comment) Data Source: 1 2 3 4 5 6 For each of the following services, gas / oil for other heating / electricity water telephone heating and internet cooking fuel cooking 28 Do you have this service? 29 What is your main source for this service? 30 How many hours per day is this service available, on average? 31 How many days last week was service interrupted? 40 Draft report: 05/08/2012 SECTION 3b: Inputs -- Personnel directly & indirectly involved in HIV intervention Data Source: To be asked of facility manager or personnel manager. Fiscal year detail column N to AC 1 2 3 4 5 6 7 8 9 10 11 12 year Monthly breakdown of payment for the past fiscal 13 . . . 14 15 16 17 18 19 20 Month 1 Month 2 Month 3 Month 4 Month 5 Month 6 Month 7 Month 8 Month 9 Month 10 Month 11 Month 12 Initials for Job Job Title What is the Sex Age When did s/he When did s/he Does s/he How much Did s/he From whom Did s/he Did s/he Does s/he How How many Is s/he If the wage in Current % allocation of personnel efforts to: each Category highest level of start working at stop working receive any did this receive did s/he receive any receive any have any many hours did working questions 8-9 is personnel, formal education this facility? at this facility? compensation? person his/her most directly training in payments other job hours did you expect today? zero what would STI VCT CSM IEC school MTCT SW RR including s/he completed? (mm/yy) (mm/yy) (i.e. salary, receive for recent receive this the most to help with outside this s/he work him / her you have to pay care volunteers bonues, per the last payment on payment? recent fiscal any facility? last to work to hire someone diem or other?) month? time? year? training? week? last week? to do the work of 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 41 Draft report: 05/08/2012 SECTION 3c: Inputs -- personnel turnover To be asked of facility manager or personnel manager. Data Source: 1 2 3 4 5 6 7 When did you What was the How long did it most recently Did you try to What is the monthly take to fill that lose replace him / monthly salary Personnel/volunteer job categories Why did s/he leave? salary of the vacancy personnel of her? (if no, of the person who (number of this type? go to Q6) replacement? left? months)? (mm/yy) 1 Senior managerial / administrative staff 2 Other managerial / administrative staff 3 Physician 4 Nurse or other medical service staff 5 Counselor 6 Medical / health support staff 7 Other support staff 8 Other (specify) 9 Other (specify) 10 Other (specify) 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 42 Draft report: 05/08/2012 SECTION 3d: Inputs -- Capital goods, equipment Data Source: To be asked of facility manager or purchasing/supply manager. 1 2 3 4 5 6 7 8 9 10 11 12 13 How did you acquire it? What is the value per What proportion What is the value Average % allocation of capital goods to: Did you share the What proportion (or Do you sometimes How old was this 1=purchased by this facility -> Q5 week of any (or how many per week of any How much did you When did you purchase of this how many days) of lend or rent out this item when 2=rented -> Q5 Price paid or compensation (in days) of each compensation (in pay for repairs and Item Description acquire it? equipment with other each week do you equipment to other STI acquired? 3=borrowed -> Q7 monthly rent money, goods or week do you lend money, etc) that upkeep in the most VCT CSM IEC schools MTCT SW RR (mm/yy) 4=received free from parent org or organizations? (if no, use/borrow this organization? (If no care (years) services) that you give / rent this you receive for recent fiscal year? other source -> Q9 skip to Q9) equipment? skip to Q12) for borrowing this equipment? renting this 1 General office 2 Computers 3 Printers 4 FAX machines 5 6 Transportation 7 Cars, Trucks 8 Motorbikes 9 Bicycles 10 11 Audio/visual 12 Televisions 13 Video players 14 Video monitors 15 Projectors 16 17 Lab/clinical 18 Generators 19 Autoclaves 20 Refrigerators 21 Freezers 22 Microscopes 23 Incubators 24 HIV testing machines 25 STI testing machines 26 Air conditioners 27 Heaters 28 29 Others 30 31 32 43 Draft report: 05/08/2012 SECTION 3e: Inputs -- Recurrent expenses, goods. NON-PERSONNEL Data Source: To be asked of facility manager or purchasing/supply manager. Fiscal year detail column J to Y 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Where did you Current % allocation of recurrent expenditure to: What did you do? How many most recently Did you How many How many did (1. buy more, 2. did you purchase them? What was the How much purchase times did you How much you receive free request from parent How many purchase in (1. local mkt, 2. average price did you these run out of did you How many do from parent When did you org or donor., 3. Redundant did you the most formal supplier, you paid for spend, in together with stock in the Item Description Category purchase last recent fiscal 3. gov't, 4. each one during spend, in total, in the another you have in organization or most recent last run out? wait for resupply, 4. STI school total, last stock? other donor in (mm/yy) borrow from another VCT CSM IEC MTCT SW RR month? year? (if 0, imported directly, the most recent most recent organization fiscal year? care s month? the most recent organization, 5. skip to 5. Other fiscal year? fiscal year? outside your (If 0, go to fiscal year? Other (Specify), DK, question 9) (Specify), DK, facility? Q14) N/A N/A 1 Methadom 2 Needles/syringes 3 Male condoms 4 Female condoms 5 Baby formula 6 HIV test kits - ELISA 7 HIV test kits - Western blot 8 STI test kits & reagents 9 STI medications 10 Nevirapine for vertical transmission 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 44 Draft report: 05/08/2012 SECTION 3f: Inputs -- Recurrent expenses, services. NON-PERSONNEL To be asked of facility manager or purchasing/supply manager. Data Source: Fiscal year detail column H to W 1 2 3 4 5 6 7 Current % allocation of recurrent expenditure to: What is the value Did you purchase of the services you How much did this service How much did received free from you spend, in together with Redundant you spend, in parent Service Description total, in the most another STI Category total, last organization or VCT CSM IEC schools MTCT SW RR recent fiscal organization care month? other donor in the year? outside your most recent fiscal facility? year? 1 Training for staff 2 Training for others 3 Radio advertisements (design, production, broadcasting) 4 Television advertisements (design, production, broadcasting) 5 Pamphlets, newsletters, etc (design, production, distribution) 6 Video and audio cassette production 7 Other materials development 8 Other publicity / IEC activities 9 Cleaning and building maintenance 10 Waste disposal 11 Electricity 12 Water 13 Telephone 14 Gas / oil for heating and cooking 15 Other heating / cooling / cooking fuel 16 Maintenance for buldings 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 45 Draft report: 05/08/2012 SECTION 4: Financing To be asked of facility manager or finance officer. Data Source 1 2 3 4 5 Please tell us the organizations from which you received funding in the How many times did What % of your What % of your intervention budget(s) did this organization most recent fiscal year, starting with the most important. (Prompt list Are you required to this organization visit facility budget did provide in the most recent fiscal year? - submit an annual your facility in the this organization below) - STI report to this funder? most recent fiscal provide in the most VCT CSM IEC schools MTCT SW RR care year? recent fiscal year? - a b c d e f g h I j Data Source 6 Please tell us the other organizations from which you sought funding in the most recent fiscal year, but did not receive it. - Sample list of funders a 1. Bilateral donor b 1.1 USAID c 1.2. DfID d 1.3. France e 1.4. Ireland f 1.5. Scandinavia g 1.6. Germany h 1.7. Other I 2. International agency j 2.1. WHO / OMS 2.2. World Bank 2.3. UNAIDS Data Source 2,4, Other UN 2.5. Other How many days did facility staff spend fundraising (not cost 3. International NGO / foundation 7 recovery ) in the most recent fiscal year? (person days) - 3.1. MSF / FSF 3.2. Soros / Open Society 8 Do you collect user fees? (If no, go to Q9) 3.3. IPPF (Ask of facilities that are not independent) How much of the 3.4. Rockefeller 9 revenue you collect in user fees are you required to remit to 3.5. Ford your parent organization? (percent) (If n/a or 0, end) 3.6. Other 4. Domestic government Of the amount that your facility keeps, what percent is allocated to: 4.1. Health ministry 1 2 3 4 5 4.2. National AIDS control program Wages / Drugs and 4.3. Other 10 Other facility Other / salaries / other essential Savings 5. Domestic NGO / foundation costs miscellaneous bonuses supplies 46 Draft report: 05/08/2012 Section 5a: Inputs -- facility expenditures To be asked of facility manager or purchasing/supply manager. Instructions Data Source: Please tell us your ACTUAL EXPENDITURES (not budgeted expenditure) in the last month and most recent fiscal year (with month or quarter detail, if available), for your FACILITY (the entire facility or a subunit, as decided for this site; see the manual.) Items on this sheet in all CAPITAL letters (and shading) are most important (all or part of Questions 1, 2, 3, 4, 6, and 7) Items in Mixed Case letters are desirable if possible, or should be completed as indicated. 1. TOTAL FACILITY EXPENDITURE LAST MONTH MOST RECENT FY (Provide category detail below if 3. VALUE OF IN-KIND 2. MOST RECENT FY TOTALS BY TIME PERIOD the total is unavailable; note type DONATIONS IN MOST TOTAL FACILITY Month Quarter Year of donations in Comments.) RECENT FY EXPENDITURE TOTAL Buildings Utilities Personnel Capital goods Recurrent goods Recurrent services Other (specify) Complete "Goods" and "Services" columns only 4. TOTAL RECURRENT EXPENDITURE if a total is unavailable. 6. TOTAL CAPITAL GOODS VALUE Goods Services TOTAL LAST MONTH EXPENDITURES (Total last FY) MOST RECENT FY ASSETS IN PLACE (total value) List of assets if register available but total value unknown 5. Most Recent FY Detail Total Recurrent Expenditure Month Quarter Year Goods Services Total 7. TOTAL PERSONNEL 9. If total personnel expenditure is unavailable for the last month, EXPENDITURE list personnel for the last month. LAST MONTH Job Title # of Personnel (FTE) MOST RECENT FY 8. Most Recent FY Detail Total Personnel Month Quarter Year Expenditure 47 Draft report: 05/08/2012 SECTION 5b:Outputs Data source(s How many To how many HIV- How many cases of needles + How many positive clients did How many clients How many VCT STIs (non-HIV), Year Month Quarter syringes were condoms were you provide post- Activity #1 Activity #2 Activity #3 were reached? were provided? total did you distributed to distributed? test counseling detect? clients? sessions? Condom HIV(+) post-test Year Month Quarter Clients Needles-Syringes VCT STI Services Activity #1 Activity #2 Activity #3 distribution counseled Jan Feb Mar Apr May Jun 2004 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun 2005 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun 2006 Jul Aug Sep Oct Nov Dec 48 Draft report: 05/08/2012 Annex 3: Description of PANCEA epidemic model By James G. Kahn Overview The PANCEA epidemic model is intended to be a flexible tool to predict the impact and cost of alternative HIV prevention scenarios. To accomplish this purpose, the model combines several major components: description of the local HIV epidemic; HIV acquisition by risk groups; specification of prevention scenarios, effectiveness, and costs; HIV risk reduction associated with prevention; and predicted epidemic course (HIV infections expected) with and without prevention. The sections below describe each model component and how a user would interact with it. The model relies on values entered by the programmer or user (“inputs�) as well as a complex set of calculations to translate those inputs into desired results. The name of the model file, an Excel workbook, is PAN_multi-int-grp_model_vN.n.xls (“N.n� specifies the version). Description of the local HIV epidemic (worksheet InitialCond) The local HIV epidemic is characterized in terms of the size and risk level of eleven “risk groups� in the population. Key inputs include the total population size, and the proportion represented by the specified risk groups: Female: Sex workers, casual, spouse, IDU, and youth Male: Client, casual, spouse, IDU, youth, and MSM. These risk groups are specified in order to portray the major levels and types of HIV risk (which determine routes of HIV transmission) and to allow targeting of HIV prevention. The current HIV status of each group is described by HIV prevalence. The risk of HIV acquisition is described by three risk factors that can be altered by HIV prevention: Proportion of sex (needle) episodes using condoms (bleach), by type of partner Number of partners per year with each other relevant risk group Prevalence of sexually transmitted infection other than HIV The values for these risk factors may differ for partner risk groups due to inconsistencies in data referenced for the local epidemic. When HIV- and HIV+ partners groups differ in condom use, the higher value is used. The number of partners is forced to be equivalent in the inputs. For prevalence of STI, the average of prevalence between the two groups is used. In addition, the model incorporates two risk factors that are not directly modified by prevention. One is the number of episodes per partner per year. The other is non-random 49 Draft report: 05/08/2012 mixing of HIV-infected and HIV-uninfected individuals, which is specified by a same- serostatus preference factor. If data were obtained on prevention effects for these risk factors, they could be varied accordingly. In addition, the non-HIV “exit rate� is estimated (e.g., the rate at which sex workers stop sex work). This allows estimates of turnover of each risk group and feeding of HIV infection into other groups by migration. Data for these inputs derives from the best available local studies and expert opinion. [for specific implementation, list key sources here] When data are sparse, best estimates from similar populations can be used, and adjusted as necessary to produce plausible epidemic patterns over time. In addition to risk group data, the model has inputs concerning a variety of biological factors. These include the risk of HIV transmission per exposure; the effectiveness of condoms; the added risk caused by having an STI; and the rate of progression from HIV infection to symptomatic disease, AIDS, and death. These values are estimated from the scientific literature. The use of base case or intervention risk parameters is specified in cell B58 as “B� or “I� respectively. HIV acquisition by risk groups (worksheets RG1, RG2, etc) For each risk group, the model translates the risk factors into an annual risk of HIV acquisition. In general, having more partners, lower condom use, and higher STI prevalence increases risk. Specifically, the risk equation is of the form: N * (1 - (1 – R *(1 - Ec)) ^ (e*C) * (1 - (1 – R) ^ (e*(1-C)) * Srr where: N = number of HIV+ partners of that type per year R = risk of HIV acquisition per unprotected exposure Ce = effectiveness of condoms in reducing risk Ec = episodes of sex per partner per year C = proportion of sex episodes using a condom Srr = increase in risk conferred by STI prevalence The values of italicized parameters (N, e, C, Srr) reflect inputs representing the local HIV epidemic. The value of upright font parameters (R, Ce) derive directly from scientific studies. The parameter Srr also reflects scientific data on the added risk conferred by having an STI. 50 Draft report: 05/08/2012 The value of N and C may become inconsistent between risk groups, due to unequal changes in group size or risk behavior over time (e.g., due to different effects of prevention). In those cases, the model uses the lower risk value (i.e., higher condom use, fewer partners). The value of Srr remains the average of the two groups, as at baseline. The model tracks the evolution of the risk group over time. The time period for this adjustment is one month. Changes include new HIV infections, deaths, exit due to age or behavior change, and entry from other population groups. The HIV prevalence is recalculated for each time period. The risk behaviors are tracked separately for HIV+ and HIV- individuals, since prevention may differentially affect these groups (see below). Specification of prevention scenarios, effectiveness, and costs (worksheet Intrvns + Resource_Allocation_v3.xls) The model examines the individual and combined effects of 8 prevention strategies. These are: voluntary counseling and testing (VCT) sex worker programs (SW) risk reduction for IDUs (RR) information education and communication (IEC) condom social marketing (CSM) treatment of sexually transmitted infections (STI) schools curricula prevention of mother-to-child transmission (PMTCT). For each intervention, the model permits the specification for each risk group of start and stop time and coverage (proportion reached per year) – i.e., prevention program scenarios. It allows the first year coverage to be lower, due to start-up issues. The model also permits specification of the maximum effectiveness in reducing each of the 3 risk behaviors and the average duration of those reductions, derived from scientific literature. These inputs are specified in full detail in the Intrvns worksheet. The aspects that reflect prevention program scenarios are specified more concisely in a resource allocation worksheet (currently in draft as Resource_Allocation_v3.xls). All these inputs are specified separately for HIV+ and HIV- individuals. This is important to examine the potential value of targeting by serostatus and also since there is evidence that prevention works differently according to serostatus. The conceptual approach to the effectiveness side of the prevention program scenarios is summarized in the following box. 51 Draft report: 05/08/2012 Prevention program scenario (coverage by strategy by risk group and time)  Strategy-specific reduction in 3 risk factors by risk group, and time  Aggregate reduction in 3 risk factors by risk group and time  Reduction in HIV incidence The resource allocation worksheet also calculates the cost of prevention, based on data from PANCEA. It allows the specification of prevention scenarios that respect explicit constraints on funding amount and allocation rules (e.g., minimum spending on IEC). It also allows greater or lower spending in the first year. Pending: The model saves the inputs and results from each prevention scenario for later reference. Pending: dealing with PMTCT HIV risk reduction associated with prevention (worksheets RR-RG1, etc) The model integrates inputs on prevention program scenarios and risk reduction effects to estimate the risk reduction expected for the three risk factors in each risk group. For each time period (one month), a certain portion of the risk group is exposed to the prevention intervention, and achieves the maximum reduction in each risk factor. In the next time period, those individuals retain most of the risk reduction effect (the maximum minus 1/average duration), and if the prevention intervention continues another portion of individuals receives the intervention. This results in rising cumulative risk reduction in the risk group, due to that intervention. Once the prevention intervention stops, risk reduction decays at the rate specified by the average duration. Since the risk group may be exposed to multiple interventions, with poorly understood interactions, the model permits alternate assumptions about intervention synergies. These include multiplicative effect, i.e., that each prevention intervention acts on the risk that remains after other interventions have had their effects. This is “negative synergy� because it reduces the absolute effects of interventions. Another assumption is additive effect, i.e., that each prevention intervention has the same absolute effect regardless of other interventions. Needs programming – Finally, the model allows for potentiation – that one intervention (such as IEC) makes another intervention (such as SW) more effective. This is “positive synergy.� The model also permits the specification of community coverage effects, such as increasing effectiveness once a critical threshold of community penetration is reached. These effects may be important but are not well understood from empirical data. The model allows exploring if different community coverage effects alter the relative value of different prevention scenarios. 52 Draft report: 05/08/2012 The resulting risk reductions for the three risk factors, for HIV+ and HIV- individuals, are specified for each time period. These values are used, when the prevention scenario is in place, to calculate HIV acquisition in the RG worksheets. Pending: The model allows the user to specify that a smaller number of individuals are reached at the beginning of program scale-up, reflecting delays in the capacity to serve clients and/or in the build-up of demand for services. Pending: dealing with PMTCT Predicted epidemic course (HIV infections expected) with and without prevention (worksheets Figs-Incid and Figs-RGs) The model presents predictions of epidemic course over time graphically, by risk group, as follows: Figs-Incid: HIV incidence rate in susceptibles HIV incidence – new infections per year HIV incidence – cumulative new infections Figs-RGs: Population size Number of susceptibles HIV prevalence (proportion) These graphs allow the user to understand and react to predictions of epidemic course in the base case, and to examine the effects of alternative prevention scenarios. 53 Draft report: 05/08/2012 Annex 4: Proposed Workshop Plan Proposed Agenda 1. Introductions and orientation to the briefing a) Introductions b) Goals and expectations; special interests and concerns of participants 2. Program design a) Solicit from program managers about their program focus and learn about their understanding of program design issues and trends b) Discuss implications 3. Orientation to local epidemic and resource allocation issues a) What do participants see as the key problems in responding to the HIV/AIDS epidemic? b) What are the possible implications of these problems for resource allocation? 4. Study orientation and tools a) General orientation to study o The potential contribution of cost-effectiveness data for allocating limited HIV prevention resources o Limitations of current cost-effectiveness information o Objectives and methods o General study findings b) Address resource allocation issues identified in 3 above c) Discuss how political or cultural factors may require deviation from the technically „optimal‟ solution d) Site reports for all intervention programs o Financial costs by category (personnel, buildings, etc.) o Economic costs by category o Output(s) o Cost/output(s) o Solicit response – Do these data match participants‟ own data? In what ways do the two diverge? Participants 1. Harm reduction program managers 2. Health bureau and provincial/local CDC staff 3. MOH and other government officials, etc. 54