Perioperative acute myocardial infarction associated with non-cardiac surgery

Abstract

Aims

Acute myocardial infarction (AMI) is a significant cardiovascular complication following non-cardiac surgery. We sought to evaluate national trends in perioperative AMI, its management, and outcomes.

Methods and results

Patients who underwent non-cardiac surgery from 2005 to 2013 were identified using the United States National Inpatient Sample. Perioperative AMI was evaluated over time. Propensity score matching was used to compile a cohort of AMI patients managed invasively (defined as cardiac catheterization or coronary revascularization) vs. conservatively. The primary outcome was in-hospital all-cause mortality. Among 9 566 277 hospitalizations for major non-cardiac surgery, perioperative AMI occurred in 84 093 (0.88%). Over time, the rate of perioperative AMI per 100 000 surgeries declined by 170 [95% confidence intervals (95% CI) 158–181], from 898 in 2005 to 729 in 2013 (P for trend <0.0001). Perioperative AMI occurred most frequently in patients undergoing vascular (2.0%), transplant (1.6%), and thoracic (1.5%) surgery. In-hospital mortality was higher in patients with perioperative AMI than those without AMI [18.0% vs. 1.5%, P < 0.0001; adjusted odds ratio (OR) 5.76, 95% CI 5.65–5.88]. Mortality associated with perioperative AMI declined over time (adjusted OR 0.86, 95% CI 0.84–0.88). In a propensity-matched cohort of 34 650 patients with perioperative AMI, invasive management was associated with lower mortality than conservative management (8.9% vs. 18.1%, P < 0.001; OR 0.44, 95% CI 0.41–0.47).

Conclusion

In an observational cohort study from the USA, perioperative AMI occurs in 0.9% of patients undergoing major non-cardiac surgery and is strongly associated with in-hospital mortality. Invasive management of such patients may mitigate some of this excess risk, and further research on the management of perioperative AMI is warranted.

Introduction

Acute myocardial infarction (AMI) is a significant perioperative cardiovascular complication for the 300 million patients worldwide undergoing major non-cardiac surgery each year. Over the past decades, efforts to identify patients at risk for perioperative AMI have led to the development of a number of cardiovascular risk prediction models.^1–4 However, despite careful risk stratification, efforts to prevent perioperative AMI have yielded few successes.^5–9 Furthermore, optimal treatment of perioperative AMI is uncertain due to the competing risks of post-operative bleeding and thrombosis. Therefore, current clinical practice guidelines recommend conventional therapies for perioperative AMI, but with ‘modifications imposed by the specific non-cardiac surgical procedure’.^{10  ,  11}

Despite a number of large clinical trials focused on perioperative care, contemporary management and outcomes of perioperative AMI after major non-cardiac surgery are not well defined. The objective of the present study was to evaluate trends in the incidence of perioperative AMI, describe contemporary management, and report outcomes of patients with AMI after non-cardiac surgery in a large administrative database of US hospital admissions.

Methods

Study population

Patients ≥45 years of age undergoing major non-cardiac surgery from 2005 to 2013 were identified from the Healthcare Cost and Utilization Project’s (HCUP) National Inpatient Sample (NIS), as previously described.¹² The NIS is a national administrative database of discharge-level data from a 20% stratified sample of all US hospitals.¹³ Patients were included if they had a principal International Classification of Diseases, Ninth Revision (ICD-9) procedure code for a major therapeutic operating room procedure (HCUP Procedure Class 4) during the hospital admission. Principal Clinical Classifications Software (CCS) procedure codes, aggregates of related ICD-9 procedure codes, were used to stratify cases by surgical subtype. Patients who underwent cardiac procedures (n = 1 465 792), cardiac surgery and cardiac transplantation (n = 522 635), bone marrow transplantation (n = 16 934), ophthalmologic surgery (n = 11 458), radiation therapy (n = 8216), dental surgery (n = 1570), and non-operating room procedures (n = 311) were excluded from the analysis. Major non-cardiac surgery CCS procedure codes were clustered into 13 major surgical subtypes: endocrine, general, genitourinary, gynaecologic, neurosurgery, obstetric, orthopaedic, otolaryngology, skin and breast, thoracic, non-cardiac solid organ transplant, and vascular surgery. Perioperative AMI was identified using ICD-9 diagnosis codes for acute ST-segment elevation myocardial infarction (STEMI) (410.11 to 410.61, 410.81, and 410.91) and non-ST-segment elevation myocardial infarction (NSTEMI) (410.71).

In-hospital management and outcomes

Invasive management of AMI was defined as any in-hospital coronary angiography, percutaneous coronary intervention (PCI), or coronary artery bypass grafting (CABG), as reported by ICD-9 and CCS procedure codes during inpatient hospitalization. Patients who did not undergo an invasive approach to AMI were considered to have been managed conservatively. The primary study outcome was in-hospital all-cause mortality.

Statistical analysis

Continuous variables were reported as mean ± standard deviation (SD) and compared using the Student’s t-test. Categorical variables were reported as percentages and compared by χ² tests. Analyses of proportions over time were performed using the Cochran–Armitage test for trend. Multivariable logistic regression models were generated to estimate odds ratios (OR) adjusted for patient demographics, cardiovascular risk factors, and comorbidities. Models included age, sex, race/ethnicity, obesity, tobacco use, hypertension, hyperlipidaemia, diabetes mellitus, chronic kidney disease, end stage renal disease, coronary artery disease, prior revascularization with either PCI or CABG, peripheral arterial disease, congestive heart failure, valvular heart disease, prior venous thrombo-embolism, chronic lung disease, malignancy, anaemia, alcohol abuse, elective/urgent hospitalization, surgery type, and year of hospitalization as covariates for adjustment.

To evaluate the association between conservative and invasive management strategies for perioperative AMI and in-hospital mortality, propensity score matching was used to generate cohorts of patients with perioperative AMI and similar baseline characteristics who underwent invasive vs. conservative management. Covariates in the model included baseline demographics, cardiovascular risk factors, relevant comorbidities, surgery type, and presentation of AMI. Propensity score matching was performed using a 1:1 matching protocol (without replacement) with a calliper width of 0.2 of the SD of the logit of the propensity score. Absolute standardized differences (ASD) were estimated before and after propensity score matching to assess pre-match imbalance and post-match balance. Absolute standardized differences ≤10% indicate relatively small imbalances in baseline characteristics.¹⁴ To exclude cases in which AMI may have preceded non-cardiac surgery, a sensitivity analysis was also performed excluding patients who were urgently or emergently hospitalized. In order to account for declines in hospital length of stay over time among patients undergoing non-cardiac surgery, rates of AMI per 1000 inpatient hospital days were also determined.

Sampling weights were applied to calculate rates for trend analyses and to determine national incidence estimates according to HCUP guidance.¹⁵ Unweighted data were used in all other analyses, unless otherwise specified. Statistical analyses were performed using SPSS 20 (IBM SPSS Statistics, Armonk, NY, USA) and R (R Foundation for Statistical Computing, Vienna, Austria). Statistical tests are two-sided and P-values <0.05 were considered to be statistically significant.

Patient involvement

Patients were not involved in developing the research question, study outcome measures, study design, or conduct of the study. No patients provided input into the data analysis or interpretation of the results. There are no plans to disseminate the results of the research to study participants. No patients served as authors or contributors to this work.

Results

Study population

A total of 9 566 277 hospitalizations for major non-cardiac surgery satisfied our inclusion criteria. This corresponded to an estimated 45 778 053 surgical hospitalizations in the USA, after applying sampling weights. Acute myocardial infarction occurred in 84 093 patients (0.88%) hospitalized for major non-cardiac surgery, corresponding to an estimated 402 969 perioperative AMI in the USA during this time period, after applying sampling weights. Baseline characteristics of patients with and without perioperative AMI are shown in Table 1. Patients with AMI were older, more likely to be male, and had a greater burden of conventional risk factors for CAD. Among patients with perioperative AMI, STEMI occurred in 17 854 (21.2%) and NSTEMI in 66 239 (78.8%).

Over time, the number of AMI per 100 000 surgeries declined by 170 [95% confidence intervals (95% CI 158–181)], from 898 in 2005 to 729 in 2013 (P for trend <0.0001; adjusted OR 0.91, 95% CI 0.88–0.94), (Figure 1). This trend was largely driven by a decrease in the number of STEMI per 100 000 surgeries, from 261 in 2005 to 116 in 2013 (P for trend <0.001). Trends in AMI by sex are shown in the Supplementary material online, Figure S1. Rates of AMI per 1000 inpatient hospital days also declined over time, from 1.55 in 2005 to 1.42 in 2013 (P < 0.001).

Surgery-specific risks

In unadjusted analyses, vascular (2.0%), non-cardiac transplant (1.6%), and thoracic (1.5%) surgeries were associated with the highest risk of perioperative AMI (Figure 2). In a model adjusted for demographics and clinical covariates to assess the association between surgery type and perioperative AMI, non-cardiac transplant surgery (adjusted OR 1.99; 95% CI 1.81–2.19), thoracic surgery (OR 1.63; 95% CI 1.57–1.69), and vascular surgery (adjusted OR 1.56; 95% CI 1.52–1.59) were independent predictors of perioperative AMI (Table 2). Risks of AMI associated with other types of surgery are outlined in Table 2. Perioperative AMI was more common among patients who were urgently hospitalized for non-cardiac surgery (n = 3 782 565) in comparison to elective hospital admissions (n = 5 783 712) (1.62% vs. 0.40%, P < 0.0001; adjusted OR 2.38, 95% CI 2.34–2.41). Among patients who were urgently hospitalized, AMI was most common in patients undergoing vascular (3.2%), thoracic (3.0%), and non-cardiac transplant surgery (1.7%).

In-hospital management

Among patients with perioperative AMI, 17 511 patients (20.8%) underwent invasive management, the proportion of which increased over time, from 20.2% in 2005 to 23.7% in 2013 (P-value for trend <0.001) (Figure 3). Baseline characteristics of patients with perioperative AMI who were treated invasively vs. conservatively are shown in the Supplementary material online, Table S1. Patients with STEMI were more likely to undergo invasive management than patients with NSTEMI (23.6% vs. 20.1%, P < 0.001).

Among patients managed invasively, 6526 (37.3%) underwent coronary revascularization [PCI in 5376 (82.4%), CABG in 1080 (16.5%), and hybrid CABG with PCI in 70 (1.1%)]. Coronary revascularization was performed in 55.3% of patients with STEMI and 31.5% of patients with NSTEMI. Among patients undergoing PCI, 74.5% underwent stent placement [drug-eluting stents (DES) in 57.3%]. Reperfusion of perioperative STEMI with systemic thrombolysis was reported in 1.8% of cases.

Hospital length of stay was significantly longer for patients with perioperative AMI vs. those who did not develop perioperative AMI {11 days [interquartile range (IQR) 7–18] vs. 3 days (IQR 2–6), P < 0.001}. Length of stay for patients with perioperative AMI decreased over time, from a median of 11 days (IQR 7–19 days) in 2005 to 10 days (IQR 6–16 days) in 2013 (P < 0.0001).

Outcomes of perioperative acute myocardial infarction

Patients with perioperative AMI were more likely to have life-threatening cardiovascular complications, including cardiogenic shock (4.7% vs. 0.1%, P < 0.0001) and cardiac arrest (5.2% vs. 0.3%, P < 0.0001), than patients without AMI. In-hospital mortality was higher in patients with perioperative AMI vs. patients without AMI in unadjusted analysis (18.0% vs. 1.5%, P < 0.0001) and after adjustment for demographic variables and comorbidities (adjusted OR 5.76, 95% CI 5.65–5.88). Over time, in-hospital mortality associated with perioperative AMI declined from 20.1% in 2005 to 15.5% in 2013 (P for trend <0.0001), as shown in Figure 4.

Patients with perioperative AMI who underwent invasive management had lower in-hospital mortality than those who were managed conservatively (8.9% vs. 20.5%, P < 0.001; OR 0.38, 95% CI 0.36–0.40), despite higher rates of post-operative haemorrhage associated with invasive management of AMI (8.1% vs. 5.3%, P < 0.001). Patients who underwent revascularization also had lower mortality than patients managed conservatively (10.5% vs. 18.7%, P < 0.001; OR 0.51, 95% CI 0.47–0.55). In-hospital mortality was 9.0% among patients who underwent CABG for perioperative AMI (n = 1150) and 10.8% for patients who underwent PCI for perioperative AMI (n = 5376). Mortality was 8.0% among patients who underwent invasive management but did not undergo coronary revascularization (n = 10 985).

In a propensity-matched cohort of 34 650 patients with perioperative AMI (Supplementary material online, Table S1), invasive management was associated with lower in-hospital mortality than conservative management (8.9% vs. 18.1%, P < 0.001; OR 0.44, 95% CI 0.41–0.47). Similar associations between invasive management and reduced mortality were observed in propensity-matched subgroups of patients with STEMI and NSTEMI (Table 3). In a sensitivity analysis of patients electively admitted for non-cardiac surgery, an invasive management strategy was also associated with lower in-hospital mortality than conservative management after propensity matching (7.3% vs. 14.8%, P < 0.0001; OR 0.45, 95% CI 0.40–0.51).

Discussion

In a large national cohort of adults ≥45 years of age, perioperative AMI occurred in approximately 0.9% of hospitalizations for non-cardiac surgery. Patients with AMI were older and had a greater burden of cardiovascular risk factors. In adjusted analyses, patients undergoing non-cardiac vascular, thoracic, and transplant surgery had the highest risks of perioperative AMI.

Trends in acute myocardial infarction

The rate of perioperative AMI declined from 2005 to 2013, driven by a marked decline in the rate of perioperative STEMI. Declines in AMI, and STEMI in particular, may be attributed to careful pre-operative risk stratification and improved pharmacologic and percutaneous strategies in the management of coronary artery disease, including administration of appropriate antiplatelet therapy, statin use, pre-operative smoking cessation, and appropriate revascularization of high-risk patients prior to surgical clearance. Furthermore, there has been increasing recognition of thrombotic risks in the first year after PCI, and providers nationwide may have incorporated longer delays to non-cardiac surgery into routine practice.^16–18 Finally, increasing use of minimally invasive surgical approaches may also contribute to declines in perioperative STEMI. In contrast, no reductions in rates of perioperative NSTEMI were observed during the study timeframe. Trends in the rates of diagnosis of NSTEMI may reflect increases in the sensitivity of modern cardiac biomarkers necessary for the diagnosis of AMI.¹⁹

Outcomes of perioperative myocardial infarction

Patients with perioperative AMI had longer hospital length of stay, more life-threatening cardiovascular complications such as shock and cardiac arrest, and greater in-hospital mortality than patients without AMI. Prolonged hospital length of stay associated with perioperative AMI, while expected, is important given the potential for additional in-hospital complications, such as hospital-acquired infections, venous thrombo-embolism, and physical deconditioning. Mortality associated with perioperative AMI in this analysis is consistent with prior studies of patients undergoing non-cardiac surgery. A single-centre retrospective observational study of 281 patients who underwent PCI for perioperative AMI reported a 30-day mortality of 11%²⁰; in the corresponding subset of 5376 patients from this analysis, in-hospital mortality was 10.8%. Mortality associated with AMI over time declined for patients with NSTEMI, but not STEMI, in the present study. Although adherence to guideline-directed medical therapy and increases in rates of invasive management of AMI may have contributed to this encouraging trend, declines in AMI mortality may also be related to greater numbers of low-risk NSTEMI identified in the later years of the analysis period. Despite declines in perioperative AMI mortality, these data still highlight the importance of additional efforts to reduce morbidity and mortality associated with perioperative AMI after non-cardiac surgery.

Invasive management of AMI following major non-cardiac surgery is a challenging clinical decision due to the competing risks of bleeding and infarction. Antiplatelet therapies and anticoagulants necessary for cardiac catheterization and PCI may be contraindicated in the perioperative period, when surgical site bleeding complications are potentially devastating (e.g. following neurosurgery). Conservative management of perioperative AMI may exclude antiplatelet therapy entirely and potentially predispose patients to a high burden of ischaemic complications. In this analysis, only one fifth of patients with AMI were referred for invasive management, highlighting barriers to coronary revascularization in the perioperative setting for patients with STEMI and NSTEMI. The risk of bleeding is likely a major factor in provider reluctance to refer patients with perioperative AMI for invasive angiography. Patients referred for invasive management of AMI had lower mortality than those who were conservatively managed in unadjusted and propensity-matched cohorts. However, bleeding may be an important unmeasured confounder that contributes to the observed association between invasive therapy and reduced in-hospital mortality.

Study limitations

There are a number of limitations of this analysis. First, the present data are derived from diagnosis codes recorded in a large administrative database and are subject to reporting bias and/or errors in coding. Furthermore, the diagnosis of perioperative AMI may be under-reported or missed entirely as the practice of routine biomarker collection after non-cardiac surgery is uncommon. In large perioperative trials with standardized screening for AMI, 65% of patients with perioperative AMI did not experience ischaemic symptoms.²¹ Providers in certain surgical subspecialties, such as vascular and thoracic surgery, may be more aware of potential perioperative cardiovascular complications than providers in other specialties. Therefore, failure to diagnose perioperative AMI may impact the reported AMI rates. Results of perioperative laboratory testing, including cardiac biomarkers, were not available from this administrative dataset. Second, the present analysis was limited to adults age ≥45, and the inclusion of large numbers of relatively low-risk patients from the NIS dataset may contribute to a low incidence of perioperative AMI observed in this dataset. In comparison, prior clinical trials of patients undergoing non-cardiac surgery have typically enrolled older patients with at least ≥1 cardiovascular risk factor, a population enriched for cardiovascular events.⁵ Third, measures of left ventricular function and diagnostic coronary angiographic findings were not recorded in this administrative dataset. Among patients with prior coronary revascularization, the timing of non-cardiac surgery relative to PCI was not available for this analysis. However, in prior studies, patients requiring non-cardiac surgery early after DES PCI had increased risk of perioperative MI and cardiac death.²² Similarly, details of the in-hospital medical management were not available for review, and use of antiplatelet agents, statins, beta-blockers, and other cardiovascular therapies could not be determined. In recent observational studies, aspirin and statin therapy have been associated with improved short and long-term outcomes following perioperative myocardial injury and AMI.^{21  ,  23} Although multiple relevant clinical covariates were included to adjust for the assigned treatment strategy in propensity-score models in this analysis, unmeasured confounders such as medical therapy may account for the lower mortality associated with invasive management. Fourth, the sequence of non-cardiac surgery preceding AMI cannot be definitively established for patients who underwent urgent hospitalization. However, major non-cardiac surgery is contraindicated early after AMI, and patients presenting with AMI were highly unlikely to undergo non-cardiac surgery during the index hospital admission. Fifth, data were analysed as a simple random sample for many of the analyses (i.e. without using weights). This may introduce empirical weighting and thus may not be reflective of the overall population. Finally, long-term outcomes after perioperative AMI were not available from this in-hospital dataset. However, data from prior studies suggest that perioperative myocardial injury and AMI confer significant long-term risks in addition to increased short-term mortality.²⁴

Conclusions

To our knowledge, this is the largest analysis of perioperative AMI in hospitalized patients undergoing non-cardiac surgery in the USA. Despite declines in the incidence of perioperative AMI and improved outcomes over time, AMI remains strongly associated with mortality. Invasive management of such patients may mitigate some of this excess risk, and further research on the management of perioperative AMI is warranted.

Supplementary material

Supplementary material is available at European Heart Journal online.

Conflict of interest: none declared.

References

Author notes