Study population

The study included all persons aged 0–95 years living in Denmark at some point between 1 January 2000 and 31 December 2015. Each individual was followed from birth, immigration to Denmark or 1 January 2000 (whichever happened last) until death, 95th birthday, emigration from Denmark or 31 December 2015 (whichever happened first). All dates were obtained from the Danish Civil Registration System,^{Reference Pedersen9} which has maintained information on all residents since 1968, including date of birth, gender, continuously updated information on vital status and a unique personal identification number that can be used to link information from various registers. This population register allowed us to include all individuals living in Denmark regardless of whether they received a diagnosis during the study period.

Assessment of mental and substance disorders

Diagnoses of MSDs were retrieved from the Danish Psychiatric Central Research Register,^{Reference Mors, Perto and Mortensen10} which contains data on all admission to psychiatric in-patient facilities since 1969, and visits to out-patient psychiatric departments and emergency departments since 1995. The diagnostic system used was the Danish modification of the ICD-10,¹¹ from 1994 onward. MSDs are described in Supplementary Table 1 available at https://doi.org/10.1192/bjo.2022.583, and are based on specific diagnoses used in GBD study publications¹² and a previous publication of YLDs of mental disorders in Denmark.^{Reference Weye, Santomauro, Agerbo, Christensen, Iburg and Momen3} MSDs were identified from 1995, and the 5-year period before start of follow-up on 1 January 2000 was used to identify prevalent cases at the beginning of follow-up. For each individual, date of diagnosis for each MSD was defined as admission date for the specific disorder (from in-patient, out-patient or emergency visit).

Assessment of GMCs

We identified GMCs with an algorithm on data from the Danish National Patient Register^{Reference Lynge, Sandegaard and Rebolj13} and the Danish National Prescription Registry,^{Reference Wallach Kildemoes, Toft Sørensen and Hallas14} developed to identify the list of chronic GMCs presented in Supplementary Table 2, which consists of 32 diseases and conditions classified into nine broad categories based on the registers.^{Reference Momen, Plana-Ripoll, Agerbo, Benros, Borglum and Christensen6,Reference Prior, Fenger-Gron, Larsen, Larsen, Robinson and Nielsen15,Reference Prior, Vestergaard, Larsen and Fenger-Gron16} The GMC categories were circulatory disorders, endocrine disorders, pulmonary disorders and allergies, gastrointestinal disorders, urogenital disorders, musculoskeletal disorders, haematological disorders, cancers and neurological disorders. For each individual, date of diagnosis of each GMC was defined as first admission date and/or relevant prescription date for the specific disorder.

Because of requirements related to data privacy, comorbid pairs of MSDs and GMCs within a given index that had fewer than five cases were combined with other comorbid conditions within the same broad category (e.g. disorders ischemic stroke and heart failure combined into ‘other’ circulatory diseases) until the combined group (labelled ‘other categories’) consisted of at least five different individuals. In instances where a disease category (e.g. circulatory diseases) consisted of fewer than five cases, the category was grouped with another disease category until the combined group consisted of at least five different individuals (e.g. circulatory diseases and cancers combined into an ‘Other’ category). Combined comorbidities and categories are listed in Supplementary Tables 4–22.

Remission and duration of each disorder

For this study, we assumed that most GMCs were chronic disorders, and thus we assumed no remission after onset of a GMC. However, when allocating a disability weight to migraine, the GBD studies make a correction for the intermittent nature of this condition (migraine was considered to be symptomatic 8.5% (95% CI 5.8–11.2%) of the year). Date of remission for MSDs was not available through hospital registers. Consistent with our previous publication,^{Reference Weye, Santomauro, Agerbo, Christensen, Iburg and Momen3} recovery rates were modelled using observed contacts with the healthcare system and gender-, age- and year-specific remission rates for each MSDs in Demark, following GBD study estimates.¹⁷ In brief, for each date of diagnosis of an MSD, an expected date of remission for that episode was calculated based on the observed date of diagnosis and disorder duration in years following disease-specific remission rates from the GBD studies. In persons with multiple diagnoses for the same disorder, two (or more) episodes were combined into a single one if remission date for the first episode occurred after the onset of the second episode. A detailed description of the application of remission has been published elsewhere.^{Reference Weye, Santomauro, Agerbo, Christensen, Iburg and Momen3}

Disability weights for each disorder

We used disability weights developed by GBD to quantify health loss associated with a given disease.^{Reference Salomon, Haagsma, Davis, de Noordhout, Polinder and Havelaar18} Disability weights represent the severity of short- and long-term health loss associated with a given disorder, and were developed based on household and internet-based surveys from nine countries. In these surveys, descriptions of two hypothetical individuals with different health conditions were described briefly in terms of functional effects and symptoms, and respondents were asked to consider which of these individuals the respondent considered to be healthier.^{Reference Salomon, Haagsma, Davis, de Noordhout, Polinder and Havelaar18} The weights are measured on a scale of 0 to 1, with 0 being a health state that is equivalent to full health and 1 being a health state equivalent to death. For disorders with multiple severity states, a weighted average based on the proportion of cases in each severity state was used. The proportion of cases in each severity was primarily retrieved from the GBD studies.^{Reference Vos, Lim, Abbafati, Abbas, Abbasi and Abbasifard2} However, for five GMCs (i.e. diabetes mellitus, chronic obstructive pulmonary disease, cancer, vision problems and hearing problems), proportions were identified in Scottish Burden of Disease publications,¹⁹ and for three GMCs (ulcer/gastritis, chronic kidney disease and epilepsy), proportions were determined based on other publications.^{Reference Hill, Fatoba, Oke, Hirst, O'Callaghan and Lasserson20–Reference Lassen, Hallas and Schaffalitzky de Muckadell22} The weighted average disability weight was calculated by summing the product of the multiplication between the proportion of cases for each health state by their corresponding health state disability weight, across all health states within a disorder, and then dividing the sum by the proportions of cases across all health states. As the cases were sampled from hospital registers or based on retrieving prescription drugs, no cases were considered asymptomatic, thus the sum of the proportions could be <1. The weighted disability weight for each MSD and GMC included in this study are described in Supplementary Tables 1 and 2.

Statistical analysis

The main outcomes of this study were YLDs and HeLP associated with each MSD. This study is a descriptive study and there were no formal tests between specific variables. YLDs and HeLP associated with specific disorders is the share of disability arising from experiencing these disorders. YLDs were estimated as the duration of a disorder (in years) multiplied by the disorder-specific disability weight (summing all individuals at the population level). Rates of YLDs were calculated as the absolute YLDs divided by person-years in the entire population. YLDs were calculated from date of onset of each episode until date of remission, death or 31 December 2015, whichever came first. To account for the fact that some individuals experienced more than one disorder simultaneously, comorbidities were estimated with a multiplicative function, so that no combinations of diseases had a disability weight equal to or greater than 1, which is in line with methods in the GBD studies.^{Reference Vos, Lim, Abbafati, Abbas, Abbasi and Abbasifard2,Reference McDonald, Haagsma, Cassini and Devleesschauwer23}

HeLP was estimated as YLDs among those experiencing a specific index disorder divided by person-years in those exposed to the index disorder. This estimate can be interpreted as the average proportion of health loss that individuals diagnosed with a specific index MSD experience because of the index disorder and additional comorbid psychiatric and GMCs.

Confidence intervals were estimated by bootstrap with 1000 iterations, which additionally incorporated uncertainty in the disability weights and remission rates, by selecting a random number from a triangular distribution based on these two distributions for each iteration.

The study was approved by the Danish Health Data Authority, Danish Data Protection Agency and Statistics Denmark. All data were de-identified and not recognisable at an individual level. No informed consent is required for a register-based study based on anonymised data, according to Danish law.