Abstract
Aims:
Current European Society of Cardiology guidelines state that repetitive monitoring and feedback should be implemented for ST-elevation myocardial infarction (STEMI) treatment, but no evidence is available supporting this recommendation. We aimed to analyze the long-term effects of a formalized data assessment and systematic feedback on performance and mortality within the prospective, multicenter Feedback Intervention and Treatment Times in STEMI (FITT-STEMI) study.
Methods:
Regular interactive feedback sessions with local STEMI management teams were performed at six participating German percutaneous coronary intervention (PCI) centers over a 10-year period starting from October 2007.
Results:
From the first to the 10th year of study participation, all predefined key-quality indicators for performance measurement used for feedback improved significantly in all 4926 consecutive PCI-treated patients – namely, the percentages of patients with pre-hospital electrocardiogram (ECG) recordings (83.3% vs 97.1%, p < 0.0001) and ECG recordings within 10 minutes after first medical contact (41.7% vs 63.8%, p < 0.0001), pre-announcement by telephone (77.0% vs 85.4%, p = 0.0007), direct transfer to the catheterization laboratory bypassing the emergency department (29.4% vs 64.2%, p < 0.0001), and contact-to-balloon times of less than 90 minutes (37.2% vs 53.7%, p < 0.0001). Moreover, this feedback-related continuous improvement of key-quality indicators was linked to a significant reduction in in-hospital mortality from 10.8% to 6.8% (p = 0.0244). Logistic regression models confirmed an independent beneficial effect of duration of study participation on hospital mortality (odds ratio = 0.986, 95% confidence interval = 0.976–0.996, p = 0.0087). In contrast, data from a nationwide PCI registry showed a continuous increase in in-hospital mortality in all PCI-treated STEMI patients in Germany from 2008 to 2015 (n = 398,027; 6.7% to 9.2%, p < 0.0001).
Conclusions:
Our results indicate that systematic data assessment and regular feedback is a feasible long-term strategy and may be linked to improved performance and a reduction in mortality in STEMI management.
Keywords
Introduction
The current European Society of Cardiology (ESC) guidelines state that “repetitive monitoring and feedback” should be implemented in local ST-segment elevation myocardial infarction (STEMI) management networks. 1 However, no reference is given for this strong recommendation, since no study has systematically investigated this important issue so far. In order to improve quality control initiatives in the STEMI treatment, all attempts at implementing standardized data assessment and continuous feedback interventions should be encouraged.2–4 Achieving optimal treatment goals and adherence to guideline-recommended treatment strategies require the establishment of predefined key-quality management instruments in routine clinical practice. 5 As ischemic time duration is a crucial determinant of infarct size in STEMI patients, early reperfusion therapy is critical in reducing morbidity and mortality.6–10 Although the importance of key-quality indicators for assuring optimal treatment results has already been suggested, clinical studies assessing the impact of feedback interventions are rare and, in particular, there is a lack of robust data on the long-term effects of feedback. Data from the Euro Heart Survey ACS III as well as the French FAST-MI Program have demonstrated that performance measures for timely reperfusion in STEMI patients improved significantly over time with greater use of primary percutaneous coronary intervention (PCI).11–13 Recently, a report from the large American Heart Association Mission: Lifeline Program has demonstrated that, despite improved quality of care in STEMI patients with time including a reduced first medical contact-to-device time from 93 to 84 minutes, there was a significant increase in in-hospital mortality from 5.7% to 6.3%. 14
Given the importance of early recognition and timely PCI treatment of STEMI patients for hospital survival, we addressed the feasibility and prognostic significance of a quality improvement initiative based on a regular feedback intervention in the setting of regional STEMI treatment networks within the Feedback Intervention and Treatment Times in STEMI (FITT-STEMI) trial.15,16 In an initial monocenter analysis, we demonstrated impressive improvements in treatment times (contact-to-balloon time and door-to-balloon time) after initiation of a formalized data feedback within a single, regional STEMI treatment network. 3 Transferring the intervention concept to a group of six other regional STEMI care networks confirmed the short-term effects observed in the monocenter analysis. 15 In recent years, the feedback system has increased to more than 50 PCI hospitals participating in the FITT-STEMI program.16–18 In the present paper, we aimed to analyze the long-term effects of regular feedback on performance measures and mortality, reporting the first six hospitals participating more than 10 years following study entrance.
Methods
Study design
The FITT-STEMI trial (NCT00794001) was designed as a multicenter study to prospectively determine the additional benefit of systematic and formalized data assessment and interactive feedback on times to reperfusion therapy for STEMI patients. The study protocol includes the implementation of a standardized feedback-driven quality management initiative for timely reperfusion therapy in regional cardiac care networks for the treatment of STEMI patients.15–18 Enrollment in the FITT-STEMI study consortium requires participating PCI-capable hospitals to prospectively collect data on treatment times and performance measurements of all consecutive patients presenting within 24 hours of symptom onset of STEMI.
After a monocenter pre-test phase conducted in a regional STEMI treatment network with a single PCI hospital not included in the present analysis, the study was commenced in October 2007. The recruitment phase ended in April 2008, when the last of six consecutively enrolled PCI centers had joined the study. These six PCI hospitals from the initial cohort (including two university hospitals and four community PCI hospitals from all over Germany) have now collected data on treatment times and received regular annual feedback interventions for more than 10 years.
The FITT-STEMI study protocol was approved by the ethics committee of the Medical Faculty at the University of Göttingen and by the local ethics committees of all participating PCI centers.
Data assessment
For each patient, clinical information on comorbid diagnoses, Killip classification, 19 Thrombolysis in Myocardial Infarction (TIMI) risk score for STEMI, 20 blood pressure, and heart rate were documented on a case-report form, as described. 15 In addition, times of symptom onset and arrival of the emergency medical service (EMS) on the scene, the duration of out-of-hospital treatment, and the transport time to the PCI center were documented, as were intra-hospital times for the transfer to the catheterization laboratory and the start of puncture and first balloon inflation. After web-based electronic data transfer, regular data quality checks (“source data verification”) were performed and allowed for an independent monitoring of all patient data. 15
Site-specific descriptive statistics on times of transportation from the scene as well as procedural delays during the time from the first medical contact to PCI treatment were assessed. In addition to time intervals of the treatment chain, procedural data were obtained. All data were presented and discussed during regular feedback meetings with members of the participating interdisciplinary treatment teams at each local study site, as described earlier.15–17 The feedback sessions were organized by the local principal investigator at each study center. In the first year after study entrance, meetings were performed quarterly and from the second year after joining the FITT-STEMI consortium annually.
Outcome measures
The presented analysis reports the results from the subgroup of the six initial FITT-STEMI hospitals by investigating the long-term effects of feedback interventions on performance measures, times to reperfusion, and overall hospital mortality. The following clinical data were obtained as parameters for cardiac care and outcome measures: percentages of patients with pre-hospital electrocardiogram (ECG) recordings, ECG recordings within 10 minutes after first medical contact, pre-announcement of STEMI diagnosis by telephone, direct transfer to the catheterization laboratory bypassing the emergency department, and contact-to-balloon time less than 90 minutes. 15
German Hospital Quality Report on PCI procedures
In addition, we considered unpublished outcome data from the German Hospital Quality Report on PCI procedures. The participation in this nationwide PCI registry was mandatory up to the year 2015 for all certified PCI-capable catheterization laboratories in Germany, which performed PCI procedures in patients with STEMI within 24 hours after ECG diagnosis.
Statistical analysis
Descriptive statistics were summarized as means with standard deviations for continuous variables and frequencies with percentages for categorical variables. For each study patient, the interval between first medical contact and hospital arrival as well the intra-hospital times to balloon inflation during PCI were calculated based on the timings reported. No formal adjustment for multiple testing was carried out. The time from initiation of the program at a center was divided into 3-month intervals, referred to as quarters. Two independent logistic regression models were computed to assess the predictive effect of a quarter as a proxy for the duration of feedback interventions on in-hospital mortality. The first model was adjusted for the following known prognosis-relevant confounders: age, TIMI risk score, and hemodynamic instability (cardiogenic shock and out-of-hospital cardiac arrest). 16 In a second model, contact-to-door and door-to-balloon time were additionally entered as dependent variables. For the purpose of illustration, splines were computed to capture trends in important procedural variables. The reported p-values are all two-sided, and p-values ≤0.05 were considered statistically significant. All statistical analyses were performed on a personal computer using the statistical software SAS version 9.4 (SAS Institute Inc., Cary, USA).
Results
Basic characteristics of the study cohort
In total, 7745 patients with acute STEMI (5660 patients with direct transport via EMS; 1237 transfer patients from other hospitals; 665 patients with self-referral to the PCI hospital; 183 patients with in-hospital myocardial infarction) were included during the 10-year study period. Among these, 124 patients (1.6%) had received pre-hospital thrombolytic therapy. The majority of patients were transported directly from the scene via EMS transportation and treated by primary PCI (4962 patients). Reasons for not performing acute coronary angiography were ongoing resuscitation (n = 29), comorbid stroke (n = 2) or cancer (n = 9), no informed consent (n = 15), no catheterization due to failure in vascular access (n = 2), ECG normalization (n = 10), missing STEMI diagnosis (n = 5), and others (n = 30). After angiography, PCI was not performed in 596 patients and the reasons for this included takotsubo cardiomyopathy (n = 43), myocarditis (n = 47), vasospasms (n = 19), or it was because of an unidentified target vessel (n = 207) or ineffective wiring (n = 55). A total of 36 patients with times from first medical contact to PCI longer than 6 hours were excluded from the analysis. A characterization of this final cohort of 4926 STEMI patients is given in Table 1. The majority of study patients were male (75%), and 9% were octogenarians. Cardiogenic shock was diagnosed in 631 patients (12.8%), and 529 patients (10.7%) had out-of-hospital cardiac arrest with return of spontaneous circulation. In total, there were 386 deaths (7.8%) during the hospital stay. Over the 10-year study period, both the number of cases per 3-month intervals (range 118–155; Figure 1(a)) and the percentage of STEMI patients treated with PCI (range 77.5–94.8%; Figure 1(b)) stayed relatively stable.
Baseline characteristics of STEMI patients transported by emergency medical service and treated within 360 minutes after first medical contact with PCI. Data are presented as percentages or means and standard deviations for the total study cohort and the comparison between the subgroups of patients treated in the first and 10th year after study entrance. The p-values are for the comparisons between the first and 10th year.
CABG: coronary artery bypass grafting; CTO: chronic total occlusion; ECG: electrocardiogram; LAD: left anterior descending artery; LBBB: left bundle branch block; LCX: left circumflex artery; LMCA: left main coronary artery; NIRA: non-infarct-related arteries: PCI: percutaneous coronary intervention; RCA: right coronary artery; SD: standard deviation; STEMI: ST-elevation myocardial infarction; TIMI: thrombolysis in myocardial infarction.
Long-term trends in (a) the number of PCI-treated STEMI patients and (b) the percentage of PCI-based revascularizations over 10 years of participation in the FITT-STEMI study.
Long-term changes in time components to reperfusion therapy
The results showed a continuous decrease in treatment times for all three important intervals: contact-to-door time (Figure 2(a)), door-to-balloon time (Figure 2(b)), and their combination as contact-to-balloon time (Figure 2(c)). Quantitative changes in the time components to reperfusion therapy were most prominent from the first to the second 3-month interval, as in the meantime the native clinics had their first feedback intervention. The reductions in treatment times from first medical contact to reperfusion were most prominent during the first 5 years after study inclusion, but also in the following 5 years we still observed substantial improvements in preventing non-essential delays in the pre- and intra-hospital management of STEMI patients.
Time-dependent changes in three important procedural variables for PCI treatment: (a) mean contact-to-door time (C2D); (b) door-to-balloon time (D2B); and (c) contact-to-balloon time (C2B).
Changes in predefined key-quality indicators for timely reperfusion therapy
All predefined key-quality indicators improved significantly from the first to the 10th year of study participation (Figure 3, Table 2). There was a significant increase in the percentage of patients with pre-hospital ECG recordings (83.3% vs 97.1%, p < 0.0001; Figure 3(a), Table 2). In addition, the ratio of pre-hospital ECG recording within 10 minutes after first medical contact was significantly higher in the last as compared with the first year of study entrance (63.8% vs 41.7%, p < 0.0001; Figure 3(b), Table 2). During the first 5 years of study inclusion, the ratio of patients preannounced by the EMS to the intervention cardiologist increased constantly and remained stable thereafter (Figure 3(c)). Nevertheless, this increase of a quality indicator from the first to the last year was again significant (77.0% vs 85.4%, p = 0.0007; Table 2).
Trends in five major key-quality indicators for STEMI treatment: (a) percentages of patients with pre-hospital ECG recordings; (b) ECG recordings within 10 minutes after first medical contact; (c) pre-announcement of STEMI diagnosis by telephone; (d) direct transfer to the catheterization laboratory bypassing the emergency department (ED); and (e) contact-to-balloon time (C2B) equal or less than 90 minutes.
Changes in predefined key-quality indicators from the first to the last year of study inclusion in a program aimed at improving STEMI treatment by feedback-based quality interventions.
ECG: electrocardiogram; STEMI: ST-Elevation Myocardial Infarction.
The proportion of patients transported directly to the catheterization laboratory without a delay at the emergency department increased continuously over time (Figure 3(d)). The lowest rate of direct transfer was observed in the first 3 months after study enrolment (20.3%), in which due to the study design no feedback intervention was performed. In contrast, the highest proportion of direct transfers was registered at the end of the assessment period (71.1%), which resulted in a more than 3.5-fold increase in the relative numbers of patients transported directly from the field to the catheterization laboratory (Table 2). This considerable increase in the proportion of direct transports from the first to the 10th year (29.4% vs 64.2%, p < 0.0001) indicated that the PCI hospitals participating in our quality improvement initiative had successively reorganized their pre- and intra-hospital pathways for STEMI management and benefited particularly at the end of the 10-year assessment period from the feedback interventions.
Improved guideline adherence for early PCI treatment
The proportion of patients achieving the guideline-recommended 90-minute limit from first medical contact to reperfusion increased continuously over the 10-year study period (Figure 3(e), Table 2). In the first 3 months after study inclusion before the regular feedback intervention sessions started, one out of four patients reached this key-quality criterion (25.0%). The percentage of patients with contact-to-balloon times of less than 90 minutes significantly increased from 37.2% in the first year to more than half of all STEMI patients (53.7%) in the 10th year of study participation (p < 0.0001).
Reduced mortality in STEMI patients treated in PCI centers participating in feedback intervention
Hospital mortality constantly decreased over time, as depicted in Figure 4(a). The overall reduced time to reperfusion was linked to a decline in mortality from 10.8% in the first year after study entrance to 6.8% in the last year (p = 0.0244; Table 2). A logistic regression model with mortality as dependent variable adjusted for known prognosis-relevant confounders confirmed that the quarter after study enrolment used as a proxy for the duration of participation in the FITT-STEMI program was a statistically significant and independent predictor for mortality (odds ratio (OR) = 0.986, 95% confidence interval (CI) = 0.976–0.996, p = 0.0087; Table 3). When we additionally entered contact-to-door and door-to-balloon time as confounding procedural variables, which were both highly associated with mortality (for both p < 0.0001), also in this model the duration of participation in the study significantly predicted mortality (OR = 0.989, 95% CI = 0.978–0.999, p = 0.0366; Table 3).
In-hospital mortality in STEMI patients (a) from the FITT-STEMI study (n = 4926) and (b) the nationwide German Hospital Quality Report on PCI procedures for the indication “STEMI within 24 hours after ECG diagnosis” (n = 398,027). There was a decrease in mortality in FITT-STEMI study participants with time, whereas the mortality rate continuously increased every year in the German registry.
Results from two logistic regression models, with in-hospital mortality as dependent variable and time since study entrance in three-month intervals (quarter) as independent variable adjusted for the following confounders: age, Thrombolysis in Myocardial Infarction (TIMI) risk score, cardiogenic shock (CS), and out-of-hospital cardiac arrest (OHCA) (Model 1), and, additionally, contact-to-door and door-to-balloon time (Model 2). Data are presented as odds ratios (ORs) and 95% confidence intervals (CIs), including their p-values.
Increasing mortality in STEMI patients within the nationwide German PCI registry
In contrast to the decrease in mortality in the six regional STEMI care networks in FITT-STEMI, the results from the nationwide PCI registry showed a significant and continuous increase in in-hospitality mortality in all PCI-treated German STEMI patients from 6.7% (3290 deaths out of 48,922) to 9.2% (4453 out of 48,652) over an 8-year period from 2008 to 2015 (p < 0.0001; Table 4, Figure 4(b)).
Increased mortality in German STEMI patients during the years 2008 to 2015, as demonstrated by the German Hospital Quality Report on PCI procedures for the indication “ST-segment-elevation myocardial infarction within 24 hours after ECG diagnosis”. The results from this nationwide, mandatory PCI registry showed a continuous increase in in-hospital mortality per year.
ECG: electrocardiogram; PCI: percutaneous coronary intervention; STEMI: ST-elevation myocardial infarction.
Discussion
In this article presenting outcome data from the ongoing, prospective, and multicenter FITT-STEMI study, we report the long-term effects of implementing systematic data assessment and standardized feedback on improving treatment processes and shortening times from first medical contact to PCI-based early revascularization. Repeated feedback meetings were associated with a significant improvement in predefined quality indicators for guideline-recommended performance assessment, including percentages of timely pre-hospital ECG recordings and direct transportation to the catheterization laboratory bypassing the emergency department. All three important process-related time components, namely contact-to-door, door-to-balloon, and contact-to-balloon time, decreased significantly over time, depending on how long the centers had been participating in the study and had received regular feedback. Moreover, this continuous improvement over the decade in key-quality indicators for performance measurement was linked to a significant reduction in mortality. Our results showed that the most striking impact of feedback interventions had occurred in the immediate phase following study inclusion, when native clinics received their first interactive feedback sessions. The best performance results were achieved in the second half of the period, most probably because changes associated with the reorganization of the treatment chain had to be sustainably implemented. Also, in their 10th year of continuous and regular participation in this quality improvement program, the PCI hospitals were still able to significantly shorten those treatment times, which are known to be associated with reduced mortality.
In addition to the substantial improvements in performance parameters, the better clinical outcomes in the FITT-STEMI study may be due to changes in other domains of care not related to timely reperfusion, such as periprocedural administration of antithrombotic agents, pharmacological treatment of heart failure, and the use of mechanical assist devices. Therefore, we used data from the German Hospital Quality Report on PCI procedures performed in STEMI patients from 2008 to 2015, which served as an external reference. Unexpectedly, in the total population of all German PCI-treated STEMI patients, there was a continuous increase in mortality over time, which may reflect changes in the indication for reperfusion therapy. 21 However, important clinical data on out-of-hospital cardiac arrest, physical comorbidity, angiographical results, pre- and intra-hospital performance, as well as treatment times were not available in this registry. Similarly, the American Heart Association Mission: Lifeline Program also found no evidence of a reduced mortality over time. 14 The findings from these two large observational studies are in clear contrast to the results from our interventional study, suggesting that systematic feedback had contributed to the improved outcome.
Previous research has shown that increasing the rate of direct transports to the catheterization laboratory by preventing delays in the emergency department is a key element with respect to additional time savings in the intra-hospital treatment arm.17,22–28 Our results show that, even after 10 years of regular study participation, the maximal proportion of patients bypassing the emergency department has not yet been achieved, as no saturating effect was observed for the corresponding fitting curve (Figure 3(d)). Despite the endorsement for pre-arrival notification of the catheterization staff by the responsible emergency physician and promoting direct transports to the catheterization laboratory, more than one third of all STEMI patients (35.8%) failed to be directly transported and had a time-delaying stop at the emergency department. As has been reported recently, reasons for failed direct transfer were often multi-factorial, including diagnostic problems in the pre-hospital STEMI identification, difficulties in the readiness of the catheterization laboratory team, or the occupancy of the catheterization laboratory by other patients at the time of arrival. 17
In general, significant time savings were harder to achieve for PCI centers during the last 5 years of participation in the feedback project as compared with the first half of the decade. Moreover, it appeared that some essential components in the system of care were more resistant to structural changes over time than others which were readily optimized with little effort. For instance, the contact-to-door time remained relatively stable in the second half of the decade, suggesting that it may not be easily shortened with the available technical resources. Thus, further attempts to shorten treatment times in STEMI management should focus preferentially on structural changes in the organization of the catheterization laboratory readiness on a 7 day/24 hour basis and not in the organization of EMS transportation, which has a high quality in hospitals participating in the FITT-STEMI program. It is unclear whether innovative technological changes in EMS systems, such as a higher rate of helicopter transportation, will lead to further improvements. 29
Our findings from this prospective study demonstrating sustained trends for participating in a long-lasting quality improvement initiative need to be carefully interpreted in the light of several methodological limitations. First, since the recruitment strategy of the FITT-STEMI trial was not designed as a cluster-randomized approach and a direct control group of PCI hospitals not participating in this interactive quality improvement project was not included, any statement on causality regarding the effects of feedback intervention on outcome is formally not allowed. However, the fact that major improvements in the achievement of rapid revascularization were observed already in the second 3-month interval after the onset of feedback interventions as compared with the generally low performance in the first quarter without any feedback suggests some beneficial interventional effects. Second, unidentified changes in different domains of care not directly related to reperfusion therapy may additionally affect the better clinical outcomes shown in the FITT-STEMI study. Third, another limitation is that only in-hospital mortality data were available, but no information about long-term outcome. Fourth, the first six participating FITT-STEMI centers, having completed the 10-year study period, were most likely hospitals with a high performance on the time-to-PCI quality measure and, thus, our findings may not be generalized to other PCI centers with lower hospital volume. Fifth, due to methodological limitations in the study design and the presence of unknown confounders, any conclusions on causality based on the direct comparison between our data from the FITT-STEMI study and the nationwide German registry cannot be drawn.
In summary, long-term outcome data from the prospective FITT-STEMI trial indicated the feasibility and clinical utility of implementing high-quality improvement standards based on regular and sustained feedback interventions in the setting of regional STEMI care networks. Trend analyses facilitated the identification of essential components for further structural improvements, such as a desirable increase in the proportion of direct transfers to the catheterization laboratory. Based on these findings, we encourage existing cardiac care networks to improve their healthcare system readiness by promoting systematic data assessment and regular feedback. Further studies using a cluster-randomization approach for newly participating PCI centers are urgently needed to indeed confirm the beneficial prognostic impact of long-lasting quality improvement programs on both treatment times and mortality.
Footnotes
Conflict of interest
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Tim Friede reports personal fees for consultancies (including data monitoring committees) from AstraZeneca, Bayer, Biosense Webster, Boehringer Ingelheim, CSL Behring, Daiichi Sankyo, Enanta, Feldmann Patent Attorneys, Fresenius Kabi, Galapagos, IQVIA, Janssen, Mediconomics, Novartis, Penumbra, Parexel, Pharmalog, and Roche, all outside the submitted work. Furthermore, he has received research funding by the European Commission for statistical analyses on the EUTrigTreat (NCT01209494) and EU-CERT-ICD (NCT02064192) clinical studies. All relationships declared are modest. All other authors declare no conflict of interest.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by a grant from the German Heart Foundation and the Arbeitsgemeinschaft Leitender Kardiologischer Krankenhausärzte to KHS.
List of contributors
(in order of number of patients recruited)
Universitätsklinikum Würzburg (Irina Davydenko, Björn Lengenfelder, Sebastian K.G. Maier, Verena Reinhart), Universitätsklinikum Göttingen (Svetlana Hartmann, Claudius Jacobshagen, Lars S. Maier, Kristina Schröder), Klinikum Wolfsburg (Rüdiger Becker, Rolf Engberding, Claus Fleischmann), Klinikum Darmstadt (Hiller Moehlis, Gerald S. Werner), Asklepios Klinik Langen (Kerstin Eck, Hans G. Olbrich), Klinikum Worms (Jens Jung, Birgit Nicklas)