Statistics from Altmetric.com
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.
The recent proposals for smoke-free legislation in many countries have spawned a multitude of studies which attempt to predict the financial impact of such legislation. As described by Scollo et al1 in this issue of Tobacco Control, many of these studies fail to achieve basic quality standards and this is more likely when the tobacco industry funds the study. However, findings from such flawed studies can influence policy makers and it is essential that public health advocates have strategies to counter their impact.
In Hong Kong in 2001, the government proposed to make all workplaces, including catering venues, smoke-free.2 A consultancy report for the catering industry, funded by the tobacco industry, was published shortly after and concluded that the legislation would cause catering industry revenues to drop by 10.6% leading to job losses. This report was based on a survey of customers to catering venues, self reported spend on eating and drinking out, and self predicted changes in the event that catering venues were made smoke-free. Since the methods used were not made clear in the report, we had to attempt to validate or refute the report mainly by an assessment of its findings. We found the following questions useful:
(1) Was the sample used for the consultant’s survey representative of the population being studied (customers of catering venues)? Since we could not determine if sample selection was done properly, we had to look at sample characteristics. The prevalence of smokers was much higher than in other survey data indicating a bias in the sample.
(2) Did the consultant’s data, when extrapolated/aggregated, agree with other standard data sources—for example, government statistics? Much of the basic data collected by the consultants was not disclosed in their report but, to make their case, they had to present some—for example, average weekly spends in the different types of catering venues. From these data we could estimate (a) expected weekly revenue in the catering industry, (b) approximate market shares for the different types of venue, and (c) weekly spend on eating out per household if the consultant’s data were valid. Each of these estimates was quite implausible when compared with data from the census and other government sources.
(3) Could the consultant’s findings be reproduced to shed light on the methods used? Using a new set of data based on random sampling, we tried to recreate the consultant’s findings by deliberately introducing biases and incorrect aggregations which we suspected were present in the consultant’s methods. In this way we were able to produce an almost identical set of results from the new data. On the other hand, when we analysed the new data in an appropriate fashion, we predicted a rise of 5% rather than a drop of nearly 11% in catering revenues.
The best means of influencing policy on smoke-free catering venues is to use objective outcome measures and data collected both before and after the intervention, as recommended by Siegel and listed by Scollo et al.1 The study we were able to refute would have failed Siegel’s quality criteria. However, since much of the lobbying against smoke-free legislation is done before such policies are put in place, local objective, before and after data are inevitably not available. In our case, presenting our rebuttal of the consultant’s findings along with the evidence accumulated from overseas studies that smoke-free policies do not harm catering industry revenues, greatly reduced the harm that the consultant’s report could have done to the proposed legislative process. Our approach may be helpful to policy makers faced with a similar situation in their own locality.