Background:   The use of population-based estimates of HIV incidence can be an important supplement to existing surveillance data for monitoring changes in HIV transmission over time.  The Serologic Testing Algorithm for Recent HIV Seroconversion (STARHS) method can distinguish between recent and long-term HIV infections. We applied STARHS to cases reported to HIV surveillance and examined trends in HIV incidence over the first 4.5 years of named HIV reporting in NYC.

Methods:  Persons diagnosed during June 1, 2000 through December 31, 2004, reported to NYC HIV surveillance through September 30, 2005, and who had remnant serum available at the NYC Public Health Laboratories (n = 3464 or 20.4% of all non-AIDS HIV diagnoses) were re-tested using a less-sensitive bioMerieux Vironostika enzyme immunoassay. The population consisted of all testers at HIV testing sites submitting specimens to the NYC public health laboratory for a voluntary, confidential HIV test during this period (n = 428,264). Incidence rates among the testing population for each semester were calculated with 95% confidence intervals.

Results:  Incidence was 0.48%/year (95%CI 0.36 to 0.60%) in June to December 2000.  It was then stable in 2001-2003 (0.31%/year, 0.22 to 0.40%) in the first half of 2001, 0.38%/year (0.28 to 0.49%) in the second half of 2003. There was a significant decrease in 2004 to 0.24%/year (0.16 to 0.32%). Incidence remained significantly higher among males than females throughout, starting at 0.82%/year (0.65 to 0.99%) for males vs 0.35% (0.25 to 0.45%) for females in 2000 and ending at 0.60% (0.46 to 0.74%) for males vs 0.07% (0.03 to 0.11%) for females in the second half of 2004. Incidence in men who have sex with men (MSM) remained stable beginning at 4.3%/year (3.7 to 4.8%) in 2000 and ending at 4.2%/year (3.7 to 4.7%) in June through December 2004. Among injecting drug users (IDU), incidence significantly decreased from a high of 2.7%/year (2.3 to 3.0%) in June to December of 2000 to a low of 0.9%/year (0.7 to 1.0%) during July to December 2004.

Conclusions:  These data suggest that the overall HIV incidence rate has decreased over the first 4.5 years of named HIV reporting in NYC. These data also suggest a stable incidence rate among MSM testers at around 4% per year. Incidence among IDU testers in NYC appears to have significantly decreased throughout this period. It is too soon to ascertain whether the changes observed in 2004 signal the beginning of a new trend or are a reflection of a changing testing population with the growing use of rapid testing at that time.