Study design and setting
This prospective study was conducted at the Skåne University Hospital in Lund and Malmö, Sweden. We included patients ≥ 40 years old presenting in the ED with a primary diagnosis of AF or AFL and a heart rate ≥ 110 beats/min and who had at least one hsTnT sample analysed. The patients had to be discharged in sinus rhythm and be able to perform a bicycle exercise stress test within 30 days.
We excluded patients with other possible explanations for hsTnT elevation than tachycardia related to AF/AFL. Patients with history of CAD, heart failure (HF), significant anaemia, hypotension (systolic BP < 90 mmHg), hypoxia (saO2 < 90%), renal failure (creatinine > 200 µg/L), acute infection, rhabdomyolysis (myoglobin > 250 µg/L), thyrotoxicosis, acute ischemic stroke or intracranial haemorrhage or acute pulmonary embolism, were excluded. History of CAD was defined as prior percutaneous coronary intervention (PCI), coronary artery bypass grafting (CABG), a > 50 % stenosis on coronary angiography or having had a myocardial infarction. Known HF was defined as documented ejection fraction < 45%, moderate to severe valvular heart disease, hypertrophic cardiomyopathy or dilated cardiomyopathy. Significant anaemia was defined as haemoglobin values < 90 g/L for males and < 80 g/L for females at inclusion. Acute infection was defined as CRP > 50 mg/L and/or temperature > 38 °C. Patients with left bundle branch block (LBBB) on ECG were also excluded because of the difficulty to interpret ST-T changes during stress testing in these patients.
In line with the power calculation (see below), we included 45 patients with elevated hsTnT (cases) and 45 patients with hsTnT values below the 99th percentile (controls). Patients eligible for inclusion were identified and enrolled after ED triage when the corresponding author or research nurses were available (convenience sample), during the time period 14th April, 2015 to 22nd October, 2018.
Data including echocardiography examinations, ECG, laboratory results, clinical and background variables was extracted from medical records according to a predefined protocol and the patients also completed a questionnaire at inclusion.
To differ between acute and chronic hsTnT elevations we analysed a second hsTnT sample after about a week in sinus rhythm. Patients were scheduled for an outpatient bicycle exercise stress test within 30 days after inclusion. If a single-photon emission computed tomography (SPECT) myocardial perfusion imaging already was scheduled or performed before the planed bicycle exercise test, the bicycle exercise test was considered unnecessary and cancelled. Patients with an inconclusive bicycle exercise test were referred to a subsequent standard SPECT myocardial perfusion imaging study [11].
The follow-up period for incidence of a first major adverse cardiac event (MACE) or death was 30 days from inclusion and data on these events were retrieved from medical records.
All patients gave their informed consent. The study was conducted according to the principles of the Declaration of Helsinki and approved by the Regional Ethics Committee in Lund, Lund University (registry number 2014/453).
Primary endpoint
The primary endpoint was a pathological stress test confirmed by a pathological SPECT myocardial perfusion imaging or a coronary angiography depending on clinical indication.
Definitions and troponin T analysis
Hypertension and hyperlipidemia were defined as ongoing medication for each condition. Data on diabetes, current smoking, prior AF or AFL and prior stroke or transient ischemic attack (TIA) were based on both patient questionnaires and medical records. Chest pain was defined as any chest discomfort described as more than mild. Significant ST depression and significant hsTnT dynamic change were defined according to current guidelines [12-14]. During tachycardia a new horizontal or down-sloping ST depression ≥ 1mm in any lead was considered significant and in patients with elevated hsTnT at inclusion a > 20% change compared to follow-up hsTnT analysis after a about a week was considered significant. The secondary endpoint of MACE during 30 days of follow-up was defined as a first ACS, PCI, CABG or death.
The hsTnT method used was Roche (Roche Diagnostics, Basel, Switzerland) high-sensitivity troponin T assay, with a detection limit of 5 ng/L and a 99th percentile cut-off point of 14 ng/L.
Bicycle exercise stress test
The bicycle exercise test was performed on a Monark 939E ergometer bicycle following the protocol used in standard care with gradually increasing work load and subsequent rest [15, 16]. During the entire exercise protocol patients were monitored with a continuous 12-lead ECG, repeated systolic blood pressure measurements and repeated registrations of symptoms (chest pain and dyspnea) according to validated rating scales [17]. The test was considered pathological if there was a planar or down-sloping ST segment depression of > 1 mm, 60 mm after the j point in any lead or if the test had to be interrupted due to severe chest pain, systolic blood pressure drop ≥ 15 mmHg, severe ventricular arrhythmia or significant ST elevations. The test was considered inconclusive if the patients were unable to reach ≥ 85% of age-predicted maximum heart rate, if the patient had significant ventricular arrhythmia (i.e. ≥ 3 consecutive ventricular ectopic beats (VEBs) or increased frequency of VEBs during exercise) or if the ST segment deviation and/or patients symptoms were judged difficult to interpret [13, 18]. The physicians who performed the bicycle stress tests were not informed about the patients’ hsTnT values in the referral texts.
Statistical analysis
A power calculation with an alpha risk of 0.05 and a power of 0.80 was performed. In studies including individuals without suspicion of CAD the prevalence of unknown significant CAD has been reported to be 7% [19] and the incidence of pathological stress test has been reported to be 8% [20]. Based on this we estimated the incidence of pathological stress test to approximately 7% in the control group and we assumed a clinically highly relevant incidence of 30% in the case group. This resulted in a required sample size of at least 45 patients in each group.
Continuous variables are presented as medians with the interquartile range (IQR) and compared with the Mann- Whitney test. Categorical variables are presented as numbers and percentages and compared using the chi-square test or Fischer´s exact test if the expected count was low < 5.
Multivariable logistic regression analyses with elevated hsTnT at inclusion as dependent variable and clinically relevant background factors as covariates were performed. The results are presented as odds ratios (OR) with 95% confidence intervals (CI).
All tests were two tailed. Data management and statistical analysis were performed using IBM SPSS Statistics, version 22.