Villén et al. BMC Geriatrics (2020) 20:206 https://doi.org/10.1186/s12877-020-01580-1 RESEARCH ARTICLE Open Access Multimorbidity patterns, polypharmacy and their association with liver and kidney abnormalities in people over 65 years of age: a longitudinal study Noemí Villén1, Marina Guisado-Clavero2,3, Sergio Fernández-Bertolín2,3, Amelia Troncoso-Mariño1, Quintí Foguet-Boreu2,3,4, Ester Amado1, Mariona Pons-Vigués2,3,5, Albert Roso-Llorach2,3*† and Concepción Violán2,3*† Abstract Background: The implementation of individual clinical practice guidelines in patients with multimorbidity often results in polypharmacy. Our aim was to analyse medication use according to longitudinal multimorbidity patterns (MP) and determine during a 5-year period (2012–16) which MP are associated with abnormal liver and kidney function in primary care patients over 65 years of age living in Catalonia. Methods: Design: Longitudinal study (years 2012 to 2016) based on the electronic health records contained in Information System for Research in Primary Care database of the Catalan Institute of Health (SIDIAP). Variables: age, sex, MP, medication and polypharmacy (drug exposure obtained from the Pharmacy Invoice Registry). Medicines were classified in accordance with the Anatomical Therapeutic Chemical Classification System (ATC). Glomerular filtration rate was used to determine abnormal kidney function, and serum levels of alkaline phosphatase, alanine transaminase and gamma-glutamyl transpeptidase were used to diagnose abnormal liver function. Statistics: For medication use in MP, we calculated annual mean packages of each drug in each MP, and observed/expected ratios were obtained by dividing mean packages in the cluster by mean packages of the same drug in the overall population. Logistic regression models were fitted to estimate the association between MP at baseline and abnormal kidney and liver function tests during follow up. (Continued on next page) * Correspondence: aroso@idiapjgol.org; cviolan@idiapjgol.org † Albert Roso-Llorach and Concepción Violán contributed equally to this work. 2 Fundació Institut Universitari per a la recerca a l’Atenció Primària de Salut Jordi Gol i Gurina (IDIAPJGol), Barcelona, Spain Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Villén et al. BMC Geriatrics (2020) 20:206 Page 2 of 13 (Continued from previous page) Results: Nine hundred sixteen thousand six hundred nineteen patients were included, and 743,827 completed the follow up. We identified one polypharmacy profile per MP, and concluded that the most prescribed drugs in each pattern corresponded to the diseases overrepresented in that specific MP. The median of drugs ranged from 3 (Cluster 1 - Non-Specific) to 8 (Cluster 10 - Multisystem Pattern). Abnormal kidney function was most commonly observed in the Cluster 4 - Cardio-Circulatory and Renal (Odds Ratio [OR] 2.19; Confidence interval [CI] 95% 2.15– 2.23) and Cluster 3 - Minority Metabolic Autoimmune-Inflammatory (OR 2.16; CI 95% 2.12–2.20) MP. A higher risk of abnormal liver function was observed in the Cluster 8 - Digestive (OR 3.39; CI 95% 3.30–3.49), and Cluster 4 - Cardio-Circulatory and Renal (OR 1.96; CI 95% 1.91–2.02) MP. Conclusions: A higher risk of abnormal kidney and liver function was observed in specific MP. The long-term characterisation of MP and polypharmacy illustrates the burden of chronic multimorbidity and polypharmacy in the elderly population. Keywords: Multimorbidity, Epidemiology, Polypharmacy, Primary health care, Ageing, Longitudinal study Background The pharmacological management of multimorbidity Multimorbidity generally refers to the co-occurrence of represents a challenge for health professionals, especially multiple chronic medical conditions in a single individ- in elderly or frail patients, since it requires a person- ual [1]. Globally, multimorbidity is growing into a major centred approach as opposed to treating multiple dis- concern due to the longer lifespan of the population, the eases as independent entities [14]. Few publications complexity of their health status and its relation to address medication use according to MP, and to our greater use of healthcare services [2, 3]. Additionally, liv- knowledge this is the first study which includes long- ing with multiple chronic diseases is commonly associ- term data on this issue [15–19]. ated with the use of numerous drugs, function decline, In another article, we described MP in people over 65 lower of quality of life and increased mortality [4, 5]. years of age and followed their trajectories during 5 Generally, prescription in patients with multimorbidity years. We used soft clustering techniques, which enabled is based on the recommendations of individual disease- us to study MP considering as the unit of analysis the specific clinical practice guidelines. This frequently re- patient plus all their diagnosed chronic diseases. We sults in polypharmacy [6], commonly defined as the identified multimorbidity trajectories which were mostly concomitant consumption of 5 or more drugs [7]. Poly- stable [20]. In addition, we observed that nine over ten pharmacy is closely related to drug-drug and drug- patterns involved a large number of chronic diseases and disease interactions, therapeutic redundancies and ad- drugs. However, that study did not analyse the charac- verse drug reactions that can cause hospital admission, teristics and use of medicines associated with MP over worsen chronic conditions and increase morbidity and time, and their association with liver and kidney function mortality, especially in the elderly and frail population abnormalities. [8–10]. Moreover, many drugs used to treat chronic dis- The main aim of this study was to analyse medication eases have a modest benefit at most and an increased use according to longitudinal MP and characterize them risk of adverse reactions in this population. annually in primary care patients over 65 years of age The dynamic changes in cellular biological processes during the 2012–16 period in Catalonia. The secondary associated with aging [11] influence the pharmacokinet- aim was to determine which patterns are more likely to ics and pharmacodynamics of drug metabolization [12]. present an abnormal liver and kidney function during Importantly, the mitochondrial function decline linked follow-up. to the aging process increases the vulnerability of organs such as heart, lungs, liver and kidney to ischemia/reper- Methods fusion injury [13]. Consequently, understanding the link Design, setting, and study population between polypharmacy and specific multimorbidity pat- A longitudinal study was carried out in Catalonia terns (MP) and their interrelation with liver and kidney (Spain), a Mediterranean region of 7,515,398 inhabitants function decline is crucial for the design of strategies (2012) [21]. The Spanish National Health Service pro- that prevent adverse effects of medication among the vides universal coverage, financed mainly by tax revenue. elderly. Additionally, there is a lack of longitudinal infor- The Catalan Health Institute (CHI) manages 285 pri- mation about the effects of polypharmacy on adverse mary health care centres (PHCCs) in Catalonia that drug reactions and dynamic cellular changes associated serve 5,501,784 people, corresponding to the 77.2% of with aging. the population [22]. Villén et al. BMC Geriatrics (2020) 20:206 Page 3 of 13 Inclusion criteria were individuals aged 65–99 years on Death and drop out 31 December 2011 that survived until 31st of December Death was measured as the occurrence of this event, re- 2012 (index date), with at least one PHCC visit during gardless of cause. If a person transferred out of the CHI the 5-year study period (2012–2016). No new entries healthcare provider during the study period, they were were allowed in the cohort. Attrition was caused by considered lost to follow-up (drop-outs). mortality or transfer to another healthcare provider. We included 916,619 people at baseline and 743,827 com- Drugs and polypharmacy pleted all follow-up (Fig. 1). Information on drug exposure was obtained from the Pharmacy Invoice Registry, which includes drugs pre- Data sources scribed by primary care and hospital physicians and sub- The Information System for Research in Primary Care sidized by the national health system. The study (SIDIAP) contains the clinical information from the included only systemic drugs, and excluded hospital electronic health records (EHR) of the CHI PHCCs since medication, drugs dispensed by hospital pharmacies, 2006 [23]. The SIDIAP database includes anonymized drugs not subsidized by public health services and top- longitudinal EHR from primary and secondary care that ical medication (e.g. ointments and lotions). Drugs were collect information on demographics, symptoms, diag- classified according to the Anatomical Therapeutic noses, prescriptions, socio-economic status and the Chemical Classification System (ATC) [25]. We used the Pharmacy Invoice Registry . 4th level (Chemical, Pharmacological or Therapeutic subgroup) to facilitate analysis and interpretation. Chronic use was considered when the person used three Variables packages of the drug during each year of the study All variables were obtained directly from the SIDIAP period. Each drug was coded into a dichotomous vari- database [23]. able. Polypharmacy was defined as the concurrent use of 5 or more drugs in the same individual for each year of Chronic diseases and multimorbidity the study [7]. In the SIDIAP, diseases are coded in accordance with the International Classification of Diseases, version 10 Kidney function (ICD-10). An operational definition of multimorbidity Kidney function was calculated each follow up year was used, i.e., the presence of two or more chronic dis- using the four-variable Modification of Diet in Renal eases, based on the selected 60 groups of chronic dis- Disease (MDRD-4) creatinine calibrated to isotope dilu- eases determined by the Swedish National study of tion mass spectrometry (IDMS10) equation to estimate Aging and Care in Kungsholmen [24], with additional glomerular filtration rate (eGFR) [26]. When one or clinical, laboratory and medication related parameters more eGFR was < 60 ml/min/1.73 m2 after the baseline for the assessment of certain conditions. year, we classified it as abnormal kidney function. Fig. 1 Longitudinal Flow Chart of study period (Years 2012-2016; N=916,619 people) Villén et al. BMC Geriatrics (2020) 20:206 Page 4 of 13 Liver function applied machine learning technique can be found else- Liver function was determined each follow up year by where [30]. analysing serum levels of alkaline phosphatase (ALP), alanine transaminase (ALT) and gamma-glutamyl trans- Medication use in multimorbidity patterns peptidase (GGT). Liver function was considered abnor- To describe chronic medication use in each multimor- mal when after baseline, at least one of the values was as bidity pattern (MP), we calculated annual mean packages follows: ALP > 2 × 129 IU/L; ALT> 5 × 41 IU/L (men) or for each drug in each cluster. Next, we defined O/E-ra- ALT> 5 × 33 IU/L (women); GGT > 61 IU/L [27]. tios, obtained by dividing mean packages in the cluster by mean packages in the overall population. We also de- Other variables fined exclusivity as the number of medication packages Demographic variables analysed at baseline and at the included in the cluster over the total number of medica- end of study were: age at baseline (years), sex (men, tion packages. A medication was considered part of a women), socio-economic status (Mortality in Spanish MP when the O/E-ratio was ≥2. As a complementary ap- small areas and Socioeconomic and Environmental In- proach, we also investigated medicines that presented an equalities [MEDEA] index); quintiles from least to most exclusivity value higher than the percentage frequency of deprived) [28] and number of total visits to PHCCs. each cluster. Logistic regression models were fitted to estimate the association between MP at baseline and abnormal kid- Statistical analysis ney and liver function values during follow up. Odds ra- Descriptive statistics were used to summarize overall in- tio (OR) and 95% confidence interval (CI) were adjusted formation. Participants’ characteristics and prevalence of for age, sex, socioeconomic status (MEDEA variable) chronic diseases and medication use were measured at and total number of packages in 2012. We used multiple baseline and final year, and during the follow-up period. imputation to minimize the selection bias resulting from MP were identified using two steps: 1) the dataset was the presence of missing values for MEDEA (7%). We pre-processed by applying a mixture of Principal Com- used multiple imputation by chained equations to obtain ponent Analysis and Multiple Correspondence Analysis seven imputed datasets [31]. The final models were fit- (PCAmix). For the grouping of participants, it was as- ted with multiple imputed datasets using Rubin’s rules sumed that the patient population was initially distrib- to combine effect estimates and standard errors to allow uted into a fuzzy set of clusters, corresponding to the for the uncertainty related to missing data. different MP [29]; 2) In order to model the temporal The analyses were carried out using R version 3.6.1 (R evolution of individuals and clusters, the sequential indi- Foundation for Statistical Computing, Vienna, Austria). vidual observations were assumed to follow a dynamic The significance level was set at α = 0.05. random process represented by a Hidden Markov Model (HMM). The set of HMM parameters was fitted to the observation data by applying the Baum-Welch algo- Results rithm. Afterwards, a validation process was applied in A total of 916,619 people over 65 years of age living in order to obtain the optimal parameters and longitudinal Catalonia were included at baseline (women: 57.8%; clusters. The best cluster trajectory is computed by mean age: 75.4; standard deviation [SD]: 7.4), 853,085 maximizing the probability of the observed sequence, (93.1%) met multimorbidity criteria, and 487,502 (53.2%) conditioned to the best computed model parameters criteria for polypharmacy. At the end of the study (Viterbi Algorithm). Next, we associated each cluster period, 743,827 participants had completed follow-up with a hidden state or MP to characterize it. (Fig. 1). The mean number of chronic diseases was 6 The final clusters were defined as MP by the research (interquartile range [IQR] 4.0–8.0) at baseline and 7 team based on the clinical significance. We eventually (IQR 5.0–10.0) in the final year, whereas mean number obtained ten MP, and also drop out and mortality clus- of medicines during follow-up was 5 (IQR baseline 2.0– ters. To describe the MP, prevalences of diseases in each 8.0 and IQR at final year 3.0–8.0). The proportion of pa- cluster were calculated. Observed/expected ratios (O/E- tients with multimorbidity and polypharmacy was higher ratios) were obtained by dividing disease prevalence in at the end of the study period than at baseline (93.1 and the cluster by disease prevalence in the overall popula- 53.3% vs 97.2 and 56.8%, respectively) (Table 1). tion. Exclusivity, defined as the proportion of patients The most prevalent invoiced drugs, from highest to with the disease included in the cluster over the total lowest, were proton pump inhibitors (PPI), 3-hidroxi-3- number of patients with the disease, was also calculated. metil-glutaril-CoA reductase (HMGCoA-reductase) in- A disease was considered to be part of a MP when the hibitors and anilides. During follow-up, the most com- O/E-ratio was ≥2. A more detailed description of the mon drugs remained unchanged, but the prescription Table 1 Variables characterizing each cluster in the study at baseline and at the end of the study (N = 916,619) Variables Cluster 1 - Non- Cluster 2 - Eye Cluster 3 - Minority Cluster 4 - Cluster 5 - Cardio- Cluster 6 - Cluster 7 - Cluster 8 - Cluster 9 - Cluster 10 - Overall (patients Specific Impairment and Metabolic Cardio- Circulatory, Mental, Neurological, Respiratory Digestive Neurological, Multisystem alive) Mental Autoimmune- Circulatory and Respiratory and Digestive and and Ear Musculoskeletal Pattern Inflammatory Renal Genitourinary Circulatory and minor N2012=384787 N2012=177087 N2012=72190 (7.9%) - N2012=60192 N2012=54324 N2012=42422 N2012=41644 N2012=35692 N2012=33283 N2012=14998 N2012=916619 - (42.0%) - N2016= (19.3%) - N2016= N2016=73237 (9.8%) (6.6%) - N2016= (5.9%) - N2016= (4.6%) - N2016= (4.5%) - N2016= (3.9%) - N2016= (3.6%) - N2016= (1.6%) - N2016= N2016= 743827 258700 (34.8%) 155008 (20.8%) 46244 (6.2%) 45463 (6.1%) 41120 (5.5%) 38961 (5.2%) 31744 (4.3%) 36522 (4.9%) 16828 (2.3%) Villén et al. BMC Geriatrics Sex: Female 201958 (52.5%) - 130572 (73.7%) - 31395 (43.5%) - 42638 (70.8%) 3618 (6.7%) - 23577 (55.6%) - 31143 (74.8%) - 19915 (55.8%) - 30830 (92.6%) - 13485 (89.9%) 529131 (57.8%) - 137931 (53.3%) 113574 (73.3%) 31490 (43.0%) - 32786 (70.9%) 2902 (6.4%) 22601 (55.0%) 28838 (74.0%) 17846 (56.2%) 33718 (92.3%) - 15005 (89.2%) 436691 (58.7%) Male 182829 (47.5%) 46515 (26.3%) - 40795 (56.5%) - 17554 (29.2%) - 50706 (93.3%) - 18845 (44.4%) - 10501 (25.2%) 15777 (44.2%) 2453 (7.4%) - 1513 (10.1%) - 387488 (42.2%) - - 120769 (46.7%) 41434 (26.7%) 41747 (57.0%) 13458 (29.1%) 42561 (93.6%) 18519 (45.0%) - 10123 (26.0%) - 13898 (43.8%) 2804 (7.7%) 1823 (10.8%) 307136 (41.3%) (2020) 20:206 Patients with 323153 (84.0%) - 176734 (99.8%) 71562 (99.1%) - 60192 (100.0%) 54318 (100.0%) - 42126 (99.3%) - 41413 (99.4%) - 35306 (98.9%) - 33283 (100.0%) - 14998 (100.0%) 853085 (93.1%) - Multimorbidity: 238466 (92.2%) - 154922 (99.9%) 73039 (99.7%) - 46243 (100.0%) 45463 (100.0%) 41020 (99.8%) 38885 (99.8%) 31628 (99.6%) 36522 (100.0%) - 16828 (100.0%) 723016 (97.2%) Patients with 122956 (32.0%) - 110504 (62.4%) - 43115 (59.7%) - 51759 (86.0%) - 43829 (80.7%) - 30644 (72.2%) - 29416 (70.6%) - 20336 (57.0%) - 22739 (68.3%) - 12204 (81.4%) - 487502 (53.2%) - Polypharmacy: 84206 (32.5%) 99185 (64.0%) 44500 (60.8%) 40528 (87.6%) 38159 (83.9%) 29788 (72.4%) 28254 (72.5%) 18717 (59.0%) 25338 (69.4%) 13710 (81.5%) 422385 (56.8%) Age (Mean): 74.8 - 73.3 74.2 - 73.5 75.9 - 74.5 80.6 - 78.3 76.7 - 75.0 78.7 - 76.3 75.3 - 74.0 74.8 - 73.6 72.8 - 72.5 76.9 - 75.8 75.5 - 74.1 Age (n,%): [65,70) 121226 (31.5%) - 55687 (31.4%) - 17825 (24.7%) - 4711 (7.8%) - 10906 (20.1%) - 5578 (13.1%) - 10981 (26.4%) - 10135 (28.4%) - 12210 (36.7%) - 2919 (19.5%) - 252178 (27.5%) - 96106 (37.1%) 52749 (34.0%) 21292 (29.1%) 5461 (11.8%) 11395 (25.1%) 7904 (19.2%) 12061 (31.0%) 10472 (33.0%) 13773 (37.7%) 3788 (22.5%) 235001 (31.6%) [70,80) 162009 (42.1%) - 81567 (46.1%) - 31466 (43.6%) - 20490 (34.0%) - 24074 (44.3%) - 16968 (40.0%) - 19002 (45.6%) - 16359 (45.8%) - 16152 (48.5%) - 6499 (43.3%) - 394586 (43.1%) - 113532 (43.9%) 72827 (47.0%) 34163 (46.6%) 20320 (43.9%) 22198 (48.8%) 19730 (48.0%) 18753 (48.1%) 15101 (47.6%) 17933 (49.1%) 7863 (46.7%) 342420 (46%) [80,90) 86495 (22.5%) - 36744 (20.7%) - 20059 (27.8%) - 29143 (48.4%) - 17052 (31.4%) - 16937 (39.9%) - 10335 (24.8%) - 8316 (23.3%) - 4747 (14.3%) - 4916 (32.8%) - 234744 (25.6%) - 44749 (17.3%) 27946 (18.0%) 16495 (22.5%) 18693 (40.4%) 11113 (24.4%) 12481 (30.4%) 7616 (19.5%) 5851 (18.4%) 4682 (12.8%) 4798 (28.5%) 154424 (20.8%) [90,99] 15057 (3.9%) - 3089 (1.7%) - 2840 (3.9%) - 5848 (9.7%) - 2292 (4.2%) - 2939 (6.9%) - 1326 (3.2%) - 882 (2.5%) - 174 (0.5%) - 664 (4.4%) - 35111 (3.8%) - 4313 (1.7%) 1486 (1.0%) 1287 (1.8%) 1770 (3.8%) 757 (1.7%) 1005 (2.4%) 531 (1.4%) 320 (1.0%) 134 (0.4%) 379 (2.3%) 11982 (1.6%) MEDEA: R 82025 (22.7%) - 33553 (20.0%) - 15314 (22.7%) - 12276 (23.5%) - 9282 (19.1%) - 8888 (23.7%) - 7512 (19.4%) - 6396 (19.8%) - 4888 (15.4%) - 2115 (15.3%) - 182249 (21.4%) - 54244 (21.6%) 29758 (19.8%) 15667 (22.0%) 9279 (20.7%) 7490 (17.1%) 8089 (20.4%) 7006 (18.6%) 5650 (18.4%) 5456 (15.4%) 2350 (14.5%) 144989 (20.1%) U1 67060 (18.6%) - 26625 (15.9%) - 10987 (16.3%) - 8247 (15.8%) - 7144 (14.7%) - 6691 (17.8%) - 5657 (14.6%) - 5534 (17.1%) - 4306 (13.5%) - 2540 (18.4%) - 144791 (17.0%) - 47089 (18.8%) 23782 (15.8%) 11839 (16.6%) 7255 (16.2%) 6673 (15.2%) 7247 (18.3%) 5535 (14.7%) 5400 (17.6%) 4983 (14.1%) 2975 (18.3%) 122778 (17.0%) U2 58542 (16.2%) - 27027 (16.1%) - 10976 (16.3%) - 8048 (15.4%) - 7442 (15.3%) - 5862 (15.6%) - 6062 (15.6%) - 5202 (16.1%) - 4964 (15.6%) - 2306 (16.7%) - 136431 (16.0%) - 41383 (16.5%) 24586 (16.3%) 11809 (16.6%) 7160 (16.0%) 7043 (16.0%) 6555 (16.6%) 5999 (15.9%) 5056 (16.4%) 5641 (15.9%) 2806 (17.3%) 118038 (16.4%) U3 57967 (16.0%) - 27628 (16.5%) - 10665 (15.8%) - 8076 (15.5%) - 8077 (16.6%) - 6116 (16.3%) - 6409 (16.5%) - 5192 (16.1%) - 5745 (18.1%) - 2347 (17.0%) - 138222 (16.2%) - 40622 (16.2%) 25111 (16.7%) 11607 (16.3%) 7179 (16.0%) 7486 (17.1%) 6568 (16.6%) 6301 (16.7%) 5090 (16.6%) 6362 (18.0%) 2796 (17.2%) 119122 (16.5%) U4 52828 (14.6%) - 27931 (16.6%) - 10526 (15.6%) - 7832 (15.0%) - 8361 (17.2%) - 5329 (14.2%) - 6730 (17.4%) - 5194 (16.1%) - 6003 (18.9%) - 2254 (16.3%) - 132988 (15.6%) - 37452 (14.9%) 25006 (16.6%) 10980 (15.4%) 6990 (15.6%) 7777 (17.7%) 5952 (15.0%) 6644 (17.6%) 4974 (16.2%) 6580 (18.6%) 2724 (16.8%) 115079 (16.0%) U5 42956 (11.9%) - 25050 (14.9%) - 8855 (13.2%) - 7775 (14.9%) - 8236 (17.0%) - 4686 (12.5%) - 6378 (16.5%) - 4805 (14.9%) - 5894 (18.5%) - 2248 (16.3%) - 116883 (13.7%) - 30178 (12.0%) 22140 (14.7%) 9253 (13.0%) 6880 (15.4%) 7432 (16.9%) 5164 (13.0%) 6282 (16.6%) 4572 (14.9%) 6379 (18.0%) 2593 (16.0%) 100873 (14.0%) Page 5 of 13 Table 1 Variables characterizing each cluster in the study at baseline and at the end of the study (N = 916,619) (Continued) Variables Cluster 1 - Non- Cluster 2 - Eye Cluster 3 - Minority Cluster 4 - Cluster 5 - Cardio- Cluster 6 - Cluster 7 - Cluster 8 - Cluster 9 - Cluster 10 - Overall (patients Specific Impairment and Metabolic Cardio- Circulatory, Mental, Neurological, Respiratory Digestive Neurological, Multisystem alive) Mental Autoimmune- Circulatory and Respiratory and Digestive and and Ear Musculoskeletal Pattern Inflammatory Renal Genitourinary Circulatory and minor N2012=384787 N2012=177087 N2012=72190 (7.9%) - N2012=60192 N2012=54324 N2012=42422 N2012=41644 N2012=35692 N2012=33283 N2012=14998 N2012=916619 - (42.0%) - N2016= (19.3%) - N2016= N2016=73237 (9.8%) (6.6%) - N2016= (5.9%) - N2016= (4.6%) - N2016= (4.5%) - N2016= (3.9%) - N2016= (3.6%) - N2016= (1.6%) - N2016= N2016= 743827 258700 (34.8%) 155008 (20.8%) 46244 (6.2%) 45463 (6.1%) 41120 (5.5%) 38961 (5.2%) 31744 (4.3%) 36522 (4.9%) 16828 (2.3%) Villén et al. BMC Geriatrics N. Chronic 4.0 [2.0;5.0] - 6.0 [5.0;8.0] - 8.0 7.0 [5.0;8.0] - 9.0 [7.0;11.0] - 8.0 [7.0;10.0] - 7.0 [5.0;9.0] - 9.0 7.0 [5.0;9.0] - 7.0 [5.0;9.0] - 9.0 [7.0;10.0] - 11.0 [10.0;13.0] - 6.0 [4.0;8.0] - 7.0 diseases, 5.0 [3.0;6.0] [6.0;9.0] 8.0 [6.0;10.0] 10.0 [8.0;12.0] 10.0 [8.0;12.0] [6.0;11.0] 9.0 [7.0;11.0] 8.0 [6.0;11.0] 10.0 [9.0;12.0] 13.0 [11.0;15.0] [5.0;10.0] median [IQR]: N. Chronic diseases (n,%): 0 25380 (6.6%) - 0 (0.0%) - 0 0 (0.0%) - 0 (0.0%) - 0 0 (0.0%) - 0 (0.0%) 0 (0.0%) - 0 (0.0%) 0 (0.0%) - 0 (0.0%) - 0 (0.0%) - 0 (0.0%) - 25380 (2.8%) - (2020) 20:206 6584 (2.5%) (0.0%) 0 (0.0%) (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 6584 (0.9%) 1 36254 (9.4%) - 353 (0.2%) - 86 628 (0.9%) - 0 (0.0%) - 1 6 (0.0%) - 0 (0.0%) 296 (0.7%) - 100 231 (0.6%) - 386 (1.1%) - 0 (0.0%) - 0 0 (0.0%) - 0 38154 (4.2%) - 13650 (5.3%) (0.1%) 198 (0.3%) (0.0%) (0.2%) 76 (0.2%) 116 (0.4%) (0.0%) (0.0%) 14227 (1.9%) [ 2, 5) 201282 (52.3%) - 30742 (17.4%) - 13785 (19.1%) - 1809 (3.0%) - 1974 (3.6%) - 6032 (14.2%) - 5751 (13.8%) - 6751 (18.9%) - 698 (2.1%) - 12 (0.1%) - 1 268836 (29.3%) - 108605 (42.0%) 11516 (7.4%) 7640 (10.4%) 549 (1.2%) 506 (1.1%) 3276 (8.0%) 2428 (6.2%) 3105 (9.8%) 173 (0.5%) (0.0%) 137799 (18.5%) [ 5,10) 120673 (31.4%) - 132238 (74.7%) - 47338 (65.6%) - 32584 (54.1%) - 34571 (63.6%) - 25615 (60.4%) - 25450 (61.1%) - 21261 (59.6%) - 20365 (61.2%) - 3614 (24.1%) - 463709 (50.6%) - 126309 (48.8%) 111530 (72.0%) 44773 (61.1%) 17095 (37.0%) 19151 (42.1%) 21911 (53.3%) 20204 (51.9%) 17305 (54.5%) 13811 (37.8%) 1583 (9.4%) 393672 (52.9%) ≥10 1198 (0.3%) - 13754 (7.8%) - 10439 (14.5%) - 25799 (42.9%) - 17773 (32.7%) - 10479 (24.7%) - 10212 (24.5%) - 7294 (20.4%) - 12220 (36.7%) - 11372 (75.8%) - 120540 (13.2%) - 3552 (1.4%) 31876 (20.6%) 20626 (28.2%) 28599 (61.8%) 25806 (56.8%) 15833 (38.5%) 16253 (41.7%) 11218 (35.3%) 22538 (61.7%) 15244 (90.6%) 191545 (25.8%) N. of drugs, 3.0 [1.0;5.0] - 6.0 [3.0;8.0] - 5.0 [3.0;8.0] - 8.0 [6.0;11.0] - 8.0 [5.0;10.0] - 7.0 [4.0;10.0] - 7.0 [4.0;10.0] - 5.0 [3.0;8.0] - 6.0 [4.0;9.0] - 8.0 [5.0;11.0] - 5.0 [2.0;8.0] - 5.0 median [IQR]: 3.0 [1.0;5.0] 6.0 [4.0;8.0] 5.0 [3.0;8.0] 8.0 [6.0;11.0] 6.0 [8.0;10.0] 7.0 [4.0;9.0] 7.0 [4.0;10.0] 5.0 [3.0;8.0] 6.0 [4.0;9.0] 8.0 [5.0;10.0] [3.0;8.0] N. of drugs (n,%): 0 82698 (21.5%) - 9688 (5.5%) - 5408 (7.5%) - 2540 (4.2%) - 2186 (4.0%) - 2815 (6.6%) - 2366 (5.7%) - 3597 (10.1%) - 1473 (4.4%) - 597 (4.0%) - 113368 (12.4%) - 41989 (16.2%) 5355 (3.5%) 3759 (5.1%) 761 (1.6%) 800 (1.8%) 1463 (3.6%) 1293 (3.3%) 2199 (6.9%) 1068 (2.9%) 265 (1.6%) 58952 (7.9%) 1 39683 (10.3%) - 6639 (3.7%) - 3467 (4.8%) - 505 (0.8%) - 353 929 (1.7%) - 1161 (2.7%) - 1246 (3.0%) - 2083 (5.8%) - 1159 (3.5%) - 210 (1.4%) - 57082 (6.2%) - 28971 (11.2%) 5728 (3.7%) 3626 (5.0%) (0.8%) 624 (1.4%) 1257 (3.1%) 1222 (3.1%) 1822 (5.7%) 1305 (3.6%) 283 (1.7%) 45191 (6.1%) [ 2, 5) 139450 (36.2%) - 50256 (28.4%) - 20200 (28.0%) - 5388 (9.0%) - 7380 (13.6%) - 7802 (18.4%) - 8616 (20.7%) - 9676 (27.1%) - 7912 (23.8%) - 1987 (13.2%) - 258667 (28.2%) - 103534 (40.0%) 44740 (28.9%) 21352 (29.2%) 4602 (10.0%) 5880 (12.9%) 8612 (20.9%) 8192 (21.0%) 9006 (28.4%) 8811 (24.1%) 2570 (15.3%) 217299 (29.2%) [ 5,10) 108886 (28.3%) - 88542 (50.0%) - 32682 (45.3%) - 28606 (47.5%) - 26833 (49.4%) - 19941 (47.0%) - 18623 (44.7%) - 14879 (41.7%) - 16599 (49.9%) - 7090 (47.3%) - 362681 (39.6%) - 76263 (29.5%) 79758 (51.5%) 34143 (46.6%) 23462 (50.7%) 23058 (50.7%) 20083 (48.8%) 18442 (47.3%) 13799 (43.5%) 18510 (50.7%) 8278 (49.2%) 315796 (42.5%) ≥10 14070 (3.7%) - 21962 (12.4%) - 10433 (14.5%) - 23153 (38.5%) - 16996 (31.3%) - 10703 (25.2%) - 10793 (25.9%) - 5457 (15.3%) - 6140 (18.4%) - 5114 (34.1%) - 124821 (13.6%) - 7943 (3.1%) 19427 (12.5%) 10357 (14.1%) 17066 (36.9%) 15101 (33.2%) 9705 (23.6%) 9812 (25.2%) 4918 (15.5%) 6828 (18.7%) 5432 (32.3%) 106589 (14.3%) N. of visits, 7.0 [3.0;12.0] - 10.0 [6.0;16.0] - 10.0 [6.0;17.0] - 20.0 [11.0;32.0] - 14.0 [9.0;23.0] - 12.0 [7.0;20.0] - 12.0 [7.0;20.0] - 11.0 [6.0;19.0] - 12.0 [8.0;19.0] - 15.0 [9.0;23.0] - 9.0 [5.0;16.0] - median [IQR]: 7.0 [3.0;12.0] 10.0 [6.0;16.0] 10.0 [6.0;18.0] 22.0 [11.0;34.0] 15.0 [9.0;25.0] 12.0 [7.0;21.0] 12.0 [7.0;20.0] 11.0 [6.0;19.0] 12.0 [8.0;19.0] 15.0 [9.0;24.0] 10.0 [5.0;17.0] N. of visits (n,%): 0 36975 (9.6%) - 3722 (2.1%) - 1839 (2.5%) - 863 (1.4%) - 390 763 (1.4%) - 1147 (2.7%) - 779 (1.9%) - 1205 (3.4%) - 390 (1.2%) - 262 (1.7%) - 47945 (5.2%) - 17397 (6.7%) 2380 (1.5%) 1335 (1.8%) (0.8%) 395 (0.9%) 780 (1.9%) 544 (1.4%) 749 (2.4%) 311 (0.9%) 131 (0.8%) 24412 (3.3%) Page 6 of 13 Table 1 Variables characterizing each cluster in the study at baseline and at the end of the study (N = 916,619) (Continued) Variables Cluster 1 - Non- Cluster 2 - Eye Cluster 3 - Minority Cluster 4 - Cluster 5 - Cardio- Cluster 6 - Cluster 7 - Cluster 8 - Cluster 9 - Cluster 10 - Overall (patients Specific Impairment and Metabolic Cardio- Circulatory, Mental, Neurological, Respiratory Digestive Neurological, Multisystem alive) Mental Autoimmune- Circulatory and Respiratory and Digestive and and Ear Musculoskeletal Pattern Inflammatory Renal Genitourinary Circulatory and minor N2012=384787 N2012=177087 N2012=72190 (7.9%) - N2012=60192 N2012=54324 N2012=42422 N2012=41644 N2012=35692 N2012=33283 N2012=14998 N2012=916619 - (42.0%) - N2016= (19.3%) - N2016= N2016=73237 (9.8%) (6.6%) - N2016= (5.9%) - N2016= (4.6%) - N2016= (4.5%) - N2016= (3.9%) - N2016= (3.6%) - N2016= (1.6%) - N2016= N2016= 743827 258700 (34.8%) 155008 (20.8%) 46244 (6.2%) 45463 (6.1%) 41120 (5.5%) 38961 (5.2%) 31744 (4.3%) 36522 (4.9%) 16828 (2.3%) Villén et al. BMC Geriatrics 1 23812 (6.2%) - 3649 (2.1%) - 1830 (2.5%) - 649 (1.1%) - 498 629 (1.2%) - 1051 (2.5%) - 749 (1.8%) - 1043 (2.9%) - 315 (0.9%) - 157 (1.0%) - 33884 (3.7%) - 14629 (5.7%) 3624 (2.3%) 1922 (2.6%) (1.1%) 568 (1.2%) 1013 (2.5%) 672 (1.7%) 866 (2.7%) 417 (1.1%) 187 (1.1%) 24396 (3.3%) [ 2, 5) 77936 (20.3%) - 20693 (11.7%) - 9076 (12.6%) - 3064 (5.1%) - 3739 (6.9%) - 4606 (10.9%) - 3793 (9.1%) - 4310 (12.1%) - 2386 (7.2%) - 836 (5.6%) - 130439 (14.2%) - 54464 (21.1%) 18928 (12.2%) 9669 (13.2%) 2534 (5.5%) 3050 (6.7%) 4510 (11.0%) 3558 (9.1%) 3938 (12.4%) 2866 (7.8%) 1104 (6.6%) 104621 (14.1%) [ 5, 10) 117379 (30.5%) - 52731 (29.8%) - 20119 (27.9%) - 8287 (13.8%) - 10988 (20.2%) - 9698 (22.9%) - 10028 (24.1%) - 9029 (25.3%) - 8278 (24.9%) - 2812 (18.7%) - 249349 (27.2%) - (2020) 20:206 82678 (32.0%) 46906 (30.3%) 20721 (28.3%) 5948 (12.9%) 9136 (20.1%) 9891 (24.1%) 9391 (24.1%) 8329 (26.2%) 8991 (24.6%) 3234 (19.2%) 205225 (27.6%) ≥10 128685 (33.4%) - 96292 (54.4%) - 39326 (54.5%) - 47329 (78.6%) - 38205 (70.3%) - 25920 (61.1%) - 26295 (63.1%) - 20105 (56.3%) - 21914 (65.8%) - 10931 (72.9%) - 455002 (49.7%) - 89532 (34.6%) 83170 (53.7%) 39590 (54.1%) 36874 (79.7%) 32314 (71.1%) 24926 (60.6%) 24796 (63.6%) 17862 (56.3%) 23937 (65.5%) 12172 (72.3%) 385173 (51.8%) Page 7 of 13 Villén et al. BMC Geriatrics (2020) 20:206 Page 8 of 13 had expanded to include more patients (Supplementary disease, heart failure, valvular heart disease, cardiac File 1). arrhythmias, diabetes, kidney failure and anaemia (Fig. 2). Ten MP were identified at baseline (see characteristics of Cluster5 (C5)-Cardio-Circulatory, Mental, Respiratory each MP in Table 1). Except for Cluster 1 (C1) and Cluster and Genitourinary: At baseline, people included in this 4 (C4), the number of people included in all clusters in- pattern had 20 different drug groups overrepresented: creased during the study period. Most people remain in the peripheral vasodilators, bronchodilators, drugs for same cluster during follow up. For all MP, the most com- ischemic cardiomyopathy, heart failure, prostate mon shift is dying. The probability of dying varies signifi- diseases and diabetes. cantly depending on which MP the patient was classified. Cluster6 (C6)-Neurological, Digestive and Circulatory: Overrepresentation in this cluster is mostly constituted Characteristics of medication use in relation to by patients with Parkinson’s disease and corresponding multimorbidity patterns treatment, mainly dopamine, dopamine derivatives and Overall, we observed that overrepresented (O/E-ratio > other medicines used in neurological disorders. These 2) drugs (measured as mean packages) in each MP were diseases and related medications are practically associated with the overrepresented diseases in that exclusive to this cluster (more than 90.0% at the same MP. Naturally, these drugs are intended to treat beginning of the study). specific chronic diseases overrepresented in their corre- Cluster7 (C7)-Respiratory and Ear: Drugs included in sponding MP (Supplementary File 2). For example, in: this cluster were glucocorticoids, adrenergics in combination with corticosteroids or other drugs Cluster2 (C2)-Eye Impairment and Mental: The most excluding anticholinergics, selective beta-2- used drugs at baseline were ophthalmologic beta adrenoreceptor agonists, anticholinergics, corticoste- blockers. It should be noted that more than half of the roids, and antihistamines for systemic use. general population with glaucoma and under treatment Cluster8 (C8)-Digestive: The most prescribed with ophthalmic beta blockers are included in this medication corresponds to drugs used for the cluster (Exclusivity 50.59 and 48.75% at baseline, treatment of chronic liver diseases. respectively). Cluster9 (C9)-Neurological, Musculoskeletal and Cluster3 (C3)-Minority Metabolic Autoimmune- Minor: In this group the most overrepresented drugs Inflammatory: At baseline, the most overrepresented are anti-inflammatory and analgesics, in agreement drugs were glucocorticoids, uric acid lowering agents, with the MP. and Vitamin D and analogues. Cluster10 (C10)-Multisystem: The least common Cluster4 (C4)-Cardio-Circulatory and Renal: At baseline, pattern, it has multiple overrepresented disorders from people included in this pattern had 17 different drug groups different systems, and the corresponding medications overrepresented, mainly to treat ischemic cardiovascular used to treat these disorders. Fig. 2 Cardio-Circulatory and Renal patterns in relation with drug prescription Villén et al. BMC Geriatrics (2020) 20:206 Page 9 of 13 As an example, Fig. 3 represents the evolution dur- Considering C1-Non-Specific as a reference, an in- ing follow up of the O/E-ratio and mean number of crease in abnormal kidney function was observed in the medication packages for the ten most used drugs in different MP over the study period. The highest propor- C4-Cardio-Circulatory and Renal and C5-Cardio-Cir- tion of abnormal kidney function was observed in the culatory, Mental, Respiratory and Genitourinary. In C4-Cardio-Circulatory and Renal (OR 2.19; CI 95% these two MP, both the O/E-ratio and mean number 2.15–2.23) and C3-Minority Metabolic Autoimmune- of medication packages increase or remain stable over Inflammatory MP (OR 2.16; CI 95% 2.12–2.20) (Table 2 time. and Supplementary file 3). The MP with a higher risk of Fig. 3 Prevalence and O/E ratio evolution over time for diseases and drugs in Cluster 4 - Cardio Circulatory and Renal and Cluster 5 - Cardio Circulatory, Mental, Respiratory and Genitourinary Villén et al. BMC Geriatrics (2020) 20:206 Page 10 of 13 Table 2 Logistic regression models for Kidney and Liver function by cluster. abnormal liver function were C8-Digestive (OR 3.39; CI In C3-Minority Metabolic Autoimmune-Inflammatory, 95% 3.30–3.49), followed by C4-Cardio-Circulatory and over 50% of diseases are exclusive of this cluster. How- Renal (OR 1.96; CI 95% 1.91–2.02) (Table 2 and Supple- ever, even if the medication is associated with overrepre- mentary File 4). sented conditions, it does not have high exclusivity since these medicines have different authorised indications, for instance glucocorticoids and Vitamin D and ana- Discussion logues, which are even indicated off-label [34, 35]. Key results The MP C4-Cardio-Circulatory and Renal and C5- This study informs on the use of medication by the eld- Cardio-Circulatory, Mental, Respiratory and Genitouri- erly population during 5 years of follow up, according to nary, illustrate the patients with multimorbidity who are the ten most common MP. Predictably, the most over- prescribed polypharmacy to treat overrepresented dis- represented drugs in each MP coincide with the most eases. In addition, some diseases as diabetes that are not overrepresented disorders in that same MP. Also, the overrepresented might need various medicines to control medicines most prescribed in the study population re- the condition. Moreover, recent intensification of treat- main unchanged throughout the follow up period. The ments has translated into an increase in drug prescrip- analysis of polypharmacy based on specific MP and their tion [36, 37]. association with abnormal liver and kidney function has In C9-Neurological, Musculoskeletal and Minor, meta- revealed that the patients included in the various MP mizole has replaced traditional nonsteroidal anti- present high rates of abnormal liver and kidney function inflammatory drugs (NSAIDs), probably due to emerging when compared to the MP C1-Non-Specific, underlining safety concerns of NSAIDs regarding the kidney [38], the need for new safety criteria in these patients [32, 33]. cardiovascular [39, 40], and digestive systems [39]. Polypharmacy and their evolution Abnormal kidney and liver failure function In C2-Eye Impairment and Mental, the medicines most This study shows that most patients with multimorbidity frequently prescribed correspond to the ocular disorders receiving polypharmacy present abnormal kidney and liver diagnosed. However, in other groups of diseases of this function. While the design of this study cannot explain MP, for instance the Neurotic, stress-related and somato- the causes of the abnormal kidney and liver function ob- form diseases, which correspond mainly to cigarette con- served, we hypothesise that they probably originate from sumption, the pharmacological treatment is not subsidised the baseline disorders of the patients, adverse effects from by the public health system and consequently does not ap- the medication, inappropriate prescribing and lack of ad- pear in the Pharmacy Invoice Registry. justment to the kidney and liver function of each patient, Villén et al. BMC Geriatrics (2020) 20:206 Page 11 of 13 which are usually affected by old age [11–13, 41–43]. This primary care EHR, which includes a large proportion of study, based on medication packages dispensed in phar- the population with multimorbidity and with polyphar- macies, underlines the importance of stringent monitoring macy [23]. Furthermore, we have used a classification of prescriptions, particularly in patients that might have for chronic diseases previously validated by a clinically abnormal kidney and liver function. We recommend the driven methodology, which allows a homogeneous as- use of prompts in the EHR to adjust medication dosage in sessment of chronic diseases and polypharmacy in a accordance with liver function and eGFR. manageable number of categories, and also the uniform Abnormal kidney function is highest in patients from evaluation of chronicity in the European Union [24]. MP C4 and C3. In MP C4 -Cardio-Circulatory and This study also presents some limitations. Firstly, we Renal, in addition to the baseline kidney and cardiovascular only considered the medications for which at least three disorders of these patients, abnormal kidney function has packages during each year of the study period had been been attributed to the hemodynamic effects of diuretics dispensed. While this could underestimate some medi- causing nephrotoxicity [41]. The MP C3-Minority Meta- cines, it is extremely unusual to dispense less than 3 bolic Autoimmune-Inflammatory includes patients with packages per year of drugs treating chronic diseases. hypothyroidism [44] and overrepresentation of allopurinol, Similarly, we excluded the medication for acute condi- both causes of abnormal kidney function. tions, some of which can cause temporary abnormalities Abnormal liver function is highest in patients from in kidney and liver function. Secondly, the SIDIAP only clusters C8 and C4. The risk is highest in patients in C8 collects information on drugs prescribed by primary care - Digestive, since this pattern represents patients with and hospital physicians which are dispensed by commu- liver disease. C4- Cardio-Circulatory and Renal has the nity pharmacies. Consequently, our analysis cannot rule oldest patients and the highest prescription of vitamin K out active ingredients included in hospital drug regimes, antagonists, which can cause cholestasis [42]. over-the-counter medicines and para-pharmacy prod- ucts, which might also impair kidney and liver function. Comparison with the literature Thirdly and finally, censoring of people who died and Most studies on MP use a cross-sectional design. Some people who transferred might have influenced estima- publications include longitudinal data, but to our know- tions of risk associations in a competing-risk scenario, ledge no data on the association of MP and abnormal which was not an objective of this study. kidney and liver function have been published [45].. In Our research shows that while prescription of poly- the literature, European articles underscore medication pharmacy might be justified with regard to clinical for depression and chronic obstructive pulmonary dis- guidelines, it increases the risk of adverse drug reactions ease (COPD) in the elderly [18], which we included in and might negatively affect kidney and liver function. clusters C9 and C5, respectively. In contrast, Japanese Crucially, the risk of overtreatment, which occurs when authors observe the highest risk of polypharmacy in ma- the prescribed medications have no clinically significant lignant, digestive and urologic patterns [17]. While we benefit and when the risk of adverse effects associated excluded drugs for the treatment of malignancies in our with an additional medication outweighs the overall bene- study, we did not observe overrepresentation of drugs fit of the treatment, is high in polypharmacy [46, 47]. for the gastrointestinal system, since they are the type of medication most consumed in the general population, nor for urological diseases, which are highly prevalent in Conclusions the population over 65 years. Interestingly, in the Japa- Polypharmacy is highly prevalent in older adults. The nese study only 25% of patients were over 65 years of most frequently prescribed medicines were related to age. Ultimately, if we analysed the drugs overrepresented MP and their consumption was maintained throughout in specific patterns such as the cardiovascular (C4 and the follow-up period. C5), these medicines would practically replicate poly- Finally, this study provides real-world data on the evolu- pharmacy patterns described in other publications [18], tion of polypharmacy and multimorbidity among older i.e., drugs for cardiovascular diseases, for diabetes and adults and describes the association between polypharmacy for gout. However, our medication patterns included and abnormal liver and kidney function in this population. also the treatments for other diseases in MP C4 and C5, for instance anaemia, pain, glaucoma, COPD and benign Supplementary information prostatic hyperplasia. Supplementary information accompanies this paper at https://doi.org/10. 1186/s12877-020-01580-1. Strengths and limitations One of the major strengths of this study is the use of a Additional file 1. Prevalence of Top 30 Drugs during the follow up period in people over 65 years in Catalonia. large, high-quality database that originates from the Villén et al. BMC Geriatrics (2020) 20:206 Page 12 of 13 Additional file 2. Multimorbidity patterns with Hidden Markov models Competing interests at baseline (year 2012) and final year (2016) of the study. Drug use All authors have completed the ICMJE uniform disclosure form at www. icmje.org/coi_disclosure.pdf and declare: no support from any organisation Additional file 3. Evolution of kidney function during follow up by for the submitted work; no financial relationships with any organisations that clusters. might have an interest in the submitted work in the previous 3 years; no Additional file 4. Evolution of liver function during follow up by other relationships or activities that could have influenced the submitted clusters. work. Author details Abbreviations 1 Àrea del Medicament i Servei de Farmàcia, Gerencia Territorial de Barcelona, ALP: Alkaline phosphatase; ALT: Alanine transaminase; ATC: Anatomical Institut Català de la Salut, Barcelona, Spain. 2Fundació Institut Universitari per therapeutic chemical classification system; C1: Cluster 1; C2: Cluster 2; a la recerca a l’Atenció Primària de Salut Jordi Gol i Gurina (IDIAPJGol), C3: Cluster 3; C4: Cluster 4; C5: Cluster 5; C6: Cluster 6; C7: Cluster7; Barcelona, Spain. 3Universitat Autònoma de Barcelona, Cerdanyola del Vallès, C8: Cluster 8; C9: Cluster 9; C10: Cluster 10; CHI: Catalan health institute; Bellaterra, Spain. 4Department of Psychiatry, Vic University Hospital, Vic, Spain. CI: Confidence interval; COPD: Chronic obstructive pulmonary disease; 5 Àrea de Serveis Assistencials. Servei Català de la Salut, Barcelona, Spain. eGFR: Estimate glomerular filtration rate; EHR: Electronic health records; GGT: Gamma-glutamyl transpeptidase; HMGCoA-reductase : 3-Hidroxi-3- Received: 11 December 2019 Accepted: 7 May 2020 metil-glutaril-CoA reductase; HMM: Hidden markov model; ICD- 10: International classification of diseases, version 10; IDMS10: Creatinine calibrated to isotope dilution mass spectrometry; IQR: Interquartile range; MEDEA: Mortality in Spanish small areas and socioeconomic and References environmental inequalities; MDRD-4: Four-variable modification of diet in renal disease; MP: Multimorbidity pattern; NSAIDs: Nonsteroidal anti- 1. Valderas JM, Starfi B, Sibbald B. Understanding health and health services. inflammatory drugs; O/E-ratios: Observed/expected ratios; OR: Odds ratio; Ann Fam Med. 2009;7(4):357–63. PCAmix: Principal component analysis and multiple correspondence analysis; 2. Beard JR, Officer A, de Carvalho IA, Sadana R, Pot AM, Michel J-P, et al. The PHCCs : Primary health care centres; PPI: Proton pump inhibitors; world report on ageing and health: a policy framework for healthyageing. SD: Standard deviation; SIDIAP: Information system for research in primary Lancet. 2016;387(10033):2145–54. care 3. Salisbury C, Johnson L, Purdy S, Valderas JM, Montgomery A. Epidemiology and impact of multimorbidity in primary care: a retrospective cohort study. Acknowledgements Br J Gen Pract. 2011;61(582):e12–21. We thank the Catalan Institute of Health and the SIDIAP, which provided the 4. Marengoni A, Angleman S, Melis R, Mangialasche F, Karp A, Garmen A, et al. database for the study. The authors are also grateful to Carmen Ibáñez for Aging with multimorbidity: a systematic review of the literature. Ageing Res the administrative work. Rev. 2011;10(4):430–9. 5. Violan C, Foguet-Boreu Q, Flores-Mateo G, Salisbury C, Blom J, Freitag M, Authors’ contributions et al. Prevalence, determinants and patterns of multimorbidity in primary All authors contributed to the design of the study, revised the article and care: a systematic review of observational studies. PLoS One. 2014;9(7):3–11. approved the final version. CV, ARL, MGC, EA and SFB obtained the funding. 6. Guthrie B, Payne K, Alderson P, McMurdo MET, Mercer SW. Adapting clinical CV, MGC, QFB, ARL, AT, MPV and NVV drafted the article. NV, AT, MG, AR, CV guidelines to take account of multimorbidity. BMJ. 2012;345:e6341. and SFB proposed the methodology, programmed the tests, and 7. Masnoon N, Shakib S, Kalisch-Ellett L, Caughey GE. What is polypharmacy? contributed to the analysis and interpretation of data. NV, CV, MGC, and MCB systematic review of definitions. BMC Geriatr. 2017;17:230. wrote the first draft of the manuscript. NV, AT, ARL and SFB wrote the first 8. Tommelein E, Mehuys E, Petrovic M, Somers A, Colin P, Boussery K. draft of the supplementary file, and all authors contributed ideas, interpreted Potentially inappropriate prescribing in community-dwelling older people the findings and reviewed rough drafts of the manuscript. across Europe: a systematic literature review. Eur J Clin Pharmacol. 2015; 71(12):1415–27. Funding 9. Maher RL, Hanlon J, Hajjar ER. Clinical consequences of polypharmacy in This work was supported by a research grant from the the Instituto de Salud elderly. Expert Opin Drug Saf. 2014;13(1):57–65. Carlos III, under the auspices of the Ministry of Science, Innovation and 10. Hajjar ER, Cafiero AC, Hanlon JT. Polypharmacy in elderly patients. Am J Universities, awarded in the 2016 call under the Health Strategy Action Geriatr Pharmacother. 2007;5(4):345–51. 2013–2016, within the National Research Program oriented to Societal 11. Ren R, Deng L, Xue Y, Suzuki K, Zhang W, Yu Y, et al. Visualization of aging- Challenges, within the Technical, Scientific and Innovation Research National associated chromatin alterations with an engineered TALE system. Cell Res. Plan 2013–2016 ‘[grant number PI16/00639]’, co-funded with European 2017;27(4):483–504. Union ERDF funds (European Regional Development Fund) and the 12. Corsonello A, Pedone C, Incalzi R. Age-related pharmacokinetic and Department of Health of the Catalan Government, in the 2017 call that Pharmacodynamic changes and related risk of adverse drug reactions. Curr awarded subsidies for the Strategic Plan for Research in Health (Pla Estratègic Med Chem. 2010;17(6):571–84. de Recerca i Innovació en Salut, PERIS) 2016–2020, modality research oriented 13. Poulose N, Raju R. Aging and injury: alterations in cellular energetics and to Primary care ‘[grant number SLT002/16/00058]’ and by the Catalan organ function. Aging Dis. 2014;5(2):101–8. Government ‘[grant number AGAUR 2017 SGR 578]. 14. Rankin A, Cadogan CA, Patterson SM, Kerse N, Cardwell CR, Bradley MC, et al. Interventions to improve the appropriate use of polypharmacy for Availability of data and materials older people. Cochrane Database Syst Rev. 2018;9:CD008165. The datasets are not available, as researchers have signed an agreement with 15. Prados-Torres A, Calderón-Larrañaga A, Hancco-Saavedra J, Poblador-Plou B, the Information System for the Development of Research in Primary Care Van Den Akker M. Multimorbidity patterns: a systematic review. J Clin (SIDIAP) concerning confidentiality and security of the dataset, which forbids Epidemiol. 2014;67(3):254–66. providing data to third parties. The SIDIAP is subject to periodic audits. 16. Guisado-Clavero M, Violán C, López-Jimenez T, Roso-Llorach A, Pons-Vigués M, Muñoz MA, et al. Medication patterns in older adults with multimorbidity: a Ethics approval and consent to participate cluster analysis of primary care patients. BMC Fam Pract. 2019;20(1):1–12. The protocol of the study was approved by the Clinical Research Ethics 17. Aoki T, Yamamoto Y, Ikenoue T, Onishi Y, Fukuhara S. Multimorbidity Committee, Fundació Institut Universitari per a la recerca a l’Atenció Primària patterns in relation to polypharmacy and dosage frequency: a nationwide, de Salut Jordi Gol i Gurina (IDIAPJGol) (P16/151). All data were anonymized cross-sectional study in a Japanese population. Sci Rep. 2018;8(1):3806. in agreement with national and international law. 18. Calderón-Larrañaga A, Gimeno-Feliu LA, González-Rubio F, Poblador-Plou B, Lairla-San José M, Abad-Díez JM, et al. Polypharmacy patterns: Unravelling Consent for publication systematic associations between prescribed medications. PLoS One. 2013; Not Applicable. 8(12):e84967. Villén et al. BMC Geriatrics (2020) 20:206 Page 13 of 13 19. Kostev K, Jacob L. Multimorbidity and polypharmacy among elderly people 41. Bosch JA, Calbé JJ, Calero F, Díaz JM, Diego L, Esmatjes E, et al. Pautes per a followed in general practices in Germany. Eur J Intern Med. 2018;55:66–8. l’harmonització del tractament farmacològic de la malalƽa renal crònica. 20. Violán C, Fernández-Bertolín S, Guisado-Clavero M, Foguet-Boreu Q, Valderas Barcelona: Agència de Qualitat i Avaluació Sanitàries de Catalunya (AQuAS). JM, Vidal J, et al. Longitudinal multimorbidity patterns in elderly population Departament de Salut. Generalitat de Catalunya; 2016. using hidden Markov models: a cohort study of Mediterranean elderly 42. Morales ML, Vélez LN, Muñoz M. OG. Hepatotoxicidad: Patrón colestásico population. Sci Rep. inducido por fármacos. Rev Colomb Gastroenterol. 2016;31(1):36–47. 21. Statistical Institute of Catalonia. Idescat. Statistical Yearbook of Catalonia 43. Tejada F. Hepatotoxicidad por farmacos. Rev Clínica Med Fam. 2010;3(3): [Internet]. 2016 [cited 2019 Dec 9]. Available from: http://www.idescat.cat/ 177–91. pub/?id=aec&lang=en. 44. Iglesias P, Díez JJ. Thyroid dysfunction and kidney disease. Eur J Endocrinol. 22. Memòria d’activitats 2012. Institut Català de la Salut [Internet]. 2012 [cited 2009;160(4):503–15. 2019 Nov 18]. Available from: http://ics.gencat.cat/ca/detall/publicacio/ 45. Morin L, Johnell K, Laroche ML, Fastbom J, Wastesson JW. The memoria_2012-00007. epidemiology of polypharmacy in older adults: register-based prospective 23. García-Gil M, Hermosilla E, Prieto-Alhambra D, Fina F, Rosell M, Ramos R, cohort study. Clin Epidemiol. 2018;10:289–98. et al. Construction and validation of a scoring system for the selection of 46. Brownlee S, Chalkidou K, Doust J, Elshaug AG, Glasziou P, Heath I, et al. high-quality data in a Spanish population primary care database (SIDIAP). Evidence for overuse of medical services around the world. Lancet. 2017; Inform Prim Care. 2011;19(3):135–45. 390:156–68. 24. Calderón-Larrañaga A, Vetrano DL, Onder G, Gimeno-Feliu LA, Coscollar- 47. Benetos A, Rossignol P, Cherubini A, Joly L, Grodzicki T, Rajkumar C, et al. Santaliestra C, Carfí A, et al. Assessing and measuring chronic Polypharmacy in the aging patient: Management of Hypertension in multimorbidity in the older population: a proposal for its operationalization. octogenarians. J Am Med Assoc. 2015;314(2):170–80. J Gerontol - Ser A Biol Sci Med Sci. 2017;72(10):1417–23. 25. WHO Collaborating Centre for Drug Statistics Methodology. Guidelines for ATC classification and DDD assignment 2013. 16th ed. Oslo: 2013. https:// Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in www.whocc.no/filearchive/publications/1_2013guidelines.pdf. published maps and institutional affiliations. 26. Elorza-Ricart JM, Tovillas-Morán FJ, Oliveras-Puig A, Galcerán JM, Fina F, Dalfó-Baqué A. Estudio transversal comparativo de las fórmulas CKD-EPI y MDRD-4 a partir de la historia clínica informatizada de Atención Primaria de Barcelona. Hipertens y Riesgo Vasc. 2012;29(4):118–29. 27. Forns J, Cainzos-Achirica M, Hellfritzsch M, Morros R, Poblador-Plou B, Hallas J, et al. Validity of ICD-9 and ICD-10 codes used to identify acute liver injury: a study in three European data sources. Pharmacoepidemiol Drug Saf. 2019; 28(7):965–75. 28. Domínguez-Berjón MF, Borrell C, Cano-Serral G, Esnaola S, Nolasco A, Pasarín MI, et al. Construcción de un índice de privación a partir de datos censales en grandes ciudades españolas (Proyecto MEDEA). Gac Sanit. 2008; 22(3):179–87. 29. Bezdek JC. Pattern recognition with fuzzy objective function algorithms. 1st ed. New York. Springer: Plenum Press; 1981. 30. Rabinier LR. A tutorial on hidden Markov models and selected applications in speech recognition. Proc IEEE. 1989;77(2):257–86. 31. White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med. 2011;30(4):377–99. 32. Duerden M, Avery T, Payne R. Polypharmacy and medicines optimisation making it safe and sound [internet]. London: The King’s Fund; 2013. Available from: https://www.kingsfund.org.uk/sites/default/files/field/field_publication_ file/polypharmacy-and-medicines-optimisation-kingsfund-nov13.pdf. 33. Cooper JA, Cadogan CA, Patterson SM, Kerse N, Bradley MC, Ryan C, et al. Interventions to improve the appropriate use of polypharmacy in older people: a Cochrane systematic review. BMJ Open. 2015;5:e009235. 34. Morgan DJ, Dhruva SS, Coon ER, Wright SM, Korenstein D. 2018 update on medical overuse. JAMA Intern Med. 2019;179(2):240–6. 35. Woodford HJ, Barrett S, Pattman S. Vitamin D: too much testing and treating? Clin Med (Northfield Il). 2018;18(3):196–200. 36. Mata-Cases M, Franch-Nadal J, Real J, Mauricio D. Glycaemic control and antidiabetic treatment trends in primary care centres in patients with type 2 diabetes mellitus during 2007-2013 in Catalonia: a population-based study. BMJ Open. 2016;6(10):4–11. 37. Lipska KJ, Yao X, Herrin J, McCoy RG, Ross JS, Steinman MA, et al. Trends in drug utilization, glycemic control, and rates of severe hypoglycemia, 2006- 2013. Diabetes Care. 2017;40(4):468–75. 38. Dreischulte T, Morales DR, Bell S, Guthrie B. Combined use of nonsteroidal anti-inflammatory drugs with diuretics and/or renin-angiotensin system inhibitors in the community increases the risk of acute kidney injury. Kidney Int. 2015;88(2):396–403. 39. Baigent C, Bhala N, Emberson J, Merhi A, Abramson S, Arber N, et al. Vascular and upper gastrointestinal effects of non-steroidal anti- inflammatory drugs: meta-analyses of individual participant data from randomised trials. Lancet. 2013;382:769–79. 40. Arfè A, Scotti L, Varas-Lorenzo C, Nicotra F, Zambon A, Kollhorst B, et al. Non-steroidal anti-inflammatory drugs and risk of heart failure in four European countries: nested case-control study. BMJ. 2016;354:i4857.