Caregiver Contribution to Heart Failure Self‐Care (CACHS)

Abstract Aim While caregivers (CGs) make an important contribution to the self‐care of heart failure (HF) patients, there are no reliable and valid tools for measuring such contributions. Current interventions that strive to optimize patient outcomes through self‐care strategies neglect to account for CG contributions, a potential confounder on outcomes. The aim of the study was to develop an instrument that measures CG contributions to HF patients’ self‐care. Design The study design follows an established process for instrument development. Methods A systematic literature review and semi‐structured interviews of CGs were conducted to identify measureable CG activities. Items were derived from thematic analysis of CG narratives. A content validity index was computed for each item (I‐CVI). Items with an I‐CVI of >0·70 were retained. Items with an I‐CVI of 0·50–0·70 were revised for clarification and items with an I‐CVI <0·5 were discarded, except in instances where fulsome theoretical or empirical evidence supported their retention. Results 14 CGs completed interviews and 10 CGs with 4 expert nurses completed I‐CVI testing. Major interview themes included arranging appointments, medication adherence, monitoring, coordinating care, encouraging independence and taking action. A total of 36 items were constructed and underwent I‐CVI testing. Following I‐CVI testing, 27 items were retained, seven items were retained after revision based on CG feedback and two items were removed. This newly developed 34‐item questionnaire represents current literature, CGs’ experiences, excellent I‐CVI scores and ready for further psychometric testing.


Introduction Background
Heart failure (HF) is a complex chronic syndrome associated with high mortality rates and frequent hospitalizations. In the United States, HF hospitalizations increased from 1Á3 million in 1979-3Á9 million in 2004 (Fang et al. 2008). It is estimated that the prevalence of HF in the United States is expected to increase from approximately 5Á8 million-8Á5 million in 2030 (Heidenreich et al. 2013). The Canadian Enhanced Feedback For Effective Cardiac Treatment phase 1 study (1991-2001; n = 8543) found that following hospital discharge for heart failure, the median survival time of HF patients was 1Á75 years, with a 10-year mortality rate of 98% (Chun et al. 2012). Worldwide, it is estimated that HF consumes between 1Á1% and 1Á9% of total healthcare expenditure in developed countries, with 50-74% of the HF costs attributed to hospitalization or long term institutionalization (Liao et al. 2008).
Internationally, clinical practice guidelines recommend that promoting self-care is fundamental to patient-centred practice (Lindenfeld et al. 2010, McMurray et al. 2012, McKelvie et al. 2013. Engaging in self-care is vital to symptom stability and improved health-related quality of life (HRQL) (Buck et al. 2012, Lee et al. 2014. Optimally, comprehensive HF self-care activities include: (a) managing multiple medications; (b) adhering to diet and fluid restrictions; (c) engaging in exercise; (d) monitoring symptoms and weight; (e) responding to changes in symptoms; and (f) navigating the health care system .When HF patients do not consistently engage in self-care activities, they are vulnerable to clinical deterioration, poor HRQL and frequent re-hospitalization (Leventhal et al. 2005, Albert 2008, Riegel et al. 2009, Lee et al. 2011. HF patients routinely depend on their informal caregivers (CGs), such as family members or friends, to help them with self-care activities (Dickson et al. 2011, Davidson et al. 2013. The CG contribution to HF patient self-care has been described indirectly through reviews of the experience of CGs (Saunders 2008, Usher & Cammarata 2009, Kang et al. 2011) and family influences on HF patient self-care , Dunbar et al. 2008, Gallagher et al. 2011, Oosterom-Calo et al. 2012. Although the direct contributions of CG to HF patient outcomes have been explored broadly in several qualitative studies, they have yet to be categorized and examined quantitatively. Tools measuring the experience of CGs of patients with HF are available, but do not fully capture the activities to support patient self-care described by CGs in the qualitative literature (Harkness & Tranmer 2007, Luttik et al. 2008. Recently, the Self-care in Heart Failure Index (SCHFI), designed to measure patient engagement in self-care, was modified to measure the caregiver contribution to self-care (Vellone et al. 2013(Vellone et al. , 2014. However, the Caregiver Contribution to Self-care in Heart Failure Index (CC-SCHFI) may not capture the full CG experience (Vellone et al. 2013). For example, the CC-SCHFI is completed by patients and reflects the patient's perception of the CG contribution rather than representing the actual voice of the CG. This is important for two reasons: the literature suggests that many CG contributions are 'invisible' and may go unnoticed by patients . Secondly, the CG's voice was not included in item development in the SCHFI (Riegel et al. 2000) and, therefore, modifying the SCHFI to capture CGs contributions to self-care may not result in a comprehensive description or quantification of the actual CG-specific contributions as identified by CGs. Therefore, an instrument with first person items, developed from and answered by CGs of patients with HF, is clearly needed. Valuable and valid information collected from such a questionnaire can then be used to guide strategies that help optimize patient and CGs' collaborative approach to self-care and ultimately minimize adverse clinical outcomes such as symptom deterioration, poor HRQOL and hospitalization.

The study Aim
The purpose of this study was to develop an instrument to measure CG contributions to HF patients' self-care activities. Measurement of CG contributions to self-care in HF patients will enable the further examination and quantification of: a) the degree of CG involvement in self-care; b) the most common contributions made by CGs; c) the potential areas where CGs need assistance; and d) the impact of CGs contributions on patient outcomes.

Methodology
The study design follows instrument development processes established by Kirshner and Guyatt (1985), Guyatt et al. (1992) and Juniper et al. (1994). Three distinct steps to the instrument development process include: Step 1. item development; Step 2. item clarification and reduction; and Step 3. initial psychometric testing. This study included Steps 1 and 2. Please see Figure 1 for an illustration of the protocol. The instrument developed and reported on in this study is known as The Caregiver Contribution to Heart Failure Self-care (CACHS).
Step 1. Item Development

Review of the literature and interview guide development
We conducted a comprehensive, systematic review of the literature to identify measurable CG activities (Buck et al. 2014).Based on this work we developed a semi-structured interview guide which was organized according to major theoretical sub-domains from the Middle Range Theory of Self-Care of Chronic Illness ). In this theory, self-care is understood to include sub-domains of: (1) maintenance (behaviours that maintain, preserve, or Review of literature to identify questions for semi-structured interview with caregivers of patients with heart failure.
Semi-structured interviews exploring caregiver contributions when supporting heart failure patients with self-care until data saturation achieved.
Step 2. Item clarification and reduction   (2) monitoring (cognitive processes and behaviours that result in vigilant body surveillance); and (3) management (attention to and then thoughtful response when bodily changes are noted during monitoring). These theoretical underpinnings guided item development processes.

Sample/participants
Inclusion criteria: We included adult (aged ≥18 years) informal CGs of patients (age ≥18 years) with a documented history of HF for at least 3 months. Exclusion criteria: CGs were excluded if they were: (1) caring for a patient who was pre-or postheart transplant or left ventricular device implantation; or (2) unable to read, write and understand English.

Study recruitment and ethical considerations
Eligible CGs were approached by health care providers during a routine patient follow up appointment in the outpatient Heart Function Clinic at the Hamilton Health Sciences. Interested CGs were then contacted by the research assistant and informed of the study objectives and procedures and asked for verbal consent to be contacted for an interview. Prior to the interview, eligible potential participants reviewed study information and provided informed consent. At the time of informed written consent, CGs were also asked if they were willing to be approached for Part 2 of the study. All CGs provided written consent for both Part 1 and Part 2 of the study. All nursing staff members working in the HF clinic at the time of the study were approached for inclusion and all provided consent. This study was approved by Hamilton Integrated Research Ethics Board (REB # 13-041).

Caregiver interviews -data collection
Individual, face-to-face, semi-structured interviews were subsequently conducted by an experienced interviewer to explore participants' usual contributions to HF self-care. Interviews were audio recorded and lasted between 30-60 minutes. The digital audio-recordings of the interviews were then transcribed, verbatim, by a professional transcriptionist; accuracy of each transcript was assessed by the interviewer. Our data collection and analysis process was iterative; completed participant interviews informed subsequent interview questions and probes. Following the initial four interviews, the principal investigators (KH, HB) analysed the transcripts for the depth and richness of the data to determine possible revisions to the interview guide. Initial interviews revealed that CGs had difficulty identifying specific self-care tasks as these were part of their 'usual routine'. To help with recall, the interviewer probed by giving examples of assistance identified by the literature and asked if this was similar to their routine. If so, CGs were asked to provide further explanation. However, CGs appeared to have ongoing difficulty thinking about assistance related to self-management activities. Therefore, CGs were asked to describe a recent situation when the HF patient was not feeling well or required hospitalization. The interviewer was then able to ask for clarification and explanation about the process of symptom awareness and decision-making during that specific event. After the first 6 interviews, no major changes to the interview guide were necessary.

Interview data analysis
Coding and thematic content analysis of the data was guided according to established approaches (Lynn 1986, Sandelowski 2000, Hsieh & Shannon 2005. Constant comparative analysis techniques were used to explore and understand the CGs activities (Strauss & Corbin 1998). The NViVO qualitative data management software program (International NVIVO 2012) was used to store and organize data analysis.
The interviewer and principal investigators (KH and HB) coded the first four interview transcripts independently and then met to reach agreement on a preliminary coding scheme. Initial coding categories emerged from these four interviews and all subsequent interviews were reviewed and grouped by the initial coding scheme categories. New categories that emerged from the data were added to the original coding scheme. Once the initial coding scheme was developed, further coding was completed by the interviewer and double coding by either KH or HB. The interviewer, KH and HB discussed the codes and grouped them into overall themes. Recruitment was discontinued when no further themes were identified by CG interviews and theoretical saturation from the initial literature search was achieved.
In a face-to-face full day meeting, our entire investigator team reviewed the results of the interview data analysis to derive a list of potential instrument items, based on distillation of clear categories of CG contributions to HF patients' self-care (Lynn 1986). During this process, our aim was to assure low inference-based depictions of CG contributions to HF patients' self-care, thereby remaining as close as possible to the 'everyday' language of our participants. Potential items were then categorized into self-care maintenance, symptom monitoring, or self-management to assure that each theoretical sub-domain was addressed. Disagreements were discussed until consensus was achieved on all items to be included in our instrument and subsequently subjected to content validity assessment (Step 2).

Step 2. Item Clarification and Reduction
The content validity of the developed instrument was assessed using a content validity questionnaire (CVQ). Content validity is the degree to which items contained in the instrument address important aspects of the construct being measured (Polit & Beck 2006). Content validity assessment is a process wherein consensus is sought among experts about the overall content of the instrument and individual item levels (Polit et al. 2007).
The content validity of the instrument was assessed based on the content validity index procedure described by Lynn (1986). Included with the CVQ was a detailed letter of explanation about the evaluation requirements and how to rank the items contained in the CVQ. Each item was ranked using a 4-point rating scale to evaluate the: (a) clarity and wording of each item; (b) whether the item captured the contributions provided by CGs accurately; and (c) areas for possible modification or improvements (Lynn 1986). Response choices included: very relevant and easy to understand (4 points); relevant, but needs minor revision (3 points); unable to assess relevancy without item revision (2 points); not relevant (1 point). In this case, both formal and informal CGs were invited to complete the CVQ. Formal CGs constituted expert nurses who provide care for patients in the HF clinic; informal GCs were those who completed the initial interviews to derive item content. The research assistant was available to clarify questions while participants completed the CVQ.

Overall content validity index
An overall content validity index (CVI) based on the proportion of responses that indicated items scored as 4/4 ('very relevant') or 3/4 ('relevant but needs minor revision') was calculated for the total scale. Subscale CVIs were also calculated for each cluster of items representing theoretical sub-domains of self-care maintenance, symptom monitoring and self-care management. The minimum acceptable overall CVI is considered to range from 0Á8-0Á9, indicating an acceptable level of inter-rater agreement for item relevancy (Polit & Beck 2006, Polit et al. 2007).

Individual content validity index
A CVI for each item (I-CVI) was calculated by determining the proportion of responses per item scored 4/4 ('very relevant'). Items with an I-CVI of >0Á70 were retained. Items with an I-CVI of 0Á50-0Á70 were revised for clarification. Items with an I-CVI <0Á5 were discarded, except in instances where fulsome theoretical or empirical evidence supported their retention.

Results
Between March 2013-September 2013, 14 informal CGs completed the original semi-structured interviews. Content validity questionnaires were subsequently administered to the informal and professional CGs, between January 2014-February 2014 (Step 2 described above). Four informal CGs, who completed the initial interview, were unable to complete the CVQ due to scheduling conflicts. Therefore, 10 informal CGs completed the CVQ. All of the expert nurses from the HF clinic (n = 4) completed the CVQ.

Study sample
Those informal CGs (n = 14) who completed the original semi-structured interviews were primarily female (79%), spouses (64%), residing with the HF patient (71%) with a mean age of 66 (SD 10) years. Four caregivers were daughters of the HF patients while 1 caregiver was the HF patient's son. The majority of CGs were retired (86%) and 2 CGs were employed. The CGs contributed to the self-care of older HF patients (mean age 77 years) who were primarily male (57%) with advanced HF symptoms (71% New York Heart Association Class III symptoms). The four CGs lost to follow up were two wives and two adult children (1 son, 1 daughter) of the HF patients. The HF clinical experience for the four expert nurses (two expert nurse clinicians and two advanced practice nurses) who completed the CVQ in Step 2 ranged from 5-10 years.
Step 1-Item Development CGs described several activities and behaviours that supported and/or assisted HF patients with many subdomains of self-care. The major themes that emerged from the semi-structured interviews included: • Arranging appointments • Medication adherence • Monitoring • Coordinating Care • Encouraging Independence • Taking Action Major categories and sub-categories in these themes were organized using the theoretical self-care sub-domains from the Middle Range Theory of Self-Care of Chronic Illness to develop instrument items . Major themes and sample quotes followed by examples of instrument items are outlined in Table 1. A total of 36 items were constructed and then organized in each theoretical sub-domain: self-care mainte-  nance (16 items), symptom monitoring (9 items) and self-care management (11 items). Step

2-Item Clarification and Reduction
The CVI for the entire instrument (n = 36 items) was 0Á88. The CVI for each self-care theoretical sub-domain was as follows: 0Á88 (self-care maintenance); 0Á96 (symptom monitoring); and 0Á96 (self-care management). These data reflect a high inter-rater agreement across items and strong indicator of robust construct validity (Streiner & Norman 2008). The majority of items (n = 27) had an I-CVI >0Á70 and were retained (without revision) in the instrument. For a summary of item CVI calculations, see Table 2. Seven items had an I-CVI between 0Á50 and 0Á70 and therefore were retained after clarification, as based on CG suggestions. Item 1Á5 had a CVI score 0Á64 and was removed as it was anticipated to significantly correlate with item 1Á6, which had a CVI score of 0Á79. Item 3Á9 was found to be confusing and CGs were unable to provide suggestions for revision. Therefore, item 3Á9 was removed. Although item 1Á1 had a CVI score of 0Á5, it was retained as it is the only item that represents assistance with basic activities of daily living.

Discussion
The purpose of this study was to develop a questionnaire using established principles for instrument development. The Caregivers Contribution to Heart Failure Self-care (CACHS) questionnaire contains items that represent current evidence and the experience of CGs. Acceptable CVI scores suggest that the CACHS questionnaire has the potential to quantify key CG contributions to HF patient self-care. To our knowledge, this is the first questionnaire developed specifically for measuring CG contributions to HF patients' self-care that overcomes the limitations of existing tools which were either designed and validated to measure the CG experience or burden (Harkness & Tranmer 2007, Luttik et al. 2008, Buck et al. 2014 or adapted from a pre-existing patient instrument (Vellone et al. 2013) but are being used to quantify these contributions.

Limitations
Specific limitations should be kept in mind when evaluating the findings. While we sought to recruit the most diverse group of CGs for the study, our sample was primarily retired females spouses residing with a (male) patient. However, to help minimize this limitation, the systematic review of all of the CG literature for HF patients (Buck et al. 2014) was used to derive a wide range of activities in the initial interview guide. We also continued interviews until no new activities were identified and theoretical saturation from the literature review was achieved. We recommend that future studies do further psychometric work in other CG populations such as ethnic and culturally diverse CGs and patients, CGs who may not reside with the patient and in a sample of working and unemployed CGs. Recruitment strategies also need to target CGs of patients who do not attend a HF clinic.

Implications for research
The development of the CACHS questionnaire further advances research involving informal CGs of HF patients by capturing CG self-report of their perceived contributions to HF patient self-care. The CACHS questionnaire allows quantification of specific activities that CGs contribute to patients' self-care and will allow the impact of those activities on patient outcomes to be examined. CG activities were previously confounding variables in selfcare studies, which could only measure patient clinical variables or CG burden or mood states (Buck et al. 2014). Measurement of CGs contributions will allow assessment of the degree to which CGs engage in supporting HF patient self-care when serving as co-providers of care with clinicians. Furthermore, measuring CG contributions will help us to determine what degree of variability in patient self-care is predicted by their CGs contribution. Finally, the CACHS questionnaire will also allow for more precise economic analysis.

Implications for clinical practice
With further psychometric testing and refinement, the CACHS questionnaire may equip clinicians to assess and quantify the overall impact of the CGs on patient self-care and target specific self-care decisions or behaviours. For example, if the CG indicates on the questionnaire that he/ she takes responsibility for the patient's medications, it may be more efficient and efficacious to target interventions to the CG who oversees organizing and administering HF medications. Thus, clinicians can ensure self-care interventions are designed and supported to meet the contextual factors influencing self-care and meet CG needs . Current clinical guidelines recommend that patient (including families) centred care be evidence based. The CACHS questionnaire will contribute to building evidence for including CGs in HF patient education and decision-making. Including CGs as partners is in keeping with nursing's holistic, lifespan models of care. Data accrued from studies using the CACHS questionnaire can be included in evidence based protocols and future quality improvement guidelines assuring that the very real and instrumental contributions of CGs are acknowledged and mobilized to improve HF patient outcomes.

Conclusion
In this study, we reported on the development process and derivation of the content validity of the CACHS questionnaire to measure CGs contributions to HF patients' self-care. Thirty-four items contained in the CACHS questionnaire met the CVI standards. The CACHS questionnaire should undergo validation in larger studies involving diverse populations to assess whether it remains to demonstrate robust and stable psychometric properties. The process of content validation reported in this publication is the initial step in a rigorous process to provide evidence for the validity of the assessment of CGs contributions to HF patient self-care. Further research will be needed to establish the full psychometric properties of the CACHS questionnaire such as: reliability, predictive validity, responsiveness, sensitivity and specificity, interpretability, acceptability and feasibility. I suggest that he/she looks unwell and needs a rest. 0Á86 3Á8 I remove him/her from a situation if it looks like he/she is getting too tired or overwhelmed (e.g. social events). 0Á71 3Á9 I am able to figure out if his/her actions helped to make him/her feel better. 0Á57 3Á10 I am able to figure out if my suggestions or actions helped to make him/her feel better. 0Á79 3Á11 If we try something that doesn't help, I will try something else. 0Á71