We thank S. Stanojevic and co-workers for their letter and the opportunity it gives us to provide some clarification.
Although our paper 1 is recent, the data were collected in the European Community Respiratory Health Study (ECRHS) I, carried out from 1990–1992, by research teams who had not worked together previously and who came from diverse healthcare systems and different language groups, across four continents.
We are well aware of improvements and clarifications that could have been made to the written protocol and to training and quality-control procedures. Unfortunately, we do not have access to the “in press” paper 2 referred to in the letter from S. Stanojevic and co-workers. While biological controls may have some advantages for assessing between-centre variations, it is not at all clear to us how many would be needed and what the effects of training would be. However, almost certainly the number of biological controls that would have been needed to assess intra-subject variability between centres would not have been available.
It is not correct to say that “a quarter of centres did not measure height”. We reported that: “Out of 42 centres, it was measured in 31, self-reported in five and not recorded whether measured or asked in six.” We were being scrupulously honest and in five of the latter it is likely that height was measured. However, study personnel have moved on and definitive information could not be retrieved once we had realised that some centres had not measured height directly. In three of the centres that we classed as “self-reported height”, subjects were measured if any doubt was expressed, and gross errors are unlikely to have occurred. Although the over-estimation of height in the study by Stewart et al. 3 was nontrivial, that found by Niedhammer et al. 4 was <0.5 cm on average. The other reference does not seem relevant.
There was exclusion for non-White ethnicity in only one centre: Melbourne (Australia). Ethnicity was not recorded in any of the other centres, which are listed in table 1 of our paper 1, but this was not raised as an issue.
It was our intention in writing the paper to engender debate and our conclusions are endorsed by S. Stanojevic and co-workers. We believe that current reference curves cannot be guaranteed to give accurate norms of lung health, and that multicentre studies must invest substantially in standardised equipment. However, “statistical models which can adjust for between-centre differences”, as advocated by S. Stanojevic and co-workers, do not solve the problem, as differences may be due to genuine variation in health.
- © ERS Journals Ltd