Nonparametric Predictive Inference for Exposure Assessment

Exposure assessment for food and drink consumption requires the combining of information about people's consumption of products with concentration data sets to provide predictions for chemical intake by humans. In this article, we present a method called nonparametric predictive inference (NPI) for exposure assessment. NPI is a distribution‐free method relying only on Hill's assumption . Effectively, is a postdata exchangeability assumption, which is a natural starting point for nonparametric statistics. For further discussion we refer to works by Hill and Coolen. We illustrate how NPI can be implemented to produce predictions for an individual's exposure based on consumption, body weight, and concentration data. NPI has the advantage that we do not have to assume a distribution to implement it. There may, however, be information available to suggest a distribution for a random quantity. Therefore, we present an NPI‐Bayes hybrid method where this information can be taken into account by using Bayesian methods while using NPI for the other random quantities in the model.     

Estimation of Dietary Exposure to Chemicals: A Case Study Illustrating Methods of Distributional Analyses for Food Consumption Data

There are a number of sources of variability in food consumption patterns and residue levels of a particular chemical (e.g., pesticide, food additive) in commodities that lead to an expected high level of variability in dietary exposures across a population. This paper focuses on examples of consumption pattern survey data for specific commodities, namely that for wine and grape juice, and demonstrates how such data might be analyzed in preparation for performing stochastic analyses of dietary exposure. Data from the NIAAA/NHIS wine consumption survey were subset for gender and age group and, with matched body weight data from the survey database, were used to define empirically-based percentile estimates for wine intake (μl wine/kg body weight) for the strata of interest. The data for these two subpopulations were analyzed to estimate 14-day consumption distributional statistics and distributions for only those days on which wine was consumed. Data subsets for all wine-consuming adults and wine-consuming females ages 18 through 45, were determined to fit a lognormal distribution ( R ²= 0.99 for both datasets). Market share data were incorporated into estimation of chronic exposures to hypothetical chemical residues in imported table wine. As a separate example, treatment of grape juice consumption data for females, ages 18–40, as a simple lognormal distribution resulted in a significant underestimation of intake, and thus exposure, because the actual distribution is a mixture (i.e., multiple subpopulations of grape juice consumers exist in the parent distribution). Thus, deriving dietary intake statistics from food consumption survey data requires careful analysis of the underlying empirical distributions.     

Foodborne Exposure to Pesticides and Methylmercury in the United Arab Emirates

As part of a comprehensive environmental health strategic planning project initiated by the government of Abu Dhabi, we assessed potential dietary exposure in the United Arab Emirates (UAE) to methylmercury (in seafood) and pesticides (in fruits and vegetables) above international guideline levels. We present results for the UAE population by age, gender, and body mass index. Our results show very low daily risks of exposure to pesticides in fruits and vegetables at levels exceeding WHO guidelines even under the conservative assumption that no pesticides are removed during washing and food preparation. Thus, exposure to pesticides on fruits and vegetables does not appear to be a major public health concern in the UAE. The chances of exposure to methylmercury in seafood are much higher; our model estimates a mean 1 in 5 daily risk of exceeding the FAO/WHO provisional tolerable weekly intake. However, great caution should be used in interpreting these results, as we analyzed only the risks and not the substantial benefits of fish consumption. In fact, previous studies have demonstrated that exposure to the n‐3 polyunsaturated fatty acids in fish can increase IQ in developing children, and it can substantially decrease the risk in adults of coronary heart disease and stroke. Further research is warranted to compare the risk of Me‐Hg exposure from fish to the nutritional benefits of fish consumption in the UAE and to determine appropriate methods to communicate risk and benefit information to the UAE population.     

Health Risk Characterization of Chlorpyrifos Using Epidemiological Dose‐Response Data and Probabilistic Techniques: A Case Study with Rice Farmers in Vietnam

Various methods for risk characterization have been developed using probabilistic approaches. Data on Vietnamese farmers are available for the comparison of outcomes for risk characterization using different probabilistic methods. This article addresses the health risk characterization of chlorpyrifos using epidemiological dose‐response data and probabilistic techniques obtained from a case study with rice farmers in Vietnam. Urine samples were collected from farmers and analyzed for trichloropyridinol (TCP), which was converted into absorbed daily dose of chlorpyrifos. Adverse health response doses due to chlorpyrifos exposure were collected from epidemiological studies to develop dose‐adverse health response relationships. The health risk of chlorpyrifos was quantified using hazard quotient (HQ), Monte Carlo simulation (MCS), and overall risk probability (ORP) methods. With baseline (prior to pesticide spraying) and lifetime exposure levels (over a lifetime of pesticide spraying events), the HQ ranged from 0.06 to 7.1. The MCS method indicated less than 0.05% of the population would be affected while the ORP method indicated that less than 1.5% of the population would be adversely affected. With postapplication exposure levels, the HQ ranged from 1 to 32.5. The risk calculated by the MCS method was that 29% of the population would be affected, and the risk calculated by ORP method was 33%. The MCS and ORP methods have advantages in risk characterization due to use of the full distribution of data exposure as well as dose response, whereas HQ methods only used the exposure data distribution. These evaluations indicated that single‐event spraying is likely to have adverse effects on Vietnamese rice farmers.     

Feasibility Study on the Use of Probabilistic Migration Modeling in Support of Exposure Assessment from Food Contact Materials

The use of probabilistic approaches in exposure assessments of contaminants migrating from food packages is of increasing interest but the lack of concentration or migration data is often referred as a limitation. Data accounting for the variability and uncertainty that can be expected in migration, for example, due to heterogeneity in the packaging system, variation of the temperature along the distribution chain, and different time of consumption of each individual package, are required for probabilistic analysis. The objective of this work was to characterize quantitatively the uncertainty and variability in estimates of migration. A Monte Carlo simulation was applied to a typical solution of the Fick's law with given variability in the input parameters. The analysis was performed based on experimental data of a model system (migration of Irgafos 168 from polyethylene into isooctane) and illustrates how important sources of variability and uncertainty can be identified in order to refine analyses. For long migration times and controlled conditions of temperature the affinity of the migrant to the food can be the major factor determining the variability in the migration values (more than 70% of variance). In situations where both the time of consumption and temperature can vary, these factors can be responsible, respectively, for more than 60% and 20% of the variance in the migration estimates. The approach presented can be used with databases from consumption surveys to yield a true probabilistic estimate of exposure.     

Estimating Benchmark Exposure for Air Particulate Matter Using Latent Class Models

We performed benchmark exposure (BME) calculations for particulate matter when multiple dichotomous outcome variables are involved using latent class modeling techniques and generated separate results for both the extra risk and additional risk. The use of latent class models in this study is advantageous because it combined several outcomes into just two classes (namely, a high‐risk class and a low‐risk class) and compared these two classes to obtain the BME levels. This novel approach addresses a key problem in risk estimation—namely, the multiple comparisons problem, where separate regression models are fitted for each outcome variable and the reference exposure will rely on the results of the best‐fitting model. Because of the complex nature of the estimation process, the bootstrap approach was used to estimate the reference exposure level, thereby reducing uncertainty in the obtained values. The methodology developed in this article was applied to environmental data by identifying unmeasured class membership (e.g., morbidity vs. no morbidity class) among infants in utero using observed characteristics that included low birth weight, preterm birth, and small for gestational age.     

Combining Food Frequency and Survey Data to Quantify Long-Term Dietary Exposure: A Methyl Mercury Case Study

Twenty-four-hour recall data from the Continuing Survey of Food Intake by Individuals (CSFII) are frequently used to estimate dietary exposure for risk assessment. Food frequency questionnaires are traditional instruments of epidemiological research; however, their application in dietary exposure and risk assessment has been limited. This article presents a probabilistic method of bridging the National Health and Nutrition Examination Survey (NHANES) food frequency and the CSFII data to estimate longitudinal (usual) intake, using a case study of seafood mercury exposures for two population subgroups (females 16 to 49 years and children 1 to 5 years). Two hundred forty-nine CSFII food codes were mapped into 28 NHANES fish/shellfish categories. FDA and state/local seafood mercury data were used. A uniform distribution with minimum and maximum blood-diet ratios of 0.66 to 1.07 was assumed. A probabilistic assessment was conducted to estimate distributions of individual 30-day average daily fish/shellfish intakes, methyl mercury exposure, and blood levels. The upper percentile estimates of fish and shellfish intakes based on the 30-day daily averages were lower than those based on two- and three-day daily averages. These results support previous findings that distributions of "usual" intakes based on a small number of consumption days provide overestimates in the upper percentiles. About 10% of the females (16 to 49 years) and children (1 to 5 years) may be exposed to mercury levels above the EPA's RfD. The predicted 75th and 90th percentile blood mercury levels for the females in the 16-to-49-year group were similar to those reported by NHANES. The predicted 90th percentile blood mercury levels for children in the 1-to-5-year subgroup was similar to NHANES and the 75th percentile estimates were slightly above the NHANES.     

Use of Markov Chain Monte Carlo Analysis with a Physiologically-Based Pharmacokinetic Model of Methylmercury to Estimate Exposures in U.S. Women of Childbearing Age

A Bayesian approach, implemented using Markov Chain Monte Carlo (MCMC) analysis, was applied with a physiologically‐based pharmacokinetic (PBPK) model of methylmercury (MeHg) to evaluate the variability of MeHg exposure in women of childbearing age in the U.S. population. The analysis made use of the newly available National Health and Nutrition Survey (NHANES) blood and hair mercury concentration data for women of age 16–49 years (sample size, 1,582). Bayesian analysis was performed to estimate the population variability in MeHg exposure (daily ingestion rate) implied by the variation in blood and hair concentrations of mercury in the NHANES database. The measured variability in the NHANES blood and hair data represents the result of a process that includes interindividual variation in exposure to MeHg and interindividual variation in the pharmacokinetics (distribution, clearance) of MeHg. The PBPK model includes a number of pharmacokinetic parameters (e.g., tissue volumes, partition coefficients, rate constants for metabolism and elimination) that can vary from individual to individual within the subpopulation of interest. Using MCMC analysis, it was possible to combine prior distributions of the PBPK model parameters with the NHANES blood and hair data, as well as with kinetic data from controlled human exposures to MeHg, to derive posterior distributions that refine the estimates of both the population exposure distribution and the pharmacokinetic parameters. In general, based on the populations surveyed by NHANES, the results of the MCMC analysis indicate that a small fraction, less than 1%, of the U.S. population of women of childbearing age may have mercury exposures greater than the EPA RfD for MeHg of 0.1 μg/kgg/day, and that there are few, if any, exposures greater than the ATSDR MRL of 0.3 μgg/kgg/day. The analysis also indicates that typical exposures may be greater than previously estimated from food consumption surveys, but that the variability in exposure within the population of U.S. women of childbearing age may be less than previously assumed.     

Bayesian Stacked Parametric Survival with Frailty Components and Interval‐Censored Failure Times: An Application to Food Allergy Risk

To better understand the risk of exposure to food allergens, food challenge studies are designed to slowly increase the dose of an allergen delivered to allergic individuals until an objective reaction occurs. These dose‐to‐failure studies are used to determine acceptable intake levels and are analyzed using parametric failure time models. Though these models can provide estimates of the survival curve and risk, their parametric form may misrepresent the survival function for doses of interest. Different models that describe the data similarly may produce different dose‐to‐failure estimates. Motivated by predictive inference, we developed a Bayesian approach to combine survival estimates based on posterior predictive stacking, where the weights are formed to maximize posterior predictive accuracy. The approach defines a model space that is much larger than traditional parametric failure time modeling approaches. In our case, we use the approach to include random effects accounting for frailty components. The methodology is investigated in simulation, and is used to estimate allergic population eliciting doses for multiple food allergens.     

Comparative Probabilistic Assessment of Occupational Pesticide Exposures Based on Regulatory Assessments

Implementation of probabilistic analyses in exposure assessment can provide valuable insight into the risks of those at the extremes of population distributions, including more vulnerable or sensitive subgroups. Incorporation of these analyses into current regulatory methods for occupational pesticide exposure is enabled by the exposure data sets and associated data currently used in the risk assessment approach of the Environmental Protection Agency (EPA). Monte Carlo simulations were performed on exposure measurements from the Agricultural Handler Exposure Database and the Pesticide Handler Exposure Database along with data from the Exposure Factors Handbook and other sources to calculate exposure rates for three different neurotoxic compounds (azinphos methyl, acetamiprid, emamectin benzoate) across four pesticide‐handling scenarios. Probabilistic estimates of doses were compared with the no observable effect levels used in the EPA occupational risk assessments. Some percentage of workers were predicted to exceed the level of concern for all three compounds: 54% for azinphos methyl, 5% for acetamiprid, and 20% for emamectin benzoate. This finding has implications for pesticide risk assessment and offers an alternative procedure that may be more protective of those at the extremes of exposure than the current approach.     

基于贝叶斯因子模型金融高频波动率预测研究

构建了包含时变系数和动态方差的贝叶斯HAR潜在因子模型(DMA(DMS)-FAHAR),并对我国金融期货(主要是股指期货和国债期货)的高频已实现波动率进行预测.通过构建贝叶斯动态潜在因子模型提取包含波动率变量、跳跃变量和考虑杠杆效应的符号跳跃变量等预测变量的重要信息.同时,在模型中加入了投机活动变量,以考察市场投机活动对中国金融期货市场波动率预测的影响.预测结果表明,时变贝叶斯潜在因子模型在所有参与比较的预测模型当中具有最优的短期、中期和长期预测效果.同时,具有时变参数和时变预测变量的贝叶斯HAR族模型在很大程度上提高了固定参数HAR族模型的预测能力.在股指期货和国债期货的预测模型中加入投机活动变量可以获得更好的预测效果.     

Benzene Inhalation by Parts Washers: New Estimates Based on Measures of Occupational Exposure to Solvent Coaromatics

Questions persist regarding assessment of workers’ exposures to products containing low levels of benzene, such as mineral spirit solvent (MSS). This study summarizes previously unpublished data for parts‐washing activities, and evaluates potential daily and lifetime cumulative benzene exposures incurred by workers who used historical and current formulations of a recycled mineral spirits solvent in manual parts washers. Measured benzene concentrations in historical samples from parts‐washing operations were frequently below analytical detection limits. To better assess benzene exposure among these workers, air‐to‐solvent concentration ratios measured for toluene, ethylbenzene, and xylenes (TEX) were used to predict those for benzene based on a statistical model, conditional on physical‐chemical theory supported by new thermodynamic calculations of TEX and benzene activity coefficients in a modeled MSS‐type solvent. Using probabilistic methods, the distributions of benzene concentrations were then combined with distributions of other exposure parameters to estimate eight‐hour time‐weighted average (TWA) exposure concentration distributions and corresponding daily respiratory dose distributions for workers using these solvents in parts washers. The estimated 50th (95th) percentile of the daily respiratory dose and corresponding eight‐hour TWA air concentration for workers performing parts washing are 0.079 (0.77) mg and 0.0030 (0.028) parts per million by volume (ppm) for historical solvent, and 0.020 (0.20) mg and 0.00078 (0.0075) ppm for current solvent, respectively. Both 95th percentile eight‐hour TWA respiratory exposure estimates for solvent formulations are less than 10% of the current Occupational Safety and Health Administration permissible exposure limit of 1.0 ppm for benzene.     

Quantitative Assessment of Exposure to the Mycotoxin Ochratoxin A in Food

This article presents the methodology and the simulation results concerning the quantitative assessment of exposure to the fungus toxin named Ochratoxin A (OA) in food, in humans in France. We show that is possible to provide reliable calculations of exposure to OA with the conjugate means of a nonparametric-type method of simulation, a parametric-type method of simulation, and the use of bootstrap confidence intervals. In the context of the Monte Carlo simulation, the nonparametric method takes into account the consumptions and the contaminations in the simulations only via the raw data whereas the parametric method depends on the random samplings from distribution functions fitted to consumption and contamination data. Our conclusions are based on eight types of food only. Nevertheless, they are meaningful due to the major importance of these foodstuffs in human nourishment in France. This methodology can be applied whatever the food contaminant (pesticides, other mycotoxins, Cadmium, etc.) when data are available.     

Cigarette Smoking and Multiple Health Risk Behaviors: A Latent Class Regression Model to Identify a Profile of Young Adolescents

Cigarette smoking is often established during adolescence when other health‐related risk behaviors tend to occur. The aim of the study was to further investigate the hypothesis that risky health behaviors tend to cluster together and to identify distinctive profiles of young adolescents based on their smoking habits. To explore the idea that smoking behavior can predict membership in a specific risk profile of adolescents, with heavy smokers being more likely to exhibit other risk behaviors, we reanalyzed the data from the 2014 Health Behaviour in School‐Aged Children Italian survey of about 60,000 first‐ and third‐grade junior high school (JHS) and second‐grade high school (HS) students. A Bayesian approach was adopted for selecting the manifest variables associated with smoking; a latent class regression model was employed to identify smoking behaviors among adolescents. Finally, a health‐related risk pattern associated with different types of smoking behaviors was found. Heavy smokers engaged in higher alcohol use and abuse and experienced school failure more often than their peers. Frequent smokers reported below‐average academic achievement and self‐rated their health as fair/poor more frequently than nonsmokers. Lifetime cannabis use and early sexual intercourse were more frequent among heavy smokers. Our findings provide elements for constructing a profile of frequent adolescent smokers and for identifying behavioral risk patterns during the transition from JHS to HS. This may provide an additional opportunity to devise interventions that could be more effective to improve smoking cessation among occasional smokers and to adequately address other risk behaviors among frequent smokers.     

Cancer Risk Assessment with Intermittent Exposure

Applications of methods for carcinogenic risk assessment often focus on estimating lifetime cancer risk. With intermittent or time-dependent exposures, lifetime risk is often approximated on the basis of a lifetime average daily dose (LADD). In this article, we show that there exists a lifetime equivalent constant dose (LECD) which leads to the same lifetime risk as the actual time-dependent exposure pattern. The ratio C = LECD/LADD then provides a measure of accuracy of risk estimates based on the LADD, as well as a basis for correcting such estimates. Theoretical results derived under the classical multistage model and the two-stage birth-death-mutation model suggest that the maximum value of C, which represents the factor by which the LADD may lead to underestimates of risk, will often lie in the range of 2- to 5-fold. The practical application of these results is illustrated in the case of astronauts subjected to relatively short-term exposure to volatile organics in a closed space station environment, and in the case of the ingestion of pesticide residues in food where consumption patterns vary with age.     

A Bayesian Network Estimation of the Service‐Profit Chain for Transport Service Satisfaction

Bayesian network methodology is used to model key linkages of the service‐profit chain within the context of transportation service satisfaction. Bayesian networks offer some advantages for implementing managerially focused models over other statistical techniques designed primarily for evaluating theoretical models. These advantages are (1) providing a causal explanation using observable variables within a single multivariate model, (2) analysis of nonlinear relationships contained in ordinal measurements, (3) accommodation of branching patterns that occur in data collection, and (4) the ability to conduct probabilistic inference for prediction and diagnostics with an output metric that can be understood by managers and academics. Sample data from 1,101 recent transport service customers are utilized to select and validate a Bayesian network and conduct probabilistic inference.     

Inferring an Augmented Bayesian Network to Confront a Complex Quantitative Microbial Risk Assessment Model with Durability Studies: Application to Bacillus Cereus on a Courgette Purée Production Chain

The Monte Carlo (MC) simulation approach is traditionally used in food safety risk assessment to study quantitative microbial risk assessment (QMRA) models. When experimental data are available, performing Bayesian inference is a good alternative approach that allows backward calculation in a stochastic QMRA model to update the experts’ knowledge about the microbial dynamics of a given food‐borne pathogen. In this article, we propose a complex example where Bayesian inference is applied to a high‐dimensional second‐order QMRA model. The case study is a farm‐to‐fork QMRA model considering genetic diversity of Bacillus cereus in a cooked, pasteurized, and chilled courgette purée. Experimental data are Bacillus cereus concentrations measured in packages of courgette purées stored at different time‐temperature profiles after pasteurization. To perform a Bayesian inference, we first built an augmented Bayesian network by linking a second‐order QMRA model to the available contamination data. We then ran a Markov chain Monte Carlo (MCMC) algorithm to update all the unknown concentrations and unknown quantities of the augmented model. About 25% of the prior beliefs are strongly updated, leading to a reduction in uncertainty. Some updates interestingly question the QMRA model.     

Assessing Risk‐Based Upper Limits of Melamine Migration from Food Containers

Melamine contamination of food has become a major food safety issue because of incidents of infant disease caused by exposure to this chemical. This study was aimed at establishing a safety limit in Taiwan for the degree of melamine migration from food containers. Health risk assessment was performed for three exposure groups (preschool children, individuals who dine out, and elderly residents of nursing homes). Selected values of tolerable daily intake (TDI) for melamine were used to calculate the reference migration concentration limit (RMCL) or reference specific migration limit (RSML) for melamine food containers. The only existing values of these limits for international standards today are 1.2 mg/L (0.2 mg/dm²) in China and 30 mg/L (5 mg/dm²) in the European Union. The factors used in the calculations included the specific surface area of food containers, daily food consumption rate, body weight, TDI, and the percentile of the population protected at a given migration concentration limit (MCL). The results indicate that children are indeed at higher risk of melamine exposure at toxic levels than are other groups and that the 95th percentile of MCL (specific surface area = 5) for children aged 1–6 years should be the RMCL (0.07 mg/dm²) for protecting the sensitive and general population.     

Bayesian Hierarchical Structure for Quantifying Population Variability to Inform Probabilistic Health Risk Assessments

Human variability is a very important factor considered in human health risk assessment for protecting sensitive populations from chemical exposure. Traditionally, to account for this variability, an interhuman uncertainty factor is applied to lower the exposure limit. However, using a fixed uncertainty factor rather than probabilistically accounting for human variability can hardly support probabilistic risk assessment advocated by a number of researchers; new methods are needed to probabilistically quantify human population variability. We propose a Bayesian hierarchical model to quantify variability among different populations. This approach jointly characterizes the distribution of risk at background exposure and the sensitivity of response to exposure, which are commonly represented by model parameters. We demonstrate, through both an application to real data and a simulation study, that using the proposed hierarchical structure adequately characterizes variability across different populations.