AJPH RESEARCH
Consistent Estimates of Very Low HIV Incidence Among People Who Inject Drugs: New York City, 2005–2014 Don C. Des Jarlais, PhD, Kamyar Arasteh, PHD, Courtney McKnight, DrPH, Jonathan Feelemyer, MS, Aimée N. C. Campbell, PHD, Susan Tross, PHD, Lou Smith, MD, Hannah L. F. Cooper, PHD, Holly Hagan, PHD, and David Perlman, MD Objectives. To compare methods for estimating low HIV incidence among persons who inject drugs. Methods. We examined 4 methods in New York City, 2005 to 2014: (1) HIV seroconversions among repeat participants, (2) increase of HIV prevalence by additional years of injection among new injectors, (3) the New York State and Centers for Disease Control and Prevention stratified extrapolation algorithm, and (4) newly diagnosed HIV cases reported to the New York City Department of Health and Mental Hygiene. Results. The 4 estimates were consistent: (1) repeat participants: 0.37 per 100 person-years (PY; 95% confidence interval [CI] = 0.05/100 PY, 1.33/100 PY); (2) regression of prevalence by years injecting: 0.61 per 100 PY (95% CI = 0.36/100 PY, 0.87/100 PY); (3) stratified extrapolation algorithm: 0.32 per 100 PY (95% CI = 0.18/100 PY, 0.46/100 PY); and (4) newly diagnosed cases of HIV: 0.14 per 100 PY (95% CI = 0.11/100 PY, 0.16/100 PY). Conclusions. All methods appear to capture the same phenomenon of very low and decreasing HIV transmission among persons who inject drugs. Public Health Implications. If resources are available, the use of multiple methods would provide better information for public health purposes. (Am J Public Health. 2016;106:503– 508. doi:10.2105/AJPH.2015.303019)
“C
ombined prevention” for HIV, defined as using complementary evidence-based interventions implemented on a large scale, has generated considerable optimism for controlling the HIV epidemic. The Joint United Nations Programme on HIV/AIDS has announced an aspirational goal of “getting to zero” new HIV infections.1 In 2011, Secretary of State Hillary Clinton proclaimed a goal of achieving an “AIDS-free generation” with “very few new HIV infections.”2 This goal was reinforced in the White House directive on HIV and AIDS in the United States.3 New York State has an ending the epidemic initiative, with the goal of reducing newly diagnosed cases of HIV from 3000 in 2012 to 750 by 2020.4 Washington State has also announced an initiative to “end the AIDS epidemic” that includes reducing the rate of new HIV diagnosis by 50% by 2020.5 Although the optimism for reducing HIV transmission is justified, it would seem unlikely
March 2016, Vol 106, No. 3
AJPH
that HIV transmission could be fully eliminated in any large population of persons at high risk for HIV. Even with high levels of coverage of multiple interventions, neither HIV prevalence nor risk behavior have been completely eliminated in any high-risk population, and social structural drivers of risk behavior— including poverty and stigmatization—are difficult to modify. Thus, an appropriate public health goal may be to create a stable situation with close to zero new HIV infections.
Developing accurate measures of HIV incidence when the rate is low is important with respect to both the absolute numbers of new HIV infections and the trends in new HIV infections. Even in a low HIV incidence situation, differences in HIV incidence may be of great importance for large at-risk populations. For example, the population of persons who inject drugs (PWID) is estimated to be 1 million or more in the United States, China, and Russia.6,7 Thus, an HIV incidence of 0.5 per 100 person-years (PYs) at risk would generate 5000 new HIV infections among PWID annually, whereas an HIV incidence of 1 per 100 PYs would generate 10 000 new HIV infections per year in each of these countries. Assessing trends at low HIV incidence is also of great importance. Has a stable low incidence situation been reached, are significant further reductions in HIV incidence possible, or might incidence actually be increasing? It is very difficult to measure low rates of HIV incidence. Standard cohort studies are extremely expensive because measuring low incidence and participation in an ethically conducted cohort study would include exposure to repeated HIV testing and other risk reduction interventions, so the cohort participants would become less and less representative of the underlying population. Newly diagnosed cases of HIV infection
ABOUT THE AUTHORS Don C. Des Jarlais, Kamyar Arasteh, Courtney McKnight, Jonathan Feelemyer, and David Perlman are with the Baron Edmond de Rothschild Chemical Dependency Institute, Mount Sinai Beth Israel, New York, NY. Aime´e N. C. Campbell and Susan Tross are with the Department of Psychiatry, Columbia University, New York, NY. Lou Smith is with the New York State Department of Health, Albany. Hannah L. F. Cooper is with the Department of Behavioral Sciences and Health Education, Rolling School of Public Health, Emory University, Atlanta, GA. Holly Hagan is with the College of Nursing, New York University, New York. Correspondence should be sent to Don C. Des Jarlais, PhD, Director of Research, The Baron Edmond de Rothschild Chemical Dependency Institute, Mount Sinai Beth Israel, 39 Broadway, 5th floor, New York, NY 10006 (e-mail:
[email protected]). Reprints can be ordered at http://www.ajph.org by clicking the “Reprints” link. This article was accepted November 27, 2015. doi: 10.2105/AJPH.2015.303019
Des Jarlais et al.
Peer Reviewed
Research
503
AJPH RESEARCH
can also be used to measure trends in HIV incidence, but identifying new cases depends heavily on the HIV testing policies and practices in the community and the willingness of persons at risk to participate in HIV testing. Testing policies and practices and willingness to be tested may change over time. For example, Chapter 308 of the Laws of 2010 in New York State specifies that all persons presenting for primary care and inpatient or emergency care be offered HIV testing, with only a few exceptions.8 We examined 4 methods for estimating HIV incidence among PWID in New York City from 2005 to 2014. New York City experienced the largest HIV epidemic among PWID of any city in the world,9 but implementation of combined prevention (needle and syringe programs, medication-assisted treatment, behavior change interventions, HIV testing, and antiretroviral treatment) has effectively reduced HIV transmission to incidence rates that are now quite challenging to measure.
METHODS We examined 4 methods for measuring HIV incidence in New York City 2005– 2014: (1) HIV seroconversions among repeat participants, (2) increase of HIV prevalence by additional years of injection among new injectors, (3) the New York State and Centers for Disease Control and Prevention stratified extrapolation algorithm, and (4) newly diagnosed HIV cases reported to the New York City Department of Health and Mental Hygiene.
Incidence Methods 1 and 2 We derived the data for the HIV seroconversions among repeat PWID participants and the slope of the curve of HIV prevalence by years injecting for PWID participants from the same long-running research study that collects data from patients entering the Mount Sinai Beth Israel drug detoxification and methadone maintenance programs in New York City. The methods for this risk factors study have been described in detail previously,10,11 so we have presented only a summary. The programs serve New York City as a whole. There were no changes in the
504
Research
Peer Reviewed
Des Jarlais et al.
requirements for entrance into the program over the study period. Patients entering the detoxification and methadone maintenance programs were eligible to participate in the study if they reported injecting drugs in the 6 months before the interview. We checked hospital records and the questionnaire results for consistency regarding route of drug administration and examined participants for physical evidence of injecting. In the detoxification program, research staff visited the general admission wards of the program in a preset order and examined all intake records of a specific ward to construct lists of patients admitted within the previous 3 days. We then asked all patients on the list for the specific ward to participate in the study. After we asked all the patients admitted to a specific ward in the 3-day period to participate and conducted interviews among those who agreed to participate, the interviewer moved to the next ward in the preset order. As there was no relationship between the assignment of patients to wards and the order in which the staff rotated through the wards, these procedures should produce an unbiased sample of persons entering the detoxification program. In the methadone program, we asked newly admitted patients (those admitted in the previous month) to participate in the research. We paid participants $20 for their time and effort. In both programs, willingness to participate has been high, with approximately 95% of those asked agreeing to participate. A trained interviewer administered a structured questionnaire covering demographics, drug use, sexual risk behavior, and use of HIV prevention services. Most drug use and HIV risk behavior questions referred to the 6 months before the interview and thus before entry into the drug treatment programs. After completing the interview, each participant saw a counselor for HIV pretest counseling and serum collection. HIV testing was conducted at the New York City Department of Health Laboratory using a commercial, enzyme-linked, immunosorbent assays (ELISA) test with Western blot confirmation (BioRad Genetic Systems HIV-1–2+0 ELISA and HIV-1 Western Blot, BioRad Laboratories, Hercules, CA).
Participants were permitted to participate multiple times in the study although generally only once in any calendar year. The data used for incidence method 1 are from PWID who participated multiple times in the study between 2005 and 2014. We matched participants on name, drug treatment program identification number, gender, and date of birth. We used the number of participants who were HIV seronegative at their first study participation and then were HIV seropositive at a later participation as the numerator for calculating HIV incidence. The denominator was the total time between first and last participation for participants who remained seronegative through multiple participations plus one half of the time between the last seronegative participation and the first seropositive participation of the participants who did seroconvert. (We determined the time at risk for those who are HIV seropositive by assuming that seroconversion occurred midway between last seronegative participation and first seropositive participation.) We calculated 95% confidence intervals (CIs) for incidence values by using the binomial test for calculating exact CIs. The questionnaire included a question about age at first injection, permitting us to calculate the number of years the participant had injected drugs before the time of the interview. We used Poisson regression to model the relationship between HIV prevalence by years injecting as an estimate of the HIV incidence after first injection (incidence estimation 2). To avoid problems with disproportionate loss of those who are HIV seropositive, we restricted this estimation to participants who had been injecting for 6 years or less by the time of the interview. We estimated incidence as the increase of HIV prevalence by additional years injecting and the CIs for estimated incidence as the 95% CIs around the slope of the regression line. We transformed the values from the Poisson regression model to display them in the original units.12,13 Although we made method 1 and method 2 estimates with data from the risk factors study there are important differences in these 2 methods: the HIV seroconversions used in method 1 occurred between the first participation in the study and later (repeat) participations, whereas HIV seroconversions in the second method all occurred before the
AJPH
March 2016, Vol 106, No. 3
AJPH RESEARCH
first participation in the study. Of the 688 total participants in the 2 methods, we included only 69 (10%) in both methods and we did not include any of the seroconversions that occurred in both methods.
Incidence Method 3 The New York State Department of Health estimated annual numbers of incident cases and 95% CIs for the numbers of incident cases using the stratified extrapolation algorithm methodology developed by the Centers for Disease Control and Prevention.14–16 The Centers for Disease Control and Prevention funds the New York State Department of Health Wadsworth Laboratory to conduct national BED HIV-1 capture enzyme immunoassays of residual samples of persons who have tested HIV seropositive. These results are then returned to local jurisdictions to be used with the stratified extrapolation algorithm for estimating the numbers of incident cases of HIV in each jurisdiction in each year.17 The stratified extrapolation algorithm stratifies BED-positive cases by gender, race/ ethnicity, age, and transmission category. The large number of potential strata can, however, lead to very small numbers of events (BEDpositive tests) in some strata, so local epidemiological judgment may be needed in determining how strata should be combined. The characteristics of the BED testing and information on HIV testing behavior among members of the different strata (likelihood of testing, likelihood of testing multiple times, likelihood of having AIDS when testing) are then used to model the number of incident infections that would have been detected if BED testing had been used with all members of each stratum (including those who did not test for HIV in a particular year). Multiple imputation is used when there are insufficient data. CIs for the estimated numbers of incident HIV infections are calculated using the delta method. These CIs reflect uncertainty in the modeling calculations but do not reflect uncertainty in the model assumptions.14 To transform the number of incident infections in each year to rates per 100 PYs at risk, we divided the estimated numbers of incident infections and the 95% CIs by the
March 2016, Vol 106, No. 3
AJPH
estimated number of PWID in New York State.18 To obtain an estimate for the full period, we calculated the mean for the various years and the mean of the upper and lower 95% CIs. Because of concerns with small numbers of events in the various strata, only estimated incident infections for combined PWID and men who have sex with men who inject drugs for the state as a whole can be released publicly.
Incidence Method 4 The HIV Epidemiology and Field Services Program of the New York City Department of Health and Mental Hygiene conducts surveillance for HIV in New York City.19 The HIV Surveillance Registry contains information on HIV and AIDS diagnoses and HIV-related laboratory results ordered by New York City medical providers, including CD4 cell and viral load counts, and is continuously updated with new de-duplicated data. Surveillance field staff obtain demographic, clinical, and transmission risk information for newly diagnosed persons through chart review. The Field Services Unit attempts to interview all persons with new HIV diagnoses to assist in placing them in HIV care and to collect information on potentially exposed partners. The New York City Department of Health and Mental Hygiene publishes annual reports of the numbers of newly identified cases of HIV infection in total and by transmission risk category.19 The data we used did not include new diagnoses of HIV among PWID with concurrent AIDS (who had an AIDS diagnosis within 31 days of their HIV diagnosis), because these cases would not represent new infections. We divided the numbers of newly identified cases of HIV among PWID by the estimated number of PWID in the New York City metropolitan statistical area to generate incidence per 100 PYs at risk.18 The number of newly identified HIV infections among PWID should be considered a count within the population of PWID in the New York City metropolitan statistical area. Data for estimating SEs for newly identified cases or for the PWID population estimates were not available, so we calculated 95% CIs on the basis of the proportion of newly identified cases among the estimated
population with an assumption of a very large population.
Trend Analyses To examine possible trends over the 2005 to 2014 period, we compared the HIV incidence estimates for years 2005 to 2010 to the incidence estimates for years 2011 to 2014. We selected 2010 to 2011 as the midpoint because data for 2005 and 2006 were not available for several estimation methods. We used Stata version 12.0 for statistical analyses.20
RESULTS The 4 methods were consistent in capturing the phenomenon of very low and decreasing HIV transmission among PWID.
Incidence Method 1 Table 1 presents demographic characteristics of the PWID who participated multiple times in the risk factors study between 2005 and 2014. The majority were male (91%), more than half were Hispanic (53%), and a majority were injecting heroin (89%)
TABLE 1—Demographic and Drug Use Characteristics of People Who Inject Drugs With Repeat Interviews Entering the Drug Detoxification and Methadone Maintenance Programs at Beth Israel Medical Center: New York City, 2005–2014 Characteristic
No. (%) or Mean 6SD
Total
132 (100)
Gender Male
120 (91)
Female
12 (9)
Race/ethnicity White
30 (23)
Black Hispanic
26 (20) 70 (53) 39 69
Age, y Drug injected Heroin
118 (89)
Cocaine
56 (42)
Speedball
52 (39)
Injects daily
101 (77)
Des Jarlais et al.
Peer Reviewed
Research
505
AJPH RESEARCH
and injecting daily or more frequently (77%) in the 6 months before the interview. Substantial minorities reported injecting cocaine by itself (42%) and speedballs (cocaine and heroin injected together, 39%). A modest percentage (5%) reported being a man who has sex with men in the 5 years before the interview. For the 2005 to 2014 period, there were 2 seroconversions in 543 PY at risk for an HIV incidence rate of 0.37 per 100 PY (95% CI = 0.05/100 PY, 1.33/100 PY). For the 2005 to 2010 period, there were 2 seroconversions in 161 PY, for an incidence rate of 1.24 per 100 PY (95% CI = 0.15/100 PY, 4.42/100 PY). For the 2011 to 2014 period, there were no seroconversions in 382 PY, for an incidence rate of 0.00 per 100 PY (95% CI = 0.00/100 PY, 0.96/100 PY). There was no statistically significant difference between 2005 to 2010 and 2011 to 2014 (P = .09).
Incidence Method 2 Table 2 presents demographic characteristics of the participants who had injected drugs for 6 years or less and were interviewed between 2005 and 2014. We used Poisson regression to estimate HIV incidence by years
TABLE 2—Demographic and Drug Use Characteristics of People Who Inject Drugs With £ 6 Years of Injection Entering the Drug Detoxification and Methadone Maintenance Programs at Beth Israel Medical Center: New York City, 2005–2014 Characteristic
No. (%) or Mean 6SD
Total
556 (100)
injecting. The estimated incidence (when transformed to original prevalence units) was 0.61 per 100 PY (95% CI = 0.36/100 PY, 0.88/100 PY). There were insufficient numbers of HIV seropositive new injectors for separate analyses for 2005 to 2010 and for 2011 to 2014.
Incidence Method 3 Table 3 shows the estimated HIV incidence among PWID as assessed by the New York State Department of Health using the Centers for Disease Control and Prevention stratified extrapolation algorithm methodology. Table 3 presents the year, the estimated absolute number of incident HIV-infected PWID in New York State, and the 95% CIs for those estimates. Table 3 also shows the estimated incidence rates, assuming 136 210 PWID in New York State. These included estimates for the combined numbers of PWID in the major population centers in New York State, including the Albany, Buffalo, Long Island, New York City, Rochester, and Syracuse metropolitan statistical areas. There were 3047 estimated incident HIV infections over 7 years, an average of 435 incident cases per year or 0.32 per 100 PY (with an average 95% CI = 0.18/100 PY, 0.46/100 PY). For 2011 to 2012, there was an average of 269 HIV incidence cases, for an incidence of 0.20 per 100 PY (95% CI = 0.01/100 PY, 0.30/100 PY).
Incidence Method 4 Table 4 presents newly diagnosed cases of HIV infection among PWID reported to the New York City Department of Health and Mental Hygiene HIV/AIDS surveillance bureau. We divided the number of newly diagnosed cases by the estimated 103 877 PWID in the New York metropolitan statistical area.18 An estimate for the city alone was not available, so we used the metropolitan statistical area estimate. Over the 2007 to 2013 period, there was an average incidence of 0.140 per 100 PY, and for 2011 to 2013 an average incidence of 0.082 per 100 PY.
DISCUSSION We examined 4 methods for estimating HIV incidence among PWID. Each method has its own strengths and weaknesses and none of them uses a truly representative sample of PWID at risk for acquiring HIV, so none can be considered a gold standard. Although we did not begin the analyses with an a priori definition of consistency and inconsistency, we examined several potential indicators that lead us to conclude that the 4 estimates were, indeed, consistent. 1. There were no outliers. For the entire 2005 to 2014 period, all estimates were within a narrow range from 0.13 per 100 PY to 0.61 per 100 PY. There was
TABLE 3—Estimates of HIV Incidence Among Persons Who Inject Drugs in New York State Using the Centers for Disease Control and Prevention Stratified Estimation Algorithm: 2005– 2014
Gender Male
423 (76)
Estimate Year
Female
132 (24)
Race/ethnicity White
212 (38)
Black
77 (14)
Hispanic
236 (42) 33 68
Age, y Drug injected Heroin
531 (95)
Cocaine
210 (38)
Speedball
190 (34)
Injects daily
412 (74)
506
Research
Peer Reviewed
Des Jarlais et al.
Estimate, No. (95% CI)
Incidence/100 PY (95% CI)
2006
655 (377, 933)
0.48 (0.28, 0.69)
2007
685 (437, 934)
0.50 (0.32, 0.69)
2008
463 (272, 653)
0.34 (0.20, 0.48)
2009
415 (224, 607)
0.31 (0.17, 0.45)
2010
292 (154, 429)
0.21 (0.11, 0.32)
2011
294 (147, 440)
0.22 (0.11, 0.32)
2012
243 (116, 370)
0.18 (0.09, 0.27)
Note. CI = confidence interval; PY = person-years. We included both injection drug users and men who have sex with men who are injection drug users, and we included all cases for New York State to protect confidentiality. Incident rate in cases per 100 PY are on the basis of estimated numbers of people who inject drugs in major metropolitan statistical areas in the state.
AJPH
March 2016, Vol 106, No. 3
AJPH RESEARCH
TABLE 4—Newly Identified Cases of HIV Infection Associated With Injecting Drug Use Reported to New York City Department of Health and Mental Hygiene HIV Surveillance: 2005–2014 Year
No. PWID
Estimated Incidence/100 PY (Estimated 95% CI)
2007
195
0.188 (0.162, 0.216)
2008
124
0.119 (0.100, 0.142)
2009
135
0.130 (0.110, 0.154)
2010
128
0.123 (0.104, 0.147)
2011
110
0.106 (0.088, 0.128)
2012
110
0.106 (0.088, 0.128)
2013
35
0.033 (0.024, 0.046)
Note. CI = confidence interval; PWID = people who inject drugs; PY = person-years. We excluded newly identified cases of HIV with simultaneous diagnoses of AIDS, because such cases would not represent recent infections. We also excluded men who have sex with men who are injection drug users. We estimated 95% CIs using numbers of newly identified cases as the sample numerator for a very large population.
overlap among all the 95% CIs for the 4 estimation methods. For the most recent years, 2011 and later, there was a very narrow range from 0.00 PY per 100 to 0.20 per 100 PY among the 3 estimates, with a best estimate of 0.01 per 100 PY. 2. The 3 methods for which we were able to examine trends all indicated declines over time. If any of the methods had produced an estimate that appeared to be an outlier or an estimate that indicated an increasing trend in HIV incidence, it would have been critical to determine if this was because of methodological bias or if this method might have been detecting an unexpected incidence problem in a subpopulation of PWID. The consistency of incidence estimates generated by the different estimation procedures suggests that all of them capture a situation of very low and decreasing HIV incidence among PWID in New York City. In the absence of evidence that any single method is better than the others, we suggest using multiple methods whenever feasible for public health decision-making. It is critical to note that—in the absence of high-coverage prevention programs— outbreaks of HIV among PWID have occurred despite low background prevalence levels in Athens, Greece,21 and, more recently, in Indiana.22 Prevention and care programs for HIV among PWID in New York City, therefore, clearly need to be continued. In
March 2016, Vol 106, No. 3
AJPH
New York City, there is an emerging group of new injectors who began with using opiate analgesics and have progressed to heroin injection.23,24 Reaching these new injectors with HIV prevention is a high priority for New York City.
Limitations Several limitations to this study should be noted. We did not attempt to distinguish between injecting-related and sexual acquisition of HIV among our participants. Our previous studies of hepatitis C virus (an indication of injecting-related risk) and herpes simplex virus type 2 (an indication of sexual risk) suggest that the majority of new HIV infections among PWID in New York City occur through sexual transmission.25,26 Distinguishing between injecting-related and sexual transmission would be useful in further focusing HIV prevention efforts in New York City. We did not have data from a standard cohort to compare with the 4 methods we examined. A cohort study would have been extremely expensive, and generalizing from an ethically conducted cohort study (in which the researchers would have been obligated to link participants to needed prevention services) to the underlying PWID population would be problematic. The 4 methods were not completely independent. Methods 1 and 2 use data from the same long-running serial cross-sectional study. Data from PWID entering this
treatment program, however, do track well with HIV data from PWID recruited from the community.27–29 Both methods 3 and 4 were on the basis of PWID who sought HIV testing in each year, and changes in HIV testing policies in New York during 2005 to 2014 might have helped generate parallel trends in these 2 methods. All methods relied on PWID who received HIV testing, and thus there may be small networks of PWID who avoid HIV testing among whom HIV transmission rates are higher. Surveillance of these individuals would be particularly challenging and expensive, but there will always be concerns that HIV might disseminate from these networks to larger groups.
Conclusions We examined 4 methods for estimating the low HIV incidence among PWID in New York City. The consistency among the 4 methods indicates that all capture the same underlying phenomenon of very low and declining HIV prevalence. There does not appear to be any key substantive reason for preferring any 1 method over the others. If resources are available, however, the use of multiple methods would provide better information for public health purposes. CONTRIBUTORS D. C. Des Jarlais conceptualized the research question and primary hypothesis. K. Arasteh provided statistical analysis support. C. McKnight managed data collection for the hospital. J. Feelemyer, A. N. C. Campbell, S. Tross, H. L. F. Cooper, H. Hagan, and D. Perlman provided expertise and assisted in writing the article. L. Smith provided the stratified extrapolation algorithm data for New York State. All authors reviewed the article and approved its final form.
ACKNOWLEDGMENTS This research was supported by the National Institutes of Health (grants R01 DA003574 and R01 DA035707) and the Center for Drug Use and HIV Research (grant P30DA01104). Sarah Braunstein, MD, of the New York City Department of Health and Mental Hygiene reviewed the section on data from the New York City Department of Health and Mental Hygiene.
HUMAN PARTICIPANT PROTECTION Mount Sinai Beth Israel’s institutional review board approved this study. The participants provided written informed consent.
REFERENCES 1. World AIDS Day Report 2011: How to Get to Zero: Faster. Smarter. Better. Geneva, Switzwerland: Joint United Nations Programme on HIV/AIDS; 2011.
Des Jarlais et al.
Peer Reviewed
Research
507
AJPH RESEARCH
2. Clinton H. Creating an AIDS-free generation. 2011. Available at: http://www.state.gov/secretary/ 20092013clinton/rm/2011/11/176810.htm. Accessed June 5, 2012.
City Department of Health and Mental Hygiene; 1982– 2013.
3. PEPFAR Blueprint: Creating an AIDS-Free Generation. Washington, DC: US Department of State; 2012.
21. Paraskevis D, Nikolopoulos G, Tsiara C, et al. HIV-1 outbreak among injecting drug users in Greece, 2011: a preliminary report. Euro Surveill. 2011;16(36): 19962.
4. Hughes C. NY deal with drug maker paves way for plan to end HIV. Times Union. March 19, 2014. 5. Winans J. Governor Inslee issues proclamation to end AIDS in Washington State by 2020. 2014. Available at: http://www.llaa.org/press-release/governor-insleeissues-proclamation-end-aids-washington-state-2020. Accessed December 29, 2015. 6. Mathers BM, Degenhardt L, Phillips B, et al. Global epidemiology of injecting drug use and HIV among people who inject drugs: a systematic review. Lancet. 2008;372(9651):1733–1745. 7. United Nations Office on Drugs and Crime. UNODC updates global estimates on injecting drug use and HIV among people who inject drugs. 2013. Available at: https://www.unodc.org/lpo-brazil/en/frontpage/ 2013/06/28-unodc-updates-global-estimates-oninjecting-drug-use-and-hiv-among-people-who-injectdrugs.html. Accessed September 15, 2015. 8. New York State Department of Health. HIV testing. 2015. Available at: http://www.health.ny.gov/diseases/ aids/providers/testing/index.htm#publichealthlaw. Accessed September 29, 2015. 9. Des Jarlais DC, Arasteh K, Friedman SR. HIV among drug users at Beth Israel Medical Center, New York City, the first 25 years. Subst Use Misuse. 2011;46(2–3):131–139. 10. Des Jarlais DC, Friedman SR, Novick DM, et al. HIV1 infection among intravenous drug users in Manhattan, New York City, from 1977 through 1987. JAMA. 1989; 261(7):1008–1012. 11. Des Jarlais DC, Arasteh K, Hagan H, McKnight C, Perlman DC, Friedman SR. Persistence and change in disparities in HIV infection among injecting drug users in New York City after large-scale syringe exchange. Am J Public Health. 2009;99(suppl 2):S445–S451. 12. Helsel DR, Hirsch RM. Statistical Methods in Water Resources Techniques of Water Resources Investigations. Reston, VA: US Geological Survey; 2002.
20. Stata Statistical Software, Version 12.0 [computer program]. College Station, TX: StataCorp LP; 2012.
22. Centers for Disease Control and Prevention. Community outbreak of HIV infection linked to injection drug use of oxymorphone—Indiana, 2015. MMWR Morb Mortal Wkly Rep. 2015;64(16):443–444. 23. Lankenau SE, Teti M, Silva K, Bloom JJ, Harocopos A, Treese M. Initiation into prescription opioid misuse amongst young injection drug users. Int J Drug Policy. 2012;23(1):37–44. 24. Mateu-Gelabert P, Guarino H, Jessell L, Teper A. Injection and sexual HIV/HCV risk behaviors associated with nonmedical use of prescription opioids among young adults in New York City. J Subst Abuse Treat. 2015; 48(1):13–20. 25. Des Jarlais DC, Arasteh K, McKnight C, Hagan H, Perlman D, Friedman SR. Using hepatitis C virus and herpes simplex virus-2 to track HIV among injecting drug users in New York City. Drug Alcohol Depend. 2009;101 (1–2):88–91. 26. Des Jarlais DC, Arasteh K, McKnight C, Hagan H, Perlman DC, Semaan S. Associations between herpes simplex virus type 2 and HCV with HIV among injecting drug users in New York City: the current importance of sexual transmission of HIV. Am J Public Health. 2011;101 (7):1277–1283. 27. Des Jarlais DC, Perlis T, Arasteh K, et al. HIV incidence among injection drug users in New York City, 1990 to 2002: use of serologic test algorithm to assess expansion of HIV prevention services. Am J Public Health. 2005;95(8):1439–1444. 28. Des Jarlais DC, Arasteh K, Perlis T, et al. Convergence of HIV seroprevalence among injecting and noninjecting drug users in New York City. AIDS. 2007;21 (2):231–235. 29. Des Jarlais DC, Perlis T, Friedman SR, et al. Declining seroprevalence in a very large HIV epidemic: injecting drug users in New York City, 1991 to 1996. Am J Public Health. 1998;88(12):1801–1806.
13. Land CE. Confidence intervals for linear functions of the normal mean and variance. Ann Math Stat. 1971;42(4): 1187–1205. 14. Hall HI, Song R, Rhodes P, et al. Estimation of HIV incidence in the United States. JAMA. 2008;300(5): 520–529. 15. Karon JM, Song R, Brookmeyer R, Kaplan EH, Hall HI. Estimating HIV incidence in the United States from HIV/AIDS surveillance data and biomarker HIV test results. Stat Med. 2008;27(23):4617–4633. 16. Prejean J, Song R, Hernandez A, et al. Estimated HIV incidence in the United States, 2006–2009. PLoS One. 2011;6(8):e17502. 17. The Local HIV Estimation Guide Version 3.0. Albany, NY: New York State Department of Health; 2012. 18. Tempalski B, Pouget ER, Cleland CM, et al. Trends in the population prevalence of people who inject drugs in US metropolitan areas 1992–2007. PLoS One. 2013;8(6): e64789. 19. HIV Surveillance & Epidemiology Program—HIV/AIDS Annual Surveillance Statistics. New York, NY: New York
508
Research
Peer Reviewed
Des Jarlais et al.
AJPH
March 2016, Vol 106, No. 3