Sir,

We appreciate the letter from Ram and McDonald (2011) regarding our manuscript.1 We understand that this letter questions our statistics. We discussed this issue with our statisticians and would like to argue against the letter.

  1. 1)

    A relative risk can be obtained only by cohort study, not by case–control study as performed in our manuscript. Only ‘odds ratio’ can be obtained by case–control study. If Ram does want to work out a relative risk ratio using our data set, probably it will be better to subdivide the group into ‘smokers’ and ‘non-smokers’. Dividing into severe ROP and non-severe ROP groups is not appropriate, as they are just the outcome of the observation. Anyhow, it does not make sense in such a case–control study.

  2. 2)

    We would like to explain the meaning of 95% CI. We reported the odds ‘ratio’ for possible risk factors of severe ROP. If the 95% CI of odds ratio includes ‘1’ (not 0) in the interval, it makes the result statistically non-significant. However, the 95% CI for maternal smoking was lower than ‘1’ in our result, which clearly showed statistical significance.

Hence, we would like to say that our statistical analysis was correct. However, we never recommend maternal smoking, owing to a number of smoking-related systemic adverse events in mothers and infants. We reported our results only because they may give insight into some aspects of complicated ROP pathogenesis.