Erratum: “Indoor Radon Exposure and Risk of Lung Cancer: a Nested Case-Control Study in Finland,” by Auvinen et al. [J Natl Cancer Inst 1996; 88:966–72 (Issue 14)].

The authors report the following:

We made an error (misspecified conditional logistic regression model) that affected all the matched results in the data analysis of our previously published article (1). We have therefore reanalyzed the data. The revised results from the matched analysis are shown in the revised Tables 3–5. Unmatched results were not affected.

In general, all the odds ratios tend to be larger and confidence intervals wider compared with the incorrect results. However, the differences are not substantial (e.g., the significance levels remain unchanged); hence the conclusions are not fundamentally changed.

When the radon concentration was stratified, only the category with the highest radon concentration (⩾10.8 pCi/L or ⩾400 Bq/m³) suggested some elevation of risk (odds ratio = 2.7; 95% confidence interval = 0.8–9.2), even though the confidence intervals were wide and included unity (i.e., no risk elevation). This group also strongly influenced the linear risk estimate.

The revised risk estimates are consistent both with no effect and with the magnitude of risk suggested by studies of miners (2). One needs to be cautious in drawing conclusions because of wider confidence intervals and the possibility of residual confounding by much stronger risk factors such as smoking.

Author notes

We thank Professor Timo Hakulinen and Tadeusz Dyba for statistical consultations.