The designation into risk classes III, IV, and V are based on the whether the patient has one, two, or three of the central risk factors, respectively. We defined the occurrence of altered mental status as an initial Glasgow Coma score < 14 or a designation by the physician of disorientation, stupor, or coma.
End Points and Sensitivity Analyses
In-hospital mortality and need for MV at any point during the hospitalization served as co-primary end points. We further explored LOS and hospital costs relative to the BAP-65 score. Costs were computed by converting hospital charges by institution-specific cost-to-charge ratios obtained from the Centers for Medicare and Medicaid Services.
Because both the timing of death and issues related to limitations in the intensity of care may affect resource use and LOS in hospitalized patients, we conducted a sensitivity analysis examining LOS and costs among only hospital survivors. By completing an evaluation of resource use in only survivors we aimed to limit confounding due to these concerns. In addition, we validated the BAP-65 in patients with principal diagnosis of AECOPD only, using the same definition as the one used for the original BAP-65 study.
Stay tuned for updates and subscribe to our twitter. Canadian health care news online)!
Statistical Analysis
Categorical variables were compared with a x2 test, and continuous variables were compared via analysis of variance. The Cochrane-Armitage trending statistic was used to assess whether the risk score could differentiate low-risk from high-risk patients in a fashion reflecting a graded response based on the level of risk present for mortality and MV use. For contrast test, we used x2 test for categorical variable and Welch t test for continuous variables. We used the Bonferroni method to adjust for four follow-up tests of differences in each of the four end points from one BAP-65 class to the next. A P value of < .0125 was considered statistically significant and all tests were two-sided. We assessed the predictive ability of the BAP-65 for the primary end points using the area under the receiver operating characteristic curve (AUROC). We constructed 95% CIs for the AUROC using 1,000 bootstrap iterations. We assessed sensitivity, specificity, and positive and negative predictive values of the BAP-65 for the primary end points.