Electronic Medical Record Implementation Incentivises Accurate Patient Identifier Entry for Point-of-Care Glucose Measures: Another Step Towards Improved In-Hospital Diabetes Care
Title: Electronic Medical Record Implementation Incentivises Accurate Patient Identifier Entry for Point-of-Care Glucose Measures: Another Step Towards Improved In-Hospital Diabetes Care
Background: Clinical service delivery and research in the field of inpatient diabetes care is contingent on accurate documentation of point-of-care glucose results. Networked glucose meters require correct patient identifier entry to link glucose data with a patient’s chart.
Aims: To evaluate accurate entry of patient identifiers using a networked glucose meter system, before and after implementation of an electronic medical record.
Methods: This retrospective observational study evaluated glucose meter patient identifier entry at a quaternary hospital in Victoria, Australia from September 2019 (introduction of networked blood glucose meters), until May 2021, 9 months following hospital-wide EMR implementation. Identifier entry was considered accurate if string length matched the length of the identifier type in use. The four weeks following EMR implementation were censored to account for peri-implementation workflow disruptions.
Results: Over 79 weeks, 76.3% of glucose measures (269,199/352,841) had accurate patient identifier entry. Pre-EMR entry accuracy was 71.6% (132,918/185,677) while post-EMR accuracy was higher at 81.5% (136,281/167,164), p<0.001. Interrupted time series modelling showed the deterioration in identifier entry accuracy observed from program inception (slope -0.20 percentage points/week) transformed to a marked and sustained improvement following EMR implementation (slope +0.36 percentage points/week, p<0.001).
Conclusions: While patient identifier entry accuracy deteriorated pre-EMR, post-EMR improvements may relate to aligning incentives in this period. Pre-EMR entry accuracy was not incentivised, whereas post-EMR accurate entry resulted in automatic glucose data transfer to the clinical record, saving the frontline clinician time and effort. These results have implications for human-centred electronic workflow design and implementation, to optimise delivery of in-hospital diabetes care.