{"paper_id":"0a5a03db-fa17-4966-b419-7e90c66cdaf4","body_text":"The socioeconomic differences in morbidity and mortality, in particular across educational levels, have been widening in high-income countries for decades [ 1 – 8 ]. A wealth of research has linked socioeconomic inequality to reduced population health [ 2 ,  9 – 11 ]. Wilkinson’s (1996) influential work first demonstrated that it is not necessarily the wealthiest countries with the most healthy populations, but countries with the smallest income differences within society [ 9 ,  10 ]. More unequal societies show shorter life expectancy and higher prevalence rates of obesity,  HIV infections , and mental illness, among others [ 9 ,  12 – 23 ]. Further, social disparities in multimorbidity are inversely related to educational achievement [ 24 ,  25 ]. Recently, high-income countries have seen increased  income inequality  [ 2 ]. In Europe, one out of four adults is now at risk of poverty or social exclusion [ 2 ,  26 ,  27 ]. Moreover, the fallout from the COVID-19 pandemic has particularly affected the socioeconomic low-income groups with chronic conditions such as heart diseases, diabetes, and chronic obstructive pulmonary disease (COPD) [ 28 ].\nAlong with increased income inequality [ 2 ], the disease burden of chronic conditions and multimorbidity is increasing [ 29 ,  30 ]. Consequently, the expenditures also increase worldwide along with ageing populations [ 31 – 40 ]. These expenditures have earlier been shown to account for up to 80 per cent of healthcare costs [ 41 – 44 ]; and to be positively correlated with multimorbidity and socioeconomic factors [ 45 ]. To control the growing cost of healthcare, decision-makers require access to reliable, real-world evidence of disease and treatment patterns [ 46 ,  47 ]. Real-world evidence of disease prevalence is essential for estimating the burden of disease, cost of illness, and budget impact on new health technologies [ 48 ,  49 ]. Thus, the World Health Organization (WHO) has long recommended improvement in data surveillance of chronic diseases [ 37 ]; and other studies have criticised and recommended additional methodological improvements measuring disease burden [ 50 – 53 ]. Hvidberg et al. (2016, 2019) earlier replied to these calls [ 54 – 58 ]. While most prevalence studies usually included a single or a limited number of chronic conditions [ 44 ,  59 – 66 ], Hvidberg et al. (2019) encompassed the full-sized disease burden of chronic conditions based on uniform methodology physician-reported conditions covering the whole adult population of Denmark [ 56 ,  57 ]. Besides the completeness, the advantages of this single-study approach were reliable, and comparable prevalence rates were estimated across conditions which often cannot be achieved in other studies caused by biases from comparing different, heterogenic data [ 54 ]. The study comprised 199 chronic conditions and found that a record high of two-thirds of the population had at least one chronic condition [ 56 ]. However, while sex and age characteristics were identified, the studies did not investigate the socioeconomic inequalities across the conditions. Comparable knowledge about socioeconomic differences within and between chronic conditions would, for example, support decision-makers or healthcare professionals’ evidence according to service needs and treatment potentials for future prioritisation and socially differential clinical treatments, as socioeconomic larger disparities could indicate treatment potentials or patient needs [ 24 ,  67 – 72 ]. This has long been used to inform and target populations within the treatment of health behaviours like smoking, exercise, obesity etc. [ 67 ], but less so within chronic diseases, although numerous studies have identified disparities within single diseases [ 25 ,  56 ]. Thus, comparable estimates of socioeconomic disparities across a long range of chronic conditions could provide new insights and practical usefulness and move future prioritisation and socially differentiated treatment forward.\nThe present study estimates the national prevalence rates of socioeconomic inequalities regarding educational levels, income, and socioeconomic position of 199 chronic conditions of the whole adult Danish population. The study provides an off-the-shelf catalogue and a comparative overview of socioeconomic inequalities across chronic conditions for healthcare professionals, decision-makers, researchers, and clinicians to use, for example, in practical treatment, targeting low-income patients of interest. We also want to provide concrete information for future prioritisation for specific chronic conditions based on the identified socioeconomic differences and size of prevalence. To the best of the authors’ knowledge, the present study is the first and most comprehensive, independent register-based attempt to estimate socioeconomic inequalities associated with the disease burden of a whole nationwide population using a comparable, uniform methodology.\n\nThe nationwide study population comprised 4,555,439 Danish residents aged 16 years or older, alive on 1 January 2013. Hereof, 45.2% were 16–44 years old, 45.9% were 45–74 years old, and 8.9% were 75 years or older. The population had a mean age of 46.7, and 49.2% were men.\nSeven registers from Statistics Denmark were employed in the current study ( Table 1 ). The registers have been recorded for public administration, for example, for claims and management, surveillance, tax, and control functions by government officials at the individual level [ 73 ].\nPartly adapted from Hvidberg et al. (2019) [ 56 ].\nThe Population’s Education Register (PER) [ 74 ] and The Income Statistics Register (ISR) [ 75 ] were used for identifying socioeconomic disparities regarding educational achievements, income level, and socioeconomic position. The Danish National Patient Register (NPR) [ 76 ], the Danish Psychiatric Central Research Register [ 77 ], The National Health Service Register (NHSR) [ 78 ] and The National Prescription Register (TNPR) [ 79 ] contained information on physician-reported ICD-10 hospital diagnoses, prescriptions, and general practice services. These data were used to identify the 199 chronic conditions by clinical recommendation [ 54 ,  55 ,  58 ]. Finally, sex, birth date, and other information originated from the Danish Civil Registration System [ 80 ]. The registers were linked together by the unique personal identification number assigned to every person [ 81 ]. The employed registers and quality are described elsewhere [ 54 – 56 ].\nThree measures of socioeconomic disparities were used: educational achievements, personal income, and socioeconomic position in society. All three variables were derived from Statistics Denmark PER and ISR registers (coming variable names are in brackets) for the year 2012 on 31 December. Education was measured as the highest archived education based on the International Standard Classification of Education, ISCED2011 (AUDD). Educational achievements were divided into five levels: 1. no education or training, 2. student, 3. shorter, 4. middle, and 5. higher education. Please note that students, who did not have any other achieved education, were classified separately using the information of their socioeconomic position (SOCIO02), into category 2 (student). This was done as students are usually younger, healthier, and thus have different health conditions than people with no education. Income was measured as personal income (DKR) after tax (DISPON_NY). Income was divided into quartiles to better allow for international, relative comparisons and reported as means. Finally, the socioeconomic position was measured based on standard classification (SOCIO02) into the following seven categories: 1. Retired, age, 2. Retirement, early retirement (health), 3. Sick leave/leave, 4. Unemployed/social benefits, 5. Unemployed minimum six months, 7. In training/study, 8. Employed, and 9. Others not in the workforce.\nA detailed description of the phases and methods used to define the 199 chronic conditions are provided elsewhere [ 54 ,  55 ,  58 ]. In summary, we defined a ’chronic condition’ according to the definition 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’  [ 56 ,  82 – 84 ]. A clinical expert panel identified the ICD-10 diagnosis considered ’chronic’ based on the above definition [ 55 ]. The experts aggregated the ICD-10 diagnoses to 199 conditions, where several conditions included subgroups of different ICD-10 diagnoses; thus, some defined conditions contained several different conditions within related disease groups. Subsequently, all ICD-10 diseases considered chronic by definition included the full-population burden of chronic conditions [ 55 ,  56 ].\nSeveral chronic conditions do not last for a lifetime, although they last longer than the defined 12 months. The differing ’chronicity’ of the identified chronic conditions were divided into four groups of severity depending on how long they were expected to last [ 55 ]:\nCategory 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);\nCategory II : Stationary to diminishing chronic conditions (10 years from register inclusion time to the time of interest);\nCategory III : Diminishing chronic conditions (5 years from register inclusion time to the time of interest); and\nCategory IV : Borderline chronic conditions (2 years from register inclusion time to the time of interest).\nAdapted from Hvidberg et al. (2016, 2019) [ 55 ,  56 ].\nThis approach was developed to handle a well-known challenge of register research: if a condition is only identified once, for example, 10 or 25 years earlier, can researchers then be sure that the patient still suffers from this condition at present. Thus, to address this, all 199 chronic conditions were allocated into one of the four chronicity categories by medical experts. The allocation was based on a clinical judgment of how long the different chronic conditions with the best possible clinical certainty would still suffer from the condition from the time of interest, 1 January 2013. This new approach can also be used at different times of interest and as a proxy for severity.\nAn algorithm was employed based on the clinicians’ definitions and allocation of the 199 chronic conditions and related ICD-10 codes into one of the four categories. However, 35 chronic conditions were not judged by the medical experts to be appropriately representative using exclusively ICD-10 diagnosis. Thus, 35 more complex disease algorithms were suggested and employed using multiple registers, including medicine, hospital treatments, and general practice services. Furthermore, details of the 199 distinctive definitions, the medical experts and panel process, the allocation of conditions into the four categories, and the detailed algorithms for replication can be found in references [ 55 ,  58 ].\nDisease prevalence rates were calculated per thousand subjects, i.e., the number of conditions identified, divided by the total population (N = 4,555,439) multiplied by 1000. Hence, prevalence rates were calculated from a specific point in time, 1 January 2013, thus comprising conditions dating back from this point of time-based on the four inclusion time periods. Disease prevalence within educational levels, sex, or age were calculated as the number of conditions within the socioeconomic variable of interest, divided by the total of the socioeconomic variables, multiplied by 1000 (Tables  2  and  3  and S1 Table in  S1 File ). Ratios were calculated by dividing no educational achievement with high educational achievement prevalence rates. Per cent proportions within the diseases were calculated for levels of education, income quartiles, and socioeconomic position ( Table 4 , S1 and S2 Tables in  S1 File ) as the number of patients for each level of the socioeconomic variable of interest, divided by the total number of patients within the disease, multiplied by 100. Finally, nine multiple binary logistic regression models were carried out to control for potential confounding and residual correlations of the crude estimates ( Table 5 ). This was done for the two key socioeconomic measures, educational levels, and income, hence one binary model for each of the five education levels and the income quartiles. All regression models included the 199 chronic conditions, gender and age as predictors.\nAge and gender standardized estimates in brackets. [SD] = Standard Deviation.\nICD-10 International Statistical Classification of Diseases, 10 th  Revision\nc  = complex defined conditions, see reference for further details [ 55 ].\n* Total population frequencies and prevalence adapted from Hvidberg et al. 2019 [ 56 ]\n** 2-year prevalence.\nAge and gender standardized estimates in brackets.\nICD-10 International Statistical Classification of Diseases, 10 th  Revision\nc  = complex defined conditions, see reference for further details [ 55 ].\n* Total population frequencies and prevalence adapted from Hvidberg et al. 2019 [ 56 ].\nMissing categories not showed but can be identified as the reaming difference to 100%.\nAge and gender standardized estimates (std) in brackets. SD = Standard Deviation.\nICD-10 International Statistical Classification of Diseases, 10 th  Revision\nc  = complex defined conditions; see reference for further details [ 55 ].\nConditions no. marked ‘A’, overlap with other conditions and are thus not counted twice [ 55 ].\n* Overall population frequencies are adapted from Hvidberg et al. 2019 [ 56 ].\n** 2-year prevalence.\nICD-10 International Statistical Classification of Diseases, 10 th  Revision\nc  = complex defined conditions; see reference for further details [ 55 ]. p. = p-value.\nConditions no. marked ‘A’, overlap with other conditions and are thus not counted twice [ 55 ].\nDeclaration and approval to conduct the study were obtained from the Danish Data Protection Agency and the Secretariat for Research Processing Records, Data and Development Support, Region Zealand (REG-142-2021). No informed consent was required. Statistics Denmark anonymized all register-data before the data were made available on their server.\nThe conditions were ranked according to their proportion of patients with no educational achievement. The higher ranking, the higher proportion of patients with no educational achievement within the disease. Direct standardisation of age and sex from the national average of age and sex on 1 January 2013 was included to identify differences free of primary effects [ 85 ,  86 ].\nIn summary, the prevalence rates for the disease groups and chronic conditions were stratified and presented by educational levels, sex and age (Tables  2  and  3  and S1 Table in  S1 File ), education levels, income quartiles ( Table 4  and S2 Table in  S1 File ), and socioeconomic positions (S3 Table in  S1 File ). A ratio between disease prevalence for no and high educational levels was calculated to identify inequalities within and between conditions (Tables  2  and  3 , S1, S2 and S4 Table in  S1 File ). Ranked conditions were used to identify clusters and groups of conditions with the lowest educational levels ( Table 4  and S2 Table in  S1 File ). Moreover, mean and ratios were presented for chronic conditions ( Table 2 ) and income ( Table 4 ) with standard deviations. Standardised prevalence estimates are presented in brackets where applicable. Finally, to provide the reader with an overview of highlights, the analysis and commenting will be focussed on all the 14 overall disease groups of the 199 chronic conditions, 29 specific common conditions including overweight, regularly measured in the National Population Health Surveys every four years [ 87 ] among others; and the 50 specific chronic conditions with the highest proportions of patients with no educational achievement. However, readers can still look up other estimates of the total 199 presented chronic conditions within the tables.\nData analysis and management were carried out using SAS 9.4 at Statistics Denmark’s research servers. Means and standard deviations were calculated using SAS “proc means”, frequencies were calculated based on the “proc tabulate” function, and logistic regression was calculated using the SAS “proc logistic” function.\n\nThe following descriptions provide an overview of the 14 overall disease groups, on the 29 most common conditions as well as overweight, and the 50 conditions with the highest proportion of patients with no educational achievement.\nTable 2  shows the prevalence of the 14 overall disease groups and five common medicines associated with educational achievement, sex, and age groups for patients with no educational achievement. The prevalence of having one or more chronic conditions for patients with no educational achievement was 768 per thousand compared to 601.3 for patients with higher educational achievement. This is equivalent to an overall educational ratio of 1.3, i.e., indicating that patients with no educational achievement are 1.3 times more likely to have a chronic condition than people with high educational achievement. Women with no educational achievement had a higher prevalence (272.6 per thousand) of one or more chronic conditions than males with no educational achievement (188.2 per thousand). Increasing age for patients with no educational achievement also increased the risk of having one or more chronic conditions, ranging from 582.0 (age 16–44), 819.5 (age 45–74) to 954.6 per thousand for 75+ years old. The largest overall educational differences were found among patients using antipsychotic medicine with a ratio of 4.3. Other large differences were found for disease group F (mental and behavioural disorders; ratio 2.5), disease group E (endocrine, nutritional and metabolic diseases; ratio 2.4), disease group I (diseases of the circulatory system; ratio 2.1), and disease group K (diseases of the digestive system; ratio 2.1).\nTable 3  provides the prevalence for 29 common, highly prevalent diseases and overweight (see S1 and S4 Tables in  S1 File  for all conditions) by no and high educational achievement, sex, and age groups for people with no education (see S1 and S4 Tables in  S1 File  for all 199 conditions). The collective conditions account for a total disease burden of 563 per thousand having one or more chronic conditions. The highest educational differences are found among mental conditions like schizophrenia (ratio 5.9), hyperkinetic disorders (ratio 5.2), dementia (ratio 4.9), but also osteoporosis (ratio 3.9), type 2 diabetes (ratio 3.8), COPD (ratio 3.3), and heart conditions and stroke (ratios ranging from 2.3–3.1). The lowest educational differences are found in respiratory allergy (ratio 0.8), asthma (ratio 0.7), and type 1 diabetes (ratio 1.1) and rheumatoid arthritis (ratio 1.1). The sex and age trends indicate a higher disease prevalence among women and increasing prevalence with increasing age, except for type 1 diabetes, schizophrenia, and hyperkinetic disorders. Moreover, schizophrenia, hyperkinetic disorders, and type 1 diabetes also showed decreasing prevalence rates with increasing age, contrary to the general trend of most other conditions.\nTable 4  displays the prevalence (per cent within conditions) of the five educational levels, income quartiles, and the mean income for all the 199 chronic conditions. However, the following exclusively comments on the 50 conditions with the highest proportion of patients with no educational achievement ordered in ICD-10 overall disease groups (S2 Table in  S1 File  comprising only the 50 conditions sorted).  Mental conditions (F)  had the overall highest proportion of patients with no educational achievement with 15 conditions out of the 50 conditions. This included eight conditions in the top 10 of conditions with the highest proportion of patients with no educational achievement: mental retardation (F70-F79), disorders of psychological development (F80-F89), schizophrenia (F20), dementia (F00, G30, and other disease codes for dementia), emotionally unstable personality disorder (F603), mental and behavioural disorders due to psychoactive substance use (F11-F19), organic, incl. symptomatic, mental disorders (F04-F09), and schizotypal and delusional disorders (F21-F29).  Diseases of the respiratory system (J)  had the overall second-highest proportion of patients with no educational achievement with one condition in the top 10 and three conditions in the top 50: Bronchitis (F40-42), Emphysema (J43), and COPD (J44, J96, J13-J18).  Diseases of the circulatory system (I)  had the overall third-highest proportion of patients with no educational achievement. This included 12 conditions: atherosclerosis (I70), other peripheral vascular diseases (I73), sequelae of cerebrovascular disease (I69), aortic and mitral valve disease (I05, I06, I134, I135), heart failure (I11, I13, I13.2, I142, I142.6, I142.7, I142.9, I150, I150.1, I150.9), chronic ischaemic heart disease (I25), stroke (I60, I61, I63-64, Z501), acute myocardial infarction (I21-I22), ischaemic heart diseases (I20-I25), atrioventricular and left bundle branch block (I44), AMI complex/other (I23-I24), pulmonary heart disease and diseases of pulmonary circulation (I26-I28), and atrial fibrillation and flutter (I48).  Diseases of the nervous system (G)  had the overall fourth highest proportion of patients with no educational achievement and included four conditions: cerebral palsy and other paralytic syndromes (G80–G83), Parkinson’s disease (G20, G21, G22, F02.3), epilepsy (G40–G41), and systemic atrophies primarily affecting the central nervous system (G10–G14, G30–G32). Chronic viral hepatitis (B18) was ranked number five.  Cancers (C)  and  In situ and benign neoplasms (D)  had the overall sixth highest proportion of patients with no educational achievement. This included three conditions: other anaemias (D64), malignant neoplasms of bronchus and lung (C43), and aplastic and other anaemias (D60-D63).  Diseases of the eye and ear (H)  had the seventh highest proportion of patients with no educational achievement. This included four conditions: presbycusis (H911), disease of the eye lens (H25-H28), blindness and partial sight (H54), and other retinal disorders (H35).  Diseases of the musculoskeletal system and connective tissue (M)  had the eighth highest proportion of patients with no educational achievement and comprised four conditions: coxarthrosis (M16), rheumatism (M790), osteoporosis in disease classified elsewhere (M82), and polyarthrotis (M15). Finally, chronic renal failure (N18) and type 2 diabetes (E11) had the ninth and tenth highest proportion of patients with no educational achievement. Notably, the above-mentioned specific conditions do overlap in ranks and can therefore not be entirely categorised into ten discrete groups, although this broadly reflects the groups’ average ranks. Income inequalities across the conditions broadly follow those of education, although differences exist. Readers are encouraged to explore this further for conditions of interest. The differences in the disease prevalence for people with no educational achievement within diseases ranged from 17.8–80.3 per cent. Hereof, many chronic conditions occurred between 30 and 40 per cent ( Fig 1 ). In total, 153 conditions of the 199 chronic conditions had more patients with no education than the national average of 29.0. An extract of  Table 4  with solely the 50 chronic conditions sorted by the highest proportion of patients with no educational achievement can be found in S2 Table in  S1 File . The table also includes the ratio between prevalence for high and no educational levels and per cent proportions of high and low levels of education. Finally, an overview of the socioeconomic positions can be found in S3 Table in  S1 File .\nThe dashed line is the national average of 29.0 per cent.\nTable 5  presents an off the shelf catalogue of odds ratios (OR) from nine logistic regression models between the five educational levels and the income quartiles and all the 199 chronic conditions, gender and age to control for potential confounding and residual correlations of the crude estimates. However, the following exclusively comments the regression model for no educational achievement vs high educational achievement and overall tendencies for the conditions based on their disease groups. Overall,  Mental and behavioural disorders (F)  and  Diseases of the musculoskeletal system and connective tissue (M ) have the highest level of social disparity. For the mental conditions, significant, high OR ranging from 2.1.-6.0 (excluding mental retardation) for conditions like mental and behavioural disorders due to psychoactive substance use, schizophrenia, post-traumatic stress disorder, emotionally unstable personality disorder, and hyperkinetic disorders (ADHD). For  musculoskeletal diseases , significant high OR ranging from 1.7–4.0 were found for conditions like fibromyalgia, rheumatism unspecified, myalgia, shoulder lesions, and spinal osteochondrosis. Interesting, significant decreasing OR were found for a common condition, rheumatoid arthritis (OR = 0.7) and enthesopathies of lower limb (OR = 0.5). For  Diseases of the nervous system (G)  and  Diseases of the eye and ear (H) , some high significant OR were found within the range of 1.7–2.8 for specific conditions like corneal scars and opacities, epilepsy, mononeuropathies of upper limb, cerebral palsy and other paralytic syndromes, and other disease of the inner ear.  Diseases of the circulatory system (I) ,  Diseases of the respiratory system (J) , and  Disease of the digestive system (K)  displayed higher significant OR of 1.6–1.9 for chronic conditions like ventral hernia, ulcers, atherosclerosis, bronchitis, emphysema, and chronic obstructive disease (COPD). Particularly disease groups I and H relatively many had high, significant, decreasing OR ranging from 0.5–0.7, indicating significant correlation with higher educational achievement.  Disease group (B) , and  endocrine disorders (E)  show high significant OR for patients with Chronic viral hepatitis with odds ratio at 3.7 and diabetes type 2 with odds ratio at 1.9. Finally,  Cancers (C)  was the disease group with the lowest OR and highest number of decreasing OR, i.e., cancers displayed the reversed social disparity as patients with high educational achievement have higher risk of getting e.g., malignant neoplasm of skin, other malignant neoplasms of skin, malignant neoplasms of breast, and malignant neoplasm of prostate.\n\nFrom present findings, the overall disease prevalence is 768 per thousand for people with no educational achievement compared to 601.3 for people with higher education (ratio 1.3), which suggests that overall people with no educational achievement have a 30 per cent higher disease burden. Among patients with no educational achievement, women had higher disease prevalence than men, and the prevalence increased with age for most conditions, although variations were found across conditions. These results are comparable to those of other studies, which also find a higher disease burden for patients with no educational achievement, and a higher prevalence of women among patients with no educational achievement [ 24 ,  45 ]. Across disease groups, the largest educational differences were found within disease group F–Mental and behavioural disorders (ratio 2.5), E–Endocrine, nutritional and metabolic diseases (ratio 2.4), I–Diseases of the circulate system (ratio 2.1), and K–Diseases of the digestive system (ratio 2.1). Among specific chronic conditions, schizophrenia had the largest educational differences (ratio 5.9) followed by hyperkinetic disorders (ratio 5.2), dementia (ratio 4.9), osteoporosis (ratio 3.9), type 2 diabetes (ratio 3.8), COPD (ratio 3.3), heart conditions and stroke (ratios ranging from 2.3–3.1). The logistic regression estimates enable readers and health professionals to identify any confounding and residual correlations of the crude estimates for social disparities for their conditions of interests—adjusted not only of gender and age, but also an unprecedented total of all 199 chronic conditions. Findings on socioeconomic differences showed that showed that many of the crude estimate tendencies where also confirmed in the adjusted regression analysis, and that particular people with no educational achievement had the highest OR within disease group F–Mental and behavioural disorders and M—Diseases of the musculoskeletal system and connective tissue. Earlier studies have also shown that schizophrenia is often associated with low educational achievement and high unemployment [ 88 ,  89 ]. Low parental socioeconomic and a family history of mental disorders are in the literature two well establish risk factors of schizophrenia [ 90 ,  91 ]. However, a large Danish study showed that even though parental socioeconomic and a family history of mental disorders contribute to the development of schizophrenia, they made a very limited difference on a person with schizophrenias ability to be employed and get an education [ 88 ].\nFurthermore, present findings showed an increasing prevalence of people with no educational achievement for every number of chronic diseases they have more than one. People living with multimorbidity have to follow a trajectory for each of the single diseases they suffer from that is a potential driver for a high treatment burden [ 92 ]. Additionally, socioeconomic factors such as low educational achievement is associated with high treatment burden [ 25 ,  92 ,  93 ]. Treatment burden is all the aspects of treatment that a patient experience regarding manage of their disease(s) and the impact this has on the patient’s wellbeing [ 92 ,  94 ]. The increasing number of people living with multimorbidity is an advancing public health problem [ 24 ,  94 ]. This underlines the importance of minimizing the treatment burden for patients with multimorbidity [ 25 ]. Especially, because a high treatment burden may lead to involuntary non-adherence, adverse health outcomes and increasing health care cost [ 92 ,  93 ].\nOne major strength of the current study is the size and high quality of the data. A full nationwide register-based recorded dataset for chronic conditions and medical treatments including more than 4.5 million people. The high number of recorded chronic conditions is also a major strength. Moreover, the study is unique in terms of the big data linkage of seven different high-quality registers, combining many chronic conditions with three different socioeconomic measurements. This approach using uniform methodology within a single study has provided comparable socioeconomic data of 199 chronic conditions, and thus the potential to identify treatment potentials, optimising treatments by differentiating and targeting patients accordingly to socioeconomic differences.\nAt least three main limitations exist. First, there are  register-based issues  of report identification and data misclassification. One fundamental limitation of register studies is not identifying conditions not treated in the healthcare system, self-treatments, as it is not reported within registers, leading to underestimates (see detailed evaluation in previous studies [ 55 ,  56 ,  95 ]. On the other hand, being treated for a chronic condition is not necessarily the same as being actually or severely ill. There may be instances of defensive medicine, patients being treated on suspicion of a chronic condition, or even over-diagnosis, leading to the identification and overestimates from registers [ 56 ]. Moreover, data misclassification is also a source of bias due to different coding practices between hospitals [ 61 ], clinical disagreements, dissimilar clinical and administrative practices, and interpretation of the ICD-10 criteria [ 96 ]. These issues would naturally affect our estimates of socioeconomic disparities. However, earlier studies have not found evidence of systematic misclassification; reported diagnoses of psychiatric and somatic conditions have been validated with good results [ 95 ,  97 – 102 ].\nSecond, high ratios, i.e., large differences in disease prevalence by thousandth between high and low educational achievements for a disease, do not necessarily, although often, comprise conditions that have the overall lowest proportions of people with low educational achievement, as seen in  Table 3 , S1, S4 Tables in  S1 File . As conditions with the lowest educational levels are of interest to healthcare professionals, this poses a challenge. On the other hand, conditions with the lowest proportions of people with no educational achievement might not have a potential for improvement, as there might not be any real differences between high and no educational achievement prevalence (e.g., low ratio). Thus, the ratio was used to measure real-world treatment potential. However, we urge readers to use both the ratio and proportions of educational achievements when identifying and comparing conditions of interest.\nThird, practical difficulties in presenting an overview of the results due to the number of socioeconomic measurements and diseases. While the catalogue is easily accessible for a few or single conditions for use in specific treatments, it is difficult to maintain an overview of differences across all the conditions. This might pose an issue for the practical use by decision-makers in prioritising and comparing a larger number of conditions.\nThere is no gold standard for socioeconomic measurements. In this study, we mainly relayed in educational achievements, and future studies should aim to combine educational achievement, income, and job status to get a more accurate socioeconomic measure. This is especially challenging when handling a large chunk of data like this study. Therefore, future studies also need to consider how to handle and compare large chunks of data and conditions.\n\nThe present study provides a catalogue of diseases prevalence associated with socioeconomic differences for 199 different clinical-reported chronic conditions and conditions by sex, age, education, income, and socioeconomic position, based on a total nationwide population. To the best of the authors’ knowledge, the study provides the most comprehensive, comparable descriptive register study and catalogue of chronic conditions’ socioeconomic prevalence, characteristics, and differences.\n\n(DOCX)\nClick here for additional data file.","source_license":"CC-BY-4.0","license_restricted":false}