As previously reported,10 the CROSS-AMI trial was a multicenter randomized clinical trial of patients with STEMI and multivessel disease comparing complete anatomic revascularization of all nonculprit coronary lesions during initial hospitalization (ComR) with selective revascularization based on stress echocardiography-guided detection of ischemia (SelR).10 The trial received ethics committee approval and was conducted in accordance with the principles of the Declaration of Helsinki (reference 2010/160). It was registered at ClinicalTrials.gov under identifier number NCT01179126.

A total of 306 patients from our hospital were recruited for the trial between October 2010 and October 2015. They all provided signed informed consent. In brief, patients with STEMI who had undergone emergency PCI to treat an infarct-related artery (IRA) were eligible for inclusion if they had significant stenosis (70% by visual estimation) in at least 1 coronary artery other than the IRA (see inclusion and exclusion criteria of the supplementary data). They were randomized to the ComR or SelR arm using a pseudorandom number system within 48hours of the index PCI. Patients in the ComR group (n=154) were scheduled for a second PCI to treat all significant lesions prior to discharge. A pressure wire was not used to select which lesions to treat. Patients in the SelR group (n=152) underwent an exercise or dobutamine stress echocardiogram during the index admission. As per protocol, PCI was indicated for nonculprit lesions in patients with recurrent spontaneous myocardial ischemia, coronary lesions with evidence of low-load ischemia (≥ 120 beats per minute), or ischemia in more than 2 coronary segments (figure 1).

Figure 1.

Flow diagram of the CROSS-AMI trial design. AMI, acute myocardial infarction; IRA, infarct-related artery; PCI, percutaneous coronary intervention. aDue to single-vessel coronary artery disease (n=904), multivessel index PCI (n=82), and suboptimal index PCI (n=30). bInadequate anatomy for PCI in lesions other than the IRA (n=245), significant disease in left truncus arteriosus (n=70), cardiogenic shock (n=73), severe comorbidity (=57), previous revascularization surgery (n=4), and AMI due to coronary stent thrombosis (n=12).

(0.31MB).

We performed a cost minimization analysis comparing the ComR and SelR strategies from the CROSS-AMI trial. The assumption in such analyses is that the strategies being compared are of similar effectiveness. The aim is thus to estimate and compare respective costs.

Index admission costs were estimated from the perspective of the public health care system in the autonomous community where our hospital is based. They were calculated as the sum of costs attributable to the emergency PCI, coronary angiography, repeat PCIs, and any diagnostic cardiology tests performed during the index admission. Follow-up costs were calculated as the sum of costs attributable to hospital stays during readmissions for a cardiac or noncardiac cause in the 12 months following discharge and any diagnostic cardiology tests or additional revascularization procedures performed during admission.

To estimate costs, we analyzed lengths of stay for the index admission and readmissions during follow-up. We also calculated the number of diagnostic and therapeutic cardiology procedures performed during these hospital stays. The costs were derived from official rates established for health care services provided by hospitals belonging to the public health care system in our autonomous community11 (table 1).

We also performed a detailed analysis of disposable material used. The rate system used in our health care system is generic and covers staffing costs, maintenance costs, and average material costs. It does not cover the cost of all the devices used for PCI (eg, sheaths, catheters, guidewires, balloons, stents). As per the trial protocol, a prospective record was made of all material used for PCIs performed during the index admission. The cost of this material, obtained from the purchasing department, was the price paid by the hospital (table 2).

Quantitative variables are expressed as mean±standard deviation (SD), while categorical variables are described as absolute and relative frequencies. Costs are expressed as mean±SD costs per patient. Categorical variables were compared using the chi-square test or, where appropriate, the Fisher exact test, while quantitative variables were compared using the t test or, in the case of nonnormally distributed data, the Mann-Whitney U test. Statistical analyses were performed in SPSS version 24.0 (IBM). A P value of less than .05 was considered statistically significant for 2-tailed comparisons.

The baseline characteristics of the 306 patients randomized to the 2 trial arms are shown in table 3. There were no significant differences between the groups in terms of cardiovascular risk factors, medical history, or clinical presentation of myocardial infarction.

Of the 154 patients in the ComR arm, 152 (99%) underwent PCI for nonculprit lesions (147 required just 1 procedure, while 5 needed an additional 2). One patient requested to be switched to the SelR arm. The stress echocardiogram was negative and the patient was discharged without undergoing PCI. Another patient refused to undergo this procedure (figure 1).

Of the 152 patients in the SelR arm, one was switched to the ComR arm and treated with PCI, while another did not undergo a stress echocardiogram due to a mural thrombus. Two patients developed spontaneous recurrent ischemia before the stress echocardiogram and were treated with PCI as per protocol. Stress echocardiography was thus performed in 148 patients and showed extensive ischemia in 36, all of whom were treated with a second PCI. Of the remaining patients (64 with negative stress echocardiography results, 25 with inconclusive results, and 23 with results suggestive of mild ischemia), 1 underwent a second PCI before discharge because of spontaneous recurrent ischemia and 4 were referred for PCI by their cardiologists (figure 1).

The primary endpoint in the CROSS-AMI trial, a composite measure consisting of cardiovascular death, acute myocardial infarction, coronary revascularization, and readmission due to heart failure, occurred in 22 patients (14%) in the ComR group and 21 (14%) in the SelR group (hazard ratio, 0.95; 95% confidence interval [95%CI], 0.52-1.72; P=.85). There were no differences in risk for any of the components of the composite endpoint (table 1 of the supplementary data).10

Details of lengths of stay and procedures and tests performed during the index admission and readmissions are shown for the ComR and SelR groups in table 4. The estimated costs are provided in table 5.

No significant differences were observed between the ComR and SelR groups for length of stay during the index admission (8.46±10.32 vs 7.28±3.35 days, P=.182). Time to revascularization of nonculprit lesions was shorter in the ComR group than in the SelR group (5.9 vs 9.1 days, P<.001). In the SelR group, mean time from randomization to stress echocardiography was 5.5 days, while mean time from stress echocardiography to PCI was 4.13 days.

As mentioned, 98.7% of patients allocated to the ComR group underwent elective PCI for nonculprit lesions, compared with just 28.9% in the SelR group (P<.001). By contrast, stress echocardiography was performed in 97.4% of patients in SelR group but in just 1.3% of those in the ComR group (P<.001).

Transthoracic echocardiography was performed in 95.5% of ComR patients but in just 17.1% of SelR patients (P<.001).

Although the cost of emergency PCI during the index admission was similar for both groups (€6856.3 for ComR vs €6833 for SelR, P=.319), the cost of elective PCIs was significantly higher in the ComR group (€6856.3 vs €1975.4, P<.001). The cost of stress echocardiography, by contrast, was higher in the SelR group (€377.7 vs €4.9, P<.001).

The total estimated cost for the index admission was €19 657.9 in the ComR group vs €14 038.7 in the SelR group (P<.001). In other words, the SelR strategy yielded an estimated mean savings of €5619.13 (95%CI, €4350.6-€6887.7) per patient.

Readmission rates in the 12 months following discharge were similar in the 2 groups (23.4% for ComR vs 22.4% for SelR, P=.471). The reasons for readmission included cardiac conditions (13.6% vs 14.5%, P=.830) and noncardiac conditions (11.0% vs 10.5%, P=.890). There were no significant differences between the groups in terms of readmission length of stay (2.14 vs 2.58 days, P=.575).

A comparable proportion of patients in both groups underwent noninvasive cardiological tests (eg, transthoracic echocardiography [51.9% vs 49.3%, P=.366] and stress echocardiography [13.6% vs 13.8%, P=.548]) and coronary revascularization (10.4% vs 11.8%, P=.700).

Costs attributable to readmission stays (€1088.8 vs €1329.3, P=.553), additional tests such as echocardiography, stress echocardiography, and coronary angiographies, and new PCIs were also similar in the 2 groups.

The total estimated cost of readmission in the first year of follow-up was €2423.5 for the ComR group and €2653.9 for the SelR group (P=.697). The overall cost of the ComR strategy was therefore significantly higher than that of the SelR strategy (€22 081.3 vs €16 692.6, P<.001).

No significant differences were observed between the ComR and SelR groups for the cost of disposable material used in the index PCI (€1643 vs €1687) (table 6). The cost of elective PCI material, however, was significantly higher in the ComR group (€3166 vs €896, P<.001). The total cost of disposable material was therefore €4810.4 in the ComR group vs €2557.0 in the SelR group (difference of €2253 [95%CI, €1794.2-€2712.2] per patient).

One of the main limitations of this study is the inclusion of just 306 patients (77% of the planned study population) due to the premature interruption of the CROSS-AMI.10 Nonetheless, based on the sensitivity analyses performed and the assumption of a similar rate of events, even if they all occurred in the same group, the likelihood of observing significant differences for the main endpoint with a full sample is remote.10

Selection bias due to exclusion criteria or analysis of baseline characteristics (eg, high frequency of Killip class I) cannot be ruled out. In all, 31% of candidates assessed for inclusion in the trial were finally recruited. While this may have affected the external validity of the trial, this inclusion rate of 31% is similar to rates reported for major clinical trials.13

Stress echocardiography provides more prognostic information than a conventional exercise stress test,24 but its interpretation may be more difficult and depends on the operator. That said, the hospitals that participated in the CROSS-AMI trial have extensive experience with stress echocardiography.25

Our findings might have varied with a longer follow-up period. This is unlikely, however, as no significant differences were observed between the 2 groups after 3 years in an extended follow-up evaluation. Nonetheless, it cannot be completely ruled out that differences would not have been detected after a follow-up period of more than 5 years.

Our use of current health care service rates limits the applicability of our findings to health care systems with comparable cost estimates. It should be noted that autonomous communities in Spain use different rate systems, making comparisons difficult. Another consideration is that potential cost variations in the future would alter our results.

The costs calculated for the follow-up period included cardiology procedures and treatments administered during readmissions, but not primary care visits, emergency room care, or other outpatient tests. Nonetheless, these costs would be unlikely to have a significant bearing as in the CvLPRIT15 trial, medical visits accounted for less than 10% of total costs.15