HIV status should also be documented in patients with tuberculosis-associated symmetrical peripheral gangrene.

HIV status should also be documented in patients with tuberculosis-associated symmetrical peripheral gangrene Sir, A major difference between the reporting of tuberculosis-associated 'simultaneous quadrilateral gangrene' in the era antedating awareness of the association of tuberculosis and human immune deficiency virus (HIV) infection, 1 and the reporting of tuberculosis-associated symmetrical peripheral gangrene in the modern era 2 is that, in the latter context, evaluation of the patient also ought to encompass documentation of HIV status, 2 given the fact that, in its own right, HIV infection may be complicated by large, medium and/or small-sized vasculitis, 3 and also by gangrene. 3,4 Digital gangrene was, for example, reported in a patient with HIV-associated polyarteritis nodosa (PAN), 3 and in three other HIV-infected patients with unspecified PAN status. 4 Accordingly, it is hoped that in the patient reported by the authors, the implementation of the strategy of management of the underlying causes of symmetrical peripheral gangrene 5 included evaluation of HIV status, as was the case in a comparable patient, reported in 1986, who in the event, tested negative for HIV co-infection. A. Rheumatological manifestations in HIV positive patients referred to a tertiary care centre. Mendel, fraud and the repeated analysis of data Sir, Professor Galton reflects on R.A. Fisher's implicit accusations of fraud in the genesis of Mendel's data: the results were simply too good to be true. 1 One of the many possible explanations for this is that, understandably, Mendel had no notion of the importance of specifying analysis plans in advance. For ongoing studies, such as clinical trials with continuous data collection, this is an important issue. 2 If data are repeatedly analysed, and the analyst has an expectation as to what the 'correct' answer should be, there will be a tendency to decide to stop data collection when the findings are close to the anticipated answer. 3 There is no suggestion in Mendel's paper that there was any planned total number of crosses that he and his assistants would examine, therefore it is plausible that, as the results came in, repeated analyses of the simple ratio of characteristics, with 3 : 1 being the expected answer, were carried out. When the results were particularly close to the expected ratio the continued tabulation of data stopped, and these were reported.