508 Scoping the opportunity for pharmacogenomic testing for older people’s medicine optimisation in secondary care

Abstract

Abstract Introduction Pharmacogenomics (PGx) is the use of DNA variation to predict response to medication. This can help prescribers select medications, optimise dosage, and prevent adverse drug reactions (ADRs). Patients aged 65 or over are most likely to benefit because they are more susceptible to ADRs and many use multiple medications daily (polypharmacy). Aim We aimed to identify the prevalence of medications with PGx associations at planned or unplanned admission to hospital. Our study tested if the use of medicines with PGx associations can improve the prediction of length of hospital stay, unplanned admission, and readmission. Methods A retrospective cross-sectional study considered 59,973 hospital admissions of people aged 65 and over to a large NHS Trust over two years (2018-2019). We calculated the prevalence of 560,163 medications with PGx associations. We estimated the predictive performance of PGx medicines by comparing multivariable regression models with and without PGx medicines counts. The link between adverse outcomes and the use of PGx medicines and frailty status was also explored. Results Polypharmacy was high, with 83% (n=49,683) of patients on admission using ≥5 medications and 43% (n=25,832) using ≥10medications. Over 1 in 5 medications used by patients had known PGx associations at the point of admission to hospital and 84% (χ2(2, N=59973) = 93.459, p<0.001) of patients with unplanned admissions were using at least one medication with PGx association compared to 64% with a planned admission using one or more PGx medicines (χ2(2, N=59973) = 770.021, p<0.001). Lansoprazole, aspirin, and simvastatin were the most-prescribed medications with PGx association. High risk medicines implicated in hospital admissions from the cardiovascular, pain and psychiatric therapeutic areas, were amongst the ten most common medicines with PGx associations in the dataset (1). Multivariable prediction models that controlled for covariates and included the number of PGx medicines patients improved the prediction of length of stay in hospital, unplanned admissions, and unplanned readmissions. Predictions for patients with high frailty were also improved by including the number of PGx medicines. Conclusion The study shows that PGx medicines were more likely to be used by older patients with unplanned admissions than planned admissions. Predictive models that included the number of PGx medicines compared to models that omitted PGx medicines gave a better prediction of the length of stay in hospital, unplanned hospital admissions and readmission when covariates were controlled for. Although the study does not demonstrate that patients were at risk due to their genetics, a Europe-wide multi-centre trial, reported a 30% reduction in clinically significant adverse drug reactions when patients received PGx-guided care (2). Therefore, the results reported in this abstract, adds support to the case for development of PGx testing to improve patient outcomes, and indicates that older patients particularly at risk of ADRs could benefit from optimising medicines through PGx-guided switching or dose changes.