An overview of the study flow following the six steps of IM is displayed in Table 1.
Phase 1
Aim
Preliminary evidence suggests that both patients’ and care providers’ needs, beliefs (i.e., an idea or principle judged to be true) and perceptions (i.e., the organized cognitive representations that individuals have about a subject) of disease (self-management) can influence their display of health behaviors and uptake of (self-management) interventions [31-34]. Therefore, following step 1 of IM, we will first conduct a needs assessment and examine the needs, beliefs, perceptions of CKD patients and care providers towards disease (self-management) and the use of eHealth interventions.
Design
Intervention monitoring group
First, a multi-disciplinary intervention monitoring group including both Dutch and Chinese experts will be established. This group will consist of five members: two researchers, one nephrologist, and two experts in chronic disease care. This expert group has ample experience with CKD care and the implementation of (eHealth) self-management interventions. The intervention monitoring group will meet monthly throughout all IM steps to discuss progress and the execution of major deliverables such as the needs assessment (e.g. program goals), intervention development (e.g. intervention content, delivery strategies), and evaluation planning (e.g. inclusion, outcome choice, analysis).
Literature review
A scoping literature review will be conducted to identify relevant evidence on CKD patients’ and care providers’ needs toward disease management. The search strategy is already developed in collaboration with a certified librarian (see Additional file 2).
Mixed-method study
We will conduct a mixed-method study to gain insight into the needs, beliefs, perceptions of CKD patients, and care providers towards disease (self-management) and the use of eHealth (self-management) interventions. This study will include face to face interviews, focus group discussions, observations, and survey research. Methods will build on an adapted version of the theoretical framework on beliefs and perceptions towards chronic lung disease used in FRESH AIR (Brakema et al., submitted). This adapted framework combines the Health Belief Model [35] and the Theory of Planned Behavior [36] and focuses on individuals’ beliefs and perceptions as well as the sociocultural context in which the individual resides (see Fig. 1).
We will explore CKD patients’ and care providers’: (1) beliefs and perceptions towards CKD and disease self-management, (2) needs towards CKD self-management, and (3) needs, beliefs, perceptions towards the use of eHealth interventions in disease self-management. The survey will consist of three validated measures: (1) ‘The Brief Illness Perception Questionnaire’ (BIPQ) [37], (2) ‘Chronic Kidney Disease Self-management instrument’ (CKD-SM) [38], and (3) ‘Chinese eHealth Literacy Scale’ (C-eHEALS) [39]. For the qualitative part, following principles of “purposive and convenience sampling” [40], the inclusion of participants will be based on opportunity, willingness to participate, and creation of diversity (e.g., age, gender) in our sample. We will also use snowball sampling [41], in which participants will be asked if they know any other individuals who could participate in the study. As there are no defined rules for calculating sample size in qualitative studies, data collection will continue until no new themes are identified from the data. For the quantitative part, as a rule of thumb, the sample size should be 5-10 times the number of items in the questionnaires [42]. Therefore, our aim is to recruit at least 230 patients. Eligibility criteria are detailed in Table 2.
The methods to be used differ between care providers and patients following the relevant group- and context characteristics (see details in Table 3). For instance, focus groups cannot be held with care providers as they (1) cannot be of duty all at the same time, and (2) work with a tight schedule, and finding a time slot that suits all care providers is very difficult. Moreover, we feel that CKD patients would be comfortable discussing their needs towards eHealth self-management interventions in a focus group setting, but not their needs and beliefs towards their disease in general. Hence, we will plan to discuss this topic in face-to-face interviews. More details on the methods use and relevant research materials used are presented in Additional file 3.
Phase 2
Aim
Following step 2-5 of IM, we aim to tailor the core interventions components of the ‘Medical Dashboard’ self-management intervention to the Chinese context following the results of the needs assessment performed in Phase I.
Design
All the IM concepts used in the steps below are operationalized and further detailed in Table 4 and Figure 2.
Step 2: Preparing matrices of change objectives
First, we will formulate program outcomes [29] on all levels as defined in the socio-ecological model [43]. This model will help us to understand the complex interplay between individual, interpersonal, community, and societal outcomes. Second, we will subdivide program outcomes into performance objectives [29]. Third, as each performance objective can only be reached if matching behavioral determinants are addressed, we will break each performance objective down into key underlying determinants [29]. We will use the Theoretical Domains Framework (TDF) to support the identification and selection of relevant determinants of behavior [44]. Two researchers will independently identify the determinants, and discrepancies will be resolved through discussions. Also, the intervention monitoring group will evaluate the determinants selected based on relevance and changeability, using the four possible consensus-based recommendation levels proposed by Michie et al.[44]. Finally, based on the determinants identified, we will specify change objectives [29].
Step 3: Selecting theory-informed intervention methods and practical strategies
We will first review the literature and identify relevant theoretical methods that can potentially induce a change in the determinants identified in step 2 [29]. Second, we will match the selected methods with specific change objectives. Third, the selected methods will be translated into practical strategies to target each determinant. Finally, the intervention monitoring group will rank the practical strategies per method [44] and ensure that these methods and practical strategies match with the program goals.
Step 4: Develop a tailored ‘Medical Dashboard’ based intervention (plan)
First, we will review the results of the needs assessment, the initial program’s logic model of change, and discuss intervention objectives, theoretical methods, and practical strategies for each level (e.g., individual, organization) specified in step 1-3. Second, the intervention monitoring group will have a meeting to amend, and if necessary, reconstruct the core components of Medical Dashboard to tailor the intervention. Also, the intervention monitoring group will create a plan for developing and testing the new version of the Medical Dashboard. Third, we will recruit five care providers and five patients to discuss the acceptability and feasibility of the intervention plan (member-check). To this end, we will use the ‘think aloud’ method [45], in which care providers and patients can speak aloud any words in their mind as they read through parts of the intervention plan. The think-aloud research method has been demonstrated to provide valid data on participant thinking and was successfully used in other intervention development studies [46, 47]. Based on the results obtained, further modifications will be made, resulting in a pre-tested version of the intervention plan ready for implementation in practice. The description of the intervention plan will follow the Template for Intervention Description and Replication [48].
Step 5: Develop an adoption and implementation plan
The goal of this step is to write a detailed adoption and implementation plan, containing relevant strategies to optimize intervention delivery and implementation (fidelity). First, we will discuss results obtained from step 1-4 and inventory local resources (e.g., connections with primary care clinics) that may facilitate intervention implementation. Second, based on all results obtained from previous steps and our previous systematic review [21], the intervention monitoring group will have a meeting to pragmatically identify potential adopters and implementers. Also, this group will demonstrate program use outcomes, performance objectives and related determinants of implementation. Third, the intervention monitoring group will design the implementation plan following Figure 3 [49] based on Expert Recommendations for Implementing Change list of strategies [50]. Then, we will use the ‘think aloud’ method to obtain feedback from CKD patients and care providers on the implementation plan. Finally, the adoption and implementation plan will be finalized with further modifications.
Phase 3
Aim
Following step 6 of IM, we will establish an intervention evaluation plan. Our evaluation will follow a hybrid type 2 trial design, comprising of (1) a pilot RCT with two parallel arms to study effectiveness, and (2) a process evaluation to evaluate implementation integrity (fidelity) and determinants of implementation.
Design
This study will be a 9-month, two-arm, hybrid type 2 trial [51]. The trial design and corresponding study elements are detailed in Figure 4. The Standard Protocol Items: Recommendations for Interventional Trials 2013 Statement is used to report the pilot RCT protocol [52] (see Additional file 4), and the Standards for Reporting Implementation Studies will be followed for reporting the implementation study [53].
Intervention
CKD patients in the comparison group will receive usual care consisting of personalized in- and outpatient treatment based on symptoms experienced and disease severity, as outlined in the Kidney Disease Improving Global Outcomes [6]. CKD patients in the intervention group will receive the usual care plus the culturally tailored ‘Medical Dashboard’ based self-management intervention for nine months. Before the start of the intervention, CKD patients and care providers will receive a face-to-face training session on the use of Medical Dashboard. To avoid contamination, Medical Dashboard will only be made accessible for participants in the intervention group via a secure password-protected registration process.
Study population, recruitment & randomization
Effectiveness; Pilot RCT
CKD patients will be recruited from the First Affiliated Hospital of Zhengzhou University. Recruitment strategies, inclusion, and exclusion criteria are identical to those in phase 1 (see Table 2 and Additional file 3). In pilot studies, Sim and Lewis [54] recommend 55 or more patients in total. Thus, we aim to recruit 60 patients in total for randomization in our study. We summarize the participant flow through the study in Figure 5. The outcomes for effectiveness are presented in Table 5.
A biostatistician blind to the study conditions will complete the random allocation sequence using a computer random number generator, allocating equal numbers of patients in the intervention (group 1) and comparison group (group 2). The care providers delivering the intervention cannot be blind to the intervention, but will not collect data or analyze outcomes. Research assistants blind to group membership will perform all face to face assessments and will not be involved in intervention delivery. Those conducting statistical analyses will be blind to group allocation until the evaluation is completed.
Implementation study
CKD patients, as well as care providers in the intervention group, will participate in the process evaluation to evaluate implementation integrity (fidelity) and determinants of implementation.
Implementation outcomes on the patient level as well as care provider level will be evaluated, see the further paragraph about details of outcomes of implementation. All CKD patients in the intervention group will be invited to complete the survey. Also, following principles of “purposive and convenience sampling”, CKD patients and care providers in the intervention group will be invited and interviewed either face to face or by telephone for the process evaluation. A research assistant who will not involve in the pilot RCT study will collect data within process evaluation.
Outcomes measures & data collection
Outcomes for the pilot RCT evaluating the effectiveness
We plan to evaluate:
- patients’ physical outcomes including biomedical measures,
- patients’ lifestyle and psychosocial functioning including self-efficacy, perceptions about CKD, quality of life, anxiety and depression status,
- hospital admission, health care utilization, and cost-benefit
A trained research assistant will conduct data collection, and the intervention monitoring group will supervise the data collection process. We will invite participants in both the intervention and comparison group to visit the Department of Nephrology at the First Affiliated Hospital of Zhengzhou University for data collection at baseline (T0), three months (T1), six months (T2) and nine months (T3) post-randomization. At baseline, we will collect demographic data, including age, race, income, education, marital status, work type of participants. To avoid dropping out of participants, if participants cannot come to the hospital, data will then be collected via telephone interview. Table 5 provides details on the proposed outcome measures and timing of the measures. The operationalization of outcomes and descriptions of the measurement tools used are detailed in Additional file 5.
Outcomes for implementation integrity (fidelity), and determinants of implementation
The process evaluation will be based on the RE-AIM framework [55]. The RE-AIM model is used to comprehensively measure the public health impact of research conducted in real-world settings [56]. Four dimensions (with the Effectiveness domain being applicable above)—Reach (refers to the proportion of CKD patients reached by our program), Adoption (refers to the proportion of participants who use our intervention), Implementation (refer to completion as well as fidelity to the protocol), and Maintenance will be used to evaluate the implementation only in the intervention group. We will collect the implementation outcome measurements throughout the 9-month trial. The outcome measures for each dimension of the RE-AIM model are as described in Table 6.
We will use the Measurement Instrument for Determinants of Innovations questionnaire [57, 58] to evaluate the determinants of implementation. Also, individual interviews with stakeholders (e.g. patients, care providers) will be conducted to learn more about the usability and feasibility of Medical Dashboard, its potential for wide-scale implementation, and barriers and facilitators to implementation. We will categorize the determinants identified from this mixed-method study according to Fleuren Framework [59].
Data Analysis
Qualitative data analysis
A Framework Method [69] will be used to guide our qualitative analysis. We will structure the qualitative data in a matrix output formed by rows (cases), columns (codes), and ‘cells’ (summarized data). We will follow the consolidated criteria for reporting qualitative research to ensure quality and validity.
Stage A: Transcribing: All audio-taped interviews will be anonymized and transcribed verbatim in Chinese. Long pauses and interruptions (relevant to the study subject) will be noted within the text. Additionally, all participants’ names will be replaced by an ID number. Any names mentioned during the interview will not be transcribed. One researcher will perform transcription, and another will check them to ensure content accuracy.
Stage B: Familiarization: Two researchers will independently read all transcriptions and make contextual/reflective notes to become familiar with the whole data set.
Stage C: development of an analytical framework& coding: Atlas.ti for Windows version 7.5.18 (Scientific Software development, Berlin) will be used to analyze our data. Our study includes four qualitative research parts. These are research into the (1) needs, beliefs, perceptions toward CKD and self-management (phase 1); (2) needs, beliefs, perceptions toward eHealth self-management interventions in CKD (phase 1); (3) the acceptability and usability of intervention components (phase 3); (4) determinants of implementation of eHealth self-management interventions (phase 3). Therefore, based on prior literature in which specific theoretical frameworks were used for similar research questions [70-74], we will develop four distinct initial coding trees. For the first and second research questions, we will develop two coding trees based on the adapted version of the theoretical framework of Brakema et al., (submitted) and the TDF [75]. The Technology Acceptance Model [76] will be used to develop the coding tree for evaluating the acceptability and usability of intervention components. Also, the Fleuren framework [59] will be used to develop the coding tree for determinants of implementation of eHealth self-management interventions. The second researcher and third researcher will check the coding tree developed and make amendments if necessary. One researcher will then independently code two or three transcripts using the coding tree, and add new codes if the textual abstracts identified do not fit with the existing set of codes. Then, this researcher will meet with the second researcher and discuss the newly added codes. New codes will be added into the coding tree, and if needed, related codes will be grouped into categories. Thus, the process will be repeated until no new codes arise.
The final coding tree will be checked and approved by the second researcher and the third researcher. This coding tree will include codes and categories; all codes and categories will be operationalized, and relevant examples will be provided.
The finalized coding tree will then be applied to each transcript. One researcher will go through each transcript, highlight the meaningful textual abstracts, and assign the appropriate code from the final coding tree. Then, all codes assigned will be verified by the second researcher. All coding differences will be discussed until consensus is reached.
Stage D: Charting data into the framework matrix: Data will be charted into matrices per research question identified by two researchers using Microsoft Excel 2010. The matrix will comprise of one row per participant and one column per code. Interesting or illustrative quotations will be added to the matrices.
Stage E: Interpreting the data: Overarching themes will be generated from codes derived from the data set by reviewing the matrix and making connections within and between participants and codes. Relations, connections, and causality will be further explored and interpreted, and conclusions will be drawn.
As for data derived from observations, all checklists will be digitalized and transported to Microsoft Excel 2010. Also, all written filed notes will be digitalized and will be taken into account to triangulate data collected from other methods. For instance, observation data obtained in phase 3 will support the analysis of implementation integrity (fidelity).
Quantitative data analysis
All quantitative analyses will be performed using SPSS version 23 (IBM, Armonk, NY, USA). We will enter the quantitative data into Microsoft Excel 2010 and calculate descriptive statistics such as the mean, standard deviation, median, and range of linear variables, and frequencies and percentages of categorical variables.
To gain insight into the needs, beliefs, perceptions of CKD patients towards disease (self-management) and the use of eHealth interventions in phase 1, we will use the descriptive statistics to describe CKD patients’ demographic characteristics, BIPQ scores, CKD-SM score, and C-eHEALS scores. Also, we will conduct secondary analysis using (1) independent t-tests for normally distributed continuous variables, (2) Mann–Whitney U-tests for nonnormally distributed variables and (3) Chi-squared or Fisher’s exact tests for categorical variables to compare the difference between certain types of different groups of CKD patients (e.g., age, gender, disease stage) and BIPQ scores, CKD SM score and C-eHEALS scores. P-values <0.05 and odds ratios with a 95% confidence interval excluding one will be considered statistically significant.
In phase 3, one of the primary hypothesis is that the intervention group, when compared to the comparison group, will demonstrate (statistically) significant improvement in self-management behavior at 3, 6 and 9 months post-randomization. Secondary hypotheses are that the intervention group when compared to patients in the comparison group, will demonstrate (statistically) significant improvement in biomedical status, self-efficacy, illness perception, mental health, quality of life, hospital admission, healthcare utilization and cost-benefit analysis at the timing of measurement. All primary statistical analyses will be conducted using intent-to-treat methods. The primary goal of statistical analyses is to examine and compare trends over time in the primary outcome. We will replicate this analytic approach for other secondary outcomes; secondary analyses will examine trends over time for biomedical status, self-efficacy, illness perception, mental health, quality of life, hospital admission, healthcare utilization, and cost-benefit analysis. We will use longitudinal, mixed-model analyses to test the hypotheses. Exploratory analyses will assess the impact of the intervention on primary and secondary outcomes for patients.
Mixed analysis of qualitative and quantitative data by triangulation
We will conduct a combined analysis by merging the quantitative and qualitative results after separate analyses have been carried out [77]. In phase 1, the quantitative results will triangulate the qualitative results of the perception of disease, self-management behavior, and eHealth literacy. To this end, we will develop a thematic matrix [78] that includes participants’ characteristics and data derived from surveys and emerging themes from our qualitative results to summarize CKD patients’ illness perception, self-management behavior, and eHealth literacy. In phase 3, we will use the results collected from the qualitative interviews to help interpret the quantitative results from the pilot trial. Qualitative results will, therefore, be used to expand upon the results of this trial to understand the implementation process as experienced by participants. For instance, the questionnaire of determinants of implementation will be matched with the qualitative research on determinants of implementation.