Catalogue of multimorbidity mean based severity and associational prevalence rates between 199+ chronic conditions—A nationwide register-based population study

Background Real-world data on multimorbidity represents an important but underutilised source of evidence for the planning of healthcare services, including prevention, treatments, and health economic modelling. Aims This study aimed to estimate means of multimorbidity and provide associated prevalence rates and frequencies between 199 x 199 chronic conditions and disease groups based on the total adult Danish population and sex, age, and educational attainment. Thus, this study provides an off-the-shelf catalogue for use in treatments and planning by clinicians, decision-makers and researchers. Methods The study population contained all Danish residents above 16 years on 1 January 2013 (n = 4,555,439). The data was based on the linkage of six national registers covering hospital contacts, services in general practice, filled-in out-of-hospital prescriptions, and educational attainments. The health registers were used to identify the 199 chronic conditions based on the ICD-10 classification system. Results The mean number of chronic conditions (NCC) was 2.2. The mean increased with age, women had a higher mean than men, and there was a social gradient with the mean increasing with lower educational attainment. The mean NCC varied from 3.3–9.8 among all conditions. Across disease groups, the highest mean NCC were found within disease group N (chronic renal failure, mean = 8.8), D (in situ and benign neoplasms; mean = 6.5), K (diseases of the digestive system; mean = 5.7), and H (diseases of the eye and the ear; mean = 5.6). The highest mean NCC among the 29 common diseases was heart failure, ischemic heart diseases, angina pectoris, stroke, and dementia, with a mean above 6.5. Several prevalent conditions like hypertension, arthritis, chronic lower respiratory diseases, depression, type 2 diabetes, and overweight transcended other conditions regarding the associated prevalence rates. As one of the most frequent, hypertensive diseases were highly associated with arthritis (50.4%), depression (37.4%), type 2 diabetes (75.4%), cancers (49.7%), and being overweight (39.7%)–meaning that 50.4% of people with arthritis, 37.4% of people with depression and so on also had hypertensive diseases. The largest differences in means between individuals with no educational attainment and individuals with high educational attainment were found within disease groups J (diseases of the respiratory system, ratio = 1.8), Q (congenital malformations, deformations, and chromosomal abnormalities, ratio = 1.7), and B (viral hepatitis and human immunodeficiency virus disease, ratio = 1.7). Conclusions The current study provides a nationwide off-the-shelf catalogue of multimorbidity means and real-world associations estimates of 199+ chronic conditions for future clinical treatments and health care systems planning. The findings described are just one example of numerous results and underline that multimorbidity is highly prevalent in the adult Danish population and that it is a vital condition transcending all future medical treatment. The data offer essential information on the multimorbidity burden of disease in future differentiated treatments, healthcare planning, and economic, aetiological, and other research.


Introduction
Patients with multimorbidity, frequently defined as the coexistence of two or more chronic conditions within the same individual [1,2], have a lower health-related quality of life [3,4], higher mortality rates [5], decreased functional competence [6], and make more use of healthcare resources [7,8]. Numerous studies have identified the unequal distribution of diseases across socioeconomic groups [7,9,10]. Moreover, a growing proportion of the worldwide population lives with chronic disease and multimorbidity due to ageing populations, better living conditions, and improved healthcare technology [11,12]. For example, a recent study identified that 54.3 per cent of the Danish adult population had at least two chronic conditions and that 87.6 per cent of citizens above the age of 75 had multimorbidity with an average of 5.3 chronic conditions [12]. In addition, the disease-related-and increasing costs [13][14][15][16][17][18][19][20][21][22][23]-have been estimated to account for up to 80 per cent of the total healthcare expenditures for chronic conditions and multimorbidity [24][25][26][27]. Consequently, the challenges of multimorbidity are already high; and are only expected to rise in the decades to come [7,11,12].
The high prevalence of multimorbidity is particularly challenging for governments worldwide due to healthcare treatment structures. Healthcare systems worldwide are set up mainly for treating patients with single diseases; thus, most disease guidelines in the health system focus on single diseases [28]. This is contrary to an integrated approach and may amplify the risk of iatrogenic harm, increased drug interaction effects, and undesirable deficiencies in treatments and coordination for patients with multimorbidity [29]. Thus, any efforts to design future healthcare organisations to accommodate the growing number of patients with multimorbidity require detailed epidemiological data on multimorbidity and disease patterns. Moreover, decision-makers need access to reliable, real-world evidence of treatment patterns to handle the growing cost of healthcare [30,31]. Hence, real-world evidence of disease burden, prevalence, and correlational patterns are crucial for accurate estimates, cost of illness, and budget-impact analysis on novel health care technologies [32,33].
Multimorbidity is, however, a multifaceted, entangled, challenging subject to analyse. The Charlson Comorbidity Index [34], or simply counting conditions, may not provide sufficient details to understand complex disease patterns. Hence, much literature has investigated disease patterns using complex statistical methods [35][36][37][38][39][40][41][42]. For example, one study by Larsen et al. (2017) identified 6-7 disease groups from 15 conditions using latent class analysis [11]. Nonetheless, this illustrates some statistical difficulties in sufficiently describing disease patterns, as a reduction from 15 conditions to 6-7 disease groups might be considered relatively small; and researchers would most likely be able to find the same patterns by using simple prevalence estimates. Statistical pattern reduction is also particularly problematic as different statistical methods provide different results, are challenging to interpret and use, and there is no consensus on which statistical methods to use [11]. Thus, although statistical pattern recognition methods are useful for broad pattern recognition, further methodological work is needed.
Furthermore, for health professionals, raw, real-life, non-statistically reduced estimates are useful to obtain all details of the disease population of interest for either health care planning or clinical treatment. Another related study has, however, reported the prevalence rates of disease combinations but used self-reported conditions and was limited to 17 conditions [42]. Including a limited number of chronic conditions or using self-reported conditions is a limitation of many disease studies [7,11,27,[42][43][44][45][46][47][48][49][50]. This provides a boundary for real-world estimates of the full disease burden experienced by patients. But also as, different study methodologies limit the comparability of diseases prevalence estimates needed for decisionmakers and others; thus, researchers and authorities have recommended using a uniform study methodology in disease burden studies across conditions for decades [14,15,[51][52][53][54][55][56].
The current study aimed to estimate basic, descriptive, nationally representative means of multimorbidity and associated prevalence rates and frequencies of 199 x 199 chronic conditions of the total adult Danish population according to sex, age, and educational attainment. As one measure of severity, the mean NCC will enable researchers, health professionals, health economists, and decision-makers to identify, access, and compare the disease burden of the 199 chronic conditions. The correlational prevalence estimates between the 199 x 199 conditions will give real-world, detailed, unbiased, self-report estimates of the concrete multimorbidity for each of the 199 chronic conditions used in treatments and health care planning. Thus, the study provides an off-the-shelf catalogue and a comparative overview of multimorbidity across 199 chronic conditions. To the best of the authors' knowledge, the current study provides the most comprehensive descriptive estimates of multimorbidity means and correlational prevalence of chronic conditions based on an entire country's population, a uniform, comparable methodology and an exceptionally high number of chronic conditions.

Study population
The nationwide study population included 4,555,439 Danish residents aged 16 years or older alive on 1 January 2013. The study population consisted of 49.2% men, and the mean age was 46.7 years. Forty-five per cent were between 16-44 years old, 46% were between 45-74 years old, and 9% were 75 years old or older.

The registers
In Denmark, there is a long tradition of reporting diseases, treatments, medications, and contact with the healthcare system, in national health registers. The registers were originally intended for data collection by government officials in public administration at the individual level [57]. All registers have a unique civil registration number that enables individual linkage across registers by the distinct personal identification number assigned to every resident in Denmark [58].
In the current study, six registers were applied and linked from Statistics Denmark. The National Patient Register (NPR) [59], the Danish Psychiatric Central Research Register (PCRR) [60], the National Prescription Register (TNPR) [61], and the National Health Service Register (NHSR) [62] held information on ICD-10 diagnoses, medicine prescriptions, and services in general practice. Educational attainments were obtained from the Population's Education Register (PER) [63] based on the International Standard Classification of Education (ISCED2011). Sex and age originated from the Danish Civil Registration System [64]. The utilised registers and characteristics are described elsewhere [12,65,66].

Defining 'chronic condition'
A 'chronic condition' was defined in line with former studies if the '. . .condition had lasted or was expected to last twelve or more months and resulted in functional limitations and/or the need for functional limitations and/or the need for ongoing medical care' [12,[67][68][69]. Using the Delphi method, a medical expert panel decided which ICD-10 diagnosis out of around 22,000 ICD-10 codes to be considered 'chronic' from the above definition [65]. The experts grouped the chosen chronic ICD-10 diagnosis into 199 conditions, of which some conditions encompassed subgroups of ICD-10 diagnosis. Hence, some identified conditions contained multiple different conditions within interrelated disease groups. Consequently, all ICD-10 conditions considered chronic based on the definition was contained in pursuit of including the full-population burden of chronic conditions [12]. A detailed description of the definitions, distinct phases and methodology are provided elsewhere [15,65,66].

The data register algorithms used to identify the chronic conditions
Since numerous chronic conditions last longer than the 12 months used in the definition but do not persist for a lifetime, the 'severity of chronicity' was categorised into four categories depending on how long the conditions were expected to last [65]: 1. Category I: Stationary to progressive chronic conditions (no time limit equals inclusion time going back from the time of interest for as long as valid data were available. In the current study, this starting point was defined by the introduction of the ICD-10 diagnosis coding in Denmark in 1994); then expected that the patient still suffers from the condition? Hence, the expert panel assigned all of the 199 chronic conditions into one of the four categories . The allocation into one of the  four categories was based on a medical judgement on how long time the various ICD-10 diagnoses identified as 'chronic', with the best possible clinical conviction, would still have the disease from a time of interest. This systematic approach was employed as a proxy for disease severity. An algorithm based on the medical experts' definitions identified ICD-10 codes and allocated each of the 199 chronic conditions into the four chronicity categories that were utilised for data collection. However, for 35 of the 199 chronic conditions, the medical experts did not expect the ICD-10 diagnosis to be representative alone. Thus, 35 algorithms were developed based on multiple registers comprising medicine, hospital treatments, and services in general practice [12,65,66]. Additional details of the 199 distinctive definitions, including the 35 diagnostic algorithms, the medical experts and the panel process, and the four categories' assignment, are described earlier [65,66].

Statistical analysis
Means of chronic conditions and per cent prevalence were calculated for each of the 199 chronic conditions. Means were calculated as the sum of all subjects' multimorbidity within the disease of interest, divided by the number of subjects within the disease group and elaborating variables of interest. We used the following elaborating variables: sex, age groups (16-44, 45-74, and 75+) and educational attainment (no education vs higher education). Per cent prevalence was calculated within diseases of elaborating variables as the number of subjects of the elaborating variable of interest, divided by the total subjects of the disease, multiplied by a hundred. Direct standardised means and prevalence estimates were presented and calculated based on the national proportion of sex and age on 1 January 2013, as referenced [70,71] were applicable. Ratios as a measure of social disparity in multimorbidity were calculated by dividing the mean number of chronic conditions (NCCs) of individuals with no education by the means among individuals with high education attainment for all conditions. Standard deviations (SD) of means were provided. All conditions were given ranks according to their NCCs, with one indicating the highest NCCs based on the unstandardised means. To provide the reader with an overview of the comprehensive material, 14 disease groups referring to the ICD-10 system and described in detail elsewhere [12] and 29 common conditions plus overweight are presented and commented on in the result section. The common conditions comprise the conditions measured in the National Population Health Surveys every fourth year [72], among others. "Overweight" is included due to its general importance, although not consistently considered a chronic condition in the literature.
Data management and analysis were done using SAS 9.4 from Statistics Denmark's remote research servers. or more chronic conditions-e.g. 6.6% had seven or more chronic conditions, and 1.9% had ten or more chronic conditions (S1 Table).
Overall, the mean NCCs in the population was 2.2 -with a mean of 2.4 among women and 2.0 among men. The mean NCCs increased by age, and women had a higher mean of chronic conditions than men, although this gap narrowed with age (Fig 2). We found a social gradient with the mean of chronic conditions increasing with lower educational attainment. Thus, individuals with no education had the highest mean of chronic conditions (mean = 3.1), and individuals with higher education had the smallest mean (mean = 1.6)-except for the student category, where the mean was 0.5 (Fig 2).
The mean NCCs across the 199 chronic conditions and the disease groups range from around 3 to 9, with the main proportion of conditions having a mean between 5 and 7 chronic conditions (Fig 3). Table 1 displays the mean NCC for the 14 disease groups and sex and age. Disease group N (chronic renal failure) had the highest mean NCCs (mean = 8.8), followed by disease group D (in situ and benign neoplasms; mean = 6.5), K (diseases of the digestive system; mean = 5.7), and H (diseases of the eye and adnexa and diseases of the ear and mastoid process; mean = 5.6). Disease group C (cancers), followed by disease group E (endocrine, nutritional and metabolic diseases), G (diseases of the nervous system), I (diseases of the circulatory system), and F (mental and behavioural disorders), had a mean of NCCs ranging from 4. 8-5.4. Finally, disease group L (diseases of the skin and subcutaneous tissue), M (diseases of the musculoskeletal system and connective tissue), J (diseases of the respiratory system), and Q (congenital malformations) had a mean of NCC ranging from 4.0-4.7. Of the 14 disease groups, sex differences were among others found in disease group D (female = 6.2 vs male = 7.0), K  Table. Among the 29 most common chronic conditions and overweight, heart disease, stroke, and dementia had more than seven other chronic conditions (Fig 4). Further, chronic obstructive    I05-I06; I10-28; I30-33;  I36-141; I44-I52; I60-I88; I90-I94; I96- pulmonary disease (COPD), cataracts, osteoporosis, type 2 diabetes, anxiety disorders, and inflammatory polyarthropathy had relatively high NCCs, with a mean above six. The 29 common conditions and overweight had a mean of four chronic conditions or more. S2 and S3 Tables show the prevalence (N), overall mean NCCs, means by sex and age, of the total 199 chronic conditions and the 29 common conditions. Table 2 shows the 30 x 30 cross-tabulated prevalence rates between the 29 most common chronic conditions and overweight. The prevalence rates indicate how many per cent within the row condition have the condition in the column. The following commentary is delimited to selected, highly prevalent conditions: hypertensive diseases, arthritis, chronic lower respiratory diseases, depression, type 2 diabetes, cancers and overweight. Hypertensive diseases were associated with respiratory allergy (25.3%), arthritis (24.0%) and ischemic heart diseases    Table shows the frequencies and percentages of the 199 conditions cross-tabulated with the 29 most common conditions. S5 Table (spreadsheet) shows the frequencies and percentages of the 199 conditions cross-tabulated with all 199 chronic conditions, disease groups, and common medicines. Table 3 presents the overall mean NCCs, the mean NCCs for patients with no education and patients with higher educational attainment, and the prevalence of having 1, 2, 3, 4, 5, 6, or 7+ conditions within each disease across all 199 chronic conditions, disease groups, and overweight. S6 Table shows the means of the 199 conditions and all five levels of educational attainment.
Finally, endometriosis (N80) and overweight (E66), which were not defined as chronic conditions, had the highest-and third-highest ratios (2.1 and 2.0) among individuals with no educational attainment and individuals with higher educational attainment.

Discussion
To the best of the authors' knowledge, the present study is the first and most comprehensive register-based attempt to estimate the multimorbidity disease burden of chronic conditions from a nationwide population using a comparable, uniform methodology across a larger number of chronic conditions. The result section only shows a small fraction of possible examples for some multimorbid data that can be extracted from the catalogues or tables. For instance, and in summary, we investigated the mean NCCs and associations of 14 disease groups, 29 common chronic conditions (to provide the reader with an overview), and 199 chronic conditions for the entire Danish adult population, including differences in sex, age groups, and educational attainments. Our study showed that most people in the Danish population had one or more chronic conditions and that multimorbidity is common. This is in line with previous national and international research [7,12,73]. The overall mean of NCCs for the population was 2.2 and 3.4 for patients with one or more chronic conditions. The mean NCCs increased by age, and women had a higher mean of chronic conditions than men.
Furthermore, we found a social gradient in the mean of NCCs-with individuals with lower educational attainments having a higher mean. For instance, the largest difference in means of NCCs between individuals with no education and individuals with higher educational attainment was found in disease group J (diseases of the respiratory system). This, increasing NCC by age, higher rates in women and increasing rates of NCC with lower educational attainment are also following earlier studies [7]. We found large variations in the mean of NCCs between conditions ranging from a mean of 3.3 to a mean of 9.8 in chronic conditions. The diseases with the highest NCC were overall found within disease groups N-diseases of the genitourinary system (mean = 8.8), D-in situ and benign neoplasms (6.5), K-diseases in the digestive system (5.7), and H-a disease of the eye and the ear (5.6). The most common chronic conditions are also complicated by high mean rates of multimorbidity, including hypertensive diseases, respiratory allergy, chronic lower respiratory diseases, type 2 diabetes, and depression. Persons with heart failure, ischemic heart diseases, angina pectoris, and stroke had the highest NCC, all with a mean above 6.5 chronic conditions but less hypertension. Furthermore, individuals with COPD, cataracts, osteoporosis, type 2 diabetes, anxiety disorders, and inflammatory polyartropathies had high NCCs-above six chronic conditions. Most individuals with one of the 29 common conditions had above five NCCs. Other conditions are characterised by little multimorbidity rates, such as type 1 diabetes, tinnitus, and other headache syndromes, most likely because the conditions are typically diagnosed in younger patients [74].
When looking into some examples of associations between the chronic conditions, we found that conditions, not surprisingly, seem to be particularly associated with other diseases within the same disease groups; for example, chronic lower respiratory diseases were highly associated with asthma and respiratory allergy. However, conditions also often transcend disease groups. For instance, hypertensive diseases were also associated with type 2 diabetes and depression, which might be explained by the high prevalence of the three conditions. We found that prevalence rates of depression varied between 14.5% to 51.0% in the 29 conditionsfollowing other evidence showing that depression is a common comorbidity to several chronic conditions [75,76]. Further, type 2 diabetes is associated with ischaemic heart diseases-both common conditions and also linked to the same underlying pathology. This is consistent with common medical knowledge and another study by Breinholt et al. (2017) looking at correlations of 15 chronic diseases. Six disease classes were identified here, and heart diseases, particularly hypertension, were associated with at least four other conditions [11]. Other prevalent conditions like arthritis, chronic lower respiratory diseases, depression, and overweight also transcended to other disease groups.

The catalogue in summary-and future use
The above results, underlined, are not exhaustive but just a few of many possible data extractions. Hence, the main aim of this study was to provide a detailed off-the-shelf catalogue for others to explore their specific interests and needs. In summary, we provided nine comprehensive catalogues (Tables 1-3 and S2-S7 Tables) that can be used to explore how the severity and associations of multimorbidity are distributed, including differences in age, sex, and educational attainment across the 199 chronic conditions as described below: We measured disease burden severity in terms of the crude mean NCCs. Here, the following tables provide overall mean NCCs, including differences in age groups and sex: Table 1 by disease groups and medicines, S2 Table for all the 199 conditions, S3 Table by the 29 common  conditions and overweight, and S7 Table show the raw mean NCCs of 199 conditions by age and sex in 14 categories for further detailed analysis. Moreover, Table 3 provides an overview of mean NCCs and prevalence for the 199 chronic conditions regarding the overall population and means by social equality measured by high and low educational attainments. Finally, the S6 Table shows the mean NCCs of the 199 chronic conditions by all five educational attainments.
While the means provide a crude estimate of severity, the following tables provide the associational prevalence of the chronic conditions. Table 2 shows the comorbidity prevalence between the 29 common conditions and overweight, and the S4 Table shows the correlational prevalence rates of the 199 chronic conditions, disease groups, medicines, and overweight by the 29 common conditions and overweight. Finally, the S5 Table shows the correlational prevalence rates of all 53,361 combinations between the 199 chronic conditions, disease groups, medicines, and overweight.
We see three main potential uses of the catalogue: First, it can support and inform on-thefloor health care specialists of possible multimorbidity concerns to be considered within treatments. Although knowledge about possible multimorbidity is not unknown within medical practice, healthcare systems worldwide are constructed to treat patients with single diseases [28]. This is a fact even though multimorbidity is the norm for 69.7 per cent of patients with a chronic disease or 45.7 per cent of the adult population within the present study (see S1  Table). In contrast, socioeconomic disparities within health behaviours like smoking, drinking, and exercise routines have long been used to differentiate treatments [77] and, for example, to a lesser extent, within the rehabilitation of cardiac diseases [78][79][80]. We propose that future treatments are, to a greater extent, also differentiated a prior due to the multimorbidity severity, disease associations, and clusters of common comorbidities, using evidence like the current catalogue. This will make future interventions more data-driven in real-world evidence and multimorbidity directly embedded in medical practice.
Second, we propose that the catalogues are also used to identify and prioritise diseases for treatment based on severity, related disease associations, and clusters of high severity conditions. However, as multimorbidity is only one facet of disease burden, prioritisation should be done in conjunction with other aspects of disease burden, including health-related quality of life [15], overall disease prevalence, socioeconomic characteristics [12], and socioeconomic disparities [9].
Third, we propose that knowledge of multimorbidity severity, chronic condition associations, and clusters of common comorbidities are also used by health care planners to model the future health care systems. We suggest that diseases are seen in a more holistic view, comprising clusters of conditions and that interventions are set up systemically to threaten known and firstly prevalent clusters of conditions; moreover, high severity, multimorbidity, and less prevalent conditions known to be costly with low patient outcomes should be addressed in specialised centres. We propose that the current catalogue is used further to identify relevant clusters of diseases within medical specialities. For instance, the detailed spreadsheet in the S5 Table provides aggregated, detailed data of multimorbidity for all 199 x 199 chronic conditions that can be used to identify clusters. As health care systems are currently mainly set up to treat single diseases, future health care planning needs to address and incorporate the real-world norm of multimorbidity.

Strengths and limitations
One of the main strengths of this study is the data, e.i. the application of data from six nationwide, high-quality registers and the use of the total nationwide population. A second strength is the application of a uniform and comparable methodology as recommended by WHO and researchers [20,52,53,81,82], e.i. the use of medical ratified definitions and algorithms applied to the unique data and the high number of chronic conditions comprised within a single study. This enables reliable comparisons across an extraordinary number of conditions. A third strength is the identified variation in the means and types of comorbidities. For example, the prevalence of overweight differs within the same disease groups and across individual chronic conditions. Some conditions have a high prevalence of overweight within the same disease group (Schizophrenia); others do not (Dementia). This and similar information could prove crucial in planning future health care interventions across different diseases, targeting different issues dependent on disease. This detailed variation might be lost using classical statistical methods like latent class analysis, factor analysis, or correspondence analysis. However, as we provide detailed, raw descriptive data, the current study can be used to identify such detailed differences useful in concrete interventions.
There are, however, also some methodological limitations in the present study. One limitation concerns the methodological issue of defining 'chronic'. Should 'chronic' be understood literally as 'forever', and should only 'severe' (not in the sense of high mean NCC) diseases be included as suggested by critics [83,84]? These choices impact the size of the disease burden and include conditions. However, defining 'severe' possess some of the same issues as defining 'chronic'. And defining chronic strictly as 'forever' would lead to the exclusion of many diseases, such as type 2 diabetes, some heart diseases, and cancers, broadly accepted as chronic diseases, as, in fact, many commonly perceived chronic conditions do not last forever. This was why earlier studies suggested a differentiated approach based on the previously mentioned four categorisations of chronicity or severity [15,65,66].
Moreover, labelling ICD-10 conditions as chronic or something else to not change the realworld disease burden but merely how we conceptualise it. However, the debate and varying severity highlight the complexity of chronic conditions. Notably, non-communicable diseases or long-term illness may be a better term than 'chronic condition', as 'chronic' is often understood 'forever' in everyday understanding, thus causing confusion or even reluctance.
Furthermore, our study showed a lower mean of NCCs for mental conditions like schizophrenia and ADHD and a lower prevalence of cardiovascular diseases like hypertension. There is, however, no clinical reason why mental conditions should have a mean and prevalence below the national averages for hypertension. This indicates that the comorbidities regarding, for example, heart diseases are underreported, and comorbidity could be even higher for conditions within disease group F. Other studies have already discussed similar limitations in underreports of diseases in register data [12,65].
Finally, we recognize that there are other ways to measure disease burden severity than in terms of crude NCCs [15]. And that NCCs and associations are merely a proxy of severity regarding health-related quality of life, death risks, and disabilities and should not stand alone but be used with a range of different disease burden measures, including the earlier mentioned and health behaviours. Nevertheless, it is a way to provide indications and an overview of possible disease severities quickly.

Implications for research
It is challenging to provide a broad overview of tendencies and clusters of conditions using solely raw descriptive statistics, particularly for a large number of conditions as in the present study. However, this study provides real-life, detailed estimates without statistical loss of data, particularly for ground health professionals, health care planners and clinicians who need to know their detailed disease population as a first step. Statistical methods, nonetheless, such as latent cluster analysis, factor analysis, multiple pattern analysis and artificial intelligence (AI), might provide a clearer overview as a second step. Although we recognize these statistical methods might have trouble identifying detailed variations and thus identifying subtle tendencies within data, they are still useful to supplement the current catalogue with broader, reduced statistical estimates for overall planning and research purposes. Hence, there is a need for future research to the use and develop consensus on more advanced methods and thereby identify broader clusters of comorbidities, and subtle, possible non-statistical tendencies across conditions and disease groups. Finally, future studies could also investigate how the classic statistical methods perform when identifying clusters and tendencies and comparing these.

Conclusions
The current study provides an off-the-shelf catalogue of multimorbidity means, correlational disease prevalence showing the specific disease proportions for 199 different chronic conditions and groups of conditions by gender, age, and educational attainments, based on a complete nationwide population sample. The findings underline that multimorbidity is the rule and not the exception and that multimorbidity is a fundamental condition transcending disease burden and impacting all future treatments. However, current disease guidelines only include multimorbidity at a sporadic level. We argue that having reliable, real-world evidence of multimorbid disease burden is crucial for on-the-floor interventions and health care planners as provided within current study in a raw, descriptive format for others to use. We further suggest that future research identify multimorbidity clusters and investigate how these could best be identified. To the best of the authors' knowledge, the present study provides the most comprehensive descriptive register study of the means of multimorbidity and correlational prevalence of chronic conditions. Supporting information S1 Table. Frequency table of