A prospective study evaluating the effect of a "Diagnostic Stewardship Care-Bundle" for automated blood culture diagnostics.

Objectives: We prospectively implemented a diagnostic stewardship care-bundle checklist, 'Sepsis-48 DSB', with the aim of reducing intervening duration of key steps of automated blood culture diagnostics (aBCD).

Methods: Sepsis-48 DSB was implemented for automated blood culture bottles (BCBs) received from adult intensive care units (AICUs) during the intervention period (P2; July 2020-June 2021) and intervening durations were compared with those during the retrospective, pre-intervention period (P1; March-June 2020). During both periods, provisional blood culture reports (pBCR) were issued wherein direct microbial identification (dID) was performed in BCBs with Gram-negatives by directly inoculating conventional biochemical tests and direct antimicrobial susceptibility testing (dAST) using EUCAST RAST method. The results were compared with the standard of care (SoC) method (i.e. full incubation followed by identification and AST by VITEK-2 Compact).

Results: During P2, significant reductions in loading time (LT; median: 63.5 vs. 32 minutes, P < 0.001), time to dID+dAST performance (TTD; 186 vs. 115 minutes, P = 0.0018) and an increase in compliance to bundle targets (LT ≤45: 44% vs. 66%, P = 0.006 and TTD ≤120: 34% vs. 51.7%, P = 0.03) were observed. Using dID+dAST method, results were read 694 minutes earlier than SoC method. Of 176 pBCR, 165 (94%) were concordant with SoC in microbial identification of species. Categorical agreement for any drug-bug combination was 92.7% (1079/1164) and corresponding major, very major, and minor error rates were 8.8% (19/216), 4.9% (45/921), and 1.8% (21/1164), respectively.

Conclusion: The 'diagnostic stewardship care-bundle' strategy was successfully implemented with considerable diagnostic accuracy leading to significant reductions in duration of targeted steps of aBCD.