Metabolic syndrome in primary care: A latent variable analysis of electronic medical records

Abstract

BackgroundThe best definition of the metabolic syndrome (MetSyn) remains controversial. The aim was to develop a latent variable measurement model for the MetSyn using primary care data. MethodsThis retrospective case-control study was based on the data from the IQVIA Disease Analyzer (DA) database including a representative panel of general and internal medicine practices in Germany. 23,020 adult patients, who were newly diagnosed with ≥1 component of the MetSyn (cases; median age [IQR]: 60 [50–72] years; 54 % males) between January 2016 to December 2019, and 2,144 patients without any MetSyn component (controls; 48 [38–61] years; 45 % males) were included. MetSyn was modeled by ICD-10 codes (E66: obesity, I10: hypertension, E11: type 2 diabetes, E78: lipid disorders) and physiological measures (BMI, HbA1c, blood pressure, HDL-cholesterol, LDL-cholesterol, triglycerides). ResultsThe covariance structure of the ICD codes and physiological measures could not be accounted for by a single MetSyn factor (root mean square error of approximation, RMSEA=0.085). Enrichment by another factor “Likelihood to Diagnose” significantly increased model fit with the observed data (RMSEA=0.054). “Likelihood to Diagnose” hardly correlated with MetSyn (r = 0.06), and was differentially associated with male sex, older age, and corresponding cardio- and cerebrovascular diseases. MetSyn showed the highest loading on four subordinate factors “hyperlipidemia” (0.70), “obesity” (0.66), “type 2 diabetes” (0.60), followed by “hypertension” (0.42). The ICD codes E66 and E78 were less indicative of these factors relative to BMI and triglycerides/HDL-cholesterol, respectively, whereas E11 and I10 were similarly indicative as their corresponding physiological measures. ConclusionsMetSyn was best indicated by BMI, triglycerides (an indirect marker of insulin sensitivity), HbA1c, and ICD-10 diagnosis E11 (type 2 diabetes). The factor “Likelihood to diagnose” probably constitutes a method factor of differential diagnosis systematically contributing to the common consistency of the MetSyn diagnosis in primary care.