Certain methodological limitations of the study must be considered. Several important factors that have been suggested as risk factors for TB mortality, such as homelessness, history of incarceration, injection drug use, multidrug-resistant TB, compliance with DOT, and delay in diagnosis, could not be ascertained directly by this administrative database. Specifically, missed diagnosis and delayed treatment after hospitalization have been shown to occur more often in hospitals with low TB hospital admission rates and were strongly associated with in-hospital death in Canada. The measure of income was an ecologic rather than patient-specific measure, which can lead to the misclassification of income status. Given that persons with TB tend to have relatively lower socioeconomic status (SES) than their demographic or geographic counterparts, assigning SES values to individuals based on geographic means may actually overestimate their income and, in turn, underestimate the association between SES and TB outcomes. In addition, the results of this study rely on the accuracy of the diagnosis codes. The NIS database does not include patient identifiers, thus validation of the accuracy of the hospital discharge records was not feasible. However, several stud-ies have successfully assessed patient outcomes using the NIS database.
The prevalence of HIV in our study population of hospitalized TB patients was 3.3%, which is lower than the national estimate of the HIV-TB coinfectiv-ity rate of 10% published by the Centers for Disease Control and Prevention. This finding may be due to a tendency to list HIV as the primary diagnosis when TB and HIV were coincident. In fact, this appears to be a plausible explanation, as the prevalence of HIV-TB coinfectivity reached 11.3% when we included patients with a primary diagnosis of HIV and a secondary diagnosis of TB. We focused on TB as a primary diagnosis in order to capture outcomes that are most likely attributable to TB illness. Nevertheless, in a secondary analysis (data not presented) we calculated the mortality rate of patients with a primary diagnosis of HIV and a secondary diagnosis of TB to assess whether they had different mortality rates than those patients in our study. The mortality rate was 4.8%, which is similar to that for patients with a primary diagnosis of TB. Thus, we do not believe that including these patients in our analysis would have changed our results significantly. Continue reading
Older age was a strong, independent predictor of mortality. Given the aging of the American population, mortality from TB in the elderly is an enormous concern. Mortality may be higher in older adults because they may receive less vigorous care, or older persons may have more severe disease because of a decreased immunologic status and decreased baseline functional status. Older people with TB have been shown to have more extensive disease, based on chest radiograph findings at presentation. Importantly, age may have a modifying effect on TB illness itself, making the diagnosis of TB more difficult. Older TB patients have a higher prevalence of nonspecific symptoms, a lower prevalence of fever, and less frequently manifest a positive tuberculin skin test. This less classic presentation may contribute to the longer delay in presentation and initiation of treatment, and may lead to a higher risk of death. Continue reading
While all people are susceptible to infection with TB, the majority of cases occur in men, minorities, and the socially disadvantaged. Our results are consistent with these previous epidemiologic data.
This study extends previous insights, and shows that men continue to be at greater risk and that most of the patients admitted to the hospital with TB were racial minorities, residents of regions with lowhouse-hold incomes, and had publicly funded or no health insurance. Continue reading
When hospital admission source was added to the multivariate model, the associations with patient characteristics and mortality were not significantly changed. However, patients admitted to the hospital through the emergency department were more than twice as likely to die during their hospitalization compared to those with routine hospital admissions (OR, 2.38; p = 0.001). Hospital characteristics were not significant in multivariate analysis and, when included in the multivariate model, did not significantly change the results (data not shown). Continue reading
Comorbid illness was common in patients with TB, with 29% having a DCI score of 1, 9% having a score of 2, and 4% having a score of > 3. Seventy-five patients (3.3%) were identified as having HIV infection.
Over half of the patients with TB (57%) were admitted to the hospital through the emergency department, and Table 2 shows the characteristics of the hospitals to which TB patients were admitted. Compared to non-TB patients, those with TB were more likely to be admitted to urban hospitals (92% vs 84%, respectively; p < 0.001) and those designated as teaching hospitals (61% vs 43%, respectively; p < 0.001). Continue reading
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). Continue reading
We used the 2000 Nationwide Inpatient Sample (NIS), a database of hospital inpatient stays, developed by the Healthcare Cost and Utilization Project. As the largest all-payer inpatient care database that is publicly available in the United States, the NIS data set represents 20% of non-federal US hospitals. These data include a stratified random sample of 994 hospitals in 28 states, encompassing approximately 7.5 million inpatient stays. Continue reading
Tuberculosis (TB) is an infectious disease caused by Mycobacterium tuberculosis and transmitted by aerosolized droplet nuclei, infecting approximately one third of the world population. In 1997, it was estimated that almost 2 million people died of TB worldwide, with a case fatality rate as high as 23%. In the United States, active TB disease usually can be treated successfully, with an extended therapeutic course of a combination of antibiotics, often using directly observed therapy (DOT). According to a recent expert consensus statement,2 “it is well established that appropriate therapy of TB rapidly renders the patient noninfectious… minimizes the risk of disability or death from TB and nearly eliminates the possibility of relapse.” Despite the availability of curative therapy, TB affects the quality of life of the people infected. A large proportion of patients with TB are being hospitalized, and inhospital mortality remains high, with estimates of mortality rates ranging widely from 2 to 12%. Some studies have examined the costs of TB hospitalizations, however, few investigations have addressed the poor outcomes of hospitalized patients with TB. Continue reading