ABSTRACT:

Introduction:

The quality control of data entry in computerized questionnaires is an important step in the validation of new instruments. The study assessed the consistency of recorded weight and height on the Food Intake and Physical Activity of School Children (Web-CAAFE) between repeated measures and against directly measured data.

Methods:

Students from the 2^nd to the 5^th grade (n = 390) had their weight and height directly measured and then filled out the Web-CAAFE. A subsample (n = 92) filled out the Web-CAAFE twice, three hours apart. The analysis included hierarchical linear regression, mixed linear regression model, to evaluate the bias, and intraclass correlation coefficient (ICC), to assess consistency. Univariate linear regression assessed the effect of gender, reading/writing performance, and computer/internet use and possession on residuals of fixed and random effects.

Results:

The Web-CAAFE showed high values of ICC between repeated measures (body weight = 0.996, height = 0.937, body mass index - BMI = 0.972), and regarding the checked measures (body weight = 0.962, height = 0.882, BMI = 0.828). The difference between means of body weight, height, and BMI directly measured and recorded was 208 g, -2 mm, and 0.238 kg/m², respectively, indicating slight BMI underestimation due to underestimation of weight and overestimation of height. This trend was related to body weight and age.

Conclusion:

Height and weight data entered in the Web-CAAFE by children were highly correlated with direct measurements and with the repeated entry. The bias found was similar to validation studies of self-reported weight and height in comparison to direct measurements.

Keywords:
Surveys and questionnaires; Body height; Body weight; Body mass index; Child; Adolescent

INTRODUCTION

The search for the development of questionnaires applied by using computers in epidemiological studies is growing. There are instruments addressed to children and adolescents, aiming at assessing dietary intake¹1. Diep CS, Hingle M, Chen TA, Dadabhoy HR, Beltran A, Baranowski J, et al. The automated self-administered 24-hour dietary recall for children, 2012 version, for youth aged 9 to 11 years: a validation study. J Acad Nutr Diet 2015; 115(10): 1591-8. DOI: 10.1016/j.jand.2015.02.021
https://doi.org/10.1016/j.jand.2015.02.0... ^,22. Medin AC, Astrup H, Kåsin BM, Andersen LF. Evaluation of a web-based food record for children using direct unobtrusive lunch observations: a validation study. J Med Internet Res 2015; 17(12): e273. DOI: 10.2196/jmir.5031
https://doi.org/10.2196/jmir.5031... ^,33. Carvalho MA, Baranowski T, Foster E, Santos O, Cardoso B, Rito A, Pereira MJ. Validation of the Portuguese self-administered computerised 24-hour dietary recall among second-, third and fourth-grade children. J Hum Nutr Diet 2015; 28(6): 666-74. DOI: 10.1111/jhn.12280
https://doi.org/10.1111/jhn.12280... ^,44. Biltoft-Jensen A, Bysted A, Trolle E, Christensen T, Knuthsen P, Damsgaard CT, et al. Evaluation of web-based dietary assessment software for children: comparing reported fruit, juice and vegetable intakes with plasma carotenoid concentration and school lunch observations. Br J Nutr 2013; 110(1): 186-95. DOI: 10.1017/S0007114512004746
https://doi.org/10.1017/S000711451200474... ^,55. Foster E, Hawkins A, Delve J, Adamson AJ. Reducing the cost of dietary assessment: self-completed recall and analysis of nutrition for use with children (SCRAN24). J Hum Nutr Diet 2014; 27(1): 26-35. DOI: 10.1111/jhn.12108
https://doi.org/10.1111/jhn.12108... , physical activity⁶6. Legnani E, Legnani RF, Rech CR, Guimarães RF, Campos W. Instrumentos eletrônicos para avaliar atividade física em crianças: uma revisão sistemática. Motricidade 2013; 9(4): 90-9., or multiple constructs, including nutritional status based on body mass index (BMI)⁷7. Storey KE, McCargar LJ. Reliability and validity of Web-SPAN, a web-based method for assessing weight status, diet and physical activity in youth. J Hum Nutr Diet 2012; 25(1): 59-68. DOI: 10.1111/j.1365-277X.2011.01181.x
https://doi.org/10.1111/j.1365-277X.2011... ^,88. Bae J, Joung H, Kim JY, Kwon KN, Kim Y, Park SW. Validity of self-reported height, weight, and body mass index of the Korea youth risk behavior web-based survey questionnaire. J Prev Med Public Health 2010; 43(5): 396-402. DOI: 10.3961/jpmph.2010.43.5.396
https://doi.org/10.3961/jpmph.2010.43.5.... ^,99. McLure SA, Reilly JJ, Crooks S, Summerbell CD. Development and evaluation of a novel computer-based tool for assessing physical activity levels in schoolchildren. Pediatr Exerc Sci 2009; 21(4): 506-19.^,1010. Moore HJ, Ells LJ, McLure SA, Crooks S, Cumbor D, Summerbell CD, et al. The development and evaluation of a novel computer program to assess previous-day dietary and physical activity behaviours in school children: the synchronised nutrition and activity program (SNAP). Br J Nutr 2008; 99(6): 1266-74. DOI: 10.1017/S0007114507862428
https://doi.org/10.1017/S000711450786242... .

Computerized questionnaires have advantages in comparison to traditional printed tools, since they reduce the costs with the reproduction of forms, allow obtaining data from large samples in several locations, simultaneously¹¹11. Rhodes S, Bowie D, Hergnrather K. Collecting behavioural data using the world wide web: considerations for researchers. J Epidemiol Community Health 2003; 57(1): 68-73. DOI: 10.1136/jech.57.1.68
https://doi.org/10.1136/jech.57.1.68... and reduce potential biases¹²12. Larson MR. Social desirability and self-reported weight and height. Int J Obes 2000; 24(5): 663-5. by eliminating the data entry stage¹³ and by providing more anonymity and privacy for the participant¹⁴14. Supple AJ, Aquilino WS, Wright DL. Collecting sensitive self-report data with laptop computers: impact on the response tendencies of adolescents in a home interview. J Res Adolescence 1999; 9(4): 467-88.^,1515. Weeb PM, Zimet GD, Fortenberry JD, Blythe MJ. Comparability of a computer-assisted versus written method for collecting health behavior information from adolescent patients. J Adolesc Health 1999; 24(6): 383-8..

The Questionnaire Dietary Intake and Physical Activity of Students (Web-CAAFE)¹⁶16. Costa FF. Desenvolvimento e avaliação de um questionário baseado na web para avaliar o consumo alimentar e a atividade física de escolares [tese]. Florianópolis - SC: Universidade Federal de Santa Catarina; 2013. was carried out for a monitoring system focusing on dietary intake and physical activities of students aged from 7 to 10 years.

In studies conducted with students from the public elementary school in Florianópolis, Santa Catarina, the Web-CAAFE had proper usability¹⁷17. Costa FF, Schmoelz CP, Davies VF, Di Pietro PF, Kupek E, Assis MA. Assessment of diet and physical activity of brazilian schoolchildren: usability testing of a web-based questionnaire. JMIR Res Protoc 2013; 2(2): e31. DOI: 10.2196/resprot.2646
https://doi.org/10.2196/resprot.2646... and validity in the evaluation of food intake¹⁸18. Davies VF, Kupek E, Assis MA, Natal S, Di Pietro PF, Baranowski T. Validation of a web-based questionnaire to assess the dietary intake of Brazilian children aged 7 - 10 years. J Hum Nutr Diet 2015; 28(1): 93-102. DOI: 10.1111/jhn.12262
https://doi.org/10.1111/jhn.12262... , and it also proved to be a viable instrument to evaluate the fulfillment of nutritional recommendations¹⁹19. Kupek E, Assis MA, Bellisle F, Lobo AS. Validity of WebCAAFE questionnaire for assessment of schoolchildren's dietary compliance with Brazilian Food Guidelines. Public Health Nutr 2016; 19(13): 2347-56. DOI: 10.1017/S1368980016000732
https://doi.org/10.1017/S136898001600073... . The instrument also had the proper validity and reproducibility in the evaluation of physical activities and dietary intake of students in Feira de Santana, Bahia²⁰20. Jesus GM, Assis MA, Kupek E, Dias LA. Avaliação da atividade física de escolares com um questionário via internet. Rev Bras Med Esporte 2016; 22(4): 261-6. DOI: 10.1590/1517-869220162204157067
https://doi.org/10.1590/1517-86922016220... ^,2121. Jesus GM, Assis MA, Kupek E. Validade e reprodutibilidade de questionário baseado na internet (Web-CAAFE) para avaliação do consumo alimentar de escolares de 7 a 15 anos. Cad Saúde Pública 2017; 33(5): e00163016. DOI: 10.1590/0102-311X00163016
https://doi.org/10.1590/0102-311X0016301... .

In the conception of Web-CAAFE, fields to fill out weight and height were also included to provide data to calculate the BMI and the diagnosis of nutritional status, by age and sex, using the reference curve from the World Health Organization (WHO)²²22. Onis M, Onyango AW, Borghi E, Siyam A, Nishida C, Siekmann J. Development of a WHO growth reference for school-aged children and adolescents. Bull World Health Organ 2007; 85(9): 660-7.. Aiming at providing information for a monitoring system, the recommendation is that the quality control regarding weight and height data be conducted by trained professionals, both in entry and in data¹³13. Levine RS, Connor AM, Feltbower RG, Robinson M, Rudolf MC. Does routine surveillance of children's height and weight provide a sufficiently reliable means for monitoring the obesity epidemic? Public Health 2008; 122(10): 1117-9. DOI: 10.1016/j.puhe.2007.11.001
https://doi.org/10.1016/j.puhe.2007.11.0... .

The objective of this study was to assess the consistency of weight and height data typed in Web-CAAFE, as well as the calculated BMI, compared to the measures checked and the repeated entry, providing information for quality control in data entry in the computerized system.

METHODS

The study to assess the validity and reproducibility of Web-CAAFE among students from elementary school was conducted from May to August, 2014, in the city of Feira de Santana, Bahia (Northeast Brazil). The study included the validation of the physical activity²⁰20. Jesus GM, Assis MA, Kupek E, Dias LA. Avaliação da atividade física de escolares com um questionário via internet. Rev Bras Med Esporte 2016; 22(4): 261-6. DOI: 10.1590/1517-869220162204157067
https://doi.org/10.1590/1517-86922016220... and dietary intake sections²¹21. Jesus GM, Assis MA, Kupek E. Validade e reprodutibilidade de questionário baseado na internet (Web-CAAFE) para avaliação do consumo alimentar de escolares de 7 a 15 anos. Cad Saúde Pública 2017; 33(5): e00163016. DOI: 10.1590/0102-311X00163016
https://doi.org/10.1590/0102-311X0016301... in Web-CAAFE. This study focused on assessing the quality of entry of weight and height data. The consistency of data was evaluated according to age, sex, academic performance and use of computers and internet.

This was a convenience sample, composed of all students from the 2nd to the 5th grades in a public, part-time school, from the state education network from Feira de Santana, Bahia. The school was selected because it met the research protocol (being a public elementary school, having the approval of the director and professors to collaborate in the evaluation of performance of the students, having a computer room, with access to internet and school meals). The target-audience to use the Web-CAAFE included students from the 2^nd to the 5^th grades of elementary school, because the instrument was built based on the cognitive skills of children aged between seven and ten years. In the school selected the age group of the students from the 2^nd to the 5^th grades ranges from 7 to 15 years.

Sample size was calculated based on a previous study conducted to validate the instrument of dietary intake for students²³23. Assis MA, Benedet J, Kerpel R, Vasconcelos FA, Di Pietro PF, Kupek E. Validação da terceira versão do questionário alimentar do dia anterior (QUADA-3) para escolares de 6 a 11 anos. Cad Saúde Pública 2009; 25(8): 1816-26. DOI: 10.1590/S0102-311X2009000800018
https://doi.org/10.1590/S0102-311X200900... , with the following parameters: expected sensitivity of 75%, margin of error of 20% for the lower limit of this sensitivity, and prevalence of 50%, therefore obtaining a minimum sample of 124 children²⁴24. Flahault A, Cadilhac M, Thomas G. Sample size calculation should be performed for design accuracy in diagnostic test studies. J Clin Epidemiol 2005; 58(8): 859-62. DOI: 10.1016/j.jclinepi.2004.12.009
https://doi.org/10.1016/j.jclinepi.2004.... .

The study was approved by the Research Ethics Committee of Universidade Estadual de Feira de Santana - CEP/UEFS (CAAE: 19499913.3.0000.0053). Participants obtained a written authorization from the tutors and signed an assent form.

In the first stage of the study, all participants took anthropometric measurements and filled out the Web-CAAFE questionnaire afterwards. The anthropometric measurements were taken in the computer room of the school, before the children used the computers, and the height and body weight values were written on a label attached to the class diary of the students for consultation (without decimal points for body weight and only two for height).

A sub-sample of 93 students was selected randomly among those who concluded the first stage, and their height and weight were measured again. On the next day, the students filled out the Web-CAAFE twice, once in the beginning of the school shift, and again at the end. The values of height and weight were also written in the class diaries, and consultation was allowed during the filling out of the Web-CAAFE. The interval in-between the repeated entry was three hours, considering the part time stay at school of approximately five hours in each shift (morning or afternoon). Therefore, two direct measurements of weight and height were taken, and there were three entries of these measurements in Web-CAAFE for each child participating in both stages of the study.

The sub-sample size was calculated based on the mean (33.3 kg) and on the standard deviation (SD) of weight (11.46) in the validity stage. The sample was calculated considering the following parameters:

The errors of types I (alpha) and II (beta) were established at 0.05 and 0.20, respectively. Therefore, the sample size for the second stage of the study was 93.

The anthropometric measurements were taken by a team of trained researchers, according to the standards in the literature²⁵25. Lohman TG, Roche AF, Martorell R. Anthropometric standardization reference manual. Champaign: Human Kinetics Books; 1988. 177 p.. Body weight was measured with a digital scale (accuracy of 100 g and maximum capacity of 180 kg, from Wiso^®, model Ultra Slim W801). To measure height, a portable, collapsible stadiometer was used, with platform and square (213 cm of maximum height and 0.1 cm of accuracy, from Altura Exata^®). The weight was measured while the children were barefoot, and wearing the school uniform. Height was measured with the children barefoot, without ornaments on the head and aligned with the Frankfurt plan.

The performance at reading and writing was assessed by the teacher in charge of the class, using a form containing a hedonic scale (0 = very poor, 1 = poor, 2 = regular, 3 = good, 4 = excellent). The criteria established for the evaluation of the reading were: fluency, intonation in paragraphs, punctuation, recognition of the theme and explicit information in the text, and identification of linguistic marks that show the speaker and interlocutor. For the performance in writing, the following items was considered: knowledge of small and capital letters, distinction of homorganic consonants, domain of the writing of the works influenced by the characteristics of speech, application of orthography rules regarding the signaling of nasalization, and ability to write the words, sentences and texts correctly.

To assess the experience of each child with the use of computers and the internet, the following questions were asked:

STATISTICAL ANALYSIS

For the variables in continuous scale and without normal distribution, the analysis included the description of the sample by using the median and minimum and maximum values, the relative frequency of categorical variables (%) and the hierarchic linear regression with three levels: student, method (measures and entry in Web-CAAFE) and repetition for both methods. To assess the biases in the records of Web-CAAFE, besides the fixed effect for the difference between methods, the regression considered the repetition of measurements and their interaction with methods such as random effects.

The intraclass correlation coefficient (ICC) was calculated to assess the reliability of the records in repeated entries in Web-CAAFE, and to assess the correlation between the measurements and the records in the questionnaire.

The ICC for the reliability of the records in repeated entries was calculated as the squared root of the variance between the direct measurement and the entry in Web-CAAFE, divided by adding that to the variance between the repeated entries. The ICC between methods was determined with the same numerator, however, excluding the variance between the subjects of the previous denominator. The model parameters, including the standard error and the corresponding 95% confidence interval (95%CI) values, were estimated by the maximum-likelihood estimation.

The analyses excluded the records of body weight and height from Web-CAAFE as follows:

This procedure was adopted to eliminate the extreme, biologically implausible values of body weight and height, which would affect the analyses.

The impact of gender, school performance in reading/writing, possession and use of a computer/internet at home on the residuals of the fixed and random effects was assessed by using the univariate linear regression, after the hierarchical analysis. The relationship between the residue and the outcomes - body weight and height registered in Web-CAAFE and BMI calculated based on these measurements - was assessed graphically. The hypothesis tests were conducted considering the significance of p = 5%.

The body weight and height values measured in the two stages among the children in the sub-sample were compared with the Student’s t-test.

RESULTS

Of a total of 453 students, 416 accepted to participate and received an authorization from parents or tutors. Of these, 390 completed the first stage of the study [mean age (years) ± DP = 9.53 ± 1.53 years; 50.3% girls]. The sub-sample of the second stage resulted on valid data of 92 students (9.39 ± 1.41 years; 51.1% girls).

In the sample of the first stage of the study, there was balance between the proportions of students as to gender, school year and age, showing 13.8% of students who were older in relation to the school year (11 - 15 years of age). Very poor and poor performances were observed in reading and writing in 28.2 and 24.2% of the sample, respectively. Almost 2/3 of the students claimed to own a computer in their household, however, less than half of them used it or had access to the internet (Table 1). The characteristics of the sub-sample were similar. Table 2 presents median, maximum and minimum body weight, height and BMI values of the students, according to sex and age.

There was no difference in the measurements of body weight (-1.01 kg; 95%CI -3.52 - 1.49) and height (-6.07 cm; 95%CI -2.88 - 0.31) measured between the two stages of the study for the children in the sub-sample.

Body weight and height typed in the Web-CAAFE and the resulted BMI presented ICC values higher than 0.90 for reliability (Table 3). The correlation between records in Web-CAAFE and the measurements was strong, higher for body weight (> 0.95) and a bit lower for height and BMI (> 0.80).

The difference between the mean values (bias) of body weight recorded in Web-CAAFE and measured was 677 g, that is, 2.0% (95%CI -0.5 - 4.6%) of the measurement (Table 3). The corresponding differences for height and BMI were 1 cm or 0.73% (95%CI 0.0 - 1.46%) and 0.39 kg/m² or 2.11% (95%CI -0.11 - 4.49%), respectively.

However, these differences did not include the random variation of the measurements, which needs to be considered to calculate the bias attributed to Web-CAAFE. With such an adjustment, the mean bias in body weight, height and BMI records made in Web-CAAFE decreased to 208 g, 2 mm and 0.238 kg/m², respectively. Therefore, the biases attributed to Web-CAAFE indicated a mild underestimation of BMI as a consequence of the underestimation of body weight and the overestimation of height. The residuals of the random effects among students represented a summary effect of the factors that influenced the records of body weight in Web-CAAFE, but were absent from the model. The residuals were related with the students’ age and with the quartiles of body weight and height typed into the Web-CAAFE (data not shown), and BMI resulting from these measurements. The negative residue meant the underestimation of BMI resulting from the typed weight and height, and the positive results indicated the overestimation.

The students with lower BMI indexes calculated based on measured weight and height (1^st quartile) were prone to the underestimation of BMI, based on the data registered in Web-CAAFE, whereas those with higher values (4^th quartile) had a strong tendency of overestimation, especially the older ones (Figure 1).

There was no statistically significant association between random residuals and gender, school performance of reading/writing, possession and use of a computer, as well as internet in the household, in the univariate linear regression analyses (details not shown).

The correlation between the residuals of the fixed effects and the outcome variables was close to zero (details not shown). The same relation was observed with the variables of the univariate analyses. However, the residuals of the random effects presented a linear relation with the outcome variables and the age of the children.

The magnitude and variation of random residuals increased with the second entry in Web-CAAFE, and reduced in the third entry. For weight, the means were -24 g (95%CI -33 - -16 g), 96 g (95%CI 66 - 127 g) and 10 g (95%CI -14 - 35 g), respectively, in the first, second and third entries. For height, the means obtained in these entries were, in the same order, -2 mm (95%CI -3 - 1 mm), 6 mm (95%CI 4 - 8 mm) and 2 mm (95%CI -1 - 6 mm).

DISCUSSION

Records of body weight and height in Web-CAAFE, as well as the resulted BMI based on these measurements, showed strong correlation between the repeated entries and in comparison with the measurements, being higher for body weight and a bit lower for height and BMI.

There was a minor bias attributed to Web-CAAFE, of 2%, in the measured value for body weight, and less than 1% for height, having an impact of 2.11% on BMI. This finding led to a mild underestimation of BMI as a consequence of the underestimation of body weight and the overestimation of height.

Direct comparisons between these results and those of similar studies are not possible to be made, considering the significant methodological differences. This study assessed the quality of entry in the system, and not the validity and reproducibility of the report of weight and height, since the anthropometric measurements were confirmed before filling out the Web-CAAFE and the students were instructed to check the values written down in the class diary, unlike what has been done in other studies⁷7. Storey KE, McCargar LJ. Reliability and validity of Web-SPAN, a web-based method for assessing weight status, diet and physical activity in youth. J Hum Nutr Diet 2012; 25(1): 59-68. DOI: 10.1111/j.1365-277X.2011.01181.x
https://doi.org/10.1111/j.1365-277X.2011... ^,2626. Baile JI, González-Calderón M. Precisión del índice de masa corporal, obtenido a partir de datos de peso y altura autoinformados en una muestra infantil española. Nutr Hosp 2014; 29(4): 829-31. DOI: 10.3305/nh.2014.29.4.7143
https://doi.org/10.3305/nh.2014.29.4.714... ^,2727. Seghers J, Claessens AL. Bias in self-reported height and weight in preadolescents. J Pediatr 2010; 157(6): 911-6. DOI: 10.1016/j.jpeds.2010.06.038
https://doi.org/10.1016/j.jpeds.2010.06.... ^,2828. Beck J, Schaefer CA, Nace H, Steffen AD, Nigg C, Brink L, et al. Accuracy of self-reported height and weight in children aged 6 to 11 years. Prev Chronic Dis 2012; 9: e119.^,2929. Tokmakidis SP, Christodoulos AD, Mantzouranis NI. Validity of self-reported anthropometric values used to assess body mass index and estimate obesity in Greek school children. J Adolesc Health 2007; 40(4): 305-10. DOI: 10.1016/j.jadohealth.2006.10.001
https://doi.org/10.1016/j.jadohealth.200... ^,3030. Abalkhail BA, Shawky S, Soliman NK. Validity of self-reported weight and height among Saudi school children and adolescents. Saudi Med J 2002; 23(7): 831-7.^,3131. Lee B, Chung SJ, Lee SK, Yoon J. Validation of self-reported height and weight in fifth-grade Korean children. Nutr Res Pract 2013; 7(4): 326-9. DOI: 10.4162/nrp.2013.7.4.326
https://doi.org/10.4162/nrp.2013.7.4.326... ^,3232. Chan NP, Choi KC, Nelson EA, Sung RY, Chan JC, Kong AP. Self-reported body weight and height: an assessment tool for identifying children with overweight/obesity status and cardiometabolic risk factors clustering. Matern Child Health J 2013; 17(2): 282-91. DOI: 10.1007/s10995-012-0972-4
https://doi.org/10.1007/s10995-012-0972-... .

Considering that, curiously, the tendency of BMI underestimation found, such as the effect of the underestimation of body weight and the overestimation of height, was similar to that reported in validations of the self-report of body weight and height among students²⁷27. Seghers J, Claessens AL. Bias in self-reported height and weight in preadolescents. J Pediatr 2010; 157(6): 911-6. DOI: 10.1016/j.jpeds.2010.06.038
https://doi.org/10.1016/j.jpeds.2010.06.... ^,2929. Tokmakidis SP, Christodoulos AD, Mantzouranis NI. Validity of self-reported anthropometric values used to assess body mass index and estimate obesity in Greek school children. J Adolesc Health 2007; 40(4): 305-10. DOI: 10.1016/j.jadohealth.2006.10.001
https://doi.org/10.1016/j.jadohealth.200... ^,3030. Abalkhail BA, Shawky S, Soliman NK. Validity of self-reported weight and height among Saudi school children and adolescents. Saudi Med J 2002; 23(7): 831-7.^,3131. Lee B, Chung SJ, Lee SK, Yoon J. Validation of self-reported height and weight in fifth-grade Korean children. Nutr Res Pract 2013; 7(4): 326-9. DOI: 10.4162/nrp.2013.7.4.326
https://doi.org/10.4162/nrp.2013.7.4.326... .

In this study, students with difficulties to read and write were also benefitted by the access to the measurements, since the awareness of their own anthropometric measures improves the quality of the report²⁵25. Lohman TG, Roche AF, Martorell R. Anthropometric standardization reference manual. Champaign: Human Kinetics Books; 1988. 177 p.. Still, there was a tendency to underestimate body weight and overestimate height. That indicated that other factors explain these biases - besides gender, age, skills to read and write, experience with the use of computers and the internet, or the knowledge of their own measurements - such as body dissatisfaction, and factors associated with bias in the report of body weight among adolescents³³33. Elgar FJ, Roberts C, Tudor-Smith C, Moore L. Validity of self-reported height and weight and predictors of bias in adolescents. J Adolesc Health 2005; 37(5): 371-5. DOI: 10.1016/j.jadohealth.2004.07.014
https://doi.org/10.1016/j.jadohealth.200... .

The reduction of the magnitude and the variation of random effects for values of weight and height, observed between the second and third application of Web-CAAFE, can be owed to three factors:

Besides, the first and the second application of Web-CAAFE occurred in a longer interval of time (one month), indicating that the increased magnitude and the variation of the random effects among them can be related with the loss of the novelty of the instrument, which may have caused fewer attention and motivation of the participants. The graphs of the residues of the random effects among the students, according to age and the quartiles of the respected measurements, revealed the tendency to underestimate weight, height and the resulted BMI by the Web-CAAFE among the students in the first quartile, whereas those in the higher quartile tended to overestimate these measures, with visible influence of age. This result is conflicting in relation to the scientific literature, because, among adolescents³⁴34. Sherry B, Jefferds ME, Grummer-Strawn LM. Accuracy of adolescent self-report of height and weight in assessing overweight status: a literature review. Arch Pediatr Adolesc Med 2007; 161(12): 1154-61. DOI: 10.1001/archpedi.161.12.1154
https://doi.org/10.1001/archpedi.161.12.... and children²⁷27. Seghers J, Claessens AL. Bias in self-reported height and weight in preadolescents. J Pediatr 2010; 157(6): 911-6. DOI: 10.1016/j.jpeds.2010.06.038
https://doi.org/10.1016/j.jpeds.2010.06.... , it is common that overweight or obesity lead to the underestimation of real weight and BMI based on self-report. This peculiar result suggests the realization of new studies, with population-based samples, in order to reach more consistent conclusions. The random residues among the evaluated subjects were not associated with the possession and use of computers/internet in the household, nor with school performance in reading/writing. This indicates that the lack of previous experience with computers and/or the fact that the child is not completely literate did not change the quality of the weight and height recorded in the analyzed sample. However, this result should be interpreted carefully, because it may have been influenced by the limited number of subjects with information available about school performance, or by the use of a learning assessment tool built especially for this study, which has not been previously validated.

The use of a computer in the household did not change the accuracy and the reliability of the dietary intake report in Web-CAAFE among the students in Feira de Santana²¹21. Jesus GM, Assis MA, Kupek E. Validade e reprodutibilidade de questionário baseado na internet (Web-CAAFE) para avaliação do consumo alimentar de escolares de 7 a 15 anos. Cad Saúde Pública 2017; 33(5): e00163016. DOI: 10.1590/0102-311X00163016
https://doi.org/10.1590/0102-311X0016301... .

On the other hand, the performance in the report of dietary intake and physical activities in Web-CAAFE was worse among the students in Florianópolis, Santa Catarina, who did not own a computer and were attending the 2^nd and 3^rd grades¹⁷17. Costa FF, Schmoelz CP, Davies VF, Di Pietro PF, Kupek E, Assis MA. Assessment of diet and physical activity of brazilian schoolchildren: usability testing of a web-based questionnaire. JMIR Res Protoc 2013; 2(2): e31. DOI: 10.2196/resprot.2646
https://doi.org/10.2196/resprot.2646... ^,1818. Davies VF, Kupek E, Assis MA, Natal S, Di Pietro PF, Baranowski T. Validation of a web-based questionnaire to assess the dietary intake of Brazilian children aged 7 - 10 years. J Hum Nutr Diet 2015; 28(1): 93-102. DOI: 10.1111/jhn.12262
https://doi.org/10.1111/jhn.12262... , that is, they are supposedly less capable of reading and writing, when compared to students in the 4^th and 5^th grades.

The record of weight and height in Web-CAAFE requires the participants to type in only two numerical digits for weight and three for height, without worrying about separating integer and decimal values, using the comma or the dot, since this is done by the system. Besides, an animated avatar instructs the person, using audio and text presented in balloons. Together, these facilities require little ability to read and write, and can explain partially the results obtained.

In a previous study about the usability of Web-CAAFE¹⁷17. Costa FF, Schmoelz CP, Davies VF, Di Pietro PF, Kupek E, Assis MA. Assessment of diet and physical activity of brazilian schoolchildren: usability testing of a web-based questionnaire. JMIR Res Protoc 2013; 2(2): e31. DOI: 10.2196/resprot.2646
https://doi.org/10.2196/resprot.2646... , it was observed that, in the first part of the questionnaire (including weight and height data, name of the child and mother or tutor), the students presented a higher mean score of errors related with inconsistent responses, especially regarding: answering that the questionnaire had been filled out before it actually was; and entry the name incorrectly. Errors in data entry about weight and height were less frequent.

A potential limitation of the current study is that the absence of information about the school performance of about 2/3 of the students analyzed may have camouflaged the relationship between the ability to read and write and the quality of the record in Web-CAAFE. Besides, the convenience sample restricts the generalization of the results found.

The strong aspects of the study include a sample reasonably large to detect small differences between direct measures of body weight and height and the record in Web-CAAFE. They also include the use of advanced statistical methods, with higher power of detection of errors types I and II, which would allow assessing the variation of the measures. This would be owed to the use of all information available for all levels of hierarchic linear regression, instead of reducing the analytical sample for the students who completed all of the repetitions of the measurements assessed and the entries in Web-CAAFE. The adjustment of the bias calculation by the variation of the measurements is an important innovation to be used in subsequent studies to validate Web-CAAFE. Therefore, even though the regression analysis used has adopted only one independent variable (record in Web-CAAFE) as a fixed effect, it is multivariate, since it also estimated the random effects between the subjects and the reproducibility of the records.

CONCLUSION

The weight and height records in Web-CAAFE showed strong correlation with the measurements assessed and the repeated entries. The bias attributable to Web-CAAFE was minor, but indicated a mild underestimation of BMI, due to the overestimation of health and the underestimation of body weight. As a measure to control the quality of data entry in the system for further studies, and for the monitoring of students at the population level, it is possible to adopt a warning informing the record of weight and height values that are biologically implausible, with subsequent evaluation of the validity and reproducibility of these values, without the individuals disposing of the information about their measurements.

References

Publication Dates

History