INVESTIGATING THE IMPACT OF CONTEMPORARY RISK FACTORS FOR DIABETES COMPLICATIONS AND THEIR EVOLUTION ON RISK PREDICTION USING THE UKPDS 82 EQUATIONS
Author(s)
McEwan P1, Foos V2, Lamotte M3
1Health Economics and Outcomes Research Ltd, Cardiff, UK, 2IMS Health, Basel, Switzerland, 3IMS Health, Vilvoorde, Belgium
OBJECTIVES: UKPDS 82 provided updated and new event equations for use in type 2 diabetes mellitus (T2DM) that include new risk factor (RF) predictors. These new RFs do not routinely get reported in clinical studies; consequently, the objective of this study research was to report plausible baseline RF values and their time-dependent trajectories and quantify their impact upon predicted complication rates. METHODS: Availability of baseline and time-dependent RF data was assessed using a pragmatic literature review. Univariate sensitivity analysis of the UKPDS 82 equations (over a 40-year horizon ) was undertaken to assess the impact of low-density lipoprotein [LDL]; microalbuminuria (MA); heart rate (HR); white blood cell count (WBC): haemoglobin (Hb) and estimated glomerular filtration rate (eGFR) on predicted diabetes complications per 1,000 patients, using UKPDS baseline values (varied within the 95% central range). RESULTS: The review identified 32 studies reporting baseline RF values typically consistent with UKPDS (review versus UKPDS): LDL (2.6-3.8 versus 3.49mmol/l); MA (11-45% versus 6.5%); HR (67-72 versus 72bpm); WBC (5.7-7.9 versus 6.6x10/ml); Hb (12.4-14 versus 14.5g/dl); the exception was eGFR where baseline values (33-101 versus 77.5ml/min/1.73m) and decline (0.3-5.2ml/min/1.73m/year) varied widely. Utilising UKPDS 82 baseline values resulted in 877 macrovascular and 133 microvascular events predicted with 691 to 1,181 and 116 to 182, respectively, predicted in sensitivity analyses; drivers of risk were LDL and eGFR. Varying eGFR decline between 0.3 to 5.2ml/min/1.73m/year resulted in annual event rates for end stage renal disease (ESRD) between 0.021 to 1.251%; holding eGFR constant over time resulted in ESRD annual event rate of 0.020%, significantly lower than observed in UKPDS (0.13%). CONCLUSIONS: Appropriate specification of RF is important in diabetes modelling. This study suggests that the UKPDS profile is generally consistent with other identified T2DM populations. Modelling a decline in eGFR improved the predictive accuracy of ESRD incidence.
Conference/Value in Health Info
2015-11, ISPOR Europe 2015, Milan, Italy
Value in Health, Vol. 18, No. 7 (November 2015)
Code
PRM98
Topic
Methodological & Statistical Research
Topic Subcategory
Modeling and simulation
Disease
Diabetes/Endocrine/Metabolic Disorders