Do risk scores improve use of faecal immunochemical testing for haemoglobin in symptomatic patients in primary care?

Abstract

AimFaecal immunochemical testing (FIT) is used in the detection of colorectal cancer (CRC). FIT is invariably used at a single faecal haemoglobin (f‐Hb) concentration threshold. The aim of this observational study was to explore risk scoring models (RSMs) with f‐Hb and other risk factors for CRC in symptomatic patients attending primary care, potentially speeding diagnosis and saving endoscopy resources.MethodRecords of patients completing FIT were linked with The Scottish Cancer Registry and with other databases with symptoms, full blood count and demographic variables, and randomized into derivation and validation cohorts. Stepwise multivariable logistic regression created RSMs assessed in the validation cohort.ResultsOf 18 805 unique patients, 9374 and 9431 were in the derivation and validation cohorts, respectively: f‐Hb, male sex, increasing age, iron deficiency anaemia and raised systemic immune inflammation index created the final RSM. A risk score threshold of ≥2.363, generating the same number of colonoscopies as a f‐Hb threshold of ≥10 μg Hb/g gave improved sensitivity for CRC in both cohorts. A RSM which excluded f‐Hb was used to investigate the effect of raising the f‐Hb threshold from ≥10 to ≥20 μg Hb/g in those with a low risk score. This approach would have generated 234 fewer colonoscopies but missed four CRCs.ConclusionThe RSM conferred no significant benefit to patients with very low f‐Hb and CRC. Alternative strategies combining FIT with other variables may be more appropriate for safety‐netting of symptomatic patients. Further work to develop and investigate the value of RSM for significant bowel disease other than CRC may also be beneficial.

COLOFIT: Development and internal-external validation of models using age, sex, faecal immunochemical and blood tests to optimise diagnosis of colorectal cancer in symptomatic patients

“…The FAST score did not incorporate blood tests into the risk prediction model, instead using f-Hb level, age and sex 26 . A further study modelling the risk of CRC using logistic regression in a population in Scotland concluded no benefit to the use of a risk score in those with a low FIT test however it suggested that modelled risk might allow the raising of the current threshold and thus reduce endoscopy demand 27 . A recent systematic review of the performance of f-Hb-based risk prediction models identified 22 studies combining FIT with one or more variable to predict the risk of CRC or advanced colonic polyps 11 .…”