Defining and measuring multimorbidity: a systematic review of systematic reviews

Defining and measuring multimorbidity: a systematic review of systematic reviews Abstract Background Multimorbidity, the coexistence of multiple health conditions, is a growing public health challenge. Research and intervention development are hampered by the lack of consensus regarding defining and measuring multimorbidity. The aim of this systematic review was to pool the findings of systematic reviews examining definitions and measures of multimorbidity. Methods Medline, Embase, PubMed and Cochrane were searched from database inception to February 2017. Two authors independently screened titles, abstracts and full texts and extracted data from the included papers. Disagreements were resolved with a third author. Reviews were quality assessed. Results Of six reviews, two focussed on definitions and four on measures. Multimorbidity was commonly defined as the presence of multiple diseases or conditions, often with a cut-off of two or more. One review developed a holistic definition including biopsychosocial and somatic factors as well as disease. Reviews recommended using measures validated for the outcome of interest. Disease counts are an alternative if no validated measure exists. Conclusions To enable comparison between studies and settings, researchers and practitioners should be explicit about their choice of definition and measure. Using a cut-off of two or more conditions as part of the definition is widely adopted. Measure selection should be based on tools validated for the outcome being considered. Where there is no validated measure, or where multiple outcomes or populations are being considered, disease counts are appropriate. Introduction Multimorbidity is commonly understood to be the coexistence of multiple health conditions in an individual.1,2 A related term, comorbidity, describes the burden of illness co-existing with a particular disease of interest.3 Multimorbidity is a growing global public health challenge as populations age and the prevalence of long-term conditions rises.1,2 Multimorbidity is associated with poorer outcomes and the increased use of health and social care services with associated costs.4,5 There is increasing awareness that healthcare services are not adequately designed to meet the challenges of multimorbidity. Secondary care services are generally single disease focussed.6,7 Practitioners, particularly in primary care, face challenges in using clinical guidelines that are generally developed for single conditions or groups of similar conditions.8 These issues bring associated risks, for example, polypharmacy,8,9 and challenges associated with managing patients with complex needs in resource limited environments.6 Multimorbidity also places a burden on individuals who face poorer quality of life and increased disability.10 It is highly correlated with frailty (an age-related decline leading to reduced reserves of physical and mental health capacity, resulting in vulnerability to stressors and an increased risk of poor health outcomes).11,12 Despite these challenges, there is no international consensus regarding the best way to define and measure multimorbidity.13 This makes carrying out and interpreting research, comparing findings across populations and developing guidelines and interventions difficult. A review of prevalence studies of multimorbidity found estimates ranging from between less than 5% to more than 95%, often due to differences in the operational definition of multimorbidity.2 The National Institute for Health and Care Excellence (NICE) recently developed a multimorbidity guideline and commented that measuring the prevalence of multimorbidity is complex due to the varying measures being used.14 A number of reviews have summarized the multimorbidity definitions or measures used in primary studies. Our aim was to build consensus on the most appropriate ways to define and measure multimorbidity by pooling the findings of these systematic reviews. Methods The PRISMA 2009 checklist guided method development and reporting of findings.15 Medline, Embase, PubMed and the Cochrane database of systematic reviews were searched from database inception to 13 February 2017. The search strategy was comparable across all databases. At the time of searching, there was no MeSH term for multimorbidity. The search terms relating to ‘multimorbidity’ and its measures were drawn from a previous systematic review of the multimorbidity literature.16 These were combined by the Boolean operator ‘AND’ with ‘review’ as a title word. The terms were searched in the title only, as an initial trial search found that widening this to the abstract or full text significantly reduced the ability to detect relevant reviews. The search strategy is in Supplementary table S1. Systematic reviews of the multimorbidity literature which examined multimorbidity definitions and/or measures as a central focus of the review were included. While comorbidity is now commonly accepted to be distinct from multimorbidity, it is known that the terms have been used synonymously in the past. Reviews of comorbidity where no specific index disease was considered were therefore eligible. Systematic reviews that did not have the primary aim to summarize multimorbidity definitions and measures were excluded. Reviews that were ‘narrative’ or ‘semi-structured’ or which otherwise were not systematic reviews were excluded. Title, abstract and full-text screening were carried out independently by two authors (MCJ and SWM). Disagreement was resolved by CB. Primary data extraction was carried out by MCJ with four others acting as independent second reviewers (CB, MC, GJP and SWM). The data extraction form was prepared and piloted by MCJ and finalized by discussion with the other reviewers. Data extraction included the review characteristics, the definition and measures of multimorbidity presented in the review and the rationale behind any recommended measures of multimorbidity (if given). Scottish Intercollegiate Guidelines Network critical appraisal checklists17 were used to assess the quality of included reviews (‘low quality’, ‘acceptable’ or ‘high quality’). The results were combined narratively. Results Figure 1 summarizes the results of the search. Out of 1051 articles sourced during the search, there were 432 duplicates. Following screening of titles, abstracts and full texts, six reviews were included.16,18–22 The characteristics of these reviews, including their stated aims, are presented in table 1. The Le Reste and Willadsen reviews focused on the definition of multimorbidity, while the remaining four focussed on measures. The number of studies included by the reviews ranged from 39 to 194. Five reviews were of ‘acceptable quality’.16,19–22 De Groot was ‘low quality’ as they did not report the literature search strategy, the results of the literature search and the identification of papers clearly.18 Table 1 Characteristics of included reviews Review reference  Stated aim of the review  Databases and dates of search undertaken by the review  Total titles/ abstracts screened  Total texts included  Quality assessmenta  De Groot et al.18  ‘Which methods are available for measuring comorbidity that can be used in RCTs and prognostic studies’  Medline: January 1966 to September 2000. Embase: January 1988 to September 2000.  Not reported  Not reported  Low quality  Diederichs et al.19  ‘Multimorbidity, defined as the coexistence of 2 or more chronic diseases, is a common phenomenon especially in older people. Numerous efforts to establish a standardized instrument to assess the level of multimorbidity have failed until now, and indices are primarily characterized by their high heterogeneity. Thus, the objective is to provide a comprehensive overview on existing instruments on the basis of a systematic literature review’.  Medline: 1 January 1960 to 31 August 2009  1120  39  Acceptable  Huntley et al.16  ‘The aims of this review were (1) to identify and describe measures of multimorbidity that are most suitable for use in research in primary care and community populations, taking into account the data and resources they require and (2) to investigate the validity of these measures in terms of whether they have demonstrated anticipated associations with patient characteristics, process measures, and health outcomes.’  Medline and Embase: database inception to December 2009  11 191  194  Acceptable  Le Reste et al.20  ‘What are the criteria for multimorbidity found in the scientific medical literature and what definition could be produced with these criteria?’  PubMed, Embase and Cochrane: 1 January 1990 to 31 December 2010  416  54  Acceptable  Yurkovich et al.21  ‘To conduct a systematic review of studies reporting on the development or validation of comorbidity indices using administrative health data and compare their ability to predict outcomes related to comorbidity (i.e., construct validity).’  Medline and Embase: 1946 to September 2012  955  76  Acceptable  Willadsen et al.22  ‘Objective is to explore how multimorbidity is defined in the scientific literature, with a focus on the roles of diseases, risk factors, and symptoms in the definitions’.  PubMed, Medline and Embase: inception to 4 October 2013. Cochrane database: inception to 10 October 2013  943  163  Acceptable  Review reference  Stated aim of the review  Databases and dates of search undertaken by the review  Total titles/ abstracts screened  Total texts included  Quality assessmenta  De Groot et al.18  ‘Which methods are available for measuring comorbidity that can be used in RCTs and prognostic studies’  Medline: January 1966 to September 2000. Embase: January 1988 to September 2000.  Not reported  Not reported  Low quality  Diederichs et al.19  ‘Multimorbidity, defined as the coexistence of 2 or more chronic diseases, is a common phenomenon especially in older people. Numerous efforts to establish a standardized instrument to assess the level of multimorbidity have failed until now, and indices are primarily characterized by their high heterogeneity. Thus, the objective is to provide a comprehensive overview on existing instruments on the basis of a systematic literature review’.  Medline: 1 January 1960 to 31 August 2009  1120  39  Acceptable  Huntley et al.16  ‘The aims of this review were (1) to identify and describe measures of multimorbidity that are most suitable for use in research in primary care and community populations, taking into account the data and resources they require and (2) to investigate the validity of these measures in terms of whether they have demonstrated anticipated associations with patient characteristics, process measures, and health outcomes.’  Medline and Embase: database inception to December 2009  11 191  194  Acceptable  Le Reste et al.20  ‘What are the criteria for multimorbidity found in the scientific medical literature and what definition could be produced with these criteria?’  PubMed, Embase and Cochrane: 1 January 1990 to 31 December 2010  416  54  Acceptable  Yurkovich et al.21  ‘To conduct a systematic review of studies reporting on the development or validation of comorbidity indices using administrative health data and compare their ability to predict outcomes related to comorbidity (i.e., construct validity).’  Medline and Embase: 1946 to September 2012  955  76  Acceptable  Willadsen et al.22  ‘Objective is to explore how multimorbidity is defined in the scientific literature, with a focus on the roles of diseases, risk factors, and symptoms in the definitions’.  PubMed, Medline and Embase: inception to 4 October 2013. Cochrane database: inception to 10 October 2013  943  163  Acceptable  Notes: SIGN, Scottish Intercollegiate Guidelines Netwok; RCT, randomized controlled trial. a Based upon SIGN categories. Table 1 Characteristics of included reviews Review reference  Stated aim of the review  Databases and dates of search undertaken by the review  Total titles/ abstracts screened  Total texts included  Quality assessmenta  De Groot et al.18  ‘Which methods are available for measuring comorbidity that can be used in RCTs and prognostic studies’  Medline: January 1966 to September 2000. Embase: January 1988 to September 2000.  Not reported  Not reported  Low quality  Diederichs et al.19  ‘Multimorbidity, defined as the coexistence of 2 or more chronic diseases, is a common phenomenon especially in older people. Numerous efforts to establish a standardized instrument to assess the level of multimorbidity have failed until now, and indices are primarily characterized by their high heterogeneity. Thus, the objective is to provide a comprehensive overview on existing instruments on the basis of a systematic literature review’.  Medline: 1 January 1960 to 31 August 2009  1120  39  Acceptable  Huntley et al.16  ‘The aims of this review were (1) to identify and describe measures of multimorbidity that are most suitable for use in research in primary care and community populations, taking into account the data and resources they require and (2) to investigate the validity of these measures in terms of whether they have demonstrated anticipated associations with patient characteristics, process measures, and health outcomes.’  Medline and Embase: database inception to December 2009  11 191  194  Acceptable  Le Reste et al.20  ‘What are the criteria for multimorbidity found in the scientific medical literature and what definition could be produced with these criteria?’  PubMed, Embase and Cochrane: 1 January 1990 to 31 December 2010  416  54  Acceptable  Yurkovich et al.21  ‘To conduct a systematic review of studies reporting on the development or validation of comorbidity indices using administrative health data and compare their ability to predict outcomes related to comorbidity (i.e., construct validity).’  Medline and Embase: 1946 to September 2012  955  76  Acceptable  Willadsen et al.22  ‘Objective is to explore how multimorbidity is defined in the scientific literature, with a focus on the roles of diseases, risk factors, and symptoms in the definitions’.  PubMed, Medline and Embase: inception to 4 October 2013. Cochrane database: inception to 10 October 2013  943  163  Acceptable  Review reference  Stated aim of the review  Databases and dates of search undertaken by the review  Total titles/ abstracts screened  Total texts included  Quality assessmenta  De Groot et al.18  ‘Which methods are available for measuring comorbidity that can be used in RCTs and prognostic studies’  Medline: January 1966 to September 2000. Embase: January 1988 to September 2000.  Not reported  Not reported  Low quality  Diederichs et al.19  ‘Multimorbidity, defined as the coexistence of 2 or more chronic diseases, is a common phenomenon especially in older people. Numerous efforts to establish a standardized instrument to assess the level of multimorbidity have failed until now, and indices are primarily characterized by their high heterogeneity. Thus, the objective is to provide a comprehensive overview on existing instruments on the basis of a systematic literature review’.  Medline: 1 January 1960 to 31 August 2009  1120  39  Acceptable  Huntley et al.16  ‘The aims of this review were (1) to identify and describe measures of multimorbidity that are most suitable for use in research in primary care and community populations, taking into account the data and resources they require and (2) to investigate the validity of these measures in terms of whether they have demonstrated anticipated associations with patient characteristics, process measures, and health outcomes.’  Medline and Embase: database inception to December 2009  11 191  194  Acceptable  Le Reste et al.20  ‘What are the criteria for multimorbidity found in the scientific medical literature and what definition could be produced with these criteria?’  PubMed, Embase and Cochrane: 1 January 1990 to 31 December 2010  416  54  Acceptable  Yurkovich et al.21  ‘To conduct a systematic review of studies reporting on the development or validation of comorbidity indices using administrative health data and compare their ability to predict outcomes related to comorbidity (i.e., construct validity).’  Medline and Embase: 1946 to September 2012  955  76  Acceptable  Willadsen et al.22  ‘Objective is to explore how multimorbidity is defined in the scientific literature, with a focus on the roles of diseases, risk factors, and symptoms in the definitions’.  PubMed, Medline and Embase: inception to 4 October 2013. Cochrane database: inception to 10 October 2013  943  163  Acceptable  Notes: SIGN, Scottish Intercollegiate Guidelines Netwok; RCT, randomized controlled trial. a Based upon SIGN categories. Figure 1 View largeDownload slide Flow chart of search strategy Figure 1 View largeDownload slide Flow chart of search strategy Definitions The multimorbidity definitions used in the included reviews are in table 2. As described earlier, Le Reste and Willadsen were the only papers focused on reviewing definitions20,22 and so the definitions provided by the other four were the authors own. Table 2 Multimorbidity definitions from included reviews Review reference  Definition given a prioria or as a result of evidence review  Definition  De Groot et al.18  a priori  ‘The co-occurrence of multiple chronic or acute diseases and medical conditions in one person’  Diederichs et al.19  a priori  ‘Multimorbidity describes “the coexistence of two or more chronic diseases” in the same individual.’  Huntley et al.,14  a priori  ‘The co-occurrence of multiple diseases or medical conditions within 1 person’.  Le Reste et al.20  Review of evidence  ‘Multimorbidity is defined as any combination of chronic disease with at least one other disease (acute or chronic) or biopsychosocial factor (associated or not) or somatic risk factor. Any biopsychosocial factor, any somatic risk factor, the social network, the burden of diseases, the health care consumption, and the patient’s coping strategies may function as modifiers (of the effects of multimorbidity). Multimorbidity may modify the health outcomes and lead to an increased disability or a decreased quality of life or frailty.’  Yurkovich et al.21  a priori  This review used the definition of comorbidity: ‘Comorbidity may be defined as the total burden of illnesses unrelated to the principal diagnosis’  Willadsen et al.22  Review of evidence  Provides no single definition. Conclusion: -Existing definitions (consisting mainly of diseases) are ‘more usable for epidemiologists than for clinicians and patients’. -Recommends definition by Le Reste et al. (above)  Review reference  Definition given a prioria or as a result of evidence review  Definition  De Groot et al.18  a priori  ‘The co-occurrence of multiple chronic or acute diseases and medical conditions in one person’  Diederichs et al.19  a priori  ‘Multimorbidity describes “the coexistence of two or more chronic diseases” in the same individual.’  Huntley et al.,14  a priori  ‘The co-occurrence of multiple diseases or medical conditions within 1 person’.  Le Reste et al.20  Review of evidence  ‘Multimorbidity is defined as any combination of chronic disease with at least one other disease (acute or chronic) or biopsychosocial factor (associated or not) or somatic risk factor. Any biopsychosocial factor, any somatic risk factor, the social network, the burden of diseases, the health care consumption, and the patient’s coping strategies may function as modifiers (of the effects of multimorbidity). Multimorbidity may modify the health outcomes and lead to an increased disability or a decreased quality of life or frailty.’  Yurkovich et al.21  a priori  This review used the definition of comorbidity: ‘Comorbidity may be defined as the total burden of illnesses unrelated to the principal diagnosis’  Willadsen et al.22  Review of evidence  Provides no single definition. Conclusion: -Existing definitions (consisting mainly of diseases) are ‘more usable for epidemiologists than for clinicians and patients’. -Recommends definition by Le Reste et al. (above)  a a priori indicates this is the reviewers own definition. Table 2 Multimorbidity definitions from included reviews Review reference  Definition given a prioria or as a result of evidence review  Definition  De Groot et al.18  a priori  ‘The co-occurrence of multiple chronic or acute diseases and medical conditions in one person’  Diederichs et al.19  a priori  ‘Multimorbidity describes “the coexistence of two or more chronic diseases” in the same individual.’  Huntley et al.,14  a priori  ‘The co-occurrence of multiple diseases or medical conditions within 1 person’.  Le Reste et al.20  Review of evidence  ‘Multimorbidity is defined as any combination of chronic disease with at least one other disease (acute or chronic) or biopsychosocial factor (associated or not) or somatic risk factor. Any biopsychosocial factor, any somatic risk factor, the social network, the burden of diseases, the health care consumption, and the patient’s coping strategies may function as modifiers (of the effects of multimorbidity). Multimorbidity may modify the health outcomes and lead to an increased disability or a decreased quality of life or frailty.’  Yurkovich et al.21  a priori  This review used the definition of comorbidity: ‘Comorbidity may be defined as the total burden of illnesses unrelated to the principal diagnosis’  Willadsen et al.22  Review of evidence  Provides no single definition. Conclusion: -Existing definitions (consisting mainly of diseases) are ‘more usable for epidemiologists than for clinicians and patients’. -Recommends definition by Le Reste et al. (above)  Review reference  Definition given a prioria or as a result of evidence review  Definition  De Groot et al.18  a priori  ‘The co-occurrence of multiple chronic or acute diseases and medical conditions in one person’  Diederichs et al.19  a priori  ‘Multimorbidity describes “the coexistence of two or more chronic diseases” in the same individual.’  Huntley et al.,14  a priori  ‘The co-occurrence of multiple diseases or medical conditions within 1 person’.  Le Reste et al.20  Review of evidence  ‘Multimorbidity is defined as any combination of chronic disease with at least one other disease (acute or chronic) or biopsychosocial factor (associated or not) or somatic risk factor. Any biopsychosocial factor, any somatic risk factor, the social network, the burden of diseases, the health care consumption, and the patient’s coping strategies may function as modifiers (of the effects of multimorbidity). Multimorbidity may modify the health outcomes and lead to an increased disability or a decreased quality of life or frailty.’  Yurkovich et al.21  a priori  This review used the definition of comorbidity: ‘Comorbidity may be defined as the total burden of illnesses unrelated to the principal diagnosis’  Willadsen et al.22  Review of evidence  Provides no single definition. Conclusion: -Existing definitions (consisting mainly of diseases) are ‘more usable for epidemiologists than for clinicians and patients’. -Recommends definition by Le Reste et al. (above)  a a priori indicates this is the reviewers own definition. Le Reste produced a new multimorbidity definition as a result of their review: “… any combination of chronic disease with at least one other disease (acute or chronic) or biopsychosocial factor (associated or not) or somatic risk factor”.20 Willadsen found that more than a third of studies used a cut-off of two or more conditions to define multimorbidity, another third did not specify any cut-off and the remainder had varying cut-off points. The authors found that less than a third of their included studies used an existing definition of multimorbidity. Additionally, definitions varied according to whether or not they specified a duration of condition (e.g. ‘occurrence in the last 5 years’ or having lasted ‘for at least 3 months’) and whether or not they specified the severity of the condition (e.g. staging of the disease). The authors state that consideration of whether included diseases clustered together was considered in only ‘a few’ articles and there was little consideration of complications of diseases. The authors concluded that the majority of existing definitions are ‘more usable for epidemiologists than for clinicians and patients’ and recommended the Le Reste definition due to its comprehensive nature for including more than just disease.22 In the remaining reviews, De Groot and Yurkovich primarily used the term ‘comorbidity’.18,21 The consensus amongst all four was that multimorbidity is the occurrence of multiple diseases or conditions. Diederichs specified that multimorbidity is two or more chronic conditions.19 Measures Commonly used measures Le Reste did not focus on multimorbidity measures.20 The measures covered by the remaining five reviews are in table 3. While the stated aim of Willadsen was to ‘explore how multimorbidity is defined in the scientific literature’, there was overlap between definitions and measures.22 Table 3 Multimorbidity measures, conditions and data sources recommended by included review Review reference  Measures included  MM measure recommended?  Rationale for MM measure recommended  Specific MM conditions recommended?  MM data sources recommended?  De Groot et al.18  Disease counts and 12 weighted measures (Burden of disease index, Charlson index, CIRS, Cornoni-Huntley index, Duke Severity of Illness index Hallstrom index, Hurwitz index, ICED, Incalzi index, Kaplan index, Lui index Shwartz index)  Concludes Charlson, CIRS, ICED and Kaplan are valid and reliable methods to measure comorbidity in clinical research.  Validity and reliability  No  No specific recommendation. Commonly used methods to obtain data in included studies: ‘interviews, questionnaires, physical examinations, medical chart reviews and coded databases’.  Diederichs et al.19  Weighted indices: Charlson Index, Comorbidity Symptom Scale, Seattle Index of Comorbidity, Medication-Based Disease Burden Index, KoMo Score, Index of Coexisting Diseases, Functional Comorbidity Index, Incalzi Index, Kaplan-Feinstein Index, Physiologic Index of Comorbidity, Geriatric Index of Comorbidity, Self-Administered Comorbidity Questionnaire, Shwartz Index and Chronic Disease Score.  Recommends a disease count of 11 conditions. Found most studies did not specify criteria for selection of diseases. If criteria given: high prevalence of the disease, using other indices as a reference point for the selection of disease, conditions which are associated with an increased mortality risk, conditions associated with impact on function and health and the need for management.  Disease count based on conditions which are the 20 most frequently listed diagnoses for people aged greater than or equal to 65 years in three data sources in Germany (the inpatient sector, the outpatient sector and mortality statistics).  Cancer, diabetes, depression, hypertension, MI, chronic ischaemic heart disease, heart arrythmias, heart insufficiency, stroke, COPD, arthritis.  No specific recommendation. Included studies used: patient self-report, physician reports, clinical examinations, medical records, and administrative data. Gives advice on self-report: 'use disease specifications that can be distinguishable by lay persons [in order to increase validity of self-report].  Huntley et al.16  Disease counts and weighted measures. Most common measures (n studies): disease count (98), Charlson (38), ACG System (25), CIRS (10), Duke Severity of Illness (6).  Care utilization: the ACG System, Charlson index, or disease counts Costs: The ACG System Mortality: Charlson index Quality of life: Disease counts or Charlson index Other outcomes: Simple counts of diseases or medications  Recommendations based upon purpose of study and evidence base behind measures used for that purpose.  No  No specific recommendation. Commonly used measures: interviews, questionnaires, physical examinations, medical chart reviews, and coded databases.  Yurkovich et al.21  Administrative data measures (n studies): Charlson and its adaptations (35); Elixhauser (2); Fleming et al. index (1); Abildstrom et al. index (1) Medication- based indices: Chronic Disease Score (9), Rx-Risk (3) and Medication Based Disease Burden Index (2)  Diagnosis-based measures, (particularly Elixhauser and the Romano adaptation of the Charlson) resulted in higher ability to predict mortality outcomes. Medication-based indices, (such as the Chronic Disease Score) demonstrated better performance for predicting health care utilization.  Recommends selection of measure to be based on ‘type of data available, the study population, and the specific outcome of interest in the study.’  No  No specific recommendation. Review was limited to administrative data indices only but the authors commented on included studies which compared data sources (all were Charlson studies): two studies found self-report and administrative data had similar ability to ‘predict various outcomes’ One review and two studies found poor agreement between case note review and administrative data  Willadsen et al.22  Charlson, Clinical Classification Software, CIRS, ACG, Aggregated Diagnosis Groups, Medication based, Expanded Diagnosis Clusters, Resource Utilization Bands, The Functional Comorbidity Index, ICED, QoF, The Registration Network Family Practices  Does not recommend a single measure. As documented in table 2, the authors state the importance of including risk factors and symptoms and severity as well as diseases if want a clinically relevant definition (and thus measure)  N/A  No  No specific recommendation. The included studies used data from administrative data and self-report  Review reference  Measures included  MM measure recommended?  Rationale for MM measure recommended  Specific MM conditions recommended?  MM data sources recommended?  De Groot et al.18  Disease counts and 12 weighted measures (Burden of disease index, Charlson index, CIRS, Cornoni-Huntley index, Duke Severity of Illness index Hallstrom index, Hurwitz index, ICED, Incalzi index, Kaplan index, Lui index Shwartz index)  Concludes Charlson, CIRS, ICED and Kaplan are valid and reliable methods to measure comorbidity in clinical research.  Validity and reliability  No  No specific recommendation. Commonly used methods to obtain data in included studies: ‘interviews, questionnaires, physical examinations, medical chart reviews and coded databases’.  Diederichs et al.19  Weighted indices: Charlson Index, Comorbidity Symptom Scale, Seattle Index of Comorbidity, Medication-Based Disease Burden Index, KoMo Score, Index of Coexisting Diseases, Functional Comorbidity Index, Incalzi Index, Kaplan-Feinstein Index, Physiologic Index of Comorbidity, Geriatric Index of Comorbidity, Self-Administered Comorbidity Questionnaire, Shwartz Index and Chronic Disease Score.  Recommends a disease count of 11 conditions. Found most studies did not specify criteria for selection of diseases. If criteria given: high prevalence of the disease, using other indices as a reference point for the selection of disease, conditions which are associated with an increased mortality risk, conditions associated with impact on function and health and the need for management.  Disease count based on conditions which are the 20 most frequently listed diagnoses for people aged greater than or equal to 65 years in three data sources in Germany (the inpatient sector, the outpatient sector and mortality statistics).  Cancer, diabetes, depression, hypertension, MI, chronic ischaemic heart disease, heart arrythmias, heart insufficiency, stroke, COPD, arthritis.  No specific recommendation. Included studies used: patient self-report, physician reports, clinical examinations, medical records, and administrative data. Gives advice on self-report: 'use disease specifications that can be distinguishable by lay persons [in order to increase validity of self-report].  Huntley et al.16  Disease counts and weighted measures. Most common measures (n studies): disease count (98), Charlson (38), ACG System (25), CIRS (10), Duke Severity of Illness (6).  Care utilization: the ACG System, Charlson index, or disease counts Costs: The ACG System Mortality: Charlson index Quality of life: Disease counts or Charlson index Other outcomes: Simple counts of diseases or medications  Recommendations based upon purpose of study and evidence base behind measures used for that purpose.  No  No specific recommendation. Commonly used measures: interviews, questionnaires, physical examinations, medical chart reviews, and coded databases.  Yurkovich et al.21  Administrative data measures (n studies): Charlson and its adaptations (35); Elixhauser (2); Fleming et al. index (1); Abildstrom et al. index (1) Medication- based indices: Chronic Disease Score (9), Rx-Risk (3) and Medication Based Disease Burden Index (2)  Diagnosis-based measures, (particularly Elixhauser and the Romano adaptation of the Charlson) resulted in higher ability to predict mortality outcomes. Medication-based indices, (such as the Chronic Disease Score) demonstrated better performance for predicting health care utilization.  Recommends selection of measure to be based on ‘type of data available, the study population, and the specific outcome of interest in the study.’  No  No specific recommendation. Review was limited to administrative data indices only but the authors commented on included studies which compared data sources (all were Charlson studies): two studies found self-report and administrative data had similar ability to ‘predict various outcomes’ One review and two studies found poor agreement between case note review and administrative data  Willadsen et al.22  Charlson, Clinical Classification Software, CIRS, ACG, Aggregated Diagnosis Groups, Medication based, Expanded Diagnosis Clusters, Resource Utilization Bands, The Functional Comorbidity Index, ICED, QoF, The Registration Network Family Practices  Does not recommend a single measure. As documented in table 2, the authors state the importance of including risk factors and symptoms and severity as well as diseases if want a clinically relevant definition (and thus measure)  N/A  No  No specific recommendation. The included studies used data from administrative data and self-report  Note: MM, multimorbidity; CIRS, Cumulative illness rating scale; ICED, Index of Coexistent disease; MI, myocardial infarction; COPD, chronic obstructive pulmonary disease; ACG, Adjusted Clinical Groups; QoF, Quality Outcomes Framework; N/A, not applicable. Table 3 Multimorbidity measures, conditions and data sources recommended by included review Review reference  Measures included  MM measure recommended?  Rationale for MM measure recommended  Specific MM conditions recommended?  MM data sources recommended?  De Groot et al.18  Disease counts and 12 weighted measures (Burden of disease index, Charlson index, CIRS, Cornoni-Huntley index, Duke Severity of Illness index Hallstrom index, Hurwitz index, ICED, Incalzi index, Kaplan index, Lui index Shwartz index)  Concludes Charlson, CIRS, ICED and Kaplan are valid and reliable methods to measure comorbidity in clinical research.  Validity and reliability  No  No specific recommendation. Commonly used methods to obtain data in included studies: ‘interviews, questionnaires, physical examinations, medical chart reviews and coded databases’.  Diederichs et al.19  Weighted indices: Charlson Index, Comorbidity Symptom Scale, Seattle Index of Comorbidity, Medication-Based Disease Burden Index, KoMo Score, Index of Coexisting Diseases, Functional Comorbidity Index, Incalzi Index, Kaplan-Feinstein Index, Physiologic Index of Comorbidity, Geriatric Index of Comorbidity, Self-Administered Comorbidity Questionnaire, Shwartz Index and Chronic Disease Score.  Recommends a disease count of 11 conditions. Found most studies did not specify criteria for selection of diseases. If criteria given: high prevalence of the disease, using other indices as a reference point for the selection of disease, conditions which are associated with an increased mortality risk, conditions associated with impact on function and health and the need for management.  Disease count based on conditions which are the 20 most frequently listed diagnoses for people aged greater than or equal to 65 years in three data sources in Germany (the inpatient sector, the outpatient sector and mortality statistics).  Cancer, diabetes, depression, hypertension, MI, chronic ischaemic heart disease, heart arrythmias, heart insufficiency, stroke, COPD, arthritis.  No specific recommendation. Included studies used: patient self-report, physician reports, clinical examinations, medical records, and administrative data. Gives advice on self-report: 'use disease specifications that can be distinguishable by lay persons [in order to increase validity of self-report].  Huntley et al.16  Disease counts and weighted measures. Most common measures (n studies): disease count (98), Charlson (38), ACG System (25), CIRS (10), Duke Severity of Illness (6).  Care utilization: the ACG System, Charlson index, or disease counts Costs: The ACG System Mortality: Charlson index Quality of life: Disease counts or Charlson index Other outcomes: Simple counts of diseases or medications  Recommendations based upon purpose of study and evidence base behind measures used for that purpose.  No  No specific recommendation. Commonly used measures: interviews, questionnaires, physical examinations, medical chart reviews, and coded databases.  Yurkovich et al.21  Administrative data measures (n studies): Charlson and its adaptations (35); Elixhauser (2); Fleming et al. index (1); Abildstrom et al. index (1) Medication- based indices: Chronic Disease Score (9), Rx-Risk (3) and Medication Based Disease Burden Index (2)  Diagnosis-based measures, (particularly Elixhauser and the Romano adaptation of the Charlson) resulted in higher ability to predict mortality outcomes. Medication-based indices, (such as the Chronic Disease Score) demonstrated better performance for predicting health care utilization.  Recommends selection of measure to be based on ‘type of data available, the study population, and the specific outcome of interest in the study.’  No  No specific recommendation. Review was limited to administrative data indices only but the authors commented on included studies which compared data sources (all were Charlson studies): two studies found self-report and administrative data had similar ability to ‘predict various outcomes’ One review and two studies found poor agreement between case note review and administrative data  Willadsen et al.22  Charlson, Clinical Classification Software, CIRS, ACG, Aggregated Diagnosis Groups, Medication based, Expanded Diagnosis Clusters, Resource Utilization Bands, The Functional Comorbidity Index, ICED, QoF, The Registration Network Family Practices  Does not recommend a single measure. As documented in table 2, the authors state the importance of including risk factors and symptoms and severity as well as diseases if want a clinically relevant definition (and thus measure)  N/A  No  No specific recommendation. The included studies used data from administrative data and self-report  Review reference  Measures included  MM measure recommended?  Rationale for MM measure recommended  Specific MM conditions recommended?  MM data sources recommended?  De Groot et al.18  Disease counts and 12 weighted measures (Burden of disease index, Charlson index, CIRS, Cornoni-Huntley index, Duke Severity of Illness index Hallstrom index, Hurwitz index, ICED, Incalzi index, Kaplan index, Lui index Shwartz index)  Concludes Charlson, CIRS, ICED and Kaplan are valid and reliable methods to measure comorbidity in clinical research.  Validity and reliability  No  No specific recommendation. Commonly used methods to obtain data in included studies: ‘interviews, questionnaires, physical examinations, medical chart reviews and coded databases’.  Diederichs et al.19  Weighted indices: Charlson Index, Comorbidity Symptom Scale, Seattle Index of Comorbidity, Medication-Based Disease Burden Index, KoMo Score, Index of Coexisting Diseases, Functional Comorbidity Index, Incalzi Index, Kaplan-Feinstein Index, Physiologic Index of Comorbidity, Geriatric Index of Comorbidity, Self-Administered Comorbidity Questionnaire, Shwartz Index and Chronic Disease Score.  Recommends a disease count of 11 conditions. Found most studies did not specify criteria for selection of diseases. If criteria given: high prevalence of the disease, using other indices as a reference point for the selection of disease, conditions which are associated with an increased mortality risk, conditions associated with impact on function and health and the need for management.  Disease count based on conditions which are the 20 most frequently listed diagnoses for people aged greater than or equal to 65 years in three data sources in Germany (the inpatient sector, the outpatient sector and mortality statistics).  Cancer, diabetes, depression, hypertension, MI, chronic ischaemic heart disease, heart arrythmias, heart insufficiency, stroke, COPD, arthritis.  No specific recommendation. Included studies used: patient self-report, physician reports, clinical examinations, medical records, and administrative data. Gives advice on self-report: 'use disease specifications that can be distinguishable by lay persons [in order to increase validity of self-report].  Huntley et al.16  Disease counts and weighted measures. Most common measures (n studies): disease count (98), Charlson (38), ACG System (25), CIRS (10), Duke Severity of Illness (6).  Care utilization: the ACG System, Charlson index, or disease counts Costs: The ACG System Mortality: Charlson index Quality of life: Disease counts or Charlson index Other outcomes: Simple counts of diseases or medications  Recommendations based upon purpose of study and evidence base behind measures used for that purpose.  No  No specific recommendation. Commonly used measures: interviews, questionnaires, physical examinations, medical chart reviews, and coded databases.  Yurkovich et al.21  Administrative data measures (n studies): Charlson and its adaptations (35); Elixhauser (2); Fleming et al. index (1); Abildstrom et al. index (1) Medication- based indices: Chronic Disease Score (9), Rx-Risk (3) and Medication Based Disease Burden Index (2)  Diagnosis-based measures, (particularly Elixhauser and the Romano adaptation of the Charlson) resulted in higher ability to predict mortality outcomes. Medication-based indices, (such as the Chronic Disease Score) demonstrated better performance for predicting health care utilization.  Recommends selection of measure to be based on ‘type of data available, the study population, and the specific outcome of interest in the study.’  No  No specific recommendation. Review was limited to administrative data indices only but the authors commented on included studies which compared data sources (all were Charlson studies): two studies found self-report and administrative data had similar ability to ‘predict various outcomes’ One review and two studies found poor agreement between case note review and administrative data  Willadsen et al.22  Charlson, Clinical Classification Software, CIRS, ACG, Aggregated Diagnosis Groups, Medication based, Expanded Diagnosis Clusters, Resource Utilization Bands, The Functional Comorbidity Index, ICED, QoF, The Registration Network Family Practices  Does not recommend a single measure. As documented in table 2, the authors state the importance of including risk factors and symptoms and severity as well as diseases if want a clinically relevant definition (and thus measure)  N/A  No  No specific recommendation. The included studies used data from administrative data and self-report  Note: MM, multimorbidity; CIRS, Cumulative illness rating scale; ICED, Index of Coexistent disease; MI, myocardial infarction; COPD, chronic obstructive pulmonary disease; ACG, Adjusted Clinical Groups; QoF, Quality Outcomes Framework; N/A, not applicable. The measures included by reviews encompassed disease counts and weighted indices such as the Charlson Index, the Cumulative Illness Rating Scale (CIRS), the Index of Coexistent Disease (ICED), the Adjusted Clinical Groups (ACG) System and the Duke Severity of Illness. Yurkovich and Huntley examined the frequency of measures. Yurkovich categorized measures as ‘administrative data’ (the most common being Charlson) and ‘medication-based’ (the most common being the Chronic Disease Score).21 Huntley categorized the most common measures as: disease counts, the Charlson index and variations, the ACG system, the CIRS and the Duke Severity Illness Check-list System.16 Despite the name, disease counts included more than just diseases (e.g. they included categories of conditions). The authors found disease counts being used in 98 studies and the number of disease ‘items’ included within counts ranged from 9 to 35.16 Willadsen found that measures included by their papers contained conditions ranging in number from 4 to 147.22 Recommended measures Yurkovich found that diagnosis-based measures such as the Elixhauser index and the Romano adaptation of the Charlson index were best able to predict mortality outcomes while the medication-based Chronic Disease Score was best able to predict health care use.21 Huntley recommended that researchers select a measure for a study based upon the measure validated for use in that scenario, for example, the Charlson index for predicting mortality. The authors also state that simple counts of diseases or medications perform almost as effectively as complex measures in predicting most outcomes.16 De Groot assessed the content, criterion and construct validity of measures. They concluded that the Charlson, CIRS, ICED and Kaplan indices are valid and reliable methods for use in clinical research but that other measures (such as disease counts) were more difficult to assess due to limited data.18 Willadsen did not recommend a single measure and instead, as described previously, stated the importance of including risk factors, symptoms and severity of diseases.22 Diederichs also did not recommend a single measure. They found studies of disease counts often did not specify the criteria for the selection of diseases, but if criteria were given these were: high prevalence of the disease, using other indices as a reference point for the selection of disease, or high impact conditions in terms of increased mortality risk, an impact on function and health and the need for management. They recommended 11 conditions selected on the basis of being the most common causes of inpatient and outpatient attendance as well as death in people aged over 64 in Germany. The conditions included cancer, depression, myocardial infarction and hypertension.19 Data sources All five reviews found patient self-report, physician reports, clinical examinations, medical record reviews and administrative data (‘coded databases’ or ‘routine data’) were common sources of multimorbidity data among their included studies.16,18,19,21,22 No review studied whether any source was superior, although Yurkovich found evidence that the Charlson index derived from self-report and that derived from administrative data had similar abilities to ‘predict various outcomes’.21 De Groot stated that medical chart reviews are preferable for use in smaller studies as they likely yield the most complete data but that this is likely impractical in larger studies and so administrative databases can be used.18 Similarly, Huntley noted that administrative data have the advantage of ease of use but may be limited by data quality issues.16 Discussion Summary of findings Our review pooled the findings of six systematic reviews. We found heterogeneity of multimorbidity definitions and measures, but there were a number of commonalities. Most reviews defined multimorbidity as the occurrence of multiple diseases or conditions, the most common cut-off being two or more. Le Reste produced a new definition that encompassed biopsychosocial factors and somatic risk factors along with disease.20 This was recommended by Willadsen as being the most clinically relevant definition of multimorbidity available.22 Common measures included the Charlson, CIRS, ICED, Kaplan, the ACG system and disease counts, with advice that measures be selected based upon the purpose of a particular study.16,18 No reviews made recommendations about the most appropriate data sources to use when measuring multimorbidity. Strengths and limitations Our systematic review provides a high-level summary of both the definition and measurement of multimorbidity in relevant systematic reviews. Ours is the first to focus upon those reviews which primarily aimed to examine multimorbidity definitions or measures. This is important given the heterogeneity in definitions and measures available and the associated complexity in developing consensus. We acknowledge that reviews such as that by Fortin et al. (of prevalence studies of multimorbidity)2 and Marengoni et al.5 (of ageing and multimorbidity) discuss recommended definitions and measures at the end of their reviews, but we have not included these as their primary aim did not meet our inclusion criteria. A limitation is that search terms were limited to the title only for practical reasons which means some relevant reviews could be missed. We conducted a test search including these terms in the abstract or full text which revealed no additional reviews in the first 100 titles screened. Additionally, as recommended by PRISMA, systematic reviews should be identified as such in the title.15 One of the included reviews (examining measures of multimorbidity) was classed as low quality. However, as there were three other reviews examining multimorbidity measures this should reduce the likelihood that this affected our findings. Comparison with literature Our findings are consistent with other systematic reviewers who have encountered challenges due to the lack of a common approach towards measuring and defining multimorbidity.2,23–25 Definitions Willadsen highlighted that many definitions and measures seem to be tailored towards use in research rather than being clinically relevant.22 It is true that traditional approaches, for example, measuring multimorbidity using the Charlson or disease counts, do not capture the holistic experience of multimorbidity. For example, we know that an individuals’ ability to cope with disease is influenced by both person factors and wider socio-environmental factors and that at a population level, multimorbidity is associated with higher levels of deprivation.4,26–30 The definition by Le Reste is more likely to capture this complexity but the multi-faceted nature of the definition makes it difficult to operationalize in practice. Instead of adding further elements to the definition and measurement of multimorbidity, it is perhaps more appropriate to ensure there is consideration of its holistic nature when studying its determinants and outcomes and when managing it clinically. This would include understanding its relationship with health inequalities in areas of high deprivation, as well as to frailty and the ageing process.11,12 The cut-off point regarding the minimum number of conditions to equate to being multimorbid needs further consideration. The most common cut-off point found by our reviews was two or more conditions and this was consistent with the findings of Fortin in their review of prevalence studies of multimorbidity.2 The prevalence of multimorbidity is inevitably affected by the cut-off selected and additionally it is likely that a higher cut-off would select a patient group with a higher burden of multimorbidity.2 This needs further research, for example, by testing the number of conditions which best identify patients at higher risk of outcomes such as hospital stay, disability, frailty or mortality. Measures When multimorbidity is defined and measured on the basis of a count of conditions the measurement of multimorbidity is closely linked to the definition. We have used the term ‘disease counts’ as this is the common phrase used in the literature, but acknowledge these measures can include a wider spectrum of health conditions (e.g. risk factors for disease). Disease counts are likely more appropriate for scenarios where multiple outcomes are being considered or in which no single weighted measure has been validated.16 They may also be a more intuitive summary of multimorbidity burden in patients, for example, when showing the link between multimorbidity and socioeconomic status.4 Additionally, reviews have found that multimorbidity may be more appropriately considered as different common clusters of conditions and this is easier to measure using counts.24,31 If researchers are selecting conditions to include in a count the purpose of the work being conducted must be considered. Some conditions, for example, depression, may have greater impact upon patients in terms of quality of life or function.32 Other conditions such as heart disease may impact more upon health services in terms of number of admissions or treatment costs.2,19 In studies using weighted measures the definition and measurement of multimorbidity are more distinct. Many weighted measures were originally developed as comorbidity measures but are increasingly being used as multimorbidity measures.18 Weighted measures, if used for appropriate outcomes, can assist in predicting patient outcome and future healthcare usage and can also provide an assessment of the burden of multimorbidity experienced by the patient, their carers or health and social care services.33 Therefore, where the aim is to examine outcomes in patients and to account for the presence of multiple conditions, a validated weighted measure may be more appropriate or informative than a disease count. Data sources No review recommended a particular data source to measure multimorbidity. In the wider literature, a number of studies and reviews have compared data sources for comorbidity and multimorbidity measures, often with conflicting findings.3,34 The availability of data and the resource implications will additionally affect the choice of data used. For example, while case-note review is viewed as being more complete than administrative data as it is more resource intensive.3,34 Another important data source is patient self-report, which may be more likely to capture conditions which may not be seen as important clinically but impact on function or quality of life.32 Regardless of measure, different data sources will affect the prevalence of multimorbidity.2,35 Implications for research and practice Our key recommendation is that researchers be explicit about the definitions and measure(s) they are using and give a rationale for their choice. This will enable comparison of findings across different settings and outcomes as well as progress the evidence base regarding the most appropriate definitions and measures for particular scenarios. Multimorbidity is an important public health challenge, which is influenced strongly by wider social and environmental factors. In this review, the paper by Le Reste highlighted the holistic nature of multimorbidity.20 In clinical and public health practice, holistic approaches that take into account more than just the medical management of disease could assist with reducing its impact. However, there is a need for more evidence on the effectiveness of primary care and community-based interventions, including those tackling the challenges experienced by individuals with socio-economic deprivation.36 Despite this, recent research in primary care in deprived areas has shown that a co-development model of intervention development for multimorbidity (CARE Plus) was feasible and may be cost-effective, thus pointing to future directions in reducing the burden of multimorbidity.37,38 Overall, a definition of multiple co-existing conditions is reasonable and a cut-off should be explicitly defined. Researchers would be consistent with others using a cut-off of two or more. Using a weighted measure validated for the outcome being considered is advised, but where evidence is weak or where multiple outcomes or populations are being considered, the use of disease counts is appropriate. There is precedence for the inclusion of conditions other than solely chronic disease in a multimorbidity measure but a rationale for included and excluded conditions should be given. Supplementary data Supplementary data are available at EURPUB online. Funding Dr Marjorie Johnston was funded by a Clinical Academic Fellowship from the Chief Scientist Office, Scotland (CAF/13/03), was affiliated with the Farr Institute of Health Informatics and Research Scotland and was an honorary Public Health Registrar at NHS Grampian. Conflicts of interest: None declared. Key points To improve consensus in defining and measuring multimorbidity, we recommend researchers and practitioners be explicit about the definitions and measure(s) they are using and give a rationale for their choice. We conclude that multimorbidity is the coexistence of multiple conditions (most commonly defined as two or more conditions). Validated multimorbidity measures for particular scenarios should be chosen if these exist. Where there is no validated measure or where multiple outcomes or populations are being considered, disease counts are appropriate. References 1 Uijen AA, Van de Lisdon EH. Multimorbidity in primary care: prevalence and trend over the last 20 years. Eur J Gen Pract  2008; 14: 28– 32. Google Scholar CrossRef Search ADS PubMed  2 Fortin M, Stewart M, Poitras M, et al.   A systematic review of prevalence studies on multimorbidity: toward a more uniform methodology. Ann Fam Med  2012; 10: 142– 51. Google Scholar CrossRef Search ADS PubMed  3 Leal JR, Laupland KB. Validity of ascertainment of co-morbid illness using administrative databases: a systematic review. Clin Microbiol Infect  2010; 16: 715– 21. Google Scholar CrossRef Search ADS PubMed  4 Barnett K, Mercer SW, Norbury M, et al.   Epidemiology of multimorbidity and implications for health care, research and medical education: a cross-sectional study. Lancet  2012; 380: 37– 43. Google Scholar CrossRef Search ADS PubMed  5 Marengoni A, Angleman S, Melis R, et al.   Aging with multimorbidity: a systematic review of the literature. Ageing Res Rev  2011; 10: 430– 9. Google Scholar CrossRef Search ADS PubMed  6 Wallace E, Salisbury C, Guthrie B, et al.   Managing patients with multimorbidity in primary care. BMJ  2015; 350: h176. Google Scholar CrossRef Search ADS PubMed  7 Farmer C, Fenu E, O'Flynn N, Guthrie B. Clinical assessment and management of multimorbidity: summary of NICE guideline. BMJ  2016; 354: i4843. Google Scholar CrossRef Search ADS PubMed  8 Hughes LD, McMurdo MET, Guthrie B. Guidelines for people not for diseases: the challenges of applying UK clinical guidelines to people with multimorbidity. Age Ageing  2013; 42: 62– 9. Google Scholar CrossRef Search ADS PubMed  9 Calderón-Larrañaga A, Poblador-Plou B, González-Rubio F, et al.   Multimorbidity, polypharmacy, referrals, and adverse drug events: are we doing things well? Br J Gen Pract  2012; 62: e821– 6. Google Scholar CrossRef Search ADS PubMed  10 McDaid O, Normand C, Kelly A, Smith S. Prevalence, patterns and healthcare burden of multimorbidity in the older irish population. Ir J Med Sci  2013; 182: S229. 11 Beard JR, Officer A, de Carvalho IA, et al.   The World report on ageing and health: a policy framework for healthy ageing. Lancet  2016; 387: 2145– 54. Google Scholar CrossRef Search ADS PubMed  12 Villacampa-Fernández P, Navarro-Pardo E, Tarín JJ, Cano A. Frailty and multimorbidity: two related yet different concepts. Maturitas  2017; 95: 31– 5. Google Scholar CrossRef Search ADS PubMed  13 Almirall J, Fortin M. The coexistence of terms to describe the presence of multiple concurrent diseases. J Comorb  2013; 3: 4– 9. Google Scholar CrossRef Search ADS PubMed  14 National Insitute for Health and Care Excellence. Multimorbidity: clinical assessment and management. 2016. Available at: https://www.nice.org.uk/guidance/ng56 (10 July 2017 date last accessed). 15 Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ  2009; 339: b2535. Google Scholar CrossRef Search ADS PubMed  16 Huntley AL, Johnson R, Purdy S, et al.   Measures of multimorbidity and morbidity burden for use in primary care and community settings: a systematic review and guide. Ann Fam Med  2012; 10: 134– 41. Google Scholar CrossRef Search ADS PubMed  17 Scottish Intercollegiate Guidelines Network. Critical appraisal: notes and checklists. 2014. Available at: http://www.sign.ac.uk/methodology/checklists.html# (20 February 2014, date last accessed). 18 De Groot V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity: a critical review of available methods. J Clin Epidemiol  2003; 56: 221– 9. Google Scholar CrossRef Search ADS PubMed  19 Diederichs C, Berger K, Bartels DB. The measurement of multiple chronic diseases- a systematic review on existing multimorbidity indices. J Gerontol A Biol Sci Med Sci  2011; 66A: 301– 11. Google Scholar CrossRef Search ADS   20 Le Reste JY, Nabbe P, Manceau B, et al.   The European General Practice Research Network presents a comprehensive definition of multimorbidity in family medicine and long term care, following a systematic review of relevant literature. J Am Med Dir Assoc  2013; 14: 319– 25. Google Scholar CrossRef Search ADS PubMed  21 Yurkovich M, Avina-Zubieta JA, Thomas J, et al.   A systematic review identifies valid comorbidity indices derived from administrative health data. J Clin Epidemiol  2015; 68: 3– 14. Google Scholar CrossRef Search ADS PubMed  22 Willadsen TG, Bebe A, Koster-Rasmussen R, et al.   The role of diseases, risk factors and symptoms in the definition of multimorbidity—a systematic review. Scand J Prim Health Care  2016; 34: 112– 21. Google Scholar CrossRef Search ADS PubMed  23 Violan C, Foguet-Boreu Q, Flores-Mateo G, et al.   Prevalence, determinants and patterns of multimorbidity in primary care: a systematic review of observational Studies. PLoS One  2014; 9: e102149. Google Scholar CrossRef Search ADS PubMed  24 Prados-Torres A, Calderon-Larranaga A, Hancco-Saavedra J, et al.   Multimorbidity patterns: a systematic review. J Clin Epidemiol  2014; 67: 254– 66. Google Scholar CrossRef Search ADS PubMed  25 Holzer BM, Siebenhuener K, Bopp M, Minder CE. Evidence-based design recommendations for prevalence studies on multimorbidity: improving comparability of estimates. Popul Health Metr  2017; 15: 9. Available at: https://doi.org/10.1186/s12963-017-0126-4. Google Scholar CrossRef Search ADS PubMed  26 Cairns JM, Curtis SE, Bambra C. Defying deprivation: a cross-sectional analysis of area level health resilience in England. Health Place  2012; 18: 928– 33. Google Scholar CrossRef Search ADS PubMed  27 Marmot M, Allen J, Goldblatt P, et al.   Fair Society, Healthy Lives: The Marmot Review. 2010; Available at: http://www.instituteofhealthequity.org/projects/fair-society-healthy-lives-the-marmot-review/fair-society-healthy-lives-full-report (1 May 2017, last date accessed). 28 Bielderman A, de Greef MH, Krijnen WP, van der Schans CP. Relationship between socioeconomic status and quality of life in older adults: a path analysis. Qual Life Res  2015; 24: 1697– 705. Google Scholar CrossRef Search ADS PubMed  29 Lawson KD, Mercer SW, Wyke S, et al.   Double trouble: the impact of multimorbidity and deprivation on preference-weighted health related quality of life a cross sectional analysis of the Scottish Health Survey. Int J Equity Health  2013; 12: 67. Google Scholar CrossRef Search ADS PubMed  30 McLean G, Gunn J, Wyke S, et al.   The influence of socioeconomic deprivation on multimorbidity at different ages: a cross-sectional study. Br J Gen Pract  2014; 64: e440– 7. Google Scholar CrossRef Search ADS PubMed  31 Sinnige J, Braspenning J, Schellevis F, et al.   The prevalence of disease clusters in older adults with multiple chronic diseases–a systematic literature review. PLoS One  2013; 8: e79641. Google Scholar CrossRef Search ADS PubMed  32 Walker V, Perret-Guillaume C, Kesse-Guyot E, et al.   Effect of multimorbidity on health-related quality of life in adults aged 55 years or older: results from the SU.VI.MAX 2 Cohort. PLoS One  2016; 11: e0169282. Google Scholar CrossRef Search ADS PubMed  33 Brilleman SL, Salisbury C. Comparing measures of multimorbidity to predict outcomes in primary care: a cross sectional study. Fam Pract  2013; 30: 172– 80. Google Scholar CrossRef Search ADS PubMed  34 Needham DM, Scales DC, Laupacis A, Pronovost PJ. A systematic review of the Charlson comorbidity index using Canadian administrative databases: a perspective on risk adjustment in critical care research. J Crit Care  2005; 20: 12– 9. Google Scholar CrossRef Search ADS PubMed  35 Mokraoui N, Haggerty J, Almirall J, Fortin M. Prevalence of self-reported multimorbidity in the general population and in primary care practices: a cross-sectional study. BMC Res Notes  2016; 9. Available at: https://doi.org/10.1186/s13104-016-2121-4. 36 Smith SM, Wallace E, O'Dowd T, Fortin M. Interventions for improving outcomes in patients with multimorbidity in primary care and community setttings. Cochrane Database Syst Rev  2016; 3: CD006560. Doi: 10.1002/14651858.CD006560.pub3. Google Scholar PubMed  37 Mercer SW, O'Brien R, Fitzpatrick B, et al.   The development and optimisation of a primary care-based whole system complex intervention (CARE Plus) for patients with multimorbidity living in areas of high socioeconomic deprivation. Chronic Illn  2016; 12: 165– 81. Google Scholar CrossRef Search ADS PubMed  38 Mercer SW, Fitzpatrick B, Guthrie B, et al.   The Care Plus study-a whole system intervention to improve quality of life of primary care patients with multimorbidity in areas of high socioeconomic deprivation: cluster randomised controlled trial. BMC Med  2016; 14: 88. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The European Journal of Public Health Oxford University Press

Defining and measuring multimorbidity: a systematic review of systematic reviews

Loading next page...
 
/lp/ou_press/defining-and-measuring-multimorbidity-a-systematic-review-of-MmzmzoIJQL
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved.
ISSN
1101-1262
eISSN
1464-360X
D.O.I.
10.1093/eurpub/cky098
Publisher site
See Article on Publisher Site

Abstract

Abstract Background Multimorbidity, the coexistence of multiple health conditions, is a growing public health challenge. Research and intervention development are hampered by the lack of consensus regarding defining and measuring multimorbidity. The aim of this systematic review was to pool the findings of systematic reviews examining definitions and measures of multimorbidity. Methods Medline, Embase, PubMed and Cochrane were searched from database inception to February 2017. Two authors independently screened titles, abstracts and full texts and extracted data from the included papers. Disagreements were resolved with a third author. Reviews were quality assessed. Results Of six reviews, two focussed on definitions and four on measures. Multimorbidity was commonly defined as the presence of multiple diseases or conditions, often with a cut-off of two or more. One review developed a holistic definition including biopsychosocial and somatic factors as well as disease. Reviews recommended using measures validated for the outcome of interest. Disease counts are an alternative if no validated measure exists. Conclusions To enable comparison between studies and settings, researchers and practitioners should be explicit about their choice of definition and measure. Using a cut-off of two or more conditions as part of the definition is widely adopted. Measure selection should be based on tools validated for the outcome being considered. Where there is no validated measure, or where multiple outcomes or populations are being considered, disease counts are appropriate. Introduction Multimorbidity is commonly understood to be the coexistence of multiple health conditions in an individual.1,2 A related term, comorbidity, describes the burden of illness co-existing with a particular disease of interest.3 Multimorbidity is a growing global public health challenge as populations age and the prevalence of long-term conditions rises.1,2 Multimorbidity is associated with poorer outcomes and the increased use of health and social care services with associated costs.4,5 There is increasing awareness that healthcare services are not adequately designed to meet the challenges of multimorbidity. Secondary care services are generally single disease focussed.6,7 Practitioners, particularly in primary care, face challenges in using clinical guidelines that are generally developed for single conditions or groups of similar conditions.8 These issues bring associated risks, for example, polypharmacy,8,9 and challenges associated with managing patients with complex needs in resource limited environments.6 Multimorbidity also places a burden on individuals who face poorer quality of life and increased disability.10 It is highly correlated with frailty (an age-related decline leading to reduced reserves of physical and mental health capacity, resulting in vulnerability to stressors and an increased risk of poor health outcomes).11,12 Despite these challenges, there is no international consensus regarding the best way to define and measure multimorbidity.13 This makes carrying out and interpreting research, comparing findings across populations and developing guidelines and interventions difficult. A review of prevalence studies of multimorbidity found estimates ranging from between less than 5% to more than 95%, often due to differences in the operational definition of multimorbidity.2 The National Institute for Health and Care Excellence (NICE) recently developed a multimorbidity guideline and commented that measuring the prevalence of multimorbidity is complex due to the varying measures being used.14 A number of reviews have summarized the multimorbidity definitions or measures used in primary studies. Our aim was to build consensus on the most appropriate ways to define and measure multimorbidity by pooling the findings of these systematic reviews. Methods The PRISMA 2009 checklist guided method development and reporting of findings.15 Medline, Embase, PubMed and the Cochrane database of systematic reviews were searched from database inception to 13 February 2017. The search strategy was comparable across all databases. At the time of searching, there was no MeSH term for multimorbidity. The search terms relating to ‘multimorbidity’ and its measures were drawn from a previous systematic review of the multimorbidity literature.16 These were combined by the Boolean operator ‘AND’ with ‘review’ as a title word. The terms were searched in the title only, as an initial trial search found that widening this to the abstract or full text significantly reduced the ability to detect relevant reviews. The search strategy is in Supplementary table S1. Systematic reviews of the multimorbidity literature which examined multimorbidity definitions and/or measures as a central focus of the review were included. While comorbidity is now commonly accepted to be distinct from multimorbidity, it is known that the terms have been used synonymously in the past. Reviews of comorbidity where no specific index disease was considered were therefore eligible. Systematic reviews that did not have the primary aim to summarize multimorbidity definitions and measures were excluded. Reviews that were ‘narrative’ or ‘semi-structured’ or which otherwise were not systematic reviews were excluded. Title, abstract and full-text screening were carried out independently by two authors (MCJ and SWM). Disagreement was resolved by CB. Primary data extraction was carried out by MCJ with four others acting as independent second reviewers (CB, MC, GJP and SWM). The data extraction form was prepared and piloted by MCJ and finalized by discussion with the other reviewers. Data extraction included the review characteristics, the definition and measures of multimorbidity presented in the review and the rationale behind any recommended measures of multimorbidity (if given). Scottish Intercollegiate Guidelines Network critical appraisal checklists17 were used to assess the quality of included reviews (‘low quality’, ‘acceptable’ or ‘high quality’). The results were combined narratively. Results Figure 1 summarizes the results of the search. Out of 1051 articles sourced during the search, there were 432 duplicates. Following screening of titles, abstracts and full texts, six reviews were included.16,18–22 The characteristics of these reviews, including their stated aims, are presented in table 1. The Le Reste and Willadsen reviews focused on the definition of multimorbidity, while the remaining four focussed on measures. The number of studies included by the reviews ranged from 39 to 194. Five reviews were of ‘acceptable quality’.16,19–22 De Groot was ‘low quality’ as they did not report the literature search strategy, the results of the literature search and the identification of papers clearly.18 Table 1 Characteristics of included reviews Review reference  Stated aim of the review  Databases and dates of search undertaken by the review  Total titles/ abstracts screened  Total texts included  Quality assessmenta  De Groot et al.18  ‘Which methods are available for measuring comorbidity that can be used in RCTs and prognostic studies’  Medline: January 1966 to September 2000. Embase: January 1988 to September 2000.  Not reported  Not reported  Low quality  Diederichs et al.19  ‘Multimorbidity, defined as the coexistence of 2 or more chronic diseases, is a common phenomenon especially in older people. Numerous efforts to establish a standardized instrument to assess the level of multimorbidity have failed until now, and indices are primarily characterized by their high heterogeneity. Thus, the objective is to provide a comprehensive overview on existing instruments on the basis of a systematic literature review’.  Medline: 1 January 1960 to 31 August 2009  1120  39  Acceptable  Huntley et al.16  ‘The aims of this review were (1) to identify and describe measures of multimorbidity that are most suitable for use in research in primary care and community populations, taking into account the data and resources they require and (2) to investigate the validity of these measures in terms of whether they have demonstrated anticipated associations with patient characteristics, process measures, and health outcomes.’  Medline and Embase: database inception to December 2009  11 191  194  Acceptable  Le Reste et al.20  ‘What are the criteria for multimorbidity found in the scientific medical literature and what definition could be produced with these criteria?’  PubMed, Embase and Cochrane: 1 January 1990 to 31 December 2010  416  54  Acceptable  Yurkovich et al.21  ‘To conduct a systematic review of studies reporting on the development or validation of comorbidity indices using administrative health data and compare their ability to predict outcomes related to comorbidity (i.e., construct validity).’  Medline and Embase: 1946 to September 2012  955  76  Acceptable  Willadsen et al.22  ‘Objective is to explore how multimorbidity is defined in the scientific literature, with a focus on the roles of diseases, risk factors, and symptoms in the definitions’.  PubMed, Medline and Embase: inception to 4 October 2013. Cochrane database: inception to 10 October 2013  943  163  Acceptable  Review reference  Stated aim of the review  Databases and dates of search undertaken by the review  Total titles/ abstracts screened  Total texts included  Quality assessmenta  De Groot et al.18  ‘Which methods are available for measuring comorbidity that can be used in RCTs and prognostic studies’  Medline: January 1966 to September 2000. Embase: January 1988 to September 2000.  Not reported  Not reported  Low quality  Diederichs et al.19  ‘Multimorbidity, defined as the coexistence of 2 or more chronic diseases, is a common phenomenon especially in older people. Numerous efforts to establish a standardized instrument to assess the level of multimorbidity have failed until now, and indices are primarily characterized by their high heterogeneity. Thus, the objective is to provide a comprehensive overview on existing instruments on the basis of a systematic literature review’.  Medline: 1 January 1960 to 31 August 2009  1120  39  Acceptable  Huntley et al.16  ‘The aims of this review were (1) to identify and describe measures of multimorbidity that are most suitable for use in research in primary care and community populations, taking into account the data and resources they require and (2) to investigate the validity of these measures in terms of whether they have demonstrated anticipated associations with patient characteristics, process measures, and health outcomes.’  Medline and Embase: database inception to December 2009  11 191  194  Acceptable  Le Reste et al.20  ‘What are the criteria for multimorbidity found in the scientific medical literature and what definition could be produced with these criteria?’  PubMed, Embase and Cochrane: 1 January 1990 to 31 December 2010  416  54  Acceptable  Yurkovich et al.21  ‘To conduct a systematic review of studies reporting on the development or validation of comorbidity indices using administrative health data and compare their ability to predict outcomes related to comorbidity (i.e., construct validity).’  Medline and Embase: 1946 to September 2012  955  76  Acceptable  Willadsen et al.22  ‘Objective is to explore how multimorbidity is defined in the scientific literature, with a focus on the roles of diseases, risk factors, and symptoms in the definitions’.  PubMed, Medline and Embase: inception to 4 October 2013. Cochrane database: inception to 10 October 2013  943  163  Acceptable  Notes: SIGN, Scottish Intercollegiate Guidelines Netwok; RCT, randomized controlled trial. a Based upon SIGN categories. Table 1 Characteristics of included reviews Review reference  Stated aim of the review  Databases and dates of search undertaken by the review  Total titles/ abstracts screened  Total texts included  Quality assessmenta  De Groot et al.18  ‘Which methods are available for measuring comorbidity that can be used in RCTs and prognostic studies’  Medline: January 1966 to September 2000. Embase: January 1988 to September 2000.  Not reported  Not reported  Low quality  Diederichs et al.19  ‘Multimorbidity, defined as the coexistence of 2 or more chronic diseases, is a common phenomenon especially in older people. Numerous efforts to establish a standardized instrument to assess the level of multimorbidity have failed until now, and indices are primarily characterized by their high heterogeneity. Thus, the objective is to provide a comprehensive overview on existing instruments on the basis of a systematic literature review’.  Medline: 1 January 1960 to 31 August 2009  1120  39  Acceptable  Huntley et al.16  ‘The aims of this review were (1) to identify and describe measures of multimorbidity that are most suitable for use in research in primary care and community populations, taking into account the data and resources they require and (2) to investigate the validity of these measures in terms of whether they have demonstrated anticipated associations with patient characteristics, process measures, and health outcomes.’  Medline and Embase: database inception to December 2009  11 191  194  Acceptable  Le Reste et al.20  ‘What are the criteria for multimorbidity found in the scientific medical literature and what definition could be produced with these criteria?’  PubMed, Embase and Cochrane: 1 January 1990 to 31 December 2010  416  54  Acceptable  Yurkovich et al.21  ‘To conduct a systematic review of studies reporting on the development or validation of comorbidity indices using administrative health data and compare their ability to predict outcomes related to comorbidity (i.e., construct validity).’  Medline and Embase: 1946 to September 2012  955  76  Acceptable  Willadsen et al.22  ‘Objective is to explore how multimorbidity is defined in the scientific literature, with a focus on the roles of diseases, risk factors, and symptoms in the definitions’.  PubMed, Medline and Embase: inception to 4 October 2013. Cochrane database: inception to 10 October 2013  943  163  Acceptable  Review reference  Stated aim of the review  Databases and dates of search undertaken by the review  Total titles/ abstracts screened  Total texts included  Quality assessmenta  De Groot et al.18  ‘Which methods are available for measuring comorbidity that can be used in RCTs and prognostic studies’  Medline: January 1966 to September 2000. Embase: January 1988 to September 2000.  Not reported  Not reported  Low quality  Diederichs et al.19  ‘Multimorbidity, defined as the coexistence of 2 or more chronic diseases, is a common phenomenon especially in older people. Numerous efforts to establish a standardized instrument to assess the level of multimorbidity have failed until now, and indices are primarily characterized by their high heterogeneity. Thus, the objective is to provide a comprehensive overview on existing instruments on the basis of a systematic literature review’.  Medline: 1 January 1960 to 31 August 2009  1120  39  Acceptable  Huntley et al.16  ‘The aims of this review were (1) to identify and describe measures of multimorbidity that are most suitable for use in research in primary care and community populations, taking into account the data and resources they require and (2) to investigate the validity of these measures in terms of whether they have demonstrated anticipated associations with patient characteristics, process measures, and health outcomes.’  Medline and Embase: database inception to December 2009  11 191  194  Acceptable  Le Reste et al.20  ‘What are the criteria for multimorbidity found in the scientific medical literature and what definition could be produced with these criteria?’  PubMed, Embase and Cochrane: 1 January 1990 to 31 December 2010  416  54  Acceptable  Yurkovich et al.21  ‘To conduct a systematic review of studies reporting on the development or validation of comorbidity indices using administrative health data and compare their ability to predict outcomes related to comorbidity (i.e., construct validity).’  Medline and Embase: 1946 to September 2012  955  76  Acceptable  Willadsen et al.22  ‘Objective is to explore how multimorbidity is defined in the scientific literature, with a focus on the roles of diseases, risk factors, and symptoms in the definitions’.  PubMed, Medline and Embase: inception to 4 October 2013. Cochrane database: inception to 10 October 2013  943  163  Acceptable  Notes: SIGN, Scottish Intercollegiate Guidelines Netwok; RCT, randomized controlled trial. a Based upon SIGN categories. Figure 1 View largeDownload slide Flow chart of search strategy Figure 1 View largeDownload slide Flow chart of search strategy Definitions The multimorbidity definitions used in the included reviews are in table 2. As described earlier, Le Reste and Willadsen were the only papers focused on reviewing definitions20,22 and so the definitions provided by the other four were the authors own. Table 2 Multimorbidity definitions from included reviews Review reference  Definition given a prioria or as a result of evidence review  Definition  De Groot et al.18  a priori  ‘The co-occurrence of multiple chronic or acute diseases and medical conditions in one person’  Diederichs et al.19  a priori  ‘Multimorbidity describes “the coexistence of two or more chronic diseases” in the same individual.’  Huntley et al.,14  a priori  ‘The co-occurrence of multiple diseases or medical conditions within 1 person’.  Le Reste et al.20  Review of evidence  ‘Multimorbidity is defined as any combination of chronic disease with at least one other disease (acute or chronic) or biopsychosocial factor (associated or not) or somatic risk factor. Any biopsychosocial factor, any somatic risk factor, the social network, the burden of diseases, the health care consumption, and the patient’s coping strategies may function as modifiers (of the effects of multimorbidity). Multimorbidity may modify the health outcomes and lead to an increased disability or a decreased quality of life or frailty.’  Yurkovich et al.21  a priori  This review used the definition of comorbidity: ‘Comorbidity may be defined as the total burden of illnesses unrelated to the principal diagnosis’  Willadsen et al.22  Review of evidence  Provides no single definition. Conclusion: -Existing definitions (consisting mainly of diseases) are ‘more usable for epidemiologists than for clinicians and patients’. -Recommends definition by Le Reste et al. (above)  Review reference  Definition given a prioria or as a result of evidence review  Definition  De Groot et al.18  a priori  ‘The co-occurrence of multiple chronic or acute diseases and medical conditions in one person’  Diederichs et al.19  a priori  ‘Multimorbidity describes “the coexistence of two or more chronic diseases” in the same individual.’  Huntley et al.,14  a priori  ‘The co-occurrence of multiple diseases or medical conditions within 1 person’.  Le Reste et al.20  Review of evidence  ‘Multimorbidity is defined as any combination of chronic disease with at least one other disease (acute or chronic) or biopsychosocial factor (associated or not) or somatic risk factor. Any biopsychosocial factor, any somatic risk factor, the social network, the burden of diseases, the health care consumption, and the patient’s coping strategies may function as modifiers (of the effects of multimorbidity). Multimorbidity may modify the health outcomes and lead to an increased disability or a decreased quality of life or frailty.’  Yurkovich et al.21  a priori  This review used the definition of comorbidity: ‘Comorbidity may be defined as the total burden of illnesses unrelated to the principal diagnosis’  Willadsen et al.22  Review of evidence  Provides no single definition. Conclusion: -Existing definitions (consisting mainly of diseases) are ‘more usable for epidemiologists than for clinicians and patients’. -Recommends definition by Le Reste et al. (above)  a a priori indicates this is the reviewers own definition. Table 2 Multimorbidity definitions from included reviews Review reference  Definition given a prioria or as a result of evidence review  Definition  De Groot et al.18  a priori  ‘The co-occurrence of multiple chronic or acute diseases and medical conditions in one person’  Diederichs et al.19  a priori  ‘Multimorbidity describes “the coexistence of two or more chronic diseases” in the same individual.’  Huntley et al.,14  a priori  ‘The co-occurrence of multiple diseases or medical conditions within 1 person’.  Le Reste et al.20  Review of evidence  ‘Multimorbidity is defined as any combination of chronic disease with at least one other disease (acute or chronic) or biopsychosocial factor (associated or not) or somatic risk factor. Any biopsychosocial factor, any somatic risk factor, the social network, the burden of diseases, the health care consumption, and the patient’s coping strategies may function as modifiers (of the effects of multimorbidity). Multimorbidity may modify the health outcomes and lead to an increased disability or a decreased quality of life or frailty.’  Yurkovich et al.21  a priori  This review used the definition of comorbidity: ‘Comorbidity may be defined as the total burden of illnesses unrelated to the principal diagnosis’  Willadsen et al.22  Review of evidence  Provides no single definition. Conclusion: -Existing definitions (consisting mainly of diseases) are ‘more usable for epidemiologists than for clinicians and patients’. -Recommends definition by Le Reste et al. (above)  Review reference  Definition given a prioria or as a result of evidence review  Definition  De Groot et al.18  a priori  ‘The co-occurrence of multiple chronic or acute diseases and medical conditions in one person’  Diederichs et al.19  a priori  ‘Multimorbidity describes “the coexistence of two or more chronic diseases” in the same individual.’  Huntley et al.,14  a priori  ‘The co-occurrence of multiple diseases or medical conditions within 1 person’.  Le Reste et al.20  Review of evidence  ‘Multimorbidity is defined as any combination of chronic disease with at least one other disease (acute or chronic) or biopsychosocial factor (associated or not) or somatic risk factor. Any biopsychosocial factor, any somatic risk factor, the social network, the burden of diseases, the health care consumption, and the patient’s coping strategies may function as modifiers (of the effects of multimorbidity). Multimorbidity may modify the health outcomes and lead to an increased disability or a decreased quality of life or frailty.’  Yurkovich et al.21  a priori  This review used the definition of comorbidity: ‘Comorbidity may be defined as the total burden of illnesses unrelated to the principal diagnosis’  Willadsen et al.22  Review of evidence  Provides no single definition. Conclusion: -Existing definitions (consisting mainly of diseases) are ‘more usable for epidemiologists than for clinicians and patients’. -Recommends definition by Le Reste et al. (above)  a a priori indicates this is the reviewers own definition. Le Reste produced a new multimorbidity definition as a result of their review: “… any combination of chronic disease with at least one other disease (acute or chronic) or biopsychosocial factor (associated or not) or somatic risk factor”.20 Willadsen found that more than a third of studies used a cut-off of two or more conditions to define multimorbidity, another third did not specify any cut-off and the remainder had varying cut-off points. The authors found that less than a third of their included studies used an existing definition of multimorbidity. Additionally, definitions varied according to whether or not they specified a duration of condition (e.g. ‘occurrence in the last 5 years’ or having lasted ‘for at least 3 months’) and whether or not they specified the severity of the condition (e.g. staging of the disease). The authors state that consideration of whether included diseases clustered together was considered in only ‘a few’ articles and there was little consideration of complications of diseases. The authors concluded that the majority of existing definitions are ‘more usable for epidemiologists than for clinicians and patients’ and recommended the Le Reste definition due to its comprehensive nature for including more than just disease.22 In the remaining reviews, De Groot and Yurkovich primarily used the term ‘comorbidity’.18,21 The consensus amongst all four was that multimorbidity is the occurrence of multiple diseases or conditions. Diederichs specified that multimorbidity is two or more chronic conditions.19 Measures Commonly used measures Le Reste did not focus on multimorbidity measures.20 The measures covered by the remaining five reviews are in table 3. While the stated aim of Willadsen was to ‘explore how multimorbidity is defined in the scientific literature’, there was overlap between definitions and measures.22 Table 3 Multimorbidity measures, conditions and data sources recommended by included review Review reference  Measures included  MM measure recommended?  Rationale for MM measure recommended  Specific MM conditions recommended?  MM data sources recommended?  De Groot et al.18  Disease counts and 12 weighted measures (Burden of disease index, Charlson index, CIRS, Cornoni-Huntley index, Duke Severity of Illness index Hallstrom index, Hurwitz index, ICED, Incalzi index, Kaplan index, Lui index Shwartz index)  Concludes Charlson, CIRS, ICED and Kaplan are valid and reliable methods to measure comorbidity in clinical research.  Validity and reliability  No  No specific recommendation. Commonly used methods to obtain data in included studies: ‘interviews, questionnaires, physical examinations, medical chart reviews and coded databases’.  Diederichs et al.19  Weighted indices: Charlson Index, Comorbidity Symptom Scale, Seattle Index of Comorbidity, Medication-Based Disease Burden Index, KoMo Score, Index of Coexisting Diseases, Functional Comorbidity Index, Incalzi Index, Kaplan-Feinstein Index, Physiologic Index of Comorbidity, Geriatric Index of Comorbidity, Self-Administered Comorbidity Questionnaire, Shwartz Index and Chronic Disease Score.  Recommends a disease count of 11 conditions. Found most studies did not specify criteria for selection of diseases. If criteria given: high prevalence of the disease, using other indices as a reference point for the selection of disease, conditions which are associated with an increased mortality risk, conditions associated with impact on function and health and the need for management.  Disease count based on conditions which are the 20 most frequently listed diagnoses for people aged greater than or equal to 65 years in three data sources in Germany (the inpatient sector, the outpatient sector and mortality statistics).  Cancer, diabetes, depression, hypertension, MI, chronic ischaemic heart disease, heart arrythmias, heart insufficiency, stroke, COPD, arthritis.  No specific recommendation. Included studies used: patient self-report, physician reports, clinical examinations, medical records, and administrative data. Gives advice on self-report: 'use disease specifications that can be distinguishable by lay persons [in order to increase validity of self-report].  Huntley et al.16  Disease counts and weighted measures. Most common measures (n studies): disease count (98), Charlson (38), ACG System (25), CIRS (10), Duke Severity of Illness (6).  Care utilization: the ACG System, Charlson index, or disease counts Costs: The ACG System Mortality: Charlson index Quality of life: Disease counts or Charlson index Other outcomes: Simple counts of diseases or medications  Recommendations based upon purpose of study and evidence base behind measures used for that purpose.  No  No specific recommendation. Commonly used measures: interviews, questionnaires, physical examinations, medical chart reviews, and coded databases.  Yurkovich et al.21  Administrative data measures (n studies): Charlson and its adaptations (35); Elixhauser (2); Fleming et al. index (1); Abildstrom et al. index (1) Medication- based indices: Chronic Disease Score (9), Rx-Risk (3) and Medication Based Disease Burden Index (2)  Diagnosis-based measures, (particularly Elixhauser and the Romano adaptation of the Charlson) resulted in higher ability to predict mortality outcomes. Medication-based indices, (such as the Chronic Disease Score) demonstrated better performance for predicting health care utilization.  Recommends selection of measure to be based on ‘type of data available, the study population, and the specific outcome of interest in the study.’  No  No specific recommendation. Review was limited to administrative data indices only but the authors commented on included studies which compared data sources (all were Charlson studies): two studies found self-report and administrative data had similar ability to ‘predict various outcomes’ One review and two studies found poor agreement between case note review and administrative data  Willadsen et al.22  Charlson, Clinical Classification Software, CIRS, ACG, Aggregated Diagnosis Groups, Medication based, Expanded Diagnosis Clusters, Resource Utilization Bands, The Functional Comorbidity Index, ICED, QoF, The Registration Network Family Practices  Does not recommend a single measure. As documented in table 2, the authors state the importance of including risk factors and symptoms and severity as well as diseases if want a clinically relevant definition (and thus measure)  N/A  No  No specific recommendation. The included studies used data from administrative data and self-report  Review reference  Measures included  MM measure recommended?  Rationale for MM measure recommended  Specific MM conditions recommended?  MM data sources recommended?  De Groot et al.18  Disease counts and 12 weighted measures (Burden of disease index, Charlson index, CIRS, Cornoni-Huntley index, Duke Severity of Illness index Hallstrom index, Hurwitz index, ICED, Incalzi index, Kaplan index, Lui index Shwartz index)  Concludes Charlson, CIRS, ICED and Kaplan are valid and reliable methods to measure comorbidity in clinical research.  Validity and reliability  No  No specific recommendation. Commonly used methods to obtain data in included studies: ‘interviews, questionnaires, physical examinations, medical chart reviews and coded databases’.  Diederichs et al.19  Weighted indices: Charlson Index, Comorbidity Symptom Scale, Seattle Index of Comorbidity, Medication-Based Disease Burden Index, KoMo Score, Index of Coexisting Diseases, Functional Comorbidity Index, Incalzi Index, Kaplan-Feinstein Index, Physiologic Index of Comorbidity, Geriatric Index of Comorbidity, Self-Administered Comorbidity Questionnaire, Shwartz Index and Chronic Disease Score.  Recommends a disease count of 11 conditions. Found most studies did not specify criteria for selection of diseases. If criteria given: high prevalence of the disease, using other indices as a reference point for the selection of disease, conditions which are associated with an increased mortality risk, conditions associated with impact on function and health and the need for management.  Disease count based on conditions which are the 20 most frequently listed diagnoses for people aged greater than or equal to 65 years in three data sources in Germany (the inpatient sector, the outpatient sector and mortality statistics).  Cancer, diabetes, depression, hypertension, MI, chronic ischaemic heart disease, heart arrythmias, heart insufficiency, stroke, COPD, arthritis.  No specific recommendation. Included studies used: patient self-report, physician reports, clinical examinations, medical records, and administrative data. Gives advice on self-report: 'use disease specifications that can be distinguishable by lay persons [in order to increase validity of self-report].  Huntley et al.16  Disease counts and weighted measures. Most common measures (n studies): disease count (98), Charlson (38), ACG System (25), CIRS (10), Duke Severity of Illness (6).  Care utilization: the ACG System, Charlson index, or disease counts Costs: The ACG System Mortality: Charlson index Quality of life: Disease counts or Charlson index Other outcomes: Simple counts of diseases or medications  Recommendations based upon purpose of study and evidence base behind measures used for that purpose.  No  No specific recommendation. Commonly used measures: interviews, questionnaires, physical examinations, medical chart reviews, and coded databases.  Yurkovich et al.21  Administrative data measures (n studies): Charlson and its adaptations (35); Elixhauser (2); Fleming et al. index (1); Abildstrom et al. index (1) Medication- based indices: Chronic Disease Score (9), Rx-Risk (3) and Medication Based Disease Burden Index (2)  Diagnosis-based measures, (particularly Elixhauser and the Romano adaptation of the Charlson) resulted in higher ability to predict mortality outcomes. Medication-based indices, (such as the Chronic Disease Score) demonstrated better performance for predicting health care utilization.  Recommends selection of measure to be based on ‘type of data available, the study population, and the specific outcome of interest in the study.’  No  No specific recommendation. Review was limited to administrative data indices only but the authors commented on included studies which compared data sources (all were Charlson studies): two studies found self-report and administrative data had similar ability to ‘predict various outcomes’ One review and two studies found poor agreement between case note review and administrative data  Willadsen et al.22  Charlson, Clinical Classification Software, CIRS, ACG, Aggregated Diagnosis Groups, Medication based, Expanded Diagnosis Clusters, Resource Utilization Bands, The Functional Comorbidity Index, ICED, QoF, The Registration Network Family Practices  Does not recommend a single measure. As documented in table 2, the authors state the importance of including risk factors and symptoms and severity as well as diseases if want a clinically relevant definition (and thus measure)  N/A  No  No specific recommendation. The included studies used data from administrative data and self-report  Note: MM, multimorbidity; CIRS, Cumulative illness rating scale; ICED, Index of Coexistent disease; MI, myocardial infarction; COPD, chronic obstructive pulmonary disease; ACG, Adjusted Clinical Groups; QoF, Quality Outcomes Framework; N/A, not applicable. Table 3 Multimorbidity measures, conditions and data sources recommended by included review Review reference  Measures included  MM measure recommended?  Rationale for MM measure recommended  Specific MM conditions recommended?  MM data sources recommended?  De Groot et al.18  Disease counts and 12 weighted measures (Burden of disease index, Charlson index, CIRS, Cornoni-Huntley index, Duke Severity of Illness index Hallstrom index, Hurwitz index, ICED, Incalzi index, Kaplan index, Lui index Shwartz index)  Concludes Charlson, CIRS, ICED and Kaplan are valid and reliable methods to measure comorbidity in clinical research.  Validity and reliability  No  No specific recommendation. Commonly used methods to obtain data in included studies: ‘interviews, questionnaires, physical examinations, medical chart reviews and coded databases’.  Diederichs et al.19  Weighted indices: Charlson Index, Comorbidity Symptom Scale, Seattle Index of Comorbidity, Medication-Based Disease Burden Index, KoMo Score, Index of Coexisting Diseases, Functional Comorbidity Index, Incalzi Index, Kaplan-Feinstein Index, Physiologic Index of Comorbidity, Geriatric Index of Comorbidity, Self-Administered Comorbidity Questionnaire, Shwartz Index and Chronic Disease Score.  Recommends a disease count of 11 conditions. Found most studies did not specify criteria for selection of diseases. If criteria given: high prevalence of the disease, using other indices as a reference point for the selection of disease, conditions which are associated with an increased mortality risk, conditions associated with impact on function and health and the need for management.  Disease count based on conditions which are the 20 most frequently listed diagnoses for people aged greater than or equal to 65 years in three data sources in Germany (the inpatient sector, the outpatient sector and mortality statistics).  Cancer, diabetes, depression, hypertension, MI, chronic ischaemic heart disease, heart arrythmias, heart insufficiency, stroke, COPD, arthritis.  No specific recommendation. Included studies used: patient self-report, physician reports, clinical examinations, medical records, and administrative data. Gives advice on self-report: 'use disease specifications that can be distinguishable by lay persons [in order to increase validity of self-report].  Huntley et al.16  Disease counts and weighted measures. Most common measures (n studies): disease count (98), Charlson (38), ACG System (25), CIRS (10), Duke Severity of Illness (6).  Care utilization: the ACG System, Charlson index, or disease counts Costs: The ACG System Mortality: Charlson index Quality of life: Disease counts or Charlson index Other outcomes: Simple counts of diseases or medications  Recommendations based upon purpose of study and evidence base behind measures used for that purpose.  No  No specific recommendation. Commonly used measures: interviews, questionnaires, physical examinations, medical chart reviews, and coded databases.  Yurkovich et al.21  Administrative data measures (n studies): Charlson and its adaptations (35); Elixhauser (2); Fleming et al. index (1); Abildstrom et al. index (1) Medication- based indices: Chronic Disease Score (9), Rx-Risk (3) and Medication Based Disease Burden Index (2)  Diagnosis-based measures, (particularly Elixhauser and the Romano adaptation of the Charlson) resulted in higher ability to predict mortality outcomes. Medication-based indices, (such as the Chronic Disease Score) demonstrated better performance for predicting health care utilization.  Recommends selection of measure to be based on ‘type of data available, the study population, and the specific outcome of interest in the study.’  No  No specific recommendation. Review was limited to administrative data indices only but the authors commented on included studies which compared data sources (all were Charlson studies): two studies found self-report and administrative data had similar ability to ‘predict various outcomes’ One review and two studies found poor agreement between case note review and administrative data  Willadsen et al.22  Charlson, Clinical Classification Software, CIRS, ACG, Aggregated Diagnosis Groups, Medication based, Expanded Diagnosis Clusters, Resource Utilization Bands, The Functional Comorbidity Index, ICED, QoF, The Registration Network Family Practices  Does not recommend a single measure. As documented in table 2, the authors state the importance of including risk factors and symptoms and severity as well as diseases if want a clinically relevant definition (and thus measure)  N/A  No  No specific recommendation. The included studies used data from administrative data and self-report  Review reference  Measures included  MM measure recommended?  Rationale for MM measure recommended  Specific MM conditions recommended?  MM data sources recommended?  De Groot et al.18  Disease counts and 12 weighted measures (Burden of disease index, Charlson index, CIRS, Cornoni-Huntley index, Duke Severity of Illness index Hallstrom index, Hurwitz index, ICED, Incalzi index, Kaplan index, Lui index Shwartz index)  Concludes Charlson, CIRS, ICED and Kaplan are valid and reliable methods to measure comorbidity in clinical research.  Validity and reliability  No  No specific recommendation. Commonly used methods to obtain data in included studies: ‘interviews, questionnaires, physical examinations, medical chart reviews and coded databases’.  Diederichs et al.19  Weighted indices: Charlson Index, Comorbidity Symptom Scale, Seattle Index of Comorbidity, Medication-Based Disease Burden Index, KoMo Score, Index of Coexisting Diseases, Functional Comorbidity Index, Incalzi Index, Kaplan-Feinstein Index, Physiologic Index of Comorbidity, Geriatric Index of Comorbidity, Self-Administered Comorbidity Questionnaire, Shwartz Index and Chronic Disease Score.  Recommends a disease count of 11 conditions. Found most studies did not specify criteria for selection of diseases. If criteria given: high prevalence of the disease, using other indices as a reference point for the selection of disease, conditions which are associated with an increased mortality risk, conditions associated with impact on function and health and the need for management.  Disease count based on conditions which are the 20 most frequently listed diagnoses for people aged greater than or equal to 65 years in three data sources in Germany (the inpatient sector, the outpatient sector and mortality statistics).  Cancer, diabetes, depression, hypertension, MI, chronic ischaemic heart disease, heart arrythmias, heart insufficiency, stroke, COPD, arthritis.  No specific recommendation. Included studies used: patient self-report, physician reports, clinical examinations, medical records, and administrative data. Gives advice on self-report: 'use disease specifications that can be distinguishable by lay persons [in order to increase validity of self-report].  Huntley et al.16  Disease counts and weighted measures. Most common measures (n studies): disease count (98), Charlson (38), ACG System (25), CIRS (10), Duke Severity of Illness (6).  Care utilization: the ACG System, Charlson index, or disease counts Costs: The ACG System Mortality: Charlson index Quality of life: Disease counts or Charlson index Other outcomes: Simple counts of diseases or medications  Recommendations based upon purpose of study and evidence base behind measures used for that purpose.  No  No specific recommendation. Commonly used measures: interviews, questionnaires, physical examinations, medical chart reviews, and coded databases.  Yurkovich et al.21  Administrative data measures (n studies): Charlson and its adaptations (35); Elixhauser (2); Fleming et al. index (1); Abildstrom et al. index (1) Medication- based indices: Chronic Disease Score (9), Rx-Risk (3) and Medication Based Disease Burden Index (2)  Diagnosis-based measures, (particularly Elixhauser and the Romano adaptation of the Charlson) resulted in higher ability to predict mortality outcomes. Medication-based indices, (such as the Chronic Disease Score) demonstrated better performance for predicting health care utilization.  Recommends selection of measure to be based on ‘type of data available, the study population, and the specific outcome of interest in the study.’  No  No specific recommendation. Review was limited to administrative data indices only but the authors commented on included studies which compared data sources (all were Charlson studies): two studies found self-report and administrative data had similar ability to ‘predict various outcomes’ One review and two studies found poor agreement between case note review and administrative data  Willadsen et al.22  Charlson, Clinical Classification Software, CIRS, ACG, Aggregated Diagnosis Groups, Medication based, Expanded Diagnosis Clusters, Resource Utilization Bands, The Functional Comorbidity Index, ICED, QoF, The Registration Network Family Practices  Does not recommend a single measure. As documented in table 2, the authors state the importance of including risk factors and symptoms and severity as well as diseases if want a clinically relevant definition (and thus measure)  N/A  No  No specific recommendation. The included studies used data from administrative data and self-report  Note: MM, multimorbidity; CIRS, Cumulative illness rating scale; ICED, Index of Coexistent disease; MI, myocardial infarction; COPD, chronic obstructive pulmonary disease; ACG, Adjusted Clinical Groups; QoF, Quality Outcomes Framework; N/A, not applicable. The measures included by reviews encompassed disease counts and weighted indices such as the Charlson Index, the Cumulative Illness Rating Scale (CIRS), the Index of Coexistent Disease (ICED), the Adjusted Clinical Groups (ACG) System and the Duke Severity of Illness. Yurkovich and Huntley examined the frequency of measures. Yurkovich categorized measures as ‘administrative data’ (the most common being Charlson) and ‘medication-based’ (the most common being the Chronic Disease Score).21 Huntley categorized the most common measures as: disease counts, the Charlson index and variations, the ACG system, the CIRS and the Duke Severity Illness Check-list System.16 Despite the name, disease counts included more than just diseases (e.g. they included categories of conditions). The authors found disease counts being used in 98 studies and the number of disease ‘items’ included within counts ranged from 9 to 35.16 Willadsen found that measures included by their papers contained conditions ranging in number from 4 to 147.22 Recommended measures Yurkovich found that diagnosis-based measures such as the Elixhauser index and the Romano adaptation of the Charlson index were best able to predict mortality outcomes while the medication-based Chronic Disease Score was best able to predict health care use.21 Huntley recommended that researchers select a measure for a study based upon the measure validated for use in that scenario, for example, the Charlson index for predicting mortality. The authors also state that simple counts of diseases or medications perform almost as effectively as complex measures in predicting most outcomes.16 De Groot assessed the content, criterion and construct validity of measures. They concluded that the Charlson, CIRS, ICED and Kaplan indices are valid and reliable methods for use in clinical research but that other measures (such as disease counts) were more difficult to assess due to limited data.18 Willadsen did not recommend a single measure and instead, as described previously, stated the importance of including risk factors, symptoms and severity of diseases.22 Diederichs also did not recommend a single measure. They found studies of disease counts often did not specify the criteria for the selection of diseases, but if criteria were given these were: high prevalence of the disease, using other indices as a reference point for the selection of disease, or high impact conditions in terms of increased mortality risk, an impact on function and health and the need for management. They recommended 11 conditions selected on the basis of being the most common causes of inpatient and outpatient attendance as well as death in people aged over 64 in Germany. The conditions included cancer, depression, myocardial infarction and hypertension.19 Data sources All five reviews found patient self-report, physician reports, clinical examinations, medical record reviews and administrative data (‘coded databases’ or ‘routine data’) were common sources of multimorbidity data among their included studies.16,18,19,21,22 No review studied whether any source was superior, although Yurkovich found evidence that the Charlson index derived from self-report and that derived from administrative data had similar abilities to ‘predict various outcomes’.21 De Groot stated that medical chart reviews are preferable for use in smaller studies as they likely yield the most complete data but that this is likely impractical in larger studies and so administrative databases can be used.18 Similarly, Huntley noted that administrative data have the advantage of ease of use but may be limited by data quality issues.16 Discussion Summary of findings Our review pooled the findings of six systematic reviews. We found heterogeneity of multimorbidity definitions and measures, but there were a number of commonalities. Most reviews defined multimorbidity as the occurrence of multiple diseases or conditions, the most common cut-off being two or more. Le Reste produced a new definition that encompassed biopsychosocial factors and somatic risk factors along with disease.20 This was recommended by Willadsen as being the most clinically relevant definition of multimorbidity available.22 Common measures included the Charlson, CIRS, ICED, Kaplan, the ACG system and disease counts, with advice that measures be selected based upon the purpose of a particular study.16,18 No reviews made recommendations about the most appropriate data sources to use when measuring multimorbidity. Strengths and limitations Our systematic review provides a high-level summary of both the definition and measurement of multimorbidity in relevant systematic reviews. Ours is the first to focus upon those reviews which primarily aimed to examine multimorbidity definitions or measures. This is important given the heterogeneity in definitions and measures available and the associated complexity in developing consensus. We acknowledge that reviews such as that by Fortin et al. (of prevalence studies of multimorbidity)2 and Marengoni et al.5 (of ageing and multimorbidity) discuss recommended definitions and measures at the end of their reviews, but we have not included these as their primary aim did not meet our inclusion criteria. A limitation is that search terms were limited to the title only for practical reasons which means some relevant reviews could be missed. We conducted a test search including these terms in the abstract or full text which revealed no additional reviews in the first 100 titles screened. Additionally, as recommended by PRISMA, systematic reviews should be identified as such in the title.15 One of the included reviews (examining measures of multimorbidity) was classed as low quality. However, as there were three other reviews examining multimorbidity measures this should reduce the likelihood that this affected our findings. Comparison with literature Our findings are consistent with other systematic reviewers who have encountered challenges due to the lack of a common approach towards measuring and defining multimorbidity.2,23–25 Definitions Willadsen highlighted that many definitions and measures seem to be tailored towards use in research rather than being clinically relevant.22 It is true that traditional approaches, for example, measuring multimorbidity using the Charlson or disease counts, do not capture the holistic experience of multimorbidity. For example, we know that an individuals’ ability to cope with disease is influenced by both person factors and wider socio-environmental factors and that at a population level, multimorbidity is associated with higher levels of deprivation.4,26–30 The definition by Le Reste is more likely to capture this complexity but the multi-faceted nature of the definition makes it difficult to operationalize in practice. Instead of adding further elements to the definition and measurement of multimorbidity, it is perhaps more appropriate to ensure there is consideration of its holistic nature when studying its determinants and outcomes and when managing it clinically. This would include understanding its relationship with health inequalities in areas of high deprivation, as well as to frailty and the ageing process.11,12 The cut-off point regarding the minimum number of conditions to equate to being multimorbid needs further consideration. The most common cut-off point found by our reviews was two or more conditions and this was consistent with the findings of Fortin in their review of prevalence studies of multimorbidity.2 The prevalence of multimorbidity is inevitably affected by the cut-off selected and additionally it is likely that a higher cut-off would select a patient group with a higher burden of multimorbidity.2 This needs further research, for example, by testing the number of conditions which best identify patients at higher risk of outcomes such as hospital stay, disability, frailty or mortality. Measures When multimorbidity is defined and measured on the basis of a count of conditions the measurement of multimorbidity is closely linked to the definition. We have used the term ‘disease counts’ as this is the common phrase used in the literature, but acknowledge these measures can include a wider spectrum of health conditions (e.g. risk factors for disease). Disease counts are likely more appropriate for scenarios where multiple outcomes are being considered or in which no single weighted measure has been validated.16 They may also be a more intuitive summary of multimorbidity burden in patients, for example, when showing the link between multimorbidity and socioeconomic status.4 Additionally, reviews have found that multimorbidity may be more appropriately considered as different common clusters of conditions and this is easier to measure using counts.24,31 If researchers are selecting conditions to include in a count the purpose of the work being conducted must be considered. Some conditions, for example, depression, may have greater impact upon patients in terms of quality of life or function.32 Other conditions such as heart disease may impact more upon health services in terms of number of admissions or treatment costs.2,19 In studies using weighted measures the definition and measurement of multimorbidity are more distinct. Many weighted measures were originally developed as comorbidity measures but are increasingly being used as multimorbidity measures.18 Weighted measures, if used for appropriate outcomes, can assist in predicting patient outcome and future healthcare usage and can also provide an assessment of the burden of multimorbidity experienced by the patient, their carers or health and social care services.33 Therefore, where the aim is to examine outcomes in patients and to account for the presence of multiple conditions, a validated weighted measure may be more appropriate or informative than a disease count. Data sources No review recommended a particular data source to measure multimorbidity. In the wider literature, a number of studies and reviews have compared data sources for comorbidity and multimorbidity measures, often with conflicting findings.3,34 The availability of data and the resource implications will additionally affect the choice of data used. For example, while case-note review is viewed as being more complete than administrative data as it is more resource intensive.3,34 Another important data source is patient self-report, which may be more likely to capture conditions which may not be seen as important clinically but impact on function or quality of life.32 Regardless of measure, different data sources will affect the prevalence of multimorbidity.2,35 Implications for research and practice Our key recommendation is that researchers be explicit about the definitions and measure(s) they are using and give a rationale for their choice. This will enable comparison of findings across different settings and outcomes as well as progress the evidence base regarding the most appropriate definitions and measures for particular scenarios. Multimorbidity is an important public health challenge, which is influenced strongly by wider social and environmental factors. In this review, the paper by Le Reste highlighted the holistic nature of multimorbidity.20 In clinical and public health practice, holistic approaches that take into account more than just the medical management of disease could assist with reducing its impact. However, there is a need for more evidence on the effectiveness of primary care and community-based interventions, including those tackling the challenges experienced by individuals with socio-economic deprivation.36 Despite this, recent research in primary care in deprived areas has shown that a co-development model of intervention development for multimorbidity (CARE Plus) was feasible and may be cost-effective, thus pointing to future directions in reducing the burden of multimorbidity.37,38 Overall, a definition of multiple co-existing conditions is reasonable and a cut-off should be explicitly defined. Researchers would be consistent with others using a cut-off of two or more. Using a weighted measure validated for the outcome being considered is advised, but where evidence is weak or where multiple outcomes or populations are being considered, the use of disease counts is appropriate. There is precedence for the inclusion of conditions other than solely chronic disease in a multimorbidity measure but a rationale for included and excluded conditions should be given. Supplementary data Supplementary data are available at EURPUB online. Funding Dr Marjorie Johnston was funded by a Clinical Academic Fellowship from the Chief Scientist Office, Scotland (CAF/13/03), was affiliated with the Farr Institute of Health Informatics and Research Scotland and was an honorary Public Health Registrar at NHS Grampian. Conflicts of interest: None declared. Key points To improve consensus in defining and measuring multimorbidity, we recommend researchers and practitioners be explicit about the definitions and measure(s) they are using and give a rationale for their choice. We conclude that multimorbidity is the coexistence of multiple conditions (most commonly defined as two or more conditions). Validated multimorbidity measures for particular scenarios should be chosen if these exist. Where there is no validated measure or where multiple outcomes or populations are being considered, disease counts are appropriate. References 1 Uijen AA, Van de Lisdon EH. Multimorbidity in primary care: prevalence and trend over the last 20 years. Eur J Gen Pract  2008; 14: 28– 32. Google Scholar CrossRef Search ADS PubMed  2 Fortin M, Stewart M, Poitras M, et al.   A systematic review of prevalence studies on multimorbidity: toward a more uniform methodology. Ann Fam Med  2012; 10: 142– 51. Google Scholar CrossRef Search ADS PubMed  3 Leal JR, Laupland KB. Validity of ascertainment of co-morbid illness using administrative databases: a systematic review. Clin Microbiol Infect  2010; 16: 715– 21. Google Scholar CrossRef Search ADS PubMed  4 Barnett K, Mercer SW, Norbury M, et al.   Epidemiology of multimorbidity and implications for health care, research and medical education: a cross-sectional study. Lancet  2012; 380: 37– 43. Google Scholar CrossRef Search ADS PubMed  5 Marengoni A, Angleman S, Melis R, et al.   Aging with multimorbidity: a systematic review of the literature. Ageing Res Rev  2011; 10: 430– 9. Google Scholar CrossRef Search ADS PubMed  6 Wallace E, Salisbury C, Guthrie B, et al.   Managing patients with multimorbidity in primary care. BMJ  2015; 350: h176. Google Scholar CrossRef Search ADS PubMed  7 Farmer C, Fenu E, O'Flynn N, Guthrie B. Clinical assessment and management of multimorbidity: summary of NICE guideline. BMJ  2016; 354: i4843. Google Scholar CrossRef Search ADS PubMed  8 Hughes LD, McMurdo MET, Guthrie B. Guidelines for people not for diseases: the challenges of applying UK clinical guidelines to people with multimorbidity. Age Ageing  2013; 42: 62– 9. Google Scholar CrossRef Search ADS PubMed  9 Calderón-Larrañaga A, Poblador-Plou B, González-Rubio F, et al.   Multimorbidity, polypharmacy, referrals, and adverse drug events: are we doing things well? Br J Gen Pract  2012; 62: e821– 6. Google Scholar CrossRef Search ADS PubMed  10 McDaid O, Normand C, Kelly A, Smith S. Prevalence, patterns and healthcare burden of multimorbidity in the older irish population. Ir J Med Sci  2013; 182: S229. 11 Beard JR, Officer A, de Carvalho IA, et al.   The World report on ageing and health: a policy framework for healthy ageing. Lancet  2016; 387: 2145– 54. Google Scholar CrossRef Search ADS PubMed  12 Villacampa-Fernández P, Navarro-Pardo E, Tarín JJ, Cano A. Frailty and multimorbidity: two related yet different concepts. Maturitas  2017; 95: 31– 5. Google Scholar CrossRef Search ADS PubMed  13 Almirall J, Fortin M. The coexistence of terms to describe the presence of multiple concurrent diseases. J Comorb  2013; 3: 4– 9. Google Scholar CrossRef Search ADS PubMed  14 National Insitute for Health and Care Excellence. Multimorbidity: clinical assessment and management. 2016. Available at: https://www.nice.org.uk/guidance/ng56 (10 July 2017 date last accessed). 15 Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ  2009; 339: b2535. Google Scholar CrossRef Search ADS PubMed  16 Huntley AL, Johnson R, Purdy S, et al.   Measures of multimorbidity and morbidity burden for use in primary care and community settings: a systematic review and guide. Ann Fam Med  2012; 10: 134– 41. Google Scholar CrossRef Search ADS PubMed  17 Scottish Intercollegiate Guidelines Network. Critical appraisal: notes and checklists. 2014. Available at: http://www.sign.ac.uk/methodology/checklists.html# (20 February 2014, date last accessed). 18 De Groot V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity: a critical review of available methods. J Clin Epidemiol  2003; 56: 221– 9. Google Scholar CrossRef Search ADS PubMed  19 Diederichs C, Berger K, Bartels DB. The measurement of multiple chronic diseases- a systematic review on existing multimorbidity indices. J Gerontol A Biol Sci Med Sci  2011; 66A: 301– 11. Google Scholar CrossRef Search ADS   20 Le Reste JY, Nabbe P, Manceau B, et al.   The European General Practice Research Network presents a comprehensive definition of multimorbidity in family medicine and long term care, following a systematic review of relevant literature. J Am Med Dir Assoc  2013; 14: 319– 25. Google Scholar CrossRef Search ADS PubMed  21 Yurkovich M, Avina-Zubieta JA, Thomas J, et al.   A systematic review identifies valid comorbidity indices derived from administrative health data. J Clin Epidemiol  2015; 68: 3– 14. Google Scholar CrossRef Search ADS PubMed  22 Willadsen TG, Bebe A, Koster-Rasmussen R, et al.   The role of diseases, risk factors and symptoms in the definition of multimorbidity—a systematic review. Scand J Prim Health Care  2016; 34: 112– 21. Google Scholar CrossRef Search ADS PubMed  23 Violan C, Foguet-Boreu Q, Flores-Mateo G, et al.   Prevalence, determinants and patterns of multimorbidity in primary care: a systematic review of observational Studies. PLoS One  2014; 9: e102149. Google Scholar CrossRef Search ADS PubMed  24 Prados-Torres A, Calderon-Larranaga A, Hancco-Saavedra J, et al.   Multimorbidity patterns: a systematic review. J Clin Epidemiol  2014; 67: 254– 66. Google Scholar CrossRef Search ADS PubMed  25 Holzer BM, Siebenhuener K, Bopp M, Minder CE. Evidence-based design recommendations for prevalence studies on multimorbidity: improving comparability of estimates. Popul Health Metr  2017; 15: 9. Available at: https://doi.org/10.1186/s12963-017-0126-4. Google Scholar CrossRef Search ADS PubMed  26 Cairns JM, Curtis SE, Bambra C. Defying deprivation: a cross-sectional analysis of area level health resilience in England. Health Place  2012; 18: 928– 33. Google Scholar CrossRef Search ADS PubMed  27 Marmot M, Allen J, Goldblatt P, et al.   Fair Society, Healthy Lives: The Marmot Review. 2010; Available at: http://www.instituteofhealthequity.org/projects/fair-society-healthy-lives-the-marmot-review/fair-society-healthy-lives-full-report (1 May 2017, last date accessed). 28 Bielderman A, de Greef MH, Krijnen WP, van der Schans CP. Relationship between socioeconomic status and quality of life in older adults: a path analysis. Qual Life Res  2015; 24: 1697– 705. Google Scholar CrossRef Search ADS PubMed  29 Lawson KD, Mercer SW, Wyke S, et al.   Double trouble: the impact of multimorbidity and deprivation on preference-weighted health related quality of life a cross sectional analysis of the Scottish Health Survey. Int J Equity Health  2013; 12: 67. Google Scholar CrossRef Search ADS PubMed  30 McLean G, Gunn J, Wyke S, et al.   The influence of socioeconomic deprivation on multimorbidity at different ages: a cross-sectional study. Br J Gen Pract  2014; 64: e440– 7. Google Scholar CrossRef Search ADS PubMed  31 Sinnige J, Braspenning J, Schellevis F, et al.   The prevalence of disease clusters in older adults with multiple chronic diseases–a systematic literature review. PLoS One  2013; 8: e79641. Google Scholar CrossRef Search ADS PubMed  32 Walker V, Perret-Guillaume C, Kesse-Guyot E, et al.   Effect of multimorbidity on health-related quality of life in adults aged 55 years or older: results from the SU.VI.MAX 2 Cohort. PLoS One  2016; 11: e0169282. Google Scholar CrossRef Search ADS PubMed  33 Brilleman SL, Salisbury C. Comparing measures of multimorbidity to predict outcomes in primary care: a cross sectional study. Fam Pract  2013; 30: 172– 80. Google Scholar CrossRef Search ADS PubMed  34 Needham DM, Scales DC, Laupacis A, Pronovost PJ. A systematic review of the Charlson comorbidity index using Canadian administrative databases: a perspective on risk adjustment in critical care research. J Crit Care  2005; 20: 12– 9. Google Scholar CrossRef Search ADS PubMed  35 Mokraoui N, Haggerty J, Almirall J, Fortin M. Prevalence of self-reported multimorbidity in the general population and in primary care practices: a cross-sectional study. BMC Res Notes  2016; 9. Available at: https://doi.org/10.1186/s13104-016-2121-4. 36 Smith SM, Wallace E, O'Dowd T, Fortin M. Interventions for improving outcomes in patients with multimorbidity in primary care and community setttings. Cochrane Database Syst Rev  2016; 3: CD006560. Doi: 10.1002/14651858.CD006560.pub3. Google Scholar PubMed  37 Mercer SW, O'Brien R, Fitzpatrick B, et al.   The development and optimisation of a primary care-based whole system complex intervention (CARE Plus) for patients with multimorbidity living in areas of high socioeconomic deprivation. Chronic Illn  2016; 12: 165– 81. Google Scholar CrossRef Search ADS PubMed  38 Mercer SW, Fitzpatrick B, Guthrie B, et al.   The Care Plus study-a whole system intervention to improve quality of life of primary care patients with multimorbidity in areas of high socioeconomic deprivation: cluster randomised controlled trial. BMC Med  2016; 14: 88. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

The European Journal of Public HealthOxford University Press

Published: Jun 5, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off