2006 OPEN FORUM Abstracts
MORTALITY PREDICTORS FOR HOSPITAL ADMISSIONS FOR COPD
A Asiimwe PhD Research Fellow1,2, D Prytherch PhD1,2, S Kilburn PhD2, AJ Chauhan PhD1, B Higgins BSc2,
Queen Alexandra Hospital, Portsmouth1; Dept of Health Sciences, University of Portsmouth, Portsmouth UK.2
Several factors including co-morbidity, severity of airflow obstruction and previous hospitalisation are known to reduce survival following admissions for acute exacerbations of COPD (AECOPD). We examined whether routinely collected data could predict mortality at admission, 30 days and 1 year after discharge and if we could construct (and later test) models of risk of death for people admitted to Portsmouth Hospitals NHS Trust for AECOPD.
Methods: We examined 22,614 COPD admissions from 12,950 patients (ICD codes 10th version ranging from J40 to J44) from 1996-2005. Clinical data were recorded immediately on admission. Generalised estimating equations (XTGEE procedure, STATA 8) allowing for lack of independence of admissions were used to construct regression models for risk of death, using routinely available biochemical and haematological data, gender and age.
Results: The range of admissions per patient were: 1(70.1%), 2(15.7%), 3(8.2%), and ≥4(12.4%). Mortality for in-patients, at 30-day and 1-year were 1216(9.4%), 1417(10.9%) and 1883(14.5%) respectively. We identified age(p=0.001), pCO2 (p=0.000), and serum urea(P=0.000) as significant predictors of mortality at each time point.
Conclusion: Effective models of mortality applicable to all COPD admissions could be constructed from routine clinical data, obtained largely from two venesections on admission.