Evidence-based application of explicit criteria to assess the appropriateness of geriatric prescriptions at admission and hospital stay

Background Inappropriate prescribing in the elderly is a critical issue in primary care, causing a higher risk of Adverse Drug Reactions (ADRs) and resulting in major patient safety concerns. At international level, many tools have been developed to identify Potentially Inappropriate Medications (PIMs). Objective The aim of this study was the application of Beers, Screening Tool of Older People's Prescriptions (STOPP)/Screening Tool to Alert to Right Treatment (START) and Improving Prescribing in the Elderly Tool (IPET) criteria as key tool to improve the quality of prescribing. Methods A retrospective study was conducted using the aforementioned criteria. Two different cohorts of elderly patients were enrolled between January 2015 and December 2016, 1800 at admission and 1466 at hospital stay. The index of each criterion divided by politherapy were correlated with comorbidities (Pearson correlation). A comparison was made between admission and hospital stay through a Student's t test of the average of the index. Results The Proton Pump Inhibitors (PPIs) were the most prescribed PIMs according Beers criteria in both patient cohorts (56%). The most detected drug-drug and drug-disease interactions at admission and at hospital stay were 3 or more drugs active on the Central Nervous System (CNS) as they can predispose to fall-risk. The most detected PIMs with STOPP criteria at admission were PPIs administered for more than 8 weeks. Inhaled β2-agonists or antimuscarinics were the most prescribed Potential Prescription Omissions (PPOs) according to START criteria. Nonsteroidal Anti-inflammatory Drugs (NSAIDs) in patients with high blood pressure were the most detected PIMs according to IPET criteria during hospital stay. A significant correlation between the comorbidities and the all index at hospital stay, while at admission there was no significant correlation for Beers and IPET index. Conclusion The prescriptive criteria were a useful tool for assessing the quality of prescriptions in the geriatric population and identifying their critical issues.


Results
The Proton Pump Inhibitors (PPIs) were the most prescribed PIMs according Beers criteria in both patient cohorts (56%). The most detected drug-drug and drug-disease interactions at admission and at hospital stay were 3 or more drugs active on the Central Nervous System (CNS) as they can predispose to fall-risk. The most detected PIMs with STOPP criteria at admission were PPIs administered for more than 8 weeks. Inhaled β2-agonists or antimuscarinics were the most prescribed Potential Prescription Omissions (PPOs) according to START criteria. Nonsteroidal Anti-inflammatory Drugs (NSAIDs) in patients with high blood pressure were the most detected PIMs according to IPET criteria during hospital stay.

Introduction
In recent decades, the demographic change in the population has resulted in an increase in life expectancy with a consequent rapid aging of the population worldwide. According to the World Health Organization's World Report on Aging and Health [1], in 2050 the global population aged 60 years or over will increase from the current 900 million to almost 2 billion. This increase in life expectancy coincides with an increase in multiple chronic diseases and conditions of non-self-sufficiency that require long-term assistance, causing a strong impact on all health systems in industrialized countries. Elderly patients, defined as patients over 65 years [2], are at increased risk of ADRs [3] due to age-related physiological changes with consequent alteration of the normal response to pharmacological therapy. These physiological changes are complex and depend on numerous factors, including the composition of the body mass, the health conditions of the various organs and the activity of the enzymatic systems [4][5][6].
All these conditions are emphasized by the use of multiple medications which makes it difficult to prescribe drugs with the right risk-benefit ratio [7]. The prescription of PIMs in the elderly is associated with an increased risk of ADR compared to the expected benefits of treatment, particularly when a safer or more effective alternative treatment is available for the same clinical condition. Inappropriate prescribing may be due to inappropriate dosage, incorrect duration of treatment, contraindicated use, drug-drug or drug-disease interactions. [8,9]. A further aspect of inappropriate prescription is the PPO of appropriate medicines based on the patient's age or the fact that he is already taking too many drugs.
At international level, many explicit criteria have been developed as tools to identify PIMs and ensure the safety of prescriptions in the elderly. These criteria are based on scientific evidence through the standard consensus search methodologies (Delphi consensus), consequently they need periodic updates to ensure their reliability. These are lists of drugs or classes of drugs to be avoided or used with caution in elderly patients [10]. The most used explicit criteria are the Beers criteria [11], the STOPP/START criteria [12] and the IPET criteria [13].
In particular, the Beers criteria were developed by the American Geriatrics Society and include the PIMs to be avoided in older adults divided into 6 lists: drugs to be avoided in most elderly or elderly patients with specific diseases, drug-disease interactions, drug-drug interactions, drugs to be used with caution, drugs for which dose adjustment is required based on kidney function and drugs with strong anticholinergic properties [11]. The STOPP/START criteria were developed in Ireland and include PIMs (STOPP criteria) and PPOs (START criteria), both grouped by anatomical classes [12]. The IPET criteria consist of a list of 14 PIMs identified by a panel of Canadian experts [13].
The choice of one criterion over another is complex and depends on numerous factors, such as drugs availability in the market and legislative measures in each country. STOPP/ START criteria, contrary to Beers and IPET criteria, take account of PPOs which may be just as important as PIMs in the overall assessment of drug appropriateness in the elderly. In this regard, it is useful to use all four criteria to detect a greater number of PIMs/PPOs and achieve more appropriate and safer therapies.
A possible implementation of these explicit criteria could be the algorithm-driven electronic medical record notification that enhances the detection of PIMs and PPOs during prescription, based on generalities of patient and taken drugs.
However, the evaluation of prescription appropriateness in the elderly represents a complex issue that should be based on clinical judgment and these criteria should be intended as a support tool in choosing the most appropriate drugs, avoiding those at greatest risk of ADR.
In this context, the aim of this study was to assess the medication use in elderly patients in terms of PIMs and PPOs applying Beers, STOPP/START and IPET criteria. In particular, a comparison was made between the PIMs and the PPOs detected at admission and those detected at the hospital stay.

Materials and methods
A retrospective observational study was conducted by the Tracer Pharmacist (TP) in 24 months at the Clinical Pharmacy Service of the Mediterranean Institute for Transplantation and Advanced Specialised Therapies (ISMETT), a Research Institute in Palermo, Italy. This study was approved by the Institutional Review Board (n. 20/17) and the Ethic Committee (Protocol No. 001-178-GEN/2018) of ISMETT and written informed consent was obtained from all patients. The data processing of the patients enrolled in this study was done in compliance with the provisions and prescriptions of the policies and procedures adopted by ISMETT and the current legislation, also with regard to the security measures adopted for the protection of the aforementioned data. All patients aged � 65 admitted to ISMETT between January 2015 and December 2016 were enrolled. Two different patient cohorts were analyzed, one relating to admission and one relating to hospital stay, and the respective information related to the clinical profile of patients and the PIMs or PPOs have been inserted into two specific databases.
In particular, for each patient, generalities such as age and gender, date of admission and discharge, diagnosis at admission and any comorbidities were detected. Clinical information such as the blood pressure, the international normalized ratio, the blood levels of potassium and sodium and the predisposing risk factors for falling were obtained through consultation of the electronic medical record Sunrise Clinical Manager (Eclipse) 1.
In order to assess the overall quality of drug prescribing in the elderly, four evidence-based criteria were used: the Beers criteria (2015 version), the STOPP/START criteria (2014 version) and the IPET criteria (2000 version). The Beers, STOPP and IPET criteria allow the detection of PIMs, while the START criteria allow the detection of PPOs.
All prescriptions once a day or of a duration not exceeding three days made during hospital stay were excluded from the analysis, because they were considered not clinically significant to justify each of the aforementioned criteria used.

Statistical analysis
GraphPad Prism 8 1 software was used for statistical analysis. Values of p <0.05 were considered statistically significant. The statistical analysis involved the evaluation of the means and standard deviations for the parametric variables. As part of the retrospective study, the criteria of prescriptive inapropriateness (Beers, STOPP and IPET) and prescriptive appropriateness (START) were expressed in the form of Index (I) of the criteria with polytherapy. In order to identify the outliers relating to the index of the criteria, the statistical analysis was preceded by a data cleaning phase, in which it was calculated the z scores [14], given by the ratio between the deviation of the average of the index (I -Ī) for standard deviation (SD): Outliers with a z > | 3 | score have been replaced with a maximum score of +3. The resulting index were correlated with the comorbidities detected at both admission and hospitalization (Pearson r correlation). The comparison of the admission and hospital stay index averages was made by estimating Cohen's d effect size [15] and Student's t test.

Results
Two different cohorts of elderly patients were enrolled in this study, 1800 at admission and 1466 at hospital stay. Characteristics of both cohorts of patients and the respective pathologies are shown in Tables 1 and 2. Table 1 shows an increase in the average consumption of drugs from admission to hospital stay (respectively from 5.39 ± 3.00 to 10.82 ± 6.09) in response to a concomitant increase in the number of registered pathologies (from 2.83 ± 1.58 to admission to 3.52 ± 2.12 to hospital stay). Hypertension, heart disease and history of falls and fractures were the most common comorbidities in both cohorts of patients ( Table 2).
As shown in the Table 3, during the hospital stay an increase of patients with PIMs was observed according to the Beers and IPET criteria, while they were reduced using the STOPP criteria. The START criteria, which instead concern the PPOs, are increased during hospitalization.
According to the Beers criteria, the type of PIMs most detected in both patient cohorts (56%) were the potentially inappropriate drugs, of which the PPIs were the most prescribed ( Table 4). The most detected drug-drug and drug-disease interactions according to the Beers criteria both at admission and at hospital stay were, instead, 3 or more drugs active on the CNS as they can predispose to risk of falling.
On the other hand, the PIMs to be used with great caution in elderly patients according to Beers' criteria decreased from 32% to admission to 23% to hospitalization; diuretics were the most prescribed drugs as they can cause or exacerbate the Syndrome of Inappropriate Antidiuretic Hormone secretion (SIADH). PIMs based on kidney function and PIMs with strong anticholinergic properties according to Beers' criteria remained almost unchanged, with low percentages both at admission and during hospitalization (respectively 2% and 1%). In particular, the PIMs most found were enoxaparin in patients with creatinine clearance less than 30 mL/min and paroxetine as anticholinergic drug. Regarding STOPP criteria, there was a reduction in PIMs at hospital stay compared to admission (86% and 85% respectively). As can be seen in Table 5, the PIMs most detected with the STOPP criteria have been PPIs for more than 8 weeks at admission, while at the hospital stay the vasodilators in patients with orthostatic hypotension and the drugs that can cause constipation were the most detected PIMs.
The application of the START criteria has shown an increase of the percentage of patients with at least one PPO, from 72% at admission to 77% at the hospital stay. This increase is probably due to the increase in prescriptions of inhaled β2-agonists or antimuscarinics alone or in association with corticosteroids for asthma or Chronic Obstructive Pulmonary Disease (COPD).
The comparison resulting from the application of the IPET criteria at admission and at hospital stay revealed an increasing trend of patient with PIMs, from 11% on admission to 17% on admission. 44% of the detected PIMs during hospital stay were related to the prescription of NSAIDs in patients with high blood pressure ( Table 6).
The statistical analysis, carried out through Pearson's correlation r of the index with the comorbidities, showed a significant correlation between the comorbidities and the index of PIMs and PPOs at hospital stay. At admission, however, there was no significant  correlation between the index relating to the Beers and IPET criteria and the comorbidities ( Table 7). From the comparison of the average index relating to each criterion, it was lower at the hospital stay than at admission (Table 3) for all criteria due to the fact that, as can be seen from Table 1, the average of the drugs registered at admission was approximately half that of the hospital stay. The calculated Coehn's d was of medium level for the STOPP and IPET criteria (respectively -0.488 and -0.401) and of small level for the Beers and START criteria (respectively -0.124 and -0.092). The two-tailed Student's t test carried out showed significant differences emerged as regards the Beers criteria (t = 3,353, df = 3258, p < 0.0001), STOPP (t = 13.30, df = 3258, p < 0.0001) and START (t = 10.88, df = 3258, p < 0.0001). However, the differences were not significant for the IPET criteria (t = 1,908, df = 3258, p < 0.0001). This could be due to the fact that the latter consist of 14 items and therefore are less frequently detected than the other criteria consisting of many more items (i.e. 260 Beers, 75 STOPP and 33 START).

PLOS ONE
Application of explicit criteria to assess the geriatric prescription appropriateness

Discussion
The analysis of geriatric therapies carried out at admission and at hospitalization showed an increase in the average consumption of drugs during the hospital stay. This propensity for polytherapy is attributable to differences often due to the changed clinical conditions of the patient during his hospitalization. In this context, the correct implementation of the medication reconciliation in the transition of patients between care settings represents an essential tool for making safe and appropriate prescriptive decisions [16]. PPIs were the most prescribed PIMs according to Beers criteria in both cohorts of patients. In fact, prolonged use of these drugs has been associated with a higher risk of osteopenia and osteoporosis, with consequent predisposition to bone fractures [17][18][19]. This significant association between proton pump inhibitors for protracted periods and osteopenia is probably due to impaired absorption of calcium, vitamin B12, iron and magnesium [20][21][22] which can change bone mineral density. Prolonged use of proton pump inhibitors can also increase the risk of Clostridium difficile infections [23]. Therefore, Beers' criteria recommend avoiding its use for more than 8 weeks, except in high-risk patients, such as patients with erosive esophagitis, Barret's esophagitis, pathological hypersecretion, ongoing therapy with corticosteroids or oral NSAIDs, or therapeutic failure of H2-antagonists. The Beers criteria also recommend minimizing the number of CNS active drugs (such as opioids, benzodiazepines and non-benzodiazepine hypnotics, antidepressants and antipsychotics) to a maximum of 3 [24], especially in elderly patients at risk of falling and fractures. This aspect is particularly important in the hospital setting, considering that falls in hospitalized patients represent the most common adverse event and can lead to an increase in hospital stay and additional diagnostic and therapeutic activities. The Standards of the Joint Commission International, of which ISMETT has been an accredited structure since 2009, also pay particular attention to the evaluation of the risk of falling for patients and a possible revaluation in case of modification of the health conditions during hospitalization [25]. PPIs administered for more than 8 weeks, detected as main PIMs according to the STOPP criteria at admission, were not detected with the same occurrence at the hospital stay, probably because the percentage of inpatients at ISMETT for more than 31 days was relatively low (Table 1). Vasodilators (α1-antagonists, Ca-antagonists, long-acting nitrates, ACE inhibitors, sartans, diazoxide, minoxidil, hydralazine) were, instead, the most detected PIMs according to the STOPP criteria during hospital stay in patients with persistent postural hypotension, as they can predispose to risk of syncope and falls. In this regard, the 2018 guidelines of the European Society of Cardiology for the diagnosis and treatment of syncope suggest that the reduction or suspension of hypotensive therapy aimed at a systolic pressure of 140 mmHg is effective in reducing syncopal recurrence in predisposed patients [26]. The other frequently reported PIM with the STOPP criteria during hospitalization was related to drugs causing constipation (anticholinergics, oral iron, opioids, verapamil, aluminum-based antacids) in patients with chronic constipation, with consequent risk of exacerbation of constipation. In this regard, during hospitalization, the percentage of PIM referring to the prescription of opioids without concomitant laxative therapy also increased. This aspect is very important considering that chronic constipation is a very common pathophysiological phenomenon in elderly patients, especially women, due to the slowing of intestinal transit, the reduction of sensorimotor functions and the atrophy of the anal sphincters. Therefore, the prevention of iatrogenic constipation is fundamental [27]. Regarding the START criteria, the most detected PPOs were the β2-agonists or antimuscarinics inhaled alone or in combination with corticosteroids for asthma or COPD. In this regard, the most recent revision of the Global Initiative for Chronic Obstructive Lung Disease (GOLD 2019) [28] reaffirms that the inhaled administration of bronchodilators is recommended over the oral route, and that associations of bronchodilators with different mechanism and duration of action can improve the degree of bronchodilation, with a lower risk of adverse events than the dosage increase of a single bronchodilator [29][30][31].
The main PIM found with the IPET criteria during hospitalization was the prescription of NSAIDs in patients with high blood pressure. In general, they can increase blood pressure due to the inhibition of the synthesis of prostaglandins, which leads to a direct effect in modulating renal blood flow [32]. In addition, NSAIDs can cause an increase in serum aldosterone, with consequent sodium retention and hypertensive effect [33].
The statistical analysis showed a significant correlation between the comorbidities and the index of prescriptive inappropriateness and appropriateness at the hospital stay. Therefore, as shown by the descriptive statistics, during the hospital stay, the impact of comorbidities on polytherapy and, consequently, also on the index relating to the criteria is significant. On admission, however, there was no significant correlation between the index relating to the Beers and IPET criteria and the comorbidities. These results are probably due to the fact that upon admission the prescribed therapies are greatly influenced by the variability conferred by the numerous general physicians who have taken care of the elderly patients.

Conclusion
The activity carried out in this retrospective study has highlighted the main PIMs and PPOs in admitted and hospitalized patients at our Institute. A propensity to polytherapy emerged during hospital stay. The prescriptive inappropriate criteria were a useful tool for assessing the quality of prescriptions in the geriatric population and identifying their critical issues. An evidence-based approach based on the application of these criteria can allow to raise the standards of care and improve decision making in geriatric prescriptions. To make physicians aware of the use of these tools and to improve care for the elderly patients an educational brochure has been created. Furthermore, the management of drug therapy in the elderly patient requires a multidisciplinary approach, in which the TP can support physicians in choosing the safest and most appropriate therapies for elderly patients. A possible limitation of our study, which will be implemented later, is that two different cohorts of elderly patients were analyzed at admission and hospitalization, the advantage is represented by the sample size. Furthermore, the quality of pharmacological prescriptions could also be assessed at discharge as a tool to prevent further hospital readmissions following drug-related events [34]. This would increase the safety and appropriateness of the drugs prescribed in all transitions of care (admission, hospital stay and discharge). Other strategies to be implemented could be the reduction of the number of taken drugs through the support algorithms for de-description [35] or the use of non-pharmacological approaches to treat common diseases of elderly patient [36,37].