Samples
This was a cross-sectional study with a convenience sampling method. For the purposes of this analysis, we estimated our desired sample size following the sample size strategy outlined in Devellis (40), which states that the ideal sample size for factor analysis should be 15:1 or 20:1 (41). A total of 420 patients with chronic disease from 3 tertiary-level hospitals in Zhengzhou, China, participated in this study. The inclusion criteria for patients were as follows: (1) the patients were chronic patients aged 65 years or older; (2) the patients could be contacted by telephone after discharge; (3) the patients received hospital care in the areas of general medicine, geriatric medicine or oncological medicine and were discharged home rather than to long-term care facilities; (4) the patients had no obvious cognitive or language disabilities; and (5) the patients were willing to participate in the survey. Of our initial study population, 12 patients dropped out midway, and 6 did not complete the survey and could not be substituted. Ultimately, 402 people completed the PACT-M1 (95.71% recovery rate), and 306 people completed the PACT-M2 (72.86% recovery rate).
The measure
The Partners at Care Transitions Measure. We used the original version of the PACT-M, which consists of 17 items in two parts, the PACT-M1 (9 items) and PACT-M2 (8 items) (16). The PACT-M1 captures the immediate postdischarge period, and the PACT-M2 is used to assess the experience of managing care at home. Each item was rated on a 5-point Likert scale from 1 to 5 (“strongly disagree” to “strongly agree”), with higher scores indicating better quality of care during transition. There is also a “not applicable” option, but it is not included in the total score. The Cronbach’s alpha coefficients for the two parts of the scale were 0.84 and 0.92, respectively. In addition, demographic characteristics such as gender, age, marital status, illness, medical insurance, and other index conditions were also collected.
Translation procedure
Authorization was obtained from the authors of the original scale, and the Brislin translation model was used to translate the PACT-M (42). First, two translators with bilingual proficiency (a nursing graduate student with an English level of 6 and a nursing doctoral student with an English level of 6) independently translated the scale. Our research team and two translators compared and discussed the differences between the two translated versions and then created the Chinese version of the PACT-M. Second, two other translators (a Chinese professor of English linguistics with more than 3 years of experience in English-speaking countries and a second graduate student in nursing science in the USA) translated the Chinese version back into English. Finally, our research team compared the backwards translation with the original English version, provided feedback and confirmed the conceptual and literal equivalence of the Chinese version.
Moreover, we recruited 30 elderly patients with chronic diseases who met the inclusion criteria for a preliminary trial. The results supported the readability, comprehensibility and cultural adaptation of the Chinese version. All the participants understood the items easily and took 15~28 minutes to complete the questionnaire.
Data collection
Seven experts were invited to evaluate the content validity of the Chinese PACT-M. These experts included two professors specializing in chronic disease management, one professor specializing in mental health, two associate professors in geriatric care, and a professor and an associate professor with experience in questionnaire development. Based on the experts’ professional theoretical knowledge and clinical work experience, they were asked to comment on the clarity of expression, language conventions, cultural background and content relevance of each item of the scale, and the item content validity index (I-CVI) and scale content validity index (S-CVI) were then calculated to evaluate the content validity using a four-point scale ranging from 1 (not relevant) to 4 (highly relevant) (43).
A survey team was formed by three postgraduate students who had undergone uniform training. After obtaining consent from the relevant hospital departments, department heads and patients, the patients were first informed of the purpose and importance of this study, and each patient was asked to sign an informed consent form. The PACT-M1 was then used to survey patients who were about to be discharged from the hospital and met the inclusion criteria. To protect patient privacy, patients completed the questionnaire anonymously but were required to provide their contact information for follow-up. The investigator used uniform instructional language to explain the purpose of the survey, the method of completing the questionnaire and precautions to the respondents. The questionnaires were distributed and collected on the spot, and the survey respondents were asked to add any missing information immediately. One month after the survey with the PACT-M1, the patients who participated in the first survey were followed up by telephone using the PACT-M2, and if they did not answer, they were considered to have dropped out halfway. In addition, 30 patients of the 402 patients were randomly selected to complete the questionnaire again after 2 weeks, and a test-retest reliability test was performed. The survey was conducted from August 1, 2020, to December 30, 2020.
Data analyses
The returned questionnaires were numbered uniformly, the data were entered by two people, and EpiData 3.1 software was used for systematic logical error checking. The data were statistically analyzed using SPSS 21.0 and AMOS 25.0. Categorical variables were statistically described as frequency and percentage; the continuous variables, which conformed to a normal distribution, were statistically described as means and standard deviations; and the scale response rate was used to test patients’ acceptance of the Chinese version of the PACT-M.
We used the critical ratio value (CR) and item total correlation (ITC) to calculate item discriminatory power. We set the respondents with the top 27% of the total score as the high group and the respondents with the bottom 27% of the total score as the low group and then conducted an extreme group comparison using a t-test. The ITC was calculated using Spearman correlation analysis. The ITC values were higher than 0.30, and statistical significance (P<0.001) was considered to indicate desirable discriminating power (44).
We estimated the test-retest reliability using Pearson correlation analysis with a two-week interval between evaluations. We used the Cronbach’s alpha value to estimate the internal consistency reliability of the Chinese version of the PACT-M.
We calculated the I-CVI as the ratio of the number of “highly relevant” and “quite relevant” expert-opinion responses to the number of experts; we calculated the S-CVI as the average of the I-CVI for all the items rated as either “highly relevant” or “quite relevant” (45).
We used exploratory factor analysis (EFA) with principal component extraction and varimax rotation to identify the factor structure of the Chinese version of the PACT-M. KMO and Bartlett's test were used to test whether the data were suitable for EFA. Based on the results of EFA, the confirmatory factor analysis (CFA) was performed using the maximum likelihood method to obtain the structural equation model.
Ethical statement
First, we received permission to translate the PACT-M into Chinese and use it from the original authors. Second, we obtained permission to conduct this study from the Zhengzhou University ethics committee. Finally, we obtained informed consent from each participant, who was guaranteed confidentiality, anonymity and the right to withdraw from this study at any time.