911 Rising Mortality Rates among Recent Entrants to Care for HIV Infection W. Mathews*, E. Barber Univ of California at San Diego Med Ctr
Background: The study aims were 1) to estimate trends in survival of adults under care for HIV infection; 2) to identify risk factors for mortality among recent entrants; and 3) to explore the effect of missing CD4 data on mortality estimation.
Methods: Retrospective analysis of sequential cohorts of patients (pts) entering care for HIV infection at UCSD Owen Clinic between 1991-2002. Mortality was ascertained by search of the Social Security death index. Temporal trends in mortality were estimated using density estimation and Kaplan Meier analysis. Risk factors were identified using Cox regression models. Predictors of missing baseline CD4 were identified using logistic regression.
Results: Between 1991-2002, 5,246 pts entered care at the study clinic. Baseline characteristics included sex: female 12%; race/ethnicity: non-white 38%; risk factor: MSM 52%, IDU 18%; antiretroviral therapy experienced: 33%; hospitalization prior to first visit: 20%; median (IQR) age: 36.8 (32-42); and median (IQR) CD4: 226 (60-442). Unadjusted death rates (per 100 person yrs), after falling from a peak of 33.5 in 1991 and reaching a nadir in 1998, have risen by year of entry, as follows: 5.0 (1998), 6.2 (1999), 5.9 (2000), 8.7 (2001), and 16.7 (2002). Analyses stratified by CD4 at entry indicated that rising death rates observed since 1998 are confined to those entering with CD4<350 (rate ratio 1.3 [95% CI:1.1-1.5] for each yr subsequent to 1998). There were 294 deaths among 2,278 pts entering after 1997. For these recent entrants, in a Cox proportional hazards model stratified on year of entry, the following baseline characteristics were independently associated (p < 0.05) with mortality: IDU risk factor (HR 1.5), Hispanic race (HR 0.7), hospitalization prior to first clinic visit (HR 2.1), age (HR 1.2 per decade), missing prior antiretroviral history (HR 3.5), and CD4 at entry (reference category > 350: < 50 [HR 13.8], 51-200 [HR4.8], 201-350 [HR 1.7], missing baseline CD4 [HR 6.2]). In a multiple logistic regression model, factors associated with missing baseline CD4 included yr of entry, age, number of clinic visits, unknown HIV risk factor, and unknown antiretroviral history.
Conclusions: 1) Mortality rates among entrants to care have risen steadily since 1998, especially among those entering with missing or low CD4 counts; and 2) missing baseline CD4 is associated with important prognostic factors and should therefore be included in mortality models.
