Statistics from Altmetric.com
Cook and Strachan are to be congratulated on their series of meta-analyses on the health effects of passive smoking. However, in their analysis of parental smoking and spirometric indices they gave as the main reason for excluding 19 out of 42 studies that met their primary criteria that they “provide some data, but insufficient to be included in the quantitative overview”.1 In the case of our own study2 they concluded that they were unable to transform our results to the desired effect measure. They used the “difference in outcome measure between the exposed and non-exposed children expressed as a percentage of the level in the non-exposed group” and reported that they were unable to do this with our results as we “reported differences in standard deviation scores with no baseline data”.
The standard deviation scores were calculated using the mean and standard deviation of the ratio of actual to lung function predicted for height, age and sex.3 Hence, the approximate percentage difference can be calculated by multiplying by the appropriate published standard deviation. Using an estimate of 15.6 cigarettes per day for the average amount smoked by parents at home, calculated from the same data for white English children in 1988,2 3 the effect of parental smoking on forced expiratory volume in one second (FEV1) was −0.37% (SE 0.51%) for boys and −0.18% (SE 0.77%) for girls. The wide confidence intervals on our estimates encompass the greater negative estimates of Cook et al,1 but inclusion of our results would have decreased their negative estimates for all four lung function parameters.
The approximation in assuming FEV1 percent predicted to be 100 for the unexposed group is no greater an assumption than combining studies using different definitions of parental smoking and different measures of mid expiratory flow. We invite Cook et al to update their estimates accordingly.
authors’ reply The omission of the study by Rona and Chinn1-1 from our meta-analyses1-2 is not an indictment of their study, but simply a reflection of the way the data were presented. It arose because the standard deviation necessary to transform the estimates in their paper to percentage deficits was not provided in that paper but published elsewhere.1-3 This is unlikely to have occurred in any of the other studies excluded. Updating our estimates to include their study serves to emphasise the robustness of our estimates to exclusion of individual studies. The fixed effects estimate for percentage reduction in FEV1amongst children in smoking households moved from −0.9% (95% CI −1.2 to −0.7) to −0.9% (95% CI −1.1 to −0.7) and the random effects estimate from −1.4% (95% CI −1.9 to −1.0) to −1.3% (95% CI −1.8 to −0.9).
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.