Use of self-reported measures of height, weight and body mass index in a rural population of Northeast Brazil.

OBJECTIVE To assess the validity of using self-reported anthropometric data for diagnosis of nutritional status of adults in a rural population of northeast Brazil. METHODS A population-based survey was conducted on a sample of 797 individuals aged 18 years or more. The proportion of individuals who knew their anthropometric measures was calculated. For agreement analysis between those who reported their measures the following indicators were obtained: differences between averages (weight, height, body mass index), intra-class correlation coefficient (ICC), Kappa statistic, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (VPN). Bland-Altman graphics were also obtained. RESULTS More than half of the respondents (58.5%) did not know their weight or height. Weight was the most known measure among all. The magnitude of the mean difference for weight, height and body mass index (BMI) (0.43 kg, 0.31 cm, 0.32 kg/m2, respectively) was small, indicating good agreement, with a trend toward overestimation. ICC for weight, height and BMI were 0.96; 0.60; and 0.53, respectively. Kappa statistic indicated good agreement in all strata. General measures of sensitivity, specificity, PPV and NPV were 84.2; 82; 90.7 and 71.3%, respectively. Elderly, those with low schooling and those who do not often weigh were less accurate on their measures. CONCLUSION The use of self-reported measures should be done with caution in epidemiological studies in rural populations.