DISCUSSION
The prevalence of prescribing issues—PIMs/PIP, PPOs, and drug interactions—reached almost 90% and is a cause for concern among older HIV-infected patients. We identified several differences according to the criteria applied to identify these issues.
Older PLWH are often affected by comorbidities and polypharmacy [24-26] and can therefore be considered an at-risk population for prescribing issues and, consequently, for ADRs. However, to our knowledge, only 3 published studies have assessed this issue in PLWH [27-29]. Our results from the Over50 cohort show that almost all of the patients aged ≥65 years presented at least 1 prescribing issue; half had received at least 1 medication that had to be stopped (PIM by Beers or PIP by STOP). This high rate is similar to that detected in San Francisco in the USA [27,28]. One study detected at least 1 PIM in 52% of patients in a group of 89 PLWH based on the Beers criteria [27]. A similar prevalence was seen in a second study [28], which included a larger sample (211 patients) and compared both the STOPP/START and the Beers criteria, as did we. Of note, prescribing issues were more prevalent when the Beers criteria were applied than when the STOPP criteria were applied (63% vs 54%, respectively). Since the population studied was American, this result was expected. Data from the Swiss cohort [29] show that the prevalence of PIMs in 122 patients aged > 65 years based only on the Beer criteria was lower than that observed among our patients based on the Beers and the STOPP criteria, around one third of the population compared with a half, respectively. More than half of the medications for which the Beers criteria apply are not available in Spain, and this could be one of the reasons why prescribing issues in Europe are more frequently detected with the STOPP criteria than with the Beers criteria [12]. In this sense, these data indicate that the STOPP/START criteria are more appropriate than the Beers criteria for evaluation of prescribing issues in a European population.
Benzodiazepines were the drugs most frequently involved with PIMs in our and the Swiss cohort [29]. These medications are widely prescribed to the patients we attend, and those with long-term effects should be avoided in elderly patients at risk of falls and in those receiving other drugs that act on the central nervous system. In the American population, however, testosterone was the most frequently involved drug, probably owing to more frequent prescription of this drug among older American PLWH. In contrast, vaccines and vitamin D supplements were the recommended drugs least often prescribed. In line with national guidelines, it is also necessary to recommend vaccines and vitamin D supplements among older PLWH. Therefore, it is very important to carefully and regularly check the concomitant treatment of elderly PLWH at each medical visit, especially in cases involving multiple comorbidities and polypharmacy where different prescribers are involved [29-31]. This routine avoids unnecessary medications, adverse events, and drug-drug interactions.
Finally, our results also illustrate the relevance of choosing the most appropriate antiretroviral drug in elderly PLWH. As expected, antiretroviral combinations including the booster cobicistat were more frequently associated with drug-drug interactions. For that reason, these combinations should be proactively monitored. Recent studies support the antiviral efficacy and safety profile of ART combinations without a booster and some dual regimens [32-35] in order to decrease the risk of drug-drug interactions associated with boosters and prescription errors, as well as the risk of ADRs. In fact, despite the lack of formal recommendations on optimal therapy for older PLWH, cohort data demonstrate that the presence of comorbidities and polypharmacy led the physician to prescribe unusual ART combinations, mostly dual regimens [36].
Our study is potentially limited by its small sample size and by the use of the previous version of the Beers criteria (although the modifications of the 2015 revision made little difference to our results) [37] . However, the data resulting from our analysis are solid and relevant and add to the scarce published data. In addition, they highlight the need for changes in our routine practice.
In summary, prescribing issues were very prevalent among our older PLWH, with the most frequent being PIMs/PIP that could lead to avoidable ADRs. This high prevalence requires us to optimize treatment strategies by implementing a multidisciplinary critical review of our treatment plan as part of routine practice [30,38]. Concomitant medications should be more carefully and regularly monitored, especially in patients with multiple comorbidities and polypharmacy. Boosted antiretroviral combinations in particular should be used with caution. Although no standardized guidelines are available for PLWH, we consider the STOPP/START criteria to be a useful tool for European populations, since they seem to better detect prescribing issues, as they do in the general population.