A clinical model to estimate the pretest probability of lung cancer, based on 1198 pedigrees in china.
ABSTRACT : Computed tomography screening can detect lung cancer that is curable. However, some studies demonstrated that the risk for false-positives was about 50%. To make screening more efficient, we sought to create a forecasting model for individuals with different risks for lung cancer.
: We used multiple logistic regression analysis to identify independent predictors and to develop a prediction model. The pathological diagnoses in Guangdong Lung Cancer Institute were consecutively chosen as probands. All first-degree relatives of probands and their spouses were included as subjects. We divided the probands and their spouses into three subgroups according to the odds ratios (ORs), and the accuracy of lung cancer predictions for patients within the subgroups increased synchronously.
: There were 633 proband pedigrees and 565 spouse pedigrees. Independent predictors of lung cancer included sex (OR, 1.6; 95% confidence interval [CI], 1.1-2.3), smoking history (light smoker: OR, 1.1; 95% CI, 0.7-1.8; heavy smoker: OR, 4.7; 95% CI, 3.1-7.1), lung disease history (OR, 5.3; 95% CI, 2.8-10.0), occupational exposure (OR, 1.6; 95% CI, 1.1-2.2), and number of affected individuals among first-degree relatives (n = 1: OR, 2.1; 95% CI, 1.3-3.4; n ≥ 2: OR, 4.7; 95% CI, 0.5-41.2). The accuracy of the pretest probability increased for those with higher ORs: low-OR subgroup, 68.3%; mid-OR subgroup, 84.0%; and high-OR subgroup, 91.9%.
: Our prediction rule is recommended for estimating the pretest probability of lung cancer, thereby facilitating early screening.
- [Show abstract] [Hide abstract]
ABSTRACT: Background It is critical to develop a non-invasive and accurate method for differentiating between malignant and benign solitary pulmonary nodule. As a tool of assessment the probability of malignancy, the effectiveness of the diagnostic prediction model is still unclear in large sample studies. A diagnostic model established based on large samples is needed. Methods3,358 patients with a solitary pulmonary nodule between January 2005 and March 2013 were enrolled. All patients have received surgery for pulmonary nodules resection. Clinical characters, preoperative biomarkers results and computed tomography scans findings were collected from the patients. All patients were randomly separated into training set (n=1,679) and test set (n=1,679), we used training set to built a diagnostic model for the malignancy probability ofpulmonary nodule, and applied test set to validate the model we built and other published diagnostic models. ResultLogistic regression analysis identified eleven clinical characteristics as independent predictors of malignancy in patients with solitary pulmonary nodule. The goodness-of-fit statistic for the model indicated that the observed proportion of malignancies did not differ from the predicted proportion (P =0.571). The area under the curves of receiver operator characteristic curve for our model in training set was 0.935. In the test set, it was higher than Swensen's and Li's model. Conclusion Since the accuracy of the model was high, we suggest that the diagnostic model can be used as a tool to help guiding the clinical decision, when the clinician found it is difficult to make definite diagnosis with a solitary pulmonary nodule.Thoracic Cancer. 10/2013;