There are differences in the output measures used by the included cost-effectiveness studies. Eight studies are using life-years gained as the measure of effectiveness {10, 12, 13, 16-19, 25}, six are using Quality adjusted life years (QALY) {14, 15, 20-24} and one is using the intermediate outcome “detected advanced tumours” {11}. Since there are different study designs (type of model, perspective, time horizon, population, etc.) and input parameters (sensitivity, specificity, compliance rate, etc.) the results of the models are not necessarily comparable.

Life-years gained when using FIT instead of gFOBT:

• In general the results of the cost-effectiveness modelling show that using FIT instead of gFOBT increases the life-years per person. The incremental effects range from 0.041 life-years gained per person over a lifetime if annual screening is performed from age 65 – 79 {10}, to 0.003 life-years-gained after 10 years if only one round of screening is performed in a cohort of 50-75 year-olds {19}.
• There are two different types of guaiac-based tests, which are frequently used as comparators in cost-effectiveness studies: Hemoccult II and Hemoccult SENSA. The effect on life-years gained of the latter, Heoccult SENSA, is estimated to be similar to that of the immunochemical tests {10, 25}.
• Most studies using life-years gained as the measure for effectiveness are based on data from the USA and France {10, 12, 13, 16-18, 25}. Only one study is based on data from the Netherlands {19}.

Table 5: Incremental effects of FIT vs. gFOBT in life-years gained (LYG):

⁽¹⁾ Per person when using FIT instead of gFOBT (over a lifetime if not stated otherwise)

⁽²⁾ Attending initial screening

⁽³⁾ Attending subsequent rescreening

QALYs gained when using FIT instead of gFOBT:

All included cost-effectiveness studies that used QALYs as effectiveness measure support the results of the studies using life-years gained as the endpoint. The results of the comparison between FIT and gFOBT range from 0.036 QALYs gained per person over a lifetime by screening a cohort of 50-75 year-olds annually with FIT instead of gFOBT {22}, to 0.01 QALYs gained over a lifetime when annually screening a population of the same age with a low performance FIT instead of a gFOBT {14}. The study with the highest QALY gain also assumes the highest compliance with the screening strategies (75%). The smallest differences between the effectiveness of FIT and gFOBT occur when the test performance (sensitivity) is assumed to be low. The studies using QALYs as effectiveness measure are based on data from Canada {14, 15, 22}, England {23}, France {21} and Ireland {20}. Wilschut et al. (2011) {24} does not give the estimated effects of FIT and gFOBT and hence does not allow an estimation of the incremental effects.

Table 6: Incremental effects of FIT vs. gFOBT in Quality Adjusted Life-years (QALYs):

⁽¹⁾ Per person when using FIT instead of gFOBT (over a lifetime).

⁽²⁾ The time horizon is not clearly stated in the published article but it seems likely to be over a lifetime.

⁽³⁾ Attending initial screening

⁽⁴⁾ Attending subsequent rescreening

Increase in detected advanced tumours when using FIT instead of gFOBT:

One cost-effectiveness study uses detected advanced tumours as effectiveness measure {11}. This is an intermediate outcome that is used in clinical studies, and can only be assumed to be correlated with the final outcomes, life-years and QALYs gained. The results of this study show that the higher the cut-off level of the FIT (ranging from 20 – 148 ng/ml) the less effective FIT becomes compared to gFOBT (incremental effects ranging from 188 more detected tumours by FIT compared to gFOBT to 99 less detected by FIT). The reason for this is that the positive predictive value increases with the cut-off level, which means that a higher cut-off level reduces unnecessary follow-up colonoscopies but it also reduces the chance that a tumour is found if the level of blood in the stools is low (e.g. 20ng/ml).

Overall, the FIT seems to be more effective than gFOBT irrespective of the outcome measure used. The only exception is Zauber (2010) who compared FIT with a specific gFOBT (Hemoccult SENSA) and estimated that this specific gFOBT is marginally more effective (at the 5^th decimal place) than FIT {25}.