We found considerable variation in the drug use among the NHs, in terms of number of drugs used on regular basis. This was in particular pronounced for the use of analgesics and psychotropic drugs where the variation was extremely large. We believe that this variation reflect local therapeutic subcultures involving inappropriate drug use. Our findings here represent an important challenge for future quality improvement measures, especially because the psychotropic drugs include risk for many and serious side effects in frail old people with dementia [11]. However, our results are generally consistent with those reported elsewhere for long-term care home residents in Norway [12, 14], Europe [27], US [15, 28] and Canada [29].
Further, the study documented that the rates of DRPs varied up to seven-fold (from 0.5 to 3.4) between the NHs. To the best of our knowledge, only two medication review studies have previously reported variation in DPRs between NHs: one in two urban NHs, from 3.0 to 5.5 mean DRPs per patient [20] and another study in four rural NHs, from 2.7 to 5.6 mean DRPs per patient [30]. The mean of 2.0 DRPs per patient found in the total cohort is below those previously reported in Norway [2, 3, 13], probably because we reported DRPs agreed upon by the team, not all DRPs suggested by the pharmacist.
The associations between the uses of opioids, antipsychotics, benzodiazepines or antidepressants and increased risk of DRPs are consistent with the fact that so many of these drugs are commonly considered potentially inappropriate and should therefore be avoided whenever possible in frail olds. In our study, psychotropic and analgesic drugs were involved in just one third of the total DRPs, and it would be expected that by including also drugs for pro re nata use (“as needed”), this would probably have increased even more the contribution of psychotropic and analgesic drugs to the numbers of DRPs [21]. The correlation between the use of many drugs and more opioids and benzodiazepines at the respective NHs might reflect local prescription cultures [28], or simply a way to relieve staff pressure [31], as prescription of psychotropic drugs and painkillers in combination is not recommended to treat neither pain nor BPSD [9, 32].
We found no difference in the levels of DRPs between the NHs with highest and lowest drug use, although using more drugs was associated with DRPs. This unexpected finding might be due to our analytic strategy by grouping the NHs into quartiles, in addition to a large variation in the levels of DRPs within each group (e.g., three high-drug use NHs with low levels of DRPs and four low-drug use NHs with high levels of DRPs). The strong correlations found between need for additional drug, use of unnecessary drug, excessive dosing and inappropriate drug choice, suggest that prescription quality is multifaceted and hence, in case it is suboptimal, e.g. due to a high rate of DRPs, this will affect several areas of drug prescription practice.
The large difference in DRP levels found between otherwise comparable NHs most probably reflect different institutional prescription cultures, with higher prescription rates at NH-level irrespective of the patient’s clinical indications [29] or different organizational initiatives for patient safety at the NH [33]. To improve the quality of drug use in the NH setting, staff should be educated in geriatric pharmacotherapy and on alternative non-pharmacological interventions [9, 10]. Other measures should include implementing educational programs on person-centred care [34] and multidisciplinary medication reviews [18], which may also include collaboration with a geriatrician [35].
Strengths and limitations
The strength of this close to practice study was the standardized procedure for MRs, with face-to-face meetings between pharmacist, physician and nurse, having access to patients’ clinical information, and agreeing on actual DRPs for each patient.
It is an important limitation that we have only recorded the DRPs that were accepted by the physicians, without recording all the DRPs that were initially suggested by the pharmacists. Hence, we do not know how the physicians’ acceptance rates varied between the different NHs and how appropriate their rejections were [30]. Some doctors may have experienced suggestions to change their treatment as a threat and criticism towards their own prescribing practice.
The explicit criteria used in this study were updated [16] and tailored for the NH-setting [36] after the study had started, however, we do not believe that using the updated criteria would have changed our results significantly. Instead, it may be questioned if the explicit criteria used were sensitive enough to detect over- and underprescription, or inappropriate medication among multimorbid, frail NH residents commonly exposed to extensive off-label pharmacological treatment for BPSD. Although DRPs, as identified in our study, might have limitations as quality indicators for drug prescription, the NHs with high levels of DRPs probably have proportionally larger potentials for quality improvement.
We believe that the sample of institutions and residents is representative for the long-term care NH-setting because the vast majority of the NHs in the municipality participated in the study. This is a cross-sectional study, and thus we are not able to draw conclusions about causal relationships for the variation. The NHs in Oslo are quite similar: They are publicly financed and administered by the same agency, are non-academic institutions operating in the same regulatory and clinical practice context. They are staffed with full-time nursing home physicians and registered nurses according to the country standard. None of them had an in-house pharmacist. The patient-mix is quite similar due to equal admission criteria. Grouping the NHs in quartiles might be challenged due to the somewhat limited number of NHs.