Variability in Reassessment of Left Ventricular Ejection Fraction After Myocardial Infarction in the Acute Myocardial Infarction Quality Assurance Canada Study

Key Points

Question  What proportion of patients with left ventricular dysfunction identified during hospitalization for acute myocardial infarction have follow-up assessment of left ventricular ejection fraction (LVEF)?

Findings  In this cohort study of 501 patients, 303 patients had undergone LVEF reassessment by 6 months, with significant variability according to patient-level and site-level characteristics.

Meaning  These findings suggest that programs to improve the quality of post–myocardial infarction care should include measures to ensure that indicated repeat cardiac imaging is performed.

Importance  Persistently depressed left ventricular ejection fraction (LVEF) after myocardial infarction (MI) is associated with adverse prognosis and directs the use of evidence-based treatments to prevent sudden cardiac death and/or progressive heart failure.

Objective  To assess adherence with guideline-recommended LVEF reassessment and to study the evolution of LVEF over 6 months of follow-up.

Design, Setting, and Participants  This was a multicenter cohort study at Canadian academic and community hospitals with on-site cardiac catheterization services. Patients with type 1 acute MI and LVEF less than or equal to 45% during the index hospitalization were enrolled between January 2018 and August 2019 and were followed-up for 6 months. Data analysis was performed from May 2020 to September 2021.

Exposures  Baseline clinical factors, in-hospital care and LVEF, and site-specific features.

Main Outcomes and Measures  The main outcomes were receipt of repeat LVEF assessment by 6 months and the presence of a persistent LVEF reduction at 2 thresholds: LVEF less than or equal to 40%, prompting consideration of additional medical therapy for heart failure, or LVEF less than or equal to 35%, prompting referral for implanted cardioverter defibrillator in addition to medical therapy.

Results  This study included 501 patients (mean [SD] age, 63.3 [13.0] years; 113 women [22.6%]). Overall, 370 patients (73.4%) presented with STEMI, and 454 (90.6%) had in-hospital revascularization. The median (IQR) baseline LVEF was 40% (34%-43%). Of 458 patients (91.4%) who completed the 6-month follow-up, 303 (66.2%; 95% CI, 61.7%-70.5%) had LVEF reassessment, with a range of 46.7% to 90.0% across sites (χ²₁₃ = 19.6; P = .11). Participants from community hospitals were more likely than those from academic hospitals to undergo LVEF reassessment (73.6% vs 63.2%; χ²₁ = 4.50; P = .03), as were those with worse LVEF at baseline. Follow-up LVEF improved by an absolute median (IQR) of 8% (3%-15%). However, 103 patients (34.1%) met the definitions of clinically relevant LVEF reduction, including 52 patients (17.2%) with LVEF less than or equal to 35% and 51 patients (16.9%) with LVEF of 35.1% to 40.0%.

Conclusions and Relevance  In this cohort study, approximately 1 in 3 patients with at least mild LVEF reduction after acute MI did not undergo indicated LVEF reassessment within 6 months, suggesting that programs to improve the quality of post-MI care should include measures to ensure that indicated repeat cardiac imaging is performed. In those with follow-up imaging, clinically relevant persistent LVEF reduction was identified in more than one-third of patients.

Numerous studies have identified persistently reduced left ventricular ejection fraction (LVEF) as the dominant factor associated with the risk of sudden cardiac death (SCD) after acute myocardial infarction (MI).^1-6 Practice guidelines recommend routine reassessment of LVEF beginning after 3 months (or after 40 days if no revascularization procedure was performed) if the in-hospital LVEF is less than or equal to 40% (US and Europe) or less than or equal to 45% (Canada).^7-9 These guidelines further recommend specific clinical actions linked to the degree of convalescent LVEF reduction: medical therapy to treat or prevent heart failure when it remains less than or equal to 40%, and further referral for consideration of an implanted cardioverter defibrillator (ICD) when the LVEF remains less than or equal to 35%. In the observational phase of a previous single-center study,¹⁰ we found that less than one-half of patients with in-hospital LVEF less than or equal to 45% had LVEF reassessment within 12 months, which improved significantly with implementation of a directed follow-up process. US-based cohorts have also documented lower than expected rates of LVEF reassessment in similar patient populations.^11,12 This is an important gap in care because, despite guideline-directed medical therapy and revascularization, up to 50% of patients with impaired LVEF after MI do not experience normalization of their ventricular function.^13,14

Previous studies on this topic have been retrospective and lacked patient-reported data on the potential reasons for lack of follow-up testing. We therefore designed the Acute Myocardial Infarction Quality Assurance (AMIQA) Canada study as a multicenter study to quantify this care gap and better understand its associated factors in the Canadian context.

This cohort study was performed at 14 hospitals in Canada. The protocol was approved by each site’s research ethics board, and all participants provided written informed consent. This report adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for cohort studies.

We had 2 predefined principal hypotheses. The first hypothesis is that less than 80% of MI survivors with an in-hospital LVEF less than or equal to 45% would have a guideline-recommended reassessment of LVEF within 6 months. The 80% target represents a conservative estimate of the upper limit of uptake. The second hypothesis is that when repeat imaging is performed in these patients, more than 20% will have a clinically relevant reduction in LVEF, defined as either LVEF less than or equal to 40%, prompting consideration of additional medical therapy to prevent or treat heart failure (including mineralocorticoid receptor antagonists, sacubitril and valsartan, and more recently sodium glucose cotransporter 2 inhibitors), or LVEF less than or equal to 35%, prompting referral for ICD in addition to medical therapy. We also sought to identify independent patient-level and site-level factors that were associated with failure of indicated LVEF reassessment.

Adult patients admitted with a primary diagnosis of spontaneous type 1 acute MI according to the third universal definition of MI¹⁵ that required cardiac catheterization during the index hospitalization were screened for eligibility. We included patients admitted with either ST elevation MI (STEMI) or non-ST elevation MI (NSTEMI) who survived to hospital discharge and had at least 1 recorded LVEF less than or equal to 45%. When multiple in-hospital LVEF tests were performed, patients were included if any 1 measure met the study criteria. Key exclusion criteria included secondary causes of MI (types 2, 4, or 5 according to the third universal definition¹⁵), preexisting ICD or development of an indication for an ICD during hospitalization, and noncardiac comorbidities limiting life expectancy to less than 1 year, as determined by the investigator. These criteria helped ensure that a spontaneous acute MI was the primary diagnosis and that the patient’s clinical situation and preferences mandated care directed at reducing early and late mortality from MI. All sites were in urban or suburban locations with full on-site cardiac catheterization services. We categorized sites as academic if they were the primary teaching hospital of an affiliated medical school (10 sites) and as community hospitals otherwise (4 sites).

Eligible and consenting patients were enrolled during their index hospitalization or within 14 days of discharge. Patients were given a pamphlet explaining the relationship between MI and SCD and the role of LVEF in prognosis, but no other study-specific prompts encouraging LVEF reassessment were provided to the patient or care team. Baseline data collection included demographic and clinical information, details of the presentation, in-hospital management and events, cardiovascular medications prescribed at discharge, and the specialty of the physician designated to provide post-MI follow-up care.

We recorded all in-hospital LVEF values using quantitative estimates from standardized measures when available.¹⁶ A standardized algorithm was used to convert qualitative LVEF descriptors to quantitative values (eAppendix in Supplement 1). When more than 1 baseline LVEF measurement was available, we used the latest dated value that was less than or equal to 45% as the qualifying value, categorized as follows: less than 30%, 30% to 35%, 35.1% to 40%, and 40.1% to 45%.

Follow-up consisted of a single telephone visit and a web-based survey 6 months after enrollment. We determined patient-reported clinical status, medication use, and whether follow-up LVEF assessment had occurred. Follow-up was supplemented by review of medical records and contact with the most responsible physician as necessary to obtain records of cardiac imaging.

On the basis of our previous study,¹⁰ we anticipated a 20% missing rate for clinical follow-up and that overall 50% of patients would have a follow-up LVEF value available. Using these assumptions, we estimated that 500 enrolled patients would provide 90% power to assess our first primary hypothesis, and greater than 80% power for the second primary hypothesis (eAppendix in Supplement 1).

Baseline and follow-up characteristics were summarized as proportions with associated 95% CIs, mean (SD), or median (IQR) as appropriate. The 2 principal hypotheses were tested with a 1-sample test of proportions, using a 2-sided P < .05 as the threshold, without adjustment for multiple testing. In a post hoc analysis, we used a generalized linear latent and mixed model procedure with a binomial distribution, logit link, and robust SEs to assess patient-level and site-level variables independently associated with LVEF reassessment.¹⁷ In this hierarchical model, patients were nested within sites. Patient-level variables include in the model were age, sex, prior cardiac events and risk factors, comorbidities, details of clinical presentation, and baseline LVEF category. Site-specific or physician-specific variables included in the model were academic vs community site, province, and physician type (cardiologist vs other) responsible for follow-up post-MI care. Because the study purpose was to describe current practice, imputation of missing LVEF data was not performed. We used Stata SE statistical software version 15 (StataCorp) for analysis. Data analysis was performed from May 2020 to September 2021.

We enrolled 511 patients at 14 participating sites between January 2018 and August 2019. The 4 community sites enrolled 141 (28.1%) patients. The eFigure in Supplement 1 shows the cohort flow. After excluding 10 patients who were found to be ineligible or withdrew before baseline data collection, the final cohort included 501 patients (mean [SD] age, 63.3 [13.0] years; 113 women [22.6%]). Table 1 summarizes baseline patient characteristics. The index MI was a STEMI in 370 patients (73.4%), and 454 patients (90.6%) had in-hospital coronary revascularization.

In-hospital LVEF measures are summarized in eTable 1 in Supplement 1. The qualifying LVEF was from echocardiography for 397 patients (79.2%), and the median (IQR) qualifying LVEF value was 40% (34%-43%). Multiple in-hospital LVEF measures were recorded for 264 patients (52.7%).

A total of 441 patients (88.0%) completed follow-up interviews. Information on vital status and follow-up cardiac imaging was obtained from medical record review and/or physician contact for an additional 17 patients, resulting in 458 patients (91.4%) available for analysis. Overall, 303 patients (66.2%; 95% CI, 61.7%-70.5%; P < .001 for first primary hypothesis that the proportion is <80%) had at least 1 LVEF reassessment during follow-up, a median (IQR) of 102 (7-138) days after enrollment. In 94 revascularized patients (21%), the follow-up LVEF was performed earlier than 90 days after MI. The LVEF value was not available for 1 patient. There were no significant differences in the proportion undergoing LVEF reassessment according to sex or age (dichotomized at 75 years). However, the proportion undergoing reassessment was higher in those with more severe LV systolic dysfunction at baseline, ranging from 85.7% in those with LVEF less than 30% to 56.9% in those with LVEF greater than 40% (Figure 1). The proportion undergoing LVEF reassessment ranged from 46.7% to 90.0% across study sites (χ²₁₃ = 19.6; P = .11), but there were no significant interprovincial differences noted or any associations noted with the specialty of the physician providing follow-up care. However, patients discharged from community vs academic hospitals were significantly more likely to undergo follow-up LVEF assessment (73.6% vs 63.2%; χ²₁ = 4.50; P = .03).

Table 2 shows the results of generalized linear latent and mixed model modeling to identify patient-level and site-level variables associated with LVEF reassessment within 6 months. In the full model, discharge from an academic site, a history of hypertension, and less severe baseline LVEF reduction were each independently associated with a lower odds of receiving follow-up LVEF assessment. Calculated intraclass correlation within sites was very low in this model (intraclass correlation coefficient, <0.01). Univariable associations between LVEF reassessment and other evidence-based post-MI follow-up processes of care are shown in eTable 2 in Supplement 1. Those who had follow-up cardiac imaging were also more likely to have had a post-MI follow-up visit with their most responsible physician and to have enrolled in cardiac rehabilitation. However, self-reported use of most evidence-based medication classes was similar in the 2 groups.

A total of 381 patients (76.0%) returned the electronic follow-up survey, 115 of whom reported not receiving a follow-up LVEF test. eTable 3 in Supplement 1 shows the distribution of patient-reported reasons for not undergoing follow-up LVEF. The most common response was that the physician had not ordered a repeat test (42%). Barriers related to accessing care or testing facilities totaled 18%.

Among the 302 patients who had paired baseline and follow-up LVEF values, 264 (87.1%) had 1 measure, 33 (10.9%) had 2 measures, and 6 (2.0%) had 3 follow-up measures. The qualifying follow-up LVEF was from an echocardiogram for 274 patients (90.4%). Most patients experienced an improvement in LVEF (absolute median [IQR], 8% [3%-15%]), although it declined in 40 patients. Figure 2 shows changes in qualifying LVEF from baseline to follow-up according to baseline LVEF. Overall, 103 patients (34.1%; 95% CI, 29.0%-40.0%; P < .001 for the second primary hypothesis that the proportion is >20%) met 1 of our prespecified definitions for clinically relevant reduction in follow-up LVEF, including 52 (17.2%) with LVEF less than or equal to 35%, indicating eligibility for consideration of ICD therapy, and an additional 51 (16.9%) with LVEF of 35.1% to 40%. As expected, there was significant association between baseline LVEF category and actionable follow-up LVEF status (Figure 2B). Among the 36 patients whose baseline LVEF was less than 30%, 24 (66.7%) had follow-up LVEF less than or equal to 35%. Conversely, among the 103 patients whose baseline LVEF was 40.1% to 45%, only 11 (10.7%) had a follow-up LVEF less than or equal to 40% and only 2 had follow-up LVEF less than or equal to 35%.

A total of 86 interval clinical cardiac events (9 recurrent MI, 11 percutaneous coronary intervention, 7 coronary artery bypass grafting, 17 cardiac hospitalizations, and 42 emergency department visits for a cardiac cause) occurred during follow-up. Figure 3 is an alluvial diagram summarizing the overall cohort progression according to the baseline and follow-up LVEF values and subsequent referral to an electrophysiology specialist regarding LVEF, whether an ICD was recommended, and whether they had an ICD implanted or scheduled. Only 12 of 48 patients (25%) with a follow-up LVEF less than or equal to 35% reported referral to an electrophysiologist, among whom 10 (83.3%) had an ICD recommended and 6 (50%) had an ICD implanted or scheduled.

In the AMIQA Canada cohort of contemporarily managed acute MI survivors with in-hospital LVEF less than or equal to 45%, we found that only two-thirds had LVEF reassessment within the recommended time period. Furthermore, when it was measured, a clinically relevant reduction in LVEF was identified in 34.1% of the cohort, including 17.2% with LVEF less than or equal to 35%. In this latter group, only one-quarter of patients were referred for consideration of ICD implantation. These findings are in keeping with our a priori hypotheses and identify gaps in post-MI follow-up care for SCD prevention.

We identified several patient-level and system-level sources of variability in LVEF reassessment. Importantly, we observed a gradient in rate of follow-up testing according to the baseline LVEF category, declining from 85.7% in those with in-hospital LVEF less than 30% to 56.9% in those with in-hospital LVEF of 40.1% to 45%. Baseline LVEF remained associated with receipt of follow-up testing after multivariable adjustment. It is reassuring that the large majority of those with the most severe in-hospital reductions in LVEF received timely repeat testing. These Canadian results in fact compare favorably with those reported by the National Cardiovascular Data Registry Acute Coronary Treatment and Intervention Outcomes Network Registry–Get With The Guidelines in a study of US Medicare-insured post-MI patients with in-hospital LVEF less than or equal to 35%, of whom 66.8% had LVEF reassessment within 1 year.¹²

Clinician and health system factors also appeared to be associated with LVEF reassessment. Patients reported that the most common reason for not having LVEF reassessment was that it was not ordered by the physician, without an apparent association with physician specialty. At the health system level, the proportion undergoing LVEF reassessment varied widely (range, 46.7%-90.0%) across study sites. Because health care in Canada is organized provincially, we wondered whether access to follow-up testing would vary by province but found no such association. Patients also reported that issues related to accessing testing were uncommon reasons for a lack of LVEF reassessment. This combination of findings suggests that the local practice conditions and culture may play a larger role than issues of access to services on the rate of LVEF reassessment. The finding that treatment at a community vs academic hospital was independently associated with LVEF reassessment was unexpected and could represent a chance finding. The 4 community sites participating in AMIQA are urban hospitals that provide comprehensive cardiac care, and so their performance may not be representative of all Canadian community hospitals. Previous studies examining MI outcomes and processes of care have found that teaching vs nonteaching hospital status is associated with increased adherence to guideline-recommended therapy in hospital and at discharge, which in some cases translated into better outcomes.^18-21 However, none of these studies investigated the quality of follow-up care, and it is possible that coordination of follow-up care and testing is easier in community as opposed to academic hospitals. Exploring this issue is part of our plan for ongoing research.

Our study’s modest sample size and lack of clinician engagement do not permit us to fully assess the barriers to LVEF reassessment. However, the patient survey data suggest that a failure to order a follow-up test is an important factor, whereas a lack of test availability was an uncommonly reported reason. LVEF reassessment was associated with a higher likelihood of patient-reported follow-up with their most responsible physician, which may provide a partial explanation. Furthermore, the observed gradient in follow-up testing according to the baseline LVEF suggests that physicians may discount moderate, in-hospital left ventricular dysfunction when considering when to order repeat testing.

Although the LVEF improved in most patients in whom it was measured, more than one-third had a persistent and potentially clinically relevant reduction in LVEF at follow-up. This includes 17.2% who met guideline indications for consideration of primary prevention ICD therapy. By the time of follow-up, only one-quarter (12 of 48) of these patients had been referred for an ICD, and only 1 in 8 (6 of 48) had an ICD implanted or scheduled. This experience is consistent with previous Canadian and international studies that have documented low rates of ICD referral in patients meeting eligibility criteria.^22-25

As expected, those with lower initial LVEF were most likely to have more severe reductions in follow-up LVEF. Interestingly, when the in-hospital qualifying LVEF was 40.1% to 45%, only 11 of 103 (10.7%) had a follow-up LVEF less than or equal to 40%, and in only 2 patients was the LVEF less than or equal to 35%. Therefore, a policy of routinely ordering follow-up LVEF in those with in-hospital LVEF less than or equal to 40%, while being more selective in those with lesser degrees of LV dysfunction could provide a good balance between underuse and overuse of follow-up cardiac imaging.

Improving the rates of LVEF reassessment in appropriate patients to identify those at risk for SCD is one of several components of post-MI follow-up that can lead to improved outcomes. Adherence to evidence-based medical therapy and provision of cardiac rehabilitation are also critical. In this study, patients reported strong use of relevant medications for secondary prevention 6 months after MI, but much lower attendance at cardiac rehabilitation. Increasing patient knowledge and engagement in their care is a potential method to improve these metrics. The Interventions Supporting Long-term Adherence And Decreasing Cardiovascular Events After MI randomized trial²⁶ used mailed reminders with or without telephone calls encourage post-MI patients to enroll in cardiac rehabilitation and to adhere with medications and demonstrated a modest improvement in attendance at rehabilitation compared with usual care. Another promising approach is to develop automated processes to ensure care delivery. A recent observational study²⁷ from Poland reported that enrollment in a post-MI managed care program with directed follow-up, echocardiography, and referral for ICD if indicated was associated with 3-fold increases in cardiac rehabilitation attendance and ICD implantation, and significant reductions in mortality.

Enrolled patients were aware of the study’s purpose, which could have introduced a bias in the proportion of patients undergoing LVEF reassessment. If present, this effect was modest as the proportions were in line with our previous retrospective study.¹⁰ Next, we could not determine the indication for follow-up cardiac imaging. Also, despite the use of broad enrollment criteria, we enrolled fewer female participants than expected. Next, although guidelines recommend that LVEF be reassessed beginning 3 months after MI (or after 40 days if no revascularization was performed), the 6-month follow-up duration was shorter than in some previous studies on this topic. It is possible that a larger proportion of patients would have had follow-up imaging and subsequent referral for ICD consideration over 12 months. In addition, the study’s primary objectives were to describe current practice patterns, and the associations observed in our post hoc multivariable analyses should be seen as hypothesis generating.

In this multicenter cohort study, we found that 1 in 3 patients with LV dysfunction after acute MI did not undergo indicated LVEF reassessment within 6 months. In those with follow-up imaging, clinically relevant persistent LVEF reduction was identified in more than one-third of patients. This report demonstrates the persistence of important gaps in efforts to reduce the risk sudden death after MI in Canada.

Accepted for Publication: September 30, 2021.

Published: December 2, 2021. doi:10.1001/jamanetworkopen.2021.36830

Open Access: This is an open access article distributed under the terms of the CC-BY-NC-ND License. © 2021 Wilton SB et al. JAMA Network Open.

Corresponding Author: Stephen B. Wilton, MD, MSc, Libin Cardiovascular Institute, Cumming School of Medicine, University of Calgary, GE64 3280 Hospital Dr NW, Calgary, AB T2N 4Z6, Canada (sbwilton@ucalgary.ca).

Author Contributions: Dr Wilton had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Wilton, Bennett, Parkash, Kavanagh, Frisbee, MacLachlan, Manlucu.

Acquisition, analysis, or interpretation of data: Wilton, Bennett, Parkash, Kavanagh, Jolicoeur, Halperin, Jolly, Leong-Sit, Sas, Chew, Singh, Frisbee, Manlucu.

Drafting of the manuscript: Wilton, Frisbee, MacLachlan.

Critical revision of the manuscript for important intellectual content: Wilton, Bennett, Parkash, Kavanagh, Jolicoeur, Halperin, Jolly, Leong-Sit, Sas, Chew, Singh, Frisbee, Manlucu.

Statistical analysis: Wilton, Singh, Frisbee.

Obtained funding: Wilton, Bennett, Manlucu.

Administrative, technical, or material support: Wilton, Bennett, Kavanagh, Jolicoeur, Jolly, Frisbee, Manlucu.

Supervision: Wilton, Bennett, Parkash, Halperin, Jolly, Frisbee.

Conflict of Interest Disclosures: Dr Wilton reported receiving research grants from Medtronic Canada, Abbott, and Boston Scientific and consultancy fees from Arca Biopharma. Dr Bennett reported participating in research collaboration with Medtronic Canada during the conduct of the study. Dr Parkash reported receiving research grants from Medtronic Canada, Abbott Canada, and Novartis Canada. Dr Kavanagh reported receiving grants from Cardiac Arrhythmia Network of Canada (CANet), Medtronic, Abbott, and Boston Scientific during the conduct of the study. Dr Jolicoeur reported receiving research grants from Jubilant Radiopharma, Abbott Cardiovascular, and Neovasc. Dr Manlucu reported receiving consultancy fees and speaker honoraria from Medtronic and Baylis Medical and grants from Medtronic, Boston Scientific, and Abbott during the conduct of the study. No other disclosures were reported.

Funding/Support: This project was funded by a CANet strategic research grant (SRG-17-P28-001 to Dr Wilton; a portion was dispersed to Dr Frisbee to fund her portion of the study), with unrestricted matching funds provided by Medtronic Canada, Boston Scientific, Abbott, and the Libin Cardiovascular Institute, University of Calgary.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Group Information: The Acute Myocardial Infarction Quality Assurance (AMIQA) Canada Investigators are listed in Supplement 2.