PURPOSE: The purpose of this study was to determine clinical factors that are able to predict the likelihood of malignancy of unexplained cervical lymphadenopathy in children. METHODS: The data of 175 cases with unexplained cervical lymphadenopathy, including sex, age, fever, number of involved regions, and location and size of the largest node, were collected. Receiver operating characteristic analysis was performed to find out the optimal parameter for size of the largest node. Logistic regression was applied to determine independent predictors for malignancy. RESULTS: On the basis of receiver operating characteristic analysis, the ratio of maximal width to maximal length (ratio) was confirmed as the optimal parameter of size for malignancy prediction, and its threshold, which maximized sensitivity and specificity, was 0.5. Multivariate binary logistic regression model indicated that number of involved regions (odds ratio [OR], 5.169; 95% confidence interval [CI], 1.291-20.691; P = .020), location of the largest node (OR, 12.423; 95% CI, 3.657-42.205; P = .000), and ratio (OR, 52.080; 95% CI, 16.089-168.588; P = .000) were significant independent predictors of malignancy. CONCLUSIONS: Higher ratio (>0.5), multiple cervical regions of adenopathy (> or =2), and region II or III location of the largest node are associated with malignancy. These data should be helpful to supplement clinical judgment in determining which enlarged cervical nodes harbor cancer. Copyright 2010 Elsevier Inc. All rights reserved.
PURPOSE: The purpose of this study was to determine clinical factors that are able to predict the likelihood of malignancy of unexplained cervical lymphadenopathy in children. METHODS: The data of 175 cases with unexplained cervical lymphadenopathy, including sex, age, fever, number of involved regions, and location and size of the largest node, were collected. Receiver operating characteristic analysis was performed to find out the optimal parameter for size of the largest node. Logistic regression was applied to determine independent predictors for malignancy. RESULTS: On the basis of receiver operating characteristic analysis, the ratio of maximal width to maximal length (ratio) was confirmed as the optimal parameter of size for malignancy prediction, and its threshold, which maximized sensitivity and specificity, was 0.5. Multivariate binary logistic regression model indicated that number of involved regions (odds ratio [OR], 5.169; 95% confidence interval [CI], 1.291-20.691; P = .020), location of the largest node (OR, 12.423; 95% CI, 3.657-42.205; P = .000), and ratio (OR, 52.080; 95% CI, 16.089-168.588; P = .000) were significant independent predictors of malignancy. CONCLUSIONS: Higher ratio (>0.5), multiple cervical regions of adenopathy (> or =2), and region II or III location of the largest node are associated with malignancy. These data should be helpful to supplement clinical judgment in determining which enlarged cervical nodes harbor cancer. Copyright 2010 Elsevier Inc. All rights reserved.