Age-stratified utility of NT-proBNP testing as part of a blind Emergency Department shortness of breath orderset (Jan-Dec 2019)

ABSTRACT

Background: NT-proBNP was added to our emergency department (ED) triage blind 'shortness of breath (SOB) orderset' for presenters aged >70. Evidence-based thresholds for acute heart failure (HF) diagnosis are >900/1800 pg/ml for ages 50-75/>75 respectively (1.2);their utility in contemporary practice is uncertain. Purpose: To assess the relation between blind NT-proBNP testing in this setting and (1) coded discharge diagnosis stratified by age, and (2) all-cause mortality at medium-term follow-up. Methods:We retrieved all ED 'SOB' blood ordersets (1.1.2019-31.12.2019), including NT-proBNP, Hb, electrolytes, creatinine, troponin, CRP, d-dimer, and coded discharge diagnoses. Multivariate logistic regression models for all-cause survival (at 9.9.2021) were assessed. Results: There were 638 presentations (median age 76.1), unexpectedly including 198 <70 years. Modal and median lengths of stay were 0 and 1 day respectively. Stratified by age (<60, 60-69, 70-74, 75-79, 80-84, ≥85y), the proportion with HF coded as primary discharge diagnosis (5, 7, 9, 17, 18, 25% respectively) and all-comer all-cause mortality at 2.2±0.3 years (13, 42, 40, 48, 48, 49%) steadily increased (Table;orderset variables presented as median (inter-quartile range)). Median NT-proBNP was 3672, 2667, and 321 pg/ml when HF was in the primary, secondary, or neither coded discharge diagnosis field respectively;2.2-year-all-cause mortality was 54%, 60%, and 35%. In those with a primary HF discharge code, 77% of 349 presenters ≥75y and 88% of 231 aged 50-74 had NT-proBNP >1800/900 pg/ml respectively. In those without an HF code, 26% in both age cohorts had NT-proBNP >1800/900 pg/ml (dotted lines in Figure, panels A/B, respectively represent NT-proBNP thresholds). Independent predictors of all-cause mortality for patients with a primary or secondary HF code were ln(NT-proBNP) (OR 1.26, 95% CI 1-1.59) and serum Na+ (OR 0.93, 0.88-0.99);for patients without an HF code, these were serum K+ (1.87, 1.21-2.88), ln(NT-proBNP) (1.35, 1.15-1.58), ln(CRP) (1.18, 1.02-1.36), length of stay (1.08, 1.03-1.12), and age (1.03, 1.01-1.06). (Figure Presented) Conclusion: HF detection with NT-proBNP in a blind SOB orderset showed increasing sensitivity with age with the best specificity >75 years. Most presenters stayed ≤1 day, so blind testing at triage facilitates HF detection. NT-proBNP independently predicted 2.2-year-all-cause mortality irrespective of discharge HF coding. This is notable as the commonest non-HF causes of acute SOB are prognostically important at >70 years and follow-up occurred through the Covid-19 pandemic. The findings may reflect disease severity in patients without HF, but also suggest that discharge HF coding status does not identify all those with prognostically relevant HF.