Re: Goetzel RZ, Kent K, Henke RM, Pack C, D’Arco M, Thomas J, Luckett J, Arthur-Hartranft, T. Prevalence of Metabolic Syndrome in an Employed Population as Determined by Analysis of Three Data Sources. J Occup Environ Med. 2017; 59:161–168.
To the Editor:
The paper published in the February 2017 issue of JOEM by Goetzel et al compared the prevalence of metabolic syndrome (MetS) in an employed population determined by biometric testing, health risk appraisal responses, and employee medical claims. [Goetzel RZ, Kent K, Henke RM, Pack C, D’Arco M, Thomas J, Luckett J, Arthur-Hartranft, T. Prevalence of Metabolic Syndrome in an employed population as determined by analysis of three data sources. J Occup Environ Med. 2017; 59: 161–168.]
The authors state that “In this study, we apply a definition of MetS that closely approximates the American Heart Association (AHA)/National Heart, Lung, Blood Institute (NHLBI) definition.”
Members of our team have published several papers on MetS in employed populations.
The AHA/NHLBI definition for MetS is clear and must include three of the following five risk factors:
Although Goetzel et al state that they base their calculation of MetS prevalence on the above criteria when analyzing biometric screening or Health Risk Appraisal (HRA) data, their manuscript indicates that significantly different criteria was used. For example in Table 1, they report using four rather than the standard five criteria. Differences from AHA/NHBLI criteria include the following:
One benefit of the HRA questionnaire is to identify individuals whose blood pressure, HDL cholesterol, or glucose levels are normal as reported on the biometric values but who are taking medications for those conditions and would therefore meet the MetS criteria. Goetzel et al make no mention of collecting those medication data on their HRA, although they do identify people who report having high blood pressure, high cholesterol (but not specifically low HDL cholesterol), or diabetes.
The criteria used in the study by Goetzel et al are not AHA/NHLBI criteria or, to our knowledge, any currently accepted criteria for MetS. Therefore, the prevalence of MetS reported by the authors does not meet the criteria that they state were used for their analysis. We request an understanding of their use of nonaccepted criteria for MetS in their manuscript and believe they should provide a reanalysis of their data using the criteria that they state was the basis for their analysis.
In addition, as the authors state themselves, it is highly unlikely that working adults would visit the doctor specifically for MetS, as evidenced by their result of finding zero MetS diagnosis codes. One major reason employers utilize the HRA and biometric screening is to identify health risks and conditions that were previously unknown to the employee so that early steps can be taken to avoid disease progression. MetS, in particular, is well suited to this framework. Expecting to find diagnoses in the insurance claims that match up with biometric values is not sensible and therefore we find no reason whatsoever to include the medical claims component in this study. Another flaw in the medical claims analysis is that they included ICD9-790.29 (abnormal glucose) as a criterion but why did they leave out the diabetes codes (250)? The authors noted the inclusion of “diabetes medicine” in the text but did not list the prescriptions in Table 1.
Table 4 is confusing due to the blanks where there are missing data: for example, how many did not have available claims data or incomplete biometrics data (if anyone had one of the biometrics missing were the other data included)? This table makes it nearly impossible to understand the intent of the authors.