Our analyses show that multimorbidity is common, especially among older persons, and that among those with a chronic disease over two thirds has other comorbidities. A more detailed look at the combinations of diseases showed a wide variety in multimorbidity. We observed that all chronic diseases tend to cluster, i.e. most disease pairs co-occur more often than expected by chance. Furthermore, the top-5 of co-occurring chronic diseases represents only a minor proportion of the comorbidity: all diseases co-occur together. These findings may have important consequences since the organization and funding of health care is organized by disease-specific programs in many countries. Such disease-specific approaches do not match the reality of most people with a chronic disease.
Strengths and weaknesses of the study
Primary care registries represent an interesting source to describe multimorbidity of chronic diseases because often most health problems are known and recorded by the general practitioner, especially in countries as the Netherlands where general practice is the entry point for health care and general practitioners act as gatekeeper for secondary care. Analyzing multimorbidity among general practice patients can be used for different research questions [3, 4, 19, 20] but ideally we need to know the underlying population or epidemiological denominator. Strength of general practice registries in the Netherlands is that almost everybody is registered within a general practice. Hence, it is possible to use morbidity data from the general practice registries to describe morbidity of the general population.
In addition, use of long-term registration data of a country-wide network of general practices for the analyses on multimorbidity has several advantages, like the distribution of the general practices over the Netherlands, the relatively long registration period, and the standardized registration procedures. A general strength of general practice medical records compared to self-reported data is the availability of diagnosed chronic diseases. A disadvantage is that diseases for which patients do not consult a general practitioner are not in the general practitioners’ medical record system. In addition, the small group of elderly in nursing homes, usually having more than one chronic disease, are generally not registered within a general practice. Owing to these exclusions the prevalence of chronic diseases in the general population may be underestimated based on general practice registrations. Underdiagnosis and underreporting of health problems like depression may also lead to underestimation of the prevalence of multimorbidity.
Length of follow-up in a registration affects the extent and the reliability of multimorbidity estimates . The frequency of general practitioner consultations for some diseases, such as osteoarthritis, are even less than once a year. The prevalence of those diseases is underestimated in a registration with a one-year follow-up, therefore the minimum patient follow-up in our study was three years. With a longer follow-up period selection may play a role since patients move away and general practices drop out from the registration.
Apparently, the prevalence of multimorbidity in this study is completely determined by the selection of 29 chronic diseases. Generally speaking, the more chronic conditions are included the higher prevalence rates of multimorbidity will be found. We presume that an important part of chronic morbidity is included in our selection of diseases. By counting the number of chronic diseases we did not take into account any differences in the severity of diseases . When interpreting the results of our and similar studies that are based on general practice registrations, it should be noticed that the varying consequences of diseases on patients’ physical and mental functioning, disability and independency are not considered [8, 23].
Comparison with literature
Registry characteristics affect the prevalence and nature of multimorbidity, especially the selection and definition of diseases affect the actual prevalence rates [21, 24]. This limits comparisons with other studies, even those based on general practice registries. However, the general patterns we found are similar to those of other studies. First, multimorbidity being rather the rule than the exception and prevalences increasing with age but also found at younger age [3–8, 25–27]. Second, clustering of diseases is commonly found in all recent studies on multimorbidity of (chronic) diseases [5, 28–32]. Third, the most prevalent diseases like heart disease, diabetes and osteoarthritis end up high in every multimorbidity rank [4, 29]. Most studies showed hypertension, obesity and hyperlipidemia to rank high in prevalence when it was defined as a chronic disease [21, 28, 29]. To our opinion these are chronic conditions that increase the risk on chronic disease but are not diseases, we therefore did not include these in our analyses. Fourth, a majority of persons with one chronic disease also have at least one (and often more) other disease and these are not limited to a few common diseases, these can be any disease [6, 28, 29]. While a descriptive approach was used in our analyses and others [6, 29] few other studies performed cluster analyses to identify comorbidity patterns [28, 30–32]. These studies identified 3 to 6 clusters of diseases and most revealed a cluster with vascular conditions and a cluster with mental diseases along with pain [28, 30, 32]. We focused on common co-occuring conditions and our findings indicate a wide variety of co-occuring conditions since only 30% of the comorbidity spectrum can be attributed to the 5 most common comorbidities. In line with this, van den Bussche described that combinations of the six most prevalent chronic conditions span only 42% of the comorbidity spectrum .
Implications of the study
A substantial proportion of the older population being characterized by multimorbidity of chronic diseases requires reconsiderations of medical research as well as the organization of care. Most research and clinical practice are based on a single disease paradigm, which may not be appropriate for patients with multiple complex health problems [33, 34]. Studies investigating the (cost-)effectiveness of new treatments commonly exclude patients with multimorbidity. Findings from these studies have a very limited reach since multimorbidity affects the majority of the aged. There is thus a clear need to shift research from disease-specific treatments and patients with single diseases to research which takes into account multimorbidity or is explicitly focused on patients with multimorbidity .
Multimorbidity leads to complex care through the use of different treatment strategies and the involvement of various health care professionals, which may lead to opposing advices (counseling) or medications . Due to fragmented expertise and focus by different care providers, patients preferences, expectancies, values and needs regarding their daily life are often overlooked . Patients’ main priorities are usually an adequate quality of life and appropriate daily functioning in addition to the improvement of disease-specific health problems.
Currently, disease management programs are implemented worldwide in order to enhance quality and continuity of care for the chronically ill [38, 39]. These programs are characterized by a patient-centered approach of coordinated multiple healthcare interventions that structure chronic care to a specific patient group [39, 40]. However, participating in multiple single-disease oriented programs in combination with regular primary care, may lead to fragmented care. In designing these programs insufficient attention is paid to multimorbid conditions. Multimorbid patients are therefore at risk for suboptimal treatment, unsafe care, inefficient use of health care services, unnecessary costs and consequently run higher risks for adverse events . Case management is a potential model which might counteract fragmented care for multimorbid patients. It is an individualized care program which coordinates all care involved for patients enrolled in different single-disease management programs, who have to adhere to various treatment protocols. It draws on evidence-based optimal care for systematically managing all existing conditions in a patient, and is tailored to the individual patients’ preferences [42–44].
The finding that multimorbidity cannot be captured with a few common combinations of diseases represents a challenge for disease-specific treatment as well as disease-specific disease management programs and rises the question for which patient group case management should be implemented. Given the enormous variety in multimorbidity, we need more knowledge about the co-occurring conditions that are leading to a high need for care, a major decline in quality of life, and/or an increase in functional limitations.