Psychotropic medication changes and the short-term risk of falls in nursing home residents: A case-crossover study

Falls among nursing home residents are common and recurrent events. Epidemiologic studies have focused generally on fixed risk factors for falls (e.g. fall history, dementia) to identify high-risk individuals. Potential transient risk factors can identify high-risk time periods for falls (e.g., within days of an acute illness onset). To investigate the role of changes in medication administration before falls in nursing home residents, we conducted a case-crossover study of 419 falls in 158 residents of three nursing homes that occurred in 2002 and 2003. Incident reports were used to determine the date and time of each fall. Medication records were abstracted to identify the date of selected medication changes (new start, dose change, as needed dose, or discontinuation) during a nine-day period before the date of each fall. An inverse weighting method was used to balance the exposure effect within persons having two or more falls with those having one. The residents were on average 80.5 years old, and 67% were males. Within 1-3 days of a change in any central nervous system medication (anti-psychotic, sedative, anti-depressant, or anti-seizure) the estimated risk of falls increased 3.4 fold (95% CI 1.5, 17.3) using 7-9 days prior as a comparable control period. The within-person study design controlled for all between-person fixed confounders such as dementia, fall history, and standing balance. Because as many as 70% of all nursing home residents may be exposed to psychotropic medications, additional fall precautions might be applied specifically when changing this class of medication.