Discussion

In this selectively tested high-risk cohort of 954 patients undergoing noncardiac surgery, myocardial injury was more often attributed to extra-cardiac than to cardiac mechanisms. Extra-cardiac myocardial injury was most commonly attributed to renal failure, trauma, stroke, and sepsis. These patients were younger, underwent more non-elective surgery, and had significantly higher in-hospital mortality than those with cardiac causes.

Within cardiac myocardial injury, type 2 myocardial injury was the predominant type. Owing to the hierarchical classification approach, patients without an identifiable preceding cause were classified in this category by exclusion. No difference in in-hospital mortality was observed between patients with and without a documented trigger. In contrast, type 1 myocardial infarction and acute heart failure were infrequent but associated with relatively high in-hospital mortality. The low incidence of type 1 myocardial infarction is consistent with previous studies indicating that postoperative myocardial infarction is often non-thrombotic.12,17

Similar to the BASEL-PMI investigators, aetiological evaluation at the patient level was performed using a protocolised hierarchical adjudication process.12 This approach acknowledged non-ischaemic mechanisms, consistent with Standardised Endpoints in Perioperative Medicine–Core Outcome Measures for Perioperative and Anaesthetic Care and in contrast to the Vascular events In non-cardiac Surgery patIents cOhort evaluatioN (VISION) and American Heart Association MINS construct.10,18, 19, 20

However, our framework differed in several respects: adjudication was performed by anaesthetists; discrepant classifications were retained deliberately rather than resolved by consensus and inter-rater agreement assessed; renal failure was considered a possible extra-cardiac cause; and an absolute postoperative hs-cTnT threshold >50 ng L−1 was used rather than an increase in troponin concentration of at least one upper limit of normal from before to after surgery. This dichotomous threshold has previously shown a rule-in value for non-operative myocardial infarction and prognostic relevance for postoperative long-term outcomes, including in patients with chronic kidney disease.21, 22, 23, 24, 25 In the sensitivity analysis excluding patients with pre-existing chronic kidney disease, extra-cardiac aetiologies remained predominant and were associated with significantly higher in-hospital mortality, consistent with the primary analysis. In the sensitivity analysis restricted to sampled patients with hs-cTnT >50 ng L−1 at the EPC unit or after vascular surgery, extra-cardiac aetiologies also remained more frequent and in-hospital mortality was 12% in the extra-cardiac group and 8% in the cardiac group, although this difference was not statistically significant (P=0.263) (Supplement 3). The high overall in-hospital mortality observed in this selected high-risk cohort, particularly in the extra-cardiac group, likely reflects the underlying severity of illness, the intrinsic mortality of several extra-cardiac conditions, and the high proportion of non-elective surgery, rather than prognostic discrimination by the adjudicated aetiology.

Management of myocardial injury remains largely empirical. One randomised trial demonstrated a reduction in major vascular complications in patients with MINS receiving 110 mg twice daily dabigatran, without a significant increase in major bleeding.26 More recently, an observational study reported that patients with PMI who were evaluated by a cardiologist had lower 1-yr mortality and fewer subsequent cardiovascular events.27 Although the mechanisms underlying this association remain uncertain, these findings suggest that outcomes after myocardial injury may be modifiable. However, intensification of β-blockers, statins, or antiplatelet therapy based solely on myocardial injury is not supported by robust evidence, and preoperative coronary revascularisation for stable symptoms in elective vascular surgery has not been shown to improve outcomes.28 Similarly, intraoperative haemodynamic strategies aimed at avoiding hypotension and hypertension have not consistently reduced complications despite observational links between hypotension, tachycardia, and MINS.29,30

Limitations

This single-centre study has several limitations. First, postoperative troponin measurements were obtained either by institutional screening practice or at the discretion of the treating clinician. Screening practice reflected local implementation of the 2014 ESC guideline class IIb recommendation applicable during the study period, supported by institutional experience and published work. In addition, we deliberately retained patients with clinical signs warranting troponin measurement, as these may be relevant not to exclude, consistent with previous research.12 Nevertheless, this may have introduced selection bias, and therefore, our findings cannot be generalised to all surgical patients. Second, preoperative troponin concentrations were rarely available, precluding distinction between acute and chronic troponin elevation. Although sensitivity analysis excluding patients with chronic kidney disease partly addressed the potential contribution of chronic troponin elevation, dynamic troponin changes could not be assessed. Third, outcomes were limited to in-hospital mortality because follow-up after discharge was unavailable, which limits comparability with other studies and may underestimate the overall prognostic impact of different aetiologies. Fourth, postoperative hs-cTnT >50 ng L−1 was used as a dichotomous inclusion criterion based on previous work. Because data extraction only included troponin measurements exceeding this threshold, testing-rate distributions could not be described. Fifth, hierarchical adjudication has inherent limitations as it requires assignment of a single dominant cause, although multiple mechanisms may coexist. The framework was applied as closely as possible to a previously published aetiological model, with limited pragmatic adaptations to the available dataset. Therefore, it should be interpreted as a pragmatic descriptive approach, not as validation against, or comparison with, established definitions, assessment of prognostic discrimination, or proposal of a new classification system. Finally, moderate inter-rater agreement suggests limited reproducibility, which may constrain clinical applicability. This may reflect differences in reviewer experience or adherence to the adjudication protocol and co-occurring clinical conditions.