“…10,19,22,30,34,36,39 The articles generally used cancer registry data as a data source (9 [69%]) 10,22,30–35,37,38 ; as a result, the sample sizes were large (>1500 patients). Six studies (46%) 10,33,36–39 included all head and neck subsites, while 7 studies (54%) were subsite specific (5 oral cavity, 1 oropharynx, and 1 larynx). 19,22,30–32,34,35 Four of the studies (31%) 22,31,33,36 did not define the rationale or methods for their definition of prolonged DTI, 4 (31%) 19,35,37,38 used calendar-based categorical definitions (eg, <30 days, 31–60 days), 3 (23%) 30,34,39 used cohort-based quartiles or medians, and 2 (15%) 10,32 used recursive partition analysis to determine an optimal DTI threshold.…”