Article Text
PDF
Abstract
Introduction Troponin elevation in patients with sepsis is associated with increased in-hospital mortality in small studies of patients requiring intensive care, and more recently in patients suffering COVID-19. Previous meta-analyses have found that a raised troponin during sepsis is associated with increased mortality, but the studies were typically small (median 40 patients leading to meta-analyses of around 1000 patients), studied only those with the most severe infections, and often excluded those with comorbidities, particularly coronary artery disease, which may be responsible for different causes of troponin elevation. Previous studies have also usually followed up patients to the point of discharge. We determined (i) whether an association between troponin elevation and mortality was seen in a large dataset without exclusion of comorbidities (ii) whether the association was seen in those with and without a diagnosis of coronary disease and (iii) whether troponin elevation was associated with mortality even in those who survived their initial illness.
Methods A retrospective cohort study was carried out using the National Institute for Health Research Health Informatics Collaborative Cardiovascular dataset of all consecutive patients who had a troponin measured at five hospitals between 2010 and 2017. We identified patients with an ICD-code indicating infection who had a troponin and c-reactive protein (CRP) measured. We compared the survival of those with a positive troponin with those with a negative troponin during and after hospital discharge. To assess the association of dichotomous troponin level (above upper limit of normal or not) with all-cause mortality, Cox proportional hazards regression modelling was used. A landmark analysis was performed in those who survived their hospital admission.
Results Of 11,132 patients studied, 4,554 (42.4%) had a positive troponin. During median follow-up of 213 days there were 4,448 (40.0%) deaths. A positive troponin was associated with a 2.41-fold increase in mortality (95% confidence interval (CI) 2.27 to 2.56) and was similar in patients known to have coronary disease (hazard ratio (HR) 2.15, 95% CI 1.87 to 2.46) and those without (HR 2.39, 95% CI 2.24 to 2.56).
The association between troponin positivity and mortality was preserved in sepsis and different CRP strata (Figure 1). There was a trend towards higher levels of troponin being associated with higher mortality (Figure 2A). A positive troponin continued to be associated with mortality in those surviving past discharge (HR 2.14, 95% CI 1.99 to 2.29), as shown in Figure 2B.
Conclusion In a large cohort of hospitalised patients suffering an infection, a raised troponin was associated with significantly increase mortality risk, regardless of known coronary status or severity of infection. A positive troponin during admission with infection continued to be associated with an ongoing increased risk of mortality, even in those surviving admission.
Preserved association of troponin and mortality in sepsis and different CRP strata
(A) Association between the extent of troponin elevation and risk of morality; (B) Ongoing impact of positive troponin on survival in those surviving to discharge from hospital. Restricted cubic spline modelling of the association between troponin level and hazard ratio using 4 unforced knots. ULN, upper limit of normal. Positive troponin (orange); Negative troponin (blue)
Conflict of Interest No conflicts of interest