Physician Experience Predicts HIV-Related Mortality

September 1, 2002

SEATTLE-The number of HIV-infected patients that a physician has treated independently predicts HIV-related mortality in patients starting anti-retroviral therapy for the first time, Robert S. Hogg, PhD, said at the 9th Conference on Retroviruses and Opportunistic Infections (abstract 749W).

SEATTLE—The number of HIV-infected patients that a physician has treated independently predicts HIV-related mortality in patients starting anti-retroviral therapy for the first time,
Robert S. Hogg, PhD, said at the 9th Conference on Retroviruses and Opportunistic Infections (abstract 749W).

Dr. Hogg is program director of population health, British Columbia Centre for Excellence in HIV/AIDS, and associate professor of health care and epidemiology, University of British Columbia, Vancouver.

Dr. Hogg and his colleagues analyzed determinants of HIV-related and all-cause mortality in 1,219 HIV-infected, antiretroviral-naïve patients who were first prescribed triple-drug therapy between August 1996 and September 1999. Data for the patients and their approximately 100 treating physicians were obtained from the database of the British Columbia HIV/AIDS Drug Treatment Program.

Physician experience was assessed from the cumulative number of HIV-infected patients treated and was classified as above or below first quartile. Adherence to therapy was expressed as the number of months of medication dispensed divided by the number of months of follow-up during the first year of therapy, and was classified as intermittent if the value was less than 75%.

During an approximate 30-month follow-up, the rates of HIV-related and all-cause mortality were 6.7% and 8.5%, respectively. All of the non-HIV-related deaths were either suicides or accidental drug overdoses. At the start of the study, the median number of HIV-infected patients previously treated by the physicians was 47 (interquartile range, 7 to 133).

In multivariate analysis of HIV-related mortality, the risk of death was significantly decreased with physician experience above the first quartile (risk ratio, 0.54) and significantly increased with intermittent adherence (risk ratio, 3.83) and CD4 cell count below 200/µL (risk ratios at 50 to 199/µL and less than 50/µL, 3.61 and 7.29, respectively).

In multivariate analysis of all-cause mortality, the risk of death was significantly increased with intermittent adherence (risk ratio, 3.19) and CD4 cell count below 200/µL (risk ratios at 50 to 199/µL and less than 50/µL, 2.25 and 4.96, respectively). However, physician experience was not a significant predictor of all-cause mortality.

"The lack of an association between physician experience and all-cause mortality likely relates to confounding with unmeasured sociodemographic characteristics as well as to the sudden nature of accidental and overdose deaths," Dr. Hogg noted.

Use of protease inhibitors, a diagnosis of AIDS before enrollment, and a baseline plasma viral RNA level greater than 100,000 copies/mL were not significant predictors of either HIV-related or all-cause mortality in multivariate analysis.

The findings identify physician experience, intermittent adherence, and CD4 cell count as strong determinants of mortality in HIV-infected patients in the general population who are starting anti-retroviral therapy, Dr. Hogg said.

"Clearly, researchers must be cognizant of relationships like these prior to building models of prognostic factors associated with disease progression," he concluded.