Constructing synthetic populations in the age of big data.

BACKGROUND: To develop public health intervention models using micro-simulations, extensive personal information about inhabitants is needed, such as socio-demographic, economic and health figures. Confidentiality is an essential characteristic of such data, while the data should reflect realistic scenarios. Collection of such data is possible only in secured environments and not directly available for open-source micro-simulation models. The aim of this paper is to illustrate a method of construction of synthetic data by predicting individual features through models based on confidential data on health and socio-economic determinants of the entire Dutch population. METHODS: Administrative records and health registry data were linked to socio-economic characteristics and self-reported lifestyle factors. For the entire Dutch population (n = 16,778,708), all socio-demographic information except lifestyle factors was available. Lifestyle factors were available from the 2012 Dutch Health Monitor (n = 370,835). Regression model was used to sequentially predict individual features. RESULTS: The synthetic population resembles the original confidential population. Features predicted in the first stages of the sequential procedure are virtually similar to those in the original population, while those predicted in later stages of the sequential procedure carry the accumulation of limitations furthered by data quality and previously modelled features. CONCLUSIONS: By combining socio-demographic, economic, health and lifestyle related data at individual level on a large scale, our method provides us with a powerful tool to construct a synthetic population of good quality and with no confidentiality issues.

Predicting self-perceived general health status using machine learning: an external exposome study.

BACKGROUND: Self-perceived general health (SPGH) is a general health indicator commonly used in epidemiological research and is associated with a wide range of exposures from different domains. However, most studies on SPGH only investigated a limited set of exposures and did not take the entire external exposome into account. We aimed to develop predictive models for SPGH based on exposome datasets using machine learning techniques and identify the most important predictors of poor SPGH status. METHODS: Random forest (RF) was used on two datasets based on personal characteristics from the 2012 and 2016 editions of the Dutch national health survey, enriched with environmental and neighborhood characteristics. Model performance was determined using the area under the curve (AUC) score. The most important predictors were identified using a variable importance procedure and individual effects of exposures using partial dependence and accumulated local effect plots. The final 2012 dataset contained information on 199,840 individuals and 81 variables, whereas the final 2016 dataset had 244,557 individuals with 91 variables. RESULTS: Our RF models had overall good predictive performance (2012: AUC = 0.864 (CI: 0.852-0.876); 2016: AUC = 0.890 (CI: 0.883-0.896)) and the most important predictors were "Control of own life", "Physical activity", "Loneliness" and "Making ends meet". Subjects who felt insufficiently in control of their own life, scored high on the De Jong-Gierveld loneliness scale or had difficulty in making ends meet were more likely to have poor SPGH status, whereas increased physical activity per week reduced the probability of poor SPGH. We observed associations between some neighborhood and environmental characteristics, but these variables did not contribute to the overall predictive strength of the models. CONCLUSIONS: This study identified that within an external exposome dataset, the most important predictors for SPGH status are related to mental wellbeing, physical exercise, loneliness, and financial status.

Mapping chronic disease prevalence based on medication use and socio-demographic variables: an application of LASSO on administrative data sources in healthcare in the Netherlands.

BACKGROUND: Policymakers generally lack sufficiently detailed health information to develop localized health policy plans. Chronic disease prevalence mapping is difficult as accurate direct sources are often lacking. Improvement is possible by adding extra information such as medication use and demographic information to identify disease. The aim of the current study was to obtain small geographic area prevalence estimates for four common chronic diseases by modelling based on medication use and socio-economic variables and next to investigate regional patterns of disease. METHODS: Administrative hospital records and general practitioner registry data were linked to medication use and socio-economic characteristics. The training set (n = 707,021) contained GP diagnosis and/or hospital admission diagnosis as the standard for disease prevalence. For the entire Dutch population (n = 16,777,888), all information except GP diagnosis and hospital admission was available. LASSO regression models for binary outcomes were used to select variables strongly associated with disease. Dutch municipality (non-)standardized prevalence estimates for stroke, CHD, COPD and diabetes were then based on averages of predicted probabilities for each individual inhabitant. RESULTS: Adding medication use data as a predictor substantially improved model performance. Estimates at the municipality level performed best for diabetes with a weighted percentage error (WPE) of 6.8%, and worst for COPD (WPE 14.5%)Disease prevalence showed clear regional patterns, also after standardization for age. CONCLUSION: Adding medication use as an indicator of disease prevalence next to socio-economic variables substantially improved estimates at the municipality level. The resulting individual disease probabilities could be aggregated into any desired regional level and provide a useful tool to identify regional patterns and inform local policy.

Potential gains in health expectancy by improving lifestyle: an application for European regions.

BACKGROUND: Prevention aiming at smoking, alcohol consumption, and BMI could potentially bring large gains in life expectancy (LE) and health expectancy measures such as Healthy Life Years (HLY) and Life Expectancy in Good Perceived Health (LEGPH) in the European Union. However, the potential gains might differ by region. METHODS: A Sullivan life table model was applied for 27 European countries to calculate the impact of alternative scenarios of lifestyle behavior on life and health expectancy. Results were then pooled over countries to present the potential gains in HLY and LEGPH for four European regions. RESULTS: Simulations show that up to 4 years of extra health expectancy can be gained by getting all countries to the healthiest levels of lifestyle observed in EU countries. This is more than the 2 years to be gained in life expectancy. Generally, Eastern Europe has the lowest LE, HLY, and LEGPH. Even though the largest gains in LEPGH and HLY can also be made in Eastern Europe, the gap in LE, HLY, and LEGPH can only in a small part be closed by changing smoking, alcohol consumption, and BMI. CONCLUSION: Based on the current data, up to 4 years of good health could be gained by adopting lifestyle as seen in the best-performing countries. Only a part of the lagging health expectancy of Eastern Europe can potentially be solved by improvements in lifestyle involving smoking and BMI. Before it is definitely concluded that lifestyle policy for alcohol use is of relatively little importance compared to smoking or BMI, as our findings suggest, better data should be gathered in all European countries concerning alcohol use and the odds ratios of overconsumption of alcohol.

Pre-to-post diagnosis weight trajectories in colorectal cancer patients with non-metastatic disease.

PURPOSE: Previous studies have shown that > 50% of colorectal cancer (CRC) patients treated with adjuvant chemotherapy gain weight after diagnosis. This may affect long-term health. Therefore, prevention of weight gain has been incorporated in oncological guidelines for CRC with a focus on patients that undergo adjuvant chemotherapy treatment. It is, however, unknown how changes in weight after diagnosis relate to weight before diagnosis and whether weight changes from pre-to-post diagnosis are restricted to chemotherapy treatment. We therefore examined pre-to-post diagnosis weight trajectories and compared them between those treated with and without adjuvant chemotherapy. METHODS: We included 1184 patients diagnosed with stages I-III CRC between 2010 and 2015 from an ongoing observational prospective study. At diagnosis, patients reported current weight and usual weight 2 years before diagnosis. In the 2 years following diagnosis, weight was self-reported repeatedly. We used linear mixed models to analyse weight trajectories. RESULTS: Mean pre-to-post diagnosis weight change was -0.8 (95% CI -1.1, -0.4) kg. Post-diagnosis weight gain was + 3.5 (95% CI 2.7, 4.3) kg in patients who had lost ≥ 5% weight before diagnosis, while on average clinically relevant weight gain after diagnosis was absent in the groups without pre-diagnosis weight loss. Pre-to-post diagnosis weight change was similar in patients treated with (-0.1 kg (95%CI -0.8, 0.6)) and without adjuvant chemotherapy (-0.9 kg (95%CI -1.4, -0.5)). CONCLUSIONS: Overall, hardly any pre-to-post diagnosis weight change was observed among CRC patients, because post-diagnosis weight gain was mainly observed in patients who lost weight before diagnosis. This was observed independent of treatment with adjuvant chemotherapy.

Importance of details in food descriptions in estimating population nutrient intake distributions.

BACKGROUND: National food consumption surveys are important policy instruments that could monitor food consumption of a certain population. To be used for multiple purposes, this type of survey usually collects comprehensive food information using dietary assessment methods like 24-h dietary recalls (24HRs). However, the collection and handling of such detailed information require tremendous efforts. We aimed to improve the efficiency of data collection and handling in 24HRs, by identifying less important characteristics of food descriptions (facets) and assessing the impact of disregarding them on energy and nutrient intake distributions. METHODS: In the Dutch National Food Consumption Survey 2007-2010, food consumption data were collected through interviewer-administered 24HRs using GloboDiet software in 3819 persons. Interviewers asked participants about the characteristics of each food item according to applicable facets. Food consumption data were subsequently linked to the food composition database. The importance of facets for predicting energy and each of the 33 nutrients was estimated using the random forest algorithm. Then a simulation study was performed to determine the influence of deleting less important facets on population nutrient intake distributions. RESULTS: We identified 35% facets as unimportant and deleted them from the total food consumption database. The majority (79.4%) of the percent difference between percentile estimates of the population nutrient intake distributions before and after facet deletion ranged from 0 to 1%, while 20% cases ranged from 1 to 5% and 0.6% cases more than 10%. CONCLUSION: We concluded that our procedure was successful in identifying less important food descriptions in estimating population nutrient intake distributions. The reduction in food descriptions has the potential to reduce the time needed for conducting interviews and data handling while maintaining the data quality of the survey.

FFQ versus repeated 24-h recalls for estimating diet-related environmental impact.

BACKGROUND: There is an increasing interest in estimating environmental impact of individuals' diets by using individual-level food consumption data. However, like assessment of nutrient intakes, these data are prone to substantial measurement errors dependent on the method of dietary assessment, and this often result in attenuation of associations. PURPOSE: To investigate the performance of a food frequency questionnaire (FFQ) for estimating the environmental impact of the diet as compared to independent 24-h recalls (24hR), and to study the association between environmental impact and dietary quality for the FFQ and 24hR. METHODS: We analysed cross-sectional data from 1169 men and women, aged 20-76 years, who participated in the NQplus study, the Netherlands. They completed a 216-item FFQ and two replicates of web-based 24hR. Life cycle assessments of 207 food products were used to calculate greenhouse gas emissions, fossil energy and land use, summarised into an aggregated score, pReCiPe. Validity of the FFQ was evaluated against 24hRs using correlation coefficients and attenuation coefficients. Associations with dietary quality were based on Dutch Healthy Diet 15-index (DHD15-index) and Nutrient Rich Diet score (NRD9.3). RESULTS: For pReCiPe, correlation coefficient between FFQ and 24hR was 0.33 when adjusted for covariates age, gender and BMI, and increased to 0.76 when de-attenuated for within-subject variation in the 24hR. Energy-adjustment slightly reduced these correlations (r = 0.71 for residuals of observed values and 0.59 for residuals of density values). Covariate-adjusted attenuation coefficient for the FFQ was 0.56 (ʎ1 = 0.56 and ʎ1 = 0.65 for observed and density residuals), slightly lower than without covariate adjustment. Diet-related environmental impact was inversely associated with the food-based DHD15-index for both FFQ and 24hR, while associations with the nutrient-based NRD9.3 were inconsistent. CONCLUSIONS: The FFQ slightly underestimated environmental impact when compared to 24hR. Associations with dietary quality are highly dependent on the diet score used, and less dependent on the method of dietary assessment.

Using enhanced regression calibration to combine dietary intake estimates from 24 h recall and FFQ reduces bias in diet-disease associations.

OBJECTIVE: To illustrate the impact of combining 24 h recall (24hR) and FFQ estimates using regression calibration (RC) and enhanced regression calibration (ERC) on diet-disease associations. SETTING: Wageningen area, the Netherlands, 2011-2013. DESIGN: Five approaches for obtaining self-reported dietary intake estimates of protein and K were compared: (i) uncorrected FFQ intakes (FFQ); (ii) uncorrected average of two 24hR ( $\overline {\rm R}$ ); (iii) average of FFQ and $\overline {\rm R}$ ( ${\overline {\rm F}}\,\overline {\rm R}}$ ); (iv) RC from regression of 24hR v. FFQ; and (v) ERC by adding individual random effects to the RC approach. Empirical attenuation factors (AF) were derived by regression of urinary biomarker measurements v. the resulting intake estimates. PARTICIPANTS: Data of 236 individuals collected within the National Dietary Assessment Reference Database. RESULTS: Both FFQ and 24hR dietary intake estimates were measured with substantial error. Using statistical techniques to correct for measurement error (i.e. RC and ERC) reduced bias in diet-disease associations as indicated by their AF approaching 1 (RC 1·14, ERC 0·95 for protein; RC 1·28, ERC 1·34 for K). The larger sd and narrower 95% CI of AF obtained with ERC compared with RC indicated that using ERC has more power than using RC. However, the difference in AF between RC and ERC was not statistically significant, indicating no significantly better de-attenuation by using ERC compared with RC. AF larger than 1, observed for the ERC for K, indicated possible overcorrection. CONCLUSIONS: Our study highlights the potential of combining FFQ and 24hR data. Using RC and ERC resulted in less biased associations for protein and K.

Estimating disease prevalence from drug utilization data using the Random Forest algorithm.

BACKGROUND: Aggregated claims data on medication are often used as a proxy for the prevalence of diseases, especially chronic diseases. However, linkage between medication and diagnosis tend to be theory based and not very precise. Modelling disease probability at an individual level using individual level data may yield more accurate results. METHODS: Individual probabilities of having a certain chronic disease were estimated using the Random Forest (RF) algorithm. A training set was created from a general practitioners database of 276 723 cases that included diagnosis and claims data on medication. Model performance for 29 chronic diseases was evaluated using Receiver-Operator Curves, by measuring the Area Under the Curve (AUC). RESULTS: The diseases for which model performance was best were Parkinson's disease (AUC = .89, 95% CI = .77-1.00), diabetes (AUC = .87, 95% CI = .85-.90), osteoporosis (AUC = .87, 95% CI = .81-.92) and heart failure (AUC = .81, 95% CI = .74-.88). Five other diseases had an AUC >.75: asthma, chronic enteritis, COPD, epilepsy and HIV/AIDS. For 16 of 17 diseases tested, the medication categories used in theory-based algorithms were also identified by our method, however the RF models included a broader range of medications as important predictors. CONCLUSION: Data on medication use can be a useful predictor when estimating the prevalence of several chronic diseases. To improve the estimates, for a broader range of chronic diseases, research should use better training data, include more details concerning dosages and duration of prescriptions, and add related predictors like hospitalizations.

Validating fatty acid intake as estimated by an FFQ: how does the 24 h recall perform as reference method compared with the duplicate portion?

OBJECTIVE: To compare the performance of the commonly used 24 h recall (24hR) with the more distinct duplicate portion (DP) as reference method for validation of fatty acid intake estimated with an FFQ. DESIGN: Intakes of SFA, MUFA, n-3 fatty acids and linoleic acid (LA) were estimated by chemical analysis of two DP and by on average five 24hR and two FFQ. Plasma n-3 fatty acids and LA were used to objectively compare ranking of individuals based on DP and 24hR. Multivariate measurement error models were used to estimate validity coefficients and attenuation factors for the FFQ with the DP and 24hR as reference methods. SETTING: Wageningen, the Netherlands. SUBJECTS: Ninety-two men and 106 women (aged 20-70 years). RESULTS: Validity coefficients for the fatty acid estimates by the FFQ tended to be lower when using the DP as reference method compared with the 24hR. Attenuation factors for the FFQ tended to be slightly higher based on the DP than those based on the 24hR as reference method. Furthermore, when using plasma fatty acids as reference, the DP showed comparable to slightly better ranking of participants according to their intake of n-3 fatty acids (0·33) and n-3:LA (0·34) than the 24hR (0·22 and 0·24, respectively). CONCLUSIONS: The 24hR gives only slightly different results compared with the distinctive but less feasible DP, therefore use of the 24hR seems appropriate as the reference method for FFQ validation of fatty acid intake.

Limited salt consumption reduces the incidence of chronic kidney disease: a modeling study.

Background: In addition to blood pressure and cardiovascular disease, high-salt intake has been associated with renal diseases. The aim of this study is to estimate the potential health impact of salt reduction on chronic kidney disease (CKD) and end-stage kidney disease (ESKD) in the Netherlands. Methods: We developed a dynamic population health modeling tool to estimate the health impact of salt reduction on CKD and ESKD. We used data from the PREVEND study and extrapolated that to the Dutch population aged 30-75 years. We estimated the potential health impact of salt reduction comparing the current situation with the health impact of the adherence to the recommended maximum salt intake of 6 g/d. Results: In the recommended maximum intake scenario, a cumulative reduction in CKD of 1.1% (N = 290 000; interquartile range (IQR) = 249 000) and in ESKD of 3.2% (N = 470; IQR = 5080) would occur over a period of 20 years. Conclusions: Our health impact estimation showed that health benefits on CKD might be achieved when salt intake is reduced to the recommended maximum intake of 6 g/d.

BMI was found to be a consistent determinant related to misreporting of energy, protein and potassium intake using self-report and duplicate portion methods.

OBJECTIVE: As misreporting, mostly under-reporting, of dietary intake is a generally known problem in nutritional research, we aimed to analyse the association between selected determinants and the extent of misreporting by the duplicate portion method (DP), 24 h recall (24hR) and FFQ by linear regression analysis using the biomarker values as unbiased estimates. DESIGN: For each individual, two DP, two 24hR, two FFQ and two 24 h urinary biomarkers were collected within 1·5 years. Also, for sixty-nine individuals one or two doubly labelled water measurements were obtained. The associations of basic determinants (BMI, gender, age and level of education) with misreporting of energy, protein and K intake of the DP, 24hR and FFQ were evaluated using linear regression analysis. Additionally, associations between other determinants, such as physical activity and smoking habits, and misreporting were investigated. SETTING: The Netherlands. SUBJECTS: One hundred and ninety-seven individuals aged 20-70 years. RESULTS: Higher BMI was associated with under-reporting of dietary intake assessed by the different dietary assessment methods for energy, protein and K, except for K by DP. Men tended to under-report protein by the DP, FFQ and 24hR, and persons of older age under-reported K but only by the 24hR and FFQ. When adjusted for the basic determinants, the other determinants did not show a consistent association with misreporting of energy or nutrients and by the different dietary assessment methods. CONCLUSIONS: As BMI was the only consistent determinant of misreporting, we conclude that BMI should always be taken into account when assessing and correcting dietary intake.

Taking multi-morbidity into account when attributing DALYs to risk factors: comparing dynamic modeling with the GBD2010 calculation method.

BACKGROUND: Disability Adjusted Life Years (DALYs) quantify the loss of healthy years of life due to dying prematurely and due to living with diseases and injuries. Current methods of attributing DALYs to underlying risk factors fall short on two main points. First, risk factor attribution methods often unjustly apply incidence-based population attributable fractions (PAFs) to prevalence-based data. Second, it mixes two conceptually distinct approaches targeting different goals, namely an attribution method aiming to attribute uniquely to a single cause, and an elimination method aiming to describe a counterfactual situation without exposure. In this paper we describe dynamic modeling as an alternative, completely counterfactual approach and compare this to the approach used in the Global Burden of Disease 2010 study (GBD2010). METHODS: Using data on smoking in the Netherlands in 2011, we demonstrate how an alternative method of risk factor attribution using a pure counterfactual approach results in different estimates for DALYs. This alternative method is carried out using the dynamic multistate disease table model DYNAMO-HIA. We investigate the differences between our alternative method and the method used by the GBD2010 by doing additional analyses using data from a synthetic population in steady state. RESULTS: We observed important differences between the outcomes of the two methods: in an artificial situation where dynamics play a limited role, DALYs are a third lower as compared to those calculated with the GBD2010 method (398,000 versus 607,000 DALYs). The most important factor is newly occurring morbidity in life years gained that is ignored in the GBD2010 approach. Age-dependent relative risks and exposures lead to additional differences between methods as they distort the results of prevalence-based DALY calculations, but the direction and magnitude of the distortions depend on the particular situation. CONCLUSIONS: We argue that the GBD2010 approach is a hybrid of an attributional and counterfactual approach, making the end result hard to understand, while dynamic modelling uses a purely counterfactual approach and thus yields better interpretable results.

Anthropometry and the Risk of Lung Cancer in EPIC.

The associations of body mass index (BMI) and other anthropometric measurements with lung cancer were examined in 348,108 participants in the European Investigation Into Cancer and Nutrition (EPIC) between 1992 and 2010. The study population included 2,400 case patients with incident lung cancer, and the average length of follow-up was 11 years. Hazard ratios were calculated using Cox proportional hazard models in which we modeled smoking variables with cubic splines. Overall, there was a significant inverse association between BMI (weight (kg)/height (m)(2)) and the risk of lung cancer after adjustment for smoking and other confounders (for BMI of 30.0-34.9 versus 18.5-25.0, hazard ratio = 0.72, 95% confidence interval: 0.62, 0.84). The strength of the association declined with increasing follow-up time. Conversely, after adjustment for BMI, waist circumference and waist-to-height ratio were significantly positively associated with lung cancer risk (for the highest category of waist circumference vs. the lowest, hazard ratio = 1.25, 95% confidence interval: 1.05, 1.50). Given the decline of the inverse association between BMI and lung cancer over time, the association is likely at least partly due to weight loss resulting from preclinical lung cancer that was present at baseline. Residual confounding by smoking could also have influenced our findings.

A method for sensitivity analysis to assess the effects of measurement error in multiple exposure variables using external validation data.

BACKGROUND: Measurement error in self-reported dietary intakes is known to bias the association between dietary intake and a health outcome of interest such as risk of a disease. The association can be distorted further by mismeasured confounders, leading to invalid results and conclusions. It is, however, difficult to adjust for the bias in the association when there is no internal validation data. METHODS: We proposed a method to adjust for the bias in the diet-disease association (hereafter, association), due to measurement error in dietary intake and a mismeasured confounder, when there is no internal validation data. The method combines prior information on the validity of the self-report instrument with the observed data to adjust for the bias in the association. We compared the proposed method with the method that ignores the confounder effect, and with the method that ignores measurement errors completely. We assessed the sensitivity of the estimates to various magnitudes of measurement error, error correlations and uncertainty in the literature-reported validation data. We applied the methods to fruits and vegetables (FV) intakes, cigarette smoking (confounder) and all-cause mortality data from the European Prospective Investigation into Cancer and Nutrition study. RESULTS: Using the proposed method resulted in about four times increase in the strength of association between FV intake and mortality. For weakly correlated errors, measurement error in the confounder minimally affected the hazard ratio estimate for FV intake. The effect was more pronounced for strong error correlations. CONCLUSIONS: The proposed method permits sensitivity analysis on measurement error structures and accounts for uncertainties in the reported validity coefficients. The method is useful in assessing the direction and quantifying the magnitude of bias in the association due to measurement errors in the confounders.

Trend in and predictors for cardiovascular mortality in patients with rheumatoid arthritis over a period of 15 years: a prospective cohort study.

OBJECTIVES: To investigate a) the cardiovascular (CV) mortality in a clinical cohort of patients with established rheumatoid arthritis (RA) in comparison with the general population over 15 years, b) the trend in this CV mortality during the study period, and c) for a broad range of predictors, which baseline variables predict CV mortality. METHODS: In 1997, a sample of 1222 patients was randomly selected from the register of a rheumatology outpatient clinic in Amsterdam. Their CV mortality between 1997 and 2012 was obtained from Statistics Netherlands. The standardised mortality ratio (SMR) for CV mortality was calculated. A linear poisson regression analysis was performed to investigate if there was a trend in SMR over time. A Cox regression analysis was performed to determine which baseline variables predicted CV mortality. RESULTS: Mean age of the population at baseline was 60.4 (SD 15.4) years and 72.6% of the patients were women. Estimated SMR (95% confidence interval) for CV mortality was 1.24 (1.05, 1.43). The SMR decreased with 3% annually (p=0.16). Higher age, higher erythrocyte sedimentation rate, having CV comorbidity and diabetes mellitus (DM) were predictors for CV mortality. CONCLUSIONS: CV mortality among patients with RA in the past 15 years was still higher than in the general population. CV mortality decrease was not statistically significant. As CV mortality in RA is still higher than in the general population, continued attention for CV diseases in RA is important. Both tight control of disease activity and good care for comorbid conditions (CV diseases and DM) are advocated.

Potential health gains and health losses in eleven EU countries attainable through feasible prevalences of the life-style related risk factors alcohol, BMI, and smoking: a quantitative health impact assessment.

BACKGROUND: Influencing the life-style risk-factors alcohol, body mass index (BMI), and smoking is an European Union (EU) wide objective of public health policy. The population-level health effects of these risk-factors depend on population specific characteristics and are difficult to quantify without dynamic population health models. METHODS: For eleven countries-approx. 80 % of the EU-27 population-we used evidence from the publicly available DYNAMO-HIA data-set. For each country the age- and sex-specific risk-factor prevalence and the incidence, prevalence, and excess mortality of nine chronic diseases are utilized; including the corresponding relative risks linking risk-factor exposure causally to disease incidence and all-cause mortality. Applying the DYNAMO-HIA tool, we dynamically project the country-wise potential health gains and losses using feasible, i.e. observed elsewhere, risk-factor prevalence rates as benchmarks. The effects of the "worst practice", "best practice", and the currently observed risk-factor prevalence on population health are quantified and expected changes in life expectancy, morbidity-free life years, disease cases, and cumulative mortality are reported. RESULTS: Applying the best practice smoking prevalence yields the largest gains in life expectancy with 0.4 years for males and 0.3 year for females (approx. 332,950 and 274,200 deaths postponed, respectively) while the worst practice smoking prevalence also leads to the largest losses with 0.7 years for males and 0.9 year for females (approx. 609,400 and 710,550 lives lost, respectively). Comparing morbidity-free life years, the best practice smoking prevalence shows the highest gains for males with 0.4 years (342,800 less disease cases), whereas for females the best practice BMI prevalence yields the largest gains with 0.7 years (1,075,200 less disease cases). CONCLUSION: Smoking is still the risk-factor with the largest potential health gains. BMI, however, has comparatively large effects on morbidity. Future research should aim to improve knowledge of how policies can influence and shape individual and aggregated life-style-related risk-factor behavior.

Evaluation of a two-part regression calibration to adjust for dietary exposure measurement error in the Cox proportional hazards model: A simulation study.

Dietary questionnaires are prone to measurement error, which bias the perceived association between dietary intake and risk of disease. Short-term measurements are required to adjust for the bias in the association. For foods that are not consumed daily, the short-term measurements are often characterized by excess zeroes. Via a simulation study, the performance of a two-part calibration model that was developed for a single-replicate study design was assessed by mimicking leafy vegetable intake reports from the multicenter European Prospective Investigation into Cancer and Nutrition (EPIC) study. In part I of the fitted two-part calibration model, a logistic distribution was assumed; in part II, a gamma distribution was assumed. The model was assessed with respect to the magnitude of the correlation between the consumption probability and the consumed amount (hereafter, cross-part correlation), the number and form of covariates in the calibration model, the percentage of zero response values, and the magnitude of the measurement error in the dietary intake. From the simulation study results, transforming the dietary variable in the regression calibration to an appropriate scale was found to be the most important factor for the model performance. Reducing the number of covariates in the model could be beneficial, but was not critical in large-sample studies. The performance was remarkably robust when fitting a one-part rather than a two-part model. The model performance was minimally affected by the cross-part correlation.

Comparison of duplicate portion and 24 h recall as reference methods for validating a FFQ using urinary markers as the estimate of true intake.

As FFQ are subject to measurement error, associations between self-reported intake by FFQ and outcome measures should be adjusted by correction factors obtained from a validation study. Whether the correction is adequate depends on the characteristics of the reference method used in the validation study. Preferably, reference methods should (1) be unbiased and (2) have uncorrelated errors with those in the FFQ. The aim of the present study was to assess the validity of the duplicate portion (DP) technique as a reference method and compare its validity with that of a commonly used reference method, the 24 h recall (24hR), for protein, K and Na using urinary markers as the unbiased reference method. For 198 subjects, two DP, two FFQ, two urinary biomarkers and between one and fifteen 24hR (web based and/or telephone based) were collected within 1·5 years. Multivariate measurement error models were used to estimate bias, error correlations between FFQ and DP or 24hR, and attenuation factors of these methods. The DP was less influenced by proportional scaling bias (0·58 for protein, 0·72 for K and 0·52 for Na), and correlated errors between DP and FFQ were lowest (protein 0·28, K 0·17 and Na 0·19) compared with the 24hR. Attenuation factors (protein 0·74, K 0·54 and Na 0·43) also indicated that the DP performed better than the 24hR. Therefore, the DP is probably the best available reference method for FFQ validation for nutrients that currently have no generally accepted recovery biomarker.

Prediction of fruit and vegetable intake from biomarkers using individual participant data of diet-controlled intervention studies.

Fruit and vegetable consumption produces changes in several biomarkers in blood. The present study aimed to examine the dose-response curve between fruit and vegetable consumption and carotenoid (α-carotene, ß-carotene, ß-cryptoxanthin, lycopene, lutein and zeaxanthin), folate and vitamin C concentrations. Furthermore, a prediction model of fruit and vegetable intake based on these biomarkers and subject characteristics (i.e. age, sex, BMI and smoking status) was established. Data from twelve diet-controlled intervention studies were obtained to develop a prediction model for fruit and vegetable intake (including and excluding fruit and vegetable juices). The study population in the present individual participant data meta-analysis consisted of 526 men and women. Carotenoid, folate and vitamin C concentrations showed a positive relationship with fruit and vegetable intake. Measures of performance for the prediction model were calculated using cross-validation. For the prediction model of fruit, vegetable and juice intake, the root mean squared error (RMSE) was 258.0 g, the correlation between observed and predicted intake was 0.78 and the mean difference between observed and predicted intake was - 1.7 g (limits of agreement: - 466.3, 462.8 g). For the prediction of fruit and vegetable intake (excluding juices), the RMSE was 201.1 g, the correlation was 0.65 and the mean bias was 2.4 g (limits of agreement: -368.2, 373.0 g). The prediction models which include the biomarkers and subject characteristics may be used to estimate average intake at the group level and to investigate the ranking of individuals with regard to their intake of fruit and vegetables when validating questionnaires that measure intake.