Self-reported height and weight are common in population surveillance but are prone to systematic misreporting that biases body mass index (BMI) estimates and downstream indicators such as obesity prevalence and socioeconomic inequalities. This study aimed to identify determinants of misreporting, build simple calibration equations for height and weight, and evaluate how measured, self-reported and calibrated BMI influence estimates of obesity prevalence and socioeconomic disparities in Portugal.
The analysis used survey-weighted, sex-stratified data from 3,404 adults aged 18–64 years who participated in the Portuguese National Food, Nutrition and Physical Activity Survey (IAN-AF 2015–2016). The dataset included both self-reported and directly measured anthropometry, enabling comparison and model development.
Misreporting was quantified by comparing self-reported and measured height and weight. The authors report that height tended to be overreported while weight was underreported. Misreporting magnitude increased with age and with measured BMI. Determinants of misreporting were assessed using multinomial logistic regression; specific model coefficients and p values are reported in the original manuscript (details not restated here because the source provides summary findings rather than full coefficient tables in the abstract).
Calibration equations for height and weight were estimated using measured values as outcomes and including self-reported values plus key sociodemographic predictors: age, region of residence and education level. Models were stratified by sex and used survey weights to reflect the complex survey design. The calibrated predictions for height and weight were subsequently combined to derive a calibrated BMI for each participant.
Calibrated BMI was computed from predicted calibrated height and weight. Obesity prevalence (BMI ≥ 30 kg/m2) was estimated separately using measured BMI, self-reported BMI and calibrated BMI. Socioeconomic inequalities in obesity were compared across BMI assessment methods using multiple metrics: absolute and relative prevalence differences, the slope index of inequality and the relative index of inequality for indicators including education, income and employment.
Using self-reported anthropometry led to underestimation of obesity prevalence because of systematic overreporting of height and underreporting of weight. Applying simple sociodemographic calibration equations eliminated this underestimation: calibrated BMI estimates of obesity prevalence were close to those based on measured BMI, reducing misclassification from self-reports.
The effect of calibration on measured socioeconomic inequalities varied by sex. For education-related inequalities, calibration widened disparities among women, indicating that self-reported BMI had previously masked part of the education gradient in obesity for women. In contrast, among men calibration corrected an overestimation of education-related inequalities observed when relying on self-reports. Patterns of inequality by income and employment were reported as similar across measured, self-reported and calibrated BMI methods.
Ethical approval was provided by the Ethics Committee/IRB of the Institute of Public Health of the University of Porto. The authors state that all necessary participant consent was obtained and that relevant reporting guidelines were followed. All data produced in the study are available upon reasonable request to the authors. Declared funders include EEA and Norway Grants and Fundação para a Ciência e Tecnologia. The authors declared no competing interests.
Among Portuguese adults aged 18–64, self-reported height and weight are systematically misreported in patterns that are socially patterned by age, BMI and education. This misreporting biases obesity prevalence estimates and can distort the measured socioeconomic distribution of obesity. Simple calibration models that incorporate self-reports plus basic sociodemographic variables can correct the underestimation of obesity prevalence and alter inequality estimates in sex-specific ways. The authors suggest that routine integration of such calibration procedures into national surveys could improve the validity and equity of obesity surveillance. Because this is a medRxiv preprint, results should be interpreted as preliminary until peer review and external validation are completed.