Table 2.

 Definitions of criteria for evaluating strength of evidence among studies quantifying overdiagnosis from cancer screening

Criterion	Definition
Risk of bias (high/moderate/low)	See table 1
Directness (good/fair/poor)	Extent to which the evidence links screening directly to differences in long term cumulative incidence between populations without making assumptions
Analysis (good/fair/poor)	Extent to which the analysis appropriately quantifies overdiagnosis, without inclusion of age groups or time frames that lack the potential to be overdiagnosed, and without statistically adjusting for lead time
Time frame (good/fair/poor)	Extent to which the time frame is sufficient to account for the effects of lead time
External validity (good/fair/poor)	Extent to which study population is similar to US general population or Western European populations in factors that are associated with cancer incidence, screening situation (such as expertise of screening radiographers, quality of screening facilities, threshold for defining a result as abnormal), medical care, and risks for competing mortality
Precision (good/fair/poor/cannot determine)	Confidence interval of difference in cumulative incidence attributable to screening between populations over an appropriate time frame should be provided. Width of confidence interval should be narrow, not wider than about 20%.
Consistency (good/fair/poor)	Extent to which the overdiagnosis measurements from the included studies have a similar magnitude