Table 1. Interrupted time-series regression results for weekly blood culture indicators before, during, and after the bottle supply shortage.
Results are based on segmented regression using the Prais–Winsten method, which accounts for first-order autocorrelation in weekly time-series data. This table summarizes changes in four blood culture indicators: (1) total number of submissions, (2) single-set submission rate, (3) number of blood culture submission episodes, (4) true-positivity rate, and (5) contamination rate, before, during, and after the blood culture bottle shortage period (epidemiological weeks 27 to 38). Immediate-level change represents the change in the outcome at the onset of the shortage period. Slope change during the shortage indicates the weekly trend change throughout the shortage period. Slope change after the shortage reflects the change in slope following the end of the shortage period. Model p-value reflects the overall statistical significance of each regression model, assessed using the F-test within the Prais-Winsten framework. R2 represents the proportion of variance in the outcome variable explained by the model. Statistical significance levels are defined as follows: *p < 0.05, **p < 0.01, and ***p < 0.001.
| Outcome variable | Immediate-level change (coefficient (p-value)) | Slope change during shortage (coefficient (p-value)) | Slope change after shortage (coefficient (p-value)) | Model p-value | R2 |
| Number of submissions | −12.088 (0.548) | −5.970 (0.005**) | −13.212 (0.052) | 0.0003*** | 0.4827 |
| Single-set submission rate | −0.079 (0.127) | +0.018 (0.025*) | +0.000 (0.998) | 0.0000*** | 0.5845 |
| Number of blood culture submission episodes | −8.803 (0.398) | -2.212 (0.035*) | -7.331 (0.034) | 0.2169 | 0.192 |
| True-positivity rate | −0.019 (0.671) | +0.013 (0.010*) | +0.038 (0.028*) | 0.2222 | 0.19 |
| Contamination rate | +0.021 (0.296) | −0.003 (0.166) | −0.004 (0.531) | 0.0366* | 0.292 |