New supporting data to guide the use of evident toxicity in acute oral toxicity studies (OECD TG 420).

Currently there are three test guidelines (TG) for acute oral toxicity studies of substances or mixtures from the Organisation for Economic Co-operation and Development (OECD). TG 423 and TG 425 use lethality as an endpoint, while TG 420 replaces death with 'evident toxicity', defined as clear signs that exposure to a higher dose would result in death. However, the perceived subjectivity of 'evident toxicity' may be preventing wider use of TG 420. To address this, the UK National Centre for the Replacement, Refinement and Reduction of Animals in Research (NC3Rs) and the European Partnership for Alternative Approaches to Animal Testing (EPAA) collaborated to provide recommendations on the recognition of 'evident toxicity'. Historical data from acute oral toxicity studies were analysed for clinical signs at the lower dose that could have predicted death at the higher dose. Several signs including ataxia, laboured respiration, and eyes partially closed, alone or in combination, are highly predictive. Others such as lethargy, decreased respiration, and loose faeces have lower but still appreciable positive predictive value (PPV). The data has been used to develop recommendations to promote use of TG 420 and thus reduce the suffering and numbers of animals used in acute oral toxicity studies.

Comparing self-reported energy intake using an online dietary tool with energy expenditure by an activity tracker.

OBJECTIVE: The aim of this study was to compare self-reported total energy intake (TEI) collected using an online multiple-pass 24-h dietary recall tool (Intake24) with total energy expenditure (TEE) estimated from Fitbit Charge 2-improved algorithms in adults from the NoHoW trial (12-mo weight maintenance after free-living weight loss). METHODS: Bland-Altman plots were used to assess the level of agreement between TEI and TEE at baseline and after 12 mo. The ratio of TEI to TEE was also calculated. RESULTS: Data from 1323 participants (71% female) was included in the analysis (mean ± SD: age 45 ± 12 y, body mass index 29.7 ± 5.4 kg/m2, initial weight loss 11.5 ± 6.5 kg). The TEI was lower than TEE on average by 33%, with limits of agreement ranging from -91% to +25%. Men, younger individuals, those with higher body mass index, those with the greater weight loss before enrollment, and those who gained weight during the study underestimated to a greater extent. CONCLUSIONS: These findings contribute to the ongoing research examining the validity of technology-based dietary assessment tools.

Consumption of foods with the highest nutritional quality, and the lowest greenhouse gas emissions and price, differs between socio-economic groups in the UK population.

OBJECTIVE: To establish a baseline understanding of whether consuming food with the highest nutritional quality, lowest greenhouse gas emissions (GHGE) and cost differs between different UK demographic and socio-economic population groups. DESIGN: Multiple linear regression models were fitted to evaluate the relationship between predictor socio-demographic variables in this study (i.e. sex, ethnic group, age, BMI and level of deprivation) and the response variables (i.e. consumption of items considered most nutritious, with a low GHGE and price, as a proportion of total items consumed). SETTING: The UK. PARTICIPANTS: 1374 adult (18-65 years) participants from the National Diet and Nutrition Survey latest waves 9-11 (2016-2017 and 2018-2019). RESULTS: Based on the total energy consumption in a day, the average diet-based GHGE was significantly higher for participants with a higher BMI. Non-white and most deprived participants spent significantly (P < 0·001) less money per total energy consumption. Participants with a BMI between 18·6 and 39·9 kg/m2 and those living in the least deprived areas consumed a significantly (P < 0·001) higher amount of those items considered the most nutritious, with the lowest GHGE and cost per 100 kcal. CONCLUSIONS: Consumption of food with the highest nutritional quality, lowest GHGE and cost in the UK varies among those with different socio-demographic characteristics, especially the deprivation level of participants. Our analysis endorses the consideration of environmental sustainability and affordability, in addition to the consideration of nutritional quality from a health perspective, to make current dietary guidelines more encompassing and equitable.

Food-Level Analysis to Identify Dietary Choices With the Highest Nutritional Quality and Lowest Greenhouse Gas Emissions and Price.

Introduction: Food systems are challenged to provide healthy, sustainable and affordable foods. From a consumer perspective, identifying healthy, sustainable and affordable choices based on individual food products rather than diets could promote better shopping choices. Objective: To identify foods and drinks with the highest nutritional quality and lowest greenhouse gas emissions (GHGE) and price. We also assessed how a combination of these indicators (e.g., nutritional quality, GHGE and price) for food categories aligned with current United Kingdom dietary recommendations. Materials and Methods: We performed a secondary analysis of the National Diet and Nutrition Survey (NDNS) nutrient databank year 11 (2018/2019). Spearman correlation coefficients were used to assess the strength of relationships between nutritional quality, environmental impact and/or prices per 100 kcal. In addition, we developed an optimized nutritional quality, GHGE and price score for each food or drink item based on the overall medians for each of these indicators. Results: Median nutritional value was highest for fruit and vegetables, whilst median GHGE and price was lower for starchy carbohydrates, fats and items of which consumption should be limited. The relative proportions of foods considered the most nutritious and with a low GHGE and price in each of the food categories, on a per 100 kcal basis, were comparable to the proportions in the Eatwell Guide, except for the proportion of fruits and vegetables being smaller and the proportion of potatoes, bread, rice, pasta, and other starchy carbohydrates being larger in our analysis. Conclusion: Public health efforts should consider the impact of dietary choices not only in terms of nutritional quality but also in terms of environmental and economic impact. Our food-based analysis shows a large variation in nutritional quality, GHGE and price within and across food categories, which provides consumers with opportunities for "food swaps" that are more nutritious and have lower GHGE and price.

Nutritional Quality, Environmental Impact and Cost of Ultra-Processed Foods: A UK Food-Based Analysis.

Food-based analyses of the healthiness, environmental sustainability and affordability of processed and ultra-processed foods are lacking. This paper aimed to determine how ultra-processed and processed foods compare to fresh and minimally processed foods in relation to nutritional quality, greenhouse gas emissions and cost on the food and food group level. Data from the National Diet and Nutrition Survey nutrient databank year 11 (2018/2019) were used for this analysis. Median and bootstrapped medians of nutritional quality (NRF8.3 index), greenhouse gas emissions (gCO2-equivalents) and cost (in GBP) were compared across processing categories. An optimal score based on the medians was created to identify the most nutritional, sustainable, and affordable options across processing categories. On a per 100 kcal basis, ultra-processed and processed foods had a lower nutritional quality, lower greenhouse gas emissions, and were cheaper than minimally processed foods, regardless of their total fat, salt and/or sugar content. The most nutritious, environmentally friendly, and affordable foods were generally lower in total fat, salt, and sugar, irrespective of processing level. The high variability in greenhouse gas emissions and cost across food groups and processing levels offer opportunities for food swaps representing the healthiest, greenest, and most affordable options.

Comparison of the Validity and Generalizability of Machine Learning Algorithms for the Prediction of Energy Expenditure: Validation Study.

BACKGROUND: Accurate solutions for the estimation of physical activity and energy expenditure at scale are needed for a range of medical and health research fields. Machine learning techniques show promise in research-grade accelerometers, and some evidence indicates that these techniques can be applied to more scalable commercial devices. OBJECTIVE: This study aims to test the validity and out-of-sample generalizability of algorithms for the prediction of energy expenditure in several wearables (ie, Fitbit Charge 2, ActiGraph GT3-x, SenseWear Armband Mini, and Polar H7) using two laboratory data sets comprising different activities. METHODS: Two laboratory studies (study 1: n=59, age 44.4 years, weight 75.7 kg; study 2: n=30, age=31.9 years, weight=70.6 kg), in which adult participants performed a sequential lab-based activity protocol consisting of resting, household, ambulatory, and nonambulatory tasks, were combined in this study. In both studies, accelerometer and physiological data were collected from the wearables alongside energy expenditure using indirect calorimetry. Three regression algorithms were used to predict metabolic equivalents (METs; ie, random forest, gradient boosting, and neural networks), and five classification algorithms (ie, k-nearest neighbor, support vector machine, random forest, gradient boosting, and neural networks) were used for physical activity intensity classification as sedentary, light, or moderate to vigorous. Algorithms were evaluated using leave-one-subject-out cross-validations and out-of-sample validations. RESULTS: The root mean square error (RMSE) was lowest for gradient boosting applied to SenseWear and Polar H7 data (0.91 METs), and in the classification task, gradient boost applied to SenseWear and Polar H7 was the most accurate (85.5%). Fitbit models achieved an RMSE of 1.36 METs and 78.2% accuracy for classification. Errors tended to increase in out-of-sample validations with the SenseWear neural network achieving RMSE values of 1.22 METs in the regression tasks and the SenseWear gradient boost and random forest achieving an accuracy of 80% in classification tasks. CONCLUSIONS: Algorithms trained on combined data sets demonstrated high predictive accuracy, with a tendency for superior performance of random forests and gradient boosting for most but not all wearable devices. Predictions were poorer in the between-study validations, which creates uncertainty regarding the generalizability of the tested algorithms.

Evidence-Based Digital Tools for Weight Loss Maintenance: The NoHoW Project.

BACKGROUND: Effective interventions and commercial programmes for weight loss (WL) are widely available, but most people regain weight. Few effective WL maintenance (WLM) solutions exist. The most promising evidence-based behaviour change techniques for WLM are self-monitoring, goal setting, action planning and control, building self-efficacy, and techniques that promote autonomous motivation (e.g., provide choice). Stress management and emotion regulation techniques show potential for prevention of relapse and weight regain. Digital technologies (including networked-wireless tracking technologies, online tools and smartphone apps, multimedia resources, and internet-based support) offer attractive tools for teaching and supporting long-term behaviour change techniques. However, many digital offerings for weight management tend not to include evidence-based content and the evidence base is still limited. The Project: First, the project examined why, when, and how many European citizens make WL and WLM attempts and how successful they are. Second, the project employed the most up-to-date behavioural science research to develop a digital toolkit for WLM based on 2 key conditions, i.e., self-management (self-regulation and motivation) of behaviour and self-management of emotional responses for WLM. Then, the NoHoW trial tested the efficacy of this digital toolkit in adults who achieved clinically significant (≥5%) WL in the previous 12 months (initial BMI ≥25). The primary outcome was change in weight (kg) at 12 months from baseline. Secondary outcomes included biological, psychological, and behavioural moderators and mediators of long-term energy balance (EB) behaviours, and user experience, acceptability, and cost-effectiveness. IMPACT: The project will directly feed results from studies on European consumer behaviour, design and evaluation of digital toolkits self-management of EB behaviours into development of new products and services for WLM and digital health. The project has developed a framework and digital architecture for interventions in the context of EB tracking and will generate results that will help inform the next generation of personalised interventions for effective self-management of weight and health.

Improving energy expenditure estimates from wearable devices: A machine learning approach.

A means of quantifying continuous, free-living energy expenditure (EE) would advance the study of bioenergetics. The aim of this study was to apply a non-linear, machine learning algorithm (random forest) to predict minute level EE for a range of activities using acceleration, physiological signals (e.g., heart rate, body temperature, galvanic skin response), and participant characteristics (e.g., sex, age, height, weight, body composition) collected from wearable devices (Fitbit charge 2, Polar H7, SenseWear Armband Mini and Actigraph GT3-x) as potential inputs. By utilising a leave-one-out cross-validation approach in 59 subjects, we investigated the predictive accuracy in sedentary, ambulatory, household, and cycling activities compared to indirect calorimetry (Vyntus CPX). Over all activities, correlations of at least r = 0.85 were achieved by the models. Root mean squared error ranged from 1 to 1.37 METs and all overall models were statistically equivalent to the criterion measure. Significantly lower error was observed for Actigraph and Sensewear models, when compared to the manufacturer provided estimates of the Sensewear Armband (p < 0.05). A high degree of accuracy in EE estimation was achieved by applying non-linear models to wearable devices which may offer a means to capture the energy cost of free-living activities.

The NoHoW protocol: a multicentre 2×2 factorial randomised controlled trial investigating an evidence-based digital toolkit for weight loss maintenance in European adults.

INTRODUCTION: Obesity and associated diseases place a severe burden on healthcare systems. Behavioural interventions for weight loss (WL) are successful in the short term but often result in weight regain over time. Self-regulation of eating and activity behaviours may significantly enhance weight loss maintenance (WLM) and may be effectively augmented by contextual behavioural approaches to emotion regulation. The NoHoW trial tests the efficacy of a theoretically informed, evidence-based digital toolkit using a mobile-enabled website, activity trackers and Wi-Fi scales for WLM aiming to target (1) self-regulation and motivation, and (2) emotion regulation in adults who achieved clinically significant (≥5%) WL in the previous 12 months (initial body mass index (BMI) ≥25 kg/m2). METHODS AND ANALYSIS: The study is an 18-month, 3-centre, 2×2 factorial single-blind, randomised controlled trial, which recruited 1627 participants achieving ≥5% WL between March 2017 and March 2018. Participants are randomly allocated to one of four arms: (1) self-monitoring only (self-weighing and activity tracker), (2) self-regulation and motivation, (3) emotion regulation or (4) combined self-regulation, motivation and emotion regulation. Participants attend four clinical investigation days at 0, 6, 12 and 18 months and are instructed to use the digital toolkit for 18 weeks during the first 6 months and at their discretion for the remaining 12 months. The primary outcome is change in weight (kg) at 12 months from baseline. Secondary outcomes are body composition (eg, bioimpedance analysis), health biomarkers (glycated haemoglobin, lipids, blood pressure, hair cortisol), dietary intake, physical activity, sleep, motivational, self-regulatory, emotion regulatory moderators/mediators of WLM, engagement, user experience, acceptability and cost-effectiveness of the interventions. ETHICS AND DISSEMINATION: Ethical approval was granted by Institutional Ethics Committees at the Universities of Leeds (17-0082; 27 February 2017), Lisbon (17/2016; 20 February 2017) and Capital Region of Denmark (H-16030495, 8 March 2017). Results will be published in scientific journals. TRIAL REGISTRATION NUMBER: ISRCTN88405328.

Healthy and sustainable diets that meet greenhouse gas emission reduction targets and are affordable for different income groups in the UK.

OBJECTIVE: To model dietary changes required to shift the UK population to diets that meet dietary recommendations for health, have lower greenhouse gas emissions (GHGE) and are affordable for different income groups. DESIGN: Linear programming was used to create diets that meet dietary requirements for health and reduced GHGE (57 and 80 % targets) by income quintile, taking account of food budgets and foods currently purchased, thereby keeping dietary change to a minimum.Setting/ParticipantsNutrient composition, GHGE and price data were mapped to 101 food groups in household food purchase data (UK Living Cost and Food Survey (2013), 5144 households). RESULTS: Current diets of all income quintiles had similar total GHGE, but the source of GHGE differed by types of meat and amount of fruit and vegetables. It was possible to create diets with a 57 % reduction in GHGE that met dietary and cost restraints in all income groups. In the optimised diets, the food sources of GHGE differed by income group due to the cost and keeping the level of deviation from current diets to a minimum. Broadly, the changes needed were similar across all groups; reducing animal-based products and increasing plant-based foods but varied by specific foods. CONCLUSIONS: Healthy and lower-GHGE diets could be created in all income quintiles but tailoring changes to income groups to minimise deviation may make dietary changes more achievable. Specific attention must be given to make interventions and policies appropriate for all income groups.

Activity energy expenditure is an independent predictor of energy intake in humans.

BACKGROUND: There is evidence that the energetic demand of metabolically active tissue is associated with day-to-day food intake (EI). However, the extent to which behavioural components of total daily energy expenditure (EE) such as activity energy expenditure (AEE) are also associated with EI is unknown. Therefore, the present study examined the cross-sectional associations between body composition, resting metabolic rate (RMR), AEE and EI. METHODS: Data for 242 individuals (114 males; 128 females; BMI = 25.7 ± 4.9 kg/m2) were collated from the baseline control conditions of five studies employing common measures of body composition (air displacement plethysmography) and RMR (indirect calorimetry). Daily EI (weighed-dietary records) and EE (FLEX heart rate) were measured over 6-7 days, and AEE was calculated as total daily EE minus RMR. RESULTS: Linear regression indicated that RMR (ß = 0.39; P < 0.001), fat mass (ß = -0.26; P < 0.001) and AEE (ß = 0.18; P = 0.002) were independent predictors of mean daily EI, with AEE adding ≈3% of variance to the model after controlling for age, sex and study (F(10, 231) = 18.532, P < 0.001; R2 = 0.445). Path analyses indicated that the effect of FFM on mean daily EI was mediated by RMR (P < 0.05), while direct (ß = 0.19; P < 0.001) and indirect (ß = 0.20; P = 0.001) associations between AEE and mean daily EI were observed. CONCLUSIONS: When physical activity was allowed to vary under free-living conditions, AEE was associated with mean daily EI independently of other biological determinants of EI arising from body composition and RMR. These data suggest that EE per se exerts influence over daily food intake, with both metabolic (RMR) and behavioral (AEE) components of total daily EE potentially influencing EI via their contribution to daily energy requirements.

An evaluation of the fixed concentration procedure for assessment of acute inhalation toxicity.

Acute inhalation studies are conducted in animals as part of chemical hazard identification and for classification and labelling. Current methods employ death as an endpoint (Organisation for Economic Co-operation and Development (OECD) test guideline (TG) 403 and TG436) while the recently approved fixed concentration procedure (FCP) (OECD TG433) uses fewer animals and replaces lethality as an endpoint with evident toxicity. Evident toxicity is the presence of clinical signs that predict that exposure to the next highest concentration will cause severe toxicity or death in most animals. Approval of TG433 was the result of an international initiative, led by the National Centre for the Replacement, Refinement & Reduction of Animals in Research (NC3Rs), which collected data from six laboratories on clinical signs recorded for inhalation studies on 172 substances. This paper summarises previously published data and describes the additional analyses of the dataset that were essential for approval of the TG.

Buying less and wasting less food. Changes in household food energy purchases, energy intakes and energy density between 2007 and 2012 with and without adjustment for food waste.

OBJECTIVE: Consumers in the UK responded to the rapid increases in food prices between 2007 and 2009 partly by reducing the amount of food energy bought. Household food and drink waste has also decreased since 2007. The present study explored the combined effects of reductions in food purchases and waste on estimated food energy intakes and dietary energy density. DESIGN: The amount of food energy purchased per adult equivalent was calculated from Kantar Worldpanel household food and drink purchase data for 2007 and 2012. Food energy intakes were estimated by adjusting purchase data for food and drink waste, using waste factors specific to the two years and scaled for household size. SETTING: Scotland. SUBJECTS: Households in Scotland (n 2657 in 2007; n 2841 in 2012). RESULTS: The amount of food energy purchased decreased between 2007 and 2012, from 8·6 to 8·2 MJ/adult equivalent per d (P<0·001). After accounting for the decrease in food waste, estimated food energy intake was not significantly different (7·3 and 7·2 MJ/adult equivalent per d for 2007 and 2012, respectively; P=0·186). Energy density of foods purchased increased slightly from 700 to 706 kJ/100 g (P=0·010). CONCLUSIONS: While consumers in Scotland reduced the amount of food energy that they purchased between 2007 and 2012, this was balanced by reductions in household food and drink waste over the same time, resulting in no significant change in net estimated energy intake of foods brought into the home.

An evaluation of the IDEEA™ activity monitor for estimating energy expenditure.

Objective estimates of activity patterns and energy expenditure (EE) are important for the measurement of energy balance. The Intelligent Device for Energy Expenditure and Activity (IDEEA) can estimate EE from the thirty-five postures and activities it can identify and record. The present study evaluated the IDEEA system's estimation of EE using whole-body indirect calorimetry over 24 h, and in free-living subjects using doubly-labelled water (DLW) over 14 d. EE was calculated from the IDEEA data using calibration values for RMR and EE while sitting and standing, both as estimated by the IDEEA system (IDEEA(est)) and measured by indirect calorimetry (IDEEA(meas)). Subjects were seven females and seven males, mean age 38·1 and 39·7 years, mean BMI 25·2 and 26·2 kg/m2, respectively. The IDEEA(est) method produced a similar estimate of EE to the calorimeter (10·8 and 10·8 MJ, NS), while the IDEEA(meas) method underestimated EE (9·9 MJ, P < 0·001). After removing data from static cycling, which the IDEEA was unable to identify as an activity, both the IDEEA(est) and IDEEA(meas) methods overestimated EE compared to the calorimeter (9·9 MJ, P < 0·001; 9·1 MJ, P < 0·05 and 8·6 MJ, respectively). Similarly, the IDEEA system overestimated EE compared to DLW over 14 d; 12·7 MJ/d (P < 0·01), 11·5 MJ/d (P < 0·01) and 9·5 MJ/d for the IDEEA(est), IDEEA(meas) and DLW, respectively. The IDEEA system overestimated EE both in the controlled laboratory and free-living environments. Using measured EE values for RMR, sitting and standing reduced, but did not eliminate, the error in estimated EE.

Sustainable diets for the future: Can we contribute to reducing greenhouse gas emissions by eating a healthy diet?

BACKGROUND: Food systems account for 18-20% of UK annual greenhouse gas emissions (GHGEs). Recommendations for improving food choices to reduce GHGEs must be balanced against dietary requirements for health. OBJECTIVE: We assessed whether a reduction in GHGEs can be achieved while meeting dietary requirements for health. DESIGN: A database was created that linked nutrient composition and GHGE data for 82 food groups. Linear programming was used iteratively to produce a diet that met the dietary requirements of an adult woman (19-50 y old) while minimizing GHGEs. Acceptability constraints were added to the model to include foods commonly consumed in the United Kingdom in sensible quantities. A sample menu was created to ensure that the quantities and types of food generated from the model could be combined into a realistic 7-d diet. Reductions in GHGEs of the diets were set against 1990 emission values. RESULTS: The first model, without any acceptability constraints, produced a 90% reduction in GHGEs but included only 7 food items, all in unrealistic quantities. The addition of acceptability constraints gave a more realistic diet with 52 foods but reduced GHGEs by a lesser amount of 36%. This diet included meat products but in smaller amounts than in the current diet. The retail cost of the diet was comparable to the average UK expenditure on food. CONCLUSION: A sustainable diet that meets dietary requirements for health with lower GHGEs can be achieved without eliminating meat or dairy products or increasing the cost to the consumer.

Alperujo extract, hydroxytyrosol, and 3,4-dihydroxyphenylglycol are bioavailable and have antioxidant properties in vitamin E-deficient rats--a proteomics and network analysis approach.

SCOPE: Olive products are rich in phenolic compounds, which are natural antioxidants in vitro. We tested the in vivo effects of alperujo, an olive production by-product, as well as hydroxytyrosol and 3,4-dihydroxyphenylglycol (DHPG) isolated from alperujo, on indices and pathways of oxidative and metabolic stress in a vitamin E-deficient rat model. METHODS AND RESULTS: Rats were fed a vitamin E-deficient diet for 10 weeks, followed by this diet supplemented with either 100 mg/kg diet dα-tocopherol, alperujo extract, hydroxytyrosol, or 10 mg/kg diet DHPG, for a further 2 weeks. We detected alperujo phenolics in tissues and blood, indicating they are bioavailable. Alperujo extract partially ameliorated elevated plasma levels of thiobarbituric acid reactive substances and also lowered plasma cholesterol levels, whereas hydroxytyrosol increased plasma triglyceride levels. Proteomics and subsequent network analysis revealed that hepatic mitochondrial aldehyde dehydrogenase (ALDH2), of which protein and activity levels were regulated by dα-tocopherol and olive phenolics, represents a novel central regulatory protein hub affected by the dietary interventions. CONCLUSION: The in vivo free radical scavenging properties of olive phenolics appear relatively modest in our model. But alternative mechanisms, including regulation of ALDH2, may represent relevant antioxidant mechanisms by which dietary olive phenolics could have beneficial impact on cardiovascular health.

Exposure assessment for pesticide intake from multiple food products: a Bayesian latent-variable approach.

Pesticide risk assessment for food products involves combining information from consumption and concentration data sets to estimate a distribution for the pesticide intake in a human population. Using this distribution one can obtain probabilities of individuals exceeding specified levels of pesticide intake. In this article, we present a probabilistic, Bayesian approach to modeling the daily consumptions of the pesticide Iprodione though multiple food products. Modeling data on food consumption and pesticide concentration poses a variety of problems, such as the large proportions of consumptions and concentrations that are recorded as zero, and correlation between the consumptions of different foods. We consider daily food consumption data from the Netherlands National Food Consumption Survey and concentration data collected by the Netherlands Ministry of Agriculture. We develop a multivariate latent-Gaussian model for the consumption data that allows for correlated intakes between products. For the concentration data, we propose a univariate latent-t model. We then combine predicted consumptions and concentrations from these models to obtain a distribution for individual daily Iprodione exposure. The latent-variable models allow for both skewness and large numbers of zeros in the consumption and concentration data. The use of a probabilistic approach is intended to yield more robust estimates of high percentiles of the exposure distribution than an empirical approach. Bayesian inference is used to facilitate the treatment of data with a complex structure.