Associations between inappropriate prescribing and hospitalisation and mortality in adults attending a cardiovascular clinic: logistic regression analysis of the STOP-HF cohort

Abstract Introduction Potentially inappropriate prescribing (PIP) in adults is associated with adverse drug reactions, increased healthcare utilisation, and mortality. PIP occurs when the potential risk of medicines outweighs their potential benefit. Adults with cardiovascular risk factors (e.g. hypertension and hyperlipidaemia) are particularly susceptible to PIP, given the high prevalence of polypharmacy (defined as 5 or more medicines) and multimorbidity in this cohort. STOPP/START (V3.0)1 and Beers Criteria (2023)2 are widely used tools to identify PIP, but their comparative effectiveness in predicting clinical outcomes in patients with cardiovascular risk factors remains unclear. Aim This study aimed to (a) assess PIP prevalence using both STOPP/START (V3.0) and Beers Criteria (2023) and (b) determine their predictive value on hospitalisation and mortality in adults with cardiovascular risk. Methods This retrospective cross-sectional analysis included 1247 participants (mean age 65.4 ± 10.1 years, 48.8% female) from the St Vincent’s Screening TO Prevent Heart Failure (STOP-HF) clinic database. The STOP-HF programme provides structured BNP-guided screening and long-term follow-up of patients with cardiovascular risk factors, offering a well-characterised group for evaluating prescribing safety. PIP was assessed using STOPP/START (V3.0) and Beers Criteria (2023) through rule-based algorithms written in R programming language that incorporated ATC codes, ICD-10 codes, and clinical parameters, and was applied at baseline and follow up. STOPP and Beers were applied to identify potentially inappropriate medications, while START was applied to capture potential prescribing omission. Primary outcomes were all-cause hospitalisation and mortality. Multivariable logistic regression models were used to examine associations between PIP and outcomes, adjusting for age, sex, and Charlson Comorbidity Index. Results A total of 55.1% (n = 687) had at least one instance of PIP according to STOPP, 70.0% (n = 873) according to START, and 66.0% (n = 823) according to Beers. The most common PIP according to STOPP (V3.0) was aspirin for primary prevention (41.5%), while the most frequent prescribing omission according to START (V3.0) was lack of antihypertensive therapy despite elevated blood pressure (17.7%). During follow-up, 392 patients (31.4%) experienced hospitalisation and 80 (6.4%) died. STOPP-defined PIP was consistently associated with increased hospitalisation risk, with moderate exposure (2–3 criteria) conferring 75% higher odds (OR 1.75, 95% CI 1.18–2.60, p = 0.005). Comparison of those with ≥2 STOPP criteria versus 2 STOPP criteria showed further elevated risk (OR 1.57, 95% CI 1.12–2.20, p = 0.008). START omissions were inversely associated with hospitalisation (OR 0.87, 95% CI 0.77–0.97, p = 0.013). Beers Criteria showed no significant association with hospitalisation. None of the PIP tools were independently associated with mortality after adjustment. Conclusion Among the frameworks, only STOPP (V3.0) criteria were independently associated with hospitalisation risk in patients with cardiovascular risk, whereas START and Beers criteria showed no such association. These findings support integrating STOPP assessments into routine medicines optimisation strategies to improve patient safety and prevent hospital admissions in cardiovascular care settings. A key strength of this study was the use of a large, real-world cohort with comprehensive prescribing data, while a main limitation is the reliance on explicit criteria that was coded without full clinician review which may misclassify rates of PIP.