LR-PC2: a polytomous risk prediction model for differentiating between benign, borderline, and invasive adnexal tumors

This model computes the risk of a benign, borderline or invasive (primary or metastatic) tumor for women with a detected adnexal mass. It is based on Van Calster et al (BMC Medical Research Methodology 2010). It can be used for new patients using an Excel implementation.

How to use it:

• Practical instructions:
• Open the Excel implementation file .

• Enter the data of a patient into the blue cells.

• Some checks on the entered data are implemented. If all data are entered without violation of the checks, the message Data complete! appears below the data. If this does not appear, not all necessary data have been entered or a check has been violated; in this case complete data entry or correct errors.

• If all data are entered and no checks have been violated, the probabilities of a benign, borderline, and invasive tumor are given below the data entry section.

• The probability of an invasive tumor aggregates the probabilities of a primary invasive tumor and of a metastatic tumor. This is done as the analysis has shown that these tumor types cannot be discriminated.
• Because the model is based on a combination of pairwise submodels, the results for these submodels are given too (cf additional results: pairwise risk predictions). This is the risk of diagnosis A when we assume that the true diagnosis is either A or B.

• Contact Ben Van Calster in case of problems (ben.vancalster@esat.kuleuven.be)

• Consult Timmerman et al (Ultrasound in Obstetrics and Gynecology 2000; 16:500-505) for information on the terms and definitions used for standardized data collection.

• Read the important information about the model that is presented in the paper in BMC Medical Research Methodology, and that is summarized in the Excel file.