James C Boyd1, David E Bruns1. 1. Department of Pathology, University of Virginia Health System, Charlottesville, Virginia, USA.
Abstract
INTRODUCTION: Patient outcomes, such as morbidity and mortality, depend on accurate laboratory test results. Computer simulation of the effects of test performance parameters on outcome measures may represent a valuable approach to defining the quality of assay performance that is needed to provide optimal outcomes. METHODS: We carried out computer simulations of patients on intensive insulin treatment to determine the effects of glucose meter imprecision and bias on (1) the frequencies of glucose concentrations >160 mg/dL; (2) the frequencies of hypoglycemia (<60 mg/dL); (3) the mean glucose; and (4) glucose variability. For each patient, starting with a randomly selected initial glucose concentration and individualized responsiveness to insulin, hourly glucose concentrations were simulated to reflect the effects of (1) IV glucose administration, (2) gluconeogenesis, (3) insulin doses as determined using regimens from the University of Washington and Yale University, and (4) errors in glucose measurements by the meter. For each of 45 sets of glucose meter bias and imprecision conditions, 100 patients were simulated, and each patient was followed for 100 h. RESULTS: For both insulin regimens: Mean glucose was inversely related to assay bias; glucose variability increased with negative assay bias and assay imprecision; frequencies of glucose concentrations >160 mg/dL increased with negative assay bias and assay imprecision; and frequencies of hypoglycemia increased with positive assay bias and assay imprecision. Nevertheless, each regimen displayed unique sensitivity to variations in meter imprecision and bias. CONCLUSIONS: Errors in glucose measurement exert important regimen-dependent effects on glucose control in intensive IV insulin administration. The results of this proof-of-principle study suggest that simulation of the clinical effects of measurement error is an attractive approach for assessment of assay performance requirements.
INTRODUCTION:Patient outcomes, such as morbidity and mortality, depend on accurate laboratory test results. Computer simulation of the effects of test performance parameters on outcome measures may represent a valuable approach to defining the quality of assay performance that is needed to provide optimal outcomes. METHODS: We carried out computer simulations of patients on intensive insulin treatment to determine the effects of glucose meter imprecision and bias on (1) the frequencies of glucose concentrations >160 mg/dL; (2) the frequencies of hypoglycemia (<60 mg/dL); (3) the mean glucose; and (4) glucose variability. For each patient, starting with a randomly selected initial glucose concentration and individualized responsiveness to insulin, hourly glucose concentrations were simulated to reflect the effects of (1) IV glucose administration, (2) gluconeogenesis, (3) insulin doses as determined using regimens from the University of Washington and Yale University, and (4) errors in glucose measurements by the meter. For each of 45 sets of glucose meter bias and imprecision conditions, 100 patients were simulated, and each patient was followed for 100 h. RESULTS: For both insulin regimens: Mean glucose was inversely related to assay bias; glucose variability increased with negative assay bias and assay imprecision; frequencies of glucose concentrations >160 mg/dL increased with negative assay bias and assay imprecision; and frequencies of hypoglycemia increased with positive assay bias and assay imprecision. Nevertheless, each regimen displayed unique sensitivity to variations in meter imprecision and bias. CONCLUSIONS: Errors in glucose measurement exert important regimen-dependent effects on glucose control in intensive IV insulin administration. The results of this proof-of-principle study suggest that simulation of the clinical effects of measurement error is an attractive approach for assessment of assay performance requirements.
Authors: Boris P Kovatchev; Christian A Wakeman; Marc D Breton; Gerald J Kost; Richard F Louie; Nam K Tran; David C Klonoff Journal: J Diabetes Sci Technol Date: 2014-06-13