Invited commentary: confounding, measurement error, and publication bias in studies of passive smoking (original) (raw)

Evaluating the evidence for the relationship between passive smoking and lung cancer

International Journal of Cancer, 2014

Dear sir, "No Clear Link Between Passive Smoking and Lung Cancer"-this was the title of a news article published in the December, 2013 issue of the Journal of the National Cancer Institute (JNCI). 1 It was reported in the article that investigators from Stanford and other institutions examined data from the Women's Health Initiative Observational Study (WHI-OS) and observed no association between passive smoking and lung cancer in an unpublished analysis. The only level of exposure with some indication of increased risk for lung cancer was living with a smoker for at least 30 years, with a hazard ratio of 1.61 and a confidence interval (CI) including the null. However, the analysis clearly lacked statistical power to detect an association even if one existed. Of the approximately 40,000 women who had never smoked, only about 10% reported no exposure to secondhand smoke and only 152 women developed lung cancer during follow-up. Therefore, as Dr. Heather Wakelee pointed out, "It's hard to say anything conclusive with such small numbers." 1 Limited sample size is a common problem among studies on the association between passive smoking and lung cancer, because it is difficult to identify nonsmoking lung cancer patients. Therefore, it is useful to increase statistical power by combining data from multiple sources. A recent study by researchers of the International Lung Cancer Consortium using pooled data on more than 2,500 never-smoking cases showed a clear association between passive smoking and lung cancer, with an adjusted odds ratio (OR) of 1.31 (95% CI: 1.17-1.47). 2 Although some may question the validity of case-control studies due to recall bias, metaanalyses of case-control studies on the risk of lung cancer among nonsmokers exposed to secondhand smoke from their spouses produced results very similar to those of cohort studies, with summary risk ratios of 1.20 (95% CI: 1.11-1.29) and 1.29 (95% CI: 1.125-1.49), respectively. 3 Furthermore, in our unpublished analysis of 458 lung cancer cases and

The Relation of Passive Smoking to Lung Cancer1

1986

The relation of passive smoking to lung cancer

Lung Cancer, 1987

To evaluate the role of passive smoking in the development of lung cancer among nonsmokers, data were pooled from three large incident case-control interview studies. Ninety-nine lung cancer cases and 736 controls never used any form of tobacco. Overall the adjusted odds ratio for lung cancer among nonsmokers ever living with a smoker was 0.8 (95% confidence interval, 0.5-1.3) rising to 1.2 among those exposed for 40 or more years. Persons living with a spouse who smoked cigarettes were at increased risk (adjusted odds ratio, 1.5; 95% confidence interval, 0.8-2.8). When adjusted for age and gender, there was a significant trend in risk with increasing amounts smoked per week by the spouse (P = 0.05) and with cumulative pack-years of exposure (P = 0.03). This effect was limited to females, especially older women whose husbands were heavy smokers. The elevated risk associated with spouse smoking was restricted to squamous and small cell carcinomas (odds ratio, 2.9; 95% confidence interval, 0.9-9.3), which provides additional evidence linking passive smoking to lung cancer.

Dissent (B) Respiratory health effects of passive smoking: EPA's weight-of-evidence analysis

Journal of Clinical Epidemiology, 1994

After an extensive review and assessment of the scientific evidence on the respiratory health effects of passive smoking, the U.S. Environmental Protection Agency has determined that the widespread exposure to environmental tobacco smoke in the United States presents a serious and substantial public health impact. The Environmental Protection Agency concluded that environmental tobacco smoke causes lung cancer in adult nonsmokers and increases the risk for a variety of noncancer respiratory disorders, especially in children. This article reviews evidence presented in the Environmental Protection Agency's 1992 report on the respiratory health effects of passive smoking and responds to critical allegations levied by Gio Gori in his article "Science, policy, and ethics: the case of environmental tobacco smoke", appearing in the same issue of this journal. Several recent studies appearing since the cutoff date for inclusion in the EPA report are also discussed. Environmental tobacco smoke Passive smoking Lung cancer Respiratory effects Sudden infant death syndrome Asthma nia. The report estimates that 150,000 to 300,000 cases annually in infants and young children up to 18 months of age are attributable to ETS. CE 4714-B 339 340 JENNIFER JINOT and STEVEN BAYARD ?? ETS exposure is causally associated with cancer, and (3) noncancer respiratory effects. increased prevalence of fluid in the middle This article does not discuss cardiovascular ear, symptoms of upper respiratory tract disease, sick building syndrome, or any other irritation, and a small but significant reissues addressed by Gori that were beyond the duction in lung function. scope of the EPA assessment. ?? ETS exposure is causally associated with additional episodes and increased severity of symptoms in children with asthma. The report estimates that 400,000 to l,OOO,OOO asthmatic childen have their condition worsened by exposure to ETS. ?? ETS exposure is a risk factor for new cases of asthma in childen who have not previously displayed symptoms. The report estimates that ETS may be responsible for 800&26,000 new cases annually. 'A priori probability of significant (p c 0.05) test of effect when true relative risk is 1.5. *RR used for cohort studies [44,46]; OR used for case-control studies. 'One-sided p-value for test of RR = I vs RR > I. +'Relative risks corrected for smoker misclassification. 5RR shown is for all cell types with the two control groups combined. 6Years of exposure as adult (spousal plus workplace). 'Relative risks not corrected for smoker misclassification. *Data not available.

Lung cancer due to passive smoking - a review

International Archives of Occupational and Environmental Health, 2001

Objectives: Even from the scientific literature it is difficult to conclude whether the increased risk of lung cancer due to exposure to environmental tobacco smoke (ETS), as reported in many epidemiological studies, is based on sound data from reliable studies, or rather on passionate assertions derived from unsound investigations. To shed some light on this matter the differences between cigarette mainstream smoke (MS) -inhaled by the smoker -and ETS -inhaled by everyone exposed -, the concentration of ETS under real life conditions, the internal dose of toxic compounds due to ETS exposure and the risk of lung cancer as found in epidemiological studies are discussed. Results: MS and ETS differ considerably in their physical, chemical and toxicological characteristics because of the different conditions under which they are generated, the dilution in air and the degree of ageing. Based on toxicological data, a very low internal dose of potentially genotoxic compounds can be measured in people after ETS exposure. The epidemiological data suggest a slightly increased risk of lung cancer in non-smokers chronically exposed to ETS. However, it is equally well known, that none of these studies are free from bias and confounding effects. Conclusion: The average intake of toxic and genotoxic compounds due to ETS exposure is that low that it is difficult, if not impossible, to explain the increased risk of lung cancer as found in epidemiological studies. The uncertainty is further increased because the validity of epidemiological studies on passive smoking is limited severely by numerous bias and confounding factors which cannot be controlled for reliability. The question of whether or not ETS exposure is high enough to induce and/or promote the carcinogenic effects observed in epidemiological studies thus remains open, and the assumption of an increased risk of lung cancer due to ETS exposure is, at present, more a matter of opinion than of firm scientific evidence.

Passive smoking: Authors' reply

BMJ, 2003

Study was flawed from outset Editor-The study by Enstrom and Kabat has a major flaw, 1 and I urge the editors of the BMJ to consider a retraction. The study assumes a considerable difference in the exposure to environmental tobacco smoke of never smokers' spouses compared to ever smokers' spouses. This is obviously wrong. Most never smokers' spouses would have been exposed to considerable environmental tobacco smoke before the late 1990s when Californian public places became smoke-free. Thus for most of the study period, assuming the spouses are together for two to four waking hours a day, the comparison is eight to10 hours' exposure to tobacco smoke among spouses of never smokers and 12 hours' exposure to tobacco smoke among spouses of ever smokers. Assuming passive smoking increases mortality by 30%, the demonstrable difference between the groups would be about 5% ((12 − 10)/12)×30). This would be further reduced because of quitters among ever smokers and occasional smokers among never smokers. A 5% difference is extremely difficult to show in an epidemiological study, and inability to find a difference cannot be taken as absence of a difference. However flawed this study, unless it is retracted by the BMJ the tobacco industry will use it to promote their vigorous opposition to antismoking legislation in general, and anti-environmental tobacco smoke laws in particular, creating controversy where there isn't any. Of course they have an urgent and ongoing need to replace loss of their customer base-10 000-20 000 lives per daywith new recruits of young smokers.

Passive smoking: Wider evidence needs to be interpreted

BMJ, 2003

Tobacco smoking and cancer: A meta-analysis

International Journal of Cancer, 2008

We conducted a systematic meta-analysis of observational studies on cigarette smoking and cancer from 1961 to 2003. The aim was to quantify the risk for 13 cancer sites, recognized to be related to tobacco smoking by the International Agency for Research on Cancer (IARC), and to analyze the risk variation for each site in a systematic manner. We extracted data from 254 reports published between 1961 and 2003 (177 case-control studies, 75 cohorts and 2 nested case-control studies) included in the 2004 IARC Monograph on Tobacco Smoke and Involuntary Smoking. The analyses were carried out on 216 studies with reported estimates for ‘current’ and/or ‘former’ smokers. We performed sensitivity analysis, and looked for publication and other types of bias. Lung (RR = 8.96; 95% CI: 6.73–12.11), laryngeal (RR = 6.98; 95% CI: 3.14–15.52) and pharyngeal (RR = 6.76; 95% CI: 2.86–15.98) cancers presented the highest relative risks (RRs) for current smokers, followed by upper digestive tract (RR = 3.57; 95% CI: 2.63–4.84) and oral (RR = 3.43; 95% CI: 2.37–4.94) cancers. As expected, pooled RRs for respiratory cancers were greater than the pooled estimates for other sites. The analysis of heterogeneity showed that study type, gender and adjustment for confounding factors significantly influence the RRs estimates and the reliability of the studies. © 2007 Wiley-Liss, Inc.

Invited commentary: confounding, measurement error, and publication bias in studies of passive smoking (original) (raw)

Related papers