In the analysis of epidemiological studies, categorizing continuous exposures, using either data-driven quantiles or predefined thresholds, is a common practice. Although an abundant literature describes pitfalls of categorization, substantial concerns which have not been adequately addressed arise when the original continuous exposure is measured with error. Measurement error in a continuous exposure leads to misclassification following categorization. This paper aims to dispel five misconceptions regarding the impact of measurement error in a continuous exposure that is categorized. First, that categorization could help infer the functional form of an exposure-outcome association. Second, that misclassification resulting from categorizing a continuous exposure with nondifferential measurement error will also be nondifferential (i.e., independent from the outcome). Third, that categorization will necessarily mitigate the impact of measurement error compared to the continuous case. Fourth, that comparing extreme quantiles instead of comparing adjacent quantiles may reduce bias. Finally, that an estimated association is necessarily attenuated (towards the null), which we argue only holds in regressions with one mismeasured exposure (and may even be reversed in the presence of error-prone confounders). Consequently, epidemiologists who categorize continuous error-prone exposures should be aware of those misconceptions and appropriately discuss the expected impact of categorization on the estimated exposure-outcome associations.

BACKGROUND:Hispanic and Latinos living in the United States bear a disproportionate burden of chronic diseases previously associated with sugars intake. However, due to measurement error (ME) in self-reported diet, these associations have been difficult to establish. OBJECTIVES:To study ME of self-reported total sugars (TSs) intake, as well as its determinates and effect on estimating diet-disease associations in Hispanics and Latinos living in the United States, we apply the 24-h urinary sucrose and fructose predictive biomarker previously confirmed in a feeding study under United States diet. METHODS:We used data from 443 males and females aged 18 to 74 y of Central American, Cuban, Dominican, Mexican, Puerto Rican, and South American origin living across 4 United States communities who participated in the Study of Latinos: Nutrition and Physical Activity Assessment Study (SOLNAS). Participants completed 3 24-h recalls (24HR) and 1 24-h urine collection for measurement of 24-h urinary sucrose and fructose biomarker. Doubly labeled water was administered once to estimate energy intake. Measures were repeated 6 mo later in a reliability sample (n = 81). ME structure in self-reported TSs was investigated using a time-varying intake ME model adjusted for sex, age, and body mass index. RESULTS:The biomarker-based TSs intake was 171.3 g/d (95% confidence interval: 158.3-185.3) compared with 96.3 g/d (95% confidence interval: 91.7-101.2) estimated by 3 24HRs. The correlation of true intake with mean TSs from 2 24HRs as a measure of usual intake over 0 to 12 mo ranged from 0.19 to 0.21. The attenuation factors for the average of 2 24HRs ranged from 0.08 to 0.16. Age and Hispanic and Latino background were significantly associated with TSs misreporting, whereas, age, Hispanic and Latino background, being male and supplement user were associated with misreporting of TSs density. CONCLUSIONS:In this Hispanic and Latino population, we observed large ME in 24HR-measured TSs consumption that would lead to severe attenuation of estimated risk in future sugars-disease association studies. Applying biomarker-based ME correction approaches may alleviate bias.

BACKGROUND:Cancer incidence data collected by cancer registries in the United States and Canada are submitted to the North American Association of Central Cancer Registries, which publishes annual case counts for the two countries. To allow time to collect and report cases, counts for a given diagnosis year are initially published two years after the end of that year and updated annually. Initial counts typically underreport cases compared with updated counts due to reporting delays, potentially biasing estimated incidence trends. METHODS:Existing methods for estimating "delay-adjusted" counts are modified for this heterogeneous group of registries exhibiting different patterns of reporting delay. The new method can be applied to individual registries and combined to produce delay-adjusted rates for the entire population, as well as for geographic or demographic subpopulations. RESULTS:Steps involved in estimating delay-adjusted counts are illustrated for liver and intrahepatic bile duct cancer in White males, in which delay-adjusted rates exhibit a stabilized trend, in contrast to the rapid decline seen in observed (unadjusted) rates. Additionally, the new delay model reveals reporting delays varying across cancer sites, race, and ethnicity. Finally, an extended model provides validated delay-adjusted rates from preliminary data that reduces reporting time from 2 years to 1 year. CONCLUSIONS:Adjusting for reporting delay provides more accurate estimates of cancer incidence trends. The proposed method addresses practical issues of model implementation and continues the evolution of delay adjustment in cancer registries. IMPACT:The new model extends the use of delay adjustment to an important source of cancer surveillance statistics.

Physical activity reduces morbidity and mortality risk in cancer survivors, but a meaningful proportion of this vulnerable population are physically inactive. Targeted interventions can help cancer survivors adopt a more active lifestyle, but the efficacy of these interventions must be rigorously evaluated in randomized controlled intervention trials. A major barrier to such trials involves the difficulty in obtaining unbiased estimates of physical activity in free-living conditions. We conducted a randomized controlled trial of a 3-month intervention designed to increase physical activity vs. usual care in breast cancer survivors (n = 316). The primary outcome was change in physical activity as estimated by hip-worn accelerometer (MTI/Actigraph, models GT1M and GT3X). The trial included a sub-study (n = 106) wherein unbiased measures of total energy expenditure (doubly labeled water), and resting energy expenditure (indirect calorimetry) were collected. A linear mixed measurement error model characterized the structure of measurement error in accelerometry-estimated physical activity energy expenditure (PAEE), and corrected for bias in the estimated intervention effect due to measurement error. Bias in the accelerometer estimates was related to true PAEE (p < 0.001) and baseline body mass index (p < 0.001) but was not related to age (p = 0.13). After correcting for measurement error, the estimated intervention effect at 3 months (change from baseline in PAEE in the intervention arm minus change in the control arm) was 77 kcal/day (95 https://clinicaltrials.gov/study/NCT00929617

Supplementary Table 1. Validation table of delay model

Background: Sodium and potassium measured in 24-h urine collections are often used as reference measurements to validate self-reported dietary intake instruments. Objectives: To evaluate whether collection and analysis of a limited number of urine voids at specified times during the day ("timed voids") can provide alternative reference measurements, and to identify their optimal number and timing. Methods: We used data from a urine calibration study among 441 adults aged 18 - 39 y. Participants collected each urine void in a separate container for 24 h and recorded the collection time. For the same day, they reported dietary intake using a 24-h recall. Urinary sodium and potassium were analyzed in a 24-h composite sample and in 4 timed voids (morning, afternoon, evening, and overnight). Linear regression models were used to develop equations predicting log-transformed 24-h urinary sodium or potassium levels using each of the 4 single timed voids, 6 pairs, and 4 triples. The equations also included age, sex, race, BMI (kg/m(2)), and log creatinine. Optimal combinations minimizing the mean squared prediction error were selected, and the observed and predicted 24-h levels were then used as reference measures to estimate the group bias and attenuation factors of the 24-h dietary recall. These estimates were compared. Results: Optimal combinations found were as follows: single voids - evening; paired voids - afternoon + overnight (sodium) and morning + evening (potassium); and triple voids - morning + evening + overnight (sodium) and morning + afternoon + evening (potassium). Predicted 24-h urinary levels estimated 24-h recall group biases and attenuation factors without apparent bias, but with less precision than observed 24-h urinary levels. To recover lost precision, it was estimated that sample sizes need to be increased by similar to 2.6-2.7 times fora single void, 1.7-2.1 times for paired voids, and 1.5-1.6 times for triple voids. Conclusions: Our results provide the basis for further development of new reference biomarkers based on timed voids. Clinical Trial Registry: clinicaltrials.gov as NCT01631240.

We consider measurement error models for two variables observed repeatedly and subject to measurement error. One variable is continuous, while the other variable is a mixture of continuous and zero measurements. This second variable has two sources of zeros. The first source is episodic zeros, wherein some of the measurements for an individual may be zero and others positive. The second source is hard zeros, i.e., some individuals will always report zero. An example is the consumption of alcohol from alcoholic beverages: some individuals consume alcoholic beverages episodically, while others never consume alcoholic beverages. However, with a small number of repeat measurements from individuals, it is not possible to determine those who are episodic zeros and those who are hard zeros. We develop a new measurement error model for this problem, and use Bayesian methods to fit it. Simulations and data analyses are used to illustrate our methods. Extensions to parametric models and survival analysis are discussed briefly.

Background: Recently, we confirmed 24-h urinary sucrose plus fructose (24 uSF) as a predictive biomarker of total sugar intake. However, the collection of 24-h urine samples has limited feasibility in population studies.Objective: We investigated the utility of the urinary sucrose plus fructose (uSF) biomarker measured in spot urine as a measure of 24 uSF biomarker and total sugar intake. Methods: Hundred participants, 18-70 y of age, from the Phoenix Metropolitan Area completed a 15-d feeding study. For 2 of the 8 collected 24-h urine samples, each spot urine sample was collected in a separate container. We considered 4 timed voids of the day [morning (AM) void: first void 08:30-12:30; afternoon (PM) void: first void 12:31-17:30; evening (EVE) void: first void 17:31-12:00; and next-day (ND) void: first void 04:00-12:00]. We investigated the performance of uSF from 1 void, and uSF combined from 2 and 3 voids as a measure of 24 uSF and sugar intake.Results: The biomarker averaged from PM/EVE void strongly correlated with 24 uSF (partial r = 0.75). The 24 uSF predicted from the PM/ EVE combination was significantly associated with observed sugar intake and was selected for building the calibrated biomarker equation (marginal R2 = 0.36). Spot urine-based calibrated biomarker, ie, biomarker-estimated sugar intake was moderately correlated with the 15-d mean-observed sugar intake (r = 0.50).Conclusions: uSF measured from a PM and EVE void may be used to generate biomarker-based sugar intake estimate when collecting 24-h urine samples is not feasible, pending external validation.

Selecting the number of change points in segmented line regression is an important problem in trend analysis, and there have been various approaches proposed in the literature. We first study the empirical properties of several model selection procedures and propose a new method based on two Schwarz type criteria, a classical Bayes Information Criterion (BIC) and the one with a harsher penalty than BIC (BIC3). The proposed rule is designed to use the former when effect sizes are small and the latter when the effect sizes are large and employs the partial R-2 to determine the weight between BIC and BIC3. The proposed method is computationally much more efficient than the permutation test procedure that has been the default method of Joinpoint software developed for cancer trend analysis, and its satisfactory performance is observed in our simulation study. Simulations indicate that the proposed method performs well in keeping the probability of correct selection at least as large as that of BIC3, whose performance is comparable to that of the permutation test procedure, and improves BIC3 when it performs worse than BIC. The proposed method is applied to the U.S. prostate cancer incidence and mortality rates.

Regression calibration is a popular approach for correcting biases in estimated regression parameters when exposure variables are measured with error. This approach involves building a calibration equation to estimate the value of the unknown true exposure given the error-prone measurement and other covariates. The estimated, or calibrated, exposure is then substituted for the unknown true exposure in the health outcome regression model. When used properly, regression calibration can greatly reduce the bias induced by exposure measurement error. Here, we first provide an overview of the statistical framework for regression calibration, specifically discussing how a special type of error, called Berkson error, arises in the estimated exposure. We then present practical issues to consider when applying regression calibration, including: 1) how to develop the calibration equation and which covariates to include; 2) valid ways to calculate standard errors of estimated regression coefficients; and 3) problems arising if one of the covariates in the calibration model is a mediator of the relationship between the exposure and outcome. Throughout, we provide illustrative examples using data from the Hispanic Community Health Study/Study of Latinos (United States, 2008-2011) and simulations. We conclude with recommendations for how to perform regression calibration.

Supplementary Material. Section A: Excluding incomplete 24-hour urine samples based on PABA Section B: Calibrating the urinary sugars biomarker using a measurement error model that does not include age

PURPOSE:Lymphopenia is associated with poor survival outcomes in head and neck squamous cell carcinoma (HNSCC), yet there is no consensus on whether we should limit lymphopenia risks during treatment. To fully elucidate the prognostic role of baseline versus treatment-related lymphopenia, a robust analysis is necessary to investigate the relative importance of various lymphopenia metrics (LMs) in predicting survival outcomes.METHODS:In this prospective cohort study, 363 patients were eligible for analysis (patients with newly diagnosed, nonmetastatic HNSCC treated with neck radiation with or without chemotherapy in 2015-2019). Data were acquired on 28 covariates: seven baseline, five disease, seven treatment, and nine LMs, including static and time-varying features for absolute lymphocyte count (ALC), neutrophil-to-lymphocyte ratio, and immature granulocytes (IGs). IGs were included, given their hypothesized role in inhibiting lymphocyte function. Overall, there were 4.0% missing data. Median follow-up was 2.9 years. We developed a model (POTOMAC) to predict survival outcomes using a random survival forest (RSF) procedure. RSF uses an ensemble approach to reduce the risk of overfitting and provides internal validation of the model using data that are not used in model development. The ability to predict survival risk was assessed using the AUC for the predicted risk score.RESULTS:POTOMAC predicted 2-year survival with AUCs at 0.78 for overall survival (primary end point) and 0.73 for progression-free survival (secondary end point). Top modifiable risk factors included radiation dose and max ALC decrease. Top baseline risk factors included age, Charlson Comorbidity Index, Karnofsky Performance Score, and baseline IGs. Top-ranking LMs had superior prognostic performance when compared with human papillomavirus status, chemotherapy type, and dose (up to 2, 8, and 65 times higher in variable importance score).CONCLUSION:POTOMAC provides important insights into potential approaches to reduce mortality in patients with HNSCC treated by chemoradiation but needs to be validated in future studies.

Previous studies suggest that amino acid carbon stable isotope ratios (CIRAAs) may serve as biomarkers of added sugar (AS) intake, but this has not been tested in a demographically diverse population. We conducted a 15-day feeding study of U.S. adults, recruited across sex, age, and BMI groups. Participants consumed personalized diets that resembled habitual intake, assessed using two consecutive 7-day food records. We measured serum (n = 99) CIRAAs collected at the end of the feeding period and determined correlations with diet. We used forward selection to model AS intake using participant characteristics and 15 CIRAAs. This model was internally validated using bootstrap optimism correction. Median (25th, 75th percentile) AS intake was 65.2 g/day (44.7, 81.4) and 9.5% (7.2%, 12.4%) of energy. The CIR of alanine had the highest, although modest, correlation with AS intake (r = 0.32, p = 0.001). Serum CIRAAs were more highly correlated with animal food intakes, especially the ratio of animal to total protein. The AS model included sex, body weight and 6 CIRAAs. This model had modest explanatory power (multiple R2 = 0.38), and the optimism-corrected R2 was lower (R2 = 0.15). Further investigations in populations with wider ranges of AS intake are warranted.

BACKGROUND:Twenty-four-hour urinary sucrose and fructose (24uSF) has been studied as a biomarker of total sugars intake in two feeding studies conducted in the United Kingdom (UK) and Arizona (AZ). We compare the biomarker performance in these populations, testing whether it meets the criteria for a predictive biomarker.METHODS:The UK and AZ feeding studies included 13 and 98 participants, respectively, aged 18 to 70 years, consuming their usual diet under controlled conditions. Linear mixed models relating 24uSF to total sugars and personal characteristics were developed in each study and compared. The AZ calibrated biomarker equation was applied to generate biomarker-estimated total sugars intake in UK participants. Stability of the model across AZ study subpopulations was also examined.RESULTS:Model coefficients were similar between the two studies [e.g., log(total sugars): UK 0.99, AZ 1.03, P = 0.67], as was the ratio of calibrated biomarker person-specific bias to between-person variance (UK 0.32, AZ 0.25, P = 0.68). The AZ equation estimated UK log(total sugar intakes) with mean squared prediction error of 0.27, similar to the AZ study estimate (0.28). Within the AZ study, the regression coefficients of log(total sugars) were similar across age, gender, and body mass index subpopulations.CONCLUSIONS:Similar model coefficients in the two studies and good prediction of UK sugar intakes by the AZ equation suggest that 24uSF meets the criteria for a predictive biomarker. Testing the biomarker performance in other populations is advisable.IMPACT:Applications of the 24uSF biomarker will enable improved assessment of the role of sugars intake in risk of chronic disease, including cancer. See related commentary by Prentice, p. 1151.

BACKGROUND:The serum natural abundance carbon isotope ratio (CIR) was recently identified as a candidate biomarker of animal protein intake in postmenopausal women. Such a biomarker would help clarify the relation between dietary protein source (plant or animal) and chronic disease risk. OBJECTIVES:We aimed to evaluate the performance of the serum CIR as a biomarker of dietary protein source in a controlled feeding study of men and women of diverse age and BMI. METHODS:We conducted a 15-d feeding study of 100 adults (age: 18-70 y, 55% women) in Phoenix, AZ. Participants were provided individualized diets that approximated habitual food intakes. Serum was collected at the end of the feeding period for biomarker measurements. RESULTS:Median [IQR] animal protein intake was 67 g/d [55-88 g/d], which was 64% of total protein. The serum CIR was positively correlated with animal protein and inversely correlated with plant protein intake, leading to a strong correlation (r2 = 0.76) with the dietary animal protein ratio (APR; animal/total protein). Regressing serum CIR on the APR, serum nitrogen isotope ratio (NIR), gender, age, and body weight generated an R2 of 0.78. Following the measurement error model for predictive biomarkers, the resulting regression equation was then inverted to develop a calibrated biomarker equation for APR. Added sugars ratio (added/total sugars intake) and corn intakes also influenced the serum CIR but to a much lesser degree than the APR; variations in these intakes had only small effects on biomarker-estimated APR. CONCLUSIONS:Based on our findings in this US cohort of mixed sex and age, we propose the serum CIR alongside NIR as a predictive dietary biomarker of the APR. We anticipate using this biomarker to generate calibrated estimates based on self-reported intake and ultimately to obtain more precise disease risk estimates according to dietary protein source.

Abstract Objectives To evaluate an amino acid carbon stable isotope ratio (CIRAA) biomarker of added sugars (AS) intake in a controlled feeding study of men and women across age and BMI groups. Methods We conducted a 15-d feeding study in Phoenix, AZ, of men and women (N = 100, aged 18–70 y, BMI 17.9–35.0) who were recruited across sex, age, and BMI groups. Participants were provided personalized diets that resembled their habitual intakes, based on 2 consecutive 7-d food records. We measured CIRAAs in serum samples (N = 99) collected at the end of the feeding period and determined correlations with dietary intakes. We used forward selection to construct a model to explain AS intake using participant characteristics and 14 measured CIRAAs. This model was internally validated using a bootstrap optimism correction. Results Median (25th, 75th percentile) AS intake was 65.2 g/d (44.7, 81.4) and 9.5% (7.2%, 12.4%) of energy. The CIR of alanine had the highest, though still modest, correlation with AS intake (Pearson r = 0.32, P = 0.001). Serum CIRAAs were more highly correlated with animal food intakes, especially the ratio of animal to total protein intake (APR). The highest correlations were between the APR and the CIRs of phenylalanine (Pearson r = 0.85, P < 0.001) and leucine (Pearson r = 0.84, P < 0.001). The model of AS intake included participant sex and body weight and the CIRs of 6 AAs: alanine, valine, lysine, glutamic acid, serine, and glycine. This model had modest explanatory power (multiple R2 = 0.38), and the optimism-corrected R2 for the model was lower (R2 = 0.15). Conclusions The observed association between serum CIRAAs and AS intake in the U.S. diet is encouraging; however, further investigation in populations with wider ranges of AS intake is warranted. Funding Sources National Cancer Institute; Institutional Development Award (IDeA) from the National Institutes of General Medical Sciences.

BACKGROUND Developing approaches for the objective assessment of sugars intake in population research is crucial for generating reliable disease risk estimates, and evidence-based dietary guidelines. Twenty-four-hour urinary sucrose and fructose (24uSF) was developed as a predictive biomarker of total sugars intake based on 3 UK feeding studies, yet its performance as a biomarker of total sugars among US participants is unknown. OBJECTIVES To investigate the performance of 24uSF as a biomarker of sugars intake among US participants, and to characterize its use. METHODS Ninety-eight participants, aged 18-70 y, consumed their usual diet under controlled conditions of a feeding study for 15 d, and collected 8 nonconsecutive 24-h urines measured for sucrose and fructose. RESULTS A linear mixed model regressing log 24uSF biomarker on log total sugars intake along with other covariates explained 56% of the biomarker variance. Total sugars intake was the strongest predictor in the model (Marginal R2 = 0.52; P <0.0001), followed by sex (P = 0.0002) and log age (P = 0.002). The equation was then inverted to solve for total sugars intake, thus generating a calibrated biomarker equation. Calibration of the biomarker produced mean biomarker-based log total sugars of 4.79 (SD = 0.59), which was similar to the observed log 15-d mean total sugars intake of 4.69 (0.35). The correlation between calibrated biomarker and usual total sugars intake was 0.59 for the calibrated biomarker based on a single biomarker measurement, and 0.76 based on 4 biomarker repeats spaced far apart. CONCLUSIONS In this controlled feeding study, total sugars intake was the main determinant of 24uSF confirming its utility as a biomarker of total sugars in this population. Next steps will include validation of stability assumptions of the biomarker calibration equation proposed here, which will allow its use as an instrument for dietary validation and measurement error correction in diet-disease association studies.