Harm reduction for people who inject drugs in New York has worked, but hasn’t reduced racial inequality

Needle and syringe exchange and opiate substitution therapy in New York City has worked in reducing HIV infection in people who inject drugs (PWIDs) to the extent of almost abolishing HIV infection in white people who inject drugs, a recent study published in PLOS One shows.

But HIV prevalence in black people who inject drugs remains high. The study finds that this is probably because now most people with HIV who present to healthcare and are categorised as people who inject drugs have actually acquired HIV through sexual contact, before they ever started injecting drugs. In other words, the HIV in non-white people who inject drugs is not due to the ‘risk class’ they are categorised in, and young African-American and Latino people who are at risk of starting to inject drugs need access to measures such as pre-exposure prophylaxis (PrEP) and increased HIV testing and treatment when they are younger.


The study is the latest from an ongoing cohort study of people who inject drugs who attend the primary drug treatment centre in New York City, at Mount Sinai Beth Israel hospital in Brooklyn. The researchers were able to compare HIV prevalence in this study, which looked at prevalence in Mount Sinai attendees between 2007 and 2014, with a previous study looking at the years 1995 to 2008, and found evidence of a year-on-year reduction in HIV prevalence.

New York City experienced a large local HIV epidemic among people who inject drugs early in the epidemic, and introduced methadone maintenance programmes early, adding in large-scale needle and syringe exchange in 1995. The number of syringes exchanged in New York increased from 250,000 in the early 1990s to three million a decade later. Annual HIV incidence declined from 4% to 1% in people who inject drugs during this period.


multivariate analysis

An extension of multivariable analysis that is used to model two or more outcomes at the same time.

prospective study

A type of longitudinal study in which people join the study and information is then collected on them for several weeks, months or years. 


When the estimate from a study differs systematically from the true state of affairs because of a feature of the design or conduct of the study.

risky behaviour

In HIV, refers to any behaviour or action that increases an individual’s probability of acquiring or transmitting HIV, such as having unprotected sex, having multiple partners or sharing drug injection equipment.

Since 2007, after it was noted, partly due to the success of syringe exchange programmes, that more people who inject drugs were acquiring HIV sexually, free distribution of condoms was added to prevention programmes for people who inject drugs.

The study

The present study took repeat cross-sectional samples of people entering drug treatment at Beth Israel who reported injecting drugs in the last six months and who reported that they first started injecting after 1995, to include only the generation who could get syringe exchange. Participants could be included in the study several times, though only once in any one year. The study was not designed as a prospective cohort study, where participants were proactively followed-up, in order to preserve confidentiality and, if desired, anonymity, and also to eliminate possible bias caused by regular medical attention.

A total of 703 people who inject drugs were included in this study, of whom 42% were white, 46% Latino, and 12% African-American. The African-Americans were older (average age 42 as opposed to 32 for white people), more likely to be women (39% versus 21% for white people and only 16% Latina) and, although almost all participants reported injecting heroin (in the last six months), black participants were less likely to report doing it every day (54% versus 77%). Black participants were more likely than others to smoke crack or snort heroin. The prevalence of multi-person use of needles was roughly the same in all groups, at 20%, and so was unprotected sex with regular partners (48%) and casual partners (19%).

HIV prevalence was far higher in African Americans: it was only 1% in white people who inject drugs, which is actually lower than the 2% prevalence in the New York adult population. It was 4% in Latinos and 17% in African-Americans. Proportionally, women were more likely to have HIV than men, and this was not because more African Americans were women. In multivariate analysis that controlled for factors independently influencing HIV prevalence, HIV-positive status was 19 times more likely in African American people, and 4.4 times more likely in Latino people, than in white people and was also 3.2 times more likely in women than men. As we can see from the previous paragraph, this cannot be explained by rates of recent HIV risk behaviour, which were the same or lower in African-American people than in other groups.

Sixty per cent of people who inject drugs who had HIV also had hepatitis C virus (HCV), and hepatitis C prevalence was 54% among the group as a whole or 13.5 times greater than HIV prevalence. Using a model that feeds in the relative per-act transmission probabilities of HIV and HCV via sex and via injecting, the researchers found that these figures indicated that 75% of HIV infections had occurred through sexual transmission.

HIV prevalence appeared to be declining over time: it was 30% lower each year of the seven-year survey. It was not significantly larger, however, in people who had been injecting longer. This apparent paradox can be explained if most people with HIV who inject actually acquired HIV sexually before they started injecting, and then, by starting to inject, entered the group of people whose risk of HIV was falling due to the introduction of needle exchange.

This also had the effect of reinforcing the racial disparity in HIV prevalence over time. In the 2008 survey, African-American people were four times, and Latino people 1.5 times, more likely to have HIV than white people in multivariate analysis. By 2014, this had grown so that African-American people were 21 times and Latino people 4.5 times more likely to have HIV than white people (these figures differ slightly from above because more people are included in the data analysis). In other words, HIV infection during the time people were injecting had become less likely over time, but already having HIV when they started injecting had not.

More evidence supported the idea that most people in this group acquired HIV before they started to inject. The average age that people first started using drugs of any kind (other than marijuana) was 19 and the average age at first injection was 28, so people had plenty of time to acquire HIV sexually.

Thirty-five per cent of those with HIV were receiving antiretroviral treatment (ART). Only five people who had been injecting for less than five years had HIV but, of these, three were on ART, implying that their CD4 count when they started ART was below 350 cells/mm3 (the local New York CD4 count threshold for starting treatment at the time of interview). This also implies that they probably acquired HIV some time before they started injecting.


This latest instalment in a long-running study suggests, as the authors say, that “HIV infection among white PWIDs has almost been eliminated”. However to eliminate HIV in black and Latino PWIDs, programmes are needed that offer HIV testing and treatment, PrEP and other prevention help to young people in communities where drug injection is common, but before they ever start injecting.

It also implies that a person’s ‘risk class’ at the time they are diagnosed with HIV may not actually indicate how they acquired HIV, if they have transitioned from a higher to a lower risk environment before diagnosis.


Des Jarlais D et al. Will “combined prevention” eliminate racial/ethnic disparities in HIV infection among persons who inject drugs in New York City? PLOS One, DOI:10.1371./journal.pone.0126180. May 2015. See http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0126180 for full text.