Survey of clinical practice pattern in Germany’s certified chest pain units

A total of 15 evaluation benchmark questions were composed in a multiple-choice manner with a variable number of answers (Supplementary Information). The answer best describing the local practice should be selected for each question. The participants were asked to answer intuitively. The following categories were addressed:

All data are provided in a descriptive approach without further statistical analysis.

A total of 35 university hospitals, 148 academic teaching hospitals, and 31 community hospitals responded to the questionnaire. A total of 52 CPUs were allocated to northern Germany, 132 to central Germany, and 105 to southern Germany. At nearly equal levels, CPUs reported mean numbers of admittance of < 5 (42%) or 5–10 (44%) chest pain patients per day. Self-admission was estimated at < 25% in 28%, 25–50% in 60%, and > 50% in 12% without relevant differences between the different regions. University hospitals reported highest rates for the 5–10 group of CPU admissions, whereas community hospitals peaked at < 5. More than half of the CPUs (56%) and 26% of CPUs reported annual PCI rates of 250–500 or 500–1000, respectively. University hospitals reported the highest number, followed by academic teaching hospitals and community hospitals (Table 1).

The ESC 0/1‑h protocol was reported as the most commonly employed troponin protocol in 20% of CPUs involved in this survey, whereas the ESC 0/3‑h protocol was chosen as the strategy of choice in the majority of CPUs (77%). Longer protocols were stated to be favored by 3% only. The highest proportions of the ESC 0/1‑h protocol were from university hospitals (34%, Fig. 1a,b). Among all CPUs, 63% reported use of a clinical score to assess the ischemic risk, whereas 42% reported use of a score to assess the risk of bleeding, both without relevant differences between the levels of healthcare providers.

Most CPUs (39%) responded to the use of objective stress test criteria to select between an invasive or conservative strategy. Among all CPUs, 21% answered that their decision is mainly based on the absence or presence of secondary risk criteria for early invasive management. Use of cardiac computed tomography was reported in 5%. Subjective clinical decision or coronary angiography within 72 h was reported by 21% or 14%, respectively. Whereas academic teaching hospitals and community hospitals seemingly favored a noninvasive approach, university hospitals more often chose invasive management (Fig. 2a). When the answer “stress testing” was chosen, both university hospitals as well as academic teaching hospitals preferred conventional ergometry or stress echocardiography at nearly equal levels, community hospitals reported conventional ergometry as the stress test of choice (77%, Fig. 2b).

In NSTE-ACS patients with criteria of very high risk, 62% CPUs reported scheduling an invasive strategy analogous to STEMI patients with coronary angiography at the earliest occasion. Hospitals in southern and central Germany as well as university hospitals and academic teaching hospitals reported the highest percentages of immediate invasive management (Fig. 3a). As much as 99% of CPUs in this survey reported scheduling coronary angiography in NSTEMI within the first 24 h of admission without local or hierarchical differences. In 89%, this planned strategy was also followed on weekends and only 11% of the CPUs would postpone an invasive strategy to Mondays (Fig. 3b). Of those, university hospitals showed the highest rate of guideline adherence (Fig. 3c).

Detailed data on the preferential choice of the different P2Y12 inhibitors across the spectrum of ACS manifestations are given in Table 2. Prasugrel was selected most frequently for STEMI patients (72%), whereas ticagrelor was more often preferred for NSTEMI patients (58%), and clopidogrel (45%) or ticagrelor (38%) for patients with unstable angina undergoing PCI. In NSTEMI, only university hospitals tended to prefer administration of prasugrel. In addition, there was a regional preference in favor of prasugrel in the south as opposed to ticagrelor in the north (Fig. 4). Among all CPUs, 54% declared that the ISAR-REACT 5 results somehow changed their prescription behaviors.

Triple therapy was selected by 88% of the CPUs of this survey. About half of the CPUs (51%) stated a duration of 1 week to 1 month as default strategy, 23% chose a reduced duration to the length of the hospital stay, whereas 14% favored a prolonged administration of more than 1 month (Fig. 5). No relevant differences were observed among the different levels of healthcare providers. Nearly all CPUs favored NOAC administration (99%) over VKA during triple therapy. A preferred administration of a reduced dosage of NOAC was reported from 39% of the CPUs. University hospitals reported the highest preference of full NOAC dosage as their default strategy, at 77%. Whereas 73% of CPUs chose clopidogrel for combination with aspirin in triple or dual therapy in NSTE-ACS with an indication for oral anticoagulation, 24% of CPUs stated sometimes switching to ticagrelor, which was similar among the levels of healthcare providers.

Accumulating evidence from several observational trials, a randomized controlled trial, and a meta-analysis supports the diagnostic accuracy and safety of the ESC 0/1‑h algorithm [15‐20]. Although the ESC 0/3‑h algorithm that is based on the 99th percentile upper limit of normal has been established for years, direct comparisons of the ESC 0/1‑h versus the ESC 0/3‑h algorithm demonstrate small benefits in favor of the ESC 0/1‑h algorithm that are largely restricted to the numbers of patients qualifying for rule-out [21]. Therefore, a class IIa recommendation in favor of the ESC 0/1‑h algorithm is not unequivocally supported by robust evidence. Moreover, the evidence supporting the ESC 0/2‑h algorithm as an alternative to the ESC 0/1‑h algorithm, and thus in preference of the ESC 0/3‑h strategy, is not unopposed. The recommendation is limited by the fact that the ESC 0/2‑h algorithm is not widely distributed, and that this strategy has been almost completely derived from the APACE registry, with subsequent external validation against the accelerated diagnostic protocol of the 0/2‑h algorithm tested in the ADAPT study cohort [16, 22]. In their recent updated version of the certification criteria, the DGK also recommended a 0/1‑h or 0/2‑h protocol over a 0/3‑h protocol whenever a validated test is available [5]. According to our data, to date, German certified CPUs still perform a 0/3‑h protocol. This affects primarily non-university hospitals, but also less than a half of university hospitals are currently using a 0/1‑h protocol. Therefore, broad adaptions will be necessary to translate the ESC recommendations to the specific local setting as we anticipate that most of the certified units already have a test that is suitable for a faster diagnostic approach. Besides its accuracy in ruling-in ACS patients to NSTEMI patients, it will enable decision-making and ruling-out MI 2 h earlier than the 0/3‑h protocol, thereby ensuring a faster workflow in the CPU and a shorter time until direct discharge of low-risk individuals. Keeping in mind that most CPUs in Germany are located within an emergency room setting, this advantage may be of additional relevance during the current COVID-19 pandemic [12].

The 2020 ESC guidelines propose a very different algorithm for the timing of the invasive strategy from the former 2015 ESC recommendations as the group of patients with intermediate-risk indicators assigned to a delayed invasive strategy is now downgraded in risk and is shifted to a selective invasive strategy. The decision for invasive angiography should be based on stress testing, preferably stress imaging and computed coronary angiography [10, 11]. Even if not strictly performed, at present, the certification update of the DGK still recommends a delayed invasive strategy for patients at intermediate risk whereas a selective invasive strategy is still only recommended for patients with unstable angina without additional risk indicators [5]. A debate has started on whether the 2020 recommendations of the ESC are supported by appropriate evidence. Regarding the latter, a recent meta-analysis on the benefits of earlier timing of invasive management did not demonstrate significant survival benefits in the overall pooled analysis but identified subgroups of patients with mortality benefits from an earlier invasive management including patients > 75 years, patients with a GRACE score > 140, those with diabetes mellitus, and patients with elevated cardiac troponin concentrations (not necessarily with a rise and/or fall of serial concentrations; [23]). However, these patients were not shifted to the early invasive but to the selective invasive group. The issue that emerged with the shift from invasive to noninvasive imaging is the anticipated increase in the number of stress imaging and/or coronary computed tomography examinations. Unfortunately, the survey disclosed a severe underutilization of stress imaging. In line with previous data from the German CPU registry, the proportion of cardiac computed tomography in unstable angina patients remains low at about 5% [8]. Interestingly, more than half of the CPUs still rely on conventional stress ergometry, although its use for diagnostic purposes was already discouraged in the 2019 ECS guidelines on chronic coronary syndrome due to inappropriate performance, and more specific tests are favored in the latest CPU certification criteria [5, 24, 25]. We anticipate that this might still be due to local barriers to more contemporary stress tests such as stress echocardiography, stress magnetic resonance tomography, or myocardial scintigraphy resulting from reduced local experience in some nonacademic hospitals, cutback in local resources for timely performance (both personnel and/or equipment), or a lack of adequate reimbursement for time- and resource-consuming alternatives. Whether or not a primarily noninvasive approach independent of the presence of secondary risk markers will also be advised to the German CPUs remains a matter of discussion and will depend on the national commentary on the ESC guidelines by the DGK.

Risk criteria for NSTE-ACS patients at very high risk as well as at high risk remained broadly unchanged. Still, criteria of very high risk should trigger immediate invasive measures just as in STEMI patients, whereas high-risk criteria should trigger early coronary angiography within 24 h [10]. For the latter category, while regular ischemic risk assessment at admission (e.g., GRACE-scoring) was weak at about 60%, our current survey demonstrates a nearly optimal in-hospital care as far as the timing of coronary angiography in NSTEMI patients is concerned. The CPU physicians indicated an optimistic expectation that about 99% of all NSTE-ACS patients at high risk are scheduled for coronary angiography within 24 h. Almost all CPUs state that early invasive strategy can be provided on weekdays and weekends, supporting previous data from the German CPU registry on on- and off-hour care in STEMI patients and single-center experience in NSTEMI patients [8, 26]. However, in 2018, CPU registry data controversially reported on a treatment paradox regarding the timing of early PCI in NSTEMI patients at very high risk. Overall, the proportion of patients who underwent coronary angiography within 24 h was almost 80% for patients with a NSTEMI diagnosis [27]. Unexpectedly, delays in coronary angiography were substantially longer as patients’ risk increased [28]. In our current survey, those patients at very high risk were identified with an indication for immediate coronary angiography within 2 h, or analogous to STEMI patients, in 62% of all CPUs. The higher the level of care, the higher the proportion of timely coronary angiography, which was best among university hospitals at about 80%. Interestingly, there was also a higher guideline adherence in southern and centrally located hospitals. This might be explained by the former-documented better CPU network coverage in southern and central/western parts of Germany [2].

According to the 2020 ESC guidelines, clopidogrel should be used when ticagrelor or prasugrel are not available, or when the novel antiplatelets are contraindicated (class IC). The selection the P2Y12 and the duration of a dual therapy should be based on the ischemic/bleeding risk balance of each patient. In agreement with precedent guidelines, prasugrel and ticagrelor are recommended with a class IB recommendation [10]. A new recommendation to prefer prasugrel over ticagrelor in patients planned for PCI (class IIa) is based on the subgroup of patients with NSTE-ACS from the ISAR-REACT 5 trial [13]. It is interesting to observe that the ISAR-REACT 5 trial already influenced the selection of antiplatelets, even before the online publication of the ESC guidelines. The current survey reveals that CPUs prefer the administration of prasugrel for STEMI whereas ticagrelor still remains the P2Y12 inhibitor of first choice for NSTEMI patients. The regional preference of prasugrel over ticagrelor in the southern parts of Germany at university and academic teaching hospitals but not at other healthcare facilities is difficult to interpret. It might be related to a selective multiplicator function of individual key opinion leaders or the effect of focused symposia.

According to the ESC guidelines, NOACs should be preferred over VKA in the absence of contraindications whenever patients with NSTE-ACS and nonvalvular AF receive PCI [10, 29]. The ESC guidelines recommend a default setting with duration of triple therapy for up to 1 week, and a switch to dual therapy with a NOAC plus clopidogrel for up to 12 months. Our survey hereby demonstrates that those studies are already integrated in the daily practice of CPUs, as nearly all CPUs are using NOACs and restricting triple therapy but 75% of CPUs prefer to administer a dual antiplatelet therapy for a maximum of 4 weeks. Interestingly, about 25% of CPUs are also using ticagrelor in combination with NOAC adjusted, a strategy that lacks broad evidence from randomized controlled trials and is not recommended by 2020 ESC guidelines [30‐33].

Estimations of the representative CPU delegates of our study suggest that nearly three quarter of all CPU patients are self-referrals, a proportion that is higher than previous data originating from the German CPU registry [8]. During the last few years and with increasing awareness, we anticipate a shift toward a lower threshold to self-refer to a certified CPU. Even though early self-referral is preferred over late activation of emergency medical service, oncoming CPU improvement efforts should address an increasing “open CPU concept,” redirecting patient flow and ensuring early risk assessment thereby underscoring the importance of the ESC 0/1-h algorithm [34, 35].

First, this survey used a standardized questionnaire to explore clinical practice patterns and information on guideline adherence. The replies reflect the subjective expectations of the lead physician responsible for each CPU, without provision of statistics or objective findings on clinical reality. As such, this survey is not a substitute for a national registry for quality assessment, guideline adherence, or benchmarking. Second, the reported clinical practice patterns indicate a high level of quality of care and very high adherence to the 2015 ESC guideline recommendations. However, the practice pattern was reported from certified CPUs that had passed audits in the past, and involved CPUs affiliated to PCI centers. Thus, a selection bias cannot be excluded. Hence, findings from this survey should be interpreted cautiously and cannot be generalized to other emergency departments or to CPUs without certification from the DGK. Third, physicians were not aware of the 2020 ESC guidelines on NSTE-ACS at the time of the survey. Therefore, physicians had no opportunity to adapt their clinical practice standards for adherence with the updated guidelines. If interpreted in a negative sense, instead of anticipating innovation and state-of-the-art medicine, one could also claim that many centers did not follow the guidelines that were valid at the time of survey.

Although we are aware of the fact that data from the current survey originate from intentional ideas rather than given scientific evidence, we strongly interpret our results as hypothesis-generating and best-practice stimulating for those certified CPUs discovering certain gaps in their daily routine as compared to general practice in other certified CPUs. Thus, in addition to its scientific character (with all the limitations of a survey), we also anticipate our data pool to have an educational character. Furthermore, the data originating from our survey should encourage periodic and obligatory patient data collection from all (or representative) certified CPUs on a regular basis addressing benchmarking and quality control. Obligation by the DGK and short-term exemplary data collection similar to the former German CPU registry might be a possible option.