Samantha Rees

Purpose: Evidence is inconsistent regarding an association between alcohol consumption and survival following a breast cancer diagnosis; therefore, we examined how change in alcohol consumption affects all-cause mortality (ACM) in a population of breast cancer survivors. Methods: We prospectively assessed pre-diagnosis, pre- to post-diagnosis, and post-diagnosis change in alcohol consumption among 1,558 postmenopausal women (mean age-at-diagnosis: 71 years) diagnosed with non-metastatic invasive breast cancer from the American Cancer Society Cancer Prevention Study-II Nutrition Cohort (baseline: 1992–1993). Five categories of alcohol intake were determined at each assessment; change in alcohol consumption was categorized as stable, quit, reduced, and increased. Change occurred at a median of two years before (pre-diagnosis) and two years after (post-diagnosis) diagnosis. Cox proportional hazards regression was used to estimate adjusted hazard ratios (HR) and 95% confidence intervals (CI) for the associations between change groups and ACM. Results: Over a median follow-up of 12.6 years, 750 women died (130 breast cancer deaths). Compared to stable consumption, the association between cessation and ACM varied by change interval: pre-diagnosis (HR:0.91, CI:0.69–1.21), pre- to post-diagnosis: (HR:1.39, CI:1.12–1.73), post-diagnosis: (HR:0.98, CI:0.70–1.35). Alcohol reduction was not associated with ACM. Among women consuming ≥ 1 drink/day pre-diagnosis, increasing consumption elevated ACM risk (HR:1.83, 95%CI:1.02–3.31). Conclusion: Alcohol reduction or cessation was not associated with a decrease in ACM among older breast cancer survivors. Among those drinking ≥ 1 drink/day, an increase in consumption pre-diagnosis was associated with higher mortality. Findings may be limited by uncontrolled confounding, a low prevalence of heavy drinking, or an insufficient duration of sustained change.

Background: Processed meat and unprocessed red meat are considered carcinogenic and ‘probably carcinogenic’ to humans, respectively, but findings for bladder cancer remain inconsistent. Objectives: We assessed associations between processed meat and unprocessed red meat intake and bladder cancer risk using data from the Pooling Project of Prospective Studies of Diet and Cancer. Methods: Associations with bladder cancer risk were assessed using individual-level data from 30 cohort studies from North America, Europe, Australia and Asia. Diet was assessed at enrolment using validated food frequency questionnaires or diet histories. Following a pre-specified analysis plan, study- and sex-specific hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using Cox regression, adjusted for smoking status and duration and other potential confounders, and pooled using inverse-variance-weighted random-effects meta-analysis. Results: During median follow-up across cohorts of 8–29 years, among 2,575,971 participants, 20,780 were diagnosed with bladder cancer. Processed meat intake was associated with increased bladder cancer risk (HR comparing ≥40 with <5 g/day = 1.08; 95% CI: 1.02–1.15; HR per 20 g/day = 1.02; 95% CI: 1.00–1.03). Weaker evidence of association with increased risk was also observed for unprocessed red meat (HR comparing ≥100 with <20 g/day = 1.06; 95% CI: 1.00–1.13; HR per 50 g/day = 1.01; 95% CI: 0.99–1.03). There was no evidence of heterogeneity by study, sex, or region. Conclusions: Higher intakes of processed meat and unprocessed red meat were associated with a small increase in risk of bladder cancer. These findings are consistent with existing public health recommendations to limit consumption of red and processed meat.

Purpose: Examine which demographic, lifestyle, and clinical risk factors differ between patients with early-onset colorectal cancers (EOCRC) compared to late-onset colorectal cancers (LOCRC). Methods: We conducted a case-case comparison of risk factors and symptoms for EOCRC and LOCRC, utilizing the Ohio Colorectal Cancer Prevention Initiative (OCCPI) data, a statewide study of newly diagnosed CRC among Ohio residents, aged 20–92 years. Unconditional logistic regression (odds ratios (OR) and 95% confidence intervals (CI)) was used to compare risk factors by age-at-diagnoses; those diagnosed < 50 years (EOCRC, N = 288) compared to those diagnosed ≥ 50 years (LOCRC, N = 1,018), adjusting for sex, race/ethnicity, education, smoking status, and family history of CRC. Results: Compared to LOCRC, EOCRC cases had higher odds of ever consuming alcohol (OR = 2.47, CI: 1.55–3.91), consuming more alcohol in their teens/twenties than in other decades (OR = 1.85, CI: 1.36–2.51), and binge drinking (OR = 3.15, CI: 2.31–4.30). EOCRC cases were more likely to have Lynch syndrome (OR = 4.61, CI: 2.72–7.84), and report experiencing pre-diagnostic CRC symptoms (OR = 6.08, CI: 3.77–9.82), including blood in stool (52.3 vs. 30.7%), change in bowel habits (37.2 vs. 19.8%), and bowel obstruction (13.9 vs. 7.7%). Birth weight, inflammatory bowel disease, irritable bowel syndrome, and sex did not differ by age-at-diagnosis. However, when birthweight was compared between the youngest EOCRC to the oldest LOCRC cases (< 40 vs. ≥ 65 years), odds of birthweight of < 6lbs was higher in the youngest cases (OR: 2.15, CI: 0.95–4.87). Conclusion: Alcohol consumption, consuming the most alcohol in one’s teens/twenties, binge drinking, having Lynch syndrome, and pre-diagnostic CRC symptoms were more associated with EOCRC than LOCRC.

Background: This study aimed to investigate the association between alcohol consumption and squamous cell cancers of the upper aerodigestive tract (UADT), using data from 28 cohorts within the Pooling Project of Prospective Studies of Diet and Cancer (DCPP). Methods: Individual-level data from 2 365 437 participants were pooled. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using Cox models to quantify the association between alcohol consumption (g/day) and UADT cancer risk, adjusting for potential confounders. Analyses were conducted by sex, smoking status, geographic region, and alcoholic beverages. Results: Over a median follow-up of 15.5 years, 6903 UADT cancer cases were identified. Alcohol consumption was positively associated with UADT cancer risk overall. Even at intakes as low as 5–<15 g/day, the HR estimate was 1.12 (95% CI = 1.03 to 1.21) compared with the reference group (0.1-<5 g/day). The HR10 g/day (95% CI) was 1.16 (1.14 to 1.18) for women and 1.12 (1.11 to 1.13) for men (Pheterogeneity < .0001). HR10 g/day estimates were 1.14 (1.13 to 1.15) in current, 1.10 (1.09 to 1.12) in former, and 1.15 (1.12 to 1.18) in never smokers. Consistent UADT HR10 g/day estimates were observed across all beverage types. HR10 g/day estimates varied across geographic regions, with HR10 g/day (95% CI) equal to 1.15 (1.14 to 1.17) in Europe–Australia, 1.13 (1.11 to 1.15) in Asia, and 1.11 (1.09 to 1.12) in North America (Pheterogeneity < .0001). Conclusion: Alcohol consumption was associated with UADT cancer risk, irrespective of smoking status or beverage type. However, due to differential baseline risks, alcohol is expected to impact the UADT cancer burden more in smokers than never smokers. These findings support public health strategies to reduce alcohol consumption.

Background: We describe the use of Apisensr, a web-based application that can be used to implement quantitative bias analysis for misclassification, selection bias, and unmeasured confounding. We apply Apisensr using an example of exposure misclassification bias due to use of self-reported body mass index (BMI) to define obesity status in an analysis of the relationship between obesity and diabetes. Methods: We used publicly available data from the National Health and Nutrition Examination Survey. The analysis consisted of: (1) estimating bias parameter values (sensitivity, specificity, negative predictive value, and positive predictive value) for self-reported obesity by sex, age, and race–ethnicity compared to obesity defined by measured BMI, and (2) using Apisensr to adjust for exposure misclassification. Results: The discrepancy between self-reported and measured obesity varied by demographic group (sensitivity range: 75%–89%; specificity range: 91%–99%). Using Apisensr for quantitative bias analysis, there was a clear pattern in the results: the relationship between obesity and diabetes was underestimated using self-report in all age, sex, and race–ethnicity categories compared to measured obesity. For example, in non-Hispanic White men aged 40–59 years, prevalence odds ratios for diabetes were 3.06 (95% confidence inerval = 1.78, 5.30) using self-reported BMI and 4.11 (95% confidence interval = 2.56, 6.75) after bias analysis adjusting for misclassification. Conclusion: Apisensr is an easy-to-use, web-based Shiny app designed to facilitate quantitative bias analysis. Our results also provide estimates of bias parameter values that can be used by other researchers interested in examining obesity defined by self-reported BMI.