Developing a core outcome set for patient-reported symptom monitoring to reduce hospital admissions for patients with heart failure

Abstract

Aims

In patients with heart failure (HF), hospitalization rates are increasing, particularly for non-HF causes and over half may be avoidable. Self-monitoring of symptoms plays a key part in the early identification of deterioration. Our objective was to develop expert consensus for a core outcome set (COS) of symptoms to be monitored by patients, using validated single-item patient-reported outcome measures (PROMs), focused on the key priority of reducing admissions in HF.

Methods and results

A rigorous COS development process incorporating systematic review, modified e-Delphi and nominal group technique (NGT) methods. Participants included 24 HF patients, 4 carers, 29 HF nurses, and 9 doctors. In three Delphi and NGT rounds, participants rated potential outcomes on their importance before a HF or a non-HF admission using a 5-point Likert scale. Opinion change between rounds was assessed and a two-thirds threshold was used for outcome selection.

Item generation using systematic review identified 100 validated single-item PROMs covering 34 symptoms or signs, relevant to admission for people with HF. De-duplication and formal consensus processes, resulted in a COS comprising eight symptoms and signs; shortness of breath, arm or leg swelling, abdomen bloating, palpitations, weight gain, chest pain, anxiety, and overall health. In the NGT, a numerical rating scale was selected as the optimal approach to symptom monitoring.

Conclusion

Recognition of a range of HF-specific and general symptoms, alongside comorbidities, is an important consideration for admission prevention. Further work is needed to validate and integrate the COS in routine care with the aim of facilitating faster identification of clinical deterioration.

Heart failure, Patient-reported outcomes measures, Consensus

Novelty

• Patients with heart failure (HF) prioritize symptoms typical of HF, as well as more generic symptoms, before admission to the hospital.

• Common symptoms typical in HF such as fatigue are not always the best predictors of deterioration.

• Generic symptoms such as chest pain and anxiety are important to patients with HF and should be monitored alongside HF symptoms.

• Symptoms typical in HF such as dyspnoea are non-specific and highlight the importance of comorbidities in symptom assessment.

Introduction

Heart failure (HF) is a global pandemic affecting over 26 million people worldwide.^1,2 Heart failure is an irreversible clinical syndrome characterized by comorbidities, declining health status, episodic exacerbation of symptoms, and frequent unplanned hospitalizations. These hospitalizations may be for HF or comorbidities and over half may be avoidable.³ Costs associated with hospitalization account for over two-thirds of the estimated US$108bn spent on HF each year,⁴ a figure set to rise alongside a projected 46% increase in HF prevalence over the next decade.⁵

Early identification of people with chronic HF who are at high risk of hospitalization is a critical challenge. However, these patients are infrequently monitored, acute deterioration is poorly predicted,⁶ and patients can delay seeking help for up to 3 months before admission.⁷ Whilst monitoring and management of HF patients’ symptoms and signs is key to preventing deterioration and hospitalization,⁸ limited resources mean that most HF services can only provide short-term clinical or remote monitoring for newly diagnosed or early post-discharge patients.⁹ Tele-monitoring technologies have been used to remotely monitor physiological symptoms, signs, and biometrics in HF patients showing some success in improving self-care and reducing HF hospitalization and mortality.⁸ Implanted tele-monitoring systems such as CardioMEMS™ have also shown early promise.¹⁰ However, tele-monitoring usually focuses on cardiovascular status and evidence of effectiveness for reducing all-cause hospitalization remains inconclusive.¹¹ Furthermore, current tele-monitoring approaches are also resource intensive, often requiring expensive electronic technologies for physiological measurements and specialist input for risk assessment. Consequently, tele-monitoring is only used ad hoc for selected patients and often for a limited time period.

Improving patient self-monitoring of symptoms and self-management is, therefore, a core component of HF care pathways. Patient-reported outcome measures (PROMs) are validated measures that can be used by patients to monitor a range of HF-specific and general symptoms that are sensitive to changes in physiological and clinical status^12,13 and predictive of hospitalization.^14–16 There have been recent policy calls including in the recent recommendations by the American College of Cardiology and American Heart Association on ‘Key Data Elements and Definitions for Measuring the Clinical Management and Outcomes of Patients With Chronic Heart Failure’,¹⁷to increase the use of PROMS in routine HF care. An International Consortium for Health Outcomes Measurement also recently recommended the inclusion of 6-monthly monitoring of HF and mental health symptoms alongside health-related quality of life, to facilitate research and quality of care assessment at an organizational level.¹⁸ However, there is no current consensus on which symptom PROMs should be used in routine patient self-monitoring to help prevent unplanned admission for people with HF. This is important, given the increasing number of hospitalizations for non-cardiovascular causes in this population.¹⁹ Furthermore, prior to hospitalizations for HF, change in cardiovascular symptoms may occur later than the patient’s experience of deteriorating symptoms for a non-HF condition. Recently, there has been a move to using single-item ‘global’ questions for symptom and health assessment, which are highly correlated with outcomes and substantially reduce patient and healthcare resource burden.²⁰ Patient-reported outcome measures, therefore, provide a potentially cost-effective and inclusive approach to clinical monitoring in HF patients. We used consensus methods with HF patients, carers, and clinicians, to establish a core set of outcomes (CSO) comprising the most important symptoms and signs occurring prior to hospital admission that could be monitored by HF patients at home using single-item PROMs.

Methods

Study design

Consensus study using Delphi and nominal group technique (NGT) methods.

Participants

The study was conducted between 3 December 2020 and 27 April 2021. Participants included patients with HF, carers, and clinicians from the hospital and community setting in the UK and international clinical and patient networks.

Patients were included if they were aged 40 or above, had a new or prior diagnosis of HF at any stage of New York Heart Association classification, had experienced a prior unplanned hospital admission for any cause at or following their diagnosis of HF, and who were willing and able to give informed consent. Patients were identified by a HF outpatient clinic and by a community HF team at a multi-ethnic city in central England. Patient members from a patient-led HF network charity were also identified by the charity’s chief executive officer. Patients identified from the hospital and community trusts had their HF diagnosis verified by their clinical team as part of routine care, based on a combination of signs, symptoms, and objective evidence of cardiac abnormality using European Society Cardiology criteria.¹ Patients from the national patient network had self-reported HF and were under the care of a HF specialist. Patients were excluded if they were unlikely to survive hospital admission or in the terminal phase of HF. Carers were identified by invited patients, were aged 18 or over, and caring for a patient with HF fitting the inclusion criteria for patient participants.

Clinicians were registered nurses or doctors (medical doctor, cardiologist, or general practitioner) with a minimum of 1 year of clinical experience following registration and were directly involved with the care of patients with HF. Clinicians were identified by service managers from the same hospital and community trusts and from an open invite to a national HF clinical network in the UK, through personal contacts to known international HF experts and through HF specialist Twitter accounts.

Recruitment and consent

Letters of invitation and study information were sent to potential participants. Consent forms were then sent to participants who expressed an interest in participation. Participants were informed that they could withdraw from the study at any time, but data collected up to their withdrawal would need to be retained as it would form part of the consensus findings.

Ethical review

The investigation conforms with the principles outlined in the Declaration of Helsinki. Ethics approval for the study was granted by Cambridgeshire and Hertfordshire Research Ethics Committee (20/EE/0162).

Data and materials access

All data were pseudo-anonymized before being transferred to a secure University network with a data custodian with access under current governance procedures. Contact email addresses were kept separate from the main study data and destroyed immediately following study feedback to the participants.

Baseline characteristics

From patients and carers, we collected sociodemographic data on sex, age, ethnicity, living environment, and care needs of the patient; clinical data on HF duration, comorbidities, symptoms (type, frequency, recording, and reporting behaviour), and admission history. In addition, we asked patients about smoking and alcohol use, weight, and height, and we asked carers about their relationship to the patient and frequency of care. From clinicians, we collected data on the country, sex, professional registration, work setting, and experience of HF care.

Delphi questionnaire design

We conducted a systematic review in April 2020, to identify HF-specific and generic symptoms or signs for inclusion in the Delphi survey. The review comprised a two-stage search to identify symptoms or signs that had validated single-item PROM scales, that are responsive to change, and that could be monitored by patients or carers at home. Several scales are commonly used for assessing symptom intensity. Amongst them, the numerical rating scale (NRS) and visual analogue scale (VAS) are often used in clinical practice and form the basis of single-item measurement tools. These tools may form part of multi-item questionnaires or be stand alone. In the first stage of the review, we searched PubMed and COSMIN Database for systematic reviews of PROM multi-item questionnaires or stand-alone tools to identify relevant single-item symptom or sign scales. In the second stage, we searched for any additional individual symptom studies including stand-alone NRS or VAS tools. We also conducted an additional search for qualitative or quantitative studies identifying common symptoms or signs that are known to be relevant to admission for patients with HF. This final stage was conducted to cross-reference any of the symptoms or signs identified in the systematic search, for relevance and inclusion in the study (see Supplementary material online, Table S1 for all search strategies).

Modified e-Delphi process and statistical analysis

The Delphi method is an approach to addressing consensus using individual participant feedback to derive agreement where a range of opinions are likely to exist.²¹ A modified e-Delphi method^18,22 was chosen for this study, to achieve group consensus in three rounds, on important symptoms and signs occurring prior to hospital admission for addition to the core outcome set (COS). A web-based e-survey was used to present the set of symptoms to participants.

Round 1

Participants were asked to rate their agreement with three statements: (i) the following symptoms are an important trigger for a person to be admitted into hospital for their HF, (ii) the following symptoms are an important trigger for a person to be admitted into hospital for another cause (not their HF), and (iii) if the following symptoms changed (got better or worse), it would be easy for a person with HF to detect this change themselves or with help from a carer. Importance ratings were based on 5-point Likert scales of 1 ‘Strongly disagree’, 2 ‘Disagree’, 3 ‘Undecided’, 4 ‘Agree’, and 5 ‘Strongly agree’. Participants were invited to suggest additional symptoms if important ones were missing. Symptoms were carried forward to Round 2 if they scored a mean of >3 (agree or strongly agree) in either the patient/carer or the clinician group, for either type of admission, and >3 for ease of detection.

Round 2

Participants were presented with their previous score for each symptom and the mean group score from Round 1. They were then asked to re-evaluate and re-rate their responses. A McNemar test was used to evaluate opinion change for each mean symptom score between the rounds.²³ Participants were also presented with the 10 highest scoring symptoms from Round 1 and asked to rank the most important five symptoms for each admission type. In Round 2, symptoms were defined as achieving consensus and added to the COS, if they met all three criteria for at least one admission type: (i) a two-thirds threshold, meaning that at least 66% of participants either agreed or strongly agreed with the symptom’s importance, (ii) the symptom was stable between rounds (McNemar P ≥ 0.05), and (iii) more than one-third of participants had selected the symptom as one of the top five important symptoms. The remaining symptoms carried forward to Round 3, scored a mean of >3, or showed inconsistency in scores between rounds (McNemar P < 0.05).

Round 3

Participants were asked to score the remaining set of symptoms again. Symptoms were added to the COS if they met the two-thirds threshold and were stable. Any symptoms that were just under the threshold or that showed significant instability (McNemar P < 0.05) were taken forward to the Nominal Group Meeting for discussion.

Nominal group technique study

Following final selection of COS in the Delphi study, patients with HF and their carers were invited to attend an NGT consensus e-meeting,^24,25 with three rounds and feedback in between. The advantage of the NGT meeting is that agreement or disagreement can be debated. In Round 1, participants were presented with undecided symptoms in turn. They were first asked to write down one or two of their thoughts or ideas about the importance of the symptom. Next, in Round 2, each participant was given the opportunity to state one of their ideas. The ideas were then summarized and presented back to the group for discussion, before a final vote in Round 3. The NGT meeting was also used to seek agreement on the most appropriate type of measurement scale to be used by patients for monitoring symptoms at home. Participants were presented with a range of different types of validated measurement scales (such as Likert scales, VASs, or NRSs) and response types (such as numbers, statements, or pictures) relating to the symptoms selected in the COS. Each participant was asked to rate each type of measurement and response for its ease of use, taking account of time to complete, clarity, the range for detecting change, and different patient characteristics and abilities.

Sample size

Previous evidence suggests that 11–25 participants are required for consensus-derived agreement,²⁶ so recruitment aimed to sample at least 25 clinicians and 25 patients and carers.

Results

Item generation via systematic review

We identified 315 unique studies including 9 relevant literature reviews of PROMs in HF patients, 1 review of generic PROMs, and 11 stand-alone VAS or NRS tools used in HF (Figure 1; Supplementary material online, Table S1). From the reviews, we identified 26 unique HF and 30 unique generic multi-item PROM questionnaires of which six HF and seven generic questionnaires had relevant single-item VAS or NRS component scales. One HF multi-item questionnaire had to be excluded as it was not readily available (see Supplementary material online, Table S2). In all, from the combined searches we extracted 100 single-item scales (VAS or NRS) covering 34 symptoms, which were relevant to admission in patients with HF (see Supplementary material online, Table S3). These were condensed to 24 symptoms by grouping similar symptoms together, to be included in the Delphi questionnaire (Table 1).

Figure 1

PRISMA flow chart.

Table 1

Symptoms included in Delphi survey

Symptom or signs
Shortness of breath at anytime	Chest pain
Shortness of breath at night	Numbness or tingling
Cough at anytime	Nausea
Cough at night	Vomiting
Tiredness/fatigue	Diarrhoea
Change in taste	Constipation
Drowsy	Anxiety
Abdomen bloating	Sad or depressed
Feet swelling	Overall health
Arm swelling	Disturbed sleep
Palpitations	Urination problems
General pain	Forgetfulness

Participants

From 31 patients and carers and 45 clinicians that consented to participate, 24 patients, 4 carers, and 35 clinicians returned the first Delphi survey (Table 2 and Supplementary material online, Figure S1). The four carers reported information on behalf of a patient and are included in the patient summary. The mean age of patient participants was 59 (12.1) years, 12 (43%) were female, 22 (79%) were White, and 6 (21%) were South Asian. Most patients lived with a family member or friend (n = 23, 82%) and 10 (36%) received some level of care. Twenty-one (75%) had one or more comorbidities including diabetes (n = 6, 21%), chronic obstructive pulmonary disease, and arthritis (each n = 5, 18%). Nearly all patients reported symptoms (n = 26, 93%) with breathlessness (86%), lack of energy (61%), fluid retention (32%), and pain (29%) being the most common. Of the patients with symptoms, half reported experiencing symptoms daily. Despite being highly symptomatic, over half of the patients never recorded their symptoms and one in five only recording symptoms monthly or less. Nearly all patients had experienced a prior admission for HF (n = 25, 93%) and over half had experienced an admission for another cause (n = 15, 56%). Most patients experiencing an admission had noticed symptom changing prior to their admission for HF (75%) and for their admission for another cause (50%). Around half of the patients experiencing symptom change had noticed symptom change for weeks or months prior to both types of admission.

Table 2

Participants characteristics

Patients (n = 28; self-report n = 24, information from carers n = 4)		Missing data (n)
Age, mean (SD)	59.1 (12.1)	0
Female	12 (43)	0
South Asian	6 (21)	0
White	22 (79)	0
Lives alone	5 (18)	0
Lives with family or a friend	23 (82)	0
Independent	18 (64)	0
Receives care from family/friend or organization	10 (36)	0
HF for <3 years	19 (68)	0
HF for 3 or more years	9 (32)	0
Smokes or has smoked previously	13 (54)	4
Never smoked	11 (46)	4
Drinks alcohol currently	13 (54)	4
Does not drink alcohol currently	11 (46)	4
BMI, mean (SD)	31.6 (9.5)	4
Comorbidities
None	7 (25)	0
1–3	14 (50)	0
>3	7 (25)	0
Diabetes	6 (21)	0
Chronic obstructive pulmonary disease	5 (18)	0
Arthritis	5 (18)	0
Symptoms
Have symptoms	26 (93)	0
Top five symptoms		0
Breathlessness	24 (86)
Water retention	9 (32)
Lack of energy	17 (61)
Pain	8 (29)
Dizzyness/confusion	5 (18)
Muscle aches or cramps	5 (18)
Sleeping problems	5 (18)
Other (nausea/cough/walking difficulties/palpitations)	7 (25)
Frequency of symptoms in those with symptoms (n = 26)
Not every day	13 (50)	0
Every day or continuously	13 (50)
Symptom recording in those with symptoms (n = 26)		0
Daily	7 (27)
Monthly or less	5 (19)
Never	14 (54)
HF admission history
Prior HF admission	25 (93)	1
Prior HF admission in past 12 months)	6 (27)	6
Noticed symptoms changing prior to HF admission (n = 25)	18 (75)	1
Main symptom that changed prior to HF admission (n = 25)		7
Breathing	16 (89)
Swelling	1 (6)
Fatigue	1 (6)
Symptom change duration prior to HF admission (n = 18)		2
Minutes, hours, or days	9 (56)
Weeks or months	7 (44)
Non-HF admission history
Prior non-HF admission	15 (56)	1
Prior non-HF admission in past 12 months	7 (27)	2
Noticed symptoms changing prior to non-HF admission (n = 15)	7 (50)	1
Main symptoms that changed prior to HF admission (n = 7)		0
Breathing, chest pain, abdominal pain, water retention, swelling, bleeding	7 (100)	0
Symptoms change duration prior to non-HF admission (n = 7)
Minutes, hours, or days	4 (57)	0
Weeks or months	3 (43)	0
Clinician participants (n = 35)
Female	28 (80)
Country		2
England	26 (74)
Other	9 (26)
Profession		0
Nurse	29 (83)
Doctor	6 (17)
Clinical experience		0
Time qualified >10 years	35 (100)
Duration of HF work experience 1–10 years	19 (54)
Duration of HF work experience >10 years	16 (46)
Place of work		0
Hospital	15 (43)
Community	15 (43)
Outpatients or across both hospital and community	5 (14)
Experience of care of HF patients		0
Before admission	34 (97)
During admission	25 (71)
After admission	35 (100)

Most participating clinicians were nurses (n = 29, 83%) and 9 (17%) were doctors. All had over 10 years’ experience after qualifying and 16 (46%) had over 10 years’ experience in HF care. Just under half (n = 15, 43%) worked in the community setting and the same proportion worked in the hospital setting. The remainder worked across both settings.

Delphi rounds

Round 1

In relation to their importance prior to an admission, 19 symptoms achieved a mean score of >3.0 (agree or strongly agree) by patients/carers and/or clinicians for at least one admission type (see Supplementary material online, Table S4). Change in taste, constipation, forgetfulness, sad or depressed, and nausea all scored a mean of <3 from patients/carers and clinicians for both admission types and were eliminated. All retained symptoms achieved a mean score of >3 for ease of detection (see Supplementary material online, Table S5). Following free-text comments from participants, arm and feet swelling were combined into one symptom ‘arm or leg swelling’ and two new symptoms were added; weight gain and shortness of breath at rest.

Round 2

There were 24 patients/carers and 24 clinicians who returned the second survey. Eight symptoms (shortness of breath; at any time/at rest/at night, arm or leg swelling, abdomen bloating, palpitations, weight gain, and chest pain) met all the criteria for at least one admission type, for inclusion in the COS (more than two-thirds of participants agreed or strongly agreed with the symptom’s importance, more than one-third ranked the symptom in their top five and the symptom score was stable between rounds) (Table 3). Eight remaining symptoms achieved a mean score >3 or were unstable between rounds and were carried through to Round 3 (see ‘a’ in Table 3). Whilst diarrhoea had an unstable score between rounds, the mean score was low (<3) in both rounds and was eliminated in Round 2.

Table 3

Symptom selection; Delphi Round 2

Admission for heart failure					Admission not for heart failure
Symptom	Agree or strongly agree n (%)	Symptom rated in top five (%)	Symptom score mean (SD)	McNemar test (P-value)	Symptom	Agree or strongly agree n (%)	Symptom rated in top five (%)	Symptom score mean (SD)	McNemar test (P-value)
Palpitations	44 (92)	22 (46%)	4.0 (0.7)	0.56	Chest pain	42 (88)	35 (73%)	4.2 (0.8)	0.41
SOB at rest	43 (90)	35 (71%)	4.2 (0.8)	N/A	Palpitations	41 (85)	24 (50%)	3.9 (0.9)	0.71
Chest pain	40 (89)	36 (75%)	4.2 (0.9)	0.08	SOB at anytime	39 (81)	36 (75%)	3.9 (0.8)	0.78
SOB at night	42 (88)	26 (54%)	4.3 (0.8)	0.65	SOB at night	37 (77)	25 (52%)	3.8 (0.8)	0.06
Arm or leg swelling	42 (88)	34 (71%)	4.1 (0.8)	0.16	Arm or leg swelling	21 (45)	19 (40%)	3.3 (0.9)	0.17
SOB at anytime	40 (85)	34 (71%)	4.1 (0.9)	0.65	Abdomen bloating	18 (38)	12 (25%)	3.2 (0.9)	0.01
Abdomen bloating	32 (68)	18 (38%)	3.7 (0.9)	0.65	Numbness and tingling	16 (33)	8 (17%)	3.1 (0.8)a	0.07
Weight gain	32 (67)	35 (73%)	3.6 (0.8)	N/A	Drowsy	14 (29)	14 (29%)	3.1 (0.9)a	0.01a
Cough at night	21 (45)	9 (19%)	3.3 (0.8)a	1.00	Vomiting	14 (29)	11 (23%)	3.0 (0.9)a	0.48
Drowsy	17 (36)	5 (10%)	3.2 (0.9)a	0.03a	Urination problems	13 (27)	12 (25%)	3.0 (0.8)a	0.01a
Tiredness/fatigue	17 (35)	8 (17%)	3.1 (1.0)a	0.48	Overall health	12 (26)		2.7 (1.0)	0.32
Anxiety	14 (30)		3.0 (1.1)a	0.41	General pain	12 (25)	7 (15%)	3.0 (0.9)a	0.08
Urination problems	14 (30)		3.0 (0.8)a	0.13	Diarrhoea	10 (21)		2.8 (0.8)	0.03
Overall health	11 (23)		2.8 (1.0)	0.06	Tiredness/fatigue	8 (17)		2.8 (0.8)	0.10
Numbness and tingling	10 (21)	9 (19%)	3.0 (1.1)a	0.26	Cough at night	6 (13)		2.7 (0.8)	0.03a
Cough at anytime	9 (19)		2.9 (0.7)	0.71
Disturbed sleep	8 (17)		2.7 (0.9)	0.06

SOB, shortness of breath; N/A, symptom added in Round 2, so McNemar test not applicable. Symptoms shaded in grey were selected for the COS based on meeting the following criteria for at least one admission type (i) a two-thirds threshold, meaning that over 66% of participants either agreed or strongly agreed with the symptom’s importance, (ii) the symptom was stable between rounds (McNemar P ≥ 0.05), and (iii) more than one-third of participants had selected the symptom as one of the top five important symptoms.

a

Symptoms were carried forward to Round 3 based on (i) not already selected in the final set, (ii) a mean ≥ 3, or (iii) showing inconsistently between rounds (McNemar P < 0.05).

Round 3

There were 24 patients/carers and 20 clinicians who returned the third survey. Only one symptom, anxiety, met the two-thirds threshold, and was stable between rounds and was added to the COS. Drowsy, numbness/tingling, and urination problems were unstable and taken forward to the NGT. Fatigue was just under the threshold (63%) and so was also taken forward for discussion (see Supplementary material, Table S6). Overall, health was eliminated at Round 2, but was also taken forward given the strong supporting evidence of its association with outcomes in HF.

Nominal group technique

Eleven patients or carers participated in the NGT consensus e-meeting. After the three rounds, four out of the five symptoms were eliminated. The group agreed that fatigue was often ‘constant’ and ‘chronic’ and so ‘may not be sensitive to change’ and that ‘tiredness’ may be ‘temporary’ and not related to important clinical deterioration. They also agreed that drowsiness and numbness and tingling were more specific to individuals rather than a general presentation of people with HF and were also experienced less frequently. The group decided that urination problems may be more ‘to do with age’ or the ‘use of diuretics’ and was ‘too vague’ a term. They did agree that overall health could be related to symptom change and the feeling that ‘something is wrong on a daily basis’ or that ‘something is going wrong’. Whilst there was some discussion about the ‘vagueness’ of the term, the overall consensus was to include general health in the final COS (Central illustration). The group unanimously agreed that an NRS would be the best approach to symptom monitoring. The VAS was deemed simple to use but problematic when comparing current with previous ratings (see Figure 2 for a consensus summary).

Figure 2

Development of core outcome set.

Central illustration

Final core outcome set.

Discussion

This is the first study to establish a COS of patient-important symptoms, focused on preventing admissions in people with HF, and for use in routine monitoring. We followed a rigorous approach following COMET methods for COS development.²⁷ A systematic review of validated single-item symptom tools and published literature on symptom relevance was used to generate a comprehensive list of potential outcomes. Consensus methodology combining Delphi and NGT was used to establish eight symptoms for inclusion in the COS.

Improving the patient centeredness of care has become a healthcare priority worldwide,^28,29 with patient self-care and self-monitoring at the core of chronic disease management. In HF, symptom awareness, monitoring and management, alongside optimal medication adherence, are the key pillars of patient management.¹ Yet, symptom monitoring is currently focused on HF-specific symptoms, which do not address the increasing number of non-HF-related admissions experienced by people with HF. Self-monitoring of symptoms is also practised ad hoc with low patient adherence,³⁰ with only one-quarter of patients in this study reporting regular symptom monitoring. Validated single-item PROMs could be used by patients electronically to facilitate quick, simple, and regular monitoring and potentially improve adherence, by overcoming some of the challenges of filling out complex symptom diaries.³¹ Whilst clinical assessments of symptoms and health can be varied and non-consistently applied between clinicians or across services,³² PROMs have the added advantage of being standardized to the patient and reproducible across services. Furthermore, they have the potential to improve shared decision-making and facilitate active self-management and timely clinician interactions, to prevent deterioration.^33,34 Yet, whilst there is growing evidence in support of PROMs to inform and improve the quality of healthcare services and delivery, the increase of their use in health research has not translated into their application in routine patient self-management.³⁵

Heart failure guidelines focus on HF-specific symptoms associated with evidence of congestion, namely shortness of breath, ankle swelling, and fatigue.⁸ The COS includes four such symptoms and signs that characterize congestion: dyspnoea, arm or leg swelling, abdomen bloating, and weight gain. Their inclusion is understandable given the prior evidence on their importance in newly admitted HF patients.^36,37 Dyspnoea and arm or leg swelling were also selected for non-HF admission, indicating their non-specificity and relevance to HF comorbidities, such as respiratory and mobility disorders. Monitoring sudden weight change, alongside these symptoms, is therefore important to aid interpretation and intervention and comorbidities should be considered. Half of the symptoms selected were not directly related to HF or congestion. Palpitations and chest pain are valid choices, given that both are known predictors of health deterioration and admission for HF^38,39 and other causes.^40,41

Interestingly, anxiety and not depression was selected in the COS for HF admission, which is in contrast to prior HF evidence reporting depression but not anxiety, to be associated with mortality.⁴² In contrast to depression, which is underpinned by chronic and low positive affect, anxiety is a discrete, more fluctuating emotional state, typically containing positive and negative effects and components of fear and lack of control.⁴³ These mental components are known to influence physical functioning and quality of life⁴⁴ and may be better correlated with sudden deterioration and imminent clinical events, such as admission. General health has previously been shown to be sensitive to change in clinical status and HF severity^13,45 and predictive of HF⁴⁶ and other-cause admissions.⁴⁷ Whilst overall health was not selected in the COS in the Delphi rounds, participants in the NGT meeting agreed to its inclusion in the COS. The initial concern was that the concept may be too vague, but the group later decided that the non-specific nature of the symptom would mean that it could be used as a sign of general deterioration, relevant to any condition. Interestingly fatigue, which is a well-established symptom in HF, was not selected, demonstrating that common symptoms are not always the best predictors of deterioration. In the NGT discussion, participants commented that fatigue was often constant and chronic so not very sensitive. Some participants commented that fatigue could be a late sign of other symptom change.

This study responds to a key gap in the development of routine symptom monitoring using PROMs by patients with HF. The COS, developed using robust methodology, provides a patient-centred approach to symptom monitoring. The attrition rate of patient and carer participants was low, only 14% between the first and final Delphi rounds. The symptoms selected through expert consensus are relevant to patients prior to admission for HF and for other causes, which reflects the reality of the increasing number of non-cardiovascular admissions in this population. Whilst there are many potential causes of non-HF admission, some of the symptoms chosen, namely, anxiety, pain, and general health, are generic symptoms that apply across conditions. Study limitations include the exclusion of non-English-speaking participants and the low number of medical participants. The mean age of the patients was low compared with the national average,¹⁹ which may reflect healthy responders. That said, a wide range of ages were represented in the consensus agreement (33–85 years), with five patient participants older than 70 years, all of the patients had experienced an admission into hospital and different ethnicities were represented.

Conclusion

This study establishes a COS of symptoms focused on the key priority of reducing admissions in HF. A range of HF-specific and general symptoms were selected that should be considered, alongside comorbidities, for admission prevention. With development of the COS, we seek to promote standardized symptom monitoring by patients with HF at home. Further work is now needed to incorporate and test the use of single-item NRS in a simple and quick electronic device, for routine COS monitoring, and to facilitate faster detection of clinical deterioration.

Supplementary material

Supplementary material is available at European Journal of Cardiovascular Nursing online.

Acknowledgements

We would like to thank the patient-led HF network charity ‘Pumping Marvellous’ and the patients and clinicians who took part in this study.

Data availability

The data underlying this article cannot be shared publicly due to the privacy of individuals that participated in the study. The anonymized data will be shared on reasonable request to the corresponding author.

Funding

This work was funded by the National Institute for Health Research (NIHR-300111).

Author notes

Conflict of interest: K.K. is funded with an unrestricted educational grant from the NIHR ARC East Midlands and the NIHR Leicester BRC.