3.1 The literature research method determines the scale dimension and the result of the entry pool
3.1.1 Scale dimension build results
We retrieved TCM quality of life scale from Chinese knowledge network (CNKI), Wanfang (WANFANG), Vip (VIP), Springer, PubMed database of existing Chinese life quality scale and infertility dedicated quality of life scale by keywords "quality of life", "QOL", "traditional Chinese medicine", from building time to March 16, 2016.
So far, China has already developed TCM life quality scales of acute myocardial infarction, chronic liver disease, chronic urticaria, alopecia and acne, etc. By combing through the dimensions and entries of 41 TCM life quality scales, the Chinese medicine life quality scale covers 3 to 7 dimensions and 19 to 69 entries. After removing one scale that used different sorting technique, the 40 TCM life quality scales mainly include physiological dimension, psychological dimension, social function state dimension and symptom dimension which distribution is shown in Table 1. The Pareto chart of 40 TCM life quality scales is shown in Figure 1. Based on the retrieved 41 Chinese medicine quality of life scales and the existing 14 specificity infertility scales to comb and statistics, and combined with liver depression type of infertility patients clinical syndrome, we finally determined symptoms, emotion state, social function state, satisfaction as dimensions for liver depression type infertility quality of life scale.
Figure 1 Pareto chart of 40 TCM life quality scale
3.1.2 Results for constructing scale entry pool
The results are based on scales used to measure the anxiety level of infertile patients, such as SDS, SAS, SCL-90, and women's infertility scale developed by Qinnan (China)[4], which are according to the characteristics of TCM syndrome differentiation of infertility, and liver depression patients’ symptom, and combined with consultation with the clinical experts of TCM infertility to construct item pool of this scale. They were initially settled, excluded entries that were improper, similar and repetitive, and formed the entry pool of liver depression type infertility quality of life scale. Dimensions of Symptom, emotion state, social function state and satisfaction include 19, 12, 6, and 4 entries, respectively, with totally 41 entries, as shown in table 2.
3.2 Filter results after Delphi method
The first round of Delphi expert consultation questionnaire was started on March 10, 2016, while the second was on May 4, 2016. All the questionnaires were filled by experts independently, and returned within 7 days. Double entry and statistics were performed for the returned questionnaire data via Excel 2010.
3.2.1 Basic information statistics of experts
Based on working experience, study field, professional ranks and titles, we chose 13 experts of TCM gynecology, basic theories of Chinese medicine and quality of life, which occupied 53.85%(7/13), 30.77%(4/13)and 15.38%(2/13)respectively. More than 80% experts have medicine background. The basic information of Delphi experts are shown as Table 3.
3.2.2 Statistic result of expert authority degree coefficient
We calculated expert authority coefficient by expert familiarity and judging criteria[5]. Results showed familiarity with the maximum of 0.7, the minimum of 0.3, and a mean of 0.65, while the expert authority coefficient with the maximum of 0.85, the minimum of 0.55, and a mean of 0.79. It could be explained that Delphi consulting experts participating in this study are more authoritative. The details are shown in Table 4.
3.2.3 The concentrate degree of experts
In the first round of Delphi expert consultation, 13 experts scored the importance of 41 entries. In the second round, 13 of them rated the importance of 38 entries. The importance of entries were scored as 1, 2, 3, 4, and 5 respectively, indicating: very unimportant, unimportant, general, important and very important. By calculating the score ratio, variation coefficient and arithmetic mean of each entry, to show the concentration of expert opinions.
The results of two rounds Delphi expert consultation are shown in Table 8 and 9.
3.2.4 Coordination degree of experts’ opinions
The coordination degree of experts’ opinions is expressed by calculating the W coefficient of two rounds Delphi expert consultations. Coordination degree W valued between 0 and 1. The larger the number is, the higher the coordination degree of experts, the number of experts and the number of entries will be. In the first round, W=0.244, whereas W=0.447 in the second, meaning W increased. Results are shown in table 5.
3.2.5 An entry filtering result based on boundary value method
In the first round of expert consultation, 13 experts evaluated and recommended the dimensions of the scale, when 71.15% of experts gave the affirmation, and considered the setting of the dimension is reasonable. Three experts disagreed the dimensions of state of social function state and satisfaction, who accounted for 23.08%. No one advised to increase or reduce dimensions. After the team discussion, according to the result of the first round of consultation, we determined the first level of the dimension changing to the name of symptoms, emotion state, social function state and satisfaction, as the number of dimensions requires no increase or decrease, and basic classification is reasonable.
In Delphi expert consultation, the boundary value method was used to filter the entries, and the rules did not meet the entries of two or more boundary values. In the first round, arithmetic mean, variation coefficient and full mark rate was 2.944, 0.454 and 0.120 respectively. Entries would not be deleted when the arithmetic mean of entry importance score of higher than 2.944, the variation coefficient of lower than 0.454, and the full mark rate of above 0.120. Seven entries were removed and four were added in the first round of consultation, then the second round of Delphi expert questionnaire was launched based on first round’s result, where symptoms, emotion state, social function state and satisfaction dimensions contained 17, 12, 5 and 4 entries, respectively, with a total of 38 entries.
In the second round, the arithmetic mean, variation coefficient and full mark rate was 3.241, 0.236 and 0.009 respectively. Keeping entries when arithmetic mean of higher than 3.241, the variation coefficient of lower than 0.236, and the full mark rate of higher than 0.009. In the second round, we cut out five entries, with 33 entries remained, where symptoms, emotion state, social function state and satisfaction dimensions contained 13, 11, 5 and 4 entries, respectively. The Delphi method entry filter boundary values are shown in Table 6 and 7. The results of the entries are shown in Table 8 and 9.
3.3 Clinical questionnaire result
All the subjects of this study were from the Gynecological Outpatient Department of Dongzhimen Hospital, Beijing University of Chinese Medicine. With the consent of respondents who meet the inclusion criteria, respondents filled out questionnaires, and the investigators were able to explain the questionnaire for the respondents. A total of 171 questionnaires were distributed in the clinical survey, where 171 were recovered, 3 were deleted from the survey, thus there were 168 effective questionnaires, with 98.25% effective rate. The questionnaire includes three parts: the basic condition of the patient, the life quality questionnaire of liver depression type infertility, and the Chinese version of Ferti QOL. The basic information of the clinical investigation subjects is shown in Table 10.
Screening results were integrated for methods of frequency distribution, correlation coefficient, discrete tendency, Kronbach alpha [6] and the correlation coefficient, while eliminating the entries excluded by two methods. Entry B2 was deleted by correlation coefficient method, Kronbach alpha method and factor analysis method, whereas B12, Bl3 were deleted by correlation coefficient method and Kronbach alpha method; E3 was deleted by frequency distribution method, Kronbach alpha method and factor analysis method, so eventually deleting entry B2 "premenstrual pain at two side", B12 "menstrual cycle unstable", Bl3 "less menstrual quantity" and E3 "satisfied with your health", when 29 entries remained, as shown in Table 11.
3.4 The formation of liver depression type infertility quality of life scale
3.4.1 Formation of the scale
This study formed a formal liver depression type infertility quality of life scale based on the result of Delphi expert consultation and clinical questionnaire. The scale includes four dimensions (symptoms, emotion state, social function state and satisfaction), with a total of 29 entries and a VAS score scale, where four dimensions contain 10, 11, 5 and 3 entries respectively. The liver depression type infertility quality of life scale is shown in Appendix 1.
3.4.2 Weight of each dimension and entry
According to the results of the second round of Delphi, this study determined the weight of entries. In the second round, the experts evaluated and scored each dimension and entry according to their importance, scoring 1-5. The higher the score was, the more important the dimension or entry was. In this study, the weight of each dimension and entry was normalized by each expert. According to the statistical calculation, the weight of the symptom dimension is 0.293, of emotional state dimension is 0.270, of social function state dimension is 0.223 and of satisfaction dimension is 0.214. With the total score of 5, the higher the score was, the lower the quality of life it indicated. Entries 25, 27-29 are the reverse ones, and the calculation of the quality of life scores should be calculated with the other entry vectors. Weight of each dimension and entry is shown in Table 12.
3.5 The validity and reliability of the scale
3.5.1 Reliability analysis
This research evaluated the reliability and validity of the scale through clinical questionnaire results, and the reliability mainly used Kronbach alpha method to evaluate internal consistency of scale. We used split-half reliability method to evaluate consistency across entries, and the reliability evaluation result of liver depression type infertility quality of life scale showed that: Kronbach alpha of the whole scale is 0.884, and of dimensions of symptoms, emotion state, social function state and satisfaction are 0.742, 0.947, 0.591 and 0.742 respectively. Kronbach alpha > 0.5 for all dimensions, which shows the scale has good reliability[7]. The scale’s split-half reliability is 0.909, indicating that the cross-entries of the scale were consistent. Details are shown in Table 13 and 14.
3.5.2 validity evaluation analysis
Validity measurement mainly used the content validity to evaluate correlation between entries and the whole scale, other entries and related dimensions. We used structure confirmatory factor analysis to evaluate whether the scale is consistent with the theoretical assumptions, and used standard correlation to evaluate the correlation between the scale and results measured by standard scale. Liver depression type infertility quality of life scale validity evaluation results showed that the correlation between each dimension (symptoms, emotion state, social function state and satisfaction) and the total score were 0.619, 0.827, 0.728 and 0.619 respectively. In addition to the satisfaction dimension and whole scale correlation fair to middling, the remaining three dimensions provide good correlation with the whole scale. The evaluation of structural validity of the scale includes seven common factors, and the cumulative variance contribution rate is 66.48%. The factors contained in each dimension are in accordance with the original theoretical conception. As we have the FertiQoL International 2008 as a standard scale[8], correlation of liver depression type infertility quality of life scale and standard scale could be calculated, where the absolute correlation value was 0.792, so liver depression type infertility quality of life scale could be used as an independent tool. The results of structural validity are shown in Table 15-17.
3.5.3 Evaluation results of standard correlation degree
This study selected the FertiQoL international 2008 as the standard scale. Through calculation, the score of liver depression type infertility quality of life scale and the standard scale are under normal distribution. The results showed that Pearson correlation coefficient was -0.792. Correlative coefficient presents negative correlation because scoring methods of two scales are different. By correlation coefficient, we could conclude that the liver depression type infertility quality of life scale could be used independently as an evaluation tool to evaluate liver depression type infertility patients’ quality of life.