Prevalence estimates of multimorbidity: a comparative study of two sources

Martin Fortin 0 3 Catherine Hudon 0 3 Jeannie Haggerty 2 Marjan van den Akker 1 Jos Almirall 0 3 0 Department of Family Medicine, Sherbrooke University , Sherbrooke, Quebec , Canada 1 Department of General Practice, Care and Public Health Research Institute, Maastricht University , Maastricht , The Netherlands 2 Department of Community Sciences, Sherbrooke University , Sherbrooke, Quebec , Canada 3 Department of Family Medicine, Sherbrooke University , Sherbrooke, Quebec , Canada Background: Published prevalence studies on multimorbidity present diverse data collection methods, sources of data, targeted age groups, diagnoses considered and study populations, making the comparability of prevalence estimates questionable. The objective of this study was to compare prevalence estimates of multimorbidity derived from two sources and to examine the impact of the number of diagnoses considered in the measurement of multimorbidity. Methods: Prevalence of multimorbidity was estimated in adults over 25 years of age from two separate Canadian studies: a 2005 survey of 26,000 respondents randomly selected from the general population and a 2003 study of 980 patients from 21 family practices. We estimated the prevalence of multimorbidity based on the co-occurrence of 2 and 3 diseases of the seven diseases listed in the general population survey. For primary care patients, we also estimated multimorbidity prevalence using an open list of chronic diseases. Results: Prevalence estimates were considerably higher for each age group in the primary care sample than in the general population. For primary care patients, the number of chronic diseases considered for estimates resulted in large differences, especially in younger age groups. The prevalence of multimorbidity increased with age in both study populations. Conclusions: The prevalence of multimorbidity was substantially lower when estimated in a general population than in a family practice-based sample and was higher when the number of conditions considered increased. - Background As a result of different factors, including aging populations and advances in medical care and public health policy, a growing proportion of patients present multiple coexistent diseases, or multimorbidity [1]. To estimate the magnitude of this problem, studies about the prevalence of multimorbidity have been conducted in different parts of the world: in Europe [1-5], the Middle East [6], Australia [7], the United States [8-10], and Canada [1113]. These studies present diverse sources of data (e.g., questionnaires, chart reviews, administrative data), data collection methods, targeted age groups, diagnoses considered and study populations, making the comparability of prevalence estimates questionable. Some studies [5,6,8-11] focused on data from samples of the general population; others [1-3,7,13] from primary care practices. In a recent study [4] on the prevalence of multimorbidity in four different settings (population-based, general practice, hospital and nursing home), the authors concluded that the study population in research on the prevalence of multimorbidity required special attention. Their study, however, was limited to people 55 years of age, and the diseases considered for prevalence estimates of multimorbidity varied in the different study populations. Furthermore, information about all morbidities (complaints, diseases and disorders) was collected from medical charts for patients in general practice and from a limited list of chronic conditions for the general population [4]. We believe that differences among the diagnoses considered could have influenced estimates of the prevalence of multimorbidity. Analysis of how the prevalence of multimorbidity in the general population is reflected in clinical practice has not yet been done. To make a valid comparison of prevalence estimates, data should be obtained simultaneously or within a short time frame from both general and primary care populations. In addition, the medical conditions taken into account and the operational definition of multimorbidity used should be consistent. Data from the general population, although useful for a comparison of different geographical areas, provide an incomplete picture of where chronic diseases are dealt with, for example, in the offices of family doctors. A recent publication [14] in which prevalence estimates of multimorbidity were calculated using data obtained in 2005 from the general population of the province of Quebec, Canada, allowed us to make a comparison with prevalence estimates of multimorbidity using data from a 2003 family practice-based study [13] in the same province. The aim of the current study was to compare ageand sex-specific estimates of the prevalence of multimorbidity derived from these two populations, and to examine the impact of a variation in the number of different diagnoses considered on prevalence estimates within the practice-based population. Methods The prevalence of multimorbidity in the general population of Quebec comes from a Quebec publication [14] based on data from the Canadian Community Health Survey (CCHS). The CCHS is a series of general surveys that Statistics Canada http://www.statcan.gc.ca has carried out since 2000. Approximately 132,000 randomlyselected Canadians (26,000 from Quebec) 12 years and older participate, either in person or by telephone, in a 45-minute computer-assisted interview. For the current study, we used data collected in the 2005 CCHS [14] on the self-reported presence of seven diseases: arthritis, cancer, diabetes mellitus, hypertension, heart disease, obstructive lung disease and psychiatric problems. We limited calculations to results from participants aged 25 years and over to permit comparisons. We obtained family practice-based data from a study conducted by our research group in the Saguenay region of Quebec in 2003 [13]. In this study, trained research staff extracted diagnoses of chronic diseases from patients' medical charts. Details of the methods and sampling strategies used in this study are described elsewhere [13]. The regional ethics committee approved the original Saguenay study. In brief, 980 (90%) of 1085 consecutive adult patients solicited during successive consultation periods from 21 family physicians' practices consented to participate. For the current study, we limited chronic diseases to the same seven diseases included in the CCHS. We also calculated estimates of the prevalence of multimorbidity in the family-practice patients with an open list of chronic diseases, i.e., all the diseases that might have appeared in the patients' medical records. We retained the two operational definitions of multimorbidity used in the CCHS report [14]; that is, we defined multimorbidity as (1) 2 diseases and (2) 3 diseases. Both definitions were previously used in several other studies [13-17]. Disease complications were considered as distinct diseases. For the comp (...truncated)