Fig. 4: A targeted phenome-wide scan comparing the effect of changes in cardiac function and structure to those of a heart failure or atrial fibrillation diagnosis.

Nota bene: p-values passing the 0.05 threshold are indicated by an open diamond, with stars indicating results passing a threshold of 2.6 × 10⁻⁴. Cells were coloured by effect direction times -log₁₀(p-value); where p-values were truncated at 8 for display purposes. CMR traits were selected based on the Kolmogorov-Smirnov test for multiplicity, with effects estimated using the weighted median estimator. The following abbreviations were used, LV left-ventricle, RV right-ventricle, LA right-atrial, EF ejection fraction, SV stroke volume, EDV/ESV diastolic or systolic volumes, EDM end diastolic mass, MVR mass to volume ratio V (max): maximum volume, AF atrial fibrillation, T2DM type 2 diabetes, CKD chronic kidney disease, VTE venous thromboembolism, AAA abdominal aortic aneurysm, SBP/DBP systolic/diastolic blood pressure, BMI: body mass index, CRP c-reactive protein, FVC forced vital capacity, FEV1 forced expiratory volume, PEF peak expiratory flow, eGFR estimated glomerular filtration rate, HbA1c glycated haemoglobin, CE cardioembolic, LA large artery, IS: ischaemic, SV small vessel, AD Alzheimer’s disease. See Data 3–6 for the underlying data.