Abstract
INTRODUCTION: Blood cultures are essential for identifying bloodstream infections and guiding treatment. A global BD BACTEC™ bottle (BD, Franklin Lakes, NJ, US) shortage disrupted diagnostics, including at the Saitama Medical Center. During the shortage period, clinicians were instructed to prioritize cultures based on clinical necessity and infection focus, often opting for single-set submissions or deferring testing. This study examined changes in blood culture practices during the shortage period. METHODS: We conducted a retrospective observational study to analyze the blood culture data collected between January 1 and October 31, 2024. Five outcome indicators were evaluated: submission count, single-set submission rate, submission episode count, true-positivity rate, and contamination rate. The analyses were limited to patients aged ≥15 years to minimize pediatric-related biases. Segmented regression with Prais-Winsten adjustment was used to perform interrupted time-series analysis and account for autocorrelation and seasonal trends. Additionally, the detection frequencies of true bloodstream pathogens were compared between the pre-shortage and shortage periods using Fisher's exact test. RESULTS: Of the 7,073 total submissions, 5,592 adult blood culture sets were included in the analysis. The Durbin-Watson statistic ranged from 1.92 to 2.37 across models, indicating no significant autocorrelation in residuals. Weekly submission counts declined significantly during the shortage period (p = 0.005) and showed a marginal post-shortage decline (p = 0.052). Single-set submission rates increased significantly during the shortage (p = 0.025) but plateaued afterward. Submission episode counts declined significantly during the shortage period (p = 0.035) and continued to decrease after the shortage (p = 0.034), although no immediate-level change was observed at the shortage onset (p = 0.398). True-positivity rates increased during (p = 0.010) and after (p = 0.028) the shortage, which indicated improved diagnostic yield per test. Contamination rates remained stable throughout the experiment. No significant changes were observed in the distribution of key bloodstream pathogens, such as Escherichia coli, Staphylococcus aureus, and Enterococcus spp., which suggested that the diagnostic quality was maintained. CONCLUSIONS: The bottle shortage prompted fewer submissions, more selective sampling, and higher positivity rates without increased contamination. These findings highlight the value of targeted diagnostic stewardship and the need for adaptable, data-driven protocols to sustain care quality during future supply disruptions.