by Wu, Wen-Chi

Abstract (Summary)
Breast cancer-related lymphedema is a chronic complication of breast cancer treatment. It can result not only in physical discomfort and disfigurement but also in substantial impairment of daily activities. The public health importance of this study is to determine what, if any, factors contribute to an increased risk of lymphedema as well as to establish which subgroups of patients are at increased risk. Once the factors that influence the development of lymphedema are clarified, such findings can be used to develop preventive measures. In 2006, a 1:2 matched case-control study was carried out to determine significant predictors associated with breast cancer-related lymphedema. The results of the study showed that infection of the dominant arm, level of hand use and BMI would be significant predictors to cause lymphedema. Although the development of lymphedema still needs to be taken into account in clinical practice, this case-control study confirmed that some of risk factors can be used in prediction of lymphedema for breast cancer survivors. Because there is no precise incidence of lymphedema at present, the present study used the incidence rate from an independent study to predict probabilities of lymphedema for a group of breast cancer survivors by utilizing some confirmed risk factors. This study used Bayes Theorem to develop an estimator for the probability of lymphedema given various combinations of BMI, infection, and level of hand use. The delta method was used to estimate the variance of predicted lymphedema probabilities. The results consist of a list of lymphedema probabilities for different combinations of risk factors, as well as 95% confidence limits for these probabilities. Patients who have BMI 25kg/m2, infection, and medium/high of occupational/hobby hand use would have the highest risk of lymphedema (76.71%) after breast cancer surgery. The goal of this analysis is to address issues in lymphedema formation, to determine whether a set of confirmed risk factors can predict lymphedema, and to estimate the probability of lymphedema in the final model. A well-established lymphedema predicting system for the general breast cancer survivors should be seriously taken consideration in the future.
Bibliographical Information:

Advisor:Atilla Soran; Joseph P. Costantino; John W. Wilson

School:University of Pittsburgh

School Location:USA - Pennsylvania

Source Type:Master's Thesis



Date of Publication:09/28/2008

© 2009 All Rights Reserved.