A sensitivity analysis using information about measured confounders yielded improved uncertainty assessments for unmeasured confounding

Umbilical cord length and neurodevelopmental disorders, a national cohort study

INTRODUCTION

Adversities in fetal life are known risk factors for neurodevelopmental disorders (NDD). Despite the pivotal role of the umbilical cord, little is known about its associations to later NDD.

OBJECTIVE

To estimate the associations between umbilical cord length and NDD (Attention-deficit/hyperactivity disorder (ADHD), autism spectrum disorder (ASD), intellectual disability (ID), cerebral palsy (CP), epilepsy, impaired vision or hearing), and whether associations differed by sex.

MATERIALS AND METHODS

A prospective population-based cohort study including all liveborn singletons in Norway from 1999, through 2013 and followed up through 2019. Data were retrieved from The Medical Birth Registry of Norway and linked with other national health and administrative registries. Exposures were extreme umbilical cord length (empirical percentile <5th or ≥ 95th percentiles). Main outcome measures were NDD (ADHD, ASD, ID, CP, epilepsy, impaired vision or hearing). Associations with umbilical cord length were assessed using logistic regression.

RESULTS

The cohort consisted of 858,397 births (51.3% boys). We identified 33,370 persons with ADHD (69.8% boys), 10,818 had ASD (76.0% boys), 5538 ID (61.4% boys), 2152 with CP (59.9% boys), 8233 epilepsy (55.0% boys), 900 impaired vision (boys 55.0%), and 11,441 impaired hearing (boys 52.8%). Cord length was positively associated with ADHD (OR 1.15; 95%CI 1.09-1.22), i.e., the risk increased with long cord and decreased with short cord, regardless of sex. A short cord was positively associated with ID (OR 2.42; 95%CI 2.17-2.69), impaired hearing (OR 1.41; 95%CI 1.29-1.54), and epilepsy (OR 1.31; 95%CI 1.18-1.46). CP was associated with both short and long cord (OR 1.31; 95% CI 1.07-1.61 and 1.34, 95%CI 1.13-1.60, respectively). There was no association between cord length and impaired vision.

CONCLUSIONS

This first population study finds that umbilical cord length is associated with NDD. The findings support the hypothesis that neurodevelopment and development of the umbilical cord share pathways.

A Bayesian latent class approach to causal inference with longitudinal data

Bayesian methods are becoming increasingly in demand in clinical and public health comparative effectiveness research. Limited literature has explored parametric Bayesian causal approaches to handle time-dependent treatment and time-dependent covariates. In this article, building on to the work on Bayesian g-computation, we propose a fully Bayesian causal approach, implemented using latent confounder classes which represent the patient's disease and health status. Our setting is suitable when the latent class represents a true disease state that the physician is able to infer without misclassification based on manifest variables. We consider a causal effect that is confounded by the visit-specific latent class in a longitudinal setting and formulate the joint likelihood of the treatment, outcome and latent class models conditionally on the class indicators. The proposed causal structure with latent classes features dimension reduction of time-dependent confounders. We examine the performance of the proposed method using simulation studies and compare the proposed method to other causal methods for longitudinal data with time-dependent treatment and time-dependent confounding. Our approach is illustrated through a study of the effectiveness of intravenous immunoglobulin in treating newly diagnosed juvenile dermatomyositis.

Shoulder dystocia by severity in families: A nationwide population study

INTRODUCTION

Previous studies have established a history of shoulder dystocia as an important risk factor for shoulder dystocia, but studies on shoulder dystocia by severity are scarce. It is unknown if shoulder dystocia tends to be passed on between generations. We aimed to assess the recurrence risk of shoulder dystocia by severity in the same woman and between generations on both the maternal and paternal side. We also assessed the likelihood of a second delivery and planned cesarean section after shoulder dystocia.

MATERIAL AND METHODS

This was a population-based cohort study, using data from the Medical Birth Registry of Norway. To study recurrence in the same mother, we identified 1 091 067 pairs of first and second, second and third, and third and fourth births in the same mother. To study intergenerational recurrence, we identified an individual both as a newborn and as a mother or father in 824 323 mother-offspring pairs and 614 663 father-offspring pairs. We used Bayesian log-binomial multilevel regression to calculate relative risks (RR) with 95% credible intervals.

RESULTS

In subsequent deliveries in the same woman the unadjusted RR of recurrence was 7.05 (95% credible interval 6.39-7.79) and 2.99 (2.71-3.31) after adjusting for possible confounders, including current birthweight. The RRs were higher with severe shoulder dystocia as exposure or outcome. With severe shoulder dystocia as both exposure and outcome, unadjusted and adjusted RR was 20.42 (14.25-29.26) and 6.29 (4.41-8.99), respectively. Women with severe and mild shoulder dystocia and those without had subsequent delivery rates of 71.1, 68.9 and 69.0%, respectively. However, the rates of planned cesarean section in subsequent deliveries for those without shoulder dystocia, mild and severe were 1.3, 5.2 and 16.0%, respectively. On the maternal side the unadjusted inter-generational RR of recurrence was 2.82 (2.25-3.54) and 1.41 (1.05-1.90) on the paternal side. Corresponding adjusted RRs were 1.90 (1.51-2.40) and 1.19 (0.88-1.61), respectively.

CONCLUSIONS

We found a strong recurrence risk of shoulder dystocia, especially severe, in subsequent deliveries in the same woman. The inter-generational recurrence risk was higher on the maternal than paternal side. Women with a history of shoulder dystocia had more often planned cesarean section.

Sensitivity Analyses for Unmeasured Confounders

Purpose of Review

This review expands on sensitivity analyses for unmeasured confounding techniques, demonstrating state-of-the-art methods as well as specifying which should be used under various scenarios, depending on the information about a potential unmeasured confounder available to the researcher.

Recent Findings

Methods to assess how sensitive an observed estimate is to unmeasured confounding have been developed for decades. Recent advancements have allowed for the incorporation of measured confounders in these assessments, updating the methods used to quantify the impact of an unmeasured confounder, whether specified in terms of the magnitude of the effect from a regression standpoint, for example, as a risk ratio, or with respect to the percent of variation in the outcome or exposure explained by the unmeasured confounder. Additionally, single number summaries, such as the E-value or robustness value, have been proposed to allow for ease of computation when less is known about a specific potential unmeasured confounder.

Summary

This paper aimed to provide methods and tools to implement sensitivity to unmeasured confounder analyses appropriate for various research settings depending on what is known or assumed about a potential unmeasured confounder. We have provided mathematical justification, recommendations, as well as R code to ease the implementation of these methods.

Unmeasured confounding in nonrandomized studies: quantitative bias analysis in health technology assessment

Evidence generated from nonrandomized studies (NRS) is increasingly submitted to health technology assessment (HTA) agencies. Unmeasured confounding is a primary concern with this type of evidence, as it may result in biased treatment effect estimates, which has led to much criticism of NRS by HTA agencies. Quantitative bias analyses are a group of methods that have been developed in the epidemiological literature to quantify the impact of unmeasured confounding and adjust effect estimates from NRS. Key considerations for application in HTA proposed in this article reflect the need to balance methodological complexity with ease of application and interpretation, and the need to ensure the methods fit within the existing frameworks used to assess nonrandomized evidence by HTA bodies.

Applying the E Value to Assess the Robustness of Epidemiologic Fields of Inquiry to Unmeasured Confounding

We explored the use of the E value to gauge the robustness of fields of epidemiologic inquiry to unmeasured confounding. We surveyed nutritional and air pollution studies that found statistically significant associations between exposures and incident outcomes. For 100 studies in each field, we extracted adjusted relative effect estimates and associated confidence intervals. We inverted estimates where necessary so that all effects were greater than 1. We calculated E values for both the effect estimate and the lower limit of the 95% confidence interval. Nutritional studies were smaller than air pollution studies (median participants per study, 40,652 vs. 72,460). More than 90% of nutritional studies categorized the exposure, whereas 89% of air pollution studies analyzed the exposure as a continuous variable. The median relative effect was 1.33 in nutrition and 1.16 in air pollution. The corresponding median E values for the estimates were 2.00 and 1.59, respectively. E values for the 95% confidence intervals had median values of 1.39 and 1.26, respectively. Little to moderate unmeasured confounding could explain away most observed associations. The E value is necessarily larger for smaller studies that reach statistical significance, making cross-field comparison difficult. The E value for the 95% confidence interval might be a more useful measure in reports of epidemiologic observational studies.

Trade-offs of Personal Versus More Proxy Exposure Measures in Environmental Epidemiology

The technological ability to make personal measurements of toxicant exposures is growing rapidly. While this can decrease measurement error and therefore help reduce attenuation of effect estimates, we argue that as measures of exposure or dose become more personal, threats to validity of study findings can increase in ways that more proxy measures may avoid. We use directed acyclic graphs (DAGs) to describe conditions where confounding is introduced by use of more personal measures of exposure and avoided via more proxy measures of personal exposure or target tissue dose. As exposure or dose estimates are more removed from the individual, they become less susceptible to biases from confounding by personal factors that can often be hard to control, such as personal behaviors. Similarly, more proxy exposure estimates are less susceptible to reverse causation. We provide examples from the literature where adjustment for personal factors in analyses that use more proxy exposure estimates have little effect on study results. In conclusion, increased personalized exposure assessment has important advantages for measurement accuracy, but it can increase the possibility of biases from personal factors and reverse causation compared with more proxy exposure estimates. Understanding the relation between more and less proxy exposures, and variables that could introduce confounding are critical components to study design.

Bayesian Approaches to Racial Disparities in HIV Risk Estimation Among Men Who Have Sex with Men

BACKGROUND

Men who have sex with men (MSM) continue to be overrepresented for new HIV infections compared with non-MSM. This disparity becomes even more alarming when considering racial groups. We describe the race-specific effects in HIV prevalence among MSM relative to non-MSM and explore the causes of disagreement among estimates.

METHODS

We used data from the National Epidemiologic Survey on Alcohol and Related Conditions, a nationally representative longitudinal survey conducted in the US Bayesian learning corrected for potential misclassification of MSM status and adjusted for residual confounding, hypothesized to explain the MSM racial disparity in HIV. We articulated the structure and strength of the latent confounders that would make race-specific risk gradients equivalent.

RESULTS

Compared with non-MSM, the adjusted prevalence odds ratio (POR) and 95% credible interval for black MSM having self-reported HIV infection was 5.8 (2.0, 16), while the POR for white MSM was 12 (4.2, 31). For all MSM, the POR for HIV infection was 9.3 (3.6, 23) with black men having 2.6 times the odds of prevalent infection compared with white men.

CONCLUSIONS

The observed race-specific associations in MSM are likely not due to misclassification alone, but represent a constellation of factors that differ between racial groups. We recommend specific risk factors in surveys needed to further identify the behavioral characteristics that lead to the observed differences when the estimates are stratified by race.

Propensity Scores in Pharmacoepidemiology: Beyond the Horizon

Purpose of review

Propensity score methods have become commonplace in pharmacoepidemiology over the past decade. Their adoption has confronted formidable obstacles that arise from pharmacoepidemiology's reliance on large healthcare databases of considerable heterogeneity and complexity. These include identifying clinically meaningful samples, defining treatment comparisons, and measuring covariates in ways that respect sound epidemiologic study design. Additional complexities involve correctly modeling treatment decisions in the face of variation in healthcare practice, and dealing with missing information and unmeasured confounding. In this review, we examine the application of propensity score methods in pharmacoepidemiology with particular attention to these and other issues, with an eye towards standards of practice, recent methodological advances, and opportunities for future progress.

Recent findings

Propensity score methods have matured in ways that can advance comparative effectiveness and safety research in pharmacoepidemiology. These include natural extensions for categorical treatments, matching algorithms that can optimize sample size given design constraints, weighting estimators that asymptotically target matched and overlap samples, and the incorporation of machine learning to aid in covariate selection and model building.

Summary

These recent and encouraging advances should be further evaluated through simulation and empirical studies, but nonetheless represent a bright path ahead for the observational study of treatment benefits and harms.

Correction of odds ratios in case-control studies for exposure misclassification with partial knowledge of the degree of agreement among experts who assessed exposures

OBJECTIVES

Estimates of association between exposures and diseases are often distorted by error in exposure classification. When the validity of exposure assessment is known, this can be used to adjust these estimates. When exposure is assessed by experts, even if validity is not known, we sometimes have information about interrater reliability. We present a Bayesian method for translating the knowledge of interrater reliability, which is often available, into knowledge about validity, which is often needed but not directly available, and applying this to correct odds ratios (OR).

METHODS

The method allows for inclusion of observed potential confounders in the analysis, as is common in regression-based control for confounding. Our method uses a novel type of prior on sensitivity and specificity. The approach is illustrated with data from a case-control study of lung cancer risk and occupational exposure to diesel engine emissions, in which exposure assessment was made by detailed job history interviews with study subjects followed by expert judgement.

RESULTS

Using interrater agreement measured by kappas (κ), we estimate sensitivity and specificity of exposure assessment and derive misclassification-corrected confounder-adjusted OR. Misclassification-corrected and confounder-adjusted OR obtained with the most defensible prior had a posterior distribution centre of 1.6 with 95% credible interval (Crl) 1.1 to 2.6. This was on average greater in magnitude than frequentist point estimate of 1.3 (95% Crl 1.0 to 1.7).

CONCLUSIONS

The method yields insights into the degree of exposure misclassification and appears to reduce attenuation bias due to misclassification of exposure while the estimated uncertainty increased.

Identification of confounder in epidemiologic data contaminated by measurement error in covariates

Background

Common methods for confounder identification such as directed acyclic graphs (DAGs), hypothesis testing, or a 10 % change-in-estimate (CIE) criterion for estimated associations may not be applicable due to (a) insufficient knowledge to draw a DAG and (b) when adjustment for a true confounder produces less than 10 % change in observed estimate (e.g. in presence of measurement error).

Methods

We compare previously proposed simulation-based approach for confounder identification that can be tailored to each specific study and contrast it with commonly applied methods (significance criteria with cutoff levels of p-values of 0.05 or 0.20, and CIE criterion with a cutoff of 10 %), as well as newly proposed two-stage procedure aimed at reduction of false positives (specifically, risk factors that are not confounders). The new procedure first evaluates potential for confounding by examination of correlation of covariates and applies simulated CIE criteria only if there is evidence of correlation, while rejecting a covariate as confounder otherwise. These approaches are compared in simulations studies with binary, continuous, and survival outcomes. We illustrate the application of our proposed confounder identification strategy in examining the association of exposure to mercury in relation to depression in the presence of suspected confounding by fish intake using the National Health and Nutrition Examination Survey (NHANES) 2009–2010 data.

Results

Our simulations showed that the simulation-determined cutoff was very sensitive to measurement error in exposure and potential confounder. The analysis of NHANES data demonstrated that if the noise-to-signal ratio (error variance in confounder/variance of confounder) is at or below 0.5, roughly 80 % of the simulated analyses adjusting for fish consumption would correctly result in a null association of mercury and depression, and only an extremely poorly measured confounder is not useful to adjust for in this setting.

Conclusions

No a prior criterion developed for a specific application is guaranteed to be suitable for confounder identification in general. The customization of model-building strategies and study designs through simulations that consider the likely imperfections in the data, as well as finite-sample behavior, would constitute an important improvement on some of the currently prevailing practices in confounder identification and evaluation.

Electronic supplementary material

The online version of this article (doi:10.1186/s12874-016-0159-6) contains supplementary material, which is available to authorized users.

Pre-study feasibility and identifying sensitivity analyses for protocol pre-specification in comparative effectiveness research

The use of healthcare databases for comparative effectiveness research (CER) is increasing exponentially despite its challenges. Researchers must understand their data source and whether outcomes, exposures and confounding factors are captured sufficiently to address the research question. They must also assess whether bias and confounding can be adequately minimized. Many study design characteristics may impact on the results; however, minimal if any sensitivity analyses are typically conducted, and those performed are post hoc. We propose pre-study steps for CER feasibility assessment and to identify sensitivity analyses that might be most important to pre-specify to help ensure that CER produces valid interpretable results.

Evaluating the impact of unmeasured confounding with internal validation data: an example cost evaluation in type 2 diabetes

The quantitative assessment of the potential influence of unmeasured confounders in the analysis of observational data is rare, despite reliance on the "no unmeasured confounders" assumption. In a recent comparison of costs of care between two treatments for type 2 diabetes using a health care claims database, propensity score matching was implemented to adjust for selection bias though it was noted that information on baseline glycemic control was not available for the propensity model. Using data from a linked laboratory file, data on this potential "unmeasured confounder" were obtained for a small subset of the original sample. By using this information, we demonstrate how Bayesian modeling, propensity score calibration, and multiple imputation can utilize this additional information to perform sensitivity analyses to quantitatively assess the potential impact of unmeasured confounding. Bayesian regression models were developed to utilize the internal validation data as informative prior distributions for all parameters, retaining information on the correlation between the confounder and other covariates. While assumptions supporting the use of propensity score calibration were not met in this sample, the use of Bayesian modeling and multiple imputation provided consistent results, suggesting that the lack of data on the unmeasured confounder did not have a strong impact on the original analysis, due to the lack of strong correlation between the confounder and the cost outcome variable. Bayesian modeling with informative priors and multiple imputation may be useful tools for unmeasured confounding sensitivity analysis in these situations. Further research to understand the operating characteristics of these methods in a variety of situations, however, remains.

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.

Opium use and mortality in Golestan Cohort Study: prospective cohort study of 50,000 adults in Iran

OBJECTIVES

To investigate the association between opium use and subsequent risk of death.

DESIGN

Prospective cohort study.

SETTING

The Golestan Cohort Study in north-eastern Iran collected detailed validated data on opium use and other exposures at baseline. Participants were enrolled between January 2004 and June 2008 and were followed to May 2011, with a follow-up success rate of over 99%.

PARTICIPANTS

50,045 participants aged 40-75 at baseline.

MAIN OUTCOMES

Mortality, all cause and major subcategories.

RESULTS

17% (n = 8487) of the participants reported opium use, with a mean duration of 12.7 years. During the follow-up period 2145 deaths were reported. The adjusted hazard ratio for all cause mortality associated with ever use of opium was 1.86 (95% confidence interval 1.68 to 2.06). Opium consumption was significantly associated with increased risks of deaths from several causes including circulatory diseases (hazard ratio 1.81) and cancer (1.61). The strongest associations were seen with deaths from asthma, tuberculosis, and chronic obstructive pulmonary disease (11.0, 6.22, and 5.44, respectively). After exclusion of people who self prescribed opium after the onset of major chronic illnesses, the associations remained strong with a dose-response relation.

CONCLUSION

Opium users have an increased risk of death from multiple causes compared with non-users. Increased risks were also seen in people who used low amounts of opium for a long period and those who had no major illness before use.

Propensity score-based sensitivity analysis method for uncontrolled confounding

The authors developed a sensitivity analysis method to address the issue of uncontrolled confounding in observational studies. In this method, the authors use a 1-dimensional function of the propensity score, which they refer to as the sensitivity function (SF), to quantify the hidden bias due to unmeasured confounders. The propensity score is defined as the conditional probability of being treated given the measured covariates. Then the authors construct SF-corrected inverse-probability-weighted estimators to draw inference on the causal treatment effect. This approach allows analysts to conduct a comprehensive sensitivity analysis in a straightforward manner by varying sensitivity assumptions on both the functional form and the coefficients in the 1-dimensional SF. Furthermore, 1-dimensional continuous functions can be well approximated by low-order polynomial structures (e.g., linear, quadratic). Therefore, even if the imposed SF is practically certain to be incorrect, one can still hope to obtain valuable information on treatment effects by conducting a comprehensive sensitivity analysis using polynomial SFs with varying orders and coefficients. The authors demonstrate the new method by implementing it in an asthma study which evaluates the effect of clinician prescription patterns regarding inhaled corticosteroids for children with persistent asthma on selected clinical outcomes.

Covariate selection in high-dimensional propensity score analyses of treatment effects in small samples

To reduce bias by residual confounding in nonrandomized database studies, the high-dimensional propensity score (hd-PS) algorithm selects and adjusts for previously unmeasured confounders. The authors evaluated whether hd-PS maintains its capabilities in small cohorts that have few exposed patients or few outcome events. In 4 North American pharmacoepidemiologic cohort studies between 1995 and 2005, the authors repeatedly sampled the data to yield increasingly smaller cohorts. They identified potential confounders in each sample and estimated both an hd-PS that included 0-500 covariates and treatment effects adjusted by decile of hd-PS. For sensitivity analyses, they altered the variable selection process to use zero-cell correction and, separately, to use only the variables' exposure association. With >50 exposed patients with an outcome event, hd-PS-adjusted point estimates in the small cohorts were similar to the full-cohort values. With 25-50 exposed events, both sensitivity analyses yielded estimates closer to those obtained in the full data set. Point estimates generally did not change as compared with the full data set when selecting >300 covariates for the hd-PS. In these data, using zero-cell correction or exposure-based covariate selection allowed hd-PS to function robustly with few events. hd-PS is a flexible analytical tool for nonrandomized research across a range of study sizes and event frequencies.

Sensitivity analysis for an unmeasured confounder: a review of two independent methods

One of the main purposes of epidemiological studies is to estimate causal effects. Causal inference should be addressed by observational and experimental studies. A strong constraint for the interpretation of observational studies is the possible presence of unobserved confounders (hidden biases). An approach for assessing the possible effects of unobserved confounders may be drawn up through the use of a sensitivity analysis that determines how strong the effects of an unmeasured confounder should be to explain an apparent association, and which should be the characteristics of this confounder to exhibit such an effect. The purpose of this paper is to review and integrate two independent sensitivity analysis methods. The two methods are presented to assess the impact of an unmeasured confounder variable: one developed by Greenland under an epidemiological perspective, and the other developed from a statistical standpoint by Rosenbaum. By combining (or merging) epidemiological and statistical issues, this integration became a more complete and direct sensitivity analysis, encouraging its required diffusion and additional applications. As observational studies are more subject to biases and confounding than experimental settings, the consideration of epidemiological and statistical aspects in sensitivity analysis strengthens the causal inference.

Probabilistic approaches to better quantifying the results of epidemiologic studies

Typical statistical analysis of epidemiologic data captures uncertainty due to random sampling variation, but ignores more systematic sources of variation such as selection bias, measurement error, and unobserved confounding. Such sources are often only mentioned via qualitative caveats, perhaps under the heading of ‘study limitations.’ Recently, however, there has been considerable interest and advancement in probabilistic methodologies for more integrated statistical analysis. Such techniques hold the promise of replacing a confidence interval reflecting only random sampling variation with an interval reflecting all, or at least more, sources of uncertainty. We survey and appraise the recent literature in this area, giving some prominence to the use of Bayesian statistical methodology.

Covariate balance in a Bayesian propensity score analysis of beta blocker therapy in heart failure patients

Regression adjustment for the propensity score is a statistical method that reduces confounding from measured variables in observational data. A Bayesian propensity score analysis extends this idea by using simultaneous estimation of the propensity scores and the treatment effect. In this article, we conduct an empirical investigation of the performance of Bayesian propensity scores in the context of an observational study of the effectiveness of beta-blocker therapy in heart failure patients. We study the balancing properties of the estimated propensity scores. Traditional Frequentist propensity scores focus attention on balancing covariates that are strongly associated with treatment. In contrast, we demonstrate that Bayesian propensity scores can be used to balance the association between covariates and the outcome. This balancing property has the effect of reducing confounding bias because it reduces the degree to which covariates are outcome risk factors.

Observational epidemiologic studies of nutrition and cancer: the next generation (with better observation)

It would be of enormous public health importance if diet and physical activity, both modifiable behavioral factors, were causally related to cancer. Nevertheless, the nutritional epidemiology of cancer remains problematic, in part because of persistent concerns that standard questionnaires measure diet and physical activity with too much error. We present a new strategy for addressing this measurement error problem. First, as background, we note that food frequency and physical activity questionnaires require respondents to report "typical" diet or activity over the previous year or longer. Multiple 24-hour recalls (24HR), based on reporting only the previous day's behavior, offer potential cognitive advantages over the questionnaires, and biomarker evidence suggests the 24-hour dietary recall is more accurate than the food frequency questionnaire. The expense involved in administering multiple 24 HRs in large epidemiologic studies, however, has up to now been prohibitive. In that context, we suggest that Internet-based 24 HRs, for both diet and physical activity, represent a practical and cost-effective approach for incorporating multiple recalls in large epidemiologic studies. We discuss (1) recent efforts to develop such Internet-based instruments and their accompanying software support systems; (2) ongoing studies to evaluate the feasibility of using these new instruments in cohort studies; (3) additional investigations to gauge the accuracy of the Internet-based recalls vis-à-vis standard instruments and biomarkers; and (4) new statistical approaches for combining the new instruments with standard assessment tools and biomarkers The incorporation of Internet-based 24 HRs into large epidemiologic studies may help advance our understanding of the nutritional determinants of cancer.

Formulating tightest bounds on causal effects in studies with unmeasured confounders

This paper considers the problem of evaluating the causal effect of an exposure on an outcome in observational studies with both measured and unmeasured confounders between the exposure and the outcome. Under such a situation, MacLehose et al. (Epidemiology 2005; 16:548-555) applied linear programming optimization software to find the minimum and maximum possible values of the causal effect for specific numerical data. In this paper, we apply the symbolic Balke-Pearl linear programming method (Probabilistic counterfactuals: semantics, computation, and applications. Ph.D. Thesis, UCLA Cognitive Systems Laboratory, 1995; J. Amer. Statist. Assoc. 1997; 92:1172-1176) to derive the simple closed-form expressions for the lower and upper bounds on causal effects under various assumptions of monotonicity. These universal bounds enable epidemiologists and medical researchers to assess causal effects from observed data with minimum computational effort, and they further shed light on the accuracy of the assessment.