A psychosocial risk index for poor glycemic control in children and adolescents with type 1 diabetes

Objective The aim of this study was to develop and validate a psychosocial screening tool to predict risk for poor glycemic control in children with type 1 diabetes. Methods Participants seen for psychological screening were 196 children aged 3–18 yr at diabetes diagnosis. A psychosocial risk index was developed to predict poor glycemic control [mean hemoglobin A1c (HbA1c) ≥ 9.5%; 80 mmol/mol] 1–4 yr post diagnosis. Cutoff scores were derived for multiple levels of risk from receiver operating characteristic (ROC) curves and likelihood ratios (LRs). Discrimination and calibration were examined in the sample, and validated in 1000 bootstrap samples. Ability to predict diabetes-related emergency-room (ER) visits and diabetic ketoacidosis (DKA) was also tested. Results The risk index accounted for 16.2% of variance in mean HbA1c, discriminated between children with and without poor glycemic control [area under the receiver operating characteristic curve (AUC) = 0.814, 0.713–0.915; p < 0.001], ER visits (AUC = 0.655, 0.561–0.748; p = 0.001), and DKA(AUC = 0.709, 0.588–0.830; p = 0.001), and was well-calibrated. Every one-point increase in score was associated with an absolute increase in risk for poor glycemic control of approximately10% (LRs = 1.7, 3.2, 5.8, and 9.3). Sensitivity and specificity were 0.68 (0.43–0.86) and 0.79 (0.72–0.84) for detecting patients at moderate risk, and 0.53 (0.29–0.75) and 0.91 (0.85–0.95) for detecting high-risk patients. The index performed equally well in validation samples. Conclusions This paper presents the first psychosocial risk index for poor glycemic control in children newly diagnosed with type 1 diabetes. It is brief, easily administered, and provides a single score that translates directly into an estimate of risk that can help guide routine diabetes care.