Construction of a risk prediction model of hyperuricemia for community-dwelling residents
10.3760/cma.j.cn114798-20250113-00039
- VernacularTitle:基于饮食及合并疾病的基层高尿酸血症预测模型构建
- Author:
Cong XIE
1
;
Jinxiu ZHANG
;
Jinli RU
Author Information
1. 山西医科大学第二临床医学院，太原 030000
- Publication Type:Journal Article
- Keywords:
Hyperuricemia;
Dietary habits;
Comorbidities;
Prediction model
- From:
Chinese Journal of General Practitioners
2025;24(3):308-314
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To construct a risk prediction model of hyperuricemia (HUA) for community-dwelling residents.Methods:This cross-sectional study was conducted from March to November 2020. A total of 1 967 residents in the Nanzhai community of Taiyuan city were selected by stratified sampling method as study subjects, among whom 1 555 (80%) subjects served as the training set and the remaining 412 (20%) as the validation set. Blood uric acid was measured in all subjects and level>420 mmol/L was defined as HUA. The risk factors of HUA were determined with multivariate logistic regression analysis, and a risk prediction model was constructed. The Hosmer-Lemeshow goodness-of-fit test and the receiver operating characteristic (ROC) curve were used to evaluate the predictive performance of the model.Results:Among the 1 555 residents in the training set, HUA was detected in 285 cases (18.3%). The detection rate in men was significantly higher than that in women [29.8% (220/739) vs. 8.0% (65/816), χ 2=123.17, P<0.05]. Compared to non-HUA group, the waist circumference and BMI, and the proportion of smoking, drinking, staying up late, and prevalence of hypertension and dyslipidemia were significant higher in HUA group. There was significant difference in the frequency of drinking tea, coffee, milk, and eating fruits, as well as the amount of beverages consumed between the two groups ( P<0.05). Multivariate logistic regression analysis showed that high BMI ( OR=1.132, 95 %CI:1.070-1.197, P<0.001), coffee consumption ( OR=1.337, 95 %CI:1.027-1.742, P=0.032), and dyslipidemia ( OR=1.479, 95 %CI:1.049-2.086, P=0.025) were risk factors of HUA, while female sex ( OR=0.213, 95 %CI:0.146-0.390, P<0.001) was protective factor of HUA. When all factors were included in the logistic regression model, gender, BMI, coffee consumption, beverage intake, sodium salt intake, sleep quality, and dyslipidemia ( OR=0.213, 1.113, 1.353, 0.788, 1.320, 0.788, 1.651) were important components of the HUA prediction model. By substituting the constant, these factors, and the regression coefficients into the logistic regression equation, the Logit(P) formula was obtained: Logit(P)=-1.530-1.547×gender+0.107×BMI+0.303×coffee consumption frequency-0.238×sodium salt intake+0.278×beverage intake-0.238×sleep quality+0.502×dyslipidemia. The ROC curve showed an area under the curve ( AUC) of 0.750. The Hosmer-Lemeshow goodness-of-fit test: P=0.632, indicating a satisfactory fit. When the formula was applied to the validation set for internal validation, the AUC was 0.745. Conclusion:The occurrence of HUA is influenced by multiple factors. The prediction model constructed in this study has good prediction performance,which may be used to predict the risk of HUA in primary care settings for community-dwelling residents.
