Abstracts

INTRODUCTION

Overweight and obesity are increasing threats to the health of the world population and have been associated with several chronic diseases such as cardiovascular diseases, cancer, and diabetes¹1. World Health Organization (WHO). Obesity: preventing and managing the global epidemic. Report of a WHO Consultation (WHO Technical Report Series 894). Geneva: WHO; 2000.. In Brazil, the prevalence of overweight now reaches 50% of adults (20 - 59 years), 22% of adolescents (10 - 19 years) and 34% of children (5 - 9 years)²2. Instituto Brasileiro de Geografia e Estatística (IBGE). Indicadores IBGE: Pesquisa de orçamentos familiares 2008-2009: antropometria, estado nutricional de crianças, adolescentes e adultos no Brasil. Rio de Janeiro: IBGE; 2010.. This fact has motivated researchers to investigate factors associated (gender, age, education, income, lifestyle, etc.) to chronic diseases and identify groups at highest risk. Thus, the use of self-reported data and the use of self-administered questionnaires has been a common practice, especially in studies involving large samples³3. César CLG, Carandina L, Alves MCGP, Barros MBA, Goldbaum M. Saúde e condição de vida em São Paulo: inquérito multicêntrico de saúde no estado de São Paulo (ISA-SP). São Paulo: FSP/USP; 2005. 212 p.. This form of data collection is justified by its operational and logistical facility, in addition to the cost savings involved in training of personnel, transportation and procurement of equipment for measuring⁴4. Peixoto MRG, Benício MHD, Jardim PCBV. Validade do peso e da altura autorreferidos: o estudo de Goiânia. Rev Saúde Pública 2006; 40(6): 1065-72..

Self-reported measures of height and weight have been frequently used⁵5. Health Education Authority (HEA). Today's young adults: 16-19 year olds look at diet, alcohol, drugs and sexual behavior. London: HEA; 1992.

6. Strauss RS. Comparison of measured and self-reported weight and height in a cross-sectional sample of young adolescents. Int J Obes Relat Metab Disord 1999; 23(8): 904-8.

7. Center for Disease Control and Prevention (CDC). Youth risk behavior surveillance, United States - 2003. Surveillance summaries. Washington: CDC; 2004. ^{- 8}8. Virtuoso-Junior JS, Oliveira-Guerra R. Validade concorrente do peso e estatura auto-referidos no diagnóstico do estado nutricional em mulheres idosas. Rev Salud Pública 2010; 12(1): 71-81., However, it is common to find changes in the validity of the measurements according to characteristics of individuals, such as gender, age, nutritional status and socioeconomic (family income, education), cultural (valuation of thinness) and psychological (body satisfaction) conditions⁹9. Fonseca MJM, Faerstein E, Chor D, Lopes CS. Validade de peso e estatura informados e índice de massa corporal: estudo pró-saúde. Rev Saúde Pública 2004; 38(3): 392-8.

10. Pregnolato TS, Mesquita LM, Ferreira PG, Santos MM, Santos CC, Costa RF. Validade de medidas autorreferidas de massa e estatura e seu impacto na estimativa do estado nutricional pelo índice de massa corporal. Rev Bras Crescimento Desenvolvimento Hum 2009; 19(1): 35-41. ^{- 11}11. Silveira EA, Araújo CL, Gigante DP, Barros AJD, Lima MS. Validação do peso e altura referidos para o diagnóstico do estado nutricional em uma população de adultos no Sul do Brasil. Cad Saúde Pública 2005; 21(1): 235-45..

In the national literature, there are some studies on validation of height and weight measurements that show a greater tendency of underestimation of weight in obese individuals, adolescents and women, and overestimation of height in individuals of short stature, women, teens and seniors, which can generate a spurious estimate of body mass index (BMI), with a reduction of individuals at the extremes of the rating curve, compromising the validity of these measures⁴4. Peixoto MRG, Benício MHD, Jardim PCBV. Validade do peso e da altura autorreferidos: o estudo de Goiânia. Rev Saúde Pública 2006; 40(6): 1065-72. ^{, 12}12. Chor D, Coutinho ESF, Laurenti R. Reability of self-reported weight and among State bank employees. Rev Saúde Pública 1999; 33(1): 16-23.

13. Farias Júnior JC. Validade das medidas auto-referidas de peso e estatura para o diagnostico do estado nutricional de adolescentes. Rev Bras Saúde Matern Infant 2007; 7(2): 167-74. ^{- 14}14. Del Duca GF, González-Chica DA, Santos JV; Knuth AG, Camargo MBJ, Araújo CL. Peso e altura autorreferidos para determinação do estado nutricional de adultos e idosos: validade e implicações em análises de dados. Cad Saúde Pública 2012; 28(1): 75-85.. To correct these potential errors, methodological strategies, such as using equations for calibration of the data, can be employed in order to bring the measure provided closer to the actual values. However, few studies develop such approaches.

Thus, the present study aims to assess the validity of self-reported data on weight and height of adolescents, adults and elderly residents of São Paulo, participants of Health Survey of São Paulo (ISA-Capital 2008), as well as to assess the agreement between classifications of nutritional status by self-reported and measured data, and to present calibration coefficients for correction of data on weight, height and BMI.

METHODS

This study is part of the Health Survey of São Paulo (ISA-Capital), a cross-sectional, population-based study conducted in São Paulo between 2008 and 2010. A sample of ISA-Capital was obtained by complex probabilistic sampling, by conglomerates, in two stages: census tracts and households (n = 1,662). The planning was done to estimate the proportions of 50% (p = 0.50, which corresponds to the largest minimum sample size for estimating proportions) with sampling error of 7 percentage points (d = 0.07) with a confidence level 95% and 1.5 design effects. Inclusion criteria of the ISA-Capital study were: being a resident of the selected household in the urban area of the state capital, belonging to the domains of interest (adolescents, adults and elderly of both sexes), not being pregnant. Details can be found in another publication¹⁵15. Fisberg RM, Marchioni DML. Manual para estudos populacionais de alimentação, nutrição e saúde: a experiência do inquérito de saúde em São Paulo (ISA). Grupo de Avaliação de Consumo Alimentar (GAC). São Paulo: FSP/USP; 2012..

In 2008, data on food habits and socioeconomic factors of the ISA-Capital study participants were collected through home visits. A year later, the team returned to the individuals' home to collect data such as measured weight and height, and other measures of interest (blood collection, blood pressure, medication use, among others) by a previously trained nurse. This second home visit for anthropometric measurements was confirmed by phone a few days before. Upon confirmation, the participant was asked about their height and weight measurements (self-reported), and dietary data were also collected.

During this period between the collections there was a significant sample loss, totaling 832 individuals who had reported anthropometric data and 750 who had and measured anthropometric data. However, this loss occurred randomly in all census tracts, thus reducing the possibility of bias by differential loss.

Some participants were not found at the time of the phone call (before the home visit). In such cases, this and other information of interest to the study (dietary and lifestyle data) were collected after the home visit and assessment of the measures. These latter individuals were excluded only from the present study because their data were measured prior to the time specified for data collection and may influence the outcome of the study.

Thus, for the present study, only individuals who had all their measurements collected and whose and self-reported measurements were collected before the measured values were selected, totaling a subsample from ISA-Capital of 299 individuals (112 men and 187 women, 62 adolescents, 107 adults and 130 elderly). However, no difference between schooling of head of household, age, sex and nutritional status between the study sample and the sample of ISA-Capital was observed.

For weight measurement, an electronic platform scale with capacity for 150 kg and sensitivity of 100 g (TANITA^® was used. Subjects were weighed wearing light clothes, barefoot with erect posture, parallel feet and fully supported in the balance platform and with arms along the body¹⁶16. Lohman TG, Roche AF, Martorell R. Anthropometric Standardization Reference Manual. Champaign: Human Kinetics; 1988..

For height measurement, a stadiometer with scale in millimeters was used (Seca Bodymeter 208^®, fixed to the wall. Subjects stood up with erect posture, feet together and heels touching the wall. The apex of the ear and the outer corner of the eye remained in parallel to the ground, forming an angle of 90 degrees to the bar of the stadiometer. The horizontal bar of the stadiometer was lowered and laid on the head, allowing reading in centimeters16.

The self-reported data were obtained by the questions: "What is your weight?" And "How tall are you?".

From the measurements of weight and height, measured and reported BMI (weight/height²2. Instituto Brasileiro de Geografia e Estatística (IBGE). Indicadores IBGE: Pesquisa de orçamentos familiares 2008-2009: antropometria, estado nutricional de crianças, adolescentes e adultos no Brasil. Rio de Janeiro: IBGE; 2010.) were calculated and classified according to the ranges suggested by Cole et al.¹⁷17. Cole TJ, Flegal KM, Nicholls D, Jackson AA. Body mass index cut offs to define thinness in children and adolescents: international survey. BMJ 2007; 335(7612): 194-7. for teens, by the World Health Organization¹1. World Health Organization (WHO). Obesity: preventing and managing the global epidemic. Report of a WHO Consultation (WHO Technical Report Series 894). Geneva: WHO; 2000. for adults and by the Nutrition Screening Initiative for seniors.

The project was approved by the Research Ethics Committee of the School of Public Health, Universidade de São Paulo. All participants signed a free and informed consent form. There are no conflicts of interest.

Statistical analyses

Adherence of variables to normal distribution was checked by the Kolmogorov-Smirnov test, and presented normal distribution.

The intraclass correlation coefficients were used to assess the validity and relationship between reported and measured weight, height and BMI according to the categories of gender (male and female) and age (adolescent: 12 - 19 years, adult: 20 - 59; elderly: 60 years or older). This coefficient evaluates the correlation between the groups, considering the interpersonal variability, that is, the systematic under- or overestimation within the group¹⁹19. Szklo M, Javier Nieto F. Epidemiology: beyond the basics. Am J Epidemiol 2001; 153(8): 821-2.. Paired t-test was used to assess the differences between the means of reported and measured for each gender and age group.

The agreement between the measurements of weight, height and BMI was assessed by the strategy proposed by Bland and Altman²⁰20. Bland JM, Altman DG. Comparing methods of measurement: why plotting difference against standard method is misleading. Lancet 1995; 346(8982): 1085-7., which includes the construction of a graph of agreement (mean versus agreement) and the calculation of the limits of agreement. By this technique, the magnitude of differences can be assessed for 95% of the observations.

Calibration coefficients were obtained by linear regression models, in which the reference pattern (measured) is modeled as a function of the self-reported measure²¹21. Kaaks R, Riboli E. Validation and calibration of dietary intake measurements in the EPIC project: methodological considerations. European Prospective Investigation into Cancer and Nutrition. Int J Epidemiol 1997; 26(Suppl 1): S15-25.

22. Stram DO, Hankin JH, Wilkens LR, Pike MC, Monroe KR, Park S, et al. Calibration of the dietary questionnaire for a multiethnic cohort in Hawaii and Los Angeles. Am J Epidemiol 2000; 151(4): 358-70. ^{- 23}23. Teixeira JA, Baggio ML, Fisberg RM, Marchioni DM. Calibration of the dietary data obtained from the Brazilian center of the Natural History of HPV Infection in Men study: the HIM Study. Cad Saúde Pública 2010; 26(12): 2323-33.. Thus, the coefficients of attenuation can be estimated as the slope of the regression line of the reference values (measurements taken) in the observed data (self-reported measures). Calibration coefficients were estimated by gender and age group.

The use of calibration is important for extrapolating data; studies in the same population only with information on self-reported measures might use the data of calibration coefficients to improve estimates of means and confidence intervals for weight, height and BMI, decreasing the error in using the self-reported measure. This can be done by means of the equation: y = B₀ + B₁x, where y refers to the calibrated measure, x is the self-reported measure and B₁ and represents the increase in the calibrated measure for each unit of the self-reported measure.

We calculated sensitivity, specificity and positive predictive value (PPV) of nutritional status, obtained from the BMI classifications (with and without excess weight), calculated from the reported measurements, taking as gold standard the BMI calculated from the measured values. The sensitivity and specificity were analyzed according to gender and age.

The proportions of individuals with excess weight were calculated from seld-reported, measured and calibrated data, and the proportion test was performed to verify the difference between measured and reported data, and between measured and calibrated data.

The sampling design of the ISA-Capital was not used in this study. The power of the study was verified for each domain of interest (adolescents, adults and elderly separated by gender), according to Reichenheim24, who takes into account the confidence interval, Kappa, proportions and accuracy.

RESULTS

A total of 299 individuals living in the city of São Paulo (112 men and 187 women) were evaluated, 21% adolescents, 36% adults and 43% seniors. It was observed that 55% of the population had schooling of head of household with eight or more years of study. About 52% of people were overweight, and the prevalence was 36% for adolescents, 46% for adults and 64% for the elderly. This sample showed no statistical difference compared to the initial study sample by gender, age, education of household head and nutritional status (data not shown).

There was a high correlation (r > 0.70) between reported and measured weight, height and BMI, except for height in older women and adolescents, and adult men. There was no statistically significant difference between the mean reported and measured weight, height and BMI, except for the height of elderly females (Table 1). The Bland and Altman20 analysis showed a good agreement between reported and measured weight and height, and the mean difference in weight (self-reported measure minus measure taken) was higher in adolescent males (1.11 kg). The average height difference was greater in elderly females (0.04 m) (Figures 1 and 2).

It was found that women tend to underestimate their weight and men (except adults) tend to overestimate it. Elderly individuals tend to overestimate height, while teenagers tend to underestimate it. Means, confidence intervals and intraclass correlation coefficients by gender and age group are presented in Table 1.

It is noteworthy that the correlation coefficients for height are lower in all age groups and gender. However, most are even greater than 0.40, which is considered a good value according to Fleiss²⁵25. Fleiss J. The design and analysis of clinical experimentsed. New York: Wiley; 1986..

Calibration coefficients are presented in Table 2. Results related to height present lower coefficients of the regression models, especially among elderly women.

No significant difference was found in nutritional status according to measured and reported measurements, and according to measured and calibrated measures (Table 3). High sensitivity (> 91%) and specificity (> 83%) were found in all age groups and gender for BMI calculated based on reported data. There was also PPV > 82%, representing that 82% of individuals classified as overweight from reported measures are actually are overweight (Table 3).

DISCUSSION

This study is the first to assess the feasibility of using self-reported measures of height and weight in a subsample of a population-based study with a representative sample of adolescents, adults and elderly residents of São Paulo, as well as presenting calibration coefficients for correction of self-reported data.

The results of this study are better than those reported in the literature, as there was a good correlation between measurements of weight and height for the calculation of BMI in all age groups and gender⁴4. Peixoto MRG, Benício MHD, Jardim PCBV. Validade do peso e da altura autorreferidos: o estudo de Goiânia. Rev Saúde Pública 2006; 40(6): 1065-72. ^{, 9}9. Fonseca MJM, Faerstein E, Chor D, Lopes CS. Validade de peso e estatura informados e índice de massa corporal: estudo pró-saúde. Rev Saúde Pública 2004; 38(3): 392-8. ^{, 13}13. Farias Júnior JC. Validade das medidas auto-referidas de peso e estatura para o diagnostico do estado nutricional de adolescentes. Rev Bras Saúde Matern Infant 2007; 7(2): 167-74.. Differences observed between reported and measured weight were not significant for either gender and age groups, that is, were proved void. Some of the factors that could affect these results are the frequency with which individuals weigh themselves, the date of last measurement, type of clothing and footwear used, excessive preoccupation with body image and dissatisfaction with weight⁴4. Peixoto MRG, Benício MHD, Jardim PCBV. Validade do peso e da altura autorreferidos: o estudo de Goiânia. Rev Saúde Pública 2006; 40(6): 1065-72. ^{, 9}9. Fonseca MJM, Faerstein E, Chor D, Lopes CS. Validade de peso e estatura informados e índice de massa corporal: estudo pró-saúde. Rev Saúde Pública 2004; 38(3): 392-8.. In addition, access to and frequency of use of the primary health care network in São Paulo may have influenced the higher frequency of measurement of weight, bringing the self-reported value closer to the real one²⁶26. São Paulo. Prefeitura de São Paulo. Secretaria Municipal de Saúde. Estratégia Saúde da Família. Atenção Básica. Disponível em: http://www.prefeitura.sp.gov.br/cidade/secretarias/saude/atencao_basica/esf/. (Acessado em 28 de setembro de 2013).
http://www.prefeitura.sp.gov.br/cidade/s... .

It was observed that older women overestimated their height, in accordance with what Del Duca et al¹⁴14. Del Duca GF, González-Chica DA, Santos JV; Knuth AG, Camargo MBJ, Araújo CL. Peso e altura autorreferidos para determinação do estado nutricional de adultos e idosos: validade e implicações em análises de dados. Cad Saúde Pública 2012; 28(1): 75-85. observed in the elderly of both sexes of Pelotas, Rio Grande do Sul. This fact is possibly due to the low frequency of measurement and the natural reduction in height due to the compression of the intervertebral discs²⁷27. Brasil. Ministério da Saúde. Cadernos de Atenção Básica - Envelhecimento e saúde da pessoa idosa. Brasília: Ministério da Saúde; 2006..

It was also found that the majority of calibration coefficients was close to one, indicating that these self-reported measures have good equivalence with the measurements taken²⁸28. Lucca A, Moura EC. Validity and reliability of self-reported weight, height and body mass index from telephone interviews. Cad Saúde Pública 2010; 26(1): 110-22. ^{, 29}29. Kynast-Wolf G, Becker N, Kroke A, Brandstetter BR, Wahrendorf J, Boeing H. Linear regression calibration: theoretical framework and empirical results in EPIC, Germany. Ann Nutr Metab 2002; 46(1): 2-8.. The use of calibration coefficients can help in a more reliable prediction in studies with only self-reported measures. However, the coefficient of height proved to be distant from one, so the use of self-reported height, especially among the elderly, should be done with caution.

The prevalence of overweight according to the self-reported or calibrated measure was statistically the same as measured, showing that the use of such measures for nutritional diagnosis of overweight is valid. The prevalence from the calibrated measurements tended to be mostly closer to the measured value compared with the prevalence from the self-reported results. Thus, the use of calibration coefficients presented in this study as enhancement factors may favor the accuracy of association measurements obtained in epidemiological studies when arising from self-reported anthropometric weight and height measurements.

The sample size in stratified analyzes led to the reduction of the power of the study. However, can still be considered that the results of mean difference and correlation were homogeneous in their majority, no changes were found in BMI classifications in this study, which can be seen from the values of sensitivity, specificity and positive predictive value.

It is also noteworthy that, even though this study has a small sample, this resembles the original study population of ISA-Capital in socioeconomic data and nutritional status.

CONCLUSION

We conclude that the method of self-reported measures showed good validity with measures taken for most of the population studied and showed high sensitivity and specificity. Height measurements should be used with caution. Calibration coefficients presented can be used to improve BMI estimates and overweight prevalence in this study population. These data are important as they enable future studies to be developed with economy of resources and simplified field work.

REFERENCES

Publication Dates

History