Multimorbidity in primary care: a systematic review of prospective cohort studies (original) (raw)

Abstract

Background

Primary care increasingly deals with patients with multimorbidity, but relevant evidence-based interventions are scarce. Knowledge about multimorbidity over time is required to inform the development of effective interventions.

Aim

This review identifies prospective cohort studies of multimorbidity in primary care to determine: their nature, scope and key findings; the methodologies used; and gaps in knowledge.

Design

Systematic review.

Method

Studies were identified by searching electronic databases, reviewing citations, and writing to authors. Searches were limited to adult populations with no restrictions on publication date or language. In total, 996 articles were identified and screened.

Results

Of the 996 articles, six detailing five completed prospective cohort studies were selected as appropriate. Three of the studies were undertaken in the US and two in The Netherlands; none was nationally representative. The main focus of the studies was: healthcare utilisation and/or costs (n = 3); patients' physical functioning (n = 1); and risk factors for developing multimorbidity (n = 1). The conditions that were included varied widely. The findings of these studies showed that multimorbidity increased healthcare costs (n = 2), inpatient admission (n = 1), death rates (n = 1), and service use (n = 3), and reduced physical functioning (n = 1). One study identified psychosocial risk factors for multimorbidity. No study used random sampling, sample sizes were relatively small (414–3745 patients at baseline), and study duration was relatively short (1–4 years). No study focused on prevalence, treatment use, patient safety, service models, cultural or socioeconomic factors, and patient experience, and no study collected qualitative data.

Conclusion

Few longitudinal studies based in primary care have investigated multimorbidity. Further large, long-term prospective studies are required to inform healthcare commissioning, planning, and delivery.

Keywords: chronic disease, multimorbidity, primary care, review

INTRODUCTION

The dramatic rise in long-term conditions presents a significant challenge to healthcare systems worldwide.1 Primary care is key to the management of patients with long-term conditions2,3 but, in the main, takes a single-disease approach,4 even though multimorbidity — the co-occurrence of two or more long-term conditions within an individual — is common.5–8

Despite the high prevalence of multimorbidity, the evidence base for interventions is extremely limited.9,10 An important precursor to developing effective interventions is knowledge about multimorbidity over time in ‘real-life’ primary care settings. Prospective cohort studies are the most robust way to observe ‘real-life’ issues over time.11 They have fewer potential sources of bias than retrospective and case–control studies, and yield true incidence and relative risk compared with randomised trial data that, due to strict eligibility for the trial, low recruitment levels, or large numbers of people refusing consent, often have restricted generalisability. As such, prospective cohort studies are the ‘gold standard’ for studying and describing the natural history and development of morbidity, as well as the development and implementation of prognostic models of care.12

Although reviews of the impact of multimorbidity have been undertaken,13 there are no published reviews of cohort studies on multimorbidity in primary care. This article reports the findings of a systematic review of prospective cohort studies of multimorbidity in primary care. The aims were to determine:

• the nature, scope and key findings of the published studies;
• the methodologies used in the studies; and
• any gaps in knowledge.

METHOD

Inclusion and exclusion criteria

Multimorbidity was defined as an individual having two or more conditions, without a specific index condition being specified. Studies with a prospective, longitudinal design, whose main focus was multimorbidity in adults in primary care settings, were included. There were no restrictions on publication date or language of the full paper, but an abstract in English had to be available. As prospective cohort studies are the ‘gold standard’ for conducting such research, retrospective studies, cross-sectional study designs, evaluation studies, randomised controlled trials and intervention studies, studies that recruited only children aged <18 years, and those whose main focus was neither multimorbidity nor primary care data and/or settings were excluded.

How this fits in

Multimorbidity is becoming the norm, rather than the exception, in primary care, but evidence-based interventions are scarce. As knowledge of the effects of multimorbidity over time is a necessary precursor to developing effective interventions, a systematic review of prospective cohort studies of multimorbidity in primary care was carried out. Out of 996 articles identified, only six articles from five completed studies were found that were relevant; although the studies identified provide useful information, they also demonstrate significant gaps in knowledge. To plan future healthcare services and treatment guidelines for those with multimorbidity, a better understanding of the personal experience, treatment, and health service use, as well as the psychological, physical, and social factors that influence multimorbidity over time, is needed.

Search strategy

The following databases were searched; the corresponding start date is given in parentheses:

• PubMed (1960);
• Medline (1950);
• PsycINFO (1887);
• CINAHL (1982);
• the CSA Conference Papers Index (1982);
• the Index to Scientific and Technical Proceedings (via ISI Web of Science) (1990); and
• BioMedCentral (BMC) journal study protocols (2000).

In addition, hand searches of key journals (Family Practice, BMC Health Services Research, BMC Public Health, Chronic Illness, Journal of Clinical Epidemiology) were carried out for the 12 months preceding the start of this review. All searches were carried out by one researcher on 23 March 2010. Experts in the field of multimorbidity were also contacted to help identify relevant studies; they carried out hand searches of reference lists in included studies in an attempt to identify other relevant studies.

A mixture of Medical Subject Headings (MeSH) and key words were used to search PubMed and Medline; headings and key words for CINAHL; descriptors, key words and methodology terms for PsycINFO; and topics and keywords for ISI Web of Science. Other databases including the CSA Conference Papers Index and the BMC journals database rely on keyword searches. The exact search terms for selected databases are shown in Table 1. As comorbidity and multimorbidity are not consistently defined in the literature, articles using either term were searched for and included.

Table 1.

Search terms used in main databases

Multiple searches were performed via PubMed to identify relevant papers prior to, and after, the introduction of key MeSH terms. The term ‘cohort studies’ was only introduced as a MeSH term in 1989 and ‘comorbidity’ in 1990; to find articles prior to those dates the study used different search terms, such as the MeSH terms ‘follow-up studies’ or ‘prospective studies’ instead of ‘cohort studies’, and variations of the keywords ‘comorbid’ and ‘multimorbid’ in the title or abstract.

Data extraction and analysis

All citations (title and abstract) were screened by two different reviewers. If either reviewer could not confidently include or exclude the paper based on the abstract or citation, the full paper was obtained. In total, 27 papers were read in full. All authors contributed to the double screening exercise. If there was a disagreement about whether a paper should be included or excluded, it was read by one or more additional reviewers and an agreement was reached through discussion. A data extraction sheet was used independently by two reviewers and compared for consistency; again, any disagreements were resolved through discussion.

The study adhered to the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) statement to ensure our review was of good quality.

RESULTS

Eight prospective cohort studies on multimorbidity in primary care settings that were described in nine papers were identified from a total of 996 articles. Three protocol papers14–16 were excluded, leaving six papers, which related to five separate cohort studies (Figure 1).17–22

Nature and scope of studies

Study aims

Three studies19,21,22 focused on healthcare utilisation and/or costs, but also included some patient outcomes (severity of disease,22 new morbidity,19 and mortality22). One study focused solely on patient outcomes (physical decline),20 while another (written up in two papers17,18) looked at psychosocial risk factors. Full details are given in Table 2.

Table 2.

Study aims

Theoretical or conceptual frameworks

Two papers explicitly described a theoretical or conceptual framework for the study.18,20 Van den Akker et al's 2006 paper18 drew on a theory of general disease susceptibility; Bayliss et al's analyses20 were based on a conceptual interaction between long-term conditions and the ‘psychosocial environment’ that impacted on physical wellbeing. The aim was to aid clinical decision making and the management of physical decline by informing a generic chronic care model for patients with multimorbidities; implicitly, this relates to the cost to the healthcare system. The model implicit in Van den Akker et al's 200117 paper focused on psychosocial, as well as disease, factors impacting on the development of multimorbidity.

In the remaining three studies, no conceptual model was stated or implied. The impetus for these studies appeared to be to investigate the relationship between multimorbidity and resource use.19,21,22

Study location

Three studies were conducted in the US20–22 and two in The Netherlands17,18 (Table 3). None of the cohorts were multicountry but they were restricted to a single region of The Netherlands,17–19 three urban US cities,20 and the geographical area served by a single US primary care practice.21,22

Table 3.

Methods

Key overall study findings

Two studies (three papers) reported risk factors for the course of multimorbidity, including the type of disease20 and psychosocial characteristics.17,18 Van den Akker et al17,18 identified psychosocial risk factors — negative life events, an external health locus of control, and a social network of less than five people — for developing multimorbidity,17 which may predominantly apply in conditions that do not have a known common pathophysiological origin.18

One study20 found that certain combinations of chronic conditions — for example, chronic respiratory disease (CRD), congestive heart failure (CHF), and diabetes — presented a greater risk for physical decline than others, and some combinations — such as CRD and osteoarthritis — resulted in higher patient consultation rates.19

Three studies reported that patients with multimorbidities had higher healthcare utilisation19,21,22 than those with only a single condition. Increasing multimorbidity predicted higher healthcare charges in an outpatient setting and an increased likelihood of inpatient admission or death.21,22

One study suggested that a simple count of prescribed medications might have the greatest predictive validity for healthcare utilisation and costs, and diagnosis-based measures might be best for predicting 1-year mortality; however all measures had poor to modest predictive validity.23

No study had health inequalities or socioeconomic status as its major focus. Perkins et al's study21 did compare the impact of patients' income, sex, age, and ethnic origin on multimorbidity using five different measures of it; contradictory results were found, depending on the measure used.

Methodologies used

Study design and methods

Table 3 describes the methods used in the five studies. Studies varied widely in: their eligibility criteria for inclusion in the cohort; how multimorbidity was measured; which outcomes were assessed; the type of primary care setting/patient selection procedures used; and the type of data that were collected.

Selection of primary care settings and patients

None of the studies randomly selected primary care settings or GPs, then randomly selected patients. They recruited a volunteer sample of practices,20 GPs who had taken part in a previous study,19 practices registered on a database,18,19 or used a convenience sample of patients.21,22 The studies recruited between one and 15 practices. Between four and 42 GPs participated in three studies (four articles);17–19,22 225 GPs participated in another;20 and one did not state how many GPs took part.21 One study included all eligible patients from the study practices,19 one study (written up in two articles) randomly sampled patients (the method of randomisation was not stated),17,18 while the others used convenience samples.20–22

Four studies had a potentially biased sample due to: loss to followup;20 patient non-response;17,18 the study inclusion/exclusion criteria;17 or the method of sampling patients.17,18,20–22 Table 3 shows the characteristics of those patients excluded or lost due to non-response or attrition.

Multimorbidity definitions and measures

All studies operationalised multimorbidity as two ore more conditions within a patient, but not all limited the conditions to those that are long-term and the studies varied in the list of conditions that could be included. Only three of the studies (four articles) provided a clear definition.17,18,20,22

Two studies included people with less than one of five19 or six20 specific chronic diseases. In Schellevis et al's19 study, it is not clear why the specific diseases were chosen; Bayliss et al20 chose high-prevalence conditions that frequently appear in the research literature on multimorbidity or chronic disease management. Three studies (four articles) had broader inclusion criteria with few limitations on which conditions were included.17,18,21,22 Table 3 provides details of definitions and how multimorbidity was operationalised.

Sample size

None of the papers justified sample size. Cohort sizes ranged from 414 to 3745 patients at baseline and from 413 to 3551 patients at follow-up (Table 3). However, not all patients in the cohorts had, or developed, multimorbidity. One study did not state how many patients had multimorbidity,21 the number was relatively small in three studies (four articles) (n = 216,22 n = 268,19 and n = 30517,18), and one study had a larger number (n = 686).20 This meant that analyses by sub-group or sub-population (for example, type of condition, type of disease susceptibility, age, or deprivation level) were not possible or had very limited statistical power.

Patient follow-up

Four of the studies (five articles)17–19,21,22 carried out primary research; three analysed routinely collected data.19,21,22 Two papers17,18 drew on the same longitudinal dataset to carry out different analyses. One20 carried out a secondary analysis of 4-year follow-up data, which had been collected in 1990 (some 14 years previously), as part of a longitudinal study called the Medical Outcomes Study. Table 3 shows, in detail for each study, the data that were gathered and from which sources they derived. The range of outcomes measured was limited, with studies mainly appearing to rely on routinely collected data.19

The study follow-up times ranged from 1–4 years, with four of the five studies following patients for 12–24 months.17–19,21,22 One of the studies had only one follow-up point.20

Retention rates varied between 70% and 100% of the sample, depending on the follow-up methods; follow-up by record extraction resulted in little or no attrition.19,21,22 Loss to follow-up contributed to the sample being unrepresentative in one study.20

Inclusion criteria and screening procedures

All studies — except that by Perkins et al,21 which sampled on the basis of age — focused on identifying patients with clinically determined diagnoses of diseases; one also included self-reported diagnoses.20 Patients were identified by a variety of means including: physician reports verified by study clinical staff and through a patient questionnaire;20 searches of an electronic database;17,18 a GP search of records;19 or patient attendance at the practice during a specific time period.19,21,22 Further details of the inclusion criteria are given in Table 3.

DISCUSSION

Summary

This review identified five cohort studies of multimorbidity in primary care; these derived from two countries (The Netherlands and the US). Substantial variation occurred in the conditions included. Multimorbidity predicted increased health service use and costs, mortality rates, and reduced physical function. Psychosocial risk factors for multimorbidity included negative life events, external health locus of control, and small social networks, which may be most important in conditions that lack a common pathophysiological origin. Although these pioneering studies offer valuable insights, important gaps were also identified: none of the studies focused on mental illness and multimorbidity, or the interaction with socioeconomic deprivation, and patients' views were notably absent. Methodologically, a clear conceptual framework was not always apparent and no study used random sampling of general practices and patients.

Strengths and limitations

The main limitation of any systematic review is the difficulty in ensuring that all of the relevant literature has been identified. This was maximised by combining a variety of search strategies. Abstracts were required to be in English, which could have excluded potentially relevant papers, however only two papers originally identified did not fulfil this criterion. The absence of consistent indexing in databases due to the lack of a key indexing term for ‘multimorbidity’ posed a difficulty, so comorbidity — which is often used synonymously — was searched for and variations of these search terms were used.23

Comparison with existing literature

As far as the authors are aware, this is the first systematic review on this topic. The inconsistency in defining and measuring multimorbidity has been reported by others.23,24 Retrospective and cross-sectional studies support the findings on healthcare utilisation and costs, mortality, and physical functioning.6,8,25,26 Since conducting this review, two other relevant cohort studies have been identified; one on the influence of multimorbidity on cognition in an aging population in one region of The Netherlands,27 and the other on the impact of multimorbidity (as measured by the Ambulatory Care Group case mix system) on choice of primary care provider in two practices in one county of Sweden.28 However, these two recently published papers do not change our conclusions or the implications for future research outlined below.

Implications for research

The studies identified tended to be limited in scope and size, with questionable generalisability relating to issues of sampling, inclusion criteria, patient attrition and non-response. Causal pathways, prognostic factors, treatment use, patient safety, service models, quality of care, and patient perceptions and experiences were not well documented. A need to focus on socioeconomic factors in future cohort studies is important as retrospective and prevalence studies in The Netherlands,8 Scotland,29 England,9 and Ireland25,26 all suggest a significant link between low socioeconomic status and the amount and burden of multimorbidity. Future research must also explore the longitudinal links between mental illness and multimorbidity, given the growing evidence on their interconnectedness.29,30 Longitudinal studies on multimorbidity in primary care have important gaps in knowledge. A fuller understanding of personal experience, treatment burden and health service use, as well as the psychological, physical, and social factors that influence multimorbidity over time is needed.

Acknowledgments

We would like to thank the six international experts in multimorbidity research who sent us information regarding studies.

Funding

This systematic review was funded by a Scottish School of Primary Care Visiting Professor Award (grant number: 63124). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Provenance

Freely submitted; externally peer reviewed.

Competing interests

The authors have declared no competing interests.

Discuss this article

Contribute and read comments about this article on the Discussion Forum: http:http://www.rcgp.org.uk/bjgp-discuss

REFERENCES

• 1.Mather CD, Lancar D. Updated projections of global mortality and burden of disease, 2002–2030: data sources, methods, and results. Evidence and Information for Policy. Geneva: World Health Organization; 2005. [Google Scholar]
• 2.World Health Organization. The world health report 2008: Primary health care, now more than ever. Geneva: WHO; 2008. [Google Scholar]
• 3.Starfield B, Shi L, Mackinko J. Contribution of primary care to health systems and health. Milbank Q. 2005;83(3):457–502. doi: 10.1111/j.1468-0009.2005.00409.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 4.Van Weel C, Schellevis FG. Comorbidity and guidelines: conflicting interests. Lancet. 2006;367(9510):550–551. doi: 10.1016/S0140-6736(06)68198-1. [DOI] [PubMed] [Google Scholar]
• 5.Fortin M, Bravo G, Hudon C, et al. Prevalence of multimorbidity among adults seen in family practice. Ann Fam Med. 2005;3(3):223–228. doi: 10.1370/afm.272. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 6.Wolff JL, Starfield B, Anderson G. Prevalence, expenditures, and complications of multiple chronic conditions in the elderly. Arch Intern Med. 2002;162(20):2269–2276. doi: 10.1001/archinte.162.20.2269. [DOI] [PubMed] [Google Scholar]
• 7.Britt HC, Harrison CM, Miller GC, Knox SA. Prevalence and patterns of multimorbidity in Australia. Med J Aust. 2008;189(2):72–77. doi: 10.5694/j.1326-5377.2008.tb01919.x. [DOI] [PubMed] [Google Scholar]
• 8.Salisbury C, Johnson L, Purdy S, et al. Epidemiology and impact of multimorbidity in primary care: a retrospective cohort study. Br J Gen Pract. 2011 doi: 10.3399/bjgp11X548929. DOI: 10.3399/bjgp11X548929. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 9.Mercer SW, Smith SM, Wyke S, et al. Multimorbidity in primary care: developing the research agenda. Fam Pract. 2009;26(2):79–80. doi: 10.1093/fampra/cmp020. [DOI] [PubMed] [Google Scholar]
• 10.Smith SM, Soubhi H, Fortin M, et al. Cochrane Database of Systematic Reviews: Protocols. Chichester, UK: John Wiley & Sons, Ltd; 2007. Interventions to improve outcomes in patients with multimorbidity in primary care and community settings. Issue 2. [Google Scholar]
• 11.Gordis L. Epidemiology. 2nd edn. Philadelphia, PA: WB Saunders & Co; 2000. [Google Scholar]
• 12.Moons K, Royston P, Vergouwe Y, et al. Prognosis and prognostic research: what, why, and how? BMJ. 2009;338:b375. doi: 10.1136/bmj.b375. [DOI] [PubMed] [Google Scholar]
• 13.Fortin M, Lapointe L, Hudon C, et al. Multumorbidity and quality of life in primary care: a systematic review. Health Qual Life Outcomes. 2004;2:51. doi: 10.1186/1477-7525-2-51. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 14.Van den Akker M, Spigt M, De Raeve L, et al. The SMILE study: a study of medical information and lifestyles in Eindhoven, the rationale and contents of a large prospective dynamic cohort study. BMC Public Health. 2008;8:19. doi: 10.1186/1471-2458-8-19. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 15.Haggerty J, Fortin M, Beaulieu M, et al. At the interface of community and healthcare systems: a longitudinal cohort study on evolving health and the impact of primary healthcare from the patient's perspective. BMC Health Serv Res. 2010;10:258. doi: 10.1186/1472-6963-10-258. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 16.Schafer I, Hansen H, Schon G, et al. The German MultiCare-study: patterns of multimorbidity in primary health care — protocol of a prospective cohort study. BMC Health Serv Res. 2009;9:145. doi: 10.1186/1472-6963-9-145. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 17.Van den Akker M, Buntinx F, Metsemakers JF, et al. Psychosocial patient characteristics and GP-registered chronic morbidity: a prospective study. J Psychosom Res. 2001;50(2):95–102. doi: 10.1016/s0022-3999(00)00227-0. [DOI] [PubMed] [Google Scholar]
• 18.Van den Akker M, Vos R, Knottnerus JA. In an exploratory prospective study on multimorbidity general and disease-related susceptibility could be distinguished. J Clin Epidemiol. 2006;59(9):934–939. doi: 10.1016/j.jclinepi.2006.02.009. [DOI] [PubMed] [Google Scholar]
• 19.Schellevis FG, Van de Lisdonk EH, Van der Velden J, et al. Consultation rates and incidence of intercurrent morbidity among patients with chronic disease in general practice. Br J Gen Pract. 1994;44(383):259–262. [PMC free article] [PubMed] [Google Scholar]
• 20.Bayliss EA, Bayliss MS, Ware JE, Jr, Steiner JF. Predicting declines in physical function in persons with multiple chronic medical conditions: what we can learn from the medical problem list. Health Qual Life Outcomes. 2004;2:47. doi: 10.1186/1477-7525-2-47. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 21.Perkins AJ, Kroenke K, et al. Common comorbidity scales were similar in their ability to predict health care costs and mortality. J Clin Epidemiol. 2004;57(10):1040–1048. doi: 10.1016/j.jclinepi.2004.03.002. [DOI] [PubMed] [Google Scholar]
• 22.Parkerson GR, Jr, Broadhead WE, Tse CK. Health status and severity of illness as predictors of outcomes in primary care. Med Care. 1995;33(1):53–66. doi: 10.1097/00005650-199501000-00005. [DOI] [PubMed] [Google Scholar]
• 23.Van den Akker M, Buntinx F, Knottnerus J. Comorbidity or multimorbidity: what's in a name? A review of the literature. Eur J Gen Pract. 1996;2(2):65–70. [Google Scholar]
• 24.Valderas JM, Starfield B, Sibbald B, et al. Defining comorbidity: implications for understanding health and health services. Ann Fam Med. 2009;7(4):357–363. doi: 10.1370/afm.983. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 25.Smith S, Ferede A, O'Dowd T. Multimorbidity in younger deprived patients. An exploratory study of research and service implication in general practice. BMC Fam Pract. 2008;9:6. doi: 10.1186/1471-2296-9-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 26.Glynn LG, Valderas JM, Healy P, et al. The prevalence of multimorbidity in primary care and its effect on health care utilization and cost. Fam Pract. 2011;28(5):516–523. doi: 10.1093/fampra/cmr013. [DOI] [PubMed] [Google Scholar]
• 27.Aarts S, van den Akker M, Tan FE, et al. Influence of multimorbidity on cognition in a normal aging population: a 12-year follow-up in the Maastricht Aging Study. Int J Geriatr Psychiatry. 2011;26(10):1046–1053. doi: 10.1002/gps.2642. [DOI] [PubMed] [Google Scholar]
• 28.Zielinski A, Håkansson A, Beckman A, Halling A. Impact of comorbidity on the individual's choice of primary health care provider. Scand J Prim Health Care. 2011;29(2):104–109. doi: 10.3109/02813432.2011.562363. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 29.Mercer SW, Watt GCM. The inverse care law: clinical primary care encounters in deprived and affluent areas of Scotland. Ann Fam Med. 2007;5(6):503–510. doi: 10.1370/afm.778. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 30.Gunn JM, Ayton DR, Densley K, et al. The association between chronic illness, multimorbidity and depressive symptoms in an Australian primary care cohort. Soc Psychiatry Psychiatr Epidemiol. 2012;47(2):175–184. doi: 10.1007/s00127-010-0330-z. [DOI] [PubMed] [Google Scholar]