Feasibility and outcomes of an automated high-alert NT-proBNP pathway

Background

NT-proBNP is a powerful biomarker for risk stratification in heart failure (HF). Patients with markedly elevated values are at particularly high risk of adverse outcomes but may experience delays in access to specialist care. Evidence on the impact of an automated, high-threshold NT-proBNP referral pathway embedded in electronic patient records (EPR) is limited.

Purpose

To evaluate the feasibility, diagnostic yield, and clinical impact of an EPR-linked high alert NT-proBNP pathway for patients with NT-proBNP >5000 pg/ml, and to compare outcomes with usual care on the island prior to implementation.

Methods

All NT-proBNP results >5000 pg/ml generated over a 12-month period triggered an automatic referral to the HF team. Referrals were categorised as inpatients, outpatients already known to the HF service, outpatients not suitable for follow up, or new outpatients eligible for enrolment. Eligible outpatients were contacted and offered echocardiography and clinical review within 48 hours. Clinical data, echocardiographic findings, treatments, and outcomes were collected prospectively.

A retrospective comparator cohort included all outpatient NT-proBNP results >5000pg/ml from patients not known to the HF service in the 12 months before implementation.

Results

Over 12 months, 889 referrals were generated: 417 (47%) inpatients, 370 (42%) outpatients already known to the HF team, 59 (7%) not appropriate for follow up, and 43 (5%) enrolled in the pathway. Median time to echocardiography and review was 2 days, with 70% meeting the 48-hour target. The enrolled cohort had mean age 81 years, 56% male, mean NT-proBNP 9586 pg/ml. Echocardiography identified HFrEF in 22 (51%), HFmrEF in 6 (14%), HFpEF in 10 (23%), no HF in 1, and 4 patients were not scanned (3 died before attendance, 1 attended privately). New severe aortic stenosis in 4 (10%) and new atrial fibrillation in 4 (10%) were found. Following enrolment, 91% had medication changes at the initial consultation. For those with HFrEF, 78% of patients were on maximally tolerated guide-line directed therapy at 4 weeks, rising to 89% at 12 weeks (Figure 1). Mean NYHA class improved from 2.5 to 1.6 at 12 weeks, and mean LVEF in patients with HFrEF increased from 27% to 45% in those with HFrEF. Twelve-week mortality was 8% vs 12% in the retrospective cohort (n=60).

Conclusion

An automated high-threshold NT-proBNP pathway is feasible, identifies patients at very high risk, and enables rapid optimisation of therapy. Substantial improvements in symptoms and LV function were observed, alongside detection of other clinically important diagnoses. This model highlights the potential for NT-proBNP-driven pathways to improve outcomes in advanced HF but requires sufficient echocardiography and HF service capacity.Figure 1Figure 2