Development And Evaluation On Reliability And Validity Of Diabetes Self-management Behavior Scale Based On The Theory Of Planned Behavior

Objective:The diabetes self-management behavior predictive scale was developed based on the Theory of Planned Behavior,including four core behaviors:healthy diet,regular exercise,prescribed medication and blood glucose monitoring.By evaluating its reliability and validity,verifying its prediction efficiency and accuracy,convenient for medical staff to formulate planned and targeted intervention measures and to improve patients’ self-management behavior execution ability to provide a scientific basis for reference.Methods:By literature review and semi-structured interview,the scale item pool was constructed.Set up a focus group to discuss and analyze.18 professors in the field of diabetes education and endocrinology were invited for two rounds of Delphi consultation.Small sample pre-survey to form initial scale.Convenient sampling method was used to investigate 341 patients with type 2 diabetes mellitus in three large tertiary hospitals in Nanjing.Dispersion degree method and project decision value were used to screen the items;Cronbach’α coefficient and retest reliability were used to evaluate the reliability of the scale;the validity of the scale was evaluated by known-groups validity,aggregate discriminant validity,exploratory factor analysis,etc.to form a formal scale.Finally,confirmatory factor analysis was used to verify the goodness of fit of the constructed factor structure model and improve the scale structure.Receiver operating characteristic curve was drawn to find the critical value of the scale in predicting diabetes self-management behavior,calculate the accuracy of the four core behaviors,and then test the prediction efficiency and precision of the scale.Results:1.Item pool construction:A total of 312 items in 16 related scales were retrieved,and 38 items were selected as alternatives.Based on the question stem of TPB developed by Ajzen,10 patients with type 2 diabetes were interviewed semi structurally,the salience beliefs about attitude toward the behavior,subjective norm and perceived control behavior were extracted,and a pool of 20 items was constructed.2.Delphi consultation:The expert enthusiasm coefficient for the two rounds of experts consultation were 88.9%and 100.0%respectively;the authority coefficient of experts were 0.92 and 0.96 respectively;Kendall’s W coordination coefficients were 0.211 and 0.233 respectively of significance(P<0.05).After the two rounds of consultation,three items were added to form a pre-test version of the scale with 23 items.3.Items selection:A small sample of 15 patients with type 2 diabetes was pre-investigated to test the surface validity of the scale.After the formal investigation,the items were screened by item analysis method.No items were deleted,but 23 items were retained.Attitude toward the behavior(6 items),subjective norm(7 items),perceived control behavior dimensions(5 items)and behavioral intention dimensions(5 items).4.Reliability and validity analysis:(1)Reliability:①The Cronbach’s α coefficient of general scale was 0.830,each dimension was 0.669～0.808;②The split-half reliability coefficient of general scale was 0.874,and each dimension was 0.652～0.801;③ The test-retest reliability of the general scale was 0.950.All dimensions test-retest reliability were 0.886-0.919.(2)Validity:①Content validity:I-CVI was between 0.81 to 1.00,S-CVI was 0.94;②Focus discrimination validity:the correlation coefficients of each dimension was 0.205～0.560(P<0.001),the correlation coefficients between each dimension and the total table were 0.628～0.755(P<0.001);③Known-groups validity analyses:the scale can distinguish between patients who have received diabetes education or not;④Structural validity:Exploratory factor analysis extracted four common factors,and the cumulative explanatory variables were 59.499%,with no item cross load.5.Model structure:Confirmatory factor analysis was used to verify the scale structure,after correction,χ2/df was 1.952,RMSEA was 0.066,CFI was 0.900,GFI was 0.861 and AGFI was 0.834,all parameters meet the fitting standard,and the model fitted well with the actual data.6.Prediction efficiency:The correlation coefficient of our scale with SDSCA was 0.593(P<0.001),the correlation coefficient between each dimension and SDSCA was 0.395～0.481(P<0.001),showing a moderate correlation;the area under the ROC curve was 0.829.When the scale score was 75 points,the sensitivity and specificity for predicting the occurrence of behavior were 0.944 and 0.585;the precision of the four core self-management behaviors were 81.6%for healthy diet,73.7%for regular exercise,86.0%for prescribed medication and 48.6%for blood glucose monitoring.Conclusions:1.The diabetes self-management behavior scale developed based on the Theory of Planned Behavior includes 23 items and 4 dimensions.2.The diabetes self-management behavior scale developed based on the Theory of Planned Behavior shows good reliability and validity,prediction efficiency and precision,can be used as a predictor of self-management behavior ability of patients with type 2 diabetes,to provide support for type 2 diabetes patients to implement individualized and accurate self-management.