2.2 Eligibility for HFrEF trials 59 trial. Patient weights were also applied and these were calculated as a residual difference from a non-linear Gaussian kernel-based hypersurface plane. To determine the criterion most likely to impact eligibility, eligibility scores were inversely weighted by the frequency of occurrence in trials where the lowest weighted scores would be seen in the most restrictive criteria. Missing data in the registries ranged from 1% to 54% and were multiply imputed by chained equations.25 The number of imputations was set at 10.26 Statistical analysis was performed using R statistical software 4.1.2 (R Core Team, 2021), STATA SE 15 (StataCorp LP, College Station, Tx) and MATLAB R2021a. Statistical significance was set at 0.05. Eligibility in theoretical trials Lastly, we sought to determine how eligibility changes by each addition of commonly used eligibility criteria. Starting with a broad set of criteria including (i) age between 18 and 80 years, (ii) LVEF ≤ 40%, (iii) NYHA classes II, III and IV, (iv) double background therapy of any-dose ACEI/ ARB + BB, (v) no MI /PCI /CABG, (vi) no device therapy, (vii) no cancer/ COPD, (viii) no stroke (viii) renal function (eGFR >30ml/min/1.73m2), (ix) haemoglobin >10g/dL, (x) potassium <5.5 mmol/L. Alternative scenarios with more restrictive selection including (i) an LVEF of 35%, (ii) NYHA classes II & III, (iii) enrichment with previous hospitalization for HF and (iv) triple therapy (includes MRA) was also considered to determine the impact of stricter cut-offs impact on eligibility. RESULTS Characteristics of heart failure phase III trials As of end December 2021, 4425 studies for heart failure were identified on ClinicalTrials.gov and 375 were phase III HF trials. Of these, 163 (44%) were HFrEF trials, 9% were HFpEF trials, 30% were non-selective for LVEF and a remaining 17% enrolled hospitalized HF patients (Table 1). Within a 37-year observation period, the number of phase III trials registered per decade was increasing whereby more than half (55%) were initiated within the recent 10 years. The size of trials was also increasing with time, specifically from 2005 onwards (p<0.001). By subtype, the largest trial size was in HFpEF trials (median 336) vs overall median trial size of 170
RkJQdWJsaXNoZXIy MjY0ODMw