Intended for healthcare professionals

  1. News & Views
  2. Authors' reply
  3. Authors' reply
Letters

Authors' reply

BMJ 1996; 312 doi: https://doi.org/10.1136/bmj.312.7046.1607c (Published 22 June 1996) Cite this as: BMJ 1996;312:1607
  1. Roger Buchdahl,
  2. Tabitha Stebbings,
  3. Abdel Babiker,
  4. Alison Parker
  1. Consultant paediatrician Hillingdon Hospital, Middlesex UB8 3NN
  2. Senior environmental health officer London Borough of Hillingdon, Environmental Protection Unit, Uxbridge UB8 1UN
  3. Senior lecturer in epidemiology MRC HIV Clinical Trials Centre, University College of London Medical School, London WC1E 6AU
  4. Paediatric respiratory nurse Hillingdon Hospital, Middlesex UB8 3NN

    EDITOR,—The implied criticism by Merion R Evans and colleagues of not using an a priori hypothesis in our study perhaps fails to appreciate that the study was observational, undertaken in response to the controversy concerning air pollution and childhood asthma. The findings have not emerged as a result of performing “a series of complex statistical tests,” as can be seen from the plots of the raw data.

    The following a posteriori hypotheses may be made. Firstly, there may by a critical concentration of ozone which has a beneficial health effect. A second, and in our view a more plausible, hypothesis recognises ozone as a secondary pollutant. Under certain atmospheric conditions predisposing to very low or very high levels of ozone, other air pollutants may accumulate and these may be related more closely to incidence of wheeze. We speculate that volatile organic compounds or other hydrocarbons may be relevant. A test of the hypothesis would be to see whether these compounds accumulate during low or high ozone episodes and, if they do, whether they are of clinical relevance.

    Jon G Ayres is concerned that there was insufficient adjustment for other confounding meteorological factors and critically black smoke or particulate matter. In fact, relative humidity, wind speed, barometric pressure, and sunshine hours were examined in the statistical models, but for reasons of editorial brevity these data were not included in the paper. The effect of ozone was not appreciably altered.

    Graham Bentham and Peter Brimblecombe comment on the unexpectedly high ozone levels recorded in October and November. We have now repeated the analysis using pollutant data derived from the fixed point monitor in London Bloomsbury for the same time period as the study (courtesy of AEA Technology, National Environmental Technology Centre, Abingdon). The monitor is located some 20 km east of the Heathrow OPSIS monitor. For ozone, sulphur dioxide, and nitrogen dioxide the mean differences between the Bloomsbury and OPSIS monitors were +17, -11, and -10 µg/m3 respectively with 95% confidence intervals not exceeding 5 µg/m3 for any of these pollutants, although for the period 13 October to 15 November the OPSIS site gave significantly higher levels of ozone. Despite these differences, Poisson regression analysis using the pollutant data from Bloomsbury adjusted for season, temperature, and wind speed revealed a similar U shaped relation between wheezy incidence and ozone (fig 1). Additional adjustment using data on particles less than 10 microns in size (PM10), which was available from the Bloomsbury site, had no significant effect on this relation.

    Fig 1
    Fig 1

    Unadjusted Lowess smoothing averages for incidence of wheeze and 24 hour average ozone for two sites in London

    Back to top