We thank Alfred Koerblein and Ian Fairlie for their interest in our article.1 They particularly commented on the standardized incidence ratio (SIR) we showed in Table 2 for children less than 5-years-old over the whole study period 1990–2007 [SIR = 1.37, 95% confidence interval (CI): [0.75–2.30] within 5 km around the plants]. This result is also explicitly detailed in the text of our article, “The SIR for the closest area was slightly but not significantly higher for the 0–4 year age group (1.4 [0.8–2.3]; p-one-sided = 0.15).” However, by no means it can be deduced that the increase would have been statistically significant if our sample size had been larger (“[this SIR] was not statistically significant because of small numbers,” Koerblein and Fairlie stated). Besides, our statement of no association for the whole study fits quite well with a SIR of 1.1, 95% CI [0.7–1.7] (children 0–14-years-old, period 1990–2007, Table 2).
Their joint analysis of four studies with a SIR around 1.4 does not surprinsingly lead to an estimate of 1.4 based on larger numbers. In a letter following the publication of the Swiss Canupis study, Korblein presented almost the same table on the UK, German and Swiss studies.2 The table underlines the consistency between the four published studies.
The comment by Claudia Spix, Peter Kaatsch and Maria Blettner compares the French Geocap study1 and the German KiKK study3 very well. We may just add that we had the opportunity to adjust for proximity to high-voltage power lines individually, while socio-demographic characteristics were only available at the municipality level and therefore accounted for as contextual variables. The main differences between the German and French results are that (i) the significant increase in incidence was only observed in the 2002–2007 period, (ii) the excess was not limited to the children less than 5-years-old and (iii) our results were little impacted by the exclusion of any one of the nuclear plants, contrary to the German situation, in which the exclusion of the Krummel nuclear power plant (where a cluster was previously detected) had the most noticeable impact.
We agree that the SSK report on the KiKK study highlighted the increase in childhood leukemia risk in rural areas and confirmed that the relationship with the distance to the power plants was similar in rural and nonrural areas.4 Such an increase in rural areas was not observed in our data, and the difference reported in Germany between incidence and case–control estimates was not observed. The points related to demographic characteristics specific to countries are very important and relevant, and should be discussed and accounted for in any project of international pooled analysis.
We also agree that the estimated doses due to gaseous discharges are very low, far below the levels for which a detectable risk of leukemia is foreseeable. We found very interesting and useful to benefit from the dose-based geographic zoning performed by IRSN, first to interclass the individuals within our studies and second to put our studies into perspectives with other studies relating radiation to leukemia risk. It would be worthwhile that such an indicator be constructed in other countries, including Germany.
Finally, although we cannot exclude chance as an explanation for the German and French results, the overall convergence of the European published results prompts a search for risk factors that may be concentrated around nuclear power plants. International collaboration and careful analysis of pooled data would be the most powerful way to unravel these questioning results.
Thus, our recommendation is to perform a pooled analysis of the individual data in Europe, including all the studies complying with usual criteria of quality, even unpublished to date. Such an analysis could use common indicators of distance, account for similar potential confounders and cover areas around power plants located close to borders. Some countries might also consider using dose-based zoning to quantify radiation exposure in the vicinity of power plants.
Altogether, we think that the important issue of describing and understanding the increase in childhood leukemia risk observed around power plants deserves more thoughtful examination based on international collaboration than has been devoted to the issue so far.
Jacqueline Clavel Claire Sermage-Faure Stéphanie Goujon-Bellec Jérémie Rudant Aurélie Guyot-Goubin Michel Chartier Dominique Laurier Denis Hémon