Development And Evaluation Of The Scale Measuring Severity Of Myathenia Gravis

Objective: To develop a new scale on the severity of myasthenia gravis(MG)suitable for clinical practice in China.Methods: 1.Developing a preliminary version of a new scale with Delphi method: The item pool was established based on 4 commonly used scales(QMGs,MGC,MMS,ARS-MG)for measuring MG severity.Database consisted of the scores determined by 5independent observers on the vedios of examinations in 60 MG patients.Opinions from MG experts in China were collected with Delphi Method.32 experts from 21 MG centers from the 35 invited experts participated in this investigation.The central team were responsible for collecting experts' opinions and developing a preliminary version of a new scale.After the completion of the first round of investigation,summary of the survey and previous evaluation of the items in the source scales were sent back to the 32 experts,then the second round of investigation was conducted.Based on the consistency of expert opinions in the second round of survey and previous evaluation of items in source scales,the preliminary version of a new scale was developed.2.Developing a final version of the new scale by item optimazation: Items were selected from the preliminary version with the data of these 60 MG patients,with the following statistical methods: interobserver reliability of items,the roles of items on the Cronbach ? of the preliminary version,correlation coefficient of each items with the preliminary version and factor analysis method.The final version of the new scale was formed.3.Evaluation of this scale: Interobserver reliability of each item,internal consistency of the final version,interobserver reliability and test-retest reliability of the final version,construct validity and criterion validity of the final version was analysed.4.Developing a simplified self-rating version of the new scale: The items which can be measured by the patient himself were selected from the final version to form a self-rating scale,which was then evaluated on internal consistency,interobserver reliability,test-retest reliability,construct validity and criterion validity.Results: 1.The item pool consisted of 19 items from the above 4 scales.The enthusiasm coefficients of experts in two rounds of survey are 91.4% and 96.9%.The authoritycoefficients are 0.79 and 0.80.The Kendall's coordination coefficients of expert opinions were 0.22(P < 0.01)and 0.42(P < 0.01)respectively.The preliminary version was formed with items with highest expert support and lest difference in opinions(M is 3 and IQR ? 1)and judgments from central team(adding diplopia).The preliminary version consisted of 10 items.2.After the preliminary version was evaluated,one item(palpebral fissure)was deleted due to the poor reliability performance among observers.The remaining nine items constituted the final version.The complete version covered six commonly involved muscle groups in MG,including the ocular,bulbar,facial,limb,axial and the respiratory muscles.3.The final version was evaluated : the interobserver reliabilities of all the items were good(?=0.43 ~ 0.80,P <0.01).The final version had good internal consistency(Cronbach ?=0.65),high interobserver reliability(ICC =0.89,P ?0.01)and test-retest reliability(ICC =0.87,P?0.01).Factor analysis showed the final version had four common factors and good construct validity(cumulative proportion in ANOVA 68.46%).There were moderate correlation between the final version and MG-ADL,MGC(r=0.62~0.77,P?0.01),and strong correlation between the complete version and QMGs,ARS-MG(r = 0.86~0.89,P?0.01).4.Ocular and facial items were deleted,and the remaining six items constituted a self-rating version.The self-rating version covered four commonly involved muscle groups in MG,including the bulbar,limb,axial and the respiratory muscles.The self-rating version had good internal consistency(Cronbach ?=0.77),with high interobserver reliability and test-retest reliability(ICC=0.93,P?0.01).Factor analysis showed the self-rating version had two common factors and good construct validity(cumulative proportion in ANOVA 64.24%).There were strong correlation between the self-rating version and ARS-MG,the final version(r=0.80~0.82,P?0.01),and moderate correlation between the self-rating version and MG-ADL,QMGs or MGC(r=0.51~0.75,P?0.01).Conclusion: The preliminary version of the MG severity scale was established by combining Delphi method with the previous evaluation of items from source scales,then the final version was developed through various evaluation methods.The items which can be measured by the patient himself were selected from the final version to form a self-rating version.The final version and self-rating version of MG severity scale can be used to assess the severity of MG with good reliability and validity.