Standardizing the standard: reporting health status in clinical trials

In ∼20 BCE, Marcus Vipsanius Agrippa, a powerful Roman general and Augustus Caesar's right-hand man, made decree that the ‘foot’ would be precisely 11.65 in., or length of Agrippa's foot.1 Prior to this, foot was determined based on each individual's own body part, which can vary by several inches depending their height. The importance standardizing cannot understated, as this paved way for Romans' achievements in engineering construction, including aqueducts, roadways, monuments. This system measurement brought order randomness. It did not really matter what actual was; just it precise, reproducible, accepted standard. Although Kansas City Cardiomyopathy Questionnaire (KCCQ) reliably validly assess health status patients with heart failure, current reporting measures clinical trials is similar era before Agrippa. Stogios et al.,2 perspective piece, call need standardize these measures. We could agree more. KCCQ continuous measure scores ranging from 0 100. As any measure, interpretation outcomes, particularly average more challenging than discrete events such hospitalizations deaths. When compare means mean changes scores, significance differences may clear, large studies higher statistical power. For example, few individual patients, if any, actually change same amount population. More often, some improve lot, little, do change, others get worse. Without proportions different magnitudes difficult interpret benefits therapy patients' status.3 Of course, applies dichotomous outcomes death. If trial shows 2% difference deaths between groups, obviously does every patient worse group dead dead. Moreover, also judge whether certain percentage reduction death meaningful, especially there are risks treatment, requires judgement. developed 1996 published 2000,4 prognostic association 25-point ranges first shown. A special study designed support interpretability 2005.5 Despite attempt offer standardized approach interpreting thresholds has been wide variation observational studies.2 To approach, we recently state-of-the-art review specifically topic,3 where outlined how cross-sectionally both context; report scores; best trials; account missing data due guidance offered over past two decades, trialists often develop approaches data. rapid growth use (Figure 1) demands consistent reporting. believe factors contribute confusion results trials. cohort-specific describe participants' ‘starting point’. Using quartiles terciles make strong epidemiological sense, equalizes sample groups maximizes power, but undermines ability one another. challenges readers determine ill participants were, critical step application future patients. upper quartile total symptom score among outpatients GALACTIC-HF 91–100,6 indicating virtually no failure symptoms, whereas only 4% had overall summary ≥75 PARTNER 1B trial,7 suggesting included an exceedingly cohort Examining standard facilitate better understanding populations' illness severity responses thereby enhancing its practice. logic examining relative baseline. supporting changes, they challenge endpoint is, related regulatory suggestions needed scores.8 Accordingly, re-establishes clinically important populations, despite abundant evidence 5- (and 10- 20-) point threshold clinically5, 9, 10 prognostically important.11, 12 While other have sought KCCQ,13, 14 yet see estimate whose confidence intervals include 5-, 10-, 20-point estimates originally provided. clinicians understand those impairs translate into al.2 natural consequence inconsistent analysis cross-sectional interpretable Certainly, research define minor standards, goal communicate information providers, refinements improvements consequential? should continue refine our integrate well explore categorize enrolled at floor ceiling scores. However, most pressing consistency analysing results. Standardization, regard, alleviate many concerns colleagues providing framework plainly unambiguously treatments impact status, making interpretable. then readily practice target therapies who likely benefit. Conflict interest: none declared.