Eight studies were included in the review. Four studies had a prospective paired design (one of which used random allocation of order of tests). One study was a randomised controlled trial and three were cohort studies. Follow-up was one to two years in study designs, except for cohort studies where only detected cases were followed up.
The overall pooled estimate for differences in detection rate between full-field digital mammography and screen-film mammography was four (95% CI -3 to 11) extra breast cancers per 10,000 screening mammograms with some evidence of heterogeneity (I2=40%). Stratification of analyses by age showed higher full-field digital mammography detection rates for women aged up to 60 years with 11 (95% CI 4 to 18) extra breast cancers per 10,000 screening mammograms with no evidence of heterogeneity (I2=0%). Data on women aged over 60 was sparse, but one study found that while full-field digital mammography had higher detection rates among women under 50 with dense breasts, screen-film mammography had slightly higher rates in women aged over 64 with non-dense breasts.
There was too high a level of heterogeneity (I2=94%) in recall rates for data to be pooled; USA studies showed much higher levels than European or Japanese studies. There was also a high level of heterogeneity (I2=100%) for the positive predictive values of an abnormal mammogram; only cohort studies showed a higher positive predictive value for full-field digital mammography. Therefore, these outcomes were not combined statistically.
There was no evidence of publication bias.