The Study Objective
The objective of the study was to determine the role of active smoking in the occurrence of breast cancer among women. Specifically, the study aimed at establishing the extent to which active smoking increases the risk of breast cancer among Canadian women.
Primary Exposure and Outcome of Interest
The primary exposure of the study is tobacco because it contains carcinogens. According to Catsburg, Kirsh, Soskolne, Kreiger, and Rohan (2014), tobacco contains carcinogens such as aromatic amines, N-nitrosamines, and polycyclic aromatic hydrocarbons. Active smoking increases the concentration of these carcinogens and consequently increases the risk of breast cancer. The outcomes of interest for the study are the causes of breast cancer that associate with exposure to cigarette smoking. The number of cases of breast cancer indicates the degree of association between active cigarette smoking and the occurrence of breast cancer.
The study employed a case-cohort design in investigating the role of active smoking in the occurrence of breast cancer among women in Canada. Case-cohort design is appropriate for the study because it is cost-effective and efficient in the study of rare outcomes of interest (Catsburg et al., 2014). In the case-cohort design, the study selected 3314 women exposed to tobacco smoke and 1096 women with breast cancer.
Study Population and Sampling
The study population was Canadian women obtained from a prospective study, the Canadian Study of Diet, Lifestyle, and Health (CSDLH), which was conducted between 1995 and 1998. The study population comprises alumni of the major universities in Canada, namely, Toronto, Alberta, and Western Ontario. Another study population is women who took part in a cancer survey in 1992 and their data kept by the Canadian Cancer Registry. The sampling process entailed the selection of women who actively smoked tobacco and had breast cancer, according to the information obtained from self-administered questionnaires. In essence, the study selected 3314 women from a population of 39,532 women and further selected 1096 women who suffered from breast cancer.
Statistical Analysis and Primary Measures of Association
The study performed statistical analysis to determine the association between active smoking as exposure and the risk of breast cancer as an outcome using the Cox regression model. According to Catsburg et al. (2014), the Cox regression model considered the existence of confounding factors by adjusting for the effects of known risk factors for breast cancer. In the measurement of the association between cigarette smoking and the risk of breast cancer among women, the study employed hazard ratios. The hazard ratios are appropriate measures of the association because they assess the degree of risk that cigarette smoking poses.
Potential Confounders and the Technique to Minimize Them
The potential confounders of the study are alcohol intake, menopausal status, oral contraceptives, hormone therapy, physical activity, routine mammogram, and the number of live births. In determining the role of active smoking as a risk factor for breast cancer, the study considered the effects of these confounders. The study used the Cox regression model in minimizing the effects of these confounders while unmasking the effects of active smoking.
Potential Effect-Modifiers and the Technique to Analyze Their Effects
The potential effect modifiers are the age of women and menopausal status. The technique to analyze their effects is the Cox regression model, which allows for the adjustment of these potential effect modifiers. Moreover, stratification of women according to their age and menopausal status provides a robust way of analyzing the effect modification.
Summary of Major Results
The results indicate that there is no significant association between active smoking and the occurrence of breast cancer among women in Canada. However, the results do not negate the existence of association because effect modifiers, confounders, and early-life exposures explain the existence of a relatively weak association.
The random error might have affected the results because data collection and data analysis entailed the use of instruments and software. Given that the study relied on the data that participants provided in the self-administered questionnaires, there is a chance that the participants made some errors in filling the questionnaires. As a consequence, the data collected had random error owing to the inability of data to reflect the smoking status and diagnosis of breast cancer. Moreover, stratification of participants according to their ages, menopausal status, alcohol use, smoking status, and family history amongst other factors might have contributed to the random error. Stratification causes a random error because it complicates the sorting process and introduces research biases. Given that the Cox regression model adjusted for effect modifiers and confounders, the adjustments might have introduced random error during data analysis.
Possible Selection Bias
The sampling bias is a possible selection bias that might have affected the representation of participants and reduced the external validity of the findings. A sampling of participants from an already existing cohort study magnifies sampling error because it is prone to primary and secondary sampling biases. In sampling, it is evident that the study selected 3314 participants from the sampling frame of 39,532, and thus, increasing the probability of selecting a biased sample that does not represent the entire population. Moreover, the selection of 1096 participants from the Canadian Cancer Registry contributes to the occurrence of sampling bias because it introduces a different set of participants, who do not have the same attributes as the other sample of participants. Overall, the sample bias affected the findings of the study because the size and the composition of the participants did not effectively represent the target population, who are the Canadian women.
A possible source of miscalculation bias might have originated from data entry, calculation of hazard ratios, and adjustments for confounders. The entry of data into statistical analysis software might have caused some miscalculation bias because the data entry is a complex process that demands accuracy. Given that the calculation of hazard ratios requires expertise in data analysis and stringent caution, a careless analysis might have introduced biases that reduce the accuracy and validity of data. The existence of confounders required data analysis to make some adjustments to eliminate or minimize their effects. In this view, the process of adjusting might have weakened the association between active smoking and the occurrence of breast cancer.
Limitation of the Study
The limitation of the study is that it has low external validity because it did not consider the effects of environmental tobacco. In essence, the study excluded populations that have exposures to environmental tobacco. Although the study used a large sample, statistical analysis using the Cox regression model is not sensitive to the existence of modest associations between active smoking and breast cancer among women. Moreover, the limitation of the study is the undermatching of cases and controls leading to biased outcomes.
Critique of Discussion
The discussion section effectively discusses the findings of the study by interpreting the results obtained from data analysis. Although the article highlights the strengths and limitations of the study, it does not adequately address all of them. The article points out the prospective design, the large sample size, and the lack of recall bias as some of the strengths. The article should have also highlighted the reliability and validity of the questionnaires used and the accuracy of instruments used for screening breast cancer as strengths. Regarding limitations, the article noted the lack of data on environmental tobacco exposure and the inability to measure modest associations as weaknesses. In this view, the article did not consider sampling bias and self-administration of questionnaires as additional weaknesses.
Analysis of the study indicates that its findings have low external validity. In essence, the findings are only generalizable to Canadian women, who have exposure to active smoking and breast cancer. Given that the study analyzes data collected in the late 1980s and early 1990s, the findings apply to that period and have limited application to the current society.
The conclusion made by the authors is appropriate to the study because it neither refutes early findings nor reveals new findings. Given that limitations of the study significantly reduce the validity and reliability of findings, the assertion of new findings could have elicited contrary opinions. Thus, the authors carefully reported the findings by acknowledging the existence of weak association despite the fact data analysis indicated no statistically significant association. Fundamentally, the authors chose to report their findings in support of earlier findings.
Future studies should consider employing cohort study as a research design in determining the association between active smoking and the occurrence of breast cancer. The cohort study is appropriate because it is a prospective study, which allows researchers to monitor the progress of participants throughout their period of exposure. Moreover, future studies should consider including participants with different demographic attributes such as age and race in determining breast cancer risk posed by smoking. Given that confounders and effect modifiers distort findings, future research should consider applying numerous methods in adjusting for their distorting effects.
Catsburg, C., Kirsh, V., Soskolne, C., Kreiger, N., & Rohan, T. (2014). Active cigarette smoking and the risk of breast cancer: A cohort study. Cancer Epidemiology, 38(1), 376-381.