Application of a Web-based Tool for Quantitative Bias Analysis: The Example of Misclassification Due to Self-reported Body Mass Index.

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.

The association between reproductive history and abdominal adipose tissue among postmenopausal women: results from the Women's Health Initiative.

STUDY QUESTION: What is the association between reproductive health history (e.g. age at menarche, menopause, reproductive lifespan) with abdominal adiposity in postmenopausal women? SUMMARY ANSWER: Higher visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) tissue levels were observed among women with earlier menarche, earlier menopause, and greater parity. WHAT IS KNOWN ALREADY: Postmenopausal women are predisposed to accumulation of VAT and SAT. Reproductive health variables are known predictors of overall obesity status in women, defined by BMI. STUDY DESIGN, SIZE, DURATION: This study is a secondary analysis of data collected from the baseline visit of the Women's Health Initiative (WHI). The WHI is a large prospective study of postmenopausal women, including both a randomized trial and observational study. There were 10 184 women included in this analysis. PARTICIPANTS/MATERIALS, SETTING, METHODS: Data were collected from a reproductive health history questionnaire, dual-energy x-ray absorptiometry scans, and anthropometric measures at WHI baseline. Reproductive history was measured via self-report, and included age at menarche, variables related to pregnancy, and age at menopause. Reproductive lifespan was calculated as age at menopause minus age at menarche. Statistical analyses included descriptive analyses and multivariable linear regression models to examine the association between reproductive history with VAT, SAT, total body fat, and BMI. MAIN RESULTS AND THE ROLE OF CHANCE: Women who reported early menarche (<10 years) or early menopause (<40 years) had the highest levels of VAT. Adjusted multivariable linear regression results demonstrate women who experienced menarche >15 years had 23 cm2 less VAT (95% CI: -31.4, -14.4) and 47 cm2 less SAT (95% CI: -61.8, -33.4) than women who experienced menarche at age 10 years or earlier. A similar pattern was observed for age at menopause: compared to women who experienced menopause <40 years, menopause at 50-55 years was associated with 19.3 cm2 (95% CI: -25.4, -13.3) less VAT and 27.4 cm2 (-29.6, 10.3) less SAT. High parity (>3 pregnancies) was also associated with VAT and SAT. For example, adjusted beta coefficients for VAT were 8.36 (4.33, 12.4) and 17.9 (12.6, 23.2) comparing three to four pregnancies with the referent, one to two pregnancies. LIMITATIONS, REASONS FOR CAUTION: The WHI reproductive health history questionnaire may be subject to poor recall owing to a long look-back window. Residual confounding may be present given lack of data on early life characteristics, such as maternal and pre-menarche characteristics. WIDER IMPLICATIONS OF THE FINDINGS: This study contributes to our understanding of reproductive lifespan, including menarche and menopause, as an important predictor of late-life adiposity in women. Reproductive health has also been recognized as a sentinel marker for chronic disease in late life. Given established links between adiposity and cardiometabolic outcomes, this research has implications for future research, clinical practice, and public health policy that makes use of reproductive health history as an opportunity for chronic disease prevention. STUDY FUNDING/COMPETING INTEREST(S): HRB and AOO are supported by the National Institute of Health National Institute of Aging (R01AG055018-04). JWB reports royalties from 'ACSM'S Body Composition Assessment Book' and consulting fees from the WHI. The remaining authors have no competing interests to declare. TRIAL REGISTRATION NUMBER: N/A.

Longitudinal patterns of abdominal visceral and subcutaneous adipose tissue, total body composition, and anthropometric measures in postmenopausal women: Results from the Women's Health Initiative.

BACKGROUND: Abdominal adiposity, including visceral and subcutaneous abdominal adipose tissue (VAT and SAT), is recognized as a strong risk factor for cardiometabolic disease, cancer, and mortality. OBJECTIVE: The primary aim of this analysis is to describe longitudinal patterns of change in abdominal adipose tissue in postmenopausal women, overall and stratified by age, race/ethnicity, and years since menopause. METHODS: The data are from six years of follow up on 10,184 postmenopausal women (7828 non-Hispanic White women, 1423 non-Hispanic Black women, and 703 Hispanic women) who participated in the Women's Health Initiative (WHI). The WHI is a large prospective cohort study of postmenopausal women across the United States. All participants in this analysis had DXA scans in the 1990s as part of the WHI protocol. Hologic APEX software was used to re-analyze archived DXA scans and obtain measures of abdominal adipose tissue. Analyses examined differences in abdominal adipose tissue, overall adiposity, and anthropometric variables. RESULTS: There were important differences in VAT and SAT by age and race/ethnicity. In women <60 years, VAT increased over the follow-up period, while in women ≥70 years, VAT decreased. Non-Hispanic Black women had the highest levels of SAT. Hispanic women had the highest VAT levels. Women more than ten years since menopause had less SAT and more VAT than women less than ten years since menopause, resulting in a higher VAT/SAT ratio. There was a moderate to strong correlation between measures of abdominal adipose tissue and anthropometric measurements of body size. Still, there were substantial differences in the quantity of VAT and SAT within BMI and waist circumference categories. CONCLUSIONS: These results demonstrate differences in VAT and SAT according to age, race/ethnicity, time since menopause, and compared to standard measures of body composition in a large and diverse cohort of postmenopausal women.

Monte Carlo Simulation Approaches for Quantitative Bias Analysis: A Tutorial.

Quantitative bias analysis can be used to empirically assess how far study estimates are from the truth (i.e., an estimate that is free of bias). These methods can be used to explore the potential impact of confounding bias, selection bias (collider stratification bias), and information bias. Quantitative bias analysis includes methods that can be used to check the robustness of study findings to multiple types of bias and methods that use simulation studies to generate data and understand the hypothetical impact of specific types of bias in a simulated data set. In this article, we review 2 strategies for quantitative bias analysis: 1) traditional probabilistic quantitative bias analysis and 2) quantitative bias analysis with generated data. An important difference between the 2 strategies relates to the type of data (real vs. generated data) used in the analysis. Monte Carlo simulations are used in both approaches, but the simulation process is used for different purposes in each. For both approaches, we outline and describe the steps required to carry out the quantitative bias analysis and also present a bias-analysis tutorial demonstrating how both approaches can be applied in the context of an analysis for selection bias. Our goal is to highlight the utility of quantitative bias analysis for practicing epidemiologists and increase the use of these methods in the epidemiologic literature.

Propensity score methods for merging observational and experimental datasets.

We consider how to merge a limited amount of data from a randomized controlled trial (RCT) into a much larger set of data from an observational data base (ODB), to estimate an average causal treatment effect. Our methods are based on stratification. The strata are defined in terms of effect moderators as well as propensity scores estimated in the ODB. Data from the RCT are placed into the strata they would have occupied, had they been in the ODB instead. We assume that treatment differences are comparable in the two data sources. Our first "spiked-in" method simply inserts the RCT data into their corresponding ODB strata. We also consider a data-driven convex combination of the ODB and RCT treatment effect estimates within each stratum. Using the delta method and simulations, we identify a bias problem with the spiked-in estimator that is ameliorated by the convex combination estimator. We apply our methods to data from the Women's Health Initiative, a study of thousands of postmenopausal women which has both observational and experimental data on hormone therapy (HT). Using half of the RCT to define a gold standard, we find that a version of the spiked-in estimator yields lower-MSE estimates of the causal impact of HT on coronary heart disease than would be achieved using either a small RCT or the observational component on its own.

MRI Based Validation of Abdominal Adipose Tissue Measurements From DXA in Postmenopausal Women.

INTRODUCTION: Visceral adipose tissue (VAT) is a hypothesized driver of chronic disease. Dual-energy X-ray absorptiometry (DXA) potentially offers a lower cost and more available alternative compared to gold-standard magnetic resonance imaging (MRI) for quantification of abdominal fat sub-compartments, VAT and subcutaneous adipose tissue (SAT). We sought to validate VAT and SAT area (cm2) from historical DXA scans against MRI. METHODOLOGY: Participants (n = 69) from the Women's Health Initiative (WHI) completed a 3 T MRI scan and a whole body DXA scan (Hologic QDR2000 or QDR4500; 2004-2005). A subset of 43 participants were scanned on both DXA devices. DXA-derived VAT and SAT at the 4th lumbar vertebrae (5 cm wide) were analyzed using APEX software (v4.0, Hologic, Inc., Marlborough, MA). MRI VAT and SAT areas for the corresponding DXA region of interest were quantified using sliceOmatic software (v5.0, Tomovision, Magog, Canada). Pearson correlations between MRI and DXA-derived VAT and SAT were computed, and a Bland-Altman analysis was performed. RESULTS: Participants were primarily non-Hispanic white (86%) with a mean age of 70.51 ± 5.79 years and a mean BMI of 27.33 ± 5.40 kg/m2. Correlations between MRI and DXA measured VAT and SAT were 0.90 and 0.92, respectively (p ≤ 0.001). Bland-Altman plots showed that DXA-VAT slightly overestimated VAT on the QDR4500 (-3.31 cm2); this bias was greater in the smaller subset measured on the older DXA model (QDR2000; -30.71 cm2). The overestimation of DXA-SAT was large (-85.16 to -118.66 cm2), but differences were relatively uniform for the QDR4500. CONCLUSIONS: New software applied to historic Hologic DXA scans provide estimates of VAT and SAT that are well-correlated with criterion MRI among postmenopausal women.

Teaching Epidemiology Online (Pandemic Edition).

In response to the threat posed by the coronavirus disease 2019 (COVID-19) pandemic, many universities are encouraging or requiring online instruction. Teaching an epidemiology course online is different in many respects from teaching in person. In this article, we review specific approaches and strategies related to teaching epidemiology online during the pandemic and beyond, including a discussion of options for course format, grading and assessment approaches, pandemic-related contingencies, and the use of technology. Throughout this article we present practical, epidemiology-specific teaching examples. Moreover, we also examine 1) how the lessons learned about the practice of epidemiology during the pandemic can be integrated into the didactic content of epidemiology training programs and 2) whether epidemiologic pedagogy and teaching strategies should change in the long term, beyond the COVID-19 pandemic. The pandemic has served to heighten our awareness of concerns related to student health and safety, as well as issues of accessibility, equity, and inclusion. Our goal is to present a practical overview connecting pandemic-era online teaching with thoughts about the future of epidemiologic instruction.

Will Podcasting and Social Media Replace Journals and Traditional Science Communication? No, but...

The digital world in which we live is changing rapidly. The evolving media environment is having a direct impact on traditional forms of communication and knowledge translation in public health and epidemiology. Openly accessible digital media can be used to reach a broader and more diverse audience of trainees, scientists, and the lay public than can traditional forms of scientific communication. The new digital landscape for delivering content is vast, and new platforms are continuously being added. In this article, we focus on several, including Twitter and podcasting, and discuss their relevance to epidemiology and science communication. We highlight 3 key reasons why we think epidemiologists should be engaging with these mediums: 1) science communication, 2) career advancement, and 3) development of a community and public service. Other positive and negative consequences of engaging in these forms of new media are also discussed. The authors of this commentary are all engaged in social media and podcasting for scientific communication, and we reflect on our experiences with these mediums as tools to advance the field of epidemiology.

Cardiometabolic risk factors and survival after cancer in the Women's Health Initiative.

BACKGROUND: Cardiometabolic abnormalities are a leading cause of death among women, including women with cancer. METHODS: This study examined the association between prediagnosis cardiovascular health and total and cause-specific mortality among 12,076 postmenopausal women who developed local- or regional-stage invasive cancer in the Women's Health Initiative (WHI). Cardiovascular risk factors included waist circumference, hypertension, high cholesterol, and type 2 diabetes. Obesity-related cancers included breast cancer, colorectal cancer, endometrial cancer, kidney cancer, pancreatic cancer, ovarian cancer, stomach cancer, liver cancer, and non-Hodgkin lymphoma. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) adjusted for important predictors of survival. RESULTS: After a median follow-up of 10.0 years from the date of the cancer diagnosis, there were 3607 total deaths, with 1546 (43%) due to cancer. Most participants (62.9%) had 1 or 2 cardiometabolic risk factors, and 8.1% had 3 or 4. In adjusted models, women with 3 to 4 risk factors (vs none) had a higher risk of all-cause mortality (HR, 1.99; 95% CI, 1.73-2.30), death due to cardiovascular disease (CVD) (HR, 4.01; 95% CI, 2.88-5.57), cancer-specific mortality (HR, 1.37; 95% CI, 1.1-1.72), and other-cause mortality (HR, 2.14; 95% CI, 1.70-2.69). A higher waist circumference was associated with greater all-cause mortality (HR, 1.17; 95% CI, 1.06-1.30) and cancer-specific mortality (HR, 1.22; 95% CI, 1.04-1.42). CONCLUSIONS: Among postmenopausal women diagnosed with cancer in the WHI, cardiometabolic risk factors before the cancer diagnosis were associated with greater all-cause, CVD, cancer-specific, and other-cause mortality. These results raise hypotheses regarding potential clinical intervention strategies targeting cardiometabolic abnormalities that require future prospective studies for confirmation. LAY SUMMARY: This study uses information from the Women's Health Initiative (WHI) to find out whether cardiac risk factors are related to a greater risk of dying among older women with cancer. The WHI is the largest study of medical problems faced by older women in this country. The results show that women who have 3 or 4 risk factors are more likely to die of any cause, heart disease, or cancer in comparison with women with no risk factors. It is concluded that interventions to help to lower the burden of cardiac risk factors can have an important impact on survivorship among women with cancer.

Association between regional body fat and cardiovascular disease risk among postmenopausal women with normal body mass index.

AIMS: Central adiposity is associated with increased cardiovascular disease (CVD) risk, even among people with normal body mass index (BMI). We tested the hypothesis that regional body fat deposits (trunk or leg fat) are associated with altered risk of CVD among postmenopausal women with normal BMI. METHODS AND RESULTS: We included 2683 postmenopausal women with normal BMI (18.5 to <25 kg/m2) who participated in the Women's Health Initiative and had no known CVD at baseline. Body composition was determined by dual energy X-ray absorptiometry. Incident CVD events including coronary heart disease and stroke were ascertained through February 2017. During a median 17.9 years of follow-up, 291 incident CVD cases occurred. After adjustment for demographic, lifestyle, and clinical risk factors, neither whole-body fat mass nor fat percentage was associated with CVD risk. Higher percent trunk fat was associated with increased risk of CVD [highest vs. lowest quartile hazard ratio (HR) = 1.91, 95% confidence interval (CI) 1.33-2.74; P-trend <0.001], whereas higher percent leg fat was associated with decreased risk of CVD (highest vs. lowest quartile HR = 0.62, 95% CI 0.43-0.89; P-trend = 0.008). The association for trunk fat was attenuated yet remained significant after further adjustment for waist circumference or waist-to-hip ratio. Higher percent trunk fat combined with lower percent leg fat was associated with particularly high risk of CVD (HR comparing extreme groups = 3.33, 95% CI 1.46-7.62). CONCLUSION: Among postmenopausal women with normal BMI, both elevated trunk fat and reduced leg fat are associated with increased risk of CVD.

You Can't Drive a Car With Only Three Wheels.

Authors aiming to estimate causal effects from observational data frequently discuss 3 fundamental identifiability assumptions for causal inference: exchangeability, consistency, and positivity. However, too often, studies fail to acknowledge the importance of measurement bias in causal inference. In the presence of measurement bias, the aforementioned identifiability conditions are not sufficient to estimate a causal effect. The most fundamental requirement for estimating a causal effect is knowing who is truly exposed and unexposed. In this issue of the Journal, Caniglia et al. (Am J Epidemiol. 2019;000(00):000-000) present a thorough discussion of methodological challenges when estimating causal effects in the context of research on distance to obstetrical care. Their article highlights empirical strategies for examining nonexchangeability due to unmeasured confounding and selection bias and potential violations of the consistency assumption. In addition to the important considerations outlined by Caniglia et al., authors interested in estimating causal effects from observational data should also consider implementing quantitative strategies to examine the impact of misclassification. The objective of this commentary is to emphasize that you can't drive a car with only three wheels, and you also cannot estimate a causal effect in the presence of exposure misclassification bias.

The Effects of Reverse Causality and Selective Attrition on the Relationship Between Body Mass Index and Mortality in Postmenopausal Women.

Concerns about reverse causality and selection bias complicate the interpretation of studies of body mass index (BMI, calculated as weight (kg)/height (m)2) and mortality in older adults. The objective of this study was to investigate methodological explanations for the apparent attenuation of obesity-related risks in older adults. We used data from 68,132 participants in the Women's Health Initiative (WHI) clinical trial for this analysis. All of the participants were postmenopausal women aged 50-79 years at baseline (1993-1998). To examine reverse causality and selective attrition, we compared rate ratios from inverse probability of treatment- and censoring-weighted Poisson marginal structural models with results from an unweighted adjusted Poisson regression model. The estimated mortality rate ratios and 95% confidence intervals for BMIs of 30.0-34.9, 35.0-39.9 and ≥40.0 were 0.86 (95% confidence interval (CI): 0.77, 0.96), 0.85 (95% CI: 0.72, 0.99), and 0.88 (95% CI: 0.72, 1.07), respectively, in the unweighted model. The corresponding mortality rate ratios were 0.96 (95% CI: 0.86, 1.07), 1.12 (95% CI: 0.97, 1.29), and 1.31 95% CI: (1.08, 1.57), respectively, in the marginal structural model. Results from the inverse probability of treatment- and censoring-weighted marginal structural model were attenuated in low BMI categories and increased in high BMI categories. The results demonstrate the importance of accounting for reverse causality and selective attrition in studies of older adults.

Risk Factor Reversal in Studies of Infectious Disease: Making Counterintuitive Results Intuitive Again.

A previously published study reported the seemingly paradoxical finding that men who have sex with men status was strongly protective and recent sexual abstinence strongly deleterious in relation to mortality prognosis. We explain why these results are entirely logical and that the counterintuitive direction of the effects derives from the comparison group implied by the study design.

Composition and diversity of the subgingival microbiome and its relationship with age in postmenopausal women: an epidemiologic investigation.

BACKGROUND: The extent to which the composition and diversity of the oral microbiome varies with age is not clearly understood. METHODS: The 16S rRNA gene of subgingival plaque in 1219 women, aged 53-81 years, was sequenced and its taxonomy annotated against the Human Oral Microbiome Database (v.14.5). Composition of the subgingival microbiome was described in terms of centered log(2)-ratio (CLR) transformed OTU values, relative abundance, and prevalence. Correlations between microbiota abundance and age were evelauted using Pearson Product Moment correlations. P-values were corrected for multiple testing using the Bonferroni method. RESULTS: Of the 267 species identified overall, Veillonella dispar was the most abundant bacteria when described by CLR OTU (mean 8.3) or relative abundance (mean 8.9%); whereas Streptococcus oralis, Veillonella dispar and Veillonella parvula were most prevalent (100%, all) when described as being present at any amount. Linear correlations between age and several CLR OTUs (Pearson r = - 0.18 to 0.18), of which 82 (31%) achieved statistical significance (P < 0.05). The correlations lost significance following Bonferroni correction. Twelve species that differed across age groups (each corrected P < 0.05); 5 (42%) were higher in women ages 50-59 compared to ≥70 (corrected P < 0.05), and 7 (48%) were higher in women 70 years and older. CONCLUSIONS: We identified associations between several bacterial species and age across the age range of postmenopausal women studied. Understanding the functions of these bacteria could identify intervention targets to enhance oral health in later life.

Can Survival Bias Explain the Age Attenuation of Racial Inequalities in Stroke Incidence?: A Simulation Study.

BACKGROUND: In middle age, stroke incidence is higher among black than white Americans. For unknown reasons, this inequality decreases and reverses with age. We conducted simulations to evaluate whether selective survival could account for observed age patterning of black-white stroke inequalities. METHODS: We simulated birth cohorts of 20,000 blacks and 20,000 whites with survival distributions based on US life tables for the 1919-1921 birth cohort. We generated stroke incidence rates for ages 45-94 years using Reasons for Geographic and Racial Disparities in Stroke (REGARDS) study rates for whites and setting the effect of black race on stroke to incidence rate difference (IRD) = 20/10,000 person-years at all ages, the inequality observed at younger ages in REGARDS. We compared observed age-specific stroke incidence across scenarios, varying effects of U, representing unobserved factors influencing mortality and stroke risk. RESULTS: Despite a constant adverse effect of black race on stroke risk, the observed black-white inequality in stroke incidence attenuated at older age. When the hazard ratio for U on stroke was 1.5 for both blacks and whites, but U only directly influenced mortality for blacks (hazard ratio for U on mortality =1.5 for blacks; 1.0 for whites), stroke incidence rates in late life were lower among blacks (average observed IRD = -43/10,000 person-years at ages 85-94 years versus causal IRD = 20/10,000 person-years) and mirrored patterns observed in REGARDS. CONCLUSIONS: A relatively moderate unmeasured common cause of stroke and survival could fully account for observed age attenuation of racial inequalities in stroke.

Stratified Probabilistic Bias Analysis for Body Mass Index-related Exposure Misclassification in Postmenopausal Women.

BACKGROUND: There is widespread concern about the use of body mass index (BMI) to define obesity status in postmenopausal women because it may not accurately represent an individual's true obesity status. The objective of the present study is to examine and adjust for exposure misclassification bias from using an indirect measure of obesity (BMI) compared with a direct measure of obesity (percent body fat). METHODS: We used data from postmenopausal non-Hispanic black and non-Hispanic white women in the Women's Health Initiative (n=126,459). Within the Women's Health Initiative, a sample of 11,018 women were invited to participate in a sub-study involving dual-energy x-ray absorptiometry scans. We examined indices of validity comparing BMI-defined obesity (≥30 kg/m), with obesity defined by percent body fat. We then used probabilistic bias analysis models stratified by age and race to explore the effect of exposure misclassification on the obesity-mortality relationship. RESULTS: Validation analyses highlight that using a BMI cutpoint of 30 kg/m to define obesity in postmenopausal women is associated with poor validity. There were notable differences in sensitivity by age and race. Results from the stratified bias analysis demonstrated that failing to adjust for exposure misclassification bias results in attenuated estimates of the obesity-mortality relationship. For example, in non-Hispanic white women 50-59 years of age, the conventional risk difference was 0.017 (95% confidence interval = 0.01, 0.023) and the bias-adjusted risk difference was 0.035 (95% simulation interval = 0.028, 0.043). CONCLUSIONS: These results demonstrate the importance of using quantitative bias analysis techniques to account for nondifferential exposure misclassification of BMI-defined obesity. See video abstract at, http://links.lww.com/EDE/B385.

Methodological considerations for disentangling a risk factor's influence on disease incidence versus postdiagnosis survival: The example of obesity and breast and colorectal cancer mortality in the Women's Health Initiative.

Often, studies modeling an exposure's influence on time to disease-specific death from study enrollment are incorrectly interpreted as if based on time to death from disease diagnosis. We studied 151,996 postmenopausal women without breast or colorectal cancer in the Women's Health Initiative with weight and height measured at enrollment (1993-1998). Using Cox regression models, we contrast hazard ratios (HR) from two time-scales and corresponding study subpopulations: time to cancer death after enrollment among all women and time to cancer death after diagnosis among only cancer survivors. Median follow-up from enrollment to diagnosis/censoring was 13 years for both breast (7,633 cases) and colorectal cancer (2,290 cases). Median follow-up from diagnosis to death/censoring was 7 years for breast and 5 years for colorectal cancer. In analyses of time from enrollment to death, body mass index (BMI) ≥ 35 kg/m2 versus 18.5-<25 kg/m2 was associated with higher rates of cancer mortality: HR = 1.99; 95% CI: 1.54, 2.56 for breast cancer (p trend <0.001) and HR = 1.40; 95% CI: 1.04, 1.88 for colorectal cancer (p trend = 0.05). However, in analyses of time from diagnosis to cancer death, trends indicated no significant association (for BMI ≥ 35 kg/m2 , HR = 1.25; 95% CI: 0.94, 1.67 for breast [p trend = 0.33] and HR = 1.18; 95% CI: 0.84, 1.86 for colorectal cancer [p trend = 0.39]). We conclude that a risk factor that increases disease incidence will increase disease-specific mortality. Yet, its influence on postdiagnosis survival can vary, and requires consideration of additional design and analysis issues such as selection bias. Quantitative tools allow joint modeling to compare an exposure's influence on time from enrollment to disease incidence and time from diagnosis to death.