Reporting the decision-attributable failure rate alongside the pooled failure rate in rule-out diagnostic-management...

Rule-out diagnostic-management studies ask whether a clinical decision rule combined with a biomarker can safely spare patients an imaging test - for example, the YEARS algorithm and D-dimer versus computed tomographic pulmonary angiography (CTPA) for suspected pulmonary embolism. Safety is summarized by the failure rate among all patients in whom the condition was excluded at baseline. That pooled rate spans two groups: patients ruled out by the decision rule without imaging, and patients ruled out by a negative reference test. We consider what each group tells readers, and how both should be reported.

In randomized strategy trials the pooled rate is the appropriate measure for comparing assigned strategies, because randomization is at the level of the strategy and imaging itself alters subsequent management. The decision-attributable failure rate - the failure rate among patients managed without imaging - answers a different and complementary question: what a clinician should expect when imaging is withheld. Across four prospective YEARS studies it is reported separately in the single-arm management studies but not in the most recent randomized trial. In single-arm management studies, where there is no comparator strategy, it is the measure that should carry the safety claim and drive sample size. Both rates are shaped by disease prevalence, thrombotic burden, and outcome misclassification, and neither should be read alone.

Doctors use scoring rules and blood tests to decide when a scan for a blood clot in the lung can safely be skipped. Studies of these rules usually report one overall number: how often a clot was missed among everyone told they did not have one. That single number mixes two different groups - people who skipped the scan and people who had a normal scan. We argue that studies should also report, separately, how often a clot was missed among the people who skipped the scan, because that is the group the rule actually affects. Both numbers matter, and both need to be read alongside how common clots are in the population studied.