THE INTERNATIONAL JOURNAL OF DRUG POLICY, VOL 7, NO 2,1996
EVALUATING A NEEDLE EXCHANGE PROGRAMME
MODELS FOR TESTING HIV- I RISK REDUCTION
Robert Heimer', Kaveh Khoshnood', P. Clay Stephens', Bini Jariwala Freeman', and Edward H. Kaplan"
Yale University School of Medicine, Department of Medicine,
Department of Epidemiology and Public Health
Yale University School of Management
We have analysed of the role of a legal needle xchange programme in reducing the spread of HIV- I infection among drug injectors through the investigation of the syringes they return to the programme. The clean syringes distributed by the programme cannot transmit the virus whereas contaminated ones may remain infectious for at least three weeks. Needle exchange removes used, potentially infectious syringes proportional to the extent that the exchange is used. In New Haven programme clients obtained 70% of their syringes from the programme and, as a direct result of programme operations, mean circulation time for syringes rapidly declined from more than 2 weeks to less than 3 days. There was a concomitant decline in the percentage of syringes with evidence of use by HIV infected injectors from an initial level of more than 65% to a new steady-state level of approximately 40%. The change in circulation time, but neither a change in the client base nor a testing artefact, could be correlated with this decline. Test results have been incorporated into several models to estimate the incidence rate among clients. These estimate a minimum of a 33% decrease in new transmissions among clients. In addition, the New Haven programme actively reduces HIV transmission risks by accelerating entry into drug treatment programmes
New Haven is a small city located 120 kilometers northeast of New York. Among its 132 000 inhabitants are an estimated 2250 injection drug users (IDUs) (Kaplan and Soloshatz, 1993). For these inhabitants, the risk of needle,borne infections is high, and sampling the population has yielded HIV- I prevalence estimates that, in 1990, ranged from 24% to more than 60% (Kaplan and O'Keefe, 1993). The recognition of the large scale of the problem led to an effort that, following a three year campaign, culminated in the legalization and implementation of a city-run needle exchange programme (NEP) (O'Keefe, 1991). For the last four and a half years, the programme has been operating out of a van that travels to at least five sites on each of four days per week. From the start of the programme in November 1990 through June 1993, approximately 1500 individuals visited the programme, more than 19 700 exchanges were made, and more than 80 200 clean syringes were exchanged.
An evaluation of the programme, mandated by the state legislation, was established through collaboration with researchers at Yale University and was in place at the programme's start. Standard seroepidemiological evaluation of the programme was impossible. Absolute anonymity and the absence of serological testingwere required by city officials who felt these intrusionswould diminish programme participation. The state mandate required a determina, tionof the programme's effect within one year. Rapid evaluation of changes in HIV- I incidence rates could not be obtained using standard epidemiotogicat practices because of the small number of clients and the lack of access to a 'control'group of nonclients. To surmount criticisms leveled at earlier studies of needle exchanges, evaluators hoped to develop a sensitive instrument that relied as tittle as possible on self-reported behavioral changes. The paramount concern of the evaluators was to discern if there were any direct programmematic links between the exchange operations and its results. These considerations led evaluators to develop a syringe tracking and testing system (Kaplan, 199 1). Each distributed needle received a unique tracking code and each client chose a pseudonyrq upon enrolling in the programme. Each client co~td use that pseudonym to obtain a new syringe with non detachable cannula for each syringe returned whether distributed by the programme or not- to a maximum of 5 (later 10) syringes. When syringes were exchanged, the pseudonym, the outgoing syringe codes, the date, and the exchange location were recorded. A database containing all the information about distributed needles has been maintained. The returned syringes were placed in a canister with the pseudonym of the client, the date, and location of the exchange, and the canisters were transported to the laboratory where information on all returned needles was compiled in a second database. The two databases comprise the syringe tracking system, which allows measurement of client programme use and syringe flow.
We then developed methods to test returned syringes to detect use by an HIV- I -infected IDU. We have relied upon a 'nested' polymerase chain reaction (PCR) technique that can detect, in the blood remaining in a used syringe, as little as a single copy of HIV- I proviral DNA from the white blood cells of infected individuals (Heimer, Myers, Cadman and Kaplan, 1992; Myers, Heimer, Henrard and Liu, 1993) Through August 1993, we used PCR to test 5954 returned syringes, and 2528 (42.5%) tested positive. At the programme's start, two-thirds of non-programme syringes that were returned tested positive, as did syringes returned to a weekly needle exchange that operated prior to the start of the legal programme (Table 1). Initially, returned programme syringes were equally contaminated. However, with, in 5 months the percentage of positive programme syringes declined to approximately 40%, a level sustained throughout the period from April 1991 to August 1993. The decrease was not restricted to programme syringes; we observed a similar decline in the percentage of contaminated non-programme syringes returned to the programme.
How do we account for the rapid and sustained decline in the percentage of syringes used by HIV- 1 + injectors? We have investigated four possible explanations. First, the decrease could have been due to laboratory artifact. This possibility was excluded by using an antibody based enzyme immunoassay that is at least as sensitive as PCR (Myers, Heimer, Henrard and Liu, 1993). Prevalences of HIV- I proviral DNA and HIV, I specific antibodies were not statistically different in samples of syringes returned during: the same interval (Heimer, Kaplan, O'Keefe, Khoshnood and Altice, 1994) Second, the [email protected] could have resulted from increased use of the programme by uninfected individuals. We looked at risk behaviour as a proxy for HIV- I infection, and have ruled out that changes in the demographics and behaviours of the client population account for the decrease inthe percentage of HIV- I positive syringes (Kaplan, Khoshnood, andHeimer, 1994). Third, the risky behaviours of clients could have been reduced. We analysed changes in behaviours using selfreported data obtained from two surveys covering a convenience sample of 18% of the participants who visited more than once, one administered upon enrollment and a fottow,up obtained 4 to 33 months later (mean 15 months, median 9 months). Although we detected decreases in syringe sharing and gallery use (Khoshnood, 1995), the magnitude of the changes could in no way account for the 40% decrease in HIV- I+ syringes.
The changes that did account for the decrease were changes in the behaviors of the syringes detected by the syringe tracking system (Kaplan and Heimer, 1994). The percentage of programme syringes returned increased from 27% in November 1990 to a level exceeding 60% at the end of the five month period in which the percentage of positive programme syringes fell from 63% to 40%. Concomitantly, the average length of time between distribution and return of programme syringes - the circulation time - decreased from more than 7 to less than 3 days. Following this period of rapid changes in syringe prevalence, return rate, and circulation time, these three parameters have remained more or less stable. Mathematical models have been elaborated that examine the rote of decreased circulation time in decreasing syringe prevalence (Kaplan and Heimer, 1994). In fact, it is possible to estimate the monthly syringe prevalence using the circulation time data (Figure 1). The intuitive meaning of these findings is that as syringes circulate more quickly, theyhave less opportunity to come into contact with HIV- I + injectors. This also means that anyone using these syringes - programme and non-programme alike - have a 40% reduced risk of encountering a syringe used by an infected injector.
The decrease in circulation time has importance if syringes, once contaminated, remain so for extend, ed periods unless adequately disinfected. To investigate this question, we carried out simulations to determine the duration of HIV- I viability within syringes (Stephens, 1995). Blood from HIVI negative donors was infected in vitro with I of 2 clinical isolates of HIV- 1 and < 10pl of blood was introduced into syringes similar to those distributed by the New Haven NEP. The syringes were stored at room temperature for up to 28 days, whereupon the blood was extracted and combined in tissue culture with stimulated peripheral blood mononuclear cells which served as targets for HIV- 1 infection. The viability of virus recovered from syringes was measured by the production of HIV- 1 p24 antigen in the infected cells maintained in the cultures. The use of replicate syringes and seriat dilution of the syringe extracts has permitted the quantification of the titers of HIV- I recovered following each interval of storage (Hollinge, 1993). Results of the culture experiments revealed that the virus remained viable forperiodsfar in excess of the mean circulation times of programme syringes. Both strains of HIV- 1 remained viable for at least 21 days and, with one strain, viable HIV- 1 was recovered after 28 days of storage (Figure 2).
So far, we have been talking about syringes, but what about people? We have developed two models that employ the data from the syringe tracking and testing system to estimate the impact of the NEP on HIV, I incidence among programme participants. The first approach is a projection of circulation theory from prevalence in syringes to incidence in programme clients. We used two methods - the first using the prevalence = incidence x duration law and the second employing back-calculation - to calcu, late baseline incidence, and both yielded similar results (Kaplan, 1994). In both cases, we assumed that 30% of all infections among IDUs were sexually acquired. Then we adjusted for a one,third reduction in positive syringes and 200 clients per month to yield an absolute range of I to 3 infections prevented per 100 clients per year for the period from November 1990 through April 199 1. The 40% reduction over the period from May 1991 through June 1992 yielded a range of 1.5 to 4.5 infections prevented per year, assuming an average programme population of 250. For the 14 months beginning July 1~992, where the programme population was only 150, the range would be 0.9 to 2.7 infections prevented per year. So, the estimate of programme effect over 30 months acknowledging that sexual transmissions may not have been effected (so we assume no effect on sexual transmission) - is 3.2 to 9.7 infections averted (Table 2). This estimate is conservative in that it assumes no changes in client behaviors and imputes a substantial percentage of the new infections to sexual transmission. If clients, through access to increasing availability of clean syringes, shared less often, or if sexual transmission has been overestimated, then even more infections might be averted.
The second approach to estimate programme efficacy used only the syringe tracking and testing system to evaluate test results on sequences of client syringes. Any client who had 5 or more syringes tested during the course of random PCR testing could provide meaningful data to this analysis. The approach assumed sequential syringe testing as a surrogate for sequential serological testing. In an ideal world - one without syringe sharing and one with an infinitely sensitive syringe testing methodology sequential syringe testing would be identical to sequential serological testing. In this ideal world, an uninfected person would return negative syringes until he became infected whereafter he would return positive syringes. But syringe sharing with uninfected people and the lack of a test sensitive for cleaned syringes can introduce false negative results while syringe sharing with infected people and false positive testing can introduce false positive results into the sequence. Nevertheless, it stands to reason that the probability of returning a positive syringe increases after infection. We constructed a changepoint model in which each sequence was analysed for the appearance of a significant change in the propor, tion of positive syringes at some point in the sequence (Kaplan and Heimer, 1994). Maximum likelihood analysis was then used to estimate the incidence rate that best accounted for the number of changepoints observed. The initial analysis, involving 132 clients, 1,920 syringe tests, and 95.6 programme years, yielded a maximum likelihood incidence rate equal to 0 new infectionsper 100 persons per year (95% confidence interval = 0-10.2). We recently re-performed this analysis with 23 1 clients, 4134syringe tests, and 289 programme years; the best treatment for crack cocaine or alcohol problems.
Our evaluation, based upon a syringe tracking and testing system, has demonstrated a significant reduction in new HIV- I infections among programme partic ipants. The success of the programme goes beyond this single accomplishment. As discussed by Watters et al., 1995), there is a strong association between the provision of integrated services and the degree of harm reduction achieved. By this criterion, the cityrun programme has met larger goals in reducing the harm associated with injection drug use:
1. by slowing the spread of HIV-1;
2. by providing clients with harm reduction information and access to safer-sex and safer-injection supplies;
3. by increasing the access of clients and non clients to drug treatment,
4. by increasing community access to medical care.
Financial support was provided by theNational Institute on Drug Abuse (DA-076761 and the Robert Wood Johnson Foundation (20049). Nis. Stephens is supported by an institutional training grant from the Agency for Health Care Policy and Research.
Dr Robert Heimer, Department of Internal Medicine, Yale School ofMedicine, 333 CedarStreet, NewHaven, CT06520, USA.
Heimer R, Lopes M (1994). Syringe and needle exchange to prevent HIV infection (letter). JAMA 271:1825-6.
Heimer R, Kaplan EH, O'Keefe E, Khoshnood K, A Itice F 0 994). Three years of needle exchange in New Haven:
What have we learned? AIDS and Public Policy journal 9: 59-74.
HeimerR, Myers SS,CadmanEC,KaplanEH (I 992)Detection by polymerase chain reaction of HIV- I proviral DNA sequences in needles of injecting drug users. Journal of Infec, tious Diseases 165: 781-2.
Hollinger FB (I 993)~ ACTG Virology Manual for HIV Laboratories Division of AIDS, NIAID, NIH, Bethesda, MD, version 2.0, January 1993.
Kaplan EH (1991). Evaluating needle exchange programmes via syringe tracking and testing. AIDS and Pubic Policy journal 6:109-15.
Kaplan EH (1994) Operational modeling of needle exchange programmes, in Workshop onNeedle Exchange and Bleach Distribution Programs. Washington, DC, National Research Council and the Institute of Medicine. National Academy Press. pp. 202-49.
Kaplan EH, Heimer R (1994). A circulation theory of needle exchange. AIDS 8: 567-74,
Kaplan EH, Heimer R (1994). HIV incidence among needle exchange participants: Estimates from syringe tracking and testing. journal of Acquired Immune Deficiency Syndrome 7:182-9.
Kaplan EH, O'Keefe E (1993). Let the needles do the talking! Evaluating the New Haven needle exchange. Interfaces 23(0:7-26.
Kaplan EH, Soloshatz D (1993). How many drug injectors are there in New Haven? Answers from AIDS data. Mathematical Computer Modeling 17:109-115.
Kaplan EH, Khoshnood K, Heimer R 0 994). Client shift or needle exchange: What caused the HIV prevalence drop in needles returned to New Haven's needle exchange programme? American journal of Public Health 84:1991-4.
Khoshnood K (1995). Client-based evaluationof New Haven's needle exchange programme. PhD Thesis, Yale University, New Haven, CT.
Myers SS, Heimer R, Henrard D, Liu D (1993). HIV- I DNA and antibodies in syringes from injecting drug users: a comparison of techniques. AIDS 7: 925-31.
O'Keefe E (1991). Altering public policy on needle exchange: the Connecticut experience. AIDS and Public Policy 6: 159-64.
O'Keefe E, Kaplan EH, Khoshnood K (1991). Preliminary Report: City ofNew Haven Needle Exchange Program,City of New Haven, New Haven, CT; 31 July 199 1.
Stephens PC (1996). Studies in HIV- I Transmission in Women, PhD Thesis, Yale University.
Watters JK, Bluthethal RN, Kral AH (1995). HIV seroprevalence in injectiondrug users (letter). JAMA 273:1178