THINK-HF: development, feasibility, and usability study of an intelligent electronic health record integrated tool for early heart failure diagnosis in primary care
K Barber, C Deaton, G Mccann, L Bernhardt, S Prinjha, I Squire, C Taylor, J Cleland, K Khunti, C LawsonAbstract
Background
Heart failure (HF) is a leading cause of preventable hospital admissions in Europe, yet many patients remain undiagnosed until they present in acute decompensation. THINK-HF is a computerised decision support tool (CDST) designed to identify possible undiagnosed HF in primary care and prompt timely, guideline-recommended assessment.
Purpose
To assess the feasibility and early effects of a digital tool for earlier HF identification in primary care.
Methods
THINK-HF was developed through a structured co-design process combining epidemiological evidence, stakeholder consensus, workflow observation and iterative refinement with clinicians and informatics specialists. Integrated within SystmOne, it analyses routine primary care data (codes, medications, test results and selected free text) to detect missed or emerging HF signals. The main alert triggers when patients have ≥2 multimorbidity indicators plus an HF-suggestive symptom (Figure 1a) and opens a structured template with one-click options for natriuretic peptide (NP) testing, echocardiography, referral and coding (Figure 1b). Three additional algorithms identify unresolved investigations, coding inconsistencies and medication-based signals, generating non-interruptive flags and tailored templates (Figure 1c). A mixed-methods feasibility study across six practices assessed reach, usability, acceptability, early indicators of effectiveness and implementation mechanisms using the RE-AIM framework.
Results
Six socioeconomically and ethnically diverse practices participated (63 GPs; 78,640 patients). At baseline, 876 patients (1.1%) met the main alert criteria and 801 were on the HF register, 91% of whom had coding quality issues (Figure 2a). During four months of testing, THINK-HF generated 299 clinically relevant main trigger alerts (75% during consultations) and 2,805 additional flags. Across practices, early improvements included (i) 31 new HF diagnoses (4.2% relative increase), including higher gains (+10%) in practices with lower baseline prevalence; (ii) a 14% increase in refined HF phenotype coding, (iii) a 7% reduction in unresolved abnormal investigations and (iv) a 14% reduction in loop diuretic prescribing without NP testing, (45% reduction in one practice that used pharmacist-supported review; Figure 2b). Clinicians reported improved awareness, more systematic assessment, and better follow-up of missed investigations. Patients described delays, misattributed symptoms and inconsistent communication, challenges aligned with gaps highlighted by THINK-HF.
Conclusions
THINK-HF is feasible, acceptable and well aligned with real-world primary care workflows. Early gains in diagnostic processes and coding accuracy highlight its promise. Whole-team use, task sharing and ongoing educational support will be essential for scale-up. A larger evaluation is now warranted to determine clinical impact on early diagnosis and outcomes.For image description, please refer to the figure legend and surrounding text.For image description, please refer to the figure legend and surrounding text.