Open access Original research Multimorbidity patterns of chronic conditions and geriatric syndromes in older patients from the MoPIM multicentre cohort study Marisa Baré ‍ ‍,1,2 Susana Herranz,2,3 Albert Roso-­Llorach,4,5 Rosa Jordana,6 Concepción Violán,4 Marina Lleal ‍ ‍,1 Pere Roura-­Poch,2,7 Marta Arellano,8 Rafael Estrada,9 Gloria Julia Nazco,10 on behalf of the MoPIM study group To cite: Baré M, Herranz S, ABSTRACT Roso-­Llorach A, et al. Strengths and limitations of this study Objectives To estimate the frequency of chronic Multimorbidity patterns of conditions and geriatric syndromes in older patients chronic conditions and geriatric ►► The multimorbidity analysis in this study has been admitted to hospital because of an exacerbation of their syndromes in older patients developed considering a wide range of long-­term chronic conditions, and to identify multimorbidity clusters from the MoPIM multicentre conditions that may require healthcare in older cohort study. BMJ Open in these patients. people. 2021;11:e049334. doi:10.1136/ Design Multicentre, prospective cohort study. ►► To the best of our knowledge, this is the first pub- bmjopen-2021-049334 Setting Internal medicine or geriatric services of five lished study of multimorbidity clusters in older pa- general teaching hospitals in Spain. ►► Prepublication history and tients to include chronic diseases weighted by a Participants 740 patients aged 65 and older, hospitalised additional supplemental material clinical management score and geriatric syndromes. because of an exacerbation of their chronic conditions for this paper are available ►► Soft clustering is an innovative, methodologically ro- online. To view these files, between September 2016 and December 2018. bust technique that can lead to reliable results in the please visit the journal online Primary and secondary outcome measures Active field of multimorbidity analysis. (http://​dx.​doi.​org/​10.​1136/​ chronic conditions and geriatric syndromes (including risk ►► The list of chronic conditions and geriatric syn- bmjopen-​2021-​049334). factors) of the patient, a score about clinical management dromes used in this study is comprehensive but of chronic conditions during admission, and destination Received 23 January 2021 not standardised, thus hindering comparability with at discharge were collected, among other variables. Accepted 28 October 2021 other studies. Multimorbidity patterns were identified using fuzzy c-­ means cluster analysis, taking into account the clinical management score. Prevalence, observed/expected ratio and exclusivity of each chronic condition and geriatric Trial registration number NCT02830425. syndrome were calculated for each cluster, and the final solution was approved after clinical revision and BACKGROUND discussion among the research team. According to the most recent Eurostat base- Results 740 patients were included (mean age 84.12 line population projections, old-­age depen- years, SD 7.01; 53.24% female). Almost all patients had dency ratio (population 65 years and over two or more chronic conditions (98.65%; 95% CI 98.23% to 99.07%), the most frequent were hypertension (81.49%, divided by population 15–64 years) is about 95% CI 78.53% to 84.12%) and heart failure (59.86%, 32% in the European Union (EU) and it is 95% CI 56.29% to 63.34%). The most prevalent geriatric expected to reach 52% in 2050, meaning syndrome was polypharmacy (79.86%, 95% CI 76.82% that the EU’s population will continue to to 82.60%). Four statistically and clinically significant grow older.1 Together with the fact that multimorbidity clusters were identified: osteoarticular, chronic conditions (CC) are the main cause psychogeriatric, cardiorespiratory and minor chronic of disability and mortality in Europe, this © Author(s) (or their disease. Patient-­level variables such as sex, Barthel Index, implies that the coexistence of two or more employer(s)) 2021. Re-­use number of chronic conditions or geriatric syndromes, chronic health conditions, which constitutes permitted under CC BY-­NC. No chronic disease exacerbation 3 months prior to admission commercial re-­use. See rights the classic definition of multimorbidity, is or destination at discharge differed between clusters. and permissions. Published by becoming increasingly common.2 BMJ. Conclusions In older patients admitted to hospital because of the exacerbation of chronic health problems, Multimorbidity is therefore turning into For numbered affiliations see it is possible to define multimorbidity clusters using soft an important challenge for the health system end of article. because of the expanding proportion of older clustering techniques. These clusters are clinically relevant Correspondence to and could be the basis to reorganise healthcare circuits people with multiple CC and treatments as Dr Marisa Baré; or processes to tackle the increasing number of older, well as the difficulties associated with their ​mbare@​tauli.​cat multimorbid patients. clinical management (CM).3 4 Most clinical Baré M, et al. BMJ Open 2021;11:e049334. doi:10.1136/bmjopen-2021-049334 1 Open access practice guidelines are focused on single diseases, with December 2018 was designed. The detailed protocol was limited recommendations for multimorbid patients,5 and, previously published.18 in addition, randomised clinical trials often exclude older For the purposes of the study, older patients (≥65 patients with multimorbidity.6 Despite the importance years old) admitted as a result of the exacerbation of of multimorbidity in clinical practice, different criteria their chronic pathology were included. Patients referred about which conditions should be considered and how to home hospitalisation, admitted because of an acute to aggregate them are still under debate, which makes process not related to any chronic disease or with a fatal it difficult to compare different estimations around the outcome expected at the time of admission were not world.7–9 included. One of the novel, increasingly widespread definitions No written informed consent was deemed necessary for of multimorbidity considers the non-­spurious association this study. of certain CC by sharing pathophysiological mechanisms, giving rise to disease association patterns.10 In order to Data acquisition and variables identify those patterns, different statistical methodolo- The following sociodemographic and clinical data was gies have been explored. Among these techniques, soft retrieved by the clinical team responsible for the patient: clustering allows to focus on patients rather than diag- patient’s code, date of birth, sex, functional status just noses and is a useful method when there is a high overlap before entering the hospital (Barthel Index),19 house- of diagnoses between patients, as it enables patients to hold (alone, with relatives or other people, in a nursing belong to more than one multimorbidity pattern with a home), existence of any contact with healthcare services certain probability.11 (primary care, emergencies, hospital admission, outpa- Besides CC, other clinically relevant situations such tient care, home care) in the 3 months prior to hospi- as geriatric syndromes (GS) may also be considered in talisation due to exacerbation of any chronic disease, the definition of these patterns, since they might have a and destination at discharge from the present episode great impact on the health-­related quality of life and CM of hospitalisation (home, transfer to another hospital, of old patients.12 In fact, the purpose of multimorbidity transfer to a nursing home, exitus). characterisation (ie, predicting outcomes or use of health Active CC of the patient at arrival to hospital, including services, improving quality of care, organising healthcare some risk factors, were collected (see online supplemental services, etc), will have an influence on its definition. In table 1). For this purpose, the physicians of the project order to have an accurate picture of the morbidity of each defined, on a consensual basis, a limited list of 64 CC, patient, a global consideration of all conditions that may coming from the 114 groups defined by Salisbury et al20 require healthcare attention is necessary, even if they and including the 19 categories of the Charlson Index.21 are not the reason for hospitalisation. Along these lines, Following the same criteria as Salisbury et al, a condition some countries have explicitly recommended to acknowl- was considered to be chronic when it lasted for at least 6 edge all long-­term conditions for optimising care of adult months, including past conditions that require ongoing patients by reducing, for example, possible inappropriate disease or risk management, important conditions with treatments, multiple healthcare appointments or poor a significant risk of recurrence, or past conditions that health-­related quality of life.13–15 have continuing implications for patient management.20 During the past decade, there has been an increasing Drug-­related conditions of this list refer to poor manage- amount of publications that consider multimorbidity,16 ment of medication related to a chronic disease that has but few have focused on multimorbidity patterns in older clinical implications in that hospitalisation (such as any patients and even fewer take into account GS.17 For this drug intolerance or an excess drug poisoning). reason, we launched a multicentre study in 2016 with Additionally, for each of the CC, it was also recorded multiple aims related to multimorbidity, appropriateness if they had required CM (both at admission and during of chronic treatments and adverse drug reactions in older hospitalisation) by assigning a (subjective) correlative patients.18 The objectives of the present analyses were to score (CM=1, 2, 3…) to each one, according to their estimate the frequency of CC and GS in older patients clinical importance during the attention process. Thus, admitted to hospital because of an exacerbation of their CC that did not have any significance during hospitalisa- CC, and to identify possible multimorbidity patterns in tion, although recorded, had a score equal to zero. This these patients. correlative score was later used to compute a ratio to reflect the weight of each CC in each patient in the index hospitalisation. Specific GS and risk factors (acute confusional METHODS syndrome/delirium, chronic pain, cognitive/intellec- Design and setting tual impairment, constipation, depression or anxiety, A multicentre, prospective cohort study including older dysphagia, frailty, immobility, incontinence (urinary/ patients hospitalised at the internal medicine or geri- faecal), instability/falls, malnutrition, polypharmacy, atric services at five general teaching hospitals in three pressure ulcers, sensorial deficit, sleep disorders/ different regions of Spain between September 2016 and insomnia) were also recorded. Two of the departments 2 Baré M, et al. BMJ Open 2021;11:e049334. doi:10.1136/bmjopen-2021-049334 Open access systematically apply a recently developed scale for frailty,22 pattern. Through this technique, we obtained clusters of while the others consider clinical judgement (although individuals and a membership matrix, which indicates based on the same variables). the degree of participation of each subject in each cluster. In order to address potential sources of bias, a pilot study As a first step, and similarly to Violán et al,24 the PCAmix was conducted with the first 10 admissions per centre to algorithm for categorical and continuous data (GS and R validate the data collection process and identify problems variables, respectively) was implemented to reduce and that could arise. After that, proper changes were made transform the dataset to all continuous data.25 To decide in the protocol and questionnaire. All available informa- the number of retained dimensions, the Karlis-­Saporta-­ tion sources were consulted in order to register CC and Spinaki rule was used.26 Then, a soft clustering algorithm GS, and the defined list was not closed. Nonetheless, the was applied to fuzzily distribute the population into a set registration of CC and GS was based on clinical criteria. of clusters, corresponding to the different multimorbidity patterns. We computed three validation indices to obtain Sampling and analysis the optimal number of clusters (K) and the optimal value A consecutive sample of 740 patients meeting the inclu- of the fuzziness parameter (m): the partition coefficient sion criteria were included, proportionally distributed whose optimal choice for coefficient is at the maximum to the annual volume of hospitalisations of the medi- and the Xie-­Beni and the partition entropy validation cine and/or geriatric services of each centre. The esti- indices, whose optimal indices are at the minimum.27 mated sample of 800 patients (see protocol18) could not Considering the stochastic nature of the clusters, and the be reached due to organisational reasons in one of the requirement of stable multimorbidity clusters, 100 inde- participating centres. pendent clustering repetitions were applied to obtain the For the purposes of the analyses, some CC were grouped stable final solution. according to clinical criteria: Hemiplegia was included To describe each identified cluster of individuals, the in cerebrovascular disease; metastatic solid tumour, prevalence of CC and GS in each one was calculated. leukaemia, lymphoma and any malignancy were grouped Observed/expected (O/E) ratios were calculated by into ‘neoplasia’; hepatitis B and C were included in mild dividing the prevalence of a given disease within a cluster liver disease, and both congestive and non-­ congestive by its prevalence in the overall population. The exclusivity heart failure were grouped into ‘heart failure’. Other of CC and GS, defined as the fraction of patients with the diseases were finally excluded of the analyses considering disease in the cluster over the total number of patients that they have no impact on acute healthcare (cataract, with the disease, was also calculated. A CC or a GS was dermatitis, diverticular disease of the colon, glaucoma, considered to be relevant in a given cluster of individuals haemorrhoids, other vascular diseases and prostatic when its O/E ratio was >1 and its exclusivity was >25%.28–30 benign hypertrophy). In the end, 51 CC and 15 GS were The statistical significant final solution ranged from 4 to analysed. 8 clusters. After clinical revision and discussion among The updated Charlson Comorbidity Index,23 age the research team, four different clusters were consid- adjusted, was computed and categorised according to ered to be consistent with the clinical observations as well tertiles distribution. as the objective of the clustering. There is currently no Descriptive statistics were performed to assess patient consensus in the literature on the criteria used to select clinical and sociodemographic characteristics and to the number of clusters or the O/E ratio cut-­off point due obtain overall prevalence estimates of CC and GS, strat- to, in part, the novelty of the analysis. ified by sex. Multimorbidity was first defined as the pres- Finally, sociodemographic and clinical variables were ence of two or more CC. Cumulative number of CC and described for all patients assigned to each cluster. Anal- GS per patient were computed, respectively. yses were performed using R V.3.6.0 and SPSS V.22. Multimorbidity patterns Patient and public involvement CC or GS with a prevalence <2% were excluded to avoid Since this was an observational study with variables and statistical noise and therefore spurious findings in the outcomes related to the healthcare process, this research cluster solutions, leaving a list of 40 CC and 15 GS. In was developed without patient involvement. Patients were order to take into account if a CC had required CM, a not invited to comment on the study design and were not ratio variable (R) was computed as follows: consulted to develop patient relevant outcomes or inter- If CC=0 & CM=0 → R=0 pret the results. If CC=1 & CM=m → R=1/m; max (m)=max value (CM)=8; If CC=1 & CM=0 → R=0.1 RESULTS Multimorbidity patterns were identified using the fuzzy A total of 740 patients aged 65 years or older were c-­means cluster analysis algorithm, which belongs to the included, with a mean age of 84.12 years (SD 7.01), a family of soft clustering algorithms. The algorithm esti- 53.24% of females and a mean Barthel Index of 65.07 mates c cluster centres (similar to k-­means) but with fuzz- (median 75). Sociodemographic and clinical variables iness so that individuals may belong to more than one are summarised in table 1. Almost all patients had two Baré M, et al. BMJ Open 2021;11:e049334. doi:10.1136/bmjopen-2021-049334 3 Open access Table 1 Sociodemographic and clinical variables of the studied cohort Sociodemographic and clinical variables N % 95% CI Age <70 33 4.46 3.19 to 6.20 70–74 48 6.49 4.93 to 8.50 75–79 82 11.08 9.02 to 13.55 80–84 181 24.46 21.50 to 27.68 85–89 232 31.35 28.11 to 34.78 90–94 134 18.11 15.50 to 21.05 ≥95 30 4.05 2.85 to 5.73 Sex Female 394 53.24 49.64 to 56.81 Male 346 46.76 43.19 to 50.36 Barthel Index <20 90 12.16 10.00 to 14.71 20–35 76 10.27 8.28 to 12.67 40–55 124 16.76 14.24 to 19.62 60–95 294 39.73 36.27 to 43.30 100 156 21.08 18.30 to 24.17 Age adjusted, updated Charlson 2–5 148 20.00 17.28 to 23.03 Comorbidity Index 6–8 411 55.54 51.94 to 59.08 9–14 181 24.46 21.50 to 27.68 Household With relatives/other people 523 70.68 67.30 to 73.84 Nursing home 95 12.84 10.62 to 15.44 Alone 122 16.49 13.99 to 19.33 Chronic disease exacerbation No 225 30.41 27.20 to 33.81 3 months prior to admission Yes (total) 515 69.59 66.19 to 72.80  Primary care 342 46.22 42.65 to 49.82  Emergencies 263 35.54 32.17 to 39.06  Hospital admission 193 26.08 23.05 to 29.36  Outpatient care 8 1.08 0.55 to 2.12  Home care 14 1.89 1.13 to 3.15 Conditions requiring clinical 1 302 40.81 37.36 to 44.39 management 2 216 29.19 26.03 to 32.57 3 106 14.32 11.98 to 17.03 4 72 9.73 7.80 to 12.08 5 28 3.78 2.63 to 5.41 6 13 1.76 1.03 to 2.98 7 2 0.27 0.07 to 0.98 8 1 0.14 0.007 to 0.76 Destination at discharge Home 468 63.24 59.71 to 66.64 Nursing home 105 14.19 11.86 to 16.89 Another hospital 101 13.65 11.36 to 16.31 Exitus 66 8.92 7.07 to 11.19 Multimorbidity No 10 1.35 1.35 to 1.36 Yes 730 98.65 98.23 to 99.07 or more CC (98.65%; 95% CI 98.23% to 99.07%), with Figure 1 shows the distribution of the number of CC a median of 8 CC and 6 GS per patient. Nearly 70% had by age groups. The most frequent CC were hyperten- consulted a healthcare service in the 3 months prior to sion (81.49%, 95% CI 78.53% to 84.12%) and heart hospitalisation due to chronic disease exacerbation. failure (59.86%, 95% CI 56.29% to 63.34%) (see online 4 Baré M, et al. BMJ Open 2021;11:e049334. doi:10.1136/bmjopen-2021-049334 Open access Figure 1 Distribution of the number of chronic conditions (excluding the following risk factors: hypertension, dyslipidaemia, obesity, osteoporosis and drug-­related conditions) in relation to age groups. supplemental table 2). Heart failure was also the main cause of hospitalisation (30.7% of patients had CM score=1), followed by chronic obstructive pulmonary disease (COPD) (20.7%) (online supplemental table 3). There were some differences in CC between sexes, with females having more frequently heart failure, degener- ative arthropathy, obesity, hip fracture, thyroid disease, asthma, osteoporosis, vertigo and non-­ schizophrenic mental disorders. Males, in turn, had more frequently COPD, gout, neoplasia, peripheral arteriopathy and Figure 2 Observed/expected (O/E) ratio and prevalence of ulcerative disease. chronic conditions and geriatric syndromes/risk factors per The most prevalent GS was polypharmacy (79.86%, multimorbidity cluster. Conditions with exclusivity >25% and 95% CI 76.82% to 82.60%), followed by frailty (61.76%, O/E ratios >1 in each cluster are represented. Conditions are 95% CI 58.20% to 65.19%). Females had a significantly ordered by O/E ratio and from cluster 1 to 4. COPD, chronic higher number of GS compared with males (Wilcoxon obstructive pulmonary disease. rank sum test, p<0.001), as well as a higher prevalence of depression/anxiety, chronic pain, constipation, frailty, urinary/faecal incontinence and immobility. incontinence and frailty. Patients in this group had a Four statistically and clinically significant multimor- mean Barthel index lower than 50 and a high number bidity clusters or patterns were identified in our study of GS. Furthermore, nearly 20% of them were living in a population. For all clusters, CC and GS with an observed/ nursing home and in-­hospital mortality was about 13%. expected ratio >1 and exclusivity >25% are represented in Cluster 3, named minor chronic disease, had 179 figure 2 (see also online supplemental table 4) for all CC (24.2%) patients, and represents a group of patients and GS). Sociodemographic and clinical characteristics with a variety of conditions, such as hypertension, dyslipi- of patients in each cluster are described in table 2. daemia, anaemia, gout, chronic renal insufficiency, poly- The first cluster, named osteoarticular, included 132 pharmacy, non-­ischaemic heart disease, and diverse GS. patients (17.8%) having osteoporosis, fractures, inflam- O/E ratios were close to 1 in most cases. matory osteoarticular disease, chronic pain and degen- Finally, cluster 4, called cardiorespiratory, included erative arthropathy. Moreover, vertigo, sleep apnoea, 276 (37.3%) patients. The over-­represented diagnoses asthma, depression/anxiety and sleep disorders were also over-­represented. This cluster included patients with the were COPD, heart failure and cardiac arrhythmia, highest number of both CC and GS. About three-­quarters although the O/E ratios were very low. In this cluster, were female, and most of them (82%) accessed health- with the lowest number of CC and GS, and a Barthel care services 3 months prior to this admission. index greater than 75, nearly 40% had no healthcare Cluster 2, called psychogeriatric, had 152 patients consultation for a chronic disease exacerbation in the (20.7%) and included mostly GS: pressure ulcers, immo- previous 3 months. This group had the lowest in-­hos- bility, malnutrition, cognitive impairment, dementia, pital mortality (5%). Baré M, et al. BMJ Open 2021;11:e049334. doi:10.1136/bmjopen-2021-049334 5 6 Table 2 Sociodemographic and clinical variables of the multimorbidity clusters Minor chronic Open access Osteoarticular Psychogeriatric disease Cardiorespiratory Number of patients included, n (%) 132 (17.8) 153 (20.7) 179 (24.2) 276 (37.3) Age at admission (year, mean±SD) 84.03±6.48 84.51±7.25 83.94±7.19 84.06±7.03 Sex, n (%) Male 34 (25.7) 66 (42.8) 99 (55.5) 147 (53.4) Female 98 (74.3) 87 (57.2) 80 (44.5) 129 (46.6) Barthel Index (mean±SD) 63.06±24.78 47.62±34.94 64.96±33.56 75.76±27.52 Total no chronic conditions (mean±SD) 11.5±3.64 7.68±3.19 8.86±3.08 7.59±2.61 Total no geriatric syndromes/risk factors 7.76±2.07 8.16±2.82 6.4±3.32 4.42±2 (mean±SD) Charlson Comorbidity Index, n (%) 2–5 26 (19.9) 24 (15.6) 37 (20.4) 61 (22.2) 6–8 73 (55.1) 89 (58.5) 96 (53.8) 153 (55.3) 9–14 33 (25.0) 40 (25.9) 46 (25.7) 62 (22.6) Household, n (%) With relatives/other 91 (68.7) 103 (67.2) 133 (74.5) 196 (71.1) people Nursing home 16 (11.8) 28 (18.4) 23 (12.8) 28 (10.3) Alone 26 (19.5) 22 (14.3) 23 (12.7) 52 (18.7) Chronic disease exacerbation 3 months prior to the No 24 (18.3) 46 (30.1) 48 (27.0) 106 (38.5) index admission, n (%) Yes (total) 108 (81.7) 107 (69.9) 130 (73.0) 170 (61.5)  Primary care 83 (62.7) 66 (43.3) 92 (51.6) 101 (36.5)  Emergencies 71 (53.6) 40 (26.2) 69 (38.5) 83 (30.2)  Hospital 49 (37.1) 46 (30.1) 47 (26.1) 51 (18.6) admission  Outpatient care 1 (0.7) 0 (0.3) 3 (1.4) 4 (1.4)  Home care 2 (1.5) 2 (1.1) 3 (1.8) 7 (2.6) Destination at discharge, n (%) Home 86 (65.3) 83 (54.0) 111 (62.0) 188 (68.1) Nursing home 15 (11.1) 33 (21.3) 26 (14.5) 32 (11.5) Another hospital 15 (11.3) 17 (11.3) 26 (14.7) 42 (15.3) Exitus 16 (12.3) 20 (13.3) 16 (8.8) 14 (4.9) Baré M, et al. BMJ Open 2021;11:e049334. doi:10.1136/bmjopen-2021-049334 Open access DISCUSSION chronic pathology, and high in-­hospital mortality. Thus, This study aimed to identify multimorbidity patterns in this profile would identify a group of patients with a high patients aged 65 and above admitted to hospital because probability of decompensation and death. of an exacerbation of CC. The soft clustering technique Finally, the so-­ called minor chronic diseases cluster used, together with clinical criteria, was able to identify was not very well defined. It included some risk factors four different multimorbidity patterns, named osteo- (hypertension, dyslipidaemia, polypharmacy) as well as articular, psychogeriatric, cardiorespiratory and minor some CC and GS. Thus, it would be possible that it does chronic disease, in a patient-­ centred approach taking not represent a real cluster but either the set of cases that into account the importance of each disease in hospital did not belong anywhere else. management. Remarkably, high chronic multimorbidity was found in all patients, regardless of the cluster. To the Comparison with other studies best of our knowledge, this is the only study published to Given the type of patients under study and the method- date that has analysed multimorbidity patterns taking into ological approach to identify multimorbidity patterns, account both CC (with their weight during CM) and GS there are few publications to directly compare our in this type of patients. Hence, these identified patterns results to. Clerencia-­Sierra et al17 analysed multimorbidity allow us to take a further step towards understanding the patterns in hospitalised older patients. Their method- patients’ current or future healthcare needs. ological and analytical approach was slightly different, Two very important aspects of multimorbidity and they did not take into account the weight of the patterns analysis are the purpose for designing such diseases during the hospitalisation process; however, they patterns and the target population, which clearly found a similar percentage of multimorbidity (99.7%) condition the obtained results or conclusions. For and four patterns that partially coincide with those of instance, our aim in defining multimorbidity patterns our study: cardiovascular, induced-­dependency, falls and in this cohort was to identify profiles of patients with osteoarticular. similar clinical needs during the index hospitalisation Furthermore, several authors have published data on and even a similar short-­term prognosis at that time. patterns identified from primary care electronic records For this reason, the importance of their pathologies in different age groups, with lists of non-­ comparable in the course of hospitalisation was also taken into chronic problems and using different techniques (cluster account. Hence, the ones that tend to have a minimal analysis, exploratory factor analysis or latent class anal- impact on CM, such as risk factors like hypertension ysis).17 29–34 These results would not be directly compa- or dyslipidaemia did not have a leading role in the rable with our study, but all of them highlight the ability patterns. to identify association patterns of chronic diseases. All clusters contain coherent groups of conditions that are mostly pathophysiologically related. From the Strengths and limitations clinical point of view, these clusters resemble patient The strengths of this study are the prospective design, profiles that are intuitively perceived. Moreover, some ensuring data quality by thorough record keeping, the descriptive variables such as sex, Barthel Index, mean ascertainment of all CC and GS of the patient, as well as number of CC or GS, chronic pathology exacerba- the use of a novel clustering technique. Soft clustering tions in previous months, or hospital mortality, are is a methodologically robust technique less susceptible distributed in such a way that they may reinforce the to outliers in the data, choice of distance measure and distinction of these groups. the inclusion of inappropriate or irrelevant variables.24 Coexistence of CC and GS was observed in all clus- Besides, our approach focuses beyond organ diseases ters except for the cardiorespiratory, reinforcing the by incorporating GS, and using a comprehensive list need to consider other clinically relevant situations retrieved by the clinical team. Additionally, we have taken rather than only CC. In particular, the exclusivity and into account the relative importance of the different CC prevalence of GS such as immobility, malnutrition, in the CM of the patient during hospitalisation, thus cognitive impairment or dementia were considerable providing a better picture of the possible complexity and in the psychogeriatric cluster. needs during hospitalisation. Interestingly, highly prevalent CC, such as heart failure Furthermore, our work is not only limited to the iden- and COPD, which also frequently involve CM, only showed tification of possible patterns. We have validated them, remarkable exclusivity and O/E ratio in the cardiorespi- in some way, by analysing some of the patients’ variables ratory pattern and were not over-­represented elsewhere. such as sex, number of CC, previous contacts with the This highlights the fact that even though some CC may health system, hospital mortality or need for a nursing not be over-­represented in a cluster, they can have a high home. prevalence and therefore need to be properly addressed Nonetheless, our study presents some limitations that too. need to be considered. First, the identification of chronic With respect to the osteoarticular cluster, it displayed pathologies does not exclusively follow a validated list of a pattern of female predominance, with many CC and codes but either an adaptation of different ones, a fact GS, high healthcare needs in recent months due to their that could hinder comparability with other studies on Baré M, et al. BMJ Open 2021;11:e049334. doi:10.1136/bmjopen-2021-049334 7 Open access multimorbidity. Second, as this study is not longitudinal, patients with an increased risk of potentially inappro- the chronology in which CC or GS appear cannot be anal- priate prescription or adverse drug reactions. These ysed. It is possible for a patient to evolve from one pattern aspects will be the object of future analyses. to another throughout life, as some authors have already pointed out,35 and therefore, the results only show the present situation. However, given the purpose of the CONCLUSIONS defined patterns, this would not in itself be a limitation. In conclusion, in older patients admitted to hospital From a clinical point of view, the lack of usage of stan- because of the exacerbation of chronic health problems, dard scales or diagnostic criteria for determining all CC it is possible to define multimorbidity clusters or patterns or GS could question the validity of this information. using appropriate statistical techniques. These patterns However, the study gathered the data as it was routinely seem clinically coherent and could be the basis to reorga- registered in the different departments. Frailty should nise circuits, processes or healthcare models to tackle the derive from a comprehensive assessment of the patient increasing number of older, multimorbid patients. in a standardised way that still lacks of systematic imple- mentation in the healthcare routine.36 Nevertheless, our Author affiliations multimorbidity study wanted to go a little further, also 1 Clinical Epidemiology and Cancer Screening, Consorci Corporació Sanitària Parc considering GS (frailty included), an unusual fact in Taulí, Sabadell, Spain 2 the bibliography on clusters of multimorbidity in older REDISSEC-­Network for Research into Healthcare in Chronic Diseases, Madrid, Spain patients in spite of its importance for decision making in 3 Acute Care Geriatric Unit, Consorci Corporació Sanitària Parc Taulí, Sabadell, Spain the clinical practice. 4 IDIAP Jordi Gol, Barcelona, Spain The clinical conditions severity or other possible aggra- 5 Universitat Autònoma de Barcelona, Barcelona, Spain 6 vating factors have neither been gathered. Nevertheless, Internal Medicine, Consorci Corporació Sanitària Parc Taulí, Sabadell, Spain 7 the registration of a variable that takes into account the Epidemiology, Consorci Hospitalari de Vic, Vic, Spain 8 Geriatrics, Consorci Parc de Salut MAR de Barcelona, Barcelona, Spain relevance of each CC during the care process acts, in 9 Internal Medicine, Hospital Galdakao-­Usansolo, Galdakao, Spain some way, as a proxy of the importance of each disease 10 Hospital Pharmacy, Hospital Universitario de Canarias, La Laguna, Spain in the index hospitalisation when dealing with patients admitted because of decompensation. Considering the Twitter Marisa Baré @BareMarisa purpose of defining the patterns in the whole study, and Acknowledgements The authors acknowledge the dedication and support of the not knowing useful precedents in the consulted bibliog- entire MoPIM research group. We are grateful to Sandra Prados, Antonio Gimeno raphy, the assignment made by the medical professional and Beatriz Poblador for their insightful comments, as well as to Blanca Rovira for her literature review as part of the final project of her medical degree at UAB, and to who attended the patient was an easy, simple measure, Celia Corral, for helping in tables and figures editing. and shared by all professionals at the time of writing the Collaborators MoPIM study group: Parc Taulí University Hospital: Marisa Baré clinical course. (Clinical Epidemiology and Cancer Screening; REDISSEC), Susana Herranz (Acute care Geriatric Unit), Rosa Jordana (Department of Internal Medicine), Maria Clinical implications Queralt Gorgas (Pharmacy Department, REDISSEC), Sara Ortonobes (Pharmacy These patterns are not a picture of the community but Department), Marina Lleal (Clinical Epidemiology and Cancer Screening); University Hospital of Vic: Pere Roura-­Poch (Epidemiology Unit; REDISSEC), Daniel Sevilla, of older patients in geriatric or internal medicine depart- Núria Solà and Javier González (Pharmacy Department), Núria Molist, Mariona ments, which are generally in more need of health services Espauella (Department of Geriatrics); Oscar Mascaró (Department of Internal and more complex CM. However, not all of these patients Medicine); Hospital del Mar Medical Research Institute-­IMIM: Elisabet de Jaime have the same requirements. In fact, one in five patients (Geriatrics Department), Olivia Ferrandez (Pharmacy Department), Maria Sala (Department of Epidemiology and Evaluation, REDISSEC), Miguel Ángel Márquez, (the psychogeriatric cluster) caused a great burden to Marta Arellano, Carlos Clemente and Olga Sabartés (Department of Geriatrics), both the patient and their relatives while the patients in Núria Carballo and Marta de Antonio (Pharmacy Department); Hospital de Galdakao: the most frequent cluster (cardiorespiratory), with lower Rafael Estrada (Department of Internal Medicine), Maria Olatz Ibarra (Pharmacy dependency and less GS, seemed to have better imme- Department); Complejo Hospitalario Universitario de Canarias: Candelaria Martin (Department of Internal Medicine), Gloria Julia Nazco (Pharmacy Department), diate outcome. Therefore, it is possible that the thera- Rubén Hernández (Department of Internal Medicine). peutic objectives should be different in these patients. Contributors MB conceived and supervised the study, discussed the results, More importantly, the ability to distinguish patients more wrote the first version of the manuscript and is responsible for the overall content objectively than with the mere clinical impression may as guarantor. SH, RJ, MA, RE and GJN participated in patient inclusion, data allow the design of better processes, services or alterna- collection and discussion of the results and revised the manuscript. AR-­L executed the analysis and interpretation of multimorbidity clusters. CV collaborated in the tives to conventional hospitalisation. Indeed, the iden- execution of the analyses and in the interpretation of the multimorbidity patterns. tification of multimorbidity patterns in subsets of the ML participated in the statistical and graphical analysis of the results, in the population in order to detect underlying factors, under- discussion of the results and the revision of several manuscript versions. PR-­P stand their burden on patients and develop preventive collaborated in the questionnaires’ design, patient inclusion, data collection and discussion of results and revised the manuscript. All authors read and approved the strategies is considered a research priority. Finally, the final manuscript. development of clinical practice guidelines according to Funding This work was supported by grants from Instituto de Salud Carlos these patterns needs to be considered, although it may be III-­FEDER, (PI15/00552) and Institut d'Investigació i Innovació Parc Taulí (I3PT) difficult given the magnitude of the diseases comprised in [CIR2017/0070], and by the Network for Research into Healthcare in Chronic each pattern.4 14 In addition, some patterns may include Diseases, REDISSEC (RD16/0001/0002). 8 Baré M, et al. BMJ Open 2021;11:e049334. doi:10.1136/bmjopen-2021-049334 Open access Disclaimer These funding bodies had no role in the design of the study, nor in the octogenarians: a population-­based cohort study. Geriatr Gerontol Int collection, analysis and interpretation of data nor in writing the manuscript. 2016;16:345–51. 13 Heide vander I, Snoeijs SP, Boerma WGW. How to strengthen Competing interests None declared. patient-­centredness in caring for people with multimorbidity in Patient consent for publication Not applicable. Europe? 2017. Available: http://​europepmc.​org/​books/​NBK464537 14 Rijken M, Struckmann V, Heide VD I. How to improve care for Ethics approval This study was approved by the clinical research ethics people with multimorbidity in Europe? 2016. Available: https:// committees of each centre: Comité Ético de investigación Clínica del Parc Taulí, www.​euro.​who.​int/​en/​about-​us/​partners/​observatory/​publications/​ Comitè Ètic d'Investigació Clínica Osona per a la Recerca i Educació Sanitàries policy-​briefs-​and-​summaries/​how-​to-​improve-​care-​for-​people-​with-​ (FORES), Comité de Ètica de la Investigación con Medicamentos (CEIm)-­Parc de multimorbidity-​in-​europe 15 NICE. Multimorbidity: clinical assessment and management NICE Salut MAR, Comité Ético de Investigación Clínica de Euskadi, Comité de Ética de guideline, 2016. Available: www.​nice.​org.​uk/​guidance/​ng56 Investigación del Hospital Universitario de Canarias. No written informed consent 16 Catalá-López F, Alonso-­Arroyo A, Page MJ, et al. Mapping of global was deemed necessary for this study. scientific research in comorbidity and multimorbidity: a cross-­ Provenance and peer review Not commissioned; externally peer reviewed. sectional analysis. PLoS One 2018;13:e0189091. 17 Clerencia-­Sierra M, Calderón-­Larrañaga A, Martínez-­Velilla N, et al. Data availability statement Data are available on reasonable request. Multimorbidity patterns in hospitalized older patients: associations among chronic diseases and geriatric syndromes. PLoS One Supplemental material This content has been supplied by the author(s). It has 2015;10:e0132909. not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been 18 Baré M, Herranz S, Jordana R, et al. Multimorbidity patterns in peer-­reviewed. Any opinions or recommendations discussed are solely those chronic older patients, potentially inappropriate prescribing and of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and adverse drug reactions: protocol of the multicentre prospective responsibility arising from any reliance placed on the content. Where the content cohort study MoPIM. BMJ Open 2020;10:e033322. includes any translated material, BMJ does not warrant the accuracy and reliability 19 Shah S, Vanclay F, Cooper B. Improving the sensitivity of of the translations (including but not limited to local regulations, clinical guidelines, the Barthel index for stroke rehabilitation. J Clin Epidemiol terminology, drug names and drug dosages), and is not responsible for any error 1989;42:703–9. and/or omissions arising from translation and adaptation or otherwise. 20 Salisbury C, Johnson L, Purdy S, et al. Epidemiology and impact of multimorbidity in primary care: a retrospective cohort study. Br J Gen Open access This is an open access article distributed in accordance with the Pract 2011;61:e12–21. Creative Commons Attribution Non Commercial (CC BY-­NC 4.0) license, which 21 Charlson ME, Pompei P, Ales KL, et al. A new method of classifying permits others to distribute, remix, adapt, build upon this work non-­commercially, prognostic comorbidity in longitudinal studies: development and and license their derivative works on different terms, provided the original work is validation. J Chronic Dis 1987;40:373–83. 22 Amblàs-­Novellas J, Martori JC, Espaulella J, et al. Frail-­VIG index: properly cited, appropriate credit is given, any changes made indicated, and the use a concise frailty evaluation tool for rapid geriatric assessment. BMC is non-­commercial. See: http://​creativecommons.​org/​licenses/​by-​nc/​4.​0/. Geriatr 2018;18:29. 23 Quan H, Li B, Couris CM, et al. Updating and validating the ORCID iDs Charlson comorbidity index and score for risk adjustment in hospital Marisa Baré http://​orcid.​org/​0000-​0001-​8085-​2833 discharge Abstracts using data from 6 countries. Am J Epidemiol Marina Lleal http://​orcid.​org/​0000-​0002-​0015-​0725 2011;173:676–82. 24 Violán C, Foguet-­Boreu Q, Fernández-­Bertolín S, et al. Soft clustering using real-­world data for the identification of multimorbidity patterns in an elderly population: cross-­sectional study in a Mediterranean REFERENCES population. BMJ Open 2019;9:e029594. 1 Eurostat - Data Explorer. Available: https://​appsso.​eurostat.​ec.​ 25 Chavent M, Kuentz-­Simonet V, Labenne A. Multivariate analysis of europa.​eu/​nui/​show.​do?​dataset=​tps00200&​lang=​en [Accessed 21 mixed data: the R package PCAmixdata.. arXiv 2017 https://​arxiv.​ Oct 2020]. org/​abs/​1411.​4911v4 2 OECD/EU. Health at a glance: Europe 2018 state of health in the EU 26 Karlis D, Saporta G, Spinakis A. A simple rule for the selection cycle 2018. of principal components. Commun Stat Theory Methods 3 Palladino R, Pennino F, Finbarr M, et al. Multimorbidity and health 2003;32:643–66. outcomes in older adults in ten European health systems, 2006-­15. 27 Zhao Q. Cluster validity in clustering methods, 2012. Available: http://​ Health Aff 2019;38:613–23. cs.​joensuu.​fi/​sipu/​pub/​qinpei-​thesis.​pdf 4 The Academy of Medical Sciences. Multimorbidity: a priority for 28 Guisado-­Clavero M, Roso-­Llorach A, López-­Jimenez T, et al. global health research, 2018. Available: https://​acmedsci.​ac.​uk/​file-​ Multimorbidity patterns in the elderly: a prospective cohort study download/​82222577 with cluster analysis. BMC Geriatr 2018;18:16. 5 Lugtenberg M, Burgers JS, Clancy C, et al. Current guidelines 29 Marengoni A, Roso-­Llorach A, Vetrano DL. Patterns of multimorbidity have limited applicability to patients with comorbid conditions: in a population-­based cohort of older people: sociodemographic, a systematic analysis of evidence-­based guidelines. PLoS One lifestyle, clinical, and functional differences. J Gerontol A Biol Sci 2011;6:e25987. Med Sci 2019;75:798–805. 6 Buffel du Vaure C, Dechartres A, Battin C, et al. Exclusion of patients 30 Violán C, Roso-­Llorach A, Foguet-­Boreu Q, et al. Multimorbidity with concomitant chronic conditions in ongoing randomised patterns with k-­means nonhierarchical cluster analysis. BMC Fam controlled trials targeting 10 common chronic conditions and Pract 2018;19:108. registered at ​ClinicalTrials.​gov: a systematic review of registration 31 Juul-­Larsen HG, Christensen LD, Bandholm T, et al. Patterns details. BMJ Open 2016;6:e012265. of Multimorbidity and Differences in Healthcare Utilization and 7 Fortin M, Stewart M, Poitras M-­E, et al. A systematic review of Complexity Among Acutely Hospitalized Medical Patients (≥65 Years) prevalence studies on multimorbidity: toward a more uniform - A Latent Class Approach. Clin Epidemiol 2020;12:245–59. methodology. Ann Fam Med 2012;10:142–51. 32 Zhu Y, Edwards D, Mant J, et al. Characteristics, service use and 8 Diederichs C, Berger K, Bartels DB. The measurement of multiple mortality of clusters of multimorbid patients in England: a population-­ chronic diseases--a systematic review on existing multimorbidity based study. BMC Med 2020;18:78. indices. J Gerontol A Biol Sci Med Sci 2011;66:301–11. 33 Nguyen QD, Wu C, Odden MC, et al. Multimorbidity patterns, frailty, 9 Roso-­Llorach A, Violán C, Foguet-­Boreu Q, et al. Comparative and survival in community-­dwelling older adults. J Gerontol A Biol analysis of methods for identifying multimorbidity patterns: a study of Sci Med Sci 2019;74:1265–70. 'real-­world' data. BMJ Open 2018;8:e018986. 34 Poblador-­Plou B, van den Akker M, Vos R, et al. Similar 10 Prados-­Torres A, Calderón-­Larrañaga A, Hancco-­Saavedra J, et al. multimorbidity patterns in primary care patients from two European Multimorbidity patterns: a systematic review. J Clin Epidemiol regions: results of a factor analysis. PLoS One 2014;9:e100375. 2014;67:254–66. 35 Vetrano DL, Roso-­Llorach A, Fernández S, et al. Twelve-­year clinical 11 Jyoti Bora D, Kumar Gupta A. A comparative study between fuzzy trajectories of multimorbidity in a population of older adults. Nat clustering algorithm and hard clustering algorithm. Int J Comput Commun 2020;11:1–9. Trends Technol 2014;10. 36 Ward K, Reuben D. Comprehensive geriatric assessment. 12 Lu F-­P, Chang W-­C, Wu S-­C. Geriatric conditions, rather than In: UpToDate, 2020. https://www.​uptodate.​com/​contents/​ multimorbidity, as predictors of disability and mortality among comprehensive-​geriatric-​assessment Baré M, et al. BMJ Open 2021;11:e049334. doi:10.1136/bmjopen-2021-049334 9