Exploration of the methods of establishing the minimum clinical important difference based on anchors and their applications in the quality of life measurement scale QLICP-BR (V2.0) for breast cancer

OBJECTIVE: The measurement of the quality of life (QOL) in patients with breast cancer can evaluate the therapeutic effects of medical treatments and help to provide reference for clinical decisions. The minimum clinically important difference (MCID) can be better used in clinical interpretation than the traditional statistical significance. Based on the anchors, a variety of ways including traditional and updated anchor-based methods were used to explore most suitable MCID, so that to find better interpretation on scores of the scale QLICP-BR(V2.0) (Quality of Life Instruments for Cancer Patients-Breast cancer). METHODS: According to the investigation data of breast cancer patients before and after treatment, the most relevant indicators in various domains of QLICP-BR (V2.0) was found as an anchor to statistically analyze the value of MCID, and three analysis methods of anchors were used: Traditional anchor-based method, ROC curve method, multiple linear regression model analysis. Anchors are divided into four standards according to the degree of change in the treatment effect: one grade difference (Standard A), at least one grade difference (Standard B), one grade better (Standard C), better (Standard D). The final MCID value is selected from different statistical methods and classification standards that are most suitable for clinicians to use. RESULTS: Using Q29 of the EORTC QLQ-C30 as an anchor has the highest correlation with each domain of QLICP. The order of magnitude of MCID values among the four standard groups is: standard A< Standard C< Standard B< Standard D. The MCID value obtained by the ROC curve method is the most stable and is least affected by the sample size, and the MCID value obtained by the multiple linear regression model is the least. After comparisons and discussions, Standard C in the multiple linear regression model is used to determine the final MCID, which is the closest to other methods. After integer the MCID values of Physical domain (PHD), Psychological domain (PSD), Social domain (SOD), Common symptoms and side effect domain (SSD), Core/general module (CGD), Specific domain (SPD), Total score(TOT) can be taken as 15,10, 10, 11, 10, 9 and 9, respectively. CONCLUSION: In the evaluation of the QOL of breast cancer patients, although the results of MCID values produced by different methods are different, the results are relatively close. The anchor-based methods make the results of MCID more clinically interpretable by introducing clinical variables, and clinicians and researchers can choose the appropriate method according to the research purpose.