Discussion

In this study, we deep-dived into the medicine changes that were implemented during the medication review intervention leading to improved HRQoL and reduced mortality.3 The results in Table 1 show that even though there were three times the number of discontinuations (26% vs 8.7%) in the medication review group, the persistence of the discontinuations after 4 and 13 months were similar between groups (medication review: 91% at 4 months and 82% at 13 months; usual care: 83% at 4 months and 86% at 13 months). The high persistence in the medication review group despite the larger number of changes could be due to improved cross-sectoral communication as the patient’s general practitioner (GP) was involved before implementation of the changes and notified of the changes after implementation.
While the overall persistence of changes was high, some medicines were more prone to rebound than others as evidenced in Table 4. Of note, even the highest rebound rates were roughly 1/3 meaning that even for these medicines, discontinuation attempts were successful for 2/3 of the patients. The rebounded medicines were almost exclusively symptomatic treatments where recurrence of symptoms could easily be identified and therapy reinitiated, while the two preventive medicines with high rebound rates both have easy monitoring options (bone mineral density for alendronate and blood pressure for bendroflumethiazide). Interestingly, discontinuation of opioids and benzodiazepines, which we consider symptomatic treatments, were more often successful than the opposite despite the added risk of withdrawal symptoms.
In Figure \ref{222994}, other patterns regarding discontinuation and rebounding are apparent. For some medicine groups, the medication review results in fast and lasting reductions compared with usual care, i.e. for proton pump inhibitors (drugs for peptic ulcer and gastro-oesophagael reflux disease), opioids, paracetamol/acetaminophen (other analgesics and antipyretics), antithrombotic agents, antiepileptics, furosemide (high ceiling diuretics), and potassium (which follows furosemide prescriptions). For other medicines, the difference in discontinuations is not persistent, suggesting that the same changes will happen in the usual care group but over a longer time period, i.e. for drugs against constipation, vitamin B12 and folic acid, blood glucose-lowering drugs excl. insulins, statins (lipid modifying agents, plain), selective calcium channel blockers with many cardiovascular effects, and hypnotics and sedatives. For antidepressants and adrenergic inhalants, the medication review discontinuations seem to be unsuccessful with discontinuation of antidepressants being the least successful. Lastly, the calcium supplement discontinuations are outnumbered by new prescriptions leading to an overall and similar increase in prescriptions in both groups.
The reasons for the discontinuations in the medication review group listed in Table 2 clearly show that lack of indication is by far the most common reason to discontinue. Note that only the primary reason is listed, hence safety-related concerns may also apply when the primary reason for discontinuation was lack of indication. Nonetheless, it is evident that to reduce overprescription, a thorough review of the patient’s medical history is needed to understand the (original) indication for the patient’s treatments. In cases where the indication is not obvious, the patient or the patient’s family or GP may have this valuable information, further highlighting the need for a strong patient involvement and cross-sectoral collaboration when conducting medication reviews.
The results in Table 3 show that many of the medicines that were more often discontinued in the medication review group than in the usual care group are medicines which can negatively impact, especially older patients’, HRQoL. For example, metoclopramide, citalopram, tramadol, tiotroprium and zopiclone may commonly cause dizziness and other important adverse effects. Even low dose aspirin may cause significant bleeding15 and quinine may increase mortality risk in heart failure patients.16 So, while a medication review is a complex intervention, the changes to the medicine may very well be one of the causes of the observed positive effects on HRQoL and mortality.
Medication reviews are labor intensive as they require a comprehensive evaluation of the patient’s medical history and prescriptions to accurately evaluate their current medication use and indications. Therefore, it is important to identify the patients that would benefit most from medication reviews to optimize the allocation of health resources. Table 5 lists the factors that were statistically significantly associated with the number of overprescribed medicines for this population. Only two factors were associated with fewer overprescribed medicines; not being motivated for medicine changes or being referred from the geriatric department. Patients who were not motivated for medication changes had 26% fewer overprescribed medicines at the first visit. It is unknown whether this is due to them having fewer overprescribed medicines prescribed or an unwillingness to consent to the proposed medicine changes. Still, even patients that were not motivated for medicine changes had a median of 3 medicines discontinued or reduced in dosage. Patients referred from the geriatric department had 25% fewer overprescribed medicines, most likely due to some extent of medication reviews being performed prior to referral. Multiple factors were associated with an increased number of overprescribed medicines. The number of medicines was an important factor with 8% more overprescribed medicines per additional medicine prescribed at baseline. This is similar to the finding by Steinman et al.17 and intuitively makes sense as more medicines impair overview while offering more opportunities for mistakes. However, not only the number of medicines but also the sedative and anticholinergic burden of the medicines (as measured with the Drug Burden Index) impacted the number of overprescribed medicines with 15%–26% (full model or univariate model) more overprescribed medicines per one increase in Drug Burden Index. Besides the patient characteristics, we also identified some medicines that seem to be proxies for suboptimal medicine treatment. Thus, a prescription of metoclopramide was associated with a 42%–73% (full model or univariate model) increase in overprescribed medicines. Thus, metoclopramide, non-steroid anti-inflammatory drugs (propionic acid derivatives), and drugs for urinary frequency and incontinence are all medicines that should serve as an alarm bell that may prompt medication reviews in patients exposed to polypharmacy.

Strengths and limitations

The strength of this study is the completeness of data and long follow-up of medicine changes due to medication reviews thatoverall resulted in improved clinical outcomes for the patients. The main limitation is that we cannot correlate individual medicine changes to these outcomes due to the design of the study. Also, all analyses in this paper are post-hoc explorative studies that require further confirmative studies.