Guisado-Clavero et al. BMC Family Practice (2019) 20:82 https://doi.org/10.1186/s12875-019-0969-9 RESEARCH ARTICLE Open Access Medication patterns in older adults with multimorbidity: a cluster analysis of primary care patients Marina Guisado-Clavero1,2,3, Concepción Violán2,3* , Tomàs López-Jimenez2,3, Albert Roso-Llorach2,3, Mariona Pons-Vigués2,3,4, Miguel Angel Muñoz1,2,3,5 and Quintí Foguet-Boreu2,3,6,7 Abstract Background: Older adults suffer from various chronic conditions which make them particularly vulnerable. The proper management of multiple drug use is therefore crucial. The aim of our study was to describe drug prescription and medication patterns in this population. Methods: A cross-sectional study in Barcelona (Spain) using electronic health records from 50 primary healthcare centres. Participants were aged 65 to 94 years, presenting multimorbidity (≥2 chronic diseases), and had been prescribed at least 1 drug for 6 months or longer during 2009. We calculated the prevalence of prescribed drugs and identified medication patterns using multiple correspondence analysis and k-means clustering. Analyses were stratified by sex and age (65–79, 80–94 years). Results: We studied 164,513 patients (66.8% women) prescribed a median of 4 drugs (interquartile range [IQR] = 3–7) in the 65–79 age-group and 6 drugs (IQR = 4–8) in the 80–94 age-group. A minimum of 45.9% of patients aged 65–79 years, and 61.8% of those aged 80–94 years, were prescribed 5 or more drugs. We identified 6 medication patterns, a non-specific one and 5 encompassing 8 anatomical groups (alimentary tract and metabolism, blood, cardiovascular, dermatological, musculo-skeletal, neurological, respiratory, and sensory organ). Conclusions: Drug prescription is widespread among the elderly. Six medication patterns were identified, 5 of which were related to one or more anatomical group, with associations among drugs from different systems. Overall, guidelines do not accurately reflect the situation of the elderly multimorbid, new strategies for managing multiple drug uses are needed to optimize prescribing in these patients. Keywords: Ageing, Cluster analysis, Drugs, Electronic health records, Multimorbidity, Primary health care Introduction are more likely to require multiple drugs to achieve optimal Worldwide, individuals are living longer [1] thanks to ad- clinical (or disease) management [7, 8], indeed, a prescrip- vances in medical research and care [2]. For instance, in tion rate of over 80% for ≥5 drugs has been reported [9]. 2016, 19% of the European population was aged 65 years or Multiple drug use in older adults, however, is associated older [3], a figure that is expected to reach 30% by 2060 [4]. with overall worsening physical and psychological health as Nevertheless, a longer life span is closely related to the like- a result of age-related changes in pharmacokinetics and lihood of developing chronic disease [5] and 55–98% of pharmacodynamics [10]. In addition, it has a potential in- older adults suffer from multimorbidity [6]. Such patients fluence on aspects of safety, including inappropriate pre- scription, adverse drug reaction, risk of medication * Correspondence: cviolan@idiapjgol.org; http://www.idiapjgol.org/ 2 interaction (drug-drug or drug-disease interaction), and ad- Institut Universitari d’Investigació en Atenció Primària Jordi Gol (IDIAP Jordi Gol), Gran Via Corts Catalanes, 587 àtic, 08007 Barcelona, Spain herence [11, 12]. 3 Universitat Autònoma de Barcelona, Plaça Cívica, 08193 Bellaterra, Barcelona, Due to ageing vulnerability, multiple drug use in the Spain multimorbid elderly is a main issue of concern for the Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. 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. Guisado-Clavero et al. BMC Family Practice (2019) 20:82 Page 2 of 12 public health system. Identifying which drugs are being (PHCC) run by the public Catalan Health Institute [25, taken is crucial to define patients at risk. As a result, tools 26]. The study was performed in Barcelona (Spain) in need to be developed with the aim of decreasing prescrip- 2009 with information from 50 PHCC. The participants tion errors, drugs interactions, adverse drug reactions, and were aged 65 to 94 years, and the inclusion criteria were other consequences such as falls, hospitalization, and mor- a) to have attended a PHCC at least once during 2009; tality associated with multiple drug use [13, 14]. A recent b) to present multimorbidity, defined as the coexistence systematic review described clinical management oriented of 2 or more chronic diseases [27]; and c) to have been to multimorbidity and polymedication. Its recommenda- prescribed at least 1 drug for a period of 6 months or tions, however, were focused on the risks/benefits of each longer during 2009 (see flow chart in Fig. 1). drug individually rather than collectively [15]. To date, the The study protocol was approved by the Research Eth- limited information available in the literature is mostly de- ics Committee at IDIAPJGol (Protocol no: P15/149). All scriptive [16] and methods regarding pharmaco- data were anonymized, and the confidentiality of the epidemiology in multimorbidity have yet to be established. EHR was maintained at all times in accordance with na- Prescription groups and patterns could be of help in the tional and international law. As all data were anon- analysis of multiple drug use to create new strategies in the ymized, no consent to individuals were required. management of complexity among multimorbid patients. New techniques are being developed to create homoge- Variables neous patterns regarding the management of prescribed Prescription drugs were the main unit of measurement drugs. For instance, exploratory factor analysis (EFA) and were coded as 1 (present) or 0 (absent). Drugs in which is based on correlations between variables or fac- the SIDIAP database are classified using the Anatomical tors, and cluster analysis (CA), a technique for grouping a Therapeutic Chemical (ATC) system (Additional file 1), set of individuals in such a way that they are more similar a measuring unit recommended by the World Health to each other than those in other groups [17]. EFA has re- Organization for drug studies. To classify the drugs in cently been reported to be useful for describing correl- this study, and facilitate subsequent analysis and inter- ation between variables, while CA carries out an in-depth pretation, we used the 4th level of the ATC system examination of the pattern for non-random associations which corresponds to chemical subgroups. Proton pump between the determinant variables of an individual [18]. In inhibitors, for example, are coded as A02BC [28]. recent years, EFA has been employed to define a number The other variables recorded for each participant were: of multimorbidity patterns [19–21], and some medication number of chronic diseases coded with the International ones [22]. Nonetheless, the statistical technique employed Classification of Primary Care second edition and se- should be taken into account. EFA correlates specific vari- lected using the O’Halloran criteria [29], age (65–79 ables (e.g. diseases), but not all the variables of one unit years vs 80–94 years), and sex (male vs female). Accord- (e.g. patient), whilst CA could be helpful as the main start- ing to the chronic diseases selected, chronic medication ing point to look for dissimilarities. Irrespective of the was defined as the prescription of a drug for at least 6 methodology employed in these studies [23], there are continuous months during the period of study. Medica- common biological systems encompassing multimorbidity tion which did not fulfil this criterion was not analysed patterns: cardio-metabolic conditions, musculoskeletal as it was considered acute or not long-term. Neither diseases, and mental health problems [24]. Serious dis- were supplements included as they are not financed by eases and those with a greater prevalence according to the Spanish health system. EFA/CA should thus be represented with the correspond- ing medication. We hypothesized that prescribed drugs could be grouped Statistical analysis using CA to identify clusters of patients with similar drugs Data were extracted from the SIDIAP database after and consequently create medication patterns. The objective authorization of the study [25]. All the authors had ac- of this study was to describe prescribed drugs and identify cess to the database. There were no missing values, as medication patterns in multimorbid older adults. sex, age, chronic diseases, and drugs were recorded for all the sample. Methods Descriptive statistics were employed to summarize the Design, setting, and inclusion criteria overall data. Categorical variables were expressed as fre- We conducted a cross-sectional analysis of electronic quencies (percentage) and continuous variables as means health records (EHR) from the Information System for (standard deviation [SD]) or medians (interquartile range Research in Primary Care (SIDIAP). This is a centralized [IQR]). Prevalence of prescription drugs was calculated database that contains EHR from 2006 for all the pa- and medication patterns identified through 2 steps: 1) tients who have attended primary health care centres multiple correspondence analysis (MCA), and 2) k- Guisado-Clavero et al. BMC Family Practice (2019) 20:82 Page 3 of 12 Fig. 1 Flow chart means clustering. All analyses were stratified by sex and obtained for each cluster. The optimal number of clus- age. ters (k), which is the solution with the highest Calinski- Harabaz index value, was assessed using criteria with Multiple correspondence analysis 100 iterations. To assess internal cluster quality, cluster MCA is a data analysis technique used to detect and rep- stability of the optimal solution was computed using Jac- resent underlying structures in sets of nominal categorical card bootstrap values with 100 runs [17]. Highly stable data. It identifies groups with similar characteristics and clusters should yield average Jaccard similarities of 0.85 shows, in a multidimensional space, relationships between and above. dichotomous or categorical variables (in our case drug prescriptions) that would be difficult to observe in a con- Medication patterns tingency table [30, 31]. MCA also allows individuals to be To describe the medication patterns across the clusters, directly represented as points (coordinates) in a geometric we used three criteria: a) the prevalence of prescribed space through the transformation of original binary data drugs in each cluster; b) the observed/expected (O/E) ra- to continuous ones. The MCA was based on the indicator tios obtained by dividing the prevalence of a particular matrix. The optimal number of dimensions extracted and drug in each cluster by the prevalence of the same pre- percentages of inertia were determined by means of a scribed drug in the age and sex groups, considering scree plot. over-represented drugs when value ≥2; and c) exclusiv- ity, defined as the proportion of individuals with a par- K-means clustering ticular prescribed drug included in the cluster over the Using the geometric space created in the MCA, patients total number of individuals with a particular prescribed were classified into clusters according to proximity cri- drug in the corresponding age and sex group, consider- teria by means of the k-means algorithm, and centers ing high exclusivity when value ≥50%. Guisado-Clavero et al. BMC Family Practice (2019) 20:82 Page 4 of 12 Medication patterns were defined by considering drugs At least 45.9% of the 65–79 year and 61.8% of the 80– with a prevalence ≥20% or an O/E ratio ≥ 2. To identify the 94 year groups were prescribed 5 or more drugs. As ex- importance of each medication and, as a consequence, the pected, the use of 10 or more drugs was almost twice in amount of medication included in a cluster, we employed the 80–94 compared to the 65–79 year age group. The exclusivity. In order to facilitate the designation of a medi- number of prescribed drugs and chronic diseases did not cation pattern we named the patterns considering medica- differ between sexes (Table 1). The 10 most widely pre- tions belonging to the same ATC group with an exclusivity scribed drugs across the sample belonged to 3 ATC sys- value ≥50%, even when presenting a low prevalence. And tem groups: alimentary tract and metabolism (A), we also took into consideration to name the pattern those nervous system (N), and cardiovascular system (C). Pro- drugs over-represented by O/E ratio ≥ 2. We then described ton pump inhibitors and HMG CoA reductase inhibitors medications included in each cluster using three numbers were present in the top 3 most prescribed drugs in all of characteristics: prevalent drugs (prevalence ≥20%), drugs groups, with platelet aggregation inhibitors (excluding over-represented (O/E ratio ≥ 2) and exclusive drugs (exclu- heparin) in men and benzodiazepine derivatives (65–79 sivity ≥50%). But we considered only exclusive and over- years) and anilides (80–94 years) for women (Table 2). represented drugs to label the pattern. In addition to mathematical validation, clinical criteria Characteristics of medication patterns based on previous literature [32–34] and clinical feed- Six medication patterns for each age and sex group were back from the research team (3 family physicians and 2 identified. All the groups had a non-specific pattern con- epidemiologists) were employed to evaluate the sisting of highly prevalent drugs that were neither over- consistency and significance of the final cluster solution. represented nor exclusive. The other 5 patterns were The analyses were carried out using SPSS for Win- made up of drugs belonging to 1 or more anatomical dows, version 24 (SPSS Inc., Chicago, IL, USA) and R groups corresponding to: alimentary tract and metabol- version 3.4.2 (R Foundation for Statistical Computing, ism (A), blood and blood forming organs (B), cardiovascu- Vienna, Austria). lar system (C), dermatological (D), musculoskeletal system (M), nervous system (N), respiratory system (R), and sen- Results sory organs (S) (Table 3, Additional files 2, 3 and 4). The sample was composed of 164,513 patients aged ≥65 As an example, findings for women 65–79 years are rep- years all of whom presented multimorbidity and had at resented in Table 3. Six medication patterns were identified, least 1 drug prescribed; 66.8% were women. The group numbered according to the weight of the sample implied 65–79 years had a mean age of 72.0 years (SD = 4.3) and (descending order): non-specific (cluster 1), followed by ner- was prescribed a median of 4 (IQR = 3–7) drugs. The vous system (cluster 2), musculo-skeletal + dermatological group 80–94 years had a mean age of 84.1 years (SD = (cluster 3), alimentary tract and metabolism (cluster 4), re- 3.4) and was prescribed a median of 6 (IQR: 4–8) drugs. spiratory system (cluster 5), and cardiovascular system Table 1 Descriptive data, by sex and age groups, of the multimorbid patients (n = 164,513) aged 65–94 years attended in 2009 at primary healthcare centres located in Barcelona Women Men 65–79 years 80–94 years 65–79 years 80–94 years Participants n (%) 78,008 (47.4) 31,848 (19.4) 41,931 (25.5) 12,726 (7.7) Number drugs n (%) 1–4 40,931 (52.5) 11,374 (35.7) 22,703 (54.1) 4868 (38.3) 5–9 31,500 (40.4) 16,460 (51.7) 16,339 (39.0) 6268 (49.3) ≥10 5577 (7.1) 4014 (12.6) 2889 (6.9) 1590 (12.5) Median number of drugs (IQRa) 4 (3–7) 6 (4–8) 4 (2–6) 5 (3–8) Number of chronic diseases n (%) 2 2806 (3.6) 1125 (3.5) 1792 (4.3) 410 (3.2) [3–5] 20,301 (26.0) 7689 (24.1) 12,484 (29.8) 3090 (24.3) [6–9] 33,089 (42.4) 13,495 (42.4) 17,955 (42.8) 5562 (43.7) ≥10 21,812 (28.0) 9539 (30.0) 9700 (23.1) 3664 (28.8) Median number of chronic diseases (IQRa) 7 (5–10) 8 (5–10) 7 (5–9) 7 (5–10) IQRa Interquartile range Guisado-Clavero et al. BMC Family Practice (2019) 20:82 Page 5 of 12 Table 2 The ten most commonly prescribed drugs in 2009 for multimorbid patients (n = 164,513) aged 65–94 years, by sex and age groups, attended at primary healthcare centres located in Barcelona Women Men a ATC code Drug name N % ATC codea Drug name N % 65–79 years A02BC Proton pump inhibitors 32,634 41.8 C10AA HMG CoA reductase inhibitors 18,188 43.4 C10AA HMG CoA reductase inhibitors 32,004 41.0 A02BC Proton pump inhibitors 15,170 36.2 N05BA Benzodiazepine derivatives 20,649 26.5 B01AC Platelet aggregation inhibitors excl. Heparin 13,872 33.1 N02BE Anilides 18,434 23.6 C09AA ACE inhibitors, plain 9748 23.2 B01AC Platelet aggregation 15,338 19.7 G04CA Alpha-adrenoreceptor antagonists 7235 17.3 inhibitors excl. Heparin C09AA ACE inhibitors, plain 13,578 17.4 A10BA Biguanides 6306 15.0 M05BA Bisphosphonates 13,309 17.1 N05BA Benzodiazepine derivatives 6019 14.4 N06AB Selective serotonin reuptake inhibitors 11,522 14.8 C08CA Dihydropyridine derivatives 5996 14.3 C03AA Thiazides, plain 10,112 13.0 C07AB Beta blocking 5934 14.2 agents, selective C09CA Angiotensin II antagonists, plain 9231 11.8 N02BE Anilides 5625 13.4 80–94 years A02BC Proton pump inhibitors 16,496 51.8 A02BC Proton pump inhibitors 5877 46.2 N02BE Anilides 11,370 35.7 B01AC Platelet aggregation inhibitors excl. Heparin 5641 44.3 C10AA HMG CoA reductase inhibitors 11,222 35.2 C10AA HMG CoA reductase inhibitors 4657 36.6 B01AC Platelet aggregation 10,512 33.0 C09AA ACE inhibitors, plain 3235 25.4 inhibitors excl. Heparin N05BA Benzodiazepine derivatives 9633 30.2 N02BE Anilides 2638 20.7 C09AA ACE inhibitors, plain 7223 22.7 G04CA Alpha-adrenoreceptor antagonists 2601 20.4 C08CA Dihydropyridine derivatives 5283 16.6 N05BA Benzodiazepine derivatives 2313 18.2 C03CA Sulfonamides, plain 5265 16.5 C08CA Dihydropyridine derivatives 2260 17.8 N06AB Selective serotonin reuptake inhibitors 5258 16.5 C03CA Sulfonamides, plain 1930 15.2 C09CA Angiotensin II antagonists, plain 4502 14.1 C01DA Organic nitrates 1622 12.7 a Code : chemical subgroup, 4rt level, ATC code (Anatomical Therapeutic Chemical classification) from the World Health Organization (Additional file 1) For more details, visit webside: https://www.whocc.no/atc/structure_and_principles/ (cluster 6). For each cluster, three subgroups of prescribed It was observed that the non-specific pattern had the drugs that encompassed the pattern were defined. Three greatest number of patients for all groups and was de- kinds of data were shown for every cluster. Using the ex- fined by drugs that were neither prevalent nor over- ample of the musculo-skeletal and dermatological pattern represented. With respect to the non-specific pattern, the (cluster 3), we identified three different groups of drugs in number of patients aged 65–79 years was higher than the pattern: those aged 80–94 years for both sexes. According to the a) drugs with a high prevalence but not over- repre- frequency of patients, the next patterns were: for women sented such as proton pump inhibitors (prevalence 66%, 65–79 years “nervous system” and “musculo-skeletal + O/E ratio 1.58, exclusivity 19%) and benzodiazepine de- dermatological”, whilst for women 80–94 years they in- rivatives (prevalence 33%, O/E ratio 1.26, exclusivity cluded alimentary tract and metabolism as a drug group 15%); implied in frequency; for men 65–79 years they were b) drugs with a high/low prevalence over-represented “cardiovascular system” and “alimentary tract and me- with exclusivity < 50% such as anilides (prevalence 61%, tabolism”, and for those 80–94 years was added the drug O/E ratio 2.57, exclusivity 31%) and other opioids (preva- group related to musculo-skeletal and nervous system lence 10%, O/E ratio 3.25, exclusivity 40%); (Table 3, Additional files 2, 3 and 4). c) drugs with a high/low prevalence over-represented Comparing patterns between sexes, women had four and with exclusivity ≥ 50% such as anti-inflammatory patterns in both age groups which implied only one preparations, non-steroids for topical use (prevalence over-represented anatomical system (alimentary tract 33%, O/E ratio 5.96, exclusivity 70%) and potent cortico- and metabolism, cardiovascular system, nervous sys- steroids (group III) (prevalence 9%, O/E ratio 6.65, exclu- tem, and respiratory system), in contrast to men who sivity 81%) (Table 3). had only two patterns implying one anatomical Guisado-Clavero et al. BMC Family Practice (2019) 20:82 Page 6 of 12 Table 3 Example of medication patterns across women 65–79 years attended in primary health centres in Barcelona during 2009 (N = 78,008) Code& Drugs Preb O/E Exclus. ratioa Cluster 1 n = 39,202 (50%) Non-specific pattern C10AA HMG CoA reductase inhibitors 32% 0.78 39% A02BC Proton pump inhibitors 21% 0.51 26% Cluster 2 n = 14,604 (19%) Nervous system pattern A02BC Proton pump inhibitors 70% 1.68 31% N05BA Benzodiazepine derivatives 58% 2.18 41% C10AA HMG CoA reductase inhibitors 56% 1.36 25% N06AB Selective serotonin reuptake inhibitors 40% 2.71 51% B01AC Platelet aggregation inhibitors excl. Heparin 35% 1.76 33% M05BA Bisphosphonates 28% 1.62 30% N02BE Anilides 26% 1.10 21% N06AX Other antidepressants 14% 3.49 65% N03AX Other antiepileptics 10% 3.16 59% N05CD Benzodiazepine derivatives 9% 2.03 38% A12AA Calcium 7% 2.58 48% N06AA Non-selective monoamine reuptake inhibitors 7% 3.07 57% C01DA Organic nitrates 7% 2.32 43% N02AX Other opioids 6% 2.03 38% A11CC Vitamin D and analogues 6% 2.97 56% A06AD Osmotically acting laxatives 6% 2.04 38% N03AE Benzodiazepine derivatives 4% 3.69 69% C07AA Beta blocking agents, non-selective 4% 2.61 49% H02AB Glucocorticoids 4% 2.71 51% C10AX Other lipid modifying agents 4% 2.25 42% N06DX Other anti-dementia drugs 3% 2.44 46% A03FA Propulsive 3% 2.25 42% Cluster 3 n = 9502 (12%) “Musculo-skeletal system” and A02BC Proton pump inhibitors 66% 1.58 19% “Dermatologicals” pattern N02BE Anilides 61% 2.57 31% N05BA Benzodiazepine derivatives 33% 1.26 15% M02AA Antiinflammatory preparations, non-steroids for tropical use 31% 5.96 73% C10AA HMG CoA reductase inhibitors 30% 0.74 9% M01AE Propionic acid derivatives 27% 4.30 52% C05CA Bioflavonoids 19% 2.88 35% M01AX Other antiinflammatory and antirheumatic agents, non-steroids 17% 2.84 35% A02AD Combinations and complexes of aluminium, calcium and magnesium 15% 4.24 52% compounds M01AB Acetic acid derivatives and related substances 15% 5.17 63% D01AC Imidazole and triazole derivatives 10% 5.93 72% N02AX Other opioids 10% 3.25 40% D07AC Corticosteroidas, potent (group III) 9% 6.65 81% N02BB Pyrazolones 6% 5.30 65% Guisado-Clavero et al. BMC Family Practice (2019) 20:82 Page 7 of 12 Table 3 Example of medication patterns across women 65–79 years attended in primary health centres in Barcelona during 2009 (N = 78,008) (Continued) Code& Drugs Preb O/E Exclus. ratioa A06AD Osmotically acting laxatives 6% 2.13 26% R05CB Mucolytics 5% 2.83 34% R06AX Other antihistamines for systemic use 5% 3.34 41% N07CA Antivertigo preparations 5% 2.26 28% Cluster 4 n = 8745 (11%) Alimentary tract and metabolism pattern C10AA HMG CoA reductase inhibitors 68% 1.67 19% A10BA Biguanides 65% 5.86 66% B01AC Platelet aggregation inhibitors excl. Heparin 61% 3.12 35% A02BC Proton pump inhibitors 50% 1.19 13% A10BB Sulfonylureas 37% 6.69 75% C09AA ACE inhibitors, plain 30% 1.70 19% C08CA Dihydropyridine derivatives 29% 2.68 30% C07AB Beta blocking agents, selective 23% 2.18 24% N02BE Anilides 22% 0.94 10% A10AC Insulins and analogues for injection, intermediate-acting 10% 6.34 71% C01DA Organic nitrates 9% 3.22 36% A10AE Insulins and analogues for injection, long-acting 9% 6.04 68% M04AA Preparations inhibiting uric acid production 9% 2.97 33% C02CA Alpha-adrenoreceptor antagonists 7% 3.57 40% C10AB Fibrates 7% 2.96 33% A10AD Insulins and analogues for injection, intermediate- or long- acting 7% 6.24 70% combined with fast- acting G04CA Alpha-adrenoreceptor antagonists 4% 2.13 24% C10AX Other lipid modifying agents 4% 2.36 26% Cluster 5 n = 3275 (4%) Respiratory system pattern R03AC Selective beta-2-adrenoreceptor agonists 72% 16.88 71% A02BC Proton pump inhibitors 54% 1.30 5% R03BB Anticholinergics 54% 18.86 79% R03AK Adrenergics in combination with corticosteroids or other drugs, excl. 51% 11.87 50% Anticholinergics R03BA Glucocorticoids 40% 18.45 77% C10AA HMG CoA reductase inhibitors 38% 0.94 4% N02BE Anilides 32% 1.36 6% N05BA Benzodiazepine derivatives 27% 1.01 4% B01AC Platelet aggregation inhibitors excl. Heparin 23% 1.17 5% C03CA Sulfonamides, plain 14% 2.11 9% R05CB Mucolytics 11% 6.34 27% R06AX Other antihistamines for systemic use 6% 3.79 16% C08DB Benzothiazepine derivatives 5% 2.26 9% H02AB Glucocorticoids 4% 3.07 13% G04CA Alpha-adrenoreceptor antagonists 4% 2.02 8% Guisado-Clavero et al. BMC Family Practice (2019) 20:82 Page 8 of 12 Table 3 Example of medication patterns across women 65–79 years attended in primary health centres in Barcelona during 2009 (N = 78,008) (Continued) Code& Drugs Preb O/E Exclus. ratioa Cluster 6 n = 2680 (3%) Cardiovascular system pattern B01AA Vitamin K antagonists 78% 15.74 54% C03CA Sulfonamides, plain 66% 10.05 35% A02BC Proton pump inhibitors 58% 1.37 5% C01AA Digitalis glycosides 53% 27.60 95% C10AA HMG CoA reductase inhibitors 43% 1.06 4% N02BE Anilides 32% 1.35 5% N05BA Benzodiazepine derivatives 29% 1.11 4% C09AA ACE inhibitors, plain 27% 1.56 5% C07AB Beta blocking agents, selective 24% 2.25 8% C09CA Angiotensin II antagonists, plain 23% 1.98 7% A12BA Potassium 20% 16.35 56% C03DA Aldosterone antagonists 19% 17.02 58% C08DB Benzothiazepine derivatives 13% 6.07 21% C07AG Alpha and beta blocking agents 12% 8.91 31% M04AA Preparations inhibiting uric acid production 11% 3.73 13% C01DA Organic nitrates 9% 3.26 11% B03AA Iron bivalent, oral preparations 5% 3.60 12% A10AE Insulins and analogues for injection, long-acting 3% 2.30 8% A02BA H2 - receptor antagonists 3% 2.12 7% &: Code Chemical subgroup, 4rt level, ATC code (Anatomical Therapeutic Chemical classification) from the World Health Organization O/E ratioa: Observed/expected ratio Preb: Prevalence Exclus.: Exclusivity Selected criteria: Prevalence ≥ 20 or Observed/Expected ratio ≥ 2 system (cardiovascular and respiratory system). The neurological drugs representing the most frequently pre- other patterns were formed by two or more anatom- scribed groups. We defined 6 medication patterns which ical systems. The rest of the results are detailed in provide information about the multiple drugs grouped Table 3 and Additional files 2, 3 and 4. closely together in elderly patients. The pattern with the Comparing patterns between age groups, no significant most participants, non-specific, had up to 39% of the differences were observed for women with the exception of age-sex sample included and was composed of drugs additional drugs encompassing the non-specific pattern corresponding to specific diseases (hypertension, lipid (anilides, ACE inhibitors, benzodiazepine derivatives) (Table disorder, depressive disorder (women)) and others re- 3, Additional file 2). The men’s patterns, however, appeared lated to the secondary prevention of cardiovascular/di- more complex: to the non-specific pattern were added two gestive diseases (platelet aggregation inhibitors and drugs (platelet aggregation inhibitors excluding heparin and proton pump inhibitors). The rest of the medication pat- proton pump inhibitors), and in the 80–94 age group the terns could be linked to the multimorbidity ones defined patterns encompassed multiple anatomical groups includ- in a previous article performed in the same sample [35]. ing a sensory organs pattern (Additional files 3 and 4). Comparison with published literature Discussion Ageing is associated with functional decline, and the In this study, we present data regarding prescription prescription of multiple drugs tends to be highest in the drugs in an urban population of elderly adults with mul- oldest segments of the population [36]. Just over half the timorbidity. Prescription rates were high, particularly in patients in our study had been prescribed 5 or more the older subset of patients, probably due to the greater drugs, rates of between 45.0 and 80.0% have been previ- burden of chronic disease. Proton pump inhibitors were ously described based on primary care EHR [9, 37]. the most widely prescribed drug with cardiovascular and These results showed that the 10 most prescribed drugs Guisado-Clavero et al. BMC Family Practice (2019) 20:82 Page 9 of 12 were to treat metabolic, cardiovascular, and nervous sys- anatomical system was possibly related to the burden of tem disorders, in line with other reports for the elderly chronic disease associated with age [23]. The observed [35, 38, 39]. As expected, considering that heart disease medication patterns should coincide with the multimor- is the leading cause of death in such populations [40], bidity ones given that the former reflect the various ill- cardiovascular drugs were the main group of prescribed nesses being treated. For instance, if we compare drugs. Looking more closely, proton pump inhibitors multimorbidity and medication patterns from the same were the most widely prescribed drug in our study, con- sample, the endocrine-metabolic multimorbidity pattern trasting with findings on the prevalence of digestive tract should be related to the alimentary tract and metabol- chronic diseases conducted in the same sample [35]. ism one [35]. A concept that concurs with a number of Off-label use of proton pump inhibitors could be related publications that have reported that medication data to the prevention of adverse gastrointestinal effects, as may represent a way of identifying chronic conditions reported elsewhere [41]. In addition, a high prevalence [47]. Following this idea, medication patterns could help of lipid modifying (C10AA) agents and antithrombotic characterise individuals with multimorbidity. Finally, the drugs (B01AC) was probably linked to their use in the use of three criteria to define patterns permitted a repre- primary and secondary prevention of thrombotic events. sentation of all drugs, including those related to low We would like to point out that benzodiazepines, despite prevalence diseases. Variability between chronic diseases their potentially adverse effects for older adults (e.g, and treatments was thus respected in our results. memory impairment, delirium, falls) [42, 43], were still To the best of our knowledge, only one study has pre- frequently prescribed in our population (from 14.4% in viously defined medication patterns using EFA [22], and men 65–79 years to 30.2% in women 80–94 years), with few authors have investigated such patterns in patients a reported prevalence among the elderly from 10.0 to with multimorbidity [16]. It is difficult to draw compari- 41.6% [44, 45]. sons because of differences in drug inclusion criteria, Six patterns per group defining user profiles with pre- number of drugs considered, and especially method- scribed drugs were obtained. We took into account pre- ology. Nevertheless, some anatomical systems, including scribed drugs, instead of consumed ones, because we cardiovascular, respiratory, and neurological ones were assumed patients followed what their doctors suggested. the same. Such similarities are probably related to the As we studied patients with multimorbidity, we consid- strong prevalence of chronic conditions. Nevertheless, ered chronic drugs rather than supplements or acute with CA we obtained 6 markedly different patterns, and prescriptions. As a result, many of the defined patterns with the O/E ratio and the exclusivity criteria we could seemed logical and in concordance with chronic disease define which drugs were over-represented, playing a prevalence [35]. In addition, differences in intra- and more crucial role. inter-patterns were represented defining prevalence, O/E A recent publication has established that guidelines ratio, and exclusivity for each drug. The relevance of the addressing polymedication appear arbitrary [15]. Our re- prescribed drug was thus represented by these three search thus contributes to the definition of medication parameters. patterns which could be used to identify both user pro- The non-specific pattern had the greatest number of files and safety issues (e.g. detecting prescription errors, patients in all strata as no anatomical group was over- for instance inappropriate drugs, or drug-drug associa- represented. It could, therefore, be hypothesized that pa- tions), something that is not possible with multimorbid- tients evolve to 5 specific patterns across time, that is to ity patterns. The definition of medication patterns could say, the non-specific pattern could represent a pre-state open new paths to create instruments to prioritize of a specific one. In addition, the fact that the number of groups of individuals and permit effective prescription. patients included in the non-specific pattern was lower In addition, establishing medication patterns in accord- in the 80–94 than the 65–79 year group points to the ance to multimorbidity patterns would help to deter- hypothesis that this pattern could be a pre-specific medi- mine prognostic factors in drug safety, define possible cation one. Nevertheless, longitudinal analyses should be adverse drug reactions, and identify drug-drug and drug- conducted to substantiate this issue. With respect to disease interactions. The analysis of medication patterns specific patterns, the men’s appeared more complex than thus provides an additional perspective for interpreting women’s possibly because of the anatomical systems in- and defining the population’s health. volved and male smoking habits [46]. In concordance with this difference, more men in the 65–79 year group Strength and weakness presented cardiovascular and respiratory patterns than Our study sample is both reliable and representative of the women who showed mostly neuromuscular drug-related population, thus adding robustness to our results. Moreover, patterns. Furthermore, the fact that the patterns of the we provide an accurate reflection of real prescribing habits older participants were made up of more than one for the elderly with multimorbidity in an urban public Guisado-Clavero et al. BMC Family Practice (2019) 20:82 Page 10 of 12 primary health care setting. Analyses of individual medica- Additional files tion patterns can lead to new insights into individual pre- scription situations. We consider that complexity among Additional file 1: The main groups of the Anatomical Therapeutic Chemical (ATC) system. (DOCX 12 kb) patients is well represented in these patterns. However, some Additional file 2: Medication patterns across women 80-94 years limitations should be considered. On one hand, selected cri- attended in primary health centres in Barcelona during 2009 (N = 31,848). teria of chronicity (prescription of 6 or more months) may Selected criteria: Prevalence ≥20 or Observed/Expected ratio ≥2. (DOCX have caused a selection bias, although we followed an estab- 24 kb) lished definition [23]. In addition, we have to assume that Additional file 3: Medication patterns across men 65-79 years attended in primary health centres in Barcelona during 2009 (N = 41,931). Selected CA is inherently exploratory in nature and different cluster- criteria: Prevalence ≥20 or Observed/Expected ratio ≥ 2. (DOCX 26 kb) ing algorithms may produce varying results. The lack of Additional file 4: Medication patterns across men 80-94 years attended studies defining medication patterns also limits comparisons in primary health centres in Barcelona during 2009 (N = 12,726). Selected between results and populations. Finally, we should consider criteria: Prevalence ≥20 or Observed/Expected ratio ≥ 2. (DOCX 23 kb) as a limitation the fact that the collected data were 10 years old and may not exactly reflect current prescription patterns. Abbreviations Nevertheless, these medication patterns correspond to a six- ATC: Anatomical Therapeutic Chemical; CA: Cluster analysis; EFA: Exploratory factor analysis; EHR: Electronic health records; Exclus: Exclusivity; year longitudinal multimorbidity study [35, 48] in which it IDIAPJGol: Institut Universitari d’Investigació en Atenció Primària Jordi Gol; was observed that multimorbidity patterns did not differ at IQR: Interquartile range; MCA: Multiple correspondence analysis; O/E all during the period studied. In addition, in public primary ratios: Observed/expected ratios; PHCC: Primary health care centres; Pre: Prevalence; SD: Standard deviation; SIDIAP: System for Research in health care, the implementation of new treatments for spe- Primary Care cific diseases (for example, oral anticoagulants or oral antidi- abetic medications) are not yet generalised. For this reason, Acknowledgements We thank the Catalan Health Institute, and especially the Information System we considered that the medication patterns represented for Research in Primary Care (SIDIAP) Unit which provided the study current prescription. database. Future research Authors’ contributions All authors contributed to the design of the study, revised the article, and Medication patterns could change with time as a conse- approved the final version. MGC, CV, QFB, ARL, TLJ, MAM, MPV, drafted the quence of multimorbidity evolution and new treatments study protocol and obtained the funding. MGC, CV, ARL, TLJ contributed to applied in some chronic diseases. Our study is cross- the analysis and interpretation of data. MGC, CV, QFB, TLJ; ARL, MAM, MPV wrote the first draft, and all authors contributed ideas, interpreted the sectional, but in future research it would be advantageous findings and reviewed rough drafts of the manuscript. All the authors read to analyse large prospective cohorts with different esti- and approved the final manuscript. mates to define medication patterns and identify their sta- Funding bility or evolution. In addition, generational differences This manuscript is part of a PhD being undertaken by MGC at the Public are expected due to modified lifestyle habits. Thus, re- Health Department (Departament de Pediatria, d’Obstetrícia i Ginecologia i analyses should be considered as medication patterns are de Medicina Preventiva) at the Universitat Autònoma de Barcelona. This work was supported by a pre-doctoral grant from Catalan Health Insti- expected to alter across decades. tute in Barcelona; by the Catalan Society of General Practitioners (CAMFiC) Taking into account drug prescription and medication and by SIDIAP grant to MGC in 2015; this latter organization allowed us to patterns, improvements in guidelines for the clinical man- explore their dataset to obtain the results. The funders had no role in the study design or data collection, analysis, and interpretation, writing of the agement of elderly patients should be contemplated. In manuscript, and decision to submit for publication. addition, the methodology used for clustering could be a starting point for analysing drug safety in relation to drug Availability of data and materials interaction. The data that support the findings of this study may be obtained from SIDIAP but restrictions could apply to those used under license. Upon reasonable request and with permission of SIDIAP they may be available Conclusions from the authors. This study provides information about prescription drugs in an urban population of older adults with multimorbid- Ethics approval and consent to participate The protocol of the study was approved by the Committee of the Ethics of ity. Our results showed highly elevated prescription rates, Clinical Research, Institut Universitari d’Investigació en Atenció Primària Jordi particularly in the older subset of patients, probably due Gol (IDIAP Jordi Gol) (Protocol No: P15/149). All data were anonymized, and to the greater burden of chronic disease. Clinical practice the confidentiality of EHR was respected at all times in accordance with national and international law. As all data were anonymized, no consent to should consider reviewing off-label prescribed drugs for individuals were required. possible de-prescription. The study of medication patterns provides a method for Consent for publication Not applicable. analysing the use of multiple drugs in elderly patients. We identified 6 medication patterns in our series which could Competing interests provide new avenues for evaluating multimorbidity. The authors declare that they have no competing interests. Guisado-Clavero et al. BMC Family Practice (2019) 20:82 Page 11 of 12 Author details multimorbidity patterns: a study of ‘real-world’ data. BMJ Open. 2018;8(3): 1 Gerència d’Àmbit d’Atenció Primària Barcelona Ciutat, Institut Català de la e018986. Salut, Carrer Balmes 22, Barcelona, Spain. 2Institut Universitari d’Investigació 19. Britt HC, Harrison CM, Miller GC, Knox SA. Prevalence and patterns of en Atenció Primària Jordi Gol (IDIAP Jordi Gol), Gran Via Corts Catalanes, 587 multimorbidity in Australia. Med J Aust. 2008;189(2):72–7. àtic, 08007 Barcelona, Spain. 3Universitat Autònoma de Barcelona, Plaça 20. Steinman MA, Lee SJ, John Boscardin W, Miao Y, Fung KZ, Moore KL, et al. Patterns Cívica, 08193 Bellaterra, Barcelona, Spain. 4Facultat d’infermeria, Universitat de of multimorbidity in Eldery veterans. J Am Geriatr Soc. 2012;29(10):997–1003. Girona, Emili Grahit, 77, 17071 Girona, Spain. 5Unitat de Suport a la Recerca 21. Foguet-Boreu Q, Violán C, Rodriguez-Blanco T, Roso-Llorach A, Pons-Vigués de Barcelona, Institut Universitari d’Investigació en Atenció Primària Jordi Gol M, Pujol-Ribera E, et al. Multimorbidity patterns in elderly primary health (IDIAP Jordi Gol), Carrer Sardenya, 375, 08025 Barcelona, Spain. 6Departament care patients in a South Mediterranean European region: a cluster analysis. de psiquiatria, Hospital Universitari de Vic, Francesc Pla el Vigatà, 1, 08500 Vic, PLoS One. 2015;10(11):1–14. Barcelona, Spain. 7Facultat de Ciències de la Salut i Benestar, Universitat de 22. Calderón-Larrañaga A, Gimeno-Feliu LA, González-Rubio F, Poblador-Plou B, Lairla- Vic – Universitat Central de Catalunya, Sagrada Família, 7, 08500 Vic, Spain. san José M, Abad-Díez JM, et al. Polypharmacy patterns: unravelling systematic associations between prescribed medications. PLoS One. 2013;8(12):1–10. Received: 30 October 2018 Accepted: 23 May 2019 23. Violan C, Foguet-Boreu Q, Flores-Mateo G, Salisbury C, Blom J, Freitag M, et al. Prevalence, determinants and patterns of multimorbidity in primary care: a systematic review of observational studies. PLoS One. 2014;9(7):3–11. 24. Prados-Torres A, Calderón-Larrañaga A, Hancco-Saavedra J, Poblador-Plou B, References Van Den Akker M. Multimorbidity patterns: a systematic review. J Clin 1. Murray CJL, Barber RM, Foreman KJ, Ozgoren AA, Abd-Allah F, Abera SF, et Epidemiol. 2014;67(3):254–66. al. Global, regional, and national disability-adjusted life years (DALYs) for 306 25. García-Gil MDM, Hermosilla E, Prieto-Alhambra D, Fina F, Rosell M, Ramos R, diseases and injuries and healthy life expectancy (HALE) for 188 countries, et al. Construction and validation of a scoring system for the selection of 1990–2013: quantifying the epidemiological transition. Lancet. 2015:1990– high-quality data in a Spanish population primary care database (SIDIAP). 2013. Inform Prim Care. 2011 Jan;19(3):135–45. 2. Oeppen J, Vaupel JW. Demography. Broken Limits to Life Expectancy. 26. Sistema d'informació per al desenvolupament de la Investigació en Atenció Science (80- ). 2002;296(May):1029–31. Primària.Inicio. http://www.sidiap.org/index.php/es. Accessed 6 Feb 2018. 3. Eurostat. Population structure and ageing - statistics explained. 2016. http:// 27. Valderas JM, Starfield B, Sibbald B, Salisbury C, Roland M. Defining ec.europa.eu/eurostat/statistics-explained/index.php/Population_structure_ comorbidity: implications for understanding health and health services. Ann and_ageing. Accessed 23 Aug 2017. Fam Med. 2009 Jan;7(4):357–63. 4. Eurostat. Active ageing - employment, social affairs & inclusion - European 28. Christian Berg Ms, Hege Salvesen Blix Ms, Irene Litleskare Ms, Marit Rønning Commission. 2015. http://ec.europa.eu/social/main.jsp?catId=1062. Accessed Ms, Solveig Sakshaug Ms, Hanne Strøm Ms, et al. Guidelines for ATC 23 Aug 2017. classification and DDD assignment 2013. 16th editi. Oslo; 2013. 1-284 p. 5. Jacob L, Breuer J, Kostev K. Prevalence of chronic diseases among older https://www.whocc.no/filearchive/publications/1_2013guidelines.pdf. patients in German general practices. GMS Ger Med Sci. 2016;14:1–7. Accessed 25 Jul 2017. 6. Marengoni A, Angleman S, Melis R, Mangialasche F, Karp A, Garmen A, et al. 29. O’Halloran J, Miller GC, Britt H. Defining chronic conditions for primary care Aging with multimorbidity: a systematic review of the literature. Ageing Res with ICPC-2. Fam Pract. 2004 Aug;21(4):381–6. Rev. 2011 Mar 23;10(4):430–9. 30. Sourial N, Wolfson C, Zhu B, Quail J, Fletcher J, Karunananthan S, et al. 7. Van den AM, Buntinx F. Multimorbidity in general practice: prevalence, Correspondence analysis is a useful tool to uncover the relationships incidence, and determinants of co-occurring chronic and recurrent diseases. among categorical variables. J Clin Epidemiol. 2010 Jun;63(6):638–46. J Clin Epidemiol. 1998;51(5):367–75. 31. García-Gil M, Blanch J, Comas-Cufí M, Daunis-i-Estadella J, Bolíbar B, Martí R, 8. Wehling M. Multimorbidity and polypharmacy: how to reduce the harmful et al. Patterns of statin use and cholesterol goal attainment in a high-risk drug load and yet add needed drugs in the elderly? Proposal of a new cardiovascular population: a retrospective study of primary care electronic drug classification: fit for the aged: letters to the editor. J Am Geriatr Soc. medical records. J Clin Lipidol. 2016 Jan;10(1):134–42. 2009 Mar;57(3):560–1. 32. Evidence-based clinical decision support at the point of care | UpToDate. 9. Freund J, Meiman J, Kraus C. Using electronic medical record data to https://www.uptodate.com/home. Accessed 23 Jul 2017. characterize the level of medication use by age-groups in a network of 33. Cochrane Library. https://www.cochranelibrary.com/search. Accessed 21 primary care clinics. J Prim Care Community Heal. 2013;4(4):286–93. Jul 2017. 10. Mangoni AA, Jackson SHD. Age-related changes in pharmacokinetics and 34. Guies de pràctica clínica. Institut Català de la Salut. http://ics.gencat.cat/ca/ pharmacodynamics: basic principles and practical applications. Br J Clin assistencia/coneixement-assistencial/guies-de-practica-clinica/. Accessed 10 Pharmacol. 2003;57(1):6–14. Sept 2017. 11. Hajjar ER, Cafiero AC, Hanlon JT. Polypharmacy in elderly patients. Am J 35. Guisado-Clavero M, Roso-Llorach A, López-Jimenez T, Pons-Vigués M, Geriatr Pharmacother. 2007 Dec;5(4):345–51. Foguet-Boreu Q, Muñoz MA, et al. Multimorbidity patterns in the 12. Salazar JA, Poon I, Nair M. Clinical consequences of polypharmacy in elderly: elderly: a prospective cohort study with cluster analysis. BMC Geriatr. expect the unexpected, think the unthinkable. Expert Opin Drug Saf. 2007 2018;18(1):16. Nov 30;6(6):695–704. 36. Payne RA, Avery AJ, Duerden M, Saunders CL, Simpson CR, Abel GA. 13. Gómez C, Vega-Quiroga S, Bermejo-Pareja F, Medrano MJ, Louis ED, Benito- Prevalence of polypharmacy in a Scottish primary care population. Eur J Clin León J. Polypharmacy in the elderly: a marker of increased risk of mortality Pharmacol. 2014;70(5):575–81. in a population-based prospective study (NEDICES). Gerontology. 2015;61: 37. Buck MD, Atreja A, Brunker CP, Jain A, Suh TT, Palmer RM, et al. Potentially 301–9. inappropriate medication prescribing in outpatient practices: prevalence 14. Oscanoa TJ, Lizaraso F, Carvajal A. Hospital admissions due to adverse drug and patient characteristics based on electronic health records. Am J Geriatr reactions in the elderly. A meta-analysis. Eur J Clin Pharmacol. 2017 Jun 1; Pharmacother. 2009;7(2):84–92. 73(6):759–70. 38. Slabaugh SL, Maio V, Templin M, Abouzaid S. Prevalence and risk of 15. Muth C, Blom JW, Smith SM, Johnell K, Gonzalez-Gonzalez AI, Nguyen TS, et polypharmacy among the elderly in an outpatient setting. Drugs Aging. al. Evidence supporting the best clinical management of patients with 2010 Dec;27(12):1019–28. multimorbidity and polypharmacy: a systematic guideline review and expert 39. Mizokami F, Koide Y, Noro T, Furuta K. Polypharmacy with common consensus. J Intern Med. 2019;285(3):272–88. diseases in hospitalized elderly patients. Am J Geriatr Pharmacother. 2012 16. Doos L, Roberts EO, Corp N, Kadam UT. Multi-drug therapy in chronic Apr;10(2):123–8. condition multimorbidity: a systematic review. Fam Pract. 2014;31(6):654–63. 40. Jackson CF, Wenger NK. Cardiovascular disease in the elderly. Rev Española 17. Everitt BS, Landau S, Leese M, Stahl D. Cluster analysis. 5th ed: In John Wiley Cardiol (English Ed). 2011;64(8):697–712. & Sons, Ltd; 2011. p. 321–30. 41. George CJ, Korc B, Ross JS. Appropriate proton pump inhibitor use among 18. Roso-Llorach A, Violán C, Foguet-Boreu Q, Rodriguez-Blanco T, Pons-Vigués older adults: a retrospective chart review. Am J Geriatr Pharmacother. 2008 M, Pujol-Ribera E, et al. Comparative analysis of methods for identifying Dec 1;6(5):249–54. Guisado-Clavero et al. BMC Family Practice (2019) 20:82 Page 12 of 12 42. Gurwitz JH, Field TS, Harrold LR, Rothschild J, Debellis K, Seger AC, et al. Incidence and preventability of adverse drug events among older persons in the ambulatory setting. JAMA. 2003 Mar 5;289(9):1107–16. 43. Maher RL, Hanlon J, Hajjar ER. Clinical consequences of polypharmacy in elderly. Expert Opin Drug Saf. 2014;13(1):59. 44. Madhusoodanan S, Bogunovic OJ. Safety of benzodiazepines in the geriatric population. Expert Opin Drug Saf. 2004 Sep 23;3(5):485–93. 45. Morgan SG, Weymann D, Pratt B, Smolina K, Gladstone EJ, Raymond C, et al. Sex differences in the risk of receiving potentially inappropriate prescriptions among older adults. Age Ageing. 2016 Jul;45(4):535–42. 46. Lopez AD, Collishaw NE, Piha T. A descriptive model of the cigarette epidemic in developed countries. Tob Control. 1994 Sep 1;3(3):242–7. 47. Huber CA, Szucs TD, Rapold R, Reich O. Identifying patients with chronic conditions using pharmacy data in Switzerland: an updated mapping approach to the classification of medications. BMC Public Health. 2013;13(1). 48. Ibarra-Castillo C, Guisado-Clavero M, Violan-Fors C, Pons-Vigués M, López- Jiménez T, Roso-Llorach A. Survival in relation to multimorbidity patterns in older adults in primary care in Barcelona, Spain (2010-2014): a longitudinal study based on electronic health records. J Epidemiol Community Health. 2018 Jan 12;72(3):185–92. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.