Inappropriate medication use and risk of falls – A prospective study in a large community-dwelling elderly cohort

In the United States, more than one-third of all medications is consumed by persons aged 65 years and over, although they only represent approximately 15% of the total population [ 1]. Similar data have been reported in other developed countries. Due to concurrent prescription of several drugs, the risk of inappropriate drug combinations is increased in older persons [ 2, 3]. Moreover, medication metabolism is affected by aging-related pharmacokinetic and pharmacodynamic changes which increase both drug half-life and drug free fraction. Lastly, coexisting illnesses can interact with medications. For all these reasons, older persons are at higher risk of experiencing adverse drug effects. To reduce frequency and severity of these events is a current challenge for public health agencies.

Beers et al [ 4] have elaborated explicit criteria for determining potentially inappropriate medication consumption in older adults, using Delphi methods for obtaining a consensus of recognized experts. The Beers list, which initially defined inappropriate medication use in nursing home residents, was afterwards modified and updated for being used in general elderly populations [ 5– 8]. The lists identified specific drugs or drug prescriptions (excessive dose, excessive treatment duration, inappropriate drug combination and coexisting illness) with unfavorable benefit/risk ratio or questionable efficacy. Inappropriate medication lists have been used worldwide for evaluation of medical prescriptions [ 9– 13] but there has been little evidence from epidemiological studies regarding the negative consequences of inappropriate medication use [ 14– 17].

Previous studies have suggested that some drugs could increase the risk of falls, which are a frequent and serious problem in elderly population [ 18– 24]. In a one-year follow-up study of persons aged 75 years and over living in the community, about one-third reported at least one fall [ 25]; higher annual risk of falls, up to 50%, were reported in the oldest old or in persons living in nursing homes. The consequences of injuries and fractures due to falls (mortality, hospitalizations, disability, and institutionalization) increase with age. In 2001, the American and British 'Guidelines for the prevention of falls in older persons' therefore recommended careful review of medications taken by elderly persons [ 26].

The aim of this prospective study was to evaluate the association between the use of potentially inappropriate medications and the risk of falls in a large population-based cohort of elderly persons.

In this large community-based study, we found that use of inappropriate medications increased the risk of falls in elderly people. Consumption of long-acting benzodiazepines was responsible for the main part of this increase. But regular exposure to inappropriate psychotropic drugs or to inappropriate medication with anticholinergic properties was also associated with an increased risk of falling. Repeated assessments of medication use did not bring evidence for a dose-risk relation between the level or the duration of the exposure to inappropriate medications and the risk of falls. There was no association between use of short- or intermediate-acting benzodiazepines and occurrence of falls.

Overall, our results were consistent with those of previous studies about the relation between psychotropic medication and falls in older people [ 33– 36]. However, previous studies, did not categorize psychotropic medications as appropriate or not appropriate for elderly and most of them did not differentiate benzodiazepines from other psychotropics. A few studies compared the risk of falls in long-acting and short- or intermediate-acting benzodiazepines users. Overall, unlike our study, they found that the risk of falls was independent of the half-life of the drug [ 37, 38]. Tamblyn et al [ 39] found an association between falls and some specific benzodiazepines, but it was independent from their half-life.

The design of our multi-site study was well-suited to the exploration of the relation between drug use and falls, being a large general population-based cohort, with a high follow-up rate (92% at 4-year follow-up). Data relating to classical confounders were available and additional potential confounders, such as cognitive status, depressive symptoms, or diurnal drowsiness were assessed and taken into account in the statistical analysis. Drug consumption was assessed carefully. Compared with previous studies, one strength of the present study was the availability of measures allowing us to investigate the impact of prolonged and/or elevated exposure to inappropriate medication on the risk of falls. Lastly, we used recognized criteria for selecting medications that are considered as potentially inappropriate for elderly.

Some limitations of our study deserve, however, to be mentioned. Dosage and indication for prescription were not available. The precise date of falls was not collected. Data about some risk factors of falls, as visual acuity, were not available. Although our study population was quite large and the overall prevalence of inappropriate medication use was high (32%), [ 40– 42] numbers were too small for estimating the risk of falls for specific drug classes, except for long-acting benzodiazepines. Another limitation which is common to all studies is that self-report of current drug use, as well as data on drug delivery from the health care insurance plan are imperfect measurements of actual drug consumption. Finally, our results may be conservative because subjects, whose data on falls were not available, consumed more inappropriate medication than the study population.

Most of the drugs potentially inappropriate for the elderly have known side-effects that can contribute to the risk of falling: drowsiness, decreased postural reflexes, extra-pyramidal symptoms, drugging and myorelaxant effect [ 43, 44]. Age-related changes in drug metabolism and polypharmacy can increase the frequency and severity of these side effects, which in turn can increase the risk of falls. In the 3C study participants, the risk of falling associated with the use of inappropriate medication, although statistically significant, was moderate. We did not find any high-risk group that could be the target of possible interventions. However, a small-sized controlled intervention [ 45] in persons aged 65 years and over has shown a significant 66% decrease in the risk of falling after psychotropic medication withdrawal. Another argument in favour of such interventions is that a given drug can be associated with different health events. Intervention efficacy should not therefore be evaluated on a single end-point.

Our study showed that the risk of fall was moderately increased in elderly users of inappropriate psychotropics (long-acting benzodiazepines or others) compared to non users. However, under the hypothesis of a causal relationship, the number of falls attributable to the use of inappropriate drugs depends on the exposure prevalence. In countries with a high consumption of psychotropics, an important number of falls could be attributable to inappropriate use of psychotropics in elderly. Among European countries, France has a high consumption of psychotropic drugs, with about 50% of subjects over 70 years consuming psychotropics [ 46]. In a French elderly community-dwelling population, 31% subjects reported use of at least one benzodiazepine [ 47]. In Canada, 22.5% of elderly were prescribed benzodiazepines in 1998 [ 48].

Our results on the relation between the duration and/or intensity of the exposure to inappropriate medications and the risk of falls are complex. Overall, regular and occasional users of inappropriate medications had similar risk of falling. In long-acting benzodiazepine users, analysis did not show significant linear relation (nor even any trend) between the number of boxes bought over a 3-year period and the risk of falls, which was increased only in those who bought more than 22 boxes (last decile of the distribution). There are several non exclusive explanations for the absence of a dose-risk relationship. A first possible explanation is measurement errors. Available measurements (repeated self-reports of current use and data from the national healthcare insurance plan) are only surrogate markers of the biological exposure, which depends on actual drug consumption. Another explanation is a selection process. Given that persons experiencing drug side-effects are more likely to stop their treatment than those without any tolerance problems, it would follow that long-term users of inappropriate medication could be at lower risk of undesirable events. Another, and related, explanation could be that medication side-effects may be more pronounced at the beginning of the treatment than after some weeks of use. Under this hypothesis, the instantaneous risk of fall decreases when the treatment duration increases. In line with this explanation, a study [ 49] showed that the association between benzodiazepine use and falls was stronger within the 15 days following the treatment initiation than later on.

Data from the 3C study showed that use of inappropriate psychotropic drugs, and particularly of long-acting benzodiazepines, was associated with an increased risk of falling in persons aged 65 years and over. Our study showed that short- or intermediate-acting benzodiazepines did not increase the risk of falls and, as recommended, they should be preferred to long-acting benzodiazepines in elderly patients.