CHAPTER 2.2 58 Data sources for target population A target population or domain refers to all patients to whom trial findings can be applied whereas a trial population is a smaller subgroup within the target population. Target population data were available from two registries: the BIOlogy Study to TAilored Treatment in Chronic Heart Failure (BIOSTAT-CHF) and Asian heart failure registry (ASIAN-HF).22,23 The former consists European HF patients while the latter enrolled patients from 10 Asian countries. Both HF registries included physiciandiagnosed HF patients with a majority of patients having HFrEF. Only patients with LVEF less than 40% were included from the BIOSTAT cohort to maintain comparability with ASIAN-HF. Registry variables were screened and the following common variables across registries were used for the estimation of eligibility scores: age, anaemia, atrial fibrillation (AF), body mass index (BMI), cancer, chronic kidney disease, chronic obstructive pulmonary disease (COPD), serum creatinine, device therapy, diastolic and systolic blood pressure, estimated glomerular filtration rate (eGFR) by the CKDEPI equation, haemoglobin, heart rate, LVEF, history of myocardial infarction, serum potassium, QRS duration, revascularization by percutaneous coronary intervention (PCI) or coronary artery bypass graft (CABG), stroke, sinus rhythm, use of ACEinhibitors or angiotensin-2 receptor blockers (ARB), beta-blockers (BB), mineralocorticoid receptor antagonist (MRA), history of worsening heart failure (HF hospitalization in past 6 months for ASIAN-HF or 12 months in BIOSTAT-CHF). Because of substantial missing rates, natriuretic peptides were not analysed. Valve disease was not evaluated due to insufficient depth on severity and most trials exclude only the severe forms. Estimating eligibility in existing trials We estimated overall and single-criterion eligibility based on the generalizability index for study trait, GIST 2.0 introduced by Sen et al.21,24 The score represents an estimated proportion of the target population that is trial-eligible and have values between 0 and 1. This representativeness score is first calculated by treating each criterion independently, be it the presence or absence of patient characteristic(s) or fulfilment of defined thresholds in numeric measurements such as laboratory tests. Then, an overall weighted representativeness score is estimated based on the proportion of registry patients who fulfil all criteria. The estimation method standardises numeric data and accounts for interdependence across criteria in each
RkJQdWJsaXNoZXIy MjY0ODMw