To the Editor: We read the concerns of Mr Sabour (1) with some interest. In our study (2), we essentially analyzed the agreement of body-scanner measurements between three repetitive assessments by the same observer (intra-observer reliability) and between body-scanner measurements and corresponding measurements of classical anthropometry if available (validity).
All analyzed measurements are quantitative. Except intra/interclass correlation coefficients, all statistical measures proposed by Sabour are designed for binary/categorical variables and cannot be applied in our situation. Moreover, proposed measures like sensitivity, specificity, etc, have nothing to do with agreement.
Intra-class correlation coefficient cannot be used in our situation either, because our data are paired. We used concordance correlation coefficients (CCCs) to assess agreement in our study. This measure is maximal if there are no biases, no differences in the variances, and if there exist perfect correlations between two or more paired measurement series. The author also proposed interclass correlation as an alternative measure here, which, however, is less stringent than the CCC.
The concerns expressed by the author in paragraph 3 of his letter are obscure. There is no recognizable concept of “concordance and discordance cells” in our situation. Concordance is quantitatively assessed in our situation and not by tables. The author claims that CCC<0.5 is like “Flipping a coin.” Again, it appears that the author mixed up concepts and we would like to refer to the literature (3, 4, 5) where the statistical properties of the CCC are explained in detail. A short explanation is given in our Methods section.
In the letter by Sabour, our research paper was cited as follows: “Based on their results, intra-observer reliability of both techniques is “excellent” (OCCC≥0.9). However, neck and thigh circumference a “good” (CCC≥0.7) and head circumference a “low” (CCC<0.5; Flipping a coin!) degree of concordance over the complete study population.” This does not correctly represent the content of our paper. The statements regarding neck, thigh, and head circumference are taken from our validity analysis—i.e., agreement of conventional with laser-based anthropometry—and not from the intra-observer reliability analysis as claimed by the author.
In summary, we are confident that we used appropriate statistics for our analyses and that conclusions thereon are correct.
References
Sabour S. Validity and reliability of three-dimensional scanning compared to conventional anthropometry for children and adolescents: methodological mistake. Pediatr Res 2017 (e-pub ahead of print 31 May 2017)..
Glock F, Vogel M, Naumann S et al. Validity and intraobserver reliability of three-dimensional scanning compared with conventional anthropometry for children and adolescents from a population-based cohort study. Pediatr Res 2017;81:736–744.
Lin LI . A concordance correlation coefficient to evaluate reproducibility. Biometrics 1989;45:255–68.
Lin LI . A note on the concordance correlation coefficient. Biometrics 2000;56:324–25.
Barnhart HX, Haber M, Song J . Overall concordance correlation coefficient for evaluating agreement among multiple observers. Biometrics 2002;58:1020–27.
Acknowledgements
This publication is supported by LIFE—Leipzig Research Centre for Civilization Diseases. LIFE is supported by the European Union, by the European Regional Development Fund (ERDF), and by funds from the Free State of Saxony in the context of the excellence initiative.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Competing interests
The authors declare no conflict of interest.
Rights and permissions
About this article
Cite this article
Glock, F., Scholz, M., Kuehnapfel, A. et al. Response to Sabour. Pediatr Res 82, 177 (2017). https://doi.org/10.1038/pr.2017.76
Published:
Issue Date:
DOI: https://doi.org/10.1038/pr.2017.76