171 Variation in random capillary blood glucose and HbA1c as predictors of cystic fibrosis related diabetes (CFRD)

Abstract

Objectives The Leeds CF unit implemented electronic care records (ECR) in 2007 coded for all aspect of CF disease. Subjectively, patterns of disease progression have become evident, including increased variation in blood glucose measures prior to diagnosis of CFRD. The aim of this study was to determine objectively whether a.random capillary blood glucose (RCBG) variation and HbA1c differs between those who develop CFRD in the 3 years preceding diagnosis and controls, b.univariate and/or multivariate analysis using longitudinal datasets can predict onset of CFRD. Method Adult patients were identified using the ECR. Clinical data was extracted and interquartile range and median value for RCBG from routine clinic visits, median HbA1c and days of intravenous antibiotics in the 3 years preceding diagnosis of CFRD as well as non CFRD controls was calculated. Results 156/491 patients had a diagnosis of CFRD. Of these, 16 patients and 179 controls had 3 years of RCBG measurements [16 CFRD (6M/10F), mean age 32.3(±8.0) years v 179 controls (116 M/63F), mean age 29.5(±7.5) years]. Differences in parameters are shown in the table. 3 years preceding diagnosisCFRDControlKSSigVariation of RCBG (IQR)2.351.760.37p = 0.03Median RCBG6.85.90.54p = 0.0002Median HbAlc46400.69p Conclusion Epidemiological datasets extracted from ECR reveal early potential for prediction of CFRD from multivariate datasets. Multivariate Gaussian mixtures modelling of HbA1c and RCBG show potential as an indicator of CFRD that can be monitored over time, which might be realised with larger sample sizes.