Participants.

Based on the guidelines for cardiovascular rehabilitation and secondary prevention in China [39], this section will longitudinally track the symptoms of post-stenting patients in rehabilitation phase I (post-stenting hospitalization, T1), rehabilitation phase II (3 months after discharge, T2), and rehabilitation phase III (6 months after discharge, T3), for a total of three batches of surveys. Those that are eligible for this study recruitment are as follows: patients admitted with a confirmed diagnosis of coronary heart disease who have undergone successful PCI with stent implantation, patients who reside in a community within the urban area of Hengyang City, Hunan Province, China, disclose no reading comprehension difficulties, and voluntarily enroll in our study. Meanwhile, patients with mental abnormalities of any type and malignant disease according to clinical criteria, and those who have vague consciousness and cannot cooperate with the researcher will be excluded from participation in the study. During the following T2 and T3 surveys, patients who experience further major adverse cardiovascular events, die, refuse follow-up, or choose to withdraw from the study will also be excluded.

Sample and study setting.

There are many tertiary and teaching hospitals in Hengyang City, with high-quality service resources. These hospitals have established bases for medical research activities with community service units and undertake public health work for the resident population, providing a source of research subjects and community medical resources for conducting this study. We will select hospitals in Hengyang City using the cluster sampling method. Eventually, three hospitals will be randomly selected. All post-stenting patients in the selected three hospitals from March 1, 2023, to May 31, 2023, who meet the inclusion and exclusion standards, will be included and excluded, respectively, in this T1 survey, and we will continue to perform the T2 and T3 surveys at 3 and 6 months after discharge.

We aim to explore the symptom patterns of patients with CHD after stent implantation using Latent Transition Analysis (LTA) which is based on Latent Class Analysis (LCA) and enables the researcher to model dynamic changes. The estimation of sample size for Latent Class Model (LCM) is currently being explored. Most studies suggest 500 as the worthy goal in practice [40]. Therefore, we will recruit a total of 600 eligible patients who meet the inclusion criteria from three tertiary hospitals in Hengyang City with an expected dropout rate of 20%. Each hospital will enroll 200 patients through convenience sampling.

Data collection.

Before collecting the data, the study participants will be informed of the purpose, implementation steps, and ethical principles of the study, and their right to informed consent will be fully ensured. The research instruments will be distributed by uniformly trained staff. The study participants will be required to complete the questionnaires and submit these immediately on completion, and those who may be unable to complete the questionnaire themselves will be able to dictate their responses to a family member who will record them. During this phase, data will be collected using two instruments. A general questionnaire will be designed by the research team to collect the research general personal and health characteristics of the participants. The questions will include general information (such as age, sex, residence, marital status, educational level, occupation, economic income, medical insurance type, smoking, drinking, height, and weight) and details about the disease (such as duration of CHD, family history of CHD, cardiac function grading, comorbid chronic diseases, postoperative time, and the number of stents implanted). We will also use the following measurements to collect statistics on symptom scores.

The Patient-reported Outcomes Measurement Information System (PROMIS) was initiated by National Institutes of Health. Evon [41] conducted a study among patients diagnosed with chronic hepatitis C virus to evaluate the reliability and validity of 10 PROMIS short forms, including fatigue, depression, anxiety and sleep disturbance. The PROMIS measures exhibited favorable reliability (Cronbach’s α coefficient ≥0.87) and demonstrated moderate to strong associations with theoretically-similar items from specific scales assessing symptoms in patients with HCV (0.39–0.77). Additionally, all the data provided strong support for the structural validity of PROMIS measures using item-response-theory models. Krohe [42] performed a cognitive debriefing interview to evaluate the content validity of PROMIS short term regarding physical function, and showed that more than 90% of individuals demonstrated understanding of each item in the PROMIS. The selection of informative items for the PROMIS profile-57 consisting of seven short terms was based on a comprehensive literature review, item response theory analysis, and expert review of psychometric evaluation results from the PROMIS datasets. The original English version of the PROMIS profile-57 has excellent brevity, breadth, and robust psychometric properties. Furthermore, it has been successfully translated into over 40 cross-cultural adapted versions, all of which have shown satisfactory psychometric properties in various clinical conditions [43]. It contains seven fixed PROMIS core domains, including physical function, anxiety, depression, fatigue, sleep disturbance, ability to participate in social roles and activities, and pain interference. Each of these domains has eight items with a score of 8–40 and includes a pain intensity item measured on a scale of 0–10 [44]. All domains are scored on a 5-point Likert scale, and there are differences between the descriptions of the five options for each item, except for the pain intensity item [45]. Each short form consisting of eight items in Profile-57 was scored separately, resulting in scores for a total of seven domains. In addition, the scores were converted to standardized T-scores with a mean of 50 and a standard deviation of 10 [46]. Higher T-scores reflect higher levels of the respective measured concepts. Among them, higher scores for positive scoring domains indicate better function or fewer symptoms. In contrast, higher scores for negative scoring domains indicate impaired function or more pronounced symptoms. For example, a T-score of 70 in the short term about social roles and activities indicates that the ability of a participant to perform social roles and participate in social activities is higher than the average value for the general population, and the social function of the patient is good. Conversely, a T-score of 70 in short-term depression indicates that the depression in the patient is higher than the average value for the general population, and the patient has a higher level of depression [47, 48].

Shah indicated that self-reporting using PROMIS is recommended for patients with cardiovascular disease who have a high level of self-care awareness. It can help clinical staff to better understand the needs of patients and enhance their engagement with the disease, which has positive implications for patient health [49]. Prof. Yuan’s team developed a simplified Chinese version of the PROMIS profile-57 in accordance with the translation guidelines and conducted a psychometric evaluation in patients with breast cancer demonstrating acceptable convergent validity, discriminant validity, criterion validity, and reliability. In their study, the correlations between each item with its domains were all greater than 0.40 in multi-trait scaling analysis, the correlations between PROMIS profile-57 item scores with the corresponding domains coefficients in the Functional Assessment of Cancer Therapy-Breast ranged from 0.32–0.56 using the Pearson correlation test, and Cronbach’s α coefficients ranged from 0.85 to 0.95 [50]. We will use the Chinese version of the PROMIS profile-57 to measure the patients with post-stenting symptom scores in these seven domains and to explore the different symptom categories among patients in these three different periods (T1: post-stenting hospitalization; T2: 3 months after discharge; T3: 6 months after discharge) or in patients with different characteristics.

Analysis of quantitative data.

All data will be recorded and described using IBM SPSS version 26.0 (IBMCorp., Armonk, N.Y., USA). Descriptive statistics will be calculated for general personal and health characteristics in the sample, in which continuous variables will be analyzed by means and standard deviations, and categorical variables will be described in frequency and percentages. We will transform all PROMIS-57 raw scores of patients into standardized T-scores based on PROMIS guidelines (http://www.healthmeasures.net). In addition, Latent Class Analysis and Latent Translation Analysis will be performed using the Mplus version 8.3 (Muthén & Muthén, Los Angeles, CA) to classify patients with post-stenting into different latent class groups and reveal the dynamic shift of symptom situation according to the translation probability of each potential category. The logistic regression will be applied to explore the influencing factors of potential symptom category conversion.

Sample and study setting.

Purposive sampling will be used to select the research participants for semi-structured interviews from each phase (T1, T2, and T3) of the quantitative phase. The final sample size of the study population will be determined by the criterion that the data are saturated and no new themes emerge [51]. As the patient is still in the hospital, we will conduct a one-on-one in-depth interview in the department conference room. For patients who will be discharged from the hospital, we will choose to meet with them in their homes or community meeting rooms. In these interview settings there will be no interference from other people.

Data collection.

Data will be collected through in-depth, semi-structured interviews. The interviews will be conducted by a researcher with qualitative research experience and a long history of cardiovascular disease management, and a graduate student in the field of community-based chronic disease management. Prior to conducting the interview, we will contact the interviewee, obtain informed consent from the interviewee, and agree on a time and place for the interview. The interviews will be audio-recorded live, and the major interviewer who is responsible for communication can take notes. The secondary interviewer can carefully observe and record the expressions, tone of voice, gestures, and other non-verbal behaviors of the interviewees that can reflect psychological changes. Each interview is planned to last for 30–40 minutes.

We have developed an interview outline comprising focused open-ended questions for the semi-structured interviews: (1) What were your thoughts after you learned that you needed to undergo PCI with stent implantation? (2) What discomfort have you experienced after stenting? (3) What do you think is the most serious of these problems? (4) What is the biggest change in your life after stenting? (5) What kind of help do you want at this stage?

Analysis of qualitative data.

Interviews will be analyzed using content analysis, which can describe patient health status after stenting, explore their symptom trouble in different postoperative periods, and understand the symptom scores in the quantitative phase according to the subjective descriptions. We will use NVivo version 10 software to manage qualitative data analysis. First, within 24 hours of the interview, the interviewer will transcribe the recording into an electronic document verbatim. After transcription, the text will be checked for accuracy by two subject members to ensure accuracy and avoid the addition of the personal opinions and ideas of the interviewer. It is also necessary to refer to the recorded notes to supplement the non-verbal information from the interview. Second, we will open code the transcribed text material. Next, forming different categories or themes will be the critical process, in which researchers will further refine and elevate opinions based on the words used, including word frequency and relationships between words [52]. Finally, we can describe the feelings and experiences of patients post-stenting for a specific period through abstract categories or themes.

Integration of the data.

This is a mixed-methods research consisting of two stages. The process of replenishing or explaining each other between the qualitative and quantitative findings of a study is an important element of a mixed methods study [53].

In the quantitative study, we aim to explore symptom categories and their dynamic shift patterns in patients undergoing stent implantation through the three-stage longitudinal survey. First, latent class analysis will be conducted separately for T1, T2, and T3 to classify post-stenting patients into different homogeneous symptom categories at each postoperative stage. A series of LCA models will be performed in order to identify the optimal classification. The number of latent classes will be gradually increased until adding an additional category no longer improves the fit index of the model. Next, we will compare the evaluation criteria of all the LCA models to determine the best model, including the Akaike information criterion (AIC), Bayesian information criterion (BIC), adjusted BIC (aBIC), entropy, Lo-Mendell-Rubin likelihood-ratio test (LMR-LRT), bootstrap likelihood-ratio test (BLRT), and theoretical basis. The best model will be expected to possess lower values for the AIC, BIC, and aBIC, higher entropy, and significant LMR-LRT and BLRT results (P<0.05) [46]. The same modeling procedure will be applied to latent transition analysis using the longitudinal data. To ensure that the latent classes can be defined consistently across time, the restriction of measurement invariance across three different time points will be implemented. The full information maximum likelihood (FIML) was performed to handle missing values in our LTA model. This method is better than most alternative missing values methods used in longitudinal studies [54]. In LTA, a patient’s latent symptom category will be allowed to change over time, and the transition probabilities to different symptom categories during different periods (T1-T2, T2-T3) will be presented, revealing the dynamic changes of individual symptoms over time. Finally, logistic regression will be used to explore factors including demographic and clinical characteristics in the general questionnaire influencing the transition of latent symptom classes at T1, T2, and T3.

The qualitative data will provide an interpretation and complement the quantitative data. In the qualitative study, the descriptions of health conditions and feelings of patients at different times can explain the symptom situation at different stages well in the quantitative results. In-depth face-to-face interviews with patients will help us to determine the specific impact of surgery on their lives and to understand the possible unexpected scores in the quantitative results. Moreover, the descriptions of the management needs of patients at different stages, corresponding to the symptom scores in the quantitative data (the more severe the symptoms, the higher the management needs), can complement what is not found in the quantitative exploration.

In conclusion, it is pertinent to integrate both quantitative and qualitative data. Both types of data will describe the symptom problems of patients during different time periods after surgery. The combination of qualitative and quantitative data will help us to better understand and supplement the symptom situation of patients at different stages or the symptom variance of patients with different characteristics and further clarify the individual management needs of patients post-stenting. This provides a basis for performing accurate management at a later stage.