The burden of diabetes-associated multiple long-term conditions on years of life spent and lost

Type 2 diabetes is a major conduit for diverse morbidity through chronic hyperglycemia, insulin resistance, endothelial dysfunction and inflammation. While guideline-driven prevention and risk factor control have reduced classic microvascular and macrovascular complications in many high-income countries, diabetes is increasingly linked to broader causes of morbidity and mortality, including cancer, infections, respiratory and liver disease, and dementia. Emerging evidence suggests diversification of complications with differing drivers across the age spectrum: increased life expectancy and reduced CVD risk in older adults, and rising obesity, changing risk profiles, and earlier-onset diabetes in younger adults. Multimorbidity (multiple long-term conditions, MLTCs) is therefore growing, posing burdens on patients and health systems that are often organized around single diseases. A critical gap has been quantifying years of life spent with and lost due to diabetes-related MLTCs across the life course. This study seeks to characterize the prevalence, age of onset, and the years lived with and lost associated with diabetes-related MLTCs in England, providing individual- and community-level metrics to inform monitoring, prevention, and health service planning.

Prior research has documented strong associations between diabetes and classic complications (cardiovascular, renal, neuropathic, and ophthalmic) and noted declines in certain complication rates (particularly cardiovascular) over recent decades. However, diabetes is also associated with broader morbidity—cancers, infections, respiratory and liver disease, and dementia—with relative risks that remain elevated and in some cases rising. Clustering studies indicate that multimorbidity in diabetes may aggregate along vascular, mental health, and aging/musculoskeletal axes, but these have not clearly quantified the MLTC phenotypes driving contemporary burden or their impact on life expectancy across the life course. There is also uncertainty about the extent to which MLTCs are driven by concordant (etiologically linked) versus discordant combinations of conditions. Existing metrics (counts of conditions) are crude and overlook the diversity and severity of MLTCs, highlighting the need for time-based, precise measures (years spent with conditions and years of life lost) to support policy, research, and care models.

Study design and data sources: The study used the National Bridges to Health Segmentation Dataset (England), comprising individuals registered with a general practice (GP) since 2014 and linked across more than 15 years of national, predominantly secondary-care datasets in the NHS National Commissioning Data Repository via pseudonymized NHS numbers. Analyses focused on 46,748,714 adults aged 20+ alive as of 31 March 2019, using the NHS financial year 2019–2020 (1 April 2019 to 31 March 2020) to avoid COVID-19 distortions. The dataset includes demographics (age, sex, ethnicity, deprivation), geography, and clinical diagnoses coded primarily from hospital data (ICD-10, OPCS, SNOMED CT). Thirty-five long-term conditions were prioritized based on prior work and Delphi consensus; condition logic and codes were clinically reviewed and validated against benchmarks (e.g., Quality and Outcomes Framework). Ethical approvals were not required due to the service evaluation remit and data governance within NHS England secure environments. Prevalence analyses: Point prevalence of all dual combinations of diabetes with each of the other conditions was estimated in March 2020; observed minus expected joint prevalence (expected = product of the marginal prevalences) quantified excess co-occurrence beyond chance. Counts of comorbid conditions by age and diabetes status were also calculated. Illness-death Markov modeling: A three-state illness-death (semi-competing risks) Markov model was constructed to estimate lifetime risk, median age at onset of each diabetes–comorbidity pair, years lived with the MLTC (YLW), age at death, and years of life lost (YLL) associated with MLTCs. States: healthy, illness (presence of the diabetes–condition pair of interest, irrespective of other conditions), and death. Allowed transitions: healthy→illness, healthy→death, illness→death; remission was not allowed. Age-dependent transition probabilities were estimated from monthly observations of condition and mortality status (April 2019–March 2020). For each age a, the probability of transition from i to j, Pij(a), equals nij divided by the sum over all possible destinations from i. Monthly transition matrices were raised to the 12th power to obtain annual transitions. Modeling horizon was 0–100 years. Inclusion thresholds and exclusions: For reliability, analysis of a diabetes–condition pair required at least 1,000 observed transitions of each type. Frailty was initially excluded due to observed remission incompatible with the model. Pairs with insufficient transitions (e.g., sickle cell disease, cystic fibrosis, autism, sarcoidosis, multiple sclerosis) were excluded from YLL/YLW modeling. Community-level metrics (per 1,000 population) were derived by multiplying person-level YLL and YLW by the lifetime risk of entering the illness state and scaling to 1,000. Outputs: Key outputs included lifetime risk of each MLTC pair, median onset age, YLW, YLL, and community burden expressed as years spent and years of life lost per 1,000 individuals, stratified by sex and age where applicable.

Population and prevalence: Of 46,748,714 English adults, 3,663,429 (7.8%) had diagnosed diabetes (type 1, type 2, and other forms). Across all ages, adults with diabetes were more likely to have MLTCs than those without diabetes. By age 50, approximately one-third of adults with diabetes have at least three conditions (women 37.0%, men 29.4%) versus 17.2% of women and 16.4% of men without diabetes. The prevalence of having three conditions in the non-diabetes population does not reach these levels until after age 70 in women and 65 in men. At age 40–44, women with diabetes were >3× and men >2× as likely as peers without diabetes to have four or more conditions; by age 70, 40% of both women and men with diabetes had at least four conditions versus 20% without diabetes. Comorbidity patterns: Common coexisting conditions with diabetes included hypertension (~51% women, ~50% men), CHD (25.7% men, 18% women), osteoarthritis (22.2% women, 14.6% men), depression (16.7% women, 10.0% men), and asthma (16.2% women, 9.5% men). A second tier (5–10% prevalence) included atrial fibrillation, cancer, cerebrovascular disease, COPD, heart failure, CKD, peripheral vascular disease, frailty, osteoporosis, and chronic pain. In younger adults (20–49 years), hypertension remained common (≈16–17%), CHD was present in 5.9% of men, and mental health conditions (depression ~18% overall; serious mental illness ~5%) and asthma (~13%) were prominent. Observed joint prevalence of diabetes with many conditions exceeded expected by chance alone. Age of onset and time metrics: Median onset of having at least two conditions occurred at age 67 in women and 66 in men; onset for three or more conditions typically occurred in the 70s. Women had onset 1–2 years later than men but lost a similar number of years relative to peers. Individuals with more conditions spent fewer years in the MLTC state and died earlier versus the general population. Example: those with three conditions lived about 10 years in that state and lost ~5 years; those with five or more lived ~5 years and lost ~6 years. Burden in years for specific pairs (per person): In younger adulthood, several combinations had large individual burdens. Serious mental illness and learning disability had early onset and were associated with many years lived with the conditions (about 14.0–15.4 years in women and 14.0–14.9 years in men) and substantial years of life lost (about 9.7–10.8 years in women and 10.3–10.6 years in men). Combinations with alcohol dependence and with chronic liver disease were associated with the largest reductions in years lived after diagnosis (alcohol dependence: 15.7 years lost in men, 13.5 in women; chronic liver disease: 13.1 in men, 12.5 in women). Asthma and depression with diabetes were associated with long years lived with the conditions (for women, asthma 16.6 years and depression 16.2; for men, serious mental illness 14.9; learning disability 14.0; asthma 13.8; depression 13.0). Community-level burden (per 1,000 individuals): Years spent with conditions were dominated by prevalent combinations: hypertension, depression, osteoarthritis, asthma, and CHD accounted for approximately 1,101–4,318 years per 1,000. In women, the largest contributions were hypertension (3,908), depression (2,261), osteoarthritis (2,199), and asthma (1,681); in men, hypertension (4,318), CHD (2,057), osteoarthritis (1,617), depression (1,375), and asthma (1,101). For years of life lost, leading contributors were hypertension, cancer, heart failure, and depression, totaling roughly 788–1,417 years per 1,000. Specifically, reduced years lived per 1,000: women—hypertension (1,255), depression (1,253), cancer (1,126), osteoarthritis (821), cerebrovascular disease (788); men—hypertension (1,417), cancer (1,200), heart failure (986), depression (963), CHD (900). Life-course acceleration: Diabetes advanced the onset of severe MLTCs by approximately 20 years in women and 15 years in men compared with the general population. By age 50, those with diabetes and at least three conditions spent over 20 subsequent years with MLTCs and died about 11 years earlier than the general population; each additional condition at age 50 was associated with approximately four fewer years of life. Discordant conditions: Beyond vascular complications, discordant conditions (mental health disorders, musculoskeletal disease, cancers, asthma) substantially contributed to MLTC burden, especially in younger and middle age, indicating diversification of diabetes-related morbidity.

The study demonstrates that diabetes-related MLTCs are extensive, heterogeneous, and vary by life stage. In older age, MLTCs encompass a broad spectrum of vascular complications (CHD, heart failure, CKD, stroke) and discordant conditions (osteoarthritis, cancers, asthma). By middle age, MLTC burden is already substantial, with about one-third of 50-year-olds with diabetes having three or more additional conditions, a pattern occurring 20 years earlier in women and 15 years earlier in men than in the general population and associated with about 12 fewer years of life. In younger and middle-aged adults, depression, asthma, and serious mental illness play prominent roles, with certain combinations (serious mental illness, learning disability, alcohol dependence, chronic liver disease) having large individual impacts on both years spent and years lost, though lower community-level prevalence tempers their aggregate impact. These findings align with MLTC clusters observed in prior work and suggest that MLTC subtypes—vascular, mental health, and musculoskeletal—differ in their contributions to years lived with and years lost. The results highlight a need for integrated, generalist-led care models that bridge mental and physical health, address polypharmacy, and coordinate across specialties. They also underscore potential etiologic drivers beyond classic cardiometabolic pathways, including obesity, social determinants, and treatment effects, and the need to refine prevention strategies to mitigate progression to complex MLTCs.

This national study quantifies the life-course burden of diabetes-associated MLTCs in both individual- and community-level terms, revealing marked acceleration of MLTC onset, long durations lived with multiple conditions, and substantial years of life lost. The burden extends beyond classic vascular complications to include mental health, musculoskeletal, respiratory, and oncologic conditions, particularly in younger adults. The presented metrics provide a baseline for monitoring MLTC burden, informing resource allocation and commissioning, and guiding innovation in prevention and integrated care models. Future research should delineate the relative roles of traditional and nontraditional risk factors in MLTC progression, evaluate and optimize models of integrated care for multimorbidity, and test lifestyle and therapeutic interventions targeting MLTC prevention and mitigation.

- Data sources were predominantly hospital and community-coded datasets rather than GP records, likely under-ascertaining conditions primarily diagnosed/managed in primary care (for example, hypertension, depression, non–end-stage CKD), leading to conservative estimates of time spent with conditions. - Diabetes-related eye diseases were not included, despite their contribution to disability burden. - Median onset of MLTC combinations was defined by the second condition, potentially obscuring variability in the timing and effects of the first condition. - Age-stratified differences may reflect both age effects and birth cohort effects, which could not be disentangled. - Years of life lost associated with MLTC combinations were not adjusted for additional coexisting conditions and should be interpreted as associations, not causal effects. - Type 1 and type 2 diabetes were not differentiated; given relative numbers, impact on YLL is likely modest overall but could affect estimates at younger ages. - Analyses were limited to 35 prioritized conditions; rarer conditions and combinations were excluded due to low prevalence and modeling constraints. - Frailty was initially excluded from modeling due to remission dynamics incompatible with the Markov framework; several rare condition pairs (sickle cell disease, cystic fibrosis, autism, sarcoidosis, multiple sclerosis) were excluded due to insufficient transition observations. - The model assumes no remission from illness and relies on monthly-to-yearly transition extrapolation, which may not capture all clinical complexities.