Lost in Transcription: Frequency of Inaccurate Manually Documented Point‐of‐Care Blood Glucose and Ketone Measures in Hospital

BACKGROUND: Optimal inpatient diabetes management requires accurate blood glucose (BG) and ketone (BK) documentation. In 2019, Royal Melbourne Hospital implemented hospital-wide networked blood glucose monitoring (NBGM) (NovaBiomed StatStrip), which automatically uploads BG/BK values to point-of-care device software (Bioconnect). Until electronic health record (EHR) implementation in 2020, nurses manually recorded BG/BK data into paper-records, enabling blinded gold-standard accuracy assessment by digital software. This study evaluated the accuracy and potential clinical significance of inaccurate manual POC BG/BK values compared with NBGM records. METHODS: DINGO POC, a sub-analysis of the Diabetes IN-hospital: Glucose and Outcomes (DINGO) study, audited paper-based glucose charts against NBGM-uploaded BG/BK values and BG time-stamps to identify manual transcription inaccuracies. Discrepancy rates, magnitude and potential clinical significance were analysed. RESULTS: 4391 BG and 378 BK NBGM measures from 250 admissions over a two-month period were assessed. Of BG measures, 325 (7.4%) were not recorded in patient charts and 558 (13%) were inaccurate. 302 (54%) inaccurate BGs had potentially clinically significant discrepancies (≥ 0.4 mmol/L). 1570 (36%) BG time-stamps were recorded inaccurately, and 329 (7.5%) were not recorded. 524 (33%) inaccurate BG time-stamps had discrepancies > 15 min. Of BK measures, 153 (41%) were not recorded and 18 (8%) were transcribed inaccurately. Inaccuracy rates were similar across wards and patient groups. CONCLUSION: Manual transcription of POC BG/BK values and time-stamps, evaluated against blinded concurrent gold-standard digital software, was often inaccurate with potential for clinical harm. Implementation of hospital-wide NBGM systems with automated POC BG/BK upload to EHRs may mitigate manual transcription errors.