Author's response to reviews Title: Validation of Metabolic Syndrome using medical records in the SUN cohort Authors: Maria-Teresa Barrio-Lopez (terebarriol@gmail.com) Maira Bes-Rastrollo (mbes@unav.es) Juan-Jose Beunza (jjbeunza@unav.es) Alejandro Fernandez-Montero (afmontero@unav.es) Martin Garcia-Lopez (mglopez@unav.es) Miguel-Angel Martinez-Gonzalez (mamartinez@unav.es) Version: 3 Date: 13 October 2011 Author's response to reviews: see over
Facultad de Medicina Departamento de Medicina Preventiva y Salud Pública Victorino Silvestre Journal Editorial Office BioMed Central 12 th October 2011. Dear Victorino Silvestre, We have the pleasure of submitting a revised version of our manuscript "Validation of metabolic syndrome in the SUN cohort". We have addressed all the reviewer comments. Revisions are highlighted in the new version of the manuscript. We have answered point by point the reviewers questions below: Giuseppe La Torre's points: - In the abstract, the authors reported the proportion of confirmed MS and confirmed non-ms only for the ATP III criteria and not for the IDF criteria, while it seems that the k index was better for the latter criteria. Please add the proportions also for IDF criteria: we have included also this information in the abstract. - The authors enrolled 172 participants in this validation study. Nevertheless, they did not report any data supporting the hypothesis that this is a sufficient number to detect robust figure. They should provide power calculation: we calculated the sample size for a proportion using the next formula: n=z α/2 2 pq/w 2, for a 95% confidence interval, where w stands for the width or desired precision of the confidence interval (it was set at w=+/-0,04). We selected 336 potential participants, we obtained enough data from medical records to confirm or reject MS in 172 participants. (Harvey, M. Intuitive biostatistics. Oxford University Press, 1995: 363-366) (Lemeshow S, Hosmer DW, Klar J, Lwanga SK. Adequacy of sample size in health Studies. John Wiley & Sons, 1990:1-8) - In the statistical analysis section, the authors refer to the use of the Fisher exact test. However, the use of this test is questionable: the total population sample is less than 20, or between 20 and 40 and at least one expected frequency less than 5 (see as an example Wayne Daniel. Biostatistics. A foundation for analysis in health sciences. John Wiley & Sons, 2005). Moreover, the Fisher test cannot be applied for age group and educational level (three modalities). Please, be consistent: The different characteristics (age, sex, body mass index, university career) between the portion of cases with MS and without confirmed MS were studied using the Fisher's exact test. The Fisher's exact test is similar to a chi-square test but it can be used when one or more of the cells have an expected frequency of five or less. Irunlarrea, 1. 31080 Pamplona. España +34 948 425600 Fax +34 948 425649 mamartinez@unav.es
(Daniel, W.W: From The chi-square distribution and the analysis of frequencies. In Daniel, W.W. Biostatistics: A foundation for analysis in the health sciences, 8th edition. Wiley, 2005:629-631.) - In the statistical analysis the explanation of the way for calculating sensitivity and specificity is not clear. Did the authors used the Bayes theorem using the estimated prevalence of MS in the population of 20%? Please clarify: You were right. We used the Bayes theorem using the estimated prevalence of MS in Spanish population of 20%. We added in the manuscript the references: To calculate sensitivity and specificity of a self-reported diagnosis of MS, the expected distribution of true and false positive and negative cases in the sample were estimated through sampling fractions and observed percentage of confirmed diagnoses to estimate the prevalence of MS in our population. We estimated the prevalence of MS in our population as 20% which is similar to the prevalence reported by descriptive studies in Spain 10-13. - Moreover, in the results section the use of this test is not reported. Are people with MS and people without MS homogeneous as far as concerns age, sex, body mass index, university career? How the results can be biased by the fact that these two groups could be different?: we analyzed the differences between MS and no-ms and we observed more MS in men than in women and an increased prevalence with age. We observed no differences in MS and no-ms between different university degrees. It is possible that in the SUN cohort there are less MS incidence than in the general Spanish population because it is a young cohort (mean age was 37.5 years). We have included this point in the new version of the manuscript. - One of the main limitations of the study is the internal and external validity of the questionnaire as a tool for self-detecting MS. The authors did not address this fact in the discussion section. They must provide sufficient discussion on that point. We accept that the sample was not representative of the general population because it is a cohort of university graduates and this could be viewed as a limitation in our study. It is possible that university graduates have more knowledge about MS and about health in general that general population. However, this could be viewed also as a strength of the study because our aim was to validate self-reported diagnosis of MS and its components among highly educated subjects (i.e. in this cohort). This would allow us to use the self-reported information from these highly educated participants to conduct studies of association between various risk factors and this entity. In this context we should note that the sample is not necessary to be representative, in the statistical sense of random sampling, of the general population. Indeed, the sample of this 336 selected participants should be representative of all the SUN cohort participants. This situation contributes to increase the internal validity of our results because we are obtaining self-reported information of high quality. Moreover, uniform education of the cohort reduces the potential confounding effect by socioeconomic factors. On the other hand, a strong point of our design is that participants were unaware that we could get information from medical records after answering their questionnaire. It thus prevents them to overstate or be artificially more precise in their reported data. - The name of the first investigator, methods page 5, needs to be written using the initials only: we have changed this point in the text.
- It could be useful for the reader having the questionnaire available, even as a supplementary data file: we have included a Spanish document with the used questionnaire, the measuring tape, and measuring instructions needed to define MS. We have also included their translation to English. Rosalba Rojas's points: - The authors present an interesting issue. They have recently published a similar report: Fernández-Montero A et al. Validity of self-reported metabolic syndrome components in a cohort study. Gac Sanit. 2011 Jul-Aug;25(4):303-7. What is the difference between both papers?: In the SUN cohort we have made a previous validation of the MS components only. In the present study we wanted to validate not only the MS components, but also the MS diagnosis according to the two more important criteria (IDF and ATP III). - Further description on the study population should be made in Methods section. For example: Did they select only voluntaries? How did they invite the participants?: we added in the text the selection method: To select the voluntaries, all graduates of the University of Navarra and university graduates from other colleges and associations received a mailed questionnaire and a letter of invitation to participate in the SUN study. A response to the initial questionnaire was considered as informed consent to participate in the study. We have also clarified the methods used to select the sample of participants included in the validation study. - The sampling method is vague: How did they select the 336 participants? Did they obtain a sample size to assure the validation? Which is their non-response rate? Did they select the participants randomly?: we calculated the sample size for a proportion using the next formula: n= n=z α/2 2 pq/w 2, for a 95% confident interval, where w stands for the width or desired precision of the confidence interval (it was set at w=+/-0,04). We selected 336 selecting participants, we obtained sufficient data from medical records to confirm or reject MS in 172 participants. (Harvey, M. Intuitive biostatistics. Oxford University Press, 1995: 363-366) (Lemeshow S, Hosmer DW, Klar J, Lwanga SK. Adequacy of sample size in health Studies. Edited by John Wiley & Sons, 1990:1-8) - More detailed information should be provided on self-report questionnaires. They should include the questions about previous medical diagnosis? of MS and each of the MS components. We have written in the text: Every two years, a follow-up questionnaire is mailed to each participant, gathering information about new medical diagnoses and changes in exposures of interest. Part of this information is self-reported and participants obtain it from their blood test and medical check-ups that they routinely undergo in Spain by their occupational physicians (blood pressure, levels of triglycerides, fasting glucose, high-density lipoprotein, cholesterol, etc). The questionnaire includes questions about previous and new diagnosis and changes in medications. The diagnosis of MS was not present in the questionnaire. We preferred to ask separately about each of the MS components to avoid an
underestimation and to be able to obtain more information. In addition we have added as a supplementary data file. - Two last lines from the introduction first paragraph should say: It is estimated that MS by its consequences, including type 2 diabetes and cardiovascular disease, is the first cause of death in women and the second cause of death in men in our society: we have reworded this sentence in the manuscript according to your suggestion: It is estimated that MS, by its consequences, including type 2 diabetes and cardiovascular disease, is the first cause of death in women and the second cause of death in men in our society 2. Thanks for your comment. have. Sincerely, We look forward to your response and available for any other questions you might Miguel A. Martínez-Gonzalez Departamento de Medicina Preventiva y Salud Pública Univ. Navarra. Irunlarrea 1. 31080 Pamplona. Tel: 948425600 (6463), 636355333 Fax: 948425649 mamartinez@unav.es http://www.unav.es/preventiva