BACKGROUND: Falls are the leading cause of injury-related hospitalizations among older adults in New York City (NYC) and may increase as the older adult population grows. Electronic health record (EHR) data could provide a way to explore modifiable risk factors for falls. Our aim was to evaluate prescribing of medications that increase falls risk among older adults (aged 65-100) by various patient and practice characteristics in NYC ambulatory practices.
METHODS: We analyzed aggregate EHR data through the Hub Population Health System, representing approximately 26% of NYC adults 65-100 with a personal doctor. We identified and modeled the percentage of older patients with a visit in 2013 (n=175,304 patients) that were prescribed at least one medication from six falls-risk medication categories.
RESULTS: Each medication category had a unique prescribing profile by patient characteristic and geography. Adults ≥75 (vs. <75) had higher odds of being prescribed any of five of the six medication categories examined. Women were more often (p<0.05) prescribed anticonvulsants (OR: 1.18), antidepressants (1.65), antipsychotics (1.21), benzodiazepines (1.78), and sedatives (1.30). Black and Hispanic patients, compared to whites, had significantly higher odds of being prescribed anticonvulsants (B: 1.18; H: 1.37) and antihypertensives (B: 1.79, H: 1.51) and lower odds of being prescribed antidepressants (B: 0.37; H: 0.74) and benzodiazepines (B: 0.20; H: 0.43). Medication prescribing for all categories was highest in the Bronx and Staten Island, where fall-related hospitalization rates are highest. In certain neighborhoods, up to 16.2% and 16.5% of older adults at Hub practices were prescribed benzodiazepines and sedatives, respectively. In general, significant differences between practices with different provider specialty profiles did not persist after adjusting for other patient and practice characteristics.
CONCLUSIONS: EHR data enable exploration of modifiable fall-risk factors at the population level and identification of disparities in these risk factors by patient characteristics. Relatively high prescription percentages for benzodiazepines and sedatives in certain populations were notable and underscore the value of these data to alert health care providers about local prescribing patterns and the importance of medication review to identify possible opportunities for alternative treatment opportunities and/or adjustment in dosage levels. Since it allows data to be obtained more frequently than through surveys, the Hub affords opportunities for quick updates of regional messages to providers that reflect changes in clinical guidelines and practice. Given the unique prescribing profiles of medications considered in this study, researchers should analyze these categories separately instead of combining them into one risk category.