Background and design of the Physical Frailty and Symptom Monitoring and Management Behaviors in Heart Failure (PRISM-HF) study: A mixed methods study

Abstract

Background

Many adults with heart failure (HF) are physically frail and have worse outcomes. While the biological profile of physical frailty in HF has been examined, the behavioral profile remains unstudied. Physical frailty may impact self-care behaviors, particularly symptom monitoring and management (SMM), which in turn results in adverse outcomes. This paper describes the background and design of a study that addresses this knowledge gap, entitled “Physical Frailty and Symptom Monitoring and Management Behaviors in Heart Failure” (PRISM-HF).

Study design and methods

PRISM-HF is a sequential mixed methods study where in Phase 1, we collect quantitative data from a sex-balanced sample of 120 adults with HF, and in Phase 2, we collect qualitative data from ∼32–40 adults from this sample, aiming to: (1) quantify associations among physical frailty, SMM behaviors, and outcomes; (2) describe the experience of SMM behaviors for physically frail and non-physically frail adults with HF; and (3) identify the SMM behavioral needs of physically frail and non-physically frail adults with HF. At baseline, we measure symptoms, SMM behaviors, and physical frailty and collect clinical events at 6-months. We will use generalized linear modeling and survival analysis in Aim 1, directed content analysis in Aim 2, and triangulation analyses using an informational matrix in Aim 3.

Conclusions

This innovative study will investigate the behavioral underpinnings of physical frailty in HF, incorporate the patient's perspective of SMM behaviors in the context of physical frailty, and identify possible explanations for the effect of physical frailty on outcomes.

What is already known:

• •Frailty is highly prevalent and confers worse outcomes in heart failure
• •Some of the biological underpinnings of frailty have been examined
• •The self-care behaviors associated with frailty have not been studied

Alt-text: Unlabelled box

What this paper adds:

• •The design of a study to understand how self-care is associated with frailty
• •How a mixed methods study will help us learn about these associations

Alt-text: Unlabelled box

1. Introduction

Heart failure (HF) is estimated to affect 6.7 million adults in the U.S, (Martin et al., 2024) and about 1 of every 2 adults with HF also experiences physical frailty (Denfeld et al., 2017). There is strong evidence that physical frailty in HF is associated with worse clinical (Uchmanowicz et al., 2020) and patient-reported outcomes; (Buck and Riegel, 2011; Denfeld et al., 2018) however, we do not know the mechanism by which physical frailty develops and influences outcomes. To better understand the physical frailty phenotype in HF and why it is associated with worse outcomes, our group has worked to elucidate the biological underpinnings of physical frailty, (Denfeld et al., 2021, 2023) but the associations between physical frailty and behaviors remain unclear.

Engaging in self-care behaviors, particularly symptom monitoring and management (SMM) behaviors, is critical for adults with HF (Riegel et al., 2009, 2017). Beyond self-care maintenance (i.e. behaviors performed to maintain health and stability), adults with HF must adequately monitor for and manage common HF symptoms (e.g. recognizing dyspnea on exertion and calling a provider) (Alpert et al., 2017; Riegel et al., 2019). Effective SMM is associated with reduced hospitalizations (Lee et al., 2018) and improved quality of life (Lee et al., 2015). Physical frailty may be associated with SMM behaviors by blunting the monitoring of symptoms (e.g. too physically exhausted to notice changes in other symptoms) and hindering the management of symptoms (e.g. too weak to see a provider), which in turn impact outcomes.

The purpose of this article is to describe the background and design of an explanatory sequential mixed methods study entitled “Physical Frailty and Symptom Monitoring and Management Behaviors in Heart Failure” (PRISM-HF) designed to address this knowledge gap. Overall, we propose that physical frailty is associated with worse SMM behaviors, and in turn worse clinical event risk. Understanding this relationship will complement our emerging understanding of the biological profile of physical frailty in HF and provide the necessary data for developing and deploying effective biobehavioral interventions to address physical frailty and improve outcomes among patients with HF.

2. Background

Physical frailty, which affects almost half of adults with HF, (Denfeld et al., 2017) is defined as a biological syndrome of decreased physiological reserves across multiple systems and increased vulnerability to stressors (Fried et al., 2001). The phenotype of physical frailty includes weight loss, weakness, slowness, physical exhaustion, and low physical activity (Fried et al., 2001). The importance of physical frailty in HF has been recognized by an increasing number of HF guidelines and statements (Yancy et al., 2018; Denfeld et al., 2024) as there is strong evidence that physical frailty is associated with considerably higher mortality and hospitalization rates (Uchmanowicz et al., 2020) and worse patient-reported outcomes (Buck and Riegel, 2011; Denfeld et al., 2018).

Some work has focused on unraveling the biological profiles of physical frailty in HF, which may be linked to biological aging-related processes such as inflammation, neurohormonal activation, insulin resistance, and skeletal muscle dysfunction (Denfeld et al., 2023; Clegg et al., 2013; Boxer et al., 2014; Afilalo et al., 2014; Bellumkonda et al., 2017). Additionally, there is evidence of key sex differences in the presentation (Denfeld et al., 2021) and mechanisms (Denfeld et al., 2023) of frailty in HF with women significantly more likely to be frail than men (Davis et al., 2021). However, we lack an understanding of how physical frailty intersects with behaviors, especially self-care behaviors. Only one study has examined the relationship between frailty and self-care in HF; using a multidimensional measure of frailty, they found that those patients who scored high on the social components of frailty had better self-care capabilities (Uchmanowicz et al., 2015). No study has examined physical frailty, measured using the phenotype approach, and self-care behaviors in HF. Moreover, while there is some qualitative research on frailty among older adults, (Lekan et al., 2021) no studies have examined physical frailty in HF from the patient's perspective. In parallel with our emerging understanding of the biological profile of physical frailty in HF, there is a need to understand the behavioral profile of physical frailty in HF, with a particular focus on potential sex differences.

SMM behaviors are important facets of overall self-care, which is often defined as the naturalistic decision-making process of maintaining health and physiological stability through health-promoting practices and managing illness (Riegel et al., 2011). As opposed to the relatively little time spent engaging with providers, patients spend a large amount of time engaging in self-care (Riegel et al., 2017). As part of HF self-care, patients must engage in daily self-care maintenance behaviors such as weighing oneself, getting exercise, and avoiding getting sick. (Riegel et al., 2022) In addition, patients must monitor for changes in symptoms (e.g. dyspnea) and other physiological indicators (e.g. edema), evaluate the change, and engage in decision-making to address the change (e.g. calling a provider), as described in the Situation Specific Theory of Self-Care of HF (Riegel et al., 2022). These SMM behaviors are particularly important in HF as a means of mitigating symptom burden and avoiding adverse outcomes (Riegel et al., 2009; Lee et al., 2011, 2015, 2018). Thus, groups such as the American Heart Association recommend that patients with HF engage in SMM behaviors (Riegel et al., 2017; Heidenreich et al., 2022).

Despite knowing that SMM behaviors are important, it can be incredibly difficult for adults with HF to effectively engage in these behaviors given the physiologic and symptomatic complexity of HF coupled with multiple comorbidities (Riegel et al., 2022). For example, previous research has shown that mild cognitive dysfunction in HF is associated with reduced ability to recognize and manage HF symptoms and contact a provider. (Lee et al., 2013) In addition, there is an emerging understanding of how patients monitor symptoms through interoception, which is the set of processes through which an organism senses, interprets, integrates, and regulates signals originating from within the body (Chen et al., 2021). Interoception may play an important role in helping patients with HF recognize and respond to their symptoms (Garfinkel et al., 2015; Lee et al., 2024). Given that physical frailty reflects a phenotype of dysfunctional physiological processes, (Fried et al., 2021) physical frailty may affect multiple aspects of self-care behaviors including the ability to monitor symptoms, recognize symptoms through interoception, and respond in an appropriate manner.

3. Research design and methods

3.1. Theoretical framework

This study is based on the Middle-Range Theory of Self-Care of Chronic Illness (Riegel et al., 2019) broadly and the Situation-Specific Theory of Heart Failure Self-Care specifically (Riegel et al., 2022). In the updated Middle-Range Theory of Self-Care of Chronic Illness, symptoms were incorporated more explicitly given the integrated nature of symptoms in the self-care decision-making process. It is proposed that symptoms are most closely linked with the monitoring and management aspects of self-care. Symptom monitoring involves awareness and interpretation of bodily changes as symptoms, and symptom management involves responding to symptoms. As our previous research has found that physical frailty is associated with worse symptoms in HF, (Denfeld et al., 2018, 2021) we are focusing on symptom monitoring and management behaviors. We hypothesize that physical frailty may be a prominent factor that impacts the ability to be aware of and interpret symptoms (i.e. using interoceptive signals and processes (Chen et al., 2021)) and respond to symptoms.

This study also incorporates the “Cycle of Frailty” as described by Fried et al. (2001). The Cycle of Frailty unifies the markers of frailty that are associated with decreased physiological reserves and energetics; when there is a critical mass of these markers, the syndrome of frailty is identified. We propose that the syndrome of frailty affects the ability to perform self-care behaviors, especially symptom monitoring and management, which in turn affects outcomes. The overall framework for our study is depicted in Fig. 1.

Fig. 1.

Overview of framework for the Physical Frailty and Symptom Monitoring and Management Behaviors in Heart Failure (PRISM-HF) Study. We incorporated the middle-range theory of self-care of chronic illness and the cycle of frailty in our approach.

3.2. Study specific aims and hypotheses

The objectives of this mixed methods study are to (1) characterize the relationship between physical frailty, SMM behaviors, and outcomes, and (2) identify SMM behaviors, aligned with the patient's perspective, to target with biobehavioral interventions. This study utilizes an explanatory sequential mixed methods approach wherein in Phase 1, we collect cross-sectional quantitative (QUAN) data from a sex-balanced sample of 120 adults with HF, and in Phase 2, we collect qualitative (qual) data from ∼32–40 adults from this sample with the following aims:

• Specific Aim #1: Quantify associations among physical frailty, SMM behaviors, and outcomes in adults with HF.

• Hypothesis 1.1: Those physically frail will have significantly worse SMM behaviors compared with those not physically frail.

• Hypothesis 1.2: Those physically frail with worse SMM behaviors will have the highest clinical event risk.

• Specific Aim #2: Describe the experience of SMM behaviors among physically frail and non-physically frail adults with HF.

• Specific Aim #3: Identify the SMM behavioral needs of physically frail and non-physically frail adults with HF.

3.3. Study design

Sample. The sampling frame is adult women and men with a confirmed diagnosis of Stage C or D HF (by documented history, physical examination, and echocardiographic evidence) of ≥ 6 months who receive care at a single academic medical center and are responsible for their own health care decisions. See Table 1 for complete inclusion and exclusion criteria. All eligible patients are approached for voluntary participation using two convenience sampling approaches. First, we approach individuals from previous research studies who indicated interest in future research studies. Second, we approach new eligible individuals when they are scheduled for an out-patient cardiology visit. We plan to enroll up to 135 participants to reach our final evaluable and analytic sample of 120 (i.e. participants who complete all study visit requirements). The Oregon Health & Science University Institutional Review Board approved this study, and written informed consent is obtained from all participants.

Table 1.

Formal inclusion and exclusion criteria.

Inclusion criteria:	Exclusion criteria:
•Willing and able to provide informed consent •Age ≥ 18 years •Able to read and comprehend 5th grade English •Reachable by telephone/email/texting •Confirmed diagnosis of HF by physical exam and echocardiographic evidence (AHA/ACC Stage C or D HF) •Current HF symptoms (i.e. NYHA Class II-IV HF) •Diagnosis of HF ≥ 6 months	•Previous heart transplant or ventricular assist device •Major diagnosed cognitive impairment (e.g. Alzheimer's) •Major and uncorrected hearing dysfunction •Active psychosis that would preclude study participation •Concomitant terminal illness •Other medical conditions that would interfere with participant safety or data collection (e.g. life-threatening illness)

Abbreviations: ACC, American College of Cardiology; AHA, American Heart Association; HF, heart failure; NYHA, New York Heart Association.

Inclusion of women, minorities, and adults across the lifespan. To address critically important sex/gender differences that occur in HF, we use a 1:1 male:female enrollment strategy. To address noted health disparities by race and ethnic categories, we are aiming to oversample minorities, aiming for about 30 % minorities (the representation of minority racial/ethnic groups in Oregon is about 20 %). Additionally, to address the entire aging-HF continuum, we will enroll patients from across the adult lifespan (18+ years).

3.4. Data collection procedures

A schedule of assessments is presented in Table 2. After participants provide written informed consent, we schedule a study visit to collect baseline data collection. Data are collected from the participant using paper or electronic data collection (via the Research Electronic Data Capture (REDCap) system) and from the medical record; all data are stored in REDCap.

Table 2.

Schedule of assessments.

	Measurement	Baseline	6 months
Clinical characteristics
Socio-demographics	Socio-demographic questionnaire	x
Comorbidities	Charlson Comorbidity Index	x
HF clinical characteristics	Chart Abstraction (per SHFM (Levy et al., 2006) & AHA/ACC guidelines (Radford et al., 2005))	x
Physicalfrailty
Shrinking	Unintentional weight loss > 10lb/year	x
Weakness	Grip strength and 5-repeat chair stands	x
Slowness	4-meter gait speed	x
Physical exhaustion	FACIT-F	x
Low physical activity	CHAMPS and question about how much time was spent exercising in the past week	x
Symptommonitoring and management behaviors
Symptom monitoring	SCHFI Self-Care Monitoring Scale	x
Self-management	SCHFI Self-Care Management Scale	x
Consulting behaviors	EHFScB-9 Consulting Behaviors Scale	x
Interoception	Multidimensional Assessment of Interoceptive Awareness	x
Decision-making	Self-Care Decisions Inventory	x
Cognitivefunction	Montreal Cognitive Assessment	x
Symptoms
Dyspnea	Heart Failure Somatic Perception Scale-Dyspnea	x
Sleep problems	PROMIS Sleep-Related Impairment	x
Pain interference	PROMIS Pain Intensity and Interference
Depression	Patient Health Questionnaire-9	x
Anxiety	PROMIS Emotional Distress-Anxiety	x
Cognitive function	PROMIS Cognitive Function	x
Clinicalevents	Hospitalizations, emergency room visits, ventricular assist device, transplant, death		x

Abbreviations: ACC, American College of Cardiology; AHA, American Heart Association; CHAMPS, Community Health Activities Model Program for Seniors; FACIT-F, Functional Assessment of Chronic Illness Therapy-Fatigue Scale; PROMIS, Patient Reported Outcomes Measurement Information System; SHFM, Seattle Heart Failure Model.

3.5. Measurement

We collect socio-demographic and clinical variables, including HF history (e.g. etiology, duration of HF), diagnostics (e.g. lab values, echocardiographic metrics), medications (e.g. diuretics, beta-blockers), and treatment (e.g. implantable cardiac defibrillator placement) (Radford et al., 2005). We assess comorbidities with the Charlson Comorbidity Index (Charlson et al., 1987). We also collect data specific to the calculation of the Seattle Heart Failure Model score, (Levy et al., 2006) which will be used to account for the severity of HF in data analyses.

Physical frailty. Physical frailty is measured based on the validated Frailty Phenotype Criteria (Fried et al., 2001) using five physical frailty criteria (i.e. unintentional weight loss, weakness, slowness, physical exhaustion, and low physical activity) as adapted for the HF population (Denfeld et al., 2017). Unintentional weight loss is measured by a self-report of unintentional weight loss of ≥10 pounds in the past year. Weakness of both upper and lower extremities is assessed using grip strength and 5-repeat chair stands, respectively. Physical exhaustion is assessed using the 13-item Functional Assessment of Chronic Illness Therapy Fatigue Scale, (Yellen et al., 1997) which captures self-reported tiredness and inability to perform activities of daily living due to fatigue. Slowness (i.e. gait speed) is measured by clocking the time it takes a participant to walk 4 m at their usual pace. Physical activity is measured with one question about how much time was spent exercising in the past week; less than one hour indicates low physical activity. We also use the Community Healthy Activities Model Program for Seniors scale (Stewart et al., 2001) to capture more detailed data on activity levels. Cut points for each criteria are described in our previous study (Denfeld et al., 2017). Scoring will be as follows: no criterion met = non-physically frail, 1–2 criteria met = pre-physically frail, and ≥ 3 criteria met = physically frail.

Symptom monitoring and management behaviors. SMM is measured with the revised Self-Care of Heart Failure Index (SCHFI) v.7.2 (Riegel et al., 2019). Briefly, the SCHFI includes 29 items divided into 3 scales: self-care maintenance, self-care monitoring, and self-care management. For this study, we focus on the monitoring scale, which includes 9 items assessing frequency of behaviors and 2 items on how quickly symptoms were recognized and identified as HF-related, and the self-care management scale, which includes 8 items asking how likely the respondent would be to try behaviors commonly used to address HF symptoms. Response choices (range 1 to 5) for each item in the scale are summed and standardized to achieve a possible score of 0 to 100, with higher scores indicating better self-care. Symptom consulting behaviors are measured with the revised 9-item European Heart Failure Self-Care Behaviour (EHFScB-9) scale (Jaarsma et al., 2009). The EHFScB-9 measures both routine adherence behaviors (e.g. weighing oneself), and consulting behaviors in response to symptoms (e.g. contacting a doctor or nurse if one gains weight). For this study, we focus on the consulting behaviors subscale. Each item is rated using 5 response options; scores on the consulting behaviors subscale range from 4 to 20 with higher scores indicating a lower likelihood of contacting providers for guidance in response to changes in signs and symptoms of worsening HF.

Interoceptive sensibility is measured using the Multidimensional Assessment of Interoceptive Awareness (MAIA) Version 2 (Mehling et al., 2018). The MAIA is a 37-item self-report questionnaire with response options 0 (never) to 5 (always) with higher scores indicating more awareness of the trait being measured. The 8 interoceptive sensibility domains include Noticing, Not-Distracting, Not-Worrying, Attention Regulation, Emotional Awareness, Self-Regulation, Body Listening, and Trust. The MAIA has been used among patients with cardiovascular disease and demonstrated moderate to good reliabilities (Cronbach's α of 0.66 to 0.93) (Lee et al., 2024).

Decision-making is measured using the 27-item Self-Care Decisions Inventory (Page et al., 2022). This instrument asks about the extent to which contextual factors influence self-care decisions about symptoms of chronic illness and contains six contextual factor scales (external, urgency, uncertainty, cognitive/affective, waiting/cue competition, and concealment) that reflect naturalistic decision making. Each contextual factor is standardized to a range of 0–100 with higher values indicating that decisions about self-care are highly influenced by that factor. The reliability of this instrument has been established (Page et al., 2022).

Cognitive function and symptoms. We assess cognitive function and symptoms as important covariates. Cognitive function is assessed using the Montreal Cognitive Assessment version 8.1, (Nasreddine et al., 2005) a cognitive screening tool that can detect mild cognitive dysfunction in this population. (McLennan et al., 2011) We have previously demonstrated a strong association between MoCA scores and physical frailty. (Denfeld et al., 2017) We also assess self-reported cognitive function with the 6-item PROMIS Cognitive Function Short Form (Cella et al., 2010). The Cognitive Function Short Form assesses perceived cognitive deficits, including mental acuity, concentration, and memory with response options ranging from 5 (never) to 1 (very often, several times a day). Cronbach's α on the PROMIS Cognitive Function Short Form was 0.97 in a previous study of patients with HF (Denfeld et al., 2021).

Dyspnea is measured with the 6-item Heart Failure Somatic Perception Scale-Dyspnea Subscale (HFSPS-D) (v.3) (Jurgens et al., 2017). The HFSPS-D asks participants how much they were bothered by common HF symptoms within the past week with 6 response options (0 (not at all) to 5 (extremely bothersome)). The psychometrics of the HFSPS-D have been established previously (Jurgens et al., 2017). Sleep-related impairment is measured with the 8-item PROMIS Sleep-Related Impairment Short Form, (Cella et al., 2010) which focuses on self-reported perceptions of alertness, sleepiness, and tiredness over the past 7 days. There are 5 response options ranging from 1 (not at all) to 5 (very much). The reliability and validity of the Sleep-Related Impairment Short Form have been established; (Yu et al., 2011) Cronbach's α on the PROMIS Sleep-Related Impairment Short Form was 0.92 in a previous study of patients with HF (Denfeld et al., 2021). Pain intensity is measured with the 1-item PROMIS Pain Intensity Short Form (Cella et al., 2010). The Pain Intensity Short Form assesses how much a person hurts, on average, ranging from 0 (no pain) to 10 (worst imaginable pain). Pain interference is measured with the 4-item PROMIS Pain Interference Short Form (Cella et al., 2010). The Pain Interference Short Form measures the self-reported consequences of pain on relevant aspects of one's life, ranging from 1 (not at all) to 5 (very much). Reliability and validity of the Pain Interference Short Form have been evaluated across diverse clinical populations; (Askew et al., 2016) Cronbach's α on the PROMIS Pain Interference Short Form was 0.96 in a previous study of patients with HF (Denfeld et al., 2021).

Depression is measured with the Patient Health Questionnaire-9 (PHQ9) (Kroenke et al., 2001). The PHQ9 scores each of the 9 related DSM-IV criteria providing 4 response options (0 (not at all) to 3 (nearly every day)). The reliability and validity of the PHQ9 in the HF population have been established (Hammash et al., 2013). Anxiety is measured with the 8-item PROMIS Emotional Distress-Anxiety Short Form (Cella et al., 2010). The Anxiety Short Form measures self-reported fear, anxious misery, hyperarousal, and somatic symptoms related to arousal with response options ranging from 0 (never) to 4 (always). The reliability and validity of the PROMIS Anxiety items have been evaluated across diverse clinical populations (Schalet et al., 2016). Cronbach's α on the PROMIS Emotional Distress-Anxiety Short Form was 0.96 in a previous study of patients with HF (Denfeld et al., 2021).

Clinical outcomes. Six months after the QUAN data collection (physical frailty and SMM behaviors), we complete an electronic medical record review to collect data on clinical events, including hospitalizations, emergency visits, urgent device implantation, heart transplantation, or death. We collect data on the days between baseline data collection and events, the number of events, and the reasons for the event(s).

Semi-structured interviews. We conduct 45–60-minute semi-structured interviews with a subset of participants following the QUAN data collection at a private place convenient for participants or via an approved, secure virtual platform. Participants are selected based on their level of physical frailty (frail vs. non-frail/pre-frail) and scores on the SCHFI monitoring scale (< 70 vs. ≥ 70), ensuring equal numbers of women and men in each group. We designed the interview guide to address Specific Aims 2 and 3, drawing upon the study instruments and the updated Middle-Range Theory of Self-Care of Chronic Illness to develop interview questions. Examples of guiding questions include “What makes it more difficult to recognize or sense your symptoms?” and “What would be helpful to have in place to help track and evaluate changes in these symptoms?” We also included questions about frailty to explore what frailty or being frail means to participants within the context of their HF: “What does being frail mean to you?” and “To what extent do you think being frail (as you understand it) might be impacting your ability to recognize and track your symptoms?” The interviews are audiotaped to obtain an accurate account of study participants’ perspectives and experiences. Observations of the meeting that may help explain the audiotaped data are written as field notes after the interview has taken place.

3.6. Data management and analysis plan

Quantitative data management. Questionnaire data are entered into REDCap, a secure electronic data collection platform. Data on clinical characteristics, treatment, and physical frailty assessment undergo double entry verification. Standard descriptive statistics of frequency, central tendency, and dispersion will be used to describe the sample. We use StataMP v.17 (College Station, Texas) and R v4.4.1 (R: A language and environment for statistical computing, 2023) to perform analyses. To ensure rigor and reproducibility, the REDCap database includes an audit trail of any data entry and changes, and all code and analyses will be documented in a markdown file and verified by another member of the study team.

Quantitative data analysis (Specific Aim #1). After classifying participants by physical frailty status, we will use simple comparative statistics to estimate unadjusted differences between groups. Given our previous findings, we anticipate having few participants who are non-frail (do not meet any frailty criteria), and we will likely group non-frail and pre-frail together as one category. Before formal analyses, we will examine the proportions and likely causes of any missing data. In the case of data missing completely at random or missing at random, we will use principled methods of multiple imputation (Kenward and Carpenter, 2007) (e.g. incremental chained equations) (Royston, 2005) or full information maximum likelihood estimation to estimate the effects under Aim 1.

We will classify participants as having good or poor SMM behaviors using pre-established cut points for the SCHFI (poor = < 70 points; good = ≥ 70 points (Riegel et al., 2019)). Our primary behavior will be the SCHFI self-care monitoring scale for the detection of symptoms. For formal testing of Hypothesis 1.1, we will use multivariate logistic regression, accounting for covariates (e.g. age, HF characteristics, and cognitive function) based on previous research and bivariate testing. For formal hypothesis testing of Hypothesis 1.2, we will use continuous time survival analysis, (Rao and Schoenfeld, 2007; Denfeld et al., 2023) using guidance on competing risks, (Austin et al., 2016) to examine cardiovascular event-free survival from emergency room visit or hospitalization and all-cause mortality as a function of physical frailty and symptom monitoring. Kaplan Meier models, Mantel-Cox log-rank tests, and generalized Wilcoxon/Breslow tests will be used to evaluate differences in the distribution of time to first event. To estimate effect sizes, multivariate hazard ratios with 95 % confidence intervals will be calculated from Cox regression models.

Quantitative sample size justification. Assuming 50 % physically frail vs. non-physically frail based on our previous research, (Denfeld et al., 2017, 2021) mean SCHFI self-care management scores of 65 ± 20, (Lee et al., 2013) α of 0.05, and power of 80 %, a sample size of n = 120 will be needed to detect a difference in SCHFI self-management scores of 10.3 between two groups.

Qualitative data management. Audio recordings of interviews are transcribed verbatim and verified. We use the software program NVivo 14 (Denver, Colorado) to organize and manage the transcribed data. To ensure rigor and trustworthiness, we use the following procedures: verbatim transcript verification through audio-playback; maintaining inter-code reliability by having all coders (QED, SOH, SJR, LH) first code one transcript and then compare coding before dividing into coding pairs for the remaining subset of transcripts; an audit travel of coding and analysis decisions; and regular team meetings (Graneheim and Lundman, 2004).

Qualitative data analysis (Specific Aim #2). Qualitative directed content analysis, which is a methodology that provides knowledge and understanding of a phenomenon or process based on text data, will be used to analyze the transcribed data (Elo and Kyngäs, 2008; Hsieh and Shannon, 2005). We will develop a coding scheme based on a priori codes developed from the Middle-Range Theory of Self-Care of Chronic Illness, (Riegel et al., 2019) SCHFI, (Riegel et al., 2019) EHFScB, (Jaarsma et al., 2009) and the physical frailty measures with a focus on the experience of SMM in relation to physical frailty (Hsieh and Shannon, 2005) as well as inductive codes based on participants’ unique perspectives grounded in the data. All coders will independently read the first interview transcript and then meet to develop inductive codes. Once an initial coding scheme is developed, the transcript data will be analyzed line-by-line by each coder independently (Sandelowski, 1995). After coders have come to consensus on discrepancies and reliability of the coding scheme has been confirmed based on the first coded transcript, a codebook will be developed. Then coders will divide into pairs, independently code the next 3 transcripts, and meet to compare coding to achieve inter-rater reliability. At this point, the team will meet to refine the codebook by adding or deleting codes. The remaining transcripts will be divided among the coding pairs to code line-by-line, meeting to discuss and resolve any coding discrepancies. Throughout the iterative process of the data analysis, codes will be collapsed and categories and subcategories will be created. General thoughts and memos about the content will be maintained throughout this analysis process.

Qualitative sample size. Our goal is to describe SMM behaviors by the level of physical frailty and explain and build upon QUAN findings. Recruitment will proceed until saturation of the qual data is achieved (Guest et al., 2006). We anticipate reaching saturation with ∼32–40 participants: minimum 8 per group by level of physical frailty and good vs. poor symptom monitoring (i.e. frail/good monitoring, frail/poor monitoring, non-frail/good monitoring, non-frail/poor monitoring) (Burmeister and Aitken, 2012). We also recognize that sex may play a role in this experience given key sex differences observed in our prior research; (Denfeld et al., 2021) thus, we will balance the qual sample with equal numbers of women and men.

Mixed methods analysis (Specific Aim #3). An informational matrix will be created to compare and contrast qual findings with QUAN evidence of the relationship between SMM behaviors and physical frailty (Creswell and Plano Clark, 2018). Side-by-side comparison of the qual and QUAN findings for each group separately will be completed to assess concordance and discordance between the qual and QUAN findings. Then a comparison will be completed between the findings of the merged qual and QUAN data from each group. These findings will be interpreted to gain a more comprehensive understanding of the relationship. Finally, the integration of results will be represented in joint display tables (Guetterman et al., 2015) showing how the qual findings from Phase 2 further enhance understanding of the preceding QUAN data.

3.7. Anticipated results and implications

From the quantitative data, we anticipate finding that physically frail adults with HF will have worse SMM behaviors compared with non-physically frail adults with HF, and that those physically frail and with worse SMM behaviors will have the highest 6-month clinical event risk. From the qualitative data, we anticipate that the personal experiences of SMM will differ between those physically frail and those non-physically frail. Specifically, those who are physically frail will possibly have worse symptoms, more difficulty monitoring their symptoms (e.g. noticing, tracking symptoms), and more difficulty managing their symptoms (e.g. seeking provider input); although those who are physically frail may use different strategies to monitor and manage symptoms such as utilizing caregiver or social support systems. From the mixed methods analysis, we anticipate identifying specific SMM behavioral needs based on level of physical frailty that can be used to guide future intervention development.

This state-of-the-art study will greatly impact the field of physical frailty in HF by elucidating the behavioral profile of physically frail adults with HF using a self-care lens, identifying one possible explanation for the relationship between physical frailty and adverse outcomes in HF, and laying the foundation for biobehavioral interventions. We will use the information gained from this study, in conjunction with our past (Denfeld et al., 2017, 2018, 2021, 2023) and current research, (Denfeld et al., 2022) to develop biobehavioral frailty interventions.

4. Conclusions

The purpose of this study is to characterize the relationship between physical frailty, SMM behaviors, and outcomes, and identify SMM behaviors, aligned with the patient's perspective, to target with interventions. Leveraging the strengths of an explanatory sequential mixed methods approach, we aim to gain knowledge that is more than the sum of quantitative and qualitative methods alone. Specifically, it will allow our team to merge our understanding of physical frailty from a biological perspective with that of a behavioral perspective to develop biobehavioral interventions aligned with self-care theories and precision health frameworks to mitigate the adverse outcomes of physical frailty in HF.

Sources of funding

This work is funded by the National Institute of Health/National Institute of Nursing Research (R21NR020059). The work reported in this paper is also supported by the National Center for Advancing Translational Sciences of the NIH (UL1TR002369). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

CRediT authorship contribution statement

Quin E. Denfeld: Writing – review & editing, Writing – original draft, Visualization, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization. Shirin O. Hiatt: Writing – review & editing, Resources, Methodology, Investigation, Data curation. Susan J. Rosenkranz: Writing – review & editing, Resources, Methodology, Investigation, Data curation. S.Albert Camacho: Writing – review & editing, Resources, Investigation, Conceptualization. Christopher V. Chien: Writing – review & editing, Resources, Investigation. Nathan F. Dieckmann: Writing – review & editing, Software, Methodology, Formal analysis. Tyler B. Ramos: Writing – review & editing, Investigation, Data curation. Christopher S. Lee: Writing – review & editing, Validation, Methodology, Investigation, Conceptualization. Barbara Riegel: Writing – review & editing, Validation, Methodology, Conceptualization. Lissi Hansen: Writing – review & editing, Visualization, Validation, Supervision, Methodology, Formal analysis, Conceptualization.

Declaration of competing interest

None.