Article Text
PDF
Abstract
Introduction Multimorbidity, defined as the co-occurrence of two or more long-term medical conditions, is an increasing public health concern worldwide causing enormous burden to individuals, healthcare systems and societies. The most effective way of decreasing the burden caused by multimorbidity is to find tools for its successful prevention but gaps in research evidence limit capacities to develop prevention strategies. The aim of the MOLTO study (Multimorbidity - identifying the most burdensome patterns, risk factors and potentials to reduce future burden) is to provide novel evidence required for cost-effective prevention of multimorbidity by defining the multimorbidity patterns causing the greatest burden at the population level, by examining their risk and protective factors and by estimating the potentials to reduce the future burden.
Methods and analysis The MOLTO study is based on the data from the Finnish population-based cross-sectional (FINRISK 2002–2012, FinHealth 2017 the Migrant Health and Well-being Study 2010–2012) and longitudinal (Health 2000/2011) health examination surveys with individual-level link to administrative health registers, allowing register-based follow-up for the study participants. Both cross-sectional and longitudinal study designs will be used. Multimorbidity patterns will be defined using latent class analysis. The burden caused by multimorbidity as well as risk and protective factors for multimorbidity will be analysed by survival analysis methods such as Cox proportional hazards and Poisson regression models.
Ethics and dissemination The survey data have been collected following the legislation at the time of the survey. The ethics committee of the Hospital District of Helsinki and Uusimaa has approved the data collection and register linkages for each survey. The results will be published as peer-reviewed scientific publications.
- epidemiology
- public health
- cardiology
- diabetes & endocrinology
- mental health
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Strengths and limitations of this study
The study is based on the data from the large, population-based health examination surveys with standardised, reliable methodology and possibility to follow-up the participants via health registers.
A possibility to analyse a large variety of measures of burden caused by multimorbidity as well as potential risk and protective factors for multimorbidity.
The definition of multimorbidity will be based on the multimorbidity patterns, which cause the greatest burden at the population level to produce the results with high scientific and public health impact.
Despite the large population-based data sets, low prevalence of certain chronic diseases may limit the identification of the multimorbidity patterns, especially when studying the differences between the population subgroups.
Introduction
Multimorbidity, usually defined as the co-occurrence of two or more long-term medical conditions, is an increasing phenomenon wordwide.1–3 The prevalence of multimorbidity increases strongly with the increase in age being nearly 100% in older persons.4–6 On the other hand, multimorbidity is comparatively common also among young and middle-aged adults especially in socioeconomically deprived populations.7 Multimorbidity is a major challenge for healthcare systems causing enormous costs to societies.1 8 From an individual perspective, multimorbidity is associated with disability and functional decline as well as reduced quality of life and life expectancy.8 Further, multimorbidity is associated with higher overall vulnerability to diseases and decreased resistance to acute health threats.2
From a methodological perspective, major challenge concerning multimorbidity research is that no clear consensus exists for measuring multimorbidity: number and types of chronic conditions included as well as research settings, designs, data sources and methodology vary widely between the different studies.9 10 This impairs the comparability of the estimates on multimorbidity prevalence and burden across the countries and studies. A systematic review of over 500 studies on multimorbidity published in 2021 highlighted that reporting should be improved by stating clearly which conditions were included in multimorbidity measurement, their clinical code sets and why these conditions were chosen.9
Knowledge of the most common clusters of diseases (ie, multimorbidity patterns) and which of them are the most burdensome from the individual and societal perspectives both in general populations and in certain population groups is still limited.1 4 This restricts possibilities to identify those individuals with a single disease who are at the greatest risk to develop another one as well as development and evaluation of intervention strategies designed specifically to prevent the relevant chronic conditions simultaneously.1 While some multimorbidity patterns are well known, such as cardiorespiratory or metabolic patterns, others need further identification, especially those including both, somatic and psychiatric diseases.11–14
Well-established sociodemographic risk factors for multimorbidity include older age and lower socioeconomic status.1 6–8 14 The results concerning sex are somewhat contradictory, although, according to systematic reviews6 8 women tend to have higher risk for multimorbidity compared with men. Regarding lifestyle-related factors, the evidence is still contradictory, possibly due to methodological differences between the studies, and mainly based on cross-sectional data.1 Smoking, physical inactivity and obesity have been identified as potential risk factors for multimorbidity in several studies,1 4 but the associations between alcohol consumption or nutrition and multimorbidity have been studied less.1 Some studies have focused on individual dietary factors, like fruit and vegetable intake, but the results are mixed.15–17 The association between quality of diet and multimorbidity is still poorly known. Furthermore, despite the fact that lifestyle-related risk factors are typically clustering,18 the research focusing on the combined effects of different lifestyle-related factors is still limited and based on cross-sectional design,19 specific disease clusters16 20 or older populations.17 Furthermore, the interactions between the lifestyle-related risk factors and genetic risk of multimorbidity are poorly known.
Systematic reviews have summarised the main consequences of multimorbidity, including reduced self-rated health and quality of life, functional decline and disability, increased risk of mortality and increased use of health services and costs.1 4 6 21 Different disease combinations may affect differentially these outcomes. There is also evidence that the higher the number, and the more severe the diseases are, the greater the negative impact.21 Some studies have also indicated that certain clusters of diseases may have greater negative impact on these outcomes than could be predicted based on the sum of the individual conditions.22
The most effective way of decreasing the burden caused by multimorbidity is to find tools for its successful prevention.3 Despite the rapidly growing number of research publications on multimorbidity,23 there are still gaps in our knowledge causing challenges to identify the risk groups for multimorbidity, to develop the cost-effective prevention strategies as well as to develop healthcare services to the special needs of multimorbid patients. In addition to the lack of agreed definition of multimorbidity and other heterogeneity in measurements described above, several other research gaps have been identified.1 3 10 23 The present study, ‘Multimorbidity identifying the most burdensome patterns, risk factors and potentials to reduce future burden’ (the MOLTO study) aims to fill the following research gaps:
The majority of the previous studies have been conducted in an observational, cross-sectional design. The MOLTO study will be conducted in population-based, longitudinal design to clarify the temporal aspects of the associations and to provide nationally representative results.
Most previous studies have focused on older populations. In addition to older populations, the MOLTO study will provide more information on the causes and consequences of multimorbidity among younger age groups.
In previous research, the definition of multimorbidity is highly heterogeneous, and in many studies, based on counting of diseases. The MOLTO study will deepen the analysis by revealing the multimorbidity patterns, which cause the greatest burden in the general population and its subgroups. The burden will be defined in several perspectives including mortality, functional decline and work disability, quality of life as well as healthcare utilisation and costs. The coexistence of both physical and mental diseases will be taken into account. Identifying multimorbidity patterns, leading to the most burdensome consequences, is essential to improve health services and to develop care guidelines.
The comprehensive picture of the risk and protective factors for multimorbidity is still unclear. The MOLTO study aims to provide more evidence by identifying the genetic, sociodemographic, lifestyle-related and biological pathways to multimorbidity. Furthermore, the combined effects and the interactions between the risk factors will be studied in detail. The results will help to develop new tools for early identification of persons at risk of multimorbidity as well as determination of the factors and their combinations, which should be prioritised to improve cost-effective prevention of multimorbidity. Furthermore, the results can be used to identify vulnerable population groups to target public health interventions and health promotion actions to population subgroups, where most multimorbidity cases could potentially be prevented.
The MOLTO study will produce reliable projections of the future multimorbidity burden and the related needs for social and healthcare services to enable the development of evidence-based public health programmes to reduce the future burden of multimorbidity.
The overall aim of the study is to provide up-to-date, reliable and novel evidence required for cost-effective prevention of multimorbidity. We will focus on the multimorbidity patterns, which cause the greatest burden in the adult population to produce results with high scientific and public health impact. The study process is summarised in figure 1. Specific aims and research questions are as follows:
Aim 1: to identify the multimorbidity patterns that cause the greatest burden among Finnish adults
Study process for the MOLTO study (Multimorbidity - Identifying the most burdensome patterns, risk factors and potentials to reduce future burden).
Q1: Which chronic diseases cause the greatest burden in Finland?
Q2: What kind of multimorbidity patterns will be identified based on these diseases? Which of these multimorbidity patterns cause the greatest burden?
Q3: Is the burden caused by these multimorbidity patterns greater than could be expected on the basis of the burden caused by the individual diseases?
Q4: Does the burden caused by these multimorbidity patterns differ between the population groups?
Aim 2: to examine the main sociodemographic, lifestyle and biological risk and protective factors of the most burdensome multimorbidity patterns
Q5: Which sociodemographic, lifestyle and biological factors are associated with the risk of multimorbidity?
Q6: How do these risk factors cluster and interact?
Q8: Does the genetic risk of multimorbidity patterns defined by polygenic risk scores modify the association between other risk factors and multimorbidity?
Aim 3: to examine how to best reduce the future burden of multimorbidity
Q9: How will multimorbidity patterns evolve by 2050?
Q10: How influencing the risk factors of multimorbidity could change these projections and reduce future burden caused by multimorbidity?
Methods and analysis
Health examination surveys
The MOLTO study is conducted between 1 January 2020 and 31 December 2026. It is based on the pooled data from the following Finnish population-based health examination surveys (HESs) coordinated by Finnish Institute for Health and Welfare (THL). The sampling designs of the HESs were developed with the main aim that the individuals selected are representative of the target population. The samples have been drawn from nationwide population registers, covering all residents in Finland.24
The longitudinal Health 2000/2011 Survey.25 26 A nationally representative two-stage stratified cluster sample of the Finnish adult population examined in 2000–2001 and re-examined in 2011–2012. The target population consisted of individuals aged 18 or over and living in mainland Finland. The sample frame was regionally stratified according to five university hospital regions (strata): Helsinki, Turku, Tampere, Kuopio and Oulu. In the first stage of sampling, 80 health centre districts, 16 from each university hospital regions, were sampled as a cluster. The 15 towns with the largest populations were selected with probability 1 and the sample size for each health centre district was proportional to its population. Other 65 health centre districts were selected using systematic sampling with probabilities proportional to size (PPS-SYS design). Systematic random sampling was used to draw the sample from each health centre district using the Finnish Population Information System. This two-stage stratified cluster sample represents the adult population living in mainland Finland.
The cross-sectional FINRISK studies conducted in 2002, 2007 and 2012.27 The target population of the FINRISK studies consisted of individuals aged 25–74 years of selected areas in Finland. Each year, an independent random sample of Finnish adults from selected areas (North Karelia and Northern Savo from Eastern Finland, cities of Turku and Loimaa from Southwestern Finland, Helsinki and Vantaa from Southern Finland, Northern Ostrobothnia and Kainuu in Northern Finland) was drawn from the Finnish Population Information System. These different geographical areas in Finland cover nearly half of the population.
The cross-sectional FinHealth 2017 Study.28 A nationally representative two-stage stratified cluster sample of the Finnish adult population examined in 2017. The target population consisted of individuals aged 18 or over and living in mainland Finland. The sampling design was based on the Health 2000 sampling design. In FinHealth 2017 Study, 50 health centre districts out of the 80 health centre districts of the Health 2000 were selected.
The cross-sectional Migrant Health and Wellbeing Study (Maamu).29 A large-scale population survey on the health and well-being of adults of Russian, Somali and Kurdish origin. The three groups were selected to represent different kinds of large foreign-born groups in Finland. The sample was randomly selected from the Finnish Population Information System, including individuals from selected large Finnish cities (Helsinki, Espoo, Vantaa, Turku, Tampere and Vaasa). The cities were selected from the metropolitan area and other parts of the country, with a higher proportion of foreign-born persons than in most of the other areas. Maamu data will be used in the substudies where migrant background will be taken into account.
Sample sizes, participation rates and age ranges of the HESs are presented in table 1. All HESs included a comprehensive health examination with blood sampling, blood pressure measurements and anthropometric measurements (tables 2 and 3). All other HESs except FINRISK surveys included also functional capacity tests. Furthermore, all HESs included self-administered questionnaires and/or interviews to gather information on sociodemographic factors, lifestyle, quality of life, healthcare utilisation, functional capacity and work ability as well as diagnosed chronic diseases and medication. Measurements in these HESs were conducted using standardised30 and mainly comparable methods allowing us to combine different data sets and use pooled data.
Characteristics of the health examination survey data
Health examination survey data on mental health, quality of life, physical functioning, work ability and healthcare utilisation
Health examination survey data on lifestyle and biological risk factors
National registers
Register-based data will be used both as aggregated population level data to define the chronic diseases, which cause the greatest burden among Finnish adults and individually linked to HES data using personal identity code. The following Finnish administrative register-based data will be used:
Causes of death register: dates, primary and contributory causes of deaths (Statistics Finland).31
The Register of Completed Education and Degrees32 and socioeconomic status,33 occupation34 (Statistics Finland).
Care Register for Health Care: dates and diagnoses of hospitalisations and outpatient visits within public healthcare, including primary care since 2011 (THL).35
Finnish Cancer Registry: date and diagnoses of cancers.36
Registers of the Social Insurance Institution (Kela): entitlement to specially reimbursed medications due to specific chronic conditions, purchase of medicines, sickness absence, disability allowance, rehabilitation.37
Registers of then Finnish Centre for Pensions: earnings-related pensions.38
Population Information System maintained by Digital and population data services agency of Finland: Spatial information.39
Measurements
Chronic diseases
Prevalent chronic diseases at baseline and incident diseases during the follow-up period will be defined based on the national register data (Causes of Death,31 Care Register for Health Care,35 Cancer Registry36 and medications37), using health examination measurements and self-reported information on diagnosed chronic diseases as a complement when feasible. The classification of the chronic diseases will be based on the 10th revision of the International Classification of the Diseases (ICD-10) complemented with the International Classification of Primary Health Care coding and the Anatomical Therapeutic Chemical Classification System codes for medication. Regarding mental health, Composite International Diagnostic Interview,40 the Beck Depression Inventory41 and the Hopkins Symptom Checklist-2542 were used (see table 2).
Multimorbidity patterns
In the MOLTO study, multimorbidity will be defined as the patterns of two or more chronic conditions as defined by WHO.2 We will focus on the chronic diseases, which cause the greatest burden in Finland. We have estimated the burden of diseases based on the latest aggregated register-based data. First, the proportion of register-based outcomes caused by each chronic diseases among Finnish adult population was defined (eg, what proportion of deaths is caused by malignant neoplasms). The chronic diseases (excluding, eg, infectious and parasitic diseases, conditions related to childbirth and pregnancy, injury as well as very rare chronic diseases) were categorised according to ICD-10 chapters and blocks. The following register-based outcomes were included: mortality, the use of health services, disability pensions, sickness and disability allowances and entitlements to specially reimbursed medication (for detailed information, see tables 4 and 5).
The proportions (%) of register-based outcomes caused by chronic diseases according to ICD-10 chapters and blocks in men
The proportion (%) of register-based outcomes caused by chronic diseases according to ICD-10 chapters and blocks in women
Second, this information was summarised by choosing those chronic diseases (ie, disease blocks), which were among the 10 most common causes according to at least one of the register-based outcomes mentioned above. This summarised information is presented in tables 4 and 5, showing that diseases of circulatory system as well as mental and behavioural disorders were particularly burdensome from several different perspectives. Concerning individual diseases (ie, disease blocks), ‘diabetes mellitus’ (E10-E14) and ‘other degenerative diseases of the nervous system’ (G30-G32; including G30 Alzheimer disease) appeared to be particularly burdensome. Diseases of the musculoskeletal system and connective tissue caused remarkable burden especially in terms of use of healthcare services, disability pensions and sickness allowance. Malignant neoplasms were significant causes of deaths and use of healthcare services.
Information on the chronic diseases causing the greatest burden in Finland, presented in tables 4 and 5, will be the basis when defining multimorbidity patterns in the MOLTO—study.
The burden caused by multimorbidity patterns
When defining the most burdensome multimorbidity patterns, the burden caused by multimorbidity will be estimated using the following outcome variables: mortality, work disability, limited functioning, quality of life, healthcare utilisation and costs. Information on mortality will be obtained from Statistics Finland.31 Functional capacity, work ability and quality of life will be estimated based on HES data (table 2). When estimating work ability, also information concerning disability pensions will be used.38 Healthcare utilisation will be determined based on register-based information from the Care Register for Health Care35 as well as self-reported information from HESs (table 2).
Potential risk and protective factors of multimorbidity patterns
All HESs include comprehensive information concerning potential sociodemographic, lifestyle and biological risk and protective factors of multimorbidity (table 3). Information on sociodemographic factors will be complemented with register-based information on education, socioeconomic status and occupation32–34 as well as information on degree of urbanisation.39 Lifestyle variables (table 3) are determined by self-administered questionnaires or interviews. Data on diet have been collected by a validated self-administered food frequency questionnaire,43 44 allowing us to assess both individual dietary indicators (eg, use of vegetables) and the overall diet defined by a dietary score. Biological factors, that is, anthropometrics and blood pressure, have been measured with standardised methods by trained nurses. Blood collection, sample processing and management were performed by trained laboratory personnel. The blood samples were processed and frozen immediately after sampling. The laboratory analyses were performed at the biochemistry laboratory at THL, which has taken part in External Quality Assessment Schemes organised by Labquality, Helsinki, Finland. Health 2000/2011, FINRISK and FinHealth cohorts have also been genotyped with genome-wide genotyping arrays (Illumina Inc and Thermo Fisher Scientific) and imputed with population-specific reference panel to contain over 12 million genomic variants.
Patient and public involvement
Patient or the public were not involved in the design or conduct of the study. The MOLTO study is based on the Finnish HESs carried out among general population.
Statistical analyses
Both cross-sectional and longitudinal study designs will be used. Longitudinal designs will be based on register-based follow-up (all HESs) and repeated survey measurements (Health 2000/2011).
Aim 1
Multimorbidity patterns will be determined using latent class analysis45 based on chronic diseases, which cause the greatest burden in Finland (tables 4 and 5). The burden caused by multimorbidity will be analysed by survival analysis methods46 such as Cox proportional hazards and Poisson regression models.
Aim 2
The risk and protective factors of multimorbidity will be analysed using Cox proportional hazard models. As the multimorbidity pattern can be determined at any time point using the event times obtained from the register data for each individual, we will also apply multistate models to analyse the transitions between the different multimorbidity categories in real time.47 We will examine the combined effects of the risk factors on selected multimorbidity patterns by testing both additive (the relative excess risk due to interaction) and multiplicative interactions in the same Cox proportional hazards models.48 The clustering of the risk factors and the accumulation of multiple risk factors in the same individual will be examined using cluster and latent class analyses. To evaluate the relative importance of the risk factors at the population level, we will assess population attributable factors for the risk factors of multimorbidity incidence.49
Aim 3
For projections novel techniques using data-driven Bayesian hierarchical and/or multistate models which have been developed in the Finnish research project ‘Projections of the burden of disease and disability in Finland—health policy prospects (PoDDy- HePo)’50 will be used.
The effects of non-participation will be handled using multiple imputation, inverse probability weights51 or other suitable methods in all analyses. Complex sampling designs will be taken into account in the analyses.
Ethics and dissemination
The HES data have been collected following the legislation at the time of the survey. The following ethics committees of the Hospital District of Helsinki and Uusimaa have approved the data collection and register linkages for each survey:
Ethical committee for research in epidemiology and public health: Health 2000 (407/E3/2000) and FINRISK 2002 (558/E3/2001).
Coordinating ethics committee: FINRISK 2007 (299/EO/06), FINRISK 2012 (162/13/03/00/2011), Health 2011 (45/13/03/00/11), Migrant Health and Wellbeing study (325/13/03/00/2009), FinHealth 2017 (37/13/03/00/2016).
The participants were fully informed, and they participated in the surveys voluntarily. The participants also provided written informed consent for the use of their data and register linkage. Permissions for record linkage have been obtained from data controllers. Both survey and register data include sensitive personal information. To ensure data confidentiality, only a very limited number of persons working with raw data (the data managers) have access to personal identification information, which is used to link survey data to register information. Researchers will work on pseudonymised data sets. When processing the personal data, EU General Data Protection Regulation will be followed.
Dissemination is targeted at the scientific community, health authorities and policymakers as well as the media and general public. Dissemination platforms will include 6–8 peer-reviewed scientific publications, conference and workshop presentations, website (under THL website) and social media such as Facebook and Twitter accounts of THL.
Ethics statements
Patient consent for publication
References
Footnotes
Contributors TJ, PK, AL and SK were mainly responsible for the study concept and design. JS provided expertise on mental health and JJ on register-based data. TJ drafted the first version of the manuscript. All authors critically commented the manuscript.
Funding This work was supported by the Social Insurance Institution, Finland (grant number 65/26/2019).
Competing interests JJ is Senior Medical Officer of the Care Register for Health Care.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.