Positive publication bias in cancer screening trials