Sixteen studies met the inclusion criteria: five RCTs (n=22,709, range 1,219 to 10,993) and 11 diagnostic cohort studies (n=49,840, range 135 to 20,322). Four of the cohort studies were in screening (average risk) populations, five were in people known to have or suspected of having colorectal disease (at risk) and two were in both populations. The RCTs were considered good quality in terms of randomisation, allocation concealment and comparability at baseline. All the diagnostic cohort studies were considered good quality in terms of the patient spectrum, reference standard, incorporation bias and reporting of withdrawals. Nine studies were prone to partial and differential verification bias.
RCTs: The detection rate of advanced colorectal neoplasm was greater with iFOBTs than with gFOBTs, but the result did not reach statistical significance (OR 1.50, 95% CI 0.94 to 2.39); statistically significant heterogeneity was observed (I2=70.7%).
Cohort studies: When data from screening and at-risk populations were combined, iFOBT detected significantly more advanced colorectal neoplasm than gFOBTs (OR 1.77, 95% CI 1.31 to 2.39); statistically significant heterogeneity was observed (I2=75.6%). This significant superiority of iFOBT over gFOBT was observed in screening (OR 2.90, 95% CI 2.50 to 3.36) and at-risk (OR 1.27, 95% CI 1.01 to 1.60) populations; no statistically significant heterogeneity was observed for either analysis. Across all diagnostic cohort studies the positive predictive value ranged from 2% to 50%.
In patients scheduled for colonoscopy, sensitivity for gFOBTs ranged from 29% to 91% and specificity ranged from 59% to 93%. iFOBT sensitivity ranged from 42% to 86% and specificity from 62% to 98%.
Results were reported for a range of subgroup analyses, which included the impact of dietary restrictions. Data were limited for most subgroup analyses.
Funnel plots showed some asymmetry, but the Egger regression test detected no significant asymmetry.