Our results show the prevalence trends of multimorbidity community-dwelling adults aged 50 and over, living in 15 European countries. The observed data reveal an increase in the prevalence of multimorbidity in the countries of central Europe (Austria, Belgium, Czech Republic, France, Germany and Switzerland) and Spain in men and women, and in the Netherlands in men. Stability was observed in the analysis by periods in some southern (Portugal and Italy), northern (Denmark and Sweden) and eastern (Estonia, Poland and Slovenia) European countries. In the case of Portugal, Estonia, Poland and Slovenia, the short period limits the results of the analysis. However, in some of these countries it was possible to identify changes when analysing age groups and prevalence by disease group.
The prevalence of multimorbidity was higher in the Czech Republic, Poland and Portugal, whereas the countries with the lowest multimorbidity prevalence were Switzerland, Sweden and the Netherlands. Similar differences were found by a study using SHARE data from wave 516. However, it is necessary to make some methodological considerations when comparing data with other studies. Firstly, the definition of multimorbidity is heterogeneous, which makes it difficult to compare prevalence data17,18. The presence of two or more chronic conditions is the most common definition19, but the number of included non-communicable diseases varies between 4 and 102 in the literature20. Therefore, the changes in prevalence may occur as a result of diseases that were included in the composition of the multimorbidity definition in a time series study. Another aspect that should be considered is the method used to collect the information. Some studies have used clinical records in hospitals or primary care units, while others collected information through self-report14,21-23. Self-report is the most feasible method for population-based epidemiological studies, but it has the potential to underestimate the prevalence24.
A study conducted in Nijmegen, the Netherlands, analysed the prevalence trend of multimorbidity for the period 1985 to 2005 through a morbidity register of 13,584 primary care patients, without an age limit. The proportion of patients without chronic diseases decreased from 70% to 63%, and the proportion of individuals with one chronic disease was stable while those with two increased from 6.7 to 8%25. Two other studies also carried out in the Netherlands found an increase in the prevalence of multimorbidity. Tacken et al.7 included diabetes mellitus as well as cardiovascular and respiratory diseases, and found an increase in prevalence during the period 2003 to 2009. As did Oostrom et al., who considered the co-occurrence of two or more of 28 diseases during the period 2001-20116.
Prevalence of multimorbidity has also been increasing in non-European countries. A trend analysis was realised in Hong Kong with 69,636 adults aged 35 or over who participated in the surveys in 1999, 2001, 2005 and 2008. Multimorbidity was defined as presenting two or more chronic conditions from a list of 148. Further, a study conducted in Canada with data of 1996-97, 2005 and 2012-13 identified an increase in the prevalence of multimorbidity that may be attributed to an increase of obesity26. A pooled analysis of individuals from cohort studies from the USA and Europe shows that the risk of cardiometabolic multimorbidity increases as BMI increases27.
Overall, the increase or reduction in the prevalence of multimorbidity could be explained by two circumstances: changes in the prevalence of the main risk factors, such as tobacco and alcohol consumption, diet or practice of physical activity; or changes in the classification system and/or improvement in diagnosis. Regarding the first situation, a study that also analysed data from the SHARE project identified a non-significant increase of obesity and overweight in France, Switzerland and Denmark. On the other hand, a significant decrease in the prevalence of overweight was observed for Spain (-3.5, 95% CI: -6.1 to -0.9) and Italy (-4.3, 95% CI: -7, 3 to -1.3). For obesity, there was a significant increase in Germany (5.8, 95% CI: 1.8-9.9) and a significant reduction in Spain (-4.7, 95% CI: -8.8 to -0.5)28 These results provide clues that could contribute to explaining the differences in the multimorbidity trends that we find by countries, but not the increase in Spain, and the changes in musculoskeletal diseases trends, which has obesity as a risk factor.
The analysis by disease group showed more changes in the prevalence of multimorbidity attributed to cardiometabolic, respiratory and musculoskeletal diseases. The prevalence of cardiometabolic diseases is highest among those with multimorbidity, with a significant reduction only in Portugal. In a global study analysing trends of prevalence of diseases and years lived with disability (YLDs) by major cause groups, the main burden was attributed to musculoskeletal, mental, neurological and respiratory conditions1.
The prevalence of neurodegenerative diseases increased in four countries analysed. Neurodegenerative diseases in men increased in Estonia and Slovenia. In women, the increase was observed in Estonia, Slovenia, Italy and Spain. A longitudinal study on Alzheimer's disease verified an increase in neurodegenerative diseases in most European countries from 1994 to 2013. This study analysed mortality trends, identifying an increasing trend in both sexes in all countries except Germany and Malta, where a reduction was observed. The authors presented two main hypotheses for this finding: better diagnoses attributed to the implementation of ICD-10; and the contribution of environmental factors29.
Musculoskeletal diseases presented an increase in northern and central countries (Belgium, Czech Republic, France, Germany, Sweden and Switzerland) and a reduction in Italy and Portugal. As observed in our study, different trends of prevalence of osteoarthritis or rheumatoid arthritis can be found in the literature30. A study conducted in the United States described a reduction in the prevalence of rheumatoid arthritis and an increase for osteoarthritis31, and another identified stability for rheumatoid arthritis in Canada32. These variations are probably related to risk factors such as obesity and genetic factors33.
Several studies show that the prevalence of multimorbidity increases with age22,34,35. Stratified analysis for 10-year age groups indicates in our results a possible cohort effect in some countries, such as the reduction among men and women in Italy for the younger age groups (50-59 and 60-69) and increase in men in the 80 and over age group. Furthermore, the increase in multimorbidity occurred in all age groups in men in Austria, Czech Republic and Spain, and in women in Germany. Chatterji et al., analysing SHARE data, did not find any cohort effect when analysing limitations in the activities of daily living, a strong indicator of multimorbidity. However, the study analysed the pooled cohort effect and no stratification by country was applied36.
Concerning the limitations of our study, we emphasise that this is an ecological study, and therefore it is not possible to establish a causal relationship between the exposure to risk factors and the prevalence of each disease. Another aspect to consider is the sampling of the oldest age group in the SHARE project, which may affect comparability across countries among those aged 80 and over37. Similar to other studies, we decided to include this age group due to its relevance to the issue. Another limitation is self-reported information and sociocultural differences between countries (e.g. language). However, these difficulties have been compensated by the harmonisation and standardisation of the questionnaires and data collection procedures, guaranteed by the meticulous process of cultural adaptation, as well as by the high professionalism of highly trained researchers and interviewers.
Regarding the strengths of our work, we highlight that this study on multimorbidity prevalence trends included the largest number of countries with detailed analysis by age and disease groups. Additionally, one advantage of a longitudinal study such as SHARE is the quality and quantity of data of different waves and countries due to the standardisation of data collection.