A cross-sectional study published in BMC Public Health reports that a five-variable routine metabolic screening model can estimate the probability of current hyperuricemia among gas station workers when direct laboratory testing is temporarily inaccessible. In internal validation across operational sites, the screening model achieved a mean area under the receiver operating characteristic curve of 0.858, indicating strong discrimination.
Direct measurement of serum uric acid provides the direct diagnostic basis for identifying current hyperuricemia in occupational and clinical practice. To address this operational gap, the authors developed a non-serum uric acid screening model designed for situations where routine occupational health information is available but immediate biochemical testing is delayed or requires separate evaluation. The five-variable model calculates an individualized probability of current hyperuricemia to support the prioritization of confirmatory testing rather than serving as an outright substitute for direct laboratory measurement.
The cross-sectional investigation evaluated a study cohort of 894 workers recruited from 11 gas stations and related operational sites in Nanning, Guangxi, China. Within this occupational study population, 169 participants met the diagnostic criteria for current hyperuricemia, representing an overall disease prevalence of 18.9 percent. For this study, current hyperuricemia was defined as a same-examination serum uric acid concentration exceeding 420 micromoles per liter in both male and female workers.
To develop the screening model, the investigators considered twelve prespecified non-serum uric acid candidate variables available from routine health examinations. Prior to algorithmic feature selection, total cholesterol was deliberately excluded from the candidate pool because of high redundancy and multicollinearity with low-density lipoprotein cholesterol. This preliminary exclusion left eleven candidate variables to be screened using least absolute shrinkage and selection operator regression. Through this penalized regression procedure, candidate predictors were evaluated against a prespecified selection-stability threshold to ensure that only stable, robust parameters were retained in the final mathematical model.
Five routine metabolic and physiological variables satisfied the prespecified selection-stability threshold and formed the final screening model: sex, waist circumference, triglycerides, systolic blood pressure, and fasting blood glucose.
Internal validation of the model was conducted using 100 repetitions of site-grouped five-fold cross-validation, systematically testing predictive stability across the 11 operational sites. Across repeated validation rounds, the model demonstrated a mean area under the receiver operating characteristic curve of 0.858 with a standard deviation of 0.004. In pooled out-of-fold evaluations, the area under the receiver operating characteristic curve reached 0.866, with a 95 percent confidence interval ranging from 0.838 to 0.893. Model performance was further assessed using out-of-fold predictions, discrimination, calibration, decision curve analysis, candidate screening thresholds, and bootstrap optimism correction.
The model also exhibited favorable calibration across out-of-fold predictions, yielding a pooled Brier score of 0.108. The pooled out-of-fold calibration intercept was -0.064 with a 95 percent confidence interval of -0.322 to 0.192, and the pooled calibration slope was 0.920 with a 95 percent confidence interval of 0.791 to 1.077. When evaluating clinical utility at a 10 percent candidate probability threshold, the screening model achieved a sensitivity of 92.9 percent, a specificity of 64.6 percent, and a negative predictive value of 97.5 percent, confirming that the tool effectively identifies individuals who require confirmatory evaluation.
The study authors noted clear boundaries and limitations regarding the model’s practical application and interpretation. The five-variable tool estimates the probability of current hyperuricemia and may support the prioritization of confirmatory serum uric acid testing when contemporaneous laboratory results are unavailable. It does not replace direct serum uric acid measurement, nor does it predict future incident hyperuricemia among workers. Furthermore, because the study was cross-sectional and conducted exclusively among gas station workers in Guangxi, the authors emphasized that independent external validation in broader populations is required before the model can be implemented more widely.
Subject of Research: Routine metabolic screening model for current hyperuricemia in gas station workers
Article Title: Development and internal validation of a routine metabolic screening model for current hyperuricemia among gas station workers in Guangxi, China: a cross-sectional study
Article References: Wang, J., He, X., Lin, Y., Chen, Z., Cong, J., Pang, Y., & Qi, G. (2026). Development and internal validation of a routine metabolic screening model for current hyperuricemia among gas station workers in Guangxi, China: a cross-sectional study. BMC Public Health. https://doi.org/10.1186/s12889-026-29953-y
Image Credits: AI Generated
DOI: 10.1186/s12889-026-29953-y
Keywords: hyperuricemia, serum uric acid, gas station workers, occupational health, metabolic screening, cross-sectional study, LASSO regression, waist circumference, triglycerides, systolic blood pressure, fasting blood glucose, internal validation, cross-validation, discrimination, calibration, Nanning, Guangxi
News Source: Ophelia Keating. (November 17, 2026). Five Routine Metabolic Markers Screen for Current Hyperuricemia in Gas Station Workers. Scienmag.



