Falls were frequent among the frail LTCF residents in our study, with 56.7% of residents having fallen at least once in the prior 6 months. We found that falls were associated with cognitive impairment, male gender, visual impairment and the use of certain medications. Potential Inappropriate Medication use (according to the Beers list) and SSRI/SNRI classes were associated with increased risk of falls, while benzodiazepines appeared to be associated with a decreased risk of falls.

Our results should be interpreted with caution. This was an observational study, and sampling of LTCF residents was not random; those with emergency ambulance calls were over-sampled. Data were abstracted from LTCF charts, and thus we relied on documentation in the charts and on clinical reporting of measures like falls and dementia diagnoses. On the other hand, these results reflect the “real world” of clinical practice. Our study analyzed frailty and falls which is a relationship that has not previously been examined in Long Term Care populations to our knowledge.

Due to the retrospective nature of the study, despite the best efforts of the chart abstractors to glean the relevant data points from resident charts, some residents were missing data. For example, 110 individuals (27.8% of the sample) were missing frailty data because this had not been completed by their primary care teams. Based on prior experience in frailty studies, generally speaking frailty is more likely to be missing in the most vulnerable (who are also the most frail) [16]. Our findings may therefore be conservative, and this may also have contributed to our lack of finding of an association between frailty and falls.

Documentation and reporting of falls is a challenge. Our study did not use a standardized method of recording of falls, as the LTCF did not have a formal tool for recording falls. Falls were counted if they were documented in a resident’s chart upon review of the preceding 6 months. It is possible that fall events would have been missed if they went undocumented (for example if the resident had not reported a fall to the care staff). Potential gaps in reporting of falls are not a problem unique to our study. Reporting differences by gender have been documented in the literature [17, 18]. This is potentially relevant to our study’s finding of an association between male gender and increased fall numbers. The current literature supports that women fall more than men, but this may be due, at least in part, to gender differences in reporting. Stevens et al., examined gender differences in seeking care in the community-dwelling adults over the age of 65. They found that a significantly greater percentage of women than men sought medical care after a falls, talked with a healthcare provider about falls, talked with a provider to understand why they fell, and/or talked with a provider about fall prevention [17]. Since our study was conducted in LTCF, underreporting remains a possibility, though residents would presumably be more likely to come to staff attention when they fell. For this reason, our finding that men in our study had more documented falls appears all the more relevant. Taking our study findings in the context of the existing literature, more focused screening for falls among men may be warranted as a prevention strategy.

The relevance of medication use in relation to falls was another important finding in our study. Our study lacks information on medication doses and indications, which would be beneficial to the interpretation of our findings. In LTCF settings, mediations such as benzodiazepines, antidepressants and antipsychotics have been previously reported to be associated with increased risk of falls [6]. Our finding that SSRI/SNRI and PIMs were associated with increased risk of falls is thus in keeping with the current literature. However, our study found that benzodiazepines appeared to be protective for falls. We conducted further analysis to try and account for this surprising finding. We hypothesized that one potential explanation could be that benzodiazepines are more likely to be prescribed at end of life (when mobility might be restricted and falls risk thus attenuated). However, we found that the majority of benzodiazepine users were not classified as being at end-of life. Another explanation could be that those taking benzodiazepines were less mobile, and therefore less likely to fall, yet we found that over half (57.9%) of the residents on benzodiazepines were walking independently. A third explanation might be the increased education in appropriate prescribing of benzodiazepines among physicians. If residents at high risk of falls preferentially had benzodiazepines de-prescribed or not initiated, while those at lower risk of falls were left taking them, this could have led to an apparently inverse association. This is particularly interesting in our study as during the time period prior to our data collection, the CBD team implemented new prescribing guidelines and medication review protocols for the physicians caring for these patients. The focus was not specifically on benzodiazepines, but it still may have impacted the prescribing habits of the physicians.

Although not statistically significant, the frailty trends discovered in our study are interesting. We found that moderately and severely frail residents experienced more falls. These intermediate frailty categories were also were the highest benzodiazepine users. Moderate to severe frailty could be considered the most vulnerable group to experience falls, as they could be mobile but also dependent for most basic activities of daily living [14]. The extremes of frailty are likely to cancel each other out, with opposite influences. Very frail and terminally ill residents might tend to be bed bound (and thus at lower risk of falls) while those who with mild frailty would not be as likely to fall as they presumably have greater independence with mobility.

Generally speaking, women tend to be more frail than men [15]. Our lack of finding of an association between gender and frailty may reflect a placement bias. Less frail men may be more likely to be cared for at home by their surviving wives, moving to LTC only at with more advanced frailty. Women may have to move to LTC at lower levels of frailty if they have been widowed and have fewer options for care at home [19].

Our study also found visual impairment and impaired cognition to be correlated with increased risk of falls. The current literature lacks studies associating visual impairment and falls in residents of LTCF. Unfortunately, our study can only conclude visual impairment increased the risk of falls, but cannot comment of the specific degree or type of visual impairment, because we relied on a general clinical diagnosis of “visual impairment”. However, visual impairment is something that is potentially modifiable, so it should not be overlooked as a risk factor for falls. On the other hand, dementia is a well-documented risk factor for falls [20]. Our study findings support the importance of dementia as a risk factor for falls in LTCF. Frequent cognitive screening and staging, and falls reduction interventions targeted to residents with dementia, may help to implement preventative measures.