Abstract
Purpose :
Studies have shown in diabetics that decreased renal function is strongly associated with retinopathy. In some administrative medical claims databases, both diagnosis code and laboratory results are available to identify decreased renal function, but no study has evaluated whether there are differences in their associations with retinopathy. In this study, we compared the association of kidney disease (KD) identified through diagnosis code to lab value-based identification of KD with development of vision threatening diabetic retinopathy(VTDR).
Methods :
A national administrative medical claims database was used for this retrospective observational study. All individuals age ≥ 18 years with diagnosed nonproliferative diabetic retinopathy(NPDR) were followed longitudinally. ICD9/10 CKD diagnoses from outpatient claims were used to classify KD without or with end stage renal disease(ESRD). Serum creatinine results were used to calculate estimated glomerular filtration rates(eGFR) which were categorized based on the National Kidney Foundation’s levels of disease. VTDR was defined as any new diagnosis of diabetic macular edema or proliferative diabetic retinopathy. Multivariate Cox models were used to assess the associations of KD diagnosis and eGFR(modelled as time-dependent variables) with progression to VTDR, controlling for demographics and time-updating covariates (systemic health, laboratory values, insulin use). Hazard ratio(HR), C-statistic(a measure of model discrimination) and their 95% confidence intervals(CI) were calculated.
Results :
Among 69,982 enrollees with NPDR, 12,770(18.2%) developed VTDR during a median of 1.5 years follow-up. In multivariate analysis, lower eGFR was associated with higher risk of VTDR(eGFR 15-29:HR=1.14;95%CI:1.02-1.27,p=0.02; eGFR<15:HR:1.37;95%CI:1.25-1.50,p<0.001) compared to patients with eGFR ≥ 90. Whereas a diagnosis of ESRD was associated with higher risk of VTDR(HR=1.07,95%CI:1.01-1.13,p=0.02), but a diagnosis of KD without ESRD was not (HR=0.97,95%:0.92-1.03,p=0.35) when compared to normals. C-statistic was 0.60(95%CI:0.59-0.60) for the model with eGFR and for the model with KD diagnosis.
Conclusions :
Both diagnosis-based and lab value-based KD were associated with development of VTDR. There was no difference in their ability to predict development of VTDR.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.