Comorbidities and mortality in subgroups of adults with diabetes with up to 14 years follow-up: a prospective cohort study in Sweden.

Abstract

Subgroups of adult-onset diabetes, namely severe autoimmune diabetes (SAID), severe insulin-deficient diabetes (SIDD), severe insulin-resistant diabetes (SIRD), and mild obesity-related diabetes (MOD) or mild age-related (MARD) diabetes, have been defined with clinical variables and a machine-learning approach. Our aim was to describe their long-term outcomes and mortality. In this prospective cohort study in Sweden, we used data from two subsets of the All New Diabetics in Scania (ANDIS) project cohort of individuals diagnosed with diabetes at regional care centres and enrolled within 1 year of diagnosis. Included participants were 18 years or older, did not have pancreatitis, and had complete data for cluster variables. We used GAD antibodies, Homeostasis Model Assessment 2 β cell and insulin resistance indices, BMI, HbA_1c, and age at the diagnosis of diabetes to group individuals, and logistic and Fine-Gray proportional hazards regression to study prevalent and incident comorbidities, using the MARD group as comparator. Between Jan 1, 2008, and Nov 3, 2016, for ANDIS1 and Nov 4, 2016, and April 6, 2022, for ANDIS2, a total of 25 590 were screened for eligibility, resulting in 19 076 participants being included in the analysis (9057 from ANDIS1 and 10 019 from ANDIS2; 11 171 men and 7905 women). The median follow-up time was 9·63 years (IQR 4·05) in ANDIS1 and 2·83 years (2·76) in ANDIS2. The SAID and SIDD subgroups had the highest HbA_1c values at diagnosis and over time, and the highest age-adjusted and sex-adjusted risk of retinopathy (SAID adjusted hazard ratio [HR] 1·35 [95% CI 1·08-1·70]; SIDD 2·11 [1·82-2·44]) and neuropathy (2·58 [1·87-3·56]; 2·13 [1·69-2·70]). At the diagnosis of diabetes, SIRD had the highest prevalence of hypertension (2336 [69·8%] of 3348) and dyslipidaemia (1484 [44·4%]), and kidney (366 [10·9%]), cardiovascular (1026 [30·6%]), and steatotic liver disease (38 [1·1%]). Despite large differences in HbA_1c, both SIDD and SIRD had an increased risk of incident kidney disease, including kidney failure (SIDD adjusted HR 2·94 [1·69-5·09]; SIRD 3·41 [2·06-5·64]), and myocardial infarction (1·44 [1·13-1·82]; 1·51 [1·22-1·87]). SIRD and MOD had the highest risk of atrial fibrillation (adjusted HR 1·32 [1·13-1·54]; 1·58 [1·26-1·98]). The risk of stroke was only increased in SIDD (adjusted HR 1·32 [1·07-1·62]), and steatotic liver disease (3·29 [2·32-4·66]) and heart failure (1·55 [1·34-1·79]) in SIRD. SIDD, SIRD, and MOD had the highest risk of total mortality (adjusted HRs 1·44-1·52) even after adjustment for established risk factors (age, sex, BMI, hypertension, LDL, and smoking), largely driven by cardiovascular mortality. Diabetes subgroups could inform on outcomes, as well as guide treatment and follow-up needed for newly diagnosed individuals with diabetes. SIRD stands out as a high-risk subgroup that is not identified by conventional glycaemia-based risk factors, but bears risk of early onset end-organ damage and would benefit from identification and treatment before the diagnosis of diabetes. Swedish Research Council, Avtal om Läkarutbildning och Forskning Swedish government grants, Diabetes Wellness Sweden, the Swedish Heart and Lung Foundation, the Crafoord Foundation, the Swedish Diabetes Foundation, the Novo Nordisk Foundation, the Bo and Kerstin Hjelt Foundation, the Albert Påhlsson Research Foundation, Vinnova, and AstraZeneca.