Development and validation of the Chinese version of the self-management support scale for kidney transplant recipients

Guided by the definition of self-management support from the ADA [13], we defined the categories and sources of support based on Wagner’s Chronic Care Model [26] and the thematic synthesis of self-management support for patients with chronic conditions by Dwarswaard et al. [14]. Next, we reviewed the literature and extracted relevant items from previous literature and relevant scales, such as the Patient Assessment of Chronic Illness Care (PACIC) [27] and Assessment of Chronic Illness Care (ACIC) [28]. We also interviewed the medical staff and KTRs to draw themes from interview transcripts to supplement the item pool. Once the initial instrument was constructed, the Delphi method was employed to examine whether the preliminary items accurately reflected the attributes of SMSSKTR. For the Delphi method, the inclusion criteria were experts who (1) had work experience with KTRs for no less than five years; (2) had expertise in kidney transplantation research or scale development; (3) held associated degree or above for nurses, master’s degree or above for physicians and pharmacists.

Convenience sampling was used until data saturation was achieved. A total of 31 KTRs participated in the interview. Among them, 10 joined in the three focus group interviews and 21 joined in the one-on-one in-depth interviews. For interviewees, the inclusion criteria were KTRs (1) who were 18 years or older; (2) whose postoperative time was no less than three months, as KTRs would usually recover from the surgery and transit to the self-management phase after three months; (3) who could read and write Chinese; (4) who were oriented to person, place, time, and situation; and (5) who informed consented to participate in the study. Recipients of multiple kidney transplantation or combined multiple organ transplantations were excluded from the study, as their needs for self-management support are different from single-kidney transplantation recipients. We also selected medical staff (four doctors, seven nurses and one pharmacist) who provided follow-up services to KTRs, as they were aware of the self-support needs of those patients. They joined ten one-on-one in-depth interviews and one focus group interview.

The data were collected between September 2018 and January 2019. One-on-one in-depth interviews or focus group interviews were conducted in a quiet room in China-Japan Friendship Hospital, Beijing. After participants signed the informed consent form, semi-structured interview guides were used to obtain rich data and achieve a better understanding of participants’ self-management support. The interview guide for patients included the following questions: (1) What difficulties do you face when managing your disease? (2) What support do you need when managing your disease? Where can you receive this support? The interview guide for the medical staff included the following questions: (1) What support do you think should be provided to KTRs? (2) What are the sources of support except for the medical staff? (3) What support do KTRs express during their clinical visits? All interviews were audio-recorded with permission and transcribed verbatim, and the main themes were extracted to form scale items. After the initial items were formed, a consultation form for the Delphi study was e-mailed to collect professional opinions on the dimensions and items of the scale. The importance of each item was rated on a 5-point Likert scale (1 = not important and 5 = very important). If an item was considered unsuitable, modification suggestions were expected.

We invited six experts who had been engaged in kidney transplantation, self-management of chronic diseases, urology and other related research or clinical work for more than 10 years to evaluate the content validity of the scale. To verify construct validity using exploratory factor analysis (EFA), we selected 325 KTRs to fill in a questionnaire, as the sample size of 325 satisfied the minimum sample-to-item ratio of 5:1 for performing EFA [29]. The inclusion and exclusion criteria of the patients were consistent with those in Stage 1. Data were collected from the outpatient department of China-Japan Friendship Hospital between February 2019 and March 2019.

The content validity evaluation forms, along with the initial SMSSKTR, were e-mailed to experts. They were given two weeks to rate the forms and return them. We decided to distribute the initial SMSSKTR to KTRs who met the inclusion criteria. Before the distribution, the investigators explained the purpose and procedure of this study to the participants and obtained informed consent from them. We informed them that the questionnaire was anonymous and that no personal information (e.g., names, addresses, and telephone) would be collected. The questionnaires were collected and checked immediately after they were completed. Incomplete information was requested to be supplemented. To test the stability of the scale, 30 KTRs were invited to fill in the initial SMSSKTR again two weeks after the completion of the first scale.

We calculated the content validity index (CVI) to assess the content validity of the scale. For each item, the number of experts scored on 3 or 4 point divided by the total number of experts is the CVI value of the item, while the number of items scored on 3 or 4 point by all experts divided by the total number of items is the CVI value of the scale [30]. To screen items of the initial scale, we calculated the coefficient of variation (items with coefficient ≥ 0.3 were retained) [31], item-total correlation coefficient (items with coefficient > 0.3 were retained) [32], and critical ratio. If the difference between the highest group (top 27%) and the lowest group (bottom 27%) of the total score did not reach the significance level (p < 0.05), the item would be deleted [33]. To test construct validity, we performed EFA using principal component analysis with varimax rotation. Factors were extracted based on eigenvalues value eguale to or higher than 1.00 and factor loadings eguale to or greater than 0.40 [34]. Before that, we performed a Kaiser-Meyer-Olkin (KMO) test and Bartlett’s test of sphericity to confirm the appropriateness of the EFA. The intra-class correlation coefficient (ICC) was calculated to indicate the test-retest reliability. ICC value < 0.5 indicates poor reliability, value between 0.5 and 0.75 indicates moderate reliability, value between 0.75 and 0.9 indicates good reliability, and values > 0.90 indicates excellent reliability [35]. IBM SPSS (version 20.0) was used for all the data analyses. All tests were two-sided, and a p value of less than 5% was considered statistically significant.

Using convenience sampling, we selected KTRs from three kidney transplant follow-up centers. The inclusion and exclusion criteria for KTRs were the same as those in Stage 1. A total of 270 KTRs were recruited to complete the survey, which exceeded the suggested minimum sample size of 200 for confirmatory factor analysis (CFA) [36].

The data were collected between October 2020 and February 2021. All eligible KTRs were invited to fill in a questionnaire. After signing the informed consent form, participants completed the questionnaire. They were informed that the questionnaire was anonymous and that no personal information would be collected. The questionnaires were double checked when they were completed to minimize missing data.

We performed CFA to test the scale’s construct validity by using Mplus (Version 7.11). The model fit indices and cut-off points were selected based on Kline [39]: standardized chi-square statistics (χ² / df; ＜5.0), comparative fit index (CFI; ≥0.9), Tucker-Lewis index (TLI; ≥0.9), root mean square error of approximation (RMSEA; ≤0.08), and standardized root mean square residual (SRMSR; ≤0.08). Convergent validity was examined using Spearman’s correlation coefficient between SMSSKTR and SMSRTR scores. Cronbach’s alpha coefficient and corrected item-total correlation coefficient were used to test the reliability of SMSSKTR. SPSS (version 20.0) was used for correlation and reliability analysis. All tests were two-sided, and a p value of less than 5% was considered statistically significant.

In our study, the average age of KTRs is 44 ± 13.25 years old. More than half of the patients were males (N = 20, 64.52%), and the average postoperative time was 4.79 ± 5.85 years. The average age for medical staff was 37.42 ± 8.34 years old, most of them were females (N = 9, 75%). The average working time in the kidney transplantation ward was 13.58 ± 7.74 years.

A total of 29 experts participated in the Delphi study. The average age was 42.24 ± 7.23 years. The working time in the kidney transplantation-related field was between five and thirty-four years. Among them, 18(62.1%) were nurses. 51.8% of the experts held a master’s degree or above.

We extracted 20 items related to self-management support for renal transplantation from the relevant literature, and extracted 27 items from the relevant scales. In addition, 39 items were extracted from interview transcripts of the patients and medical staff. We combined similar items and deleted overlapping items. Items that did not conform to the research topic were also removed, resulting in 36 items. After the two-round Delphi study, four items were deleted, eight items were added, 15 items were rephrased. For example, for the item “To what extent has your community organized health knowledge lectures or physical examinations for you after kidney transplantation?”, experts believed that it was not practical for the community to provide such services. Therefore, this item was deleted. When reviewing items in the relational support, experts believed that it should also include the support from peers. Therefore, items relevant to peer support was added (i.e., To what extent have your peers maintained a partnership with you and shared each other’s experience in the management of health condition after kidney transplantation). To avoid ambiguity, we also rephrased some items based on the feedback from experts. For example, for the item “To what extent has the medical staff given you daily guidance after kidney transplantation”, the concept of “daily guidance” was too abstract to be understood. Therefore, we added examples right after the term “daily guidance”. Finally, 40 items were achieved (Supplementary file).

In this stage, 325 questionnaires were distributed, of which 12 were invalid due to missing data. Therefore, 313 questionnaires were valid and used for the analysis. The sample had an average age of 43.80 ± 11.20 years old, and an average BMI of 23.05 ± 3.54. More than half of the participants were males (N = 204, 65.18%). Most of the organs were from deceased donors (N = 256, 81.79%). Please refer to Table 1 for more details.

The six experts rated to what extent each item in the initial SMSSKTR adequately reflected the theoretical definition of self-management support using the content validity evaluation form. We obtained CVI of 0.97 for the entire scale, with CVI values for each item ranging from 0.667 to 1.000.

After we collected the responses of the participants regarding the initial SMSSKTR, we performed an analysis of the dispersion of items, correlation analysis, and discrimination analysis. Based on the criteria we set for item screening previously, we removed items 2, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 23, 31, and 32, retained 26 items.

We used data from the 26-item SMSSKTR to perform the EFA. Bartlett’s test of sphericity (χ² = 6549.871, p < 0.001) and the Kaiser-Meyer-Olkin value (0.929) attested to the adequacy of the correlation matrix for exploratory factor analysis. The scree plot analysis suggested a four-factor solution. We removed items 24, 38, 39, and 40 because their factor loadings were less than 0.40 (items 24 and 38) or because they had cross loadings (items 39 and 40). We performed the EFA again for the remaining 22 items. The KMO value was 0.929 and Bartlett’s test of sphericity was significant (χ² = 5205.176, p < 0.001). The final EFA accounted for 66% of the total variance. All the 22 items were retained. The scree plot analysis suggested a three-factor solution (Fig. 2). We labelled the first factor as “instrumental Support” (nine items), which explained 25.9% of the total variance. We labelled the second factor as “psychosocial Support” (nine items), which explained 24.5% of the total variance. We labelled the third factor as “relational support” (four items), which explained 15.6% of the total variance. Table 2 shows the second EFA results of SMSSKTR.

The intra-class correlation coefficient (ICC) for the entire scale was 0.915. The ICC for the three subscales were 0.771, 0.896, and 0.832, respectively. The results are presented in Table 3.

Of the 270 returned questionnaires, 265 were valid for analysis. The sample had an average age of 46.00 ± 10.80 years old, and an average BMI of 22.92 ± 3.15. More than half of the participants were males (N = 192, 72.45%). Most of the organs were from deceased donors (N = 241, 90.94%). Please refer to Table 1 for more details.

We used CFA to test the model with data from the 22-item SMSSKTR validated in Stage 2. The results showed that the CFA model achieved a good fit, with χ² = 480.330, df = 206, χ²/df = 2.332, p < 0.001; CFI = 0.901; TLI = 0.889; RMSEA = 0.071, with a 90% confidence interval (0.063–0.079); and SRMR = 0.065 (Fig. 3).

We found that the total score of SMSSKTR was positively associated with that of SMSRTR (r = 0.532, p < 0.01). Further details are provided in Table 4.

The internal consistency of SMSSKTR was tested by calculating Cronbach’s alpha coefficient, which was 0.959 for the entire scale and 0.956 to 0.958 for the three subscales. The corrected item-total correlation coefficient ranged from 0.62 to 0.82.