GPs can separate oncological wheat from chaff

Editor—Summerton's editorial on the problem of identifying symptoms potentially indicating an underlying cancer in primary care calls for research based theory.¹ Only community based studies will help general practitioners to decide on the importance of a symptom or physical sign reported by their patients as hospital series are unrepresentative.

Although selection bias is part of the problem, it may be comparatively minor as most patients with cancer are treated in secondary care, even though their disease is diagnosed in primary care. Perhaps more important is bias from the development of symptoms in the time from the first symptom appearing to presentation in primary care and finally hospital.

General practitioners are already quite successful in sifting out the wheat from the chaff (although the comparative rarity of cancer in primary care indicates that a better farming analogy would be finding a needle in the haystack). For example, an estimated positive predictive value for colorectal cancer of rectal bleeding in the community is 0.1%.²,³ Once it is reported to the doctor this rises to 2-3%,³,⁴ and in patients referred for investigation it is 5.2%.⁵