ADVERSE DRUG EVENTS IN HOSPITALIZED ELDERLY

Background. The study objectives were (a) to describe the occurrence, types, and preventability of adverse drug events (ADEs) in hospitalize d patients 70 years of age and older; (b) to examine the association b etween potential risk factors and ADEs; and (c) to examine the relatio nship of an ADE occurrence and hospital length of stay (LOS) and funct ional decline. Method. Consecutive general medical admissions (II = 15 7) of community-dwelling persons were prospectively monitored daily fo r ADE occurrence. Admission assessment included demographic factors, c ognition, preadmission medication use, and functional status. Discharg e assessment included functional status, LOS, discharge diagnoses, and medication use during the hospitalization. Results. Twenty-three pati ents (14.6%) experienced 28 probable ADEs, of which 54.2% (13/24) were judged to be potentially preventable. Patients experiencing an ADE ha d a significantly lower mean Mini-Mental State Examination score (23.6 +/- 4.3 vs 25.5 +/- 3.6, p=.039) and were prescribed significantly mo re new inpatient medications (4.0 +/- 2.3 vs 2.6 +/- 1.7, p = .01) com pared to non-ADE patients. Age, gender, functional status prior to adm ission, percent with more than four active diagnoses, or number of pre admission medications were not associated with ADE status. Upon discha rge, 50.0% of ADE patients experienced a decline in one or more activi ties of daily living (ADLs), compared with 24.1% of non-ADE patients ( p = .017). ADE patients had a longer LOS (8.7 +/- 4.9 vs 6.6 +/- 3.0 d ays, p = .022) compared to non-ADE patients. Conclusions, ADEs were as sociated with number of new inpatient medications and admission cognit ive status, but not demographic, disease, or physical function variabl es. Patients experiencing an ADE were more likely to experience a long er LOS and to decline in ADL function. ADEs may be one factor contribu ting to functional decline during hospitalization. Future research in this area should include larger samples and multivariable analyses con trolling for potential confounders.