Access to the full text of this article requires a subscription or payment. Please log in or subscribe below.

Did paper have statistical discrepancies?

  1. Adam Jacobs, director (ajacobs@dianthus.co.uk)
  1. Dianthus Medical, London SW19 3TZ
  2. St Thomas Research Unit, St Thomas Health Centre, Exeter EX4 1HJ
  3. Department of Emergency Medicine, Royal Melbourne Hospital, Melbourne, Australia
  4. Julius Centre for General Practice and Patient Oriented Research, University Medical Centre, Utrecht, 3584 CG Utrecht, Netherlands
  5. Department of Gastroenterology, University Medical Centre, Utrecht

    EDITOR—I am confused by Weijnen et al's description of the statistical methods used in their study and how they fit with the data presented.1 The methods section states that all variables found to be univariate predictors of peptic ulcer with P<0.25 were entered in the multivariate regression model. However, the results section says that age was included in the model, although table 2 shows that it was not predictive (P=0.67).

    Table 2 also shows that P=0.24 for both hiatal hernia and pain after meal, so these should have been included in the multivariate model, but neither of them was. Are these discrepancies due to a typing mistake, or is there another explanation?

    References

    1. 1.↵
      1. Weijnen CF,
      2. Numans ME,
      3. de Wit NJ,
      4. Smout AJPM,
      5. Moons KGM,
      6. Verheij TJM,
      7. et al
      .Testing for Helicobacter pylori in dyspeptic patients suspected of peptic ulcer disease in primary care: cross sectional study.BMJ2001;323: 71–75. (14 July.)
      OpenUrlFREE Full Text

    Authors' suggestion muddies waters in debate

    1. Alexander Williams, general practitioner (alex.jane@virgin.net)
    1. Dianthus Medical, London SW19 3TZ
    2. St Thomas Research Unit, St Thomas Health Centre, Exeter EX4 1HJ
    3. Department of Emergency Medicine, Royal Melbourne Hospital, Melbourne, Australia
    4. Julius Centre for General Practice and Patient Oriented Research, University Medical Centre, Utrecht, 3584 CG Utrecht, Netherlands
    5. Department of Gastroenterology, University Medical Centre, Utrecht

      EDITOR—Weijnen et al suggest that we should test and treat patients at high risk of peptic ulceration.1 This seems to muddy the waters in the debate about testing for Helicobacter pylori infection in primary care. Of the 38 patients they identified as having a peptic ulcer, only 22 gave a positive result to a non-invasive H pylori test, although the rate of detection overall was increased from 31% to 41% by using invasive tests of culture or histology, which suggests that serological testing is not as sensitive.

      Agreus and Talley reported that the sensitivity of H pylori enzyme linked immunosorbent assay (ELISA) kits had an average sensitivity of 85% (low and high extremes 49% and 99% respectively).2 Why are the rates of detection so low in Weijnen et al's study (33 of the 38 patients …

      Access to the full text of this article requires a subscription or payment

      Subscribe

      Article access

      Article access for 1 day

      Purchase this article for £20 $30 €32*

      The PDF version can be downloaded as your personal record

      * Prices do not include VAT

      Rapid responses

      THIS WEEK'S POLL

      BMJ most popular