Evaluating the impact of electronic alerts on guideline implementation of heart failure medical therapies in the Netherlands: the Dutch PROMPT-HF-study
A Uijl, D Collard, S Van Der Zwaard, K Mullin, Y Chen, M Van De Veerdonk, F W AsselbergsAbstract
Introduction
Despite clear guidelines, there is considerable variation in the implementation of guideline-directed medical therapy (GDMT) for heart failure (HF) in clinical practice. Electronic decision support systems, such as alerts in electronic health records (EHR), have been shown to improve guideline adherence and HF medication prescription. The US study PROMPT-HF demonstrated that alerts led to increases in medications or doses in 36% of HF patients.
Purpose
This study aims to evaluate the impact of implementing such an alert in an outpatient cardiology clinic in a tertiary center in the Netherlands to improve GDMT for HF.
Methods
This was a before–after analysis with a 6-month silent phase followed by a 9-month active phase in which the alert was visible to clinicians. When suboptimal HF with reduced ejection fraction (HFrEF) therapy was identified, clinicians received an alert with a guideline-based order set; dismissal of the alert required justification. The primary outcome was a change in GDMT in number of medications or increase in dose within 30 days after the alert. We investigated whether GDMT change differed due to duration of HF (time since recorded HF diagnosis in EHR). Relative risks (RR) with 95% confidence intervals [95%CI] were estimated with generalized linear models (binomial distribution, log link) adjusted for age and sex.
Results
405 HFrEF patients were included in the silent phase (12/2023-05/2024) and 380 HFrEF patients in active phase (with an alert) (05/2024-02/2025). These results are an interim analysis, we are currently processing the recorded alerts of another 750 HF patients. Median age was 71.6 [interquartile range (IQR) 61.7-80.0] years, 37% was female and median HF duration was 3.5 years [IQR 0.8-7.0]. Overall, baseline prescription in participants was 64% beta-blockers, 58% RAS-inhibitors, 45% mineralocorticoid receptor antagonists (MRA) and 29% sodium-glucose cotransporter inhibitors (SGLT-2i). Patients recently diagnosed with HF (<1 year), the number of HF medications increased by 25% without an alert, whereas this increase was 38% with an alert (adjusted RR 1.49 [95%CI 0.99 – 2.25], p-value = 0.057) (Figure 1). For patients with a HF duration >1 year adjusted RR was 1.06 [95%CI 0.71 – 1.60], p-value = 0.75. There was no difference in increase in dose. MRA and SGLT-2i prescription significantly increased with an alert for patients with a HF duration <1 year (Figure 2).
Conclusions
We observed increased GDMT use after implementation of an electronic alert in a tertiary cardiology outpatient clinic. The alert was associated with a trend towards higher GDMT use in patients with HF duration <1 year, but not in those with longer-standing HF. This suggests greater impact of the alert in de novo HF, while treatment intensification based on alerts in chronic HF may be less likely. While clinically relevant, results did not reach statistical significance, likely due to limited sample size.IncreaseGDMTFor image description, please refer to the figure legend and surrounding text.IncreaseTreatmentFor image description, please refer to the figure legend and surrounding text.