The Data Analysis item Screening of patient report outcome measurement scale in idiopathic pulmonary fibrosis
Description
In order to develop the PRO scale for IPF patients with Chinese characteristics, Delphi expert consultation method and IPF patients clinical questionnaire survey method will be used to screen the items of the scale. This data is the relevant data obtained from the expert importance scoring and effective clinical questionnaire. The items collected in the early stage are made into an expert consultation questionnaire, and each expert independently scores the importance of the items in the questionnaire. The critical value is determined by the average value and standard deviation of the three indexes: The average value, Coefficient of variation and unimportant percentage of each item. Deletion criteria include items with average value less than critical value, items with coefficient of variation greater than critical value, and items with unimportant percentage greater than critical value. Items meeting the above two or more deletion criteria will be deleted. The primary scale formed after expert consultation and statistics will be made into a clinical questionnaire and distributed to IPF patients. The questions of each item are mostly expressed in frequency or degree. Using Likert 5-grade scoring method, the choice grade is from light to heavy, and 1-5 points are calculated in turn. The recovered effective questionnaires will be analyzed by six statistical methods: item distribution survey method, discrete trend method, Cronbach coefficient method, correlation coefficient method, factor analysis method and multivariate linear regression analysis method, and items that met three or more deletion criteria in the six screening methods will be intended to be deleted.
Files
Steps to reproduce
Using the Delphi method [1], send the expert consultation questionnaire to the experts in the issue group. Under each item in the questionnaire, there were five options of "very unimportant", "unimportant", "dispensable", "important" and "very important", which are scored 1, 2, 3, 4 and 5 points respectively. Each expert individually grades the importance of each item, and proposes suggestions for item deletion and additions. The data will be input into the EXCEL by two people and two computers independently, and the accuracy of the data will be checked by cross checking. The critical value is determined by the average value and standard deviation of the three indexes: The average value, Coefficient of variation and unimportant percentage of each item. Deletion criteria include items with average value less than critical value, items with coefficient of variation greater than critical value, and items with unimportant percentage greater than critical value. Items meeting the above two or more deletion criteria will be deleted. In accordance with the above requirements in the combination of expert advice, the reserved items into a clinical questionnaire. 78 copies were distributed through electronic or paper questionnaires. For each problem, it is usually expressed by frequency or degree of occurrence, the scale of measurement is from light to heavy, and the international common Likert 5 rating method is adopted(Except item A31 "the color of your expectoration", normal "1" point, abnormal "2" point), such as the provision from "never" to "always" to give 1, 2, 3, 4, 5 points respectively. Statistical Package for Social Sciences (SPSS) for Windows (SPSS, Chicago, IL, USA) was used for data analysis. Using the item distribution method, generally delete the items with the option selection rate of > 80%; Using the discrete trend method, the items with standard deviation < 0.7 are deleted [2]; Using the correlation coefficient method, Pearson correlation coefficient, the items with the absolute value of the correlation coefficient of its dimension < 0.4 are deleted [3]; Using the Cronbach Coefficient method, when an item is removed, the Chronbach Coefficient of the dimension in which the item is located increases significantly, and such an item should be removed [4]; Using the factor analysis method, delete the items with factor load factor less than 0.5 or load factor greater than 0.5 on both factors at the same time; Using multiple stepwise linear regression method, taking the sum of scores of each item as dependent variable, and taking each item as independent variable, the items not entered into the equation are excluded [5]. According to the above steps, consider deleting the items that meet the criteria of 3 or more items deletion in the 6 screening methods. Combined with expert opinions, determine the final deleted items.