Evaluating the accuracy and use of continuous glucose monitoring in hospitalized patients: a systematic review and meta-analysis

Introduction:Glycemic variability in hospitalized patients is associated with poor clinical outcomes, longer length of stay, and increased cost. Advances in subcutaneous continuous glucose monitoring (CGM) technology may reduce glycemic excursions and improve clinical outcomes in hospitalized patients, yet CGM is not used due to a lack of evidence on accuracy. The objective of this study was to review and summarize the evidence on the accuracy of CGM in hospitalized, diabetic patients.Methods:A search was conducted for literature evaluating CGM accuracy in hospitalized, diabetic patients. Relevant studies were identified through full-text review and data was extracted for analysis. The primary outcome was any reported measure of accuracy. Descriptive analysis of data was performed but pooling of primary outcomes was not feasible due to the heterogeneity and nature of reported outcomes.Results:Systematic review returned 302 results, with 9 studies meeting inclusion criteria. All included studies assessed CGM accuracy using fingerstick blood glucose measurement as a reference. Reported measures of accuracy (# reporting) included mean absolute relative difference (MARD) (6), coefficient of variation (4), bias (2), and Clarke Error Grid analysis (7). MARD ranged from 9.4% to 14.9% though direction of deviation differed across studies and blood glucose ranges.Discussion:There is considerable variation in the analytic techniques used to assess CGM accuracy in the inpatient population. Outcome measures used to assess CGM data cannot be pooled using traditional meta-analysis methods. Reporting guidelines are necessary to make data comparable and determine the suitability of CGM for use in hospitalized patients.