Abstract
Dear Sir,
Although Barlow et al. 1 are right that estimates of analytical performance characteristics in external quality assessment schemes (EQAS) may not reflect those attained in practice, their contribution on faecal occult blood tests (FOBTs) contains some flaws.
Firstly, their description of ‘the NHS Bowel Cancer Screening Programme’ is inaccurate: it is applicable only to England. Health is a devolved matter and the four countries of the UK have different approaches. The Scottish Bowel Screening Programme began in 2007 and is now fully rolled out, inviting all 50–74 y olds biennially using a two-tier reflex strategy. 2,3
Secondly, their experiment to determine ‘sensitivity’, i.e. the analytical detection limit (ADL) is flawed. Their basic assumption is that an ‘occult blood-negative’ sample contains no peroxidase activity (expressed as mg Hb/g material, not mg/Hb/g). Testing as negative with hema-screen FOBT (Alpha Labs Ltd, Eastleigh, Hants, UK) does not mean that there is no such material present: it simply shows that the concentration is below the ADL. Thus, if material with x mg Hb/g was added to a base sample with any material present, but with concentration less than the ADL, the final concentration would not simply be x mg Hb/g as assumed, 1 but lower. Thus, the true ADL for hema-screen must be less than the 0.7 mg Hb/g suggested. 1 Indeed, for a dichotomous test such as FOBT, the ADL is not the lowest concentration that gives a positive result 100% of the time but that which gives 50% negative and 50% positive results. This is superbly demonstrated in the Yorkshire External Quality Assessment Scheme. For each challenge, the percentages of positive and negative results found are reported. The challenge of 9 November 2009, with 1.2 mg Hb/g matrix, had 17.2% negative and 82.8% positive results. In contrast, the challenge of 21 December 2009, with 0.2 mg Hb/g matrix, had 83.6% negative and 16.4% positive results. If over a range of concentrations, the percentage positive was plotted on the ordinate against concentration on the abscissa, the resulting relationship would rise: if the percentage negative was similarly plotted, the relationship would fall. The concentration at which the relationships intersect is the true ADL. For hema-screen, using this objective approach, 4 the ADL is 0.6 mg Hb/g and not 0.7 mg Hb/g as suggested. 1
Thirdly, it was stated 1 that ‘while newer, more specific and more sensitive immunoassays are becoming available, these are more expensive so that the conventional colorimetric tests will remain for some time’. Guaiac-based FOBT have so many disadvantages that they should now be considered obsolete. 5 Newer faecal immunochemical tests (FITs) are specific for intact haemoglobin and its early degradation products and have many advantages. 5 They are not subject to interference from dietary constituents, are clinically more specific for bleeding in the colon and rectum, require less than the two samples from three stools needed for FOBT and generally have more user-friendly collection devices. They are used in many modern screening programmes and advocated in many guidelines. Although important, cost should not be the overriding consideration in selection of laboratory tests.
Fourthly, the only evidence for use of FOBT is in screening programmes. 5 No evidence-based guideline on investigation of lower gastrointestinal tract disorders advocates use of FOBT. Moreover, there are significant well-documented preanalytical problems in the collection, transport and handling of samples for FOBT. 5 A number of laboratories in the UK have already eliminated FOBT from their repertoires 6 and have actively discouraged use of FOBT in wards, clinics and primary care. I strongly suggest that the laboratory community follow these laudable examples. Then, of course, concerns about poor performance in EQAS would not exist and time and resources would not be wasted undertaking inconclusive studies. In the few clinical situations in which investigation for the presence of blood in faeces would possibly be helpful, including the detection of human faecal blood in the paediatric setting, investigation of symptomatic individuals when bowel visualization is difficult or impossible, follow-up of those with established disease, and in the investigation of familial colorectal cancer, I strongly suggest that FIT should be used.
DECLARATIONS