SAN DIEGO — Insulin was a surprise among the medications most commonly associated with patient falls in the hospital, results from a large single-center controlled study of 230 patients showed.
“Finding an association between insulin and falls was somewhat surprising and requires further study,” said Caroline O'Neil, research coordinator in the infectious diseases division at Washington University in St. Louis. “Previous community studies have found a connection between diabetes and falling. In our study, diabetes was not associated with falling, but use of insulin was. The question is whether insulin is a marker of more severe diabetes, or if these patients have low blood sugars or peripheral neuropathy that is increasing the risk of falling.”
Patients who fell were more likely than controls who did not fall to be taking hydantoin anticonvulsants (odds ratio 3.6), haloperidol (OR 3.4), benzodiazepines (OR 2.1), or insulin (OR 1.5).
Certain combinations of medications further increased the risk, especially the combination of hydantoins and insulin (OR 11.4), benzodiazepines and haloperidol (OR 5.7), benzodiazepines and hydantoins (OR 5.0), and benzodiazepines and insulin (OR 2.6). The data were presented in poster form at the annual meeting of the Society for Healthcare Epidemiology of America.
The study, which was led by Dr. Victoria J. Fraser, Ms. O'Neil, and her colleagues, evaluated 230 inpatients at the 1,250-bed Barnes Jewish Hospital in St. Louis. The patients were aged 21 years and older and had fallen in the hospital between July 1, 2007, and November 14, 2007; the study also included 690 randomly selected control patients who did not fall and were admitted within 1 day of the index case. Emergency department patients were excluded as were those from physical therapy, obstetrics/gynecology, oncology, and psychiatry units.
They used logistic regression to determine which medications and patient characteristics were risk factors for falls. The mean age of patients who fell was 62 years, 54% were male, and their average length of stay was 12 days.
“This study is among the largest and most detailed analyses of medication and other fall risk factors in the literature to date,” the researchers wrote. “Linking medication risk with nursing fall risk assessments may provide a more comprehensive and predictive view of a patient's risk for falls.”
Multivariate analysis also showed that the following variables were linked significantly with an increased risk of falls: history of falls within 3 months (OR 3.4), use of an assistive device (OR 3.3), BMI of 18 mg/kg
The study was funded by the National Institutes of Health and the Centers for Disease Control and Prevention. Ms. O'Neil said that she had no financial conflicts to disclose.
Patients who fell were more likely to be taking hydantoins, haloperidol, benzodiazepines, or insulin. MS. O'NEIL