Discussion
In a large cohort of 10 315 consecutive patients with suspected acute coronary syndrome, we evaluated the effectiveness and safety of implementing an early rule-out pathway using an hs-cTnT concentration of <5 ng/L to risk stratify patients at presentation. Implementation of this approach was associated with a reduction in the duration of hospital stay and an increase in the proportion of patients discharged directly from the emergency department. This was achieved without any change in all-cause or cardiovascular deaths at 1 year, suggesting that the use of this approach to risk stratify patients with suspected acute coronary syndrome is both safe and effective in routine clinical practice.
There are several strengths to our study. First, our before and after study, unlike an individual patient-level randomised controlled trial, enables inclusion of consecutive patients regardless of age, sex, symptoms, time of presentation or comorbidities. Using such an approach limits selection bias and improves the generalisability of our findings. Implementation at the hospital level minimised the risk of a Hawthorn effect from researchers observing care, which may exaggerate the effectiveness of implementing early rule-out pathways. Second, we included a large population of 10 315 patients in the study with over 1000 deaths. We are not aware of any other prospective study of this size evaluating the implementation of an early rule-out pathway using hs-cTnT. Third, given the robust national databases available in Scotland, which have already been used to deliver longitudinal cohort studies and randomised controlled trials, we were able to evaluate the short-term and long-term impact of implementing an early rule-out pathway with all-cause mortality as our primary safety outcome and complete follow-up.22 24 25 This approach limits misclassification bias and we further presented cause-specific mortality evaluating cardiovascular deaths.
Our analysis showed that the implementation of an early rule-out pathway led to substantial reductions in the duration of stay in hospital. This observation was likely due to both a change in the threshold for ruling out myocardial infarction and reducing the timing between serial sampling. However, not all patients with an initial troponin concentration <5 ng/L on presentation were discharged according to the pathway. This may be due to ongoing symptoms, early presentations (<3 hours), or other medical and social reasons requiring hospitalisation. Our findings extend the current literature. Using a high-sensitivity troponin I assay, the High-Sensitivity cardiac Troponin On presentation to Rule out myocardial InfarCtion trial demonstrated that implementation of an early rule-out pathway reduced duration of stay by 3.3 hours with no increase in 1-year mortality using a stepped-wedge cluster randomised design in 31 492 patients.26 In contrast, the Limit of Detection and ECG Discharge trial showed that implementing an early rule-out pathway using the limit of detection of a range of hs-cTn assays when combined with a normal ECG increased 4-hour discharge rates, but did not significantly reduce duration of stay and the trial was not powered to evaluate safety outcomes.16 The Rapid Assessment of Possible acute coronary syndrome In the Emergency Department with high-sensitivity TnT trial compared a 1-hour pathway that incorporates the limit of detection with a 3-hour rule-out pathway using the 99th centile in 3378 patients, and reported that the 1-hour strategy reduced duration of stay by 60 min and increased discharge rates from 32% to 45%.27 The trial concluded non-inferiority for a composite primary endpoint that included index events, but there were only 6 deaths at 30 days. However, follow-up at 1 year demonstrated an increase in both subsequent myocardial infarction and death in patients with low troponin concentrations identified by the hs-cTnT assay and randomised to the 1-hour strategy compared with the 3-hour pathway using the 99th centile.28 The explanation for these unexpected findings is not clear, but they highlight the importance of conducting adequately powered studies that evaluate both the effectiveness and safety of implementing early rule-out pathways. Furthermore, one other before and after study has evaluated the implementation of an early rule out pathway with hs-cTnT demonstrating a reduction in hospital admissions with no excess in adverse events.29 However, this study evaluated the HEART score and a change in hs-cTnT of <3 ng/L to identify low-risk patients.
In our study, implementation of an early rule-out pathway using hs-cTnT reduced duration of stay and was not associated with any change in all-cause or cardiovascular mortality at 1 year. The biggest reduction was seen in the 5–14 ng/L group which likely reflects the change in timing of serial sample from 6 or 12 hours to 3 hours. There was no evidence of harm in models adjusting for age, sex and comorbidity. Studies that have evaluated early rule-out pathways have primarily looked at short-term outcomes with small number of fatal or non-fatal events16 or have evaluated hs-cTnI assays.30 We assessed safety using both short-term and long-term outcomes. Across the study cohort, we had over 1000 deaths and showed, with a significant degree of confidence, that those patients managed with an early rule-out pathway were not at higher risk overall or when stratified according to hs-cTnT concentration at presentation.
Our pathway was based on the High-STEACS early rule-out pathway20 31 32 developed using an hs-cTnI assay. Our study therefore has important clinical implications providing further confidence in early rule-out pathways using a hs-cTnT assay. The findings from our study add to a recently published systematic review of 37 studies evaluating the diagnostic performance of two-step hs-cTn pathways.33 This evidence review informed the National Institute for Health and Care Excellence guideline, which recommended pathways using hs-cTn concentrations near the limit of detection at presentation as a first step with a second step of small absolute changes to safely rule out myocardial infarction.34 This concept is also recommended by the European Society of Cardiology guidelines using a rule-out threshold of <5 ng/L and a change of <3 ng/L in 0/1-hour or 0/2-hour pathways based on these principles.35 Our findings should provide clinicians with additional confidence when adopting early rule-out pathways into clinical practice.
Study limitations
First, this was a single-centre study in a large secondary care hospital in Scotland, and the impact of adopting this approach may differ in tertiary referral centres or other healthcare settings. Second, the controlled before and after study design is quasi-experimental and less robust than a randomised trial. It is possible that secular trends in practice outside of the pathway could have impacted on duration of stay. However, our stratified analysis demonstrated that the observed reductions in duration of stay were confined to those patients where care would have been modified by the pathway. Nevertheless, we are unable to adjust for unmeasured confounding including changes in clinical practice during the study period which may have contributed to differing clinical characteristics in the standard care and intervention groups. Third, all clinical decisions were made by the attending clinician and therefore not all patients identified as low risk may have been discharged according to our protocol. This may have introduced bias, however our findings are an accurate reflection of clinical practice where judgements are made according to many factors and care pathways are not always followed. Fourth, we did not adjudicate the diagnosis of myocardial infarction or evaluate diagnostic performance. For clinical outcomes, we used routinely collected data. Although this may introduce misclassification, given the same approach was used during both the standard care and implementation phases of our study it is unlikely to have introduced bias. Fifth, while most patients who had an admission cardiac troponin requested would be in the context of suspected acute coronary syndrome, we are unable to exclude patients in whom troponin was requested for other clinical conditions such as pulmonary embolus or acute heart failure. Finally, we used the hs-cTnT thresholds recommended by the manufacturer for use in the UK and the rest of the world, but we are aware the limit of detection varies across different platforms and that the 99th centile recommended for use by the US Food and Drug Administration (USFDA) is higher. Furthermore, the USFDA does not permit the reporting of concentration below 6 ng/L. While the threshold of 5 ng/L used here is likely to have similar effectiveness and safety as using other Roche platforms, we cannot directly inform the safety of implementing this approach using the USFDA thresholds.