Handling of missing values, stratified by whether prediction was the primary or secondary study aim.

Numbers are column percentages, with absolute numbers in parentheses.aSome studies reported more than one item. Hence, percentages do not add up to 100%.bCross-sectional studies were excluded for this item (item not applicable).cMore than one method could be applied. Hence, the percentages do not add up to 100%. Items were not applicable for two primary-aim studies that had no missing values. Hence, total n = 69.dOnly participants with completely observed data were analysed.eFor example: in a diagnostic study [73], the investigators assumed that among participants who did not undergo follow-up colonoscopy, the detection rates for any adenoma and for an advanced adenoma ranged from half to twice the rates among participants who did undergo follow-up colonoscopy.