Sir,

The article by Hirabayashi et al1 is an interesting report on the inhibitory effects of maternal smoking on the development of severe retinopathy of prematurity (ROP). However, I do not believe that the conclusion derived (that maternal smoking leads to a reduction in the incidence of severe ROP) is at all supported by the results reported. There were 27 infants that developed severe ROP, of whom only a single mother smoked (and the other 26 mothers were non-smokers). The authors’ conclusion that maternal smoking reduced the incidence of severe ROP is based on a single smoker, as they ignored the 26 other non-smoking mothers. In fact, using the reported events rates for development of severe ROP (1/27 maternal smokers versus 26/27 non-smokers), one obtains a relative risk (RR) of 0.04 and 95% CI of 0.01–0.26 (P=0.0009, see Figure 1). This clearly shows that non-smoking provides protection against the development of severe ROP, with a reduction in risk of 96% compared with maternal smoking. Strangely enough, the authors reported these data using odds ratios, especially as the event rate in the maternal smoking group is low and their reported 95% CI (Table 2, p 1026) includes ‘0’ in the interval, making the result statistically non-significant. Therefore, one can only conclude that maternal smoking does not reduce the incidence of severe ROP.

Figure 1
figure 1

Risk of developing severe retinopathy of prematurity during maternal smoking.

Full size image

The authors have erroneously concluded that maternal smoking reduced the incidence of severe ROP, when in fact only 1/27 (or 4%) reported maternal smoking and 26/27 (or 96%) did not report any maternal smoking. Lack of evidence does not equate to evidence of an effect (or association in this case). In the non-severe ROP group, 15/59 (or 25%) mothers reported maternal smoking and the authors did not report a reduction in the incidence of non-severe ROP. Re-analysis of the reported data (development of non-severe ROP; smokers 15/59 versus non-smokers 44/59) provides the following: RR 0.12, 95% CI 0.05–0.27 (P<0.00001), favouring non-smokers with a reduction in the incidence of non-severe ROP of 88%.

The correct and only conclusion from this report should read as follows: No maternal smoking provides protection against the development of both severe and non-severe ROP. There is no evidence to support that maternal smoking offers any protection against the development of ROP (Figure 1).