Using routine clinical features to classify adult-onset diabetes at diagnosis: the StartRight prospective observational study

ObjectivesIt is not known which clinical features optimally differentiate type 1 and 2 diabetes at diagnosis. We aimed to determine which clinical features differentiate adult-onset type 1 and 2 diabetes at diagnosis and develop classification models combining these features with and without islet-autoantibodies. DesignA prospective cohort study with prediction model development and validation. SettingUK primary and secondary care. Participants1800 adults ([≥]18 years) diagnosed with diabetes in the previous 12 months, excluding known secondary or monogenic diabetes. Main outcome measuresType 1 and 2 diabetes defined by a combination of insulin treatment and endogenous insulin production (measured using C-peptide) assessed [≥]three years after diabetes diagnosis. ResultsEleven clinical features and routinely measured biomarkers discriminated type 1 from type 2 diabetes independently of diagnosis age and BMI. Lower age-at-diagnosis, BMI and waist-hip ratio, unintentional weight-loss, and higher presentation HbA1c or glucose were the most discriminative features, with other features only weakly discriminative. Models integrating clinical features with and without islet-autoantibodies, developed in those age 18-50 years at diabetes diagnosis, had high performance in internal validation (clinical features only: AUCROC (95% CI) 0.94 (0.93, 0.96), clinical features and islet-autoantibodies: AUCROC 0.97 (0.96, 0.98)), and maintained high discrimination in older adults (age >50 at diagnosis; clinical features only: AUCROC 0.93 (0.90, 0.96), clinical features and islet-autoantibodies: AUCROC 0.97 (0.94, 0.99)). Simplifying the models to a point-based score (the StartRight Score) resulted in similar performance. These models had higher performance than current clinical guidance. In UK primary care data models were strongly predictive of outcomes associated with type 1 diabetes, including in those initially treated as type 2 diabetes. ConclusionsLower age-at-diagnosis, BMI, and wait-hip ratio, unintentional weight-loss and high presentation glycaemia are the most discriminative features for diagnosis of type 1 diabetes in adults. Models combining routine clinical features, with or without islet-autoantibodies, have high accuracy and could assist clinical classification and prioritisation of classification biomarker testing. Study registrationhttps://clinicaltrials.gov/study/NCT03737799 Summary boxesO_ST_ABSSection 1: What is already known on this topicC_ST_ABSO_LIMost type 1 diabetes occurs in adults, but differentiating it from type 2 diabetes, which is much more common, is challenging, and misclassification is common. C_LIO_LIAge-at-diagnosis and BMI are currently the only clinical features robustly shown to distinguish between type 1 and type 2 diabetes at diagnosis; many other features included in textbooks and guidelines have little supporting evidence. C_LIO_LIGuideline bodies, including the UK National Institute for Health and Care Excellence (NICE), have identified a need for evidence on what features discriminate type 1 and 2 diabetes in adults, and how these features can be combined to improve diagnosis. C_LI Section 2: What this study addsO_LIThis is the first study to prospectively assess the utility of clinical features for diabetes subtype at diagnosis. C_LIO_LIThe five most discriminative routine clinical features for distinguishing type 1 from type 2 diabetes at diagnosis are age-at-diagnosis, BMI, waist-hip ratio, pre-diagnosis unintentional weight-loss, and presentation glycaemia (HbA1c or glucose). C_LIO_LIMany features included in current guidelines were only very weakly discriminative of subtype, and no single clinical feature was able to adequately differentiate between type 1 and type 2 diabetes alone. C_LIO_LIA clinical prediction model combining ten routinely available clinical features, with or without islet-autoantibodies, as both a prototype calculator and a points-based score (the StartRight Score), had high accuracy in differentiating type 1 from type 2 diabetes and outperforms current clinical guidance and islet-autoantibody assessment alone. C_LI