We assessed the impact of feedback and improvement activities on our primary goals of improving the rate of PH-ECG performance, frequency of documented PH-ECG interpretation by EMS providers, and aspirin administration. We also evaluated the secondary goals of addressing factors associated with PH-ECG non-performance, improving paramedic sensitivity in identifying STEMI, identifying prevalence of hospital-confirmed STEMI within EMS systems, and determining changes in hospital D2B times and EMS total run time. To determine impact of the intervention, we compared performance measure results during the Baseline and Intervention periods. Data were collected on 6,994 patients: 1,589 in the Baseline Period and 5,405 in the Intervention Period. We checked for an interaction between period and site in the final model for our key outcome, PH-ECG performance. There was a slight site-based effect, of borderline significance, and the change in PH-ECG performance was in the same direction at both sites, but of different magnitudes, so, as planned a priori, we combined data from both sites for the final analysis.
In comparing PH-ECG performance and aspirin administration between the Baseline and Intervention periods (), we found that all of our measures improved, all statistically significantly. Prior to ongoing measurement and feedback, only 57% of patients who should be evaluated for ACS had a PH-ECG, compared to 84% (p<0.0001) after receiving feedback. Moreover, in the Baseline Period, even when a PH-ECG was done, the paramedic documented their interpretation in 84% of the cases, whereas in the Intervention Period this rose to 95% (p<0.0001). Aspirin administration in patients with symptoms of ACS also increased significantly from 75% to 82% (p= 0.001) once feedback was provided. With the intervention, performance of PH-ECGs for women rose from 53% (431/813) to 82% (2,241/2,742) (p<0.0001), and for men, from 62% (480/775) to 87% (2,305/2,660) (p<0.0001). Thus, despite an increase in PH-ECG for both genders, the baseline gender difference persisted; women still had a 30% lower odds than men of having a PH-ECG performed (OR 0.69; 95% CI 0.59, 0.80). Likewise, PH-ECG performance for patients with a communication barrier increased from 42% (84/201) to 73% (395/544) (p<0.0001), but those with a communication barrier remained 55% less likely to have a PH-ECG than those without barriers. This suggests that factors leading to disparities in PH-ECG performance persisted despite focused interventions.
PH-ECG performance and aspirin guideline adherence during baseline and intervention periods.
When PH-ECG measure feedback was provided and STEMI Alert protocols were in use, patients with confirmed STEMIs received PH-ECGs more frequently, median D2B times decreased, and the proportion of patients receiving PCI within recommended timeframes increased. The proportion of patients with confirmed STEMI who received a PH-ECG increased from 77% to 99% (p<0.0001) (). The median D2B time decreased from 158 minutes to 80 minutes (p=0.04) (), and the proportion of patients with STEMI who received PCI at the initial receiving hospital within the recommended 90 minutes D2B time, rose from the baseline of 27% to 67% (p<0.0001) (). Once a STEMI Alert protocol was in place, paramedic staff reported greater attention to PH-ECGs by ED staff and the initiation of direct transport to the cardiac catheterization laboratory (however, data on these rates and on inappropriate catheterization laboratory activation were not collected).
Importantly, these improvements caused no increase in the total EMS run time. The median run time from arrival on-scene to hospital arrival remained 22 minutes throughout the project. Paramedics had initially expressed concerns about delaying hospital transport by performing PH-ECGs, but there was no increase in total run time; perhaps becoming more efficient in performing PH-ECGs as they did so more frequently.
There are several limitations in the study intrinsic to its being conducted in a “real world” setting. It was incorporated into daily operations of EMS agencies in order to acquire, as much as possible, results generalizable to other similar settings. Process measures were collected, but outcome measures were not tracked; a much larger study would be needed to be able to detect direct impact on outcomes. The project participants were not blinded to the intervention and were aware of the aims to improve the care of patients with ACS. Awareness of being observed may have changed performance; although this would be a feature of this approach in the “real world.” Of note, the before-after quasi-experimental interventional study design used in this study provides a less strong demonstration of causality than the fully randomized approach used in other kinds of interventions. The consistency of effect at the two sites and that the interventions' timing linked with the improvement in performance all help compensate for the limitations of this study design. Nonetheless, additional studies confirming our results will be helpful.
Another limitation is that assessment of performance was abstracted from patient care records, and completeness of documentation varied and may have influenced study inclusion. Additionally, documentation of EMS “STEMI Alert” notification to the ED was difficult to collect as were cardiac catheterization laboratory activation times. This prevented evaluation of the frequency and impact of the “STEMI Alert.” Also, the two participating EMS systems were private/private-public entities in suburban/urban settings, and may not be broadly representative. Performance measures were developed considering local guidelines; other agencies may choose to develop different measures relevant to their local setting. Despite these limitations, the consistency of this study's findings in improving prehospital care for ACS suggests that the results are generalizable, although further research will be needed to verify this.