Inappropriate medication use is a major health care issue for the elderly population [1–3]. Older patients are more at risk for adverse medication outcomes because they often have complex drug regimens and because of the age-related changes in drug pharmacokinetics and pharmacodynamics [1]. Potentially inappropriate prescriptions (PIPs), defined as prescriptions in which risks outweigh benefits, have been assessed in various settings using lists of explicit criteria most often based on that developed by Beers [4]. PIPs have been estimated to affect 4.8% to 45.6% of the elderly population [5–12].

Prevalence estimates of PIPs are likely to vary with the criteria that are applied. Some authors have based their assessment on the Beers criteria [5–7, 9–12]. However, in all these studies but one [7], criteria applied were a subset only of Beers criteria as dosage and duration was not evaluated. Despite controversy about which explicit criteria should be used, there is a strong body of evidence showing that suboptimal prescribing is disturbingly common in older patients.

In Canada, a list of explicit criteria was developed by a panel of experts in 1997 [13]. The Canadian criteria required diagnostic information which is not easily accessible in the long-term care setting [6, 14]. Using various methodologies, several studies have investigated the extent of the problem in Canada. A 1995 study of community-dwelling and institutionalized older patients reported large variations in PIPs among provinces, ranging from 4.8% in the prairies to 12.8% in Quebec [9]. More recently, the prevalence of PIPs in long-term care patients in Ontario was reported to range between 14.9% and 20.8% [15–17]. In Quebec, a 1990 retrospective database survey of 63,268 older Medicare patients reported that 45.6% of non-institutionalized patients received high-risk prescriptions of questionable appropriateness [8], while a recent survey of 3,400 elderly patients in the Quebec general population reported that 6.5% had a potentially inappropriate prescription (PIP) [18]. A 1995 physician survey reported that 77.1% of nursing home patients in Quebec had been taking benzodiazepine for over one year [19].

The long-term care elderly population is particularly vulnerable to inappropriate medication use; it is composed of frail older patients who typically have functional disabilities and acute and chronic medical histories that require complex medication regimens [20, 21]. Assessing PIPs using the data available in long-term care, in particular data on dosage and duration of use, may help designing efficient interventions to improve prescribing practices in one of the frailest populations. The objectives of this study were (1) to describe prescribing patterns in elderly patients residing in long-term care facilities in the Quebec metropolitan area, (2) to assess the prevalence of PIPs in this long-term care setting using published explicit criteria [4, 13, 22] adapted for this study, and (3) to identify patient-related predictors of PIPs.

The long-term care elderly population evaluated in this study was highly medicated and a majority of patients receiving medication had a PIP. These results indicate that potentially inappropriate prescribing was significant at the time of the study in institutionalized older patients in the Quebec metropolitan area.

A total of 94% of residents in this long-term care population were prescribed at least one drug, compared to 60% in community-dwelling elderly patients in Quebec [25]. The mean number of medications was also higher (4.8) than in community-dwelling individuals in Quebec (2.9) [25], but lower than in American long-term care (7.2) [7].

The total prevalence of PIPs among the population under study was high (51.5%). Estimates of PIP prevalence in the literature vary between 4.8% [6] and 45.6% [8] for both institutionalized and community-dwelling older patients. Caution must be used when comparing these results, as the delivery of care may vary from one setting and one region to another [9]. The current lack of consensus when defining lists of criteria and variations with respect to methodologies also contribute to the observed differences [26]. For example, Zhan and colleagues [5] estimated the proportion of potentially inappropriate medication use in the community-dwelling elderly in the United States. Applying criteria on the indication for the use of 33 drugs, they observed a prevalence of 21.3% for 1996. In our study, PIPs were identified using an explicit criteria list that was primarily based on Beers and McLeod criteria [4, 13, 22] and that was updated and validated by local experts to apply to the long-term care context in Quebec. As we had access to dosage and duration information, we were able to apply a broader set of criteria which can explain the higher prevalence of PIPs we have observed. Explicit criteria lists, such as those developed by Beers and McLeod, define inappropriate prescription according to the drug overall risk-benefit profile for elderly patients. These lists were previously used in studies examining inappropriate prescribing in elderly populations [3, 5, 6, 11, 15, 27–30] and undergo a continuous process of revision and updating to reflect the most current clinical information on the risks and benefits of medications [31].

A large number of patients were receiving CNS medication (85%) and the most common PIPs were related to that category of drugs. Thirty-five percent of patients were prescribed antipsychotics and 22.9% had benzodiazepine for potentially inappropriate duration, defined as more than a month [32]. A number of studies have reported the inappropriate use of CNS drugs [5, 8, 33, 34], particularly benzodiazepines [18, 19]. Many factors may contribute to the continued use of inappropriate CNS medications, including prescriber attitudes, patient demands and the design of the health care system [34]. A survey of physicians in Quebec reported that the psychological distress of aging patients and the quasi-absence of reported side-effects justified the long-term use of psychotropic medication, which was seen as the most effective way of helping the patient [35]. Moreover, side effects of psychoactive medication are often believed to be a consequence of the aging process [34]. Almost three quarters of potentially inappropriate psychoactive medications can produce a physical dependence [34]. Psychoactive pharmacotherapy increases risk of hip fractures and is advocated for use with caution to prevent falls in elderly populations [36, 37]. Anticonvulsants, antidepressants and short- and long-acting benzodiazepines were reported to increase risk of falls in older women [38].

The length of stay was positively associated with PIPs, while the prevalence of PIPs decreased with age. Although the association between length of stay and the likelihood of receiving a PIP in nursing homes was studied in the past [6], to our knowledge, this is the first time it is being shown to be a predictor of PIPs. On the other hand, the risk of receiving a PIP was previously reported to decrease with age in nursing home patients over 65 years [7, 12]. Data on clinical status was not considered in these studies and it can be hypothesized that either the oldest residents were less ill or that physicians were more cautious when prescribing to very old patients. As reported in previous studies [12, 26], the number of medications was also a predictor of PIP in older patients. Patients in long-term care frequently have multiple diseases resulting in complex medication regimens, which makes assessment of the risks versus benefits of treatments often difficult. Female gender was previously reported as a predictor of PIP [7, 12]. Although we observed gender differences in the prescribed therapeutic classes, female gender was not a predictor of PIPs in our study.

The results presented here should be viewed in light of potential limitations. As in previous studies [15], we did not abstract information on diagnoses from the patients charts and drug prescriptions were considered as surrogates for disease conditions. Thus, the explicit criteria used in this study apply to general circumstances, but may not be applicable to specific cases, since they do not consider clinical information. For example, lipid-lowering drugs may be potentially inappropriate in patients aged 75 and over, but evidence from clinical trials suggests that statins may be of benefit if the patient's life expectancy exceeds two years [39]. Thus, misidentification of potential cases of appropriate or inappropriate prescribing may have occurred, since complex medical conditions can alter the risk-benefit profile of medications. However, due to the frail condition of most patients, it is unlikely that such misidentifications have occurred frequently. Since access to clinical data is often difficult in the nursing home setting, a list of explicit criteria that does not require that type of information may be easier to apply on a larger scale.

This study evaluated prescription patterns rather than the actual consumption of medication. The low prevalence of as-needed medication (14%) and the long-term care setting, in which medication is administered to patients by a health caregiver, suggest that this limitation did not have a significant impact on the results. As-needed prescriptions may have accounted for repeat prescriptions, which may in turn have led to overestimation of the number of drug-drug interactions. However, even after excluding as-needed prescriptions from the analysis, the proportion of residents with a PIP remains high.

Predictors of PIPs were assessed using a multivariate analysis. It allowed us to adjust for potential confounding variables. However, we were not able to adjust for facility variables as those were not available.

This study is the first to describe and qualify prescribing practices in long-term care facilities in urban Quebec. In particular, it highlights the extent of potentially inappropriate prescribing in elderly long-term care patients, which are among the frailest of society [4, 21]. Inappropriate prescribing is one component of the major health care problem of suboptimal prescribing that also includes underuse of effective agents, drug-disease interactions and prescription errors. Substantial morbidity, mortality and cost are attributed to suboptimal prescribing [1, 2]. Although a decline in the prevalence of PIPs was reported in community-dwelling older patients in the United States between 1987 and 1996 [40], the continued use of inappropriate medications is a major concern. A growing body of evidence suggests that clinical pharmacy and multidisciplinary team interventions can modify suboptimal prescribing in older patients. Modern data management [15, 41] and use of the best clinical evidence could help practitioners improve the management of complex cases [40, 42]. Recent studies in long-term care settings showed that physician or pharmacist interventions reduce PIPs [1, 12, 16, 43], while a clinical review program of prescriptions for community-dwelling patients conducted by a team of physicians, pharmacists and nurses did not seem to improve prescribing practices [44].

Inappropriate prescribing is highly prevalent in the elderly long-term care population in metropolitan Quebec. The use of a explicit criteria list to identify PIPs is a first step towards identifying most critical issues and implementing strategies to improve quality of care and patient safety. Identifying predictors of PIPs may help to target problems and prioritize interventions that are most needed in the rapidly expanding older population.