Guide wire manipulation of Crosby jejunal biopsy capsule.

SIR,-Dr Greg Holdstock and others (4 February, p 362) say that the aim of their study was to determine whether smoking is associated with less severe forms of ulcerative colitis or Crohn's disease. After analysing 172 variables in three groups of patients they present seven significant differences. For three of the differences the significance levels are modest (p<0 05, 0 05, and 0 04), and in assessing the importance of their results it would be helpful to know just how many comparisons were attempted. Restricting oneself to the eight variables in the published table a minimum of 24 comparisons are possible and therefore at least one "significant" difference might be expected to occur through chance alone. It would appear also that the Student's t test has been applied to such variables as number of relapses per patient, number of hospital admissions per patient, bowel frequency, and severity of pain. For all these variables the standard deviations quoted indicate that the data are far from normally distributed. We suspect that any differences result from the behaviour of a small minority of patients, who may not be representative of patients with Crohn's disease in general. Our reason for saying this is that the prevalence of the two types of inflammatory bowel disease in their study is quite different from the two to one ratio of ulcerative colitis to Crohn's disease usually found in the United Kingdom. Dr Holdstock and his colleagues then suggest that because they found a low prevalence of smoking among patients with ulcerative colitis, "Smokers therefore appear to have a reduced risk of developing the disease." This suggestion is plainly unwarranted because they present no evidence about the smoking habits of patients before onset of disease (or indeed any control data). They will find such evidence if they consult their references, in which we show (using control data) that the association does antedate onset of disease.' They proceed to suggest that smoking leads to Crohn's disease rather than ulcerative colitis. As they produce no evidence comparing the smoking habits of patients with Crohn's disease with those of controls their speculations are ill founded. Nevertheless, they may be right, for we have found that patients with Crohn's disease are more likely to be smokers than matched controls and the association antedates onset of disease.2