Background:  Hospitalization rates for AIDS-defining illnesses decrease after HAART initiation. It is unclear how hospitalization rates for other diagnoses change after HAART.

Methods:  This is a prospective observational study of 2031 HIV-infected patients followed by the Johns Hopkins AIDS Service who first initiated HAART from 1996 to 2005. Hospitalization rates and causes were examined before and after first HAART initiation. Modified Clinical Classification Software 2008 was used to assign the primary cause of each hospitalization to 1 of 18 diagnostic categories. Relative hospitalization rates were estimated using negative binomial regression.

Results:  The cohort was 34% female, 76% black, 45% injection drug users (IDU), and 27% men who have sex with men (MSM); and had a median age of 39 years (IQR 34 to 44). About half (54%) initiated HAART in 1996 to 1998. The median CD4 count and HIV-1 RNA level at HAART initiation were 152 cells/mm³ (37 to 298) and 4.8 log₁₀ copies/mL (4.1 to 5.3) respectively. In the overall period from 180 days prior to HAART through 365 days afterward, 38% of subjects had ≥1 hospitalization. Admissions due to any cause declined by 46 to 90 days after HAART (RR vs 180 days prior, 0.76 [0.62 to 0.93]). Non-AIDS-defining infections, AIDS-defining illnesses, psychiatric, gastrointestinal and liver, endocrine/metabolic, renal, and cardiovascular were the 7 most common diagnostic categories. Compared to the 180 days prior to HAART (figure), statistically significant decreases in hospitalization rates were seen at 91 to 180 days after HAART for AIDS-defining illnesses (RR, 0.60 [0.42 to 0.85]) and psychiatric illness (0.47 [0.26 to 0.83]), and at 181 to 365 days for AIDS-defining illnesses (0.46 [0.33 to 0.63]), psychiatric illness (0.60 [0.37 to 0.98]), and non-AIDS-defining infections (0.72 [0.55 to 0.94]). Virologic data to determine whether a subject had a significant response to HAART (≥ 1 log₁₀ decrease in HIV-1 RNA within 6 months) were available for 1327 subjects. Among this group, virologic responders showed significant decreases in AIDS-defining illnesses and non-AIDS-defining infections after 90 days, while non-responders did not.

Conclusions:  Hospitalization is common, and infections (both AIDS-defining and non-AIDS-defining) are the reasons for the majority of hospitalizations within the period surrounding HAART initiation. Providers should continue to be aware of the high risk of infectious illness through at least 1 year following HAART initiation, although decreases in risk of both infectious and psychiatric admissions may occur within 3 to 6 months.