Background:  The first step of CDC’s Serostatus Approach to Fighting the Epidemic (SAFE) is diagnosis of HIV infection in all infected individuals. Subsequent steps include linking HIV infected persons to care and prevention services, and supporting them in adhering to prescribed treatment and in adopting and sustaining HIV risk reduction behavior.  Success of this approach will depend on timeliness of HIV diagnosis.  Our objective was to describe frequency and predictors of late HIV diagnosis.

Methods:  We analyzed national HIV/AIDS surveillance data from 25 states that have conducted integrated HIV/AIDS surveillance since 1994.  HIV diagnosis was considered “late” if an AIDS diagnosis was made within one year of HIV diagnosis.  Data reported to CDC through June 2001 were adjusted for delays in reporting; mode of HIV exposure for cases initially reported without information on risk was statistically redistributed.  We analyzed cases diagnosed from 1994 through December 1999 among persons aged 13 years and older by age group, sex, race/ethnicity, and HIV exposure mode.  Summary Mantel-Haenszel odds ratios (OR_MH) were calculated for stratified analyses.

Results:  Among all HIV diagnoses, percentage late (43) remained relatively stable during 1994 through 2000.  Percentage late increased with age and ranged from 11 for persons aged 13-19 years to 58 for persons 50 years and older; was greater for men (46) than women (33); greater for whites (46) and Hispanics (49) compared with blacks (40); and greater for persons whose HIV exposure mode was male-to-male sexual contact (MMS) (46) or injection drug use (IDU) (44) compared with persons whose HIV exposure mode was heterosexual contact (HC) (35).  After adjusting for age, probability of late diagnosis was greater for men than women (OR_MH=1.5); greater for whites (OR_MH=1.3) and Hispanics (OR_MH=1.5) compared with blacks; and greater for persons exposed through MMS (OR_MH=1.6) and IDU (OR_MH=1.2) compared with persons exposed through HC.

Conclusions:  A substantial proportion of persons living with HIV received their diagnosis late–within the same year AIDS was diagnosed.  After controlling for age, substantial differences in probability of late diagnosis by sex, race/ethnicity, and HIV exposure mode were still observed.  To improve timeliness of HIV diagnosis, reasons for persistent delays in testing and potential interventions to increase testing among recently infected persons need to be identified.