PURPOSE: Hospital mortality rates and resource utilizations among respiratory failure patients are high. We intended to develop a population-based risk stratification tool for this clinical condition which would facility studies on treatment and resource consumptions.
METHODS: We analyzed 24, 883 admissions with the principal diagnosis of acute or chronic respiratory failure. Patients were admitted during 2004–2005 across 79 teaching hospitals (60% case loads) and 107 non-teaching hospitals (40% case loads). We randomly split patient population into a derivation cohort (n=12,442; deaths=2,343) and a validation cohort (n=12,441; deaths=2,359). We developed a logistic regression model using age, gender, laboratory findings, vital signs, and comorbidities as covariates. We validated the model internally with boot strap and externally with the validation cohort. We used c-statistic to assess model fit. The results are presented as odds ratios and 95% confidence intervals.
RESULTS: Overall, the median age was 71 years old (inter-quartile range of 59–80) and approximately 53% were women. The crude mortality was 18.9%. The prevalence of metastatic cancer, respiratory cancer, and hematological cancer was 3.8%, 2.7%, and 1.4% respectively. Significant multivariable predictors (p <.0001) included age, severe altered mental status (2.11, 1.93–2.30), systolic BP < 80 mmHg (1.73, 1.58–1.90), temperature <95.1 F (1.36, 1.19–1.56), base excess < −11.4 (2.52, 2.14–2.97), total bilirubin >2 mg/dl (1.95, 1.56–2.44), albumin <2.5 g/dl (1.73, 1.54–1.95), BUN >40 mg/dl (1.73, 1.58–1.90), metastatic cancer (4.05, 3.48–4.73), respiratory cancer (2.60, 2.16–3.15), and hematological cancer (1.69, 1.30–2.19). The c-statistic for the model was .79. It calibrated well for both derivation and validation cohorts.
CONCLUSION: Mortality among respiratory failure patients can be well predicted by pathophysiological variables routinely available on admission.
CLINICAL IMPLICATIONS: This model is developed and validated with a large US database using information commonly available on admission. It can be used for outcome studies among respiratory failure patients.
DISCLOSURE: Ying Tabak, No Financial Disclosure Information; No Product/Research Disclosure Information