Bayesian sensitivity analysis methods to evaluate bias due to misclassification and missing data using informative priors and external validation data.

BACKGROUND: Recent research suggests that the Bayesian paradigm may be useful for modeling biases in epidemiological studies, such as those due to misclassification and missing data. We used Bayesian methods to perform sensitivity analyses for assessing the robustness of study findings to the potential effect of these two important sources of bias. METHODS: We used data from a study of the joint associations of radiotherapy and smoking with primary lung cancer among breast cancer survivors. We used Bayesian methods to provide an operational way to combine both validation data and expert opinion to account for misclassification of the two risk factors and missing data. For comparative purposes we considered a "full model" that allowed for both misclassification and missing data, along with alternative models that considered only misclassification or missing data, and the naïve model that ignored both sources of bias. RESULTS: We identified noticeable differences between the four models with respect to the posterior distributions of the odds ratios that described the joint associations of radiotherapy and smoking with primary lung cancer. Despite those differences we found that the general conclusions regarding the pattern of associations were the same regardless of the model used. Overall our results indicate a nonsignificantly decreased lung cancer risk due to radiotherapy among nonsmokers, and a mildly increased risk among smokers. CONCLUSIONS: We described easy to implement Bayesian methods to perform sensitivity analyses for assessing the robustness of study findings to misclassification and missing data.

CT quality assurance in the lung screening study component of the National Lung Screening Trial: implications for multicenter imaging trials.

OBJECTIVE: The purpose of this study was to describe the effect of implementing an imaging quality assurance program on CT image quality in the Lung Screening Study component of the National Lung Screening Trial. MATERIALS AND METHODS: The National Lung Screening Trial is a multicenter study in which 53,457 subjects at increased risk of lung cancer were randomized to undergo three annual chest CT or radiographic screenings for lung cancer to determine the relative effect of use of the two screening tests on lung cancer mortality. Of the 26,724 subjects randomized to the CT screening arm of the National Lung Screening Trial, the Lung Screening Study randomized 17,309 through 10 screening centers. The others were randomized through the American College of Radiology Imaging Network. Quality assurance procedures were implemented that included centralized review of a random sample of 1,504 Lung Screening Study CT examinations. Quality defect rates were tabulated. RESULTS: Quality defect rates ranged from 0% (section reconstruction interval) to 7.1% (reconstructed field of view), and most errors were sporadic. However, a recurrently high effective tube current-time product setting at one center, excessive streak artifact at one center, and excessive section thickness at one center were detected and corrected through the quality assurance process. Field-of-view and scan length errors were less frequent over the second half of the screening period (p < 0.01 for both parameters, two-tailed, paired Student's t test). Error rates varied among the screening centers and reviewers for most parameters evaluated. CONCLUSION: Our experience suggested that centralized monitoring of image quality is helpful for reducing quality defects in multicenter trials.

Effects of smoking and radiotherapy on lung carcinoma in breast carcinoma survivors.

BACKGROUND: The combined effects of thoracic radiotherapy (XRT) and cigarette smoking are not known with certainty, but they have important implications for lung carcinogenesis after cancer therapy in some patients. The authors analyzed smoking, radiation, and both exposures on lung carcinoma development in women who were treated previously for breast carcinoma. METHODS: Case patients (n = 280) were female residents of the United States, ages 30-89 years, with breast carcinoma prior to primary lung carcinoma diagnosed between 1960 and 1997. Control patients (n = 300) were selected randomly from 37,000 patients with breast carcinoma who were treated at The University of Texas M. D. Anderson Cancer Center and frequency matched with women in the case group based on age at diagnosis (5-year strata), ethnicity, year of breast carcinoma diagnosis (5-year strata), and survival from breast carcinoma diagnosis to lung carcinoma diagnosis. Using stratified analysis and unconditional logistic regression, the authors evaluated the main and combined effects of smoking and XRT on lung carcinoma risk. RESULTS: At the time of breast carcinoma diagnosis, 84% of case patients had ever smoked cigarettes, compared with 37% of control patients, whereas 45% of case patients and control patients received XRT for breast carcinoma. Smoking increased the odds of lung carcinoma in women without XRT (odds ratio [OR], 6.0; 95% confidence interval [95% CI], 3.6-10.1), but XRT did not increase lung carcinoma risk in nonsmoking women (OR, 0.5; 95% CI, 0.3-1.1). Overall, the OR for both XRT and smoking, compared with no XRT or smoking, was 9.0 (95% CI, 5.1-15.9). Logistic regression modeling yielded an adjusted OR of 5.6 for the smoking main effect (95% CI, 2.9-10.5), 0.6 for the XRT main effect (95% CI, 0.3-1.4), and 8.6 (P = 0.08) for the combined effect. CONCLUSIONS: Smoking was a significant independent risk factor for lung carcinoma after breast carcinoma, but XRT alone was not. Smoking and XRT combined enhanced the effect of either alone, with marked increased risks of lung carcinoma after XRT for breast carcinoma.