Medication‐related problems in older people in Catalonia: A real‐world data study

Abstract Purpose The aim of this study was to determine medication‐related problems (MRPs) in primary care patients over 65 years of age. Methods Cross‐sectional study based on the electronic health records of patients (65‐99 years of age) visited in 284 primary health care centers during 2012 in Catalonia. Variables: age, sex, sociodemographic variables, number of drugs, kidney and liver function and MRPs (duplicate therapy, drug‐drug interactions, potentially inappropriate medications [PIMs] and drugs contraindicated in chronic kidney disease and in liver diseases). Unconditional logistic regression models were used to identify the factors associated with MRPs in patients with multimorbidity. Results 916 619 older people were included and 853 085 of them met the criteria for multimorbidity. Median age was 75 years and 57.7% of them were women. High percentages of MRPs were observed: PIMs (62.8%), contraindicated drugs in chronic kidney disease (12.1%), duplicate therapy (11.1%), contraindicated drugs in liver diseases (4.2%), and drug‐drug interactions (1.0%). These numbers were higher in the subgroup of patients with ≥10 diseases. The most common PIMs were connected to drugs that increase the risk of fall (66.8%), antiulcer agents without criteria for gastroprotection (40.6%), and the combination of drugs with anticholinergic effects (39.7%). In the multivariate analysis, the variables associated with all MRPs among the patients with multimorbidity were the number of drugs and the number of visits. Conclusions The coexistence of multimorbidity and polypharmacy is associated with an elevated risk of MRPs in older people. Medication safety for older patients constitutes a pressing concern for health services.


| Data sources
The Information System for Research in Primary Care (SIDIAP) contains anonymized clinical information originating from the primary care electronic health records (EHR) since 2006. 11 The medication database includes drugs subsidized, dispensed and billed by the national health system. For the purpose of identifying problems related to the most relevant medications, the study only included systemic drugs, and excluded hospital medication, drugs dispensed by hospital pharmacies, drugs not subsidized by public health services and topical medication (eg, ointments and lotions).

| Variables
All variables were obtained directly from the SIDIAP database. 12

| Chronic diseases and multimorbidity
In the SIDIAP database, diseases are coded in accordance with the International Classification of Diseases, version 10 (ICD-10). An operational definition of multimorbidity was used, that is, the presence of two or more chronic diseases, based on the selected 60 groups of chronic diseases determined by the Swedish National study of Aging and Care in Kungsholmen (SNAC-K), 13 with additional clinical, laboratory and medication-related parameters for the assessment of certain conditions. The number of different chronic diseases per patient was classified in four categories: 0-1 (non-multimorbidity); 2-4; 5-9; ≥10 chronic diseases.
Information on drug exposure was obtained from the Pharmacy Invoice Registry, which includes drugs prescribed by primary care and

KEY POINTS
• The increase in life expectancy is associated to an increase in multimorbidity.
• Older people belong to a population subgroup that is very prone to suffering from MRPs.
• MPRs are very prevalent in older people with multimorbidity, mainly in the subgroup of patients with 10 or more comorbidities, and with polypharmacy.
• There is a positive correlation between the number of drugs and the number of visits and the risk of MRPs.
• The clinical approach in older patients with multimorbidity and the safe use of the drugs which are prescribed to them represent two main challenges for health systems in order to reach a balance between the beneficial and drug side effects. hospital physicians. Drugs were classified according to the Anatomical Therapeutic Chemical Classification System (ATC). 14 The fourth and fifth levels were used to facilitate analysis and interpretation. Chronic use was considered when the person used three packages of the drug during the study period. Each drug was coded into a dichotomous variable. Polypharmacy was defined as the concurrent use of five or more drugs by the same individual. 15

| Kidney function
Kidney function was defined by two different parameters: a. Glomerular Filtration Rate (GFR): kidney function was calculated estimating the GFR according to the MDRD-4 IDMS10 equation. 16 Abnormal kidney function was considered when one or more values of GFR were <60 mL/min/1.73 m 2 .

| Other variables
Additional variables included in the study were socio-demographic variables such as age at baseline (years), sex (men, women), socioeconomic status (MEDEA index; quintiles from least to most deprived) 18 and number of total visits to PHC in 2012.

| Medication-related problems
The following MRPs were analyzed: duplicate therapy, drug-drug interactions, potentially inappropriate medications (PIMs) in older people (≥65 years old), contraindicated drugs in chronic kidney disease and contraindicated drugs in liver diseases. To analyze the MRPs, firstly we created the tables that contain the medicines or associations of medicines with potential safety concerns: • Duplicate therapy: practice of prescribing two or more medications with the same pharmacological activity. Duplicate therapies considered to pose significant clinical risk by professional consensus were included. The associations of active principles with the same pharmacological action that were used jointly to achieve a synergistic effect or to adjust doses were not considered (see Table S3).
• Drug-drug interaction: when the activity or effect of one drug is altered by the presence or the action of another. We prioritized interactions that might threaten life due to therapeutic failure or toxicity. We used the Thesaurus des interactions médicamenteuses from France's Agence nationale de sécurité du médicament et des produits de santé with the highest level of severity (contraindication). 19 This information was contrasted with a second source of information 20,21 or by professional consensus (see Table S4).
• Contraindicated drugs in chronic kidney disease: contraindicated medications were based on a consensus recommendation of the Catalan Health Department for patients with chronic kidney disease (see Table S5). 22 • Drugs contraindicated in liver disease: were based on the database of the Spanish College of Pharmacists (see Table S6). 23 • PIMs in older people (≥65 years old): were considered when the associated risk of adverse events was higher than the expected clinical benefits and when there was not clear scientific evidence for a specific indication or cost-effectiveness. We mainly used the STOPP/START criteria, 24 complemented with Beers' criteria, 25 PRISCUS and updated with others sources. 26 Table S7).

| Ethics approval and consent to participate
The protocol of the study was approved by the Clinical Research Ethics Committee of the Fundació Institut Universitari per a la Recerca a l'Atenció Primària de Salut Jordi Gol i Gurina (IDIAPJGol) (Protocol No: P17/080). All data were anonymized and confidentiality of EHR was guaranteed at all times in accordance with national and international law.

| Statistical analysis
Descriptive statistics were used to summarize overall information.
Categorical variables were expressed as frequencies (percentage) and continuous variables as mean (SD) or median (interquartile range, IQR). We used the Chi-square test and Mann-Whitney U test to assess differences between groups (multimorbidity groups and nonmultimorbidity group).
Prevalence of MRPs and use of PIMs were compared among groups.
Logistic regression models were fitted in order to identify the factors associated with each MRP (duplicate therapy, drug-drug interactions, contraindicated drugs in chronic kidney disease, contraindicated drugs in liver disease and PIMs) in the population with multimorbidity.
Variables included in the logistic regression models were sex, age, MEDEA index, number of drugs and number of visits.
The analyses were carried out using SPSS for Windows, version

| RESULTS
A total of 916 619 patients were included in the analysis. The median age was 75 years (IQR: 69-81) and 57.7% were women.
Out of 853 085 patients, 93.1% met multimorbidity criteria, with a higher prevalence in women (93.8% vs 92.1%, P < .001). In the multimorbidity group, the mean number of diagnoses per patient was 6.3 (SD: 3.0) and the mean number of medications was 5.6 (SD: 3.9). Table 1 shows the baseline characteristics of the study population.
The most common PIMs were drugs that increase the risk of fall (66.8%), the use of antiulcer agents without criteria for gastroprotection (40.6%), and the combination of drugs with anticholinergic effect (39.7%), for which 34 052 patients (5.9%) presented high anticholinergic load (score ≥ 3). These percentages are higher in the subgroup of patients with ≥10 comorbidities (see Table 2 for more details).  Only 29.4% patients with multimorbidity were free from MRPs.
Conversely, 21.8% presented a combination of two or more MRPs.
The most common combination was contraindicated drugs in chronic kidney disease and PIMs (10.8%), closely followed by duplicate therapy and PIMs (10.6%) (Figure 1). A 6.7% of patients with multimorbidity and contraindicated drug in chronic kidney disease and a 2.4% of patients with multimorbidity and contraindicated drug in liver disease had two or more contraindicated medication. A 16.2% of patients with multimorbidity who suffered from a PIM-type MRP had four or more PIMs. These percentages are higher in the subgroup of patients with ≥10 comorbidities (see Table S8).
The most common contraindicated drugs in cases of chronic kidney disease were metformin, hydrochlorothiazide, plain or in combination with enalapril, and citalopram; in liver failure, simvastatin, metformin and furosemide. The most common PIMs were the use of omeprazole without criteria for gastroprotection, the use of lorazepam, tamsulosin, lormetazepam and alprazolam because of the increased risk of fall, citalopram because of the increased risk of QT interval alteration and anti-inflammatories (ibuprofen) (see Tables S9-S11).
In the multivariate analysis, women showed a higher risk of MRPs, with OR that varied from 1.12 (95% CI: 1.10-1.14) to 1.24 (95% CI: The result of this study with an 11.1% of patients affected by duplicated drugs is higher than the 2.5% of hospitalized Italian patients and the 4.1% of Indian patients; but lower than the 39% obtained in an American primary care study. [35][36][37] In contrast, our results showed a low percentage of drug-drug interactions, probably due to the fact that we only included drugs that were explicitly contraindicated, and missed less relevant interactions, dietary supplements and other medications that do not require medical prescription. 38,39 The likelihood of a drug-drug interaction increase with the number of medicines, and studies indicate that it might reach up to 50% when taking 5 to 9 medications, as verified by the higher number of drug-drug interactions in patients taking ≥10 drugs. 4 With regard to the impact of medication on kidney function, we should underscore that measuring kidney function is complex. Our percentage of older people with multimorbidity who take drugs that are contraindicated in chronic kidney disease concurs with the results of other studies. However, it is not possible to draw a direct comparison since these other studies used different criteria. [40][41][42] In this study, the inclusion of a comprehensive list of commonly prescribed drugs in primary care might have influenced the high number of patients taking contraindicated drugs in chronic kidney disease. Alarmingly, almost 12.1% of patients with multimorbidity take drugs that are contraindicated in chronic kidney disease; and, in the case of people with ≥10 diseases, the contraindicated drugs in chronic kidney disease are 10 times more prevalent than in the multimorbidity group with two to four diseases. We strongly recommend strengthening automated warnings in the EHR to reduce these percentages.
There is a lack of data on medication management regarding liver disease in older primary care people. In hospital-based studies, the prevalence of contraindicated drugs in liver disease ranges from 8.5% to 30.6%. 43,44 A systematic review on PIMs showed a wide range of results (11.5%-62.5%), probably attributable to the different criteria used in the various studies. 45 51 However, multimorbidity has also been involved in many of these adverse events. Finally, these results correspond to a specific point in time.
Further studies should address the impact of MRPs when patients become older and multimorbidity increases, and also the impact of MRPs on patients' health and on the health services.

| CONCLUSIONS
The results of this study underscore the clinical significance of polypharmacy and multimorbidity, which correlate with a high risk of MRPs in patients over 65 years of age. The risk of MRPs could be attenuated with periodic reviews of medication and the implementation of automated warnings in the electronic prescription systems although the clinical impact of this reduction is not known. The inappropriately prescribed drugs most commonly related to safety issues were proton T A B L E 3 Odds of each medication-related problems according to clinical and sociodemographic variables in older people with multimorbidity (N = 853 085) pump inhibitors and anxiolytics. With the aging of the population,