Development Of Sub-Health Measurement Scale

Objective:The aims of the study are to develop a sub-health rating scale with good operability, representative, validity, accuracy based on the healthy definition of the World Health Organization and the sub-health definition of national medical community. Establish a scientific sub-health evaluation approach and provide a reliable assessment tool for sub-health population's health management.Methods:1. Operational definition of sub-healthBased on the healthy definition of World Health Organization and Clinical Guidelines of Chinese Medicine on Sub-health, sub-health can be seemed as a healthy status between health and disease. People in sub-health status feeling disorder in physical, mental or social function, but no definite physical, mental or spiritual diseases can be diagnosed. There are some special sydromes of sub-health, such as infirmity in physical health, insufficiency in organ's function, low validity, disorder in mental health, disturbance cognition function and emotion, discomfort in social intercourse, lack of social supports and social resources.2. Literature analysisSearch the sub-health-related literature with key words:"sub-health", "not sick" "Chronic fatigue", "Chronic fatigue syndrome", "CFS" et al in major database such as "CNKI", "Pubmed", "EBSCO", "ScienceDirect", "Springerlink", "Wed of Science". Items of sub-health measurement in the item pool were selected from these literatures.3. Delphi method:Delphi method is an intuitive forecasting techniques developed by the Rand Corporation in 50 years of 20 century. During the Delphi consulted process, comments were collected from several rounds of anonymous experts＇inquiry. The Coordination Group of the program was in charged of summarizing the comments and feedback the results to the experts for next round's consultation. After several rounds, experts＇comments would become more consistent, and a more consistent and reliable conclusion or decision would be got. According to the method and knowledge involved in the development of the sub-health measurement scale, the selection criteria of the experts for the Delphi consultation were set down. Thirty experts were selected for the Delphi consultation at last.4. Item analysis method.The aim of Item analysis is to test the discrimination of each item. Always the respondents were divided into two extreme groups (high total score vs low total score), with 25%-33% of respondents in each group. Actually, the respondents were usually divided into two extreme groups (high total score vs low total score), with 27% of respondents in each group. Then independent samples t test was perform between two groups' scores. If the difference between two groups is significant, the item's discrimination were considered meet the requirements. And the item can be selected for the further analysis and screening. 5. Item screening methods.In order to establish a scientific, effective and representative index system, several methods were used for the item screening.The correlation coefficient method:First, analyze the correlation coefficient between the items. If the correlation coefficient between two items were bigger than usual, it may indicate that the content test by two items may be same or have large overlap. Then you can delete one or merge them. Second, analyze the correlation coefficient between items and dimensions. If the correlation coefficient between the item and its affiliated dimension were too smaller, it may indicate that the item does not belong to this dimension. And it could be considered being deleted or altered.Variation Coefficient:Analyze the sensitivity of the item by the items'variation coefficient. If the coefficient variation is too small, it may indicate the sensitivity of the item isn't big enough. And it could be considered being deleted.Cronbach's Alpha coefficient:Cronbach's Alpha is an indicator of internal consistency which often used in the rating scale. Higher Cronbach's Alpha coefficient means the items in the scale more consistent. Otherwise, it indicates that the items of the scale are too scattered. Usually, if remove a certain item from the scale, the Cronbach's Alpha coefficient increase, it may indicate that the item can improve the internal consistency of the scale. It should retain in the final revision. Otherwise it should be deleted.Factor Analysis:Analyze 59 primary indicators with factor analysis and varimax rotation method. Delete the item with low factor loadings or the item which factor loadings were not significantly different between two factors. Principal Component Analysis:Screening the primary evaluation contents of sub-health which cover with the physical health, psychological health, social health aspects by Delphi method. Analyze each aspect with principal component analysis, and then select the item with better representative.6. Scale Evaluation Methods900 universities'students in Guangzhou were randomly selected as the test population for the scale evaluation. Scale evaluation mainly through by reliability, validity, responsiveness, and demographic factors analysis. Reliability is the index which measure the scale's reliability and stability, usually reflected by test-retest reliability, split-half reliability, and internal consistency reliability. Validity was often used to reflect validity of the result which measured by the scale.. Validity was often tested by the content validity, construct validity and discriminative validity. Content validity refers to whether the selected items can represent the content or theme. Scale's content validity is often evaluated by experts. Actually the correlation coefficient also can reveal the content validity indirectly. Construct validity refers to whether the development of scale conformed to the theory and the original design, and it is often examined by confirmatory factor analysis. Criterion validity refers to whether the results tested by the scale were concordant with the criterion. In this study, the famous health scale SF-36 was selected as a criterion to test the criterion validity of the sub-health measurement scale. Discriminative validity refers to whether the scale can clearly distinguish two different populations. In this study, discriminative validity was test by the comparison among the students with different exercise habits or came from different subjects. Results:1. After reviewed and analyzed the literatures which were searched from domestic and foreign major databases, and combined with previous studies,119 items were selected for sub-health measure, included 40 physical items,41 psychological items, 35 social items and 3 total evaluation items.2. Results of Delphi experts'consultation.In this study,30 experts were selected for two rounds of formal consultation. These experts come from medical colleges or hospitals of Guangzhou, Beijing, Tianjin, Shanghai and Shenzhen, involving traditional Chinese medicine, clinical medicine, statistics, epidemiology, social medicine and health service management, psychology and another 10 subjects, with higher subjects'representative and geographical representative. During the Delphi consultation process,10,30,28 inquiry form were sent in pre-survey study, the first round formal consultation and the second round formal consultation respectively. The response rates in each round were 90%,93.33% and 89.29% respectively. Experts'consistency coefficients in each round were 0.327,0.211, and 0.258 respectively. The weight of Physical, psychological and social aspects were 40.61%,33.34% and 26.05% respectively. Finally,64 items were contained for further analysis.3. Item analysis and screening64 items'score between the upper 27% group and lower 27% group scores were significantly different. The scores of the upper 27% group were higher than the lower 27% groups (P<0.05). After 6 methods of screening and combined with the professional knowledge and our preliminary study,35 items which were selected by more than four methods were contained. Include one "sexual function" item, one general physiological sub-health evaluation item, a general mental sub-health evaluation item, a general social sub-health evaluation item and an overall evaluation items,40 items consisted the sub-health test-vision.4. Testing and evaluation for the scaleNine hundreds participants were recruited for the testing and evaluation.682 participants sent back the questionnaires and 667 questionnaires were finished and eligible for analysis. The recovery rated and completion rate were 75.78%,97.80% respectively. It took the participants 12.91(9.67) minutes to finish the questionnaires on average.Split-half reliability:The split-half reliability coefficient of the scale is 0.810.The Cronbach's alpha coefficient of two half-scale were 0.863,0.893 respectively.Internal consistency reliability:Cronbach's alpha coefficient of the scale for all items is 0.924.And the Cronbach's alpha coefficient was 0.805 for the physiological sub-health subscale,0.894 for the mental sub-health subscale and 0.805 for the social sub-health subscale.Content validity:The items of the sub-health measurment scale were selected from domestic and foreign's literatures and discussed by different subjects'experts. All of the items retained in the finnal version were screened by several methods strictly. The items of the scale can reflected the basic content of sub-health. Therefore, it is considered that the content validity of the scale was good. The correlation coefficient between the dimensions and its affiliated items were big, range from 0.609 to 0.841.The correlation coefficient between each sub-scale and its affiliated dimensions were higher than that between the dimension and other sub-scale, all greater than 0.690.Construct validity:KMO test showed that Kaiser-Meyer-Olkin index=0.935, Bartlett's test,χ2=8490.372, P<0.01, both supporting the factorability of the matrix. The principal-component factor analysis for Sub-health measurement scale revealed eight factors with an Eigenvalue>1, explaining 57.14% of the total sample variance. After the Varimax rotation of component factor analysis, results show that:the first factor could be interpreted as psychological symptom dimension (accounted for 29.29% of the variance); the second factor could be interpreted as vitality dimension and cognitive function dimension (accounted for 6.27% of the variance); The third factor could be interpreted as social adaptability dimension and social support dimensions (accounted for 5.08% of the variance); the fourth factor could be interpreted as emotional dimension(accounted for 4.05% of the variance); the fifth factor could be interpreted as organic function dimension(accounted for 3.47% of the variance); the sixth factor could be interpreted as physical mobility dimension (accounted for 3.16% of the variance); the seventh factor is defined as the physical symptoms dimension(accounted for 2.97% of the variance); the eighth factor without any item loading,(accounted for 2.86% of the variance).Criterion validity:correlation coefficient between the test-scale's total score and the SF-36's total score is 0.679, while correlation coefficient between the dimensions and the SF-36's physical and mental health domain range from 0.378 to 0.596. Correlation coefficient between each sub-scale and its corresponding single rating item's score range from 0.614 to 0.714.Discrimination validity:Result of the survey showed that the male student's scores are higher than the female students (t=4.821, P<0.01). Students of physical education have higher scores than other major students (F=4.982, P<0.01). Students come from medical college have higher scores than other college students (t=2.309, P<0.05). Students with physical training also have higher scores than those with fewer physical training or without physical training (F=15.084, P<0.01). The scores among other different groups had not significant differences. Conclusion:1. The reliability and feasibility of Delphi method in the development of scale are re-verified. The experience of the scale development and the methods used in the study may good for other studies.2. The result of the evaluation for the scale showed that the sub-health measurement scale has good reliability and validity, up to the requirements of scale development. The sub-health measurement scale were reliable, sensitive, effective enough for reflecting the characteristics of sub-health population and for testing the sub-health population's health status.