Development of biomarker combinations for postoperative acute kidney injury via Bayesian model selection in a multicenter cohort study

Table 4 Clinical trial enrichment for severe AKI

Threshold	Number to screen to identify 1 eligible patient	Combination/Marker	Severe AKI rate among screen positives (untreated)	Total sample size required
(None)	1	(None)	4.7%	8156
25th percentile	1.3	Maximum posterior probability combination	6.0%	6328
25th percentile	1.3	Median probability combination	6.1%	6166
50th percentile	2	Maximum posterior probability combination	8.0%	4600
50th percentile	2	Median probability combination	8.0%	4600
75th percentile	4	Maximum posterior probability combination	14.2%	2450
75th percentile	4	Median probability combination	15.1%	2286

Abbreviations: AKI = acute kidney injury. We calculated sample sizes for a renopreventive treatment for 90% power, alpha = 0.05, and 30% reduction in AKI risk under treatment. A trial that enrolled only “high risk” patients would require fewer patients in the trial due to a higher event rate, while requiring that more patients be screened in order to identify eligible patients

ISSN: 2050-7771