Logistic regression models were developed to evaluate the mortality outcome. Unweighted and weighted distributions for independent variables were similar, thus unweighted values were used in logistic models for simplicity. Bivariate analyses were conducted to determine the association of potential predictors of in-hospital mortality. The following multivariate models were used to adjust for potential confounding and interaction: (1) patient characteristics; (2) patient characteristics and hospital admission source; and (3) patient characteristics and hospital characteristics. The goodness of fit of the multiple logistic regression models was assessed using the Hosmer-Lemeshow test. Data from the logistic regression analyses are presented as crude and adjusted odds ratios (ORs), with corresponding 95% confidence intervals (CIs) and p values. Statistical analyses were performed using a statistical software package (Stata, version 6.0; Stata Corp; College Station, TX).
In this 20% sample of US hospitalizations, representing 7,450,992 hospital admissions, there were 2,279 hospitalizations with a primary hospital discharge diagnosis of TB. The mean patient age on hospital admission was 50.2 years. Demographic characteristics of these patients, compared to all other hospital admissions (7,448,713 hospital admissions), are shown in Table 1. Unlike hospital admissions for primary reasons other than TB, the majority of hospital admissions for TB were in men (64%) and minorities (black, 27%; Hispanic, 25%; Asian or Pacific Islander, 14%; other, 7%). Additionally, half of the people admitted to the hospital for TB resided in zip code areas with a median income of < $35,000, and most had publicly funded health insurance or no documented health insurance (66%).
Table 1—Characteristics of Hospitalized Patients, NIS, United States, 2000
|Characteristics||TB Patients (n = 2,279)||All Other Patients (n = 7,448,713)||p Value|
|Private insurance or HMO||22.4||39.2|
|Income||< 0.001 §|