Introduction
With aging, the risk of developing chronic diseases increases and often lead to the use of multiple drugs [
1], referred to as polypharmacy [
2]. The prevalence of polypharmacy ranges from 34 to 65% in older patients in different countries [
1,
3‐
5]. As polypharmacy increases, so does the use of potentially inappropriate drugs (PIDs) which leads to a higher risk of drug related problems (DRPs) [
1,
6‐
11]. Polypharmacy is also an important risk factor for preventable hospital admissions [
7,
12‐
15]. The costs associated with avoidable admissions are estimated, for the Netherlands, at over 85 million euros [
12]. It is predicted that the number of elderly with multi-morbidity will continue to increase, which makes polypharmacy an expanding societal issue [
5].
Medication reviews aim to reduce preventable admissions and related costs. In the Netherlands, the Systematic Tool to Reduce Inappropriate Prescribing (STRIP) method [
2] is used in community pharmacies to systematically assess the medication usage of an individual patient through a review by a physician, pharmacist, and the patient (and/or informal caregiver or other caretakers).
Several studies have been performed regarding the effectiveness of medication reviews. On the one hand, medication reviews show little or no effect in improving the quality of life, mortality, patients’ functioning, (re)admissions, and adverse drug events [
16‐
21]. On the other hand, in some studies a decrease was found in the number of drugs per patient, DRPs, emergency department contacts, or inappropriate drugs [
16,
17,
22‐
25].
As of now, there is limited explanation for these unexpectedly small or absent effects seen in previous studies. One explanation might be that only certain subgroups of patients might benefit from medication reviews, which does not become apparent when mainly focusing on older polypharmacy patients in general. This idea is supported by one study in high-risk patients [
17] in which an effect was observed on the total number of prescribed medication. It is, however, not clear whether there is an effect on the prescription of inappropriate medicines as well, which could further explain the limited health gains –like hospital admissions or quality of life- that were also observed. In order to draw any firm conclusions on the usefulness of medication reviews, there is a need for studies that aim to get a closer look on individual effects and on effects in high-risk groups—with e.g. impaired renal function, or poor compliance. Also, additional insight into the stability of changes in medication and the possible continuation of PIDs after a review may be required to further understand the impact of medication reviews.
Discussion
This study aimed to assess changes in number and appropriateness of prescribed medication in older polypharmacy patients following a medication review in general practice. The overall findings suggest that medication reviews may be able to reduce the number of drugs and the number of potentially inappropriate drugs and that the changes remain fairly stable at least in the first 3 months. The drug groups ‘alimentary tract and metabolism’ and ‘cardiovascular system’ consist of the leading types of drugs that undergo changes due to the medication reviews. Overall, there were more patients that stopped one or more drugs than patients that were prescribed one or more additional drugs. Reasons for continuing PIDs are mainly the (expected) negative effects of changes and the patient’s preference.
Although there were more drugs stopped than added, which is confirmed by two previous Dutch studies [
28,
29], the average decrease in number of drugs between before and after the review was rather small. Lenaghan et al. [
17] and Jódar-Sánchez et al. [
22] also showed an average reduction of less than one drug per patient, which matches the current result. There is a chance that the selected pharmacies and the corresponding general practitioners already pay much attention to the medication of the patients which results in a small difference between before and after the review. However, the ultimate goal of medication reviews is not (only) to reduce the number of drugs, but primarily to improve the medication lists when needed, so that the drug use will be safer, i.e. less PIDs are prescribed. We observed a significant decrease in PIDs from 0.6 per patient to 0.4 per patient, both 1 week and 3 months after the medication review. Although this is on average a small difference, it might be meaningful on an individual level. In some patients, the number of PIDs decreased by two. Currently, there is conflicting evidence for overall effects of PIDs on the occurrence of adverse drug reactions, hospital admissions and other clinical outcomes [
18]. Nevertheless, there may be more prominent effects in certain subgroups and therefore certain individual patients might particularly benefit from deprescription of PIDs whereas others might not. Since it is unclear which subgroups would benefit most, more research is needed to study the effects of PIDs on a variety of outcomes in high risk groups, e.g. older patients who are also frail or have specific comorbidities.
The small decreases in mean number of drugs and number of PIDs can also be explained by the finding that in some cases drugs that had to be stopped according to the guideline were actually not stopped. At first sight, it seems as if the STOPP criteria is not used well. However, for some specific patients, we identified reasons why it is better to keep using a drug, and to ignore the guidelines. Chau et al. [
29] confirm this phenomenon. The most frequent reasons are: the specialist rejected the intervention; the patient did not want to stop the drug; and sometimes it is decided to monitor the use of the drug by the patient and to intervene when really needed. Besides these reasons, we identified two additional reasons that occurred relatively frequent. First, the health status of patients prevents stopping the drug, so the negative consequences of continuing the drug are taken for granted. Possible explanations for this could be that the GP and pharmacist are scared to disrupt a fragile balance when the patient stops that drug. This reason for not stopping the drug could well be understood by the high-risk population that is being studied. Second, patients tried to stop a drug before and experienced negative effects of stopping the drug, so they keep using it.
An important finding in our study was that, after the medication list of the patient has been assessed, patients sometimes refuse to undergo any changes in medication. Since patients are not informed beforehand in the current procedure, this would imply that the time-consuming reviews can be improved by explaining the medication review to patients and ask them whether they are open minded for changes before a review by the pharmacist and GP is planned. As a result, pharmacists and GPs may need to review fewer patients.
Striking in our study were the types of drugs that were stopped most frequently. Most stopped drugs belonged to the groups ‘alimentary tract and metabolism’, ‘cardiovascular system’, and ‘nervous system’. The types of the most frequently stopped drugs in these groups were drugs for constipation, drugs used in diabetes, diuretics, agents acting on the renin-angiotensin system, lipid modifying agents, and analgesics. Looking at existing literature, Mudge et al. [
23] concluded that diuretics and analgesics were the most common deprescribed drugs. However, they also concluded that antiepileptics and psychoanaleptics were common deprescribed drugs, which differs from the results in the current study. In our study there were more cardiovascular drugs that were stopped frequently, which was also found in a previous Dutch study [
28]. Possible, well-known explanations for these differences are first that antiepileptics are typically being prescribed to patients with epilepsy. When this is the case, the drug will not be stopped easily. Second, when antiepileptics are being prescribed as painkillers, the chance of stopping this drug is higher, but this concerns a very small group in the Netherlands. Third, psychoanaleptics are not frequently being prescribed for in the Netherlands.
Finally, looking at the dose modifications after the medication review, it can be concluded that there were just a little bit more dose reductions than dose increases. Balen et al. [
26] also showed an almost equal number between dose reductions and dose increases. In their study, 57.7% of the dose modifications were reductions, which confirms the result found in the present study. This could be seen as beneficial since dose reductions will reduce costs and may imply a lower intake of drugs for a patient.
This study has some limitations. First, the underlying limitation that comes with a pre-test/post-test study design is that there is no control group. A disadvantage of this is that it is not completely certain whether drug modifications are solely the result of the medication review or would have occurred anyway, especially after 3 months. Nevertheless, since the differences between 1 week and 3 months after the review are minor, we assume this probably does not have a strong influence on the results. Still, this paper shows what changes may occur after a medication review, but we need to be careful in drawing the definitive conclusion that the changes are the result of the review. More research is needed in this respect.
Second, the GPs’ and pharmacists’ notes that are used in this study for identifying reasons for not stopping PIDs during the medication review do not cover all possible reasons. These are the only known reasons, because they are registered. Besides this, the occurrence rate of all the registered reasons could also be higher than shown in the results. Still, spontaneously registered reasons give some insight into the most important reasons for not stopping a drug.
Although we do not know how the medication reviews were carried out in practice, it is likely that there was (some) variation between pharmacists and between GPs. The results of this study therefore show what changes may occur after a medication review is performed based on a stepwise method as a guidance rather than based on a prescriptive protocol. Given the variation between pharmacists or between GPs, it is also likely that there is room for improvement of the medication reviews by providing for example a training on how to perform the reviews.
Furthermore, according to the power calculation, 156 patients were needed to detect a minimal difference of 0.28 in the number of drugs, but only 126 patients could be included in this study. However, in the 126 patients that were studied, the differences in the number of drugs were larger and a significant difference was already observed. It therefore would not have made a large difference if 30 more patients were included. Still, smaller differences that were actually present could not have been detected in this population and we therefore need to be careful to generalize the (insignificant) results to the population of elderly polypharmacy patients as a whole.
Finally, there is a possible chance that the pharmacists and GPs did not review the medicines the way they are supposed to do according to the guidelines. This can lead to a less complete review, for example, they may have overlooked some PIDs. However, we do not expect that this would affect our results significantly.