Predictors of Short- and Long-term Survival in HIV-Infected Patients Admitted to the ICU: Materials and Methods
The Clinique de Reanimation des Maladies Infectieuses is an 18-bed ICU located in the Hopital Bichat-Claude Bernard, a 1,300-bed university hospital in Paris, France. Data were recorded for all HIV-infected patients admitted to our ICU during the 3-year period from January 1, 1990, to December 31, 1992. We evaluated only the first ICU admission in each patient. Following admission, each patient was assigned by one of the staff physicians to one of the following five groups based on the cause of admission (the main organ failure requiring ICU admission was considered): respiratory failure, hypoxia requiring oxygen supplementation and/or mechanical ventilation related to a pulmonary disease; neurologic failure, altered consciousness and/or seizures not related to drug overdose or to metabolic disturbances; heart failure, refractory hypotension and/or evidence of hypoperfusion due to cardiac dysfunction as assessed based on an echocardiogram or right heart catheterization study (Swan-Ganz catheter); severe sepsis, as previously defined by Bone, ie, refractory hypotension and/or peripheral signs of hypoperfusion with microbiological evidence of infection; and miscellaneous, all other causes.
We evaluated each patient’s health status based on the functional status during the month before ICU admission as assessed by a modified Karnofsky index, the World Health Organization score (0 = Karnofsky index of 100; 1 = Karnofsky index of 80 and 90; 2 = Karnofsky index of 60 and 70; 3=Karnofsky index of 40 and 50; 4=Karnofsky index of 20 and 30; and 5=Karnofsky index of 0 and 10), and by the percent weight lost (highest weight before HIV disease onset [if not available, ideal weight on insurance tables] minus weight at admission/highest weightX100). HIV-related variables included the risk factor for HIV infection, duration of HIV infection (from known seroconversion date or suspected date of seroconversion), time since first AIDS-defining condition (according to the 1987 Centers for Disease Control and Prevention (CDC) definition), duration of antiretroviral therapy, CD4+ lymphocyte count (most recent value obtained during the 3 months before), and number of previous opportunistic infections listed among the 1987 CDC AIDS-defining conditions. For the analysis, some variables were categorized as follows: age (^25, 26 to 35, >36 years), functional status (<2, >2), weight loss (<12%, >12%), HIV disease stage (AIDS diagnosed in ICU, AIDS diagnosed prior to ICU admission, non-AIDS), duration of AIDS in patients with a previous AIDS diagnosis (<360 days, >360 days), CD4+ count (<0.100, 0.101 to 0.200, 0.201 to 0.300, >0.301 X109/L), and number of previous opportunistic infections (0, 1 to 2, >3). Category ranges were selected according to the frequency distribution of the variables and possible clinical impact.