The FAST score was proposed as a risk scoring model that would effectively “rule-out” CRC if the score was < 2.12 and therefore ease the pressure on stretched endoscopy services. The published FAST score was derived from a cohort consisting of 1572 consecutive symptomatic patients referred to colonoscopy who were included in the derivation cohort for the COLONPREDICT study.14 The validation cohort included data from three studies conducted in Scotland and two in Spain.16 However, it was acknowledged that the patients included in the studies for the derivation and validation of the score had been selected a priori by health care professionals for further evaluation. To date, only one study has previously investigated the performance characteristics of the FAST score in primary care.4
Digby et al. 4 demonstrated that 79% of the 1447 patients referred in their study cohort had a FAST score ≥ 2.12, thus, 21.0% would be deemed to have a low risk of CRC and would not require colonoscopy. The clinical sensitivity in their cohort for the FAST score ≥ 2.12 was 99.0% (95% CI: 94.3–100.0) and 96.6% (95% CI: 93.9–98.4) for CRC and SCP respectively, and the specificity was 22.4% (95% CI: 20.2–24.7) and 25.5% (95% CI: 23.1–28.2). In their cohort, eight cases of SCP would have been missed if the FAST score ≥ 2.12 was used as compared to f-Hb plus clinical data, but only one case of CRC, which would have been referred to secondary care since the patient had IDA and would have met NG12 criteria for further investigation.
In alignment with Digby et al.4 and Herrero et al.7 our analysis of the application of the FAST score in patients referred from primary care with suspected CRC symptoms, demonstrated that the colonoscopy demand would only be reduced by 9% using a FAST score ≥ 2.12 cut-off. The sensitivity of the FAST score ≥ 2.12 in the detection of CRC and SCP in our cohort was very similar to that reported by Digby et al.4 but the specificity was much lower. This may be explained by the underlying differences in the population groups, particularly the lower prevalence of CRC (6.5% vs 8.2%) and SCP (17.5% vs 25%) in our study compared to Digby et al.4 Five cases of SCP would have been missed if the FAST score ≥ 2.12 was used as compared to f-Hb plus clinical symptoms, but only one case of CRC, which would have been referred to secondary care since the patient had IDA, change in bowel habit, weight loss and abdominal pain and would have met NG12 criteria.
Further examination of the cases of SCP which would have been missed if the FAST score had been applied showed the importance of clinical rationale and safety netting, as all five patients had “red flag” symptoms that would have led to referral to secondary care. However, a limitation of our study and the study conducted by Digby et al.4 is that the data has been collected retrospectively. The FAST score has been examined as a potential follow-up investigation prior to acceptance into colonoscopy, assessing whether the score, applied after referral on the grounds of clinical symptoms and f-Hb, would lead to a reduction in colonoscopies. Whilst our study suggests that a FAST score ≥ 2.12 cut-off would reduce colonoscopy demand by 9%, several patients with CRC and SCP would be missed.
In those in group B, who were not referred, possibly in the main because 2998 (94.0)% had f-Hb < 10 µg Hb/g faeces, 2226 (69.8%) had a FAST score ≥ 2.12 and thus, if this became the criterion for referral, many more colonoscopies would be required, without much evidence of the value of further investigations in this group. However, clinical rationale led to only 149 (4.7%) being referred to secondary care for further investigation over the subsequent year and only 7 of these patients were ultimately investigated. There were no cases of CRC or SCP in group B. A clear limitation of this study is that colonoscopy was not performed on all of group B, but this reflects real practice.
Our data does not support use of the FAST score in patient triage for colonoscopy in our population.