Implementing heart failure disease management in primary care: a mixed-methods analysis of the IMPACT-B study

Abstract

Objectives

Heart failure is an important health problem and patients are generally older with several comorbidities. Multidisciplinary heart failure care is therefore recommended. However, there is little evidence in real-world settings on how to involve primary care health professionals and how to evaluate such programmes. The main objective of this study is to integrate and evaluate several disease management interventions in a primary care setting.

Design

Prospective, non-randomised, observational implementation study with a mixed-methods process evaluation conducted over 3 years (2020–2022).

Setting

Primary care practices and two regional hospitals (one tertiary, one secondary) in the Leuven region, Belgium, serving approximately 100 000 inhabitants.

Participants

100 general practitioners (GPs) from 19 practices participated. A total of 96 patients were included in the disease management programme. Inclusion criteria for patients included high-risk status for heart failure (HF) readmission, based on clinical criteria. Exclusion criteria were not explicitly defined but participation required informed consent.

Interventions

Four interventions were implemented: (1) online HF education for GPs, (2) reimbursed natriuretic peptide (NP) testing, (3) patient education by trained primary care HF educators and (4) a structured transitional care protocol posthospital discharge.

Primary and secondary outcome measures

Primary outcomes included GP self-efficacy in HF management, NP testing rates, HF registration in electronic health records and patient self-efficacy (9-item European Heart Failure Self-Care Behaviour Scale (EHFScB-9)). Secondary outcomes included patient quality of life (Short Form-12 questionnaire (SF-12)), hospital readmission rates and provider satisfaction.

Results

GPs felt more competent in the management of HF after an online education (eight point increase in self-efficacy score after 6 months follow-up, (CI 2.9 to 13, p<0.001)). GPs conducted 314 reimbursed NP tests, half of which scored above age-specific thresholds for referral. After initiation, there was a significant increase in NP testing (Bayes Factor 64), with a total rate of 5.4 tests per 1000 patient years. The proportion of registered HF in patients aged 40 years or older in GPs’ electronic health records increased from 2.27% to 2.57% between 2020 and 2022. 96 patients were included in the disease management programme, 75% of these inclusions followed the transitional protocol after discharge for HF. There was significant improvement in patients’ self-efficacy after 6-month follow-up (an EHFScB-9 score change of 2.7 points, 95% CI 0.62 to 4.8), but no significant change in quality of life (an SF-12 score change of 0.15 points, 95% CI −1.1 to 1.4). Adherence to the transitional protocol depended on the presence of a specialist HF nurse during admission, 84% of all patients were seen within 14 days after discharge by their general practitioner. Time-trend analysis revealed an increase of HF as a secondary rather than primary cause of admission. Participating healthcare professionals reported satisfaction with the programme.

Conclusions

The IMPACT-B study demonstrated that an integrated disease management programme for HF could be implemented and assessed in routine clinical practice. The programme resulted in increased awareness and registration of HF in primary care, increased self-management of patients and improved follow-up after discharge, although these results should be interpreted cautiously given the uncontrolled pre-post study design.

Trial registration

Trial registration NCT04334447 (clinicaltrials.gov).

Correspondence to Dr Willem Raat; [email protected]

STRENGTHS AND LIMITATIONS OF THIS STUDY

• This implementation study, conducted over 3 years in a real-world Western European primary care setting, offers high external validity but is context-sensitive.

• The observational design introduces potential biases, including confounding from the COVID-19 pandemic, lack of blinding for subjective outcomes and selection bias.

• This study used routinely collected aggregated data from various healthcare settings, which are potentially incomplete or inaccurate.

Introduction

Heart failure (HF) is a prevalent disease affecting 1–3% of the general population and up to 20% of patients aged 80 years and older.¹ It has a large impact on morbidity, mortality and quality of life. Patients are generally older with several comorbidities, and provision of healthcare is thus often distributed across different healthcare settings and disciplines. HF care is therefore, almost by definition, multidisciplinary and should be integrated across the spectrum of community-based and hospital-based healthcare settings. This paradigm has been the focus of research in the last decades and is unanimously recommended by current guidelines.^2–4

Current guidelines offer little guidance on the involvement of primary care healthcare providers (PCPs) in HF care. In addition, there exists a large gap in the implementation of evidence-based interventions by PCPs for the adequate diagnosis and management of HF within several European healthcare settings.⁵ In the Belgian context of our intervention, general practitioners (GPs) report lack of confidence and experience in the diagnosis and management of HF, a threshold to use natriuretic peptides (NP) due to lack of reimbursement and a desire for a multidisciplinary pathway including general practice nurses as the preferred partner to delegate HF care. ^{6 7}We addressed these needs by introducing four guideline-recommended interventions within a real-world setting. First, e-learning based on guidelines by the European Society of Cardiology and the National Institute for Health and Care Excellence regarding the diagnosis and management of HF.^{2 3} Second, reimbursement of NPs which are very effective to rule out a diagnosis of HF.⁸ Third, HF education by trained nurses to improve patient self-management.⁹ Fourth, structured follow-up after discharge for HF.^{10 11}

The main objective of the present study is twofold. First, to describe the integration of these four guideline-recommended interventions into a multidisciplinary HF disease management programme and its regional implementation in routine clinical care settings. Second, to assess the implementation process through the use of a mixed-methods process evaluation adhering to the template for intervention description and replication (TIDieR) criteria for reporting non-pharmacological interventions in health research, such as intervention adoption, adherence and modifications.¹²

Methods

Design

This implementation study was a prospective non-randomised observational study with a follow-up of 3 years. Patient recruitment ran from November 2019 to December 2022 (trial registration: NCT04334447). We followed the Standards for Quality Improvement Reporting Excellence recommendations for reporting protocols of quality improvement trials.¹³

Participants

Study setting and criteria

This study transpired in the city of Leuven, which consists of the submunicipalities, Leuven, Wijgmaal, Wilsele, Kessel-Lo and Heverlee. This region has 102 833 inhabitants as of May 2023¹⁴ and is serviced by two local hospitals: a community hospital (Heilig Hart Leuven) and a university hospital (UZ Leuven). At the start of our implementation project in 2019, there were 194 GPs working in 67 practices. Within the first year of our study, we recruited 100 GPs, working in 19 different practices.

GPs were eligible to participate if they integrated an electronic health record in their clinical practice, used the functionality of e-prescribing for medication and were willing to participate in this observational study.

With regard to HF care, there is no reimbursement for NP testing, HF education or specialist HF nurses in the Belgian health system.⁵ Primary care in Belgium is accessible without a referral, and patients have the freedom to choose their healthcare provider, get a second opinion or even consult multiple providers simultaneously. There is also direct access to specialist care, and the vast majority of physicians are self-employed.¹⁵ Patients can opt in for a global medical record in exchange for lower co-payments, thereby allowing a single GP to manage their medical information. As of 2021, 83.3% of insured citizens and 93.2% of those aged 75 years or older had such a relationship with their GP.¹⁶

Participating patients

This was a programme with interventions at the practitioner level (online education, NP testing) and interventions at the patient level (HF educator). We could only measure patient participation (ie, how many patients were included in the disease management programme) at the patient level, by the number of patients who gave informed consent to receive HF education. Only GPs who participated in the IMPACT-B study could request a HF education for patients in a community setting. With regard to the HF educational session as part of the discharge protocol, all patients with HF and target domicile could be included regardless of whether their GP participated in the programme. We used a system of risk stratification to target patients for the structured discharge protocol and HF education. Since most predictive tools for 30-day readmission perform poorly in patients with HF with preserved ejection fraction,¹² we used a system based on expert consensus by local HF cardiologists, using a number of well-known risk factors for hospital readmission within 30 days.^13–15 We classified HF patients as high-risk if they scored positive on two or more of the following criteria:

• High need for diuretics defined as a daily dose of >1 mg bumetanide or >40 mg furosemide.

• Renal insufficiency, defined as an estimated glomerular filtration rate (eGFR) <45 mL/min.

• HF hospitalisation in the last 6 months.

• NYHA (New York Heart Association) III-IV or signs of congestion at discharge.

• Geriatric profile with a high risk of falling.

• Hospitalisation for chronic obstructive lung disease (COPD) in the last year.

• High risk of non-compliance.

Participating GPs, cardiologists and HF nurses were allowed to include patients at their own discretion. During the study, GPs could contact investigators for questions or practical assistance.

Disease management interventions

GP education

Each participating GP had to follow an online HF education session (a 30-minute narrated slideshow video). This education session (online supplemental file 1) summarised current HF guidelines (based on the guidelines available in 2020) and was reviewed by GPs with specific expertise in HF and cardiologists. We also created a website with up-to-date algorithmic flowcharts on the evidence-based diagnosis and management of HF in primary care.¹⁷

Natriuretic peptide testing

We offered all participating GPs access to reimbursed NT-proBNP tests. Participating GPs could request NPs using adhesive labels on their standard laboratory application forms. The GP education explicitly stated that according to guidelines, NT-proBNP can only be used to diagnose (rule out) HF and that there is no evidence in the use of therapeutic monitoring in known HF patients (or to assess deteriorating HF). This message was reinforced by the diagnostic flowchart made available to participating GPs.

Patient HF education

We invited nurses from two large regional nursing organisations as well as independent nurses to become HF educators in primary care in Leuven. They were trained to give HF educational sessions to patients. At the onset, this was one training session per year given by a specialist HF nurse with an academic degree in HF care, but in the course of the programme this was formalised into a 3-day postgraduate course. The contents of this HF educational session for patients included several key HF topics and self-care skills.² Participating GPs could apply for a HF educational session for patients who were classified as high-risk or if they judged a patient to need additional education.

Structured transitional protocol

We implemented a transitional protocol for high-risk HF patients to guide discharge from the hospital in the cardiology wards of UZ Leuven and Heilig Hart Leuven. This protocol consisted of a phone call to the patients’ GP to establish follow-up in primary care within a week as well as an educational feedback session with the HF educator 2 weeks after discharge.

Outcomes

Measurements at baseline

We registered contact data (telephone number and/or mail address) and sociodemographic variables (age, sex and type of HF) for patients who received a HF educational session and gave informed consent. If the educational session was given as part of the discharge protocol, the educator registered whether patients had already seen their GP.

General outcomes

We recorded all yearly hospitalisations for HF as primary or secondary cause of admission of patients in Leuven as well as the number of readmissions within 30 days. We also extracted these data for the previous 5 years, starting from 01 January 2014 to establish a time-series trend. We compared the evolution in hospital admissions with those for the overall patient population in the region of Vlaams-Brabant (± 1 million inhabitants).

Process evaluation

Table 1 offers a brief overview of the process evaluation, data sources and analysis. A TIDieR overview of interventions can be found in online supplemental file 6.

Table 1

Brief overview of process evaluation outcomes, data sources and analysis

	Intervention
	General practitioner education	Natriuretic peptide testing	Patient heart failure education	Structured transitional protocol
Outcomes	Online self-efficacy scores before and after education (pre-education, post-education and after 6 months) Provider experience	Number of tests per physician (within and outside of study context) Total number of tests in regional laboratory Number of registered heart failure diagnoses in the electronic health record Provider experience	Number of educator contacts Patient self-efficacy before session and after 6 months (EHFScB-9) Patient quality of life before session and after 6 months (SF-12) Provider experience	Number of educator contacts after discharge Proportion of patients who had general practitioner follow-up after discharge Provider experience
Data source	Online platform (Qualtrics) Interviews	Participating regional laboratory (Medisch Centrum Huisartsen Leuven) Participating general practitioners’ electronic health record Interviews	Baseline: educator registration form Follow-up: telephone or mail, conducted by investigator Interviews	Baseline: educator registration form Hospital administrative data Interviews
Data analysis	Random effects modelling Thematic analysis	Changepoint analysis with Poisson regression model Thematic analysis	Random effects modelling using calculated raw-sum scores	Changepoint analysis with Poisson regression model

EHFScB-9, 9-item European Heart Failure Self-Care Behaviour Scale; SF-12, Short Form-12 Questionnaire.

Timeline

Since the goal was to have a dynamic implementation process, we logged all significant programme changes and assembled these into a timeline.

GP education

GPs needed to complete an online questionnaire before and after the web-based mandatory training session.¹⁸ The questionnaire assessed GP self-efficacy in the diagnosis and management of HF. We compared pretest and post-test scores for the web-based training as well as follow-up scores 6 months after the training.

Natriuretic peptide testing

We looked at the number of tests conducted per participating physician within and outside the study context and the total number of tests performed in the largest regional lab (Medisch Centrum Huisartsen Leuven) in a yearly time trend series. We also studied the evolution of the number of registered HF diagnoses in the electronic health record.

Patient HF education

We registered the number of conducted HF educator contacts and patient self-efficacy and quality of life at baseline and after 6 months as scored by the 9-item European HF Self-Care Behaviour Scale (EHFScB-9) and Short Form-12 Questionnaire (SF-12), respectively.^{19 20}

Structured transitional protocol

Our main outcomes were the number of educator contacts after discharge and the proportion of patients who had GP follow-up after discharge. These were registered by the HF educator at the educational feedback session.

Qualitative analysis

We performed a qualitative analysis through focus group interviews with participating care professions’ (HF nurse, HF educator and cardiologist) to study their experiences with the HF implementation.

Data collection

At the patient level

We collected individual patient data for patients referred for a primary care HF educator consultation (by the GP or as part of the structured transitional protocol). Inclusion required informed consent. The HF educator completed a registration form with level of referral (GP or HF nurse in the hospital), contact information, sociodemographic data and an EHFScB-9 questionnaire assessing self-efficacy. We repeated the questionnaire after 6 months via telephone or mail. The HF educator recorded whether high-risk patients who gave informed consent saw their GP at the education session post discharge.

At the care profession level

We collected the number of NT-proBNP tests from the registry of the participating lab. We conducted a yearly audit in the GPs’ electronic health record to collect the number of registered HF diagnoses.

At the population level

Data collection for the number of HF hospitalisations occurred at the hospital level. We defined a HF hospitalisation as a hospitalisation with ICD-10 (Internation Classification of Diseases) code I50. For the transitional contact, we collected the number of HF educator contacts after discharge.

Data analysis

We used descriptive statistics to express population characteristics where appropriate. We used longitudinal random effects modelling to calculate effect sizes on the evolution in self-efficacy scores for GPs after online education, the prevalence of registered HF and the evolution in quality-of-life scores using the ‘lme4’ package in R.²¹ We conducted changepoint analysis with the ‘mcp’ package in R²² using Poisson regression models and Bayesian changepoint estimation for time trends in NP testing and hospitalisation. We conducted two-sided hypothesis testing of changepoint estimates by calculating Bayesian Factors (BFs) that express by what value our hypotheses become more or less likely given prior credence.

Regarding patient-related outcome measures, we calculated self-efficacy scores as raw sum scores. We did the same for the quality-of-life scores since we lacked normalised weights for our population and this method has been reported previously as a more valid representation of physical and mental health.²³

For our qualitative analysis, we held separate focus groups with participating GPs, HF educators, specialist HF nurses and cardiologists between March 2023 and November 2023. These were conducted in conference rooms and were led by WR, assisted by an observer (MS). We invited all participating HF educators, specialist HF nurses and cardiologists to participate. As for the GPs, we aimed at purposive sampling by inviting a mix of practitioners with a varying background (age, sex, practice type) and degree of participation in the programme. We constructed a topic list to guide interviews (online supplemental file 2). All interviews were audio recorded and transcribed verbatim. The audio files were deleted after transcription. We defined data saturation as the absence of new themes in the last two interviews. Participant characteristics can be found in online supplemental file 3. We analysed the qualitative data using open-source computer-assisted qualitative data analysis software.²⁴ One researcher independently coded each line of the focus group interview transcripts inductively, which resulted in a hierarchical tree organisation of descriptive themes. This set of themes was discussed by the whole research team.

Patient and public involvement

Throughout this study, GPs and the public in the region were involved in the research through regular reports and conferences organised by Zorgzaam Leuven.

Results

Timeline

Figure 1 depicts the different stages of the implementation cycle. We started this project in 2019 and conducted stakeholder meetings with representatives of all relevant medical professionals in the region, which resulted in a multidisciplinary protocol for the management of HF delineating each profession’s responsibility. At the end of 2019, we launched the transitional care protocol in the two different hospitals and conducted the first educational sessions. In 2020, we recruited GPs for online education and reimbursed NP testing and HF education. In 2021, we extended the transitional protocol to include those municipalities around Leuven that were not originally included in the protocol and integrated a diagnostic HF audit in CareConnect, the largest Belgian electronic healthcare package for GPs. The initial 1-day training for HF educators was also formalised into a 3-day postgraduate course at the graduate level (available to all interested primary care nurses) during the study period. In 2022, we conducted our process evaluation and held focus group interviews to assess the impact of different study interventions.

Figure 1. Timeline of the different stages in the IMPACT-B study implementation. EHR, electronic health record.

Online education

We previously reported results on the evolution of self-efficacy scores among participating GPs.¹⁸ Figure 2 visualises the evolution of self-efficacy scores provided by 50 GPs immediately before and after completion of the online education and after 6 months.

Figure 2. Longitudinal evolution on three different domains of perceived self-efficacy for general practitioners participating in the IMPACT-B study, colour-coded according to time of response.

50 GPs completed the online education and before-and-after scores. 16 GPs completed the follow-up questionnaire after 6 months. Six months after follow-up, the mean improvement of self-efficacy scores for each questionnaire item was 8.0 points (95% CI 2.9 to 13, p<0.001) on a 100-point Likert scale.

Thematic analysis

In our thematic analysis, GPs reported that the education influenced their general practice and noted the importance of combining education with other interventions to support improved HF management in primary care:

The education probably had some effects, if even just the fact that heart failure receives more attention now. I mean, I was glad to have followed it, to refresh my knowledge, but if it really changed my clinical practice? Maybe not, but everything together, the heart failure diaries, NP testing and all else did. (GP3)

Natriuretic peptides

Evolution in NP testing

During our study period, there were 314 reimbursed study tests conducted within the study framework. These were conducted by 80 unique physicians. Two findings stand out here:

First, half (50%) of all study samples scored above age-specific thresholds for the diagnosis of HF (see online supplemental file 4). Second, participating GPs requested a large number of NTproBNP tests outside of the reimbursed study framework (see online supplemental file 4).

This might have accounted for the large increase in regional NP testing, seen in the total number of reimbursed tests in MCH Leuven, the clinical laboratory for most GP practices in Leuven (figure 3). During the 3-year study period, participating GPs conducted a total of 1046 NP tests, which translated to 5.4 NP tests per 1000 patient years.

Figure 3. Evolution in yearly number of natriuretic peptide tests requested in MCH Leuven (our clinical laboratory partner). Pictured in (A) are the total number of tests (dark blue) and (for 2020-2022) the number of reimbursed study tests (light blue). A formal Bayesian changepoint analysis of this time trend is visualised in (B). The grey lines indicate a sampling distribution of priors within a SD of 10 from the estimated changepoint. The blue lines indicate the changepoint probability.

We conducted a changepoint analysis to analyse this time trend series to see whether there was a significant trend change after 2018. The mean estimated changepoint was during our study period, halfway in 2020 (lower bound 2020, upper bound 2021). A formal test of this hypothesis (changepoint after 2020) resulted in a very high BF of 94.

Evolution in registered HF diagnoses

During our study period, there was an increase in the proportion of registered HF diagnoses in GPs’ electronic health records. The mean proportion of patients with registered HF aged 40 years or older was 2.27% in 2020 (95%CI 1.51 to 3.09, n=606/27 273) with a significant absolute increase of 0.38% (95%CI 0.03 to 0.73, p=0.04) in 2021 and an increase of 0.30% in 2022 (95%CI −0.01 to 0.60, p=0.08, n=617/29 594) (translating to a relative increase of 13%).

Figure 4 depicts this evolution for all participating practices.

Figure 4. Evolution in prevalence of registered heart failure in general practitioners' electronic health records of patients aged 40 years or older throughout the study period. HF, heart failure.

Thematic analysis

In our thematic analysis, GPs reported increased awareness for HF in general practice:

Awareness mainly, that you say, ‘Oh I can do the NP test, yes, there are some signs that it is possible.’ I’m urged to make the connection between signs and diagnostic testing earlier. I don’t wait until there is manifest or decompensated heart failure. (GP7)

On the difference between the number of reimbursed and non-reimbursed NP tests, GPs reported the burden of extra administration (paper lab request or print-out and sticker) to request NP testing outside of the study context or when on house calls.

I’ve asked my colleagues: I think other things have stood in the way (with regards to increased registration and testing, author). But I do think we have been screening more actively, and with regards to the administration for a reimbursed testing, we steered clear of it and just requested it as usual. I think we are lucky to have a patient population that does not find it a problem when you accurately inform them what is the purpose of the test. (GP 8)

GPs reported mixed experiences with regards to increased registration of HF and the use of query-based audits in their electronic health records. Some reported reluctance or lack of time to register:

Yes, I did find this challenging to keep in mind, to actually implement such a screening… I feel, maybe unjustly, some hesitance to attach the label of heart failure on patients who are asymptomatic. Because it sounds dramatic, worse than it is often. So… I can’t say we do it often enough in our practice (GP 5)

Others mentioned the importance of a registered diagnosis:

It colors your diagnostic landscape. Of course it has an impact, you are more alert to certain things. You might adapt the advice you’re giving. Also, when you are looking at medications, when you’re deprescribing in older populations you are a lot more careful, when patients have a registered heart failure diagnosis. (GP 8)

Patient HF education

During our 3-year study period, 96 patients were included in the disease management programme, as measured by referrals for HF education. Three-quarters of these were part of the transitional protocol after discharge. Half of all referrals came from the cardiology service with a specialist HF nurse. Table 2 lists referral characteristics and sources.

Table 2

Source of referral and type of hospital setting for intake educational sessions

Characteristic	n=96*
Source of referral
GP	18 (19%)
Hospital	72 (75%)
Outpatient cardiology consultation	3 (3.1%)
Unknown	3 (3.1%)
Type of hospital setting
Not applicable: GP or outpatient	24 (25%)
Unknown	12 (13%)
With HF nurse	51 (53%)
Without HF nurse	9 (9.4%)

Hospital = referral for education after a hospital discharge. Outpatient cardiology consultation = referral for education after an outpatient consultation for patients with stable heart failure.

*n (%).

GP, general practitioner; HF, heart failure.

There was a significant improvement in patients’ self-efficacy score after 6 months (2.7 points, 95% CI 0.62 to 4.8), but no significant change in quality of life (0.15 points, 95% CI −1.1 to 1.4).

Figure 5 illustrates this difference in longitudinal evolution.

Figure 5. Evolution in patient-related outcome measures for patients who had an education by the heart failure educator. Panel A demonstrates the evolution in self-management scores as measured by the EHFScB-9 Scale. Panel B demonstrates the evolution of quality of life as measured by the SF-12 questionnaire. EHFScB-9, 9-item European Heart Failure Self-Care Behaviour; SF-12, Short Form-12 Questionnaire.

Thematic analysis

We had fewer referrals for education by GPs than anticipated. In our focus groups, this seemed to be primarily due to either a disinclination to change normal practice, reluctance to engage patients on the importance of their diagnosis or a preference to keep self-management motivation within the sphere of the GP:

I do notice a reluctance, where I’m not really sure what is holding me back. I think time. The time to explain to the patient: ‘You have this problem and a heart failure education might help. Can I refer you?’ (GP2)

I did think about it once, but this was with a very anxious patient, who had a myocardial infarction. And I considered, ‘OK, she has a diagnosis of heart failure and follow-up with a cardiologist.’ But then I really wondered, ‘If I send her to an educator, she will be even more distressed.’ (laughs) (GP5)

I do not think that this (lack of time or reluctance, author) plays a role. But I do think it depends on the patient, and I… I’m not saying I do the full education such as the one a specialist nurse has the time to do, but I do try to see those patients a few times after an initial heart failure diagnosis. I do try to give that education, at least partly myself. Because you know that individual and what is going on, is there anxiety or not? How far should I go in this education? So in our practice, I don’t think anyone requested an education. (GP6)

GPs that did request an educator reported increased self-management behaviours and (remarkably, given the general fear of anxiety) less anxiety and stress in patients:

Patients weighed themselves more often. They consulted saying ‘My weight was this much’, or they called to ask ‘Can I ingest a second (diuretic) pill?’ I really felt they had more control and less anxiety. (GP2)

Regarding the evolution in quality-of-life scores, educators remarked on the limited usability of the SF-12 questionnaire in a mostly geriatric population:

(On the questionnaire item of household work) People remarked: ‘I am 96 years old, what household work am I going to do? (ED1)

I mean, I did find that questionnaire hard to administer. General well-being was very hard for patients to respond to, and it is so subjective what you then register. (ED3)

Admissions for HF and discharge protocol

Figure 6 shows the temporal and regional evolution in HF admissions. Changepoint analysis revealed no significant trend changepoint after the initiation of our programme either in primary or secondary admissions (BF<0.001, p<0.001).

Figure 6. Evolution in yearly hospital admissions in Leuven and the larger region of Vlaams-Brabant with heart failure (HF) as a primary or secondary cause of admission between 2010 and 2022 (log-transformed scale). The evolution of admissions with chronic obstructive pulmonary disease (COPD), as primary or secondary cause of admission, is plotted for reference of trend evolution (grey dotted lines).

Online supplemental file 7 gives the evolution in readmission rates within 30 days within the region. These were generally much lower in wards with a specialist HF nurse (ie, those that adopted the discharge protocol), both before and after the initiation of IMPACT-B. With regard to the impact of our discharge protocol on readmission rates within Leuven compared with the wider region of Vlaams-Brabant, the picture is less clear. This is primarily because of the relatively limited amount of primary HF index admissions in the region, making the data more prone to outlier events such as the readmission rate for primary index admissions in 2022.

There were 144 completed checklists in the cardiology service with a HF nurse and 17 in the service without. Adherence to the follow-up protocol was high, 84% of all patients who had a HF educator consultation after discharge were seen by their GP (n=52/68).

Discussion

The IMPACT-B study demonstrated that an integrated disease management programme for HF could be implemented and assessed in routine clinical practice in Belgium. The programme impacted clinical practice across different healthcare settings, as evidenced in primary care by the increased self-perceived competency in the management of HF in participating GPs, elevated rates of diagnostic NP testing and registration of HF in GPs’ electronic health records and higher self-efficacy scores at follow-up in patients who had an educational intervention. To what extent these outcomes can be attributed to the constituent study interventions is hard to assess, but they likely enforced each other. In secondary care, we observed that adherence to the transitional protocol after discharge was dependent on the intramural presence of a specialist HF nurse and that HF trended to become a secondary rather than a primary cause of admission.

This study yields four important findings when seen through the prism of previous research on HF in primary care. First, online education for GPs on the diagnosis and management of HF is effective, particularly in combination with complementary study interventions. Second, despite our efforts to promote HF case finding and identification, there is still evidence for significant underutilisation of NP testing in primary care. We observed a prescription rate of 5.4 tests per 1000 patient years, whereas Valk et al demonstrated an average rate of 14.0 per 1000 patient years in the Netherlands, and had a total practice population as denominator rather than the practice population aged 40 years or older.²⁵ Nonetheless, there was an absolute increase of 0.30% in registered HF seen in GPs’ electronic health records over time, although the average prevalence of 2.57% was not far from the 3% seen in the OSCAR-HF study, which had a more intensive case finding procedure.²⁶

Third, the results of the self-management intervention reflect those of the ETIC and HICMAN studies,^{27 28} that studied the effect of patient education by GPs or their assistants on stable HF patients in primary care (whereas the majority of our patient population was recruited after a HF admission). These demonstrated no significant effect on general and disease-specific quality of life, but in the case of HICMAN, an effect on self-efficacy scores. By contrast, the OSCAR-HF study did demonstrate a significant improvement in disease-specific quality of life for stable patients with a specialist nurse intervention after 6 months of follow-up.²⁹ We hypothesise two possible reasons for this apparently conflicting evidence. First, the poor relationship of health-related quality-of-life questionnaires with true disease disability and severity in older patient populations is evidenced by the experience of our educators and previous reviews.^30–33 Second, the targeted patient selection in OSCAR-HF, where the specialist nurse might have selected those patients who had the highest odds of benefiting from the nurse intervention. A majority of patients in our study were included after a hospital admission and had a high burden of disease. The neutral findings seen here could therefore represent a floor effect in patients with advanced disease, something also seen in the large randomised PACT-HF study.³⁴

Fourth, our results indicate that HF is increasingly a secondary reason for hospital admission and that patients who are admitted to a ward with a specialist HF nurse are much less likely to be readmitted for any cause within 30 days. This diverging trend in admissions is in accordance with previous research, which demonstrated a decrease in primary HF readmission rates,^{35 36} as well as an increase in secondary readmission rates.^{37 38} In our study, this did not appear to be a result of different coding practices. A possible reason could be an increasingly geriatric patient profile, something experienced by the specialist HF nurses in our study (online supplemental file 5) and observed in large registries.¹ As to the difference in readmissions, it is impossible to assess to what extent the specialist HF nurse contributed to the lower rate of readmissions with our study design. However, given that specialist nurse interventions are associated with lower rates of readmission,^{10 39} there is likely a causal effect.

These findings highlight three different fronts of improvement for future research and policy on integrated care infrastructure for patients with HF. First, a further integration of data exchanges within healthcare professionals’ existing electronic health record software. In our study specifically, various health disciplines lamented their inability to share medication schemes, consultation reports or patient-related outcome measures electronically. Second, improved feedback on clinical performance within these electronic health records. Although we provided individual trimestral feedback reports by e-mail on the use of NP testing and HF education to each participating practice, this proved an insufficient incentive to increase engagement, particularly with regard to referrals for education. It would be interesting to explore whether a more tailored approach with electronic health record audits aimed at the identification of eligible patient populations, as well as the evaluation of quality of care, would result in significant changes in clinical practice. Such a population-based audit approach is currently being implemented for the use of antibiotics and quality of diabetes care in Belgium.^{40 41} Third, the need to integrate care for this increasingly geriatric population within and across different healthcare settings. At the hospital level, we identified a need for increased geriatric comanagement and support for specialist HF nurses. At the community level, we identified a need for integrated financing of a diagnostic HF biomarker and for supporting primary care nurses, who can take up advanced tasks in the management of chronic disease and have been shown to improve outcomes for type II diabetes.⁴²

To what extent these proposed follow-up interventions had an impact at the population level is an open question. Although participating healthcare practitioners were generally positive and there were modest improvements in healthcare outcomes, particularly self-efficacy scores, our study design with a pre-post analysis does not allow for a definite causal assessment, which is why we focused on guideline-recommended interventions with a robust base of evidence. However, the costs of our programme were relatively limited, with an annual expenditure of less than €10 000 and might be cost-effective downstream through a reduction in HF admissions. These limited costs were a consequence of integrating interventions within the existing workflow of healthcare practitioners and limited administrative burden, two important facilitators for implementation, which make scaling up to other regions relatively straightforward. Importantly, addressing the aforementioned barriers to implementation (lack of integrated communication, actionable feedback and cardiogeriatric care and personnel) also potentially improves care for other chronic conditions with implementation gaps in care, such as type 2 diabetes or chronic kidney disease, two frequently occurring conditions in patients with HF.

This study was conducted over a 3-year period in a real-world clinical setting, which enhances its external validity and relevance to routine practice, a particular strength of this study. The implementation of the disease management programme in a Western European urban healthcare system with direct access to both primary and specialist care reflects the complexity and heterogeneity of actual clinical workflows. However, this context-specific nature also limits generalisability to other healthcare systems with different organisational structures, reimbursement models or levels of integration between care sectors. In addition, there are several limitations. First, this is an observational study design, which introduces several potential sources of bias such as the possibility of selection bias, which was apparent in the participation of GPs and uptake of study interventions. There was a relative over-representation of larger GP group practices versus GPs working alone or in duo. Although we did not investigate this more formally, the relative underrepresentation of smaller practices is likely indicative of less resources such as administrative staff and less experience with study participation. Even though there were regular updates and reminders about the study interventions, both general and tailored to the individual practices, not all practices engaged equally with the study interventions, as evidenced by the skewed distribution of NP testing rates and relatively low percentage of physicians who completed the online education. This speaks to the difficulty of implementing change in primary care settings. A factor might have been the COVID-19 pandemic which occurred at the onset of our disease management programme. The concomitant fall-out on all levels of care might have impacted our study results, particularly in 2020 and 2021. Such an asymmetric uptake of practice changes is well known in innovation theory⁴³ and might have particularly influenced the results from our thematic analysis. We therefore conducted purposive sampling and invited across the range of implementation, from early adopters to non-adopters. In addition, COVID-19 could have influenced incidence rates of HF during follow-up, as it is associated with exacerbations of cardiovascular disease.⁴⁴ Finally, the lack of blinding in outcome assessment—particularly for subjective measures such as self-efficacy and quality of life—may have introduced response bias.

Second, the study relied heavily on routinely collected aggregated data from various healthcare settings, including electronic health records and laboratory databases. While this approach supports scalability and real-world applicability, it is susceptible to issues of data completeness and accuracy. It is not impossible, for example, that GPs requested NPs from other laboratories, which might have impacted results. However, the laboratory partner has a 90% market share in processing lab tests in the region of implementation, and the expected impact is therefore small. Likewise, we analysed aggregated data on hospital admissions based on ICD codes, and differences in admission coding standards throughout time might have influenced trends in time, although we saw no difference in COPD admissions, another prevalent chronic disease. In addition, the aim of this study was not to assess differences in admission or readmission rates for patients who underwent a study intervention, which would have required another study design and for which this study is underpowered, but rather to check whether routinely collected hospital data in the Belgian healthcare system can be used for the evaluation of healthcare interventions. In this regard, we believe this study highlights the unused potential of exploring patient trajectories through linking primary and secondary care data in our health system.

Conclusion

The IMPACT-B study demonstrated that an integrated disease management programme for HF could be implemented and assessed in routine clinical practice in Belgium. The programme resulted in increased awareness and registration of HF in primary care, increased self-management of patients and improved follow-up after discharge, although these results should be interpreted cautiously given the uncontrolled pre-post study design.

The authors would like to express their gratitude to all participating general practitioners, heart failure educators, cardiologists and specialist heart failure nurses, in particular Jan Kennes from UZ Leuven, as well as to the organisations of Zorgzaam Leuven and MCH Leuven for their involvement in this programme.

Data availability statement

Data are available upon reasonable request. Computing code and de-identified data are available from the author upon reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

This study involves human participants and was approved by University Hospitals Leuven Ethics Committee - S62838 in September 2019. Participants gave informed consent to participate in the study before taking part.

Footnote

Contributors WR conducted data curation and formal analysis. WR, MS, BV, MVdP, GVP and SJ contributed to funding acquisition. All authors contributed to the study methodology. WR was responsible for writing the original draft and all authors were involved in the design and conceptual development. All authors approved the final manuscript. WR is the guarantor of this work.

Funding We received funding from EBPracticenet to employ a project coordinator in 2019 and 2020. Funding for reimbursement of natriuretic peptides and a heart failure educator in primary care was provided by the RIZIV (Rijksinstituut voor Ziekte- en Invaliditeitsverzekering) through an integrated care pilot programme: Zorgzaam Leuven. The participating university hospital received a one-time unrestricted research grant from Novartis to expand the educational intervention after a hospital discharge in the region adjacent to Leuven (which was not covered through the integrated care programme).

Competing interests WR and MS report no conflict of interest. BV is holder of a chair in Population Health Management at the University of Leuven financed by Novartis, for which GVP and MVdP are co-chair. SJ is holder of a named chair in Cardiology at the University of Leuven financed by AstraZeneca.

Patient and public involvement Patients and/or the public were involved in the design, conduct, reporting or dissemination plans of this research. Refer to the Methods section for further details.

Provenance and peer review Not commissioned; externally peer reviewed.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

References

¹ Conrad N, Judge A, Tran J, et al. Temporal trends and patterns in heart failure incidence: a population-based study of 4 million individuals. Lancet 2018; 391: 572–80. doi:10.1016/S0140-6736(17)32520-5

² McDonagh TA, Metra M, Adamo M, et al. ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure: Developed by the Task Force for the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC) With the special contribution of the Heart Failure Association (HFA) of the ESC. Eur Heart J 2021; 42: 3599–726. doi:10.1093/eurheartj/ehab368

³ NICE. Overview | Chronic heart failure in adults: diagnosis and management | Guidance, 2018. Available: https://www.nice.org.uk/guidance/ng106

⁴ Heidenreich PA, Bozkurt B, Aguilar D, et al. 2022 AHA/ACC/HFSA Guideline for the Management of Heart Failure: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation 2022; 145: e895–1032. doi:10.1161/CIR.0000000000001063

⁵ HFPN. Heart failure policy and practice in Europe. London, 2020.

⁶ Smeets M, Zervas S, Leben H, et al. General practitioners’ perceptions about their role in current and future heart failure care: an exploratory qualitative study. BMC Health Serv Res 2019; 19: 432. doi:10.1186/s12913-019-4271-2

⁷ Smeets M, De Witte P, Peters S, et al. Think-aloud study about the diagnosis of chronic heart failure in Belgian general practice. BMJ Open 2019; 9: e025922. doi:10.1136/bmjopen-2018-025922

⁸ Taylor CJ, Ordóñez-Mena JM, Lay-Flurrie SL, et al. Natriuretic peptide testing and heart failure diagnosis in primary care: diagnostic accuracy study. Br J Gen Pract 2023; 73: e1–8. doi:10.3399/BJGP.2022.0278

⁹ Jonkman NH, Westland H, Groenwold RHH, et al. Do Self-Management Interventions Work in Patients With Heart Failure? An Individual Patient Data Meta-Analysis. Circulation 2016; 133: 1189–98. doi:10.1161/CIRCULATIONAHA.115.018006

¹⁰ Van Spall HGC, Rahman T, Mytton O, et al. Comparative effectiveness of transitional care services in patients discharged from the hospital with heart failure: a systematic review and network meta-analysis. Eur J Heart Fail 2017; 19: 1427–43. doi:10.1002/ejhf.765

¹¹ Facchinetti G, D’Angelo D, Piredda M, et al. Continuity of care interventions for preventing hospital readmission of older people with chronic diseases: A meta-analysis. Int J Nurs Stud 2020; 101: S0020-7489(19)30197-X. doi:10.1016/j.ijnurstu.2019.103396

¹² Hoffmann TC, Glasziou PP, Boutron I, et al. Better reporting of interventions: template for intervention description and replication (TIDieR) checklist and guide. BMJ 2014; 348: bmj.g1687. doi:10.1136/bmj.g1687

¹³ Goodman D, Ogrinc G, Davies L, et al. Explanation and elaboration of the SQUIRE (Standards for Quality Improvement Reporting Excellence) Guidelines, V.2.0: examples of SQUIRE elements in the healthcare improvement literature. BMJ Qual Saf 2016; 25: e7. doi:10.1136/bmjqs-2015-004480

¹⁴ Stad Leuven. Stad Leuven in cijfers. 2023.

¹⁵ Gerkens S, Merkur S. Belgium: Health System Review. Health Syst Transit 2020; 22: 1–237.

¹⁶ Gerkens S, Lefèvre M, Bouckaert N. Performance of the Belgian health system: report 2024. 1st edn. Belgian Health Care Knowledge Center (KCE), 2024.

¹⁷ Hartfalen Leuven. Extra informatie, Available: https://www.hartfalenleuven.be/extra-informatie

¹⁸ Raat W, Housiaux E, Smeets M, et al. How to Evaluate Online Education for General Practitioners: Development of a Tailored Questionnaire for Heart Failure Education. J Med Educ Curric Dev 2024; 11: 23821205241232497. doi:10.1177/23821205241232497

¹⁹ Jaarsma T, Strömberg A, Mårtensson J, et al. Development and testing of the European Heart Failure Self-Care Behaviour Scale. Eur J Heart Fail 2003; 5: 363–70. doi:10.1016/s1388-9842(02)00253-2

²⁰ Mols F, Pelle AJ, Kupper N. Normative data of the SF-12 health survey with validation using postmyocardial infarction patients in the Dutch population. Qual Life Res 2009; 18: 403–14. doi:10.1007/s11136-009-9455-5

²¹ Bates D, Maechler M, Bolker B, et al. Package ‘lme4’ 2009. Available: https://lme4.r-forge.r-proj.org/

²² Lindeløv JK. Mcp: an r package for regression with multiple change points. Open Science Framework 2020. doi:10.31219/osf.io/fzqxv

²³ Hagell P, Westergren A, Årestedt K. Beware of the origin of numbers: Standard scoring of the SF-12 and SF-36 summary measures distorts measurement and score interpretations. Res Nurs Health 2017; 40: 378–86. doi:10.1002/nur.21806

²⁴ Rampin R, Rampin V. Taguette: open-source qualitative data analysis. JOSS 2021; 6: 3522. doi:10.21105/joss.03522

²⁵ Valk M, Hoes AW, Mosterd A, et al. Time trends in the use and appropriateness of natriuretic peptide testing in primary care: an observational study. BJGP Open 2020; 4: bjgpopen20X101074. doi:10.3399/bjgpopen20X101074

²⁶ Smeets M, Vaes B, Aertgeerts B, et al. Impact of an extended audit on identifying heart failure patients in general practice: baseline results of the OSCAR-HF pilot study. ESC Heart Fail 2020; 7: 3950–61. doi:10.1002/ehf2.12990

²⁷ Vaillant-Roussel H, Laporte C, Pereira B, et al. Impact of patient education on chronic heart failure in primary care (ETIC): a cluster randomised trial. BMC Fam Pract 2016; 17: 80. doi:10.1186/s12875-016-0473-4

²⁸ Peters-Klimm F, Campbell S, Hermann K, et al. Case management for patients with chronic systolic heart failure in primary care: the HICMan exploratory randomised controlled trial. Trials 2010; 11: 56. doi:10.1186/1745-6215-11-56

²⁹ Smeets M, Raat W, Aertgeerts B, et al. Mixed-methods evaluation of a multifaceted heart failure intervention in general practice: the OSCAR-HF pilot study. ESC Heart Fail 2023; 10: 907–16. doi:10.1002/ehf2.14251

³⁰ Hickey A, Barker M, McGee H, et al. Measuring health-related quality of life in older patient populations: a review of current approaches. Pharmacoeconomics 2005; 23: 971–93. doi:10.2165/00019053-200523100-00002

³¹ Rizzoli R, Reginster J-Y, Arnal J-F, et al. Quality of life in sarcopenia and frailty. Calcif Tissue Int 2013; 93: 101–20. doi:10.1007/s00223-013-9758-y

³² van Leeuwen KM, Bosmans JE, Jansen APD, et al. Comparing measurement properties of the EQ-5D-3L, ICECAP-O, and ASCOT in frail older adults. Value Health 2015; 18: 35–43. doi:10.1016/j.jval.2014.09.006

³³ van Leeuwen KM, Jansen APD, Muntinga ME, et al. Exploration of the content validity and feasibility of the EQ-5D-3L, ICECAP-O and ASCOT in older adults. BMC Health Serv Res 2015; 15: 201. doi:10.1186/s12913-015-0862-8

³⁴ Van Spall HGC, Lee SF, Xie F, et al. Effect of Patient-Centered Transitional Care Services on Clinical Outcomes in Patients Hospitalized for Heart Failure: The PACT-HF Randomized Clinical Trial. JAMA 2019; 321: 753–61. doi:10.1001/jama.2019.0710

³⁵ Kimmoun A, Takagi K, Gall E, et al. Temporal trends in mortality and readmission after acute heart failure: a systematic review and meta-regression in the past four decades. Eur J Heart Fail 2021; 23: 420–31. doi:10.1002/ejhf.2103

³⁶ Akintoye E, Briasoulis A, Egbe A, et al. National Trends in Admission and In-Hospital Mortality of Patients With Heart Failure in the United States (2001-2014). J Am Heart Assoc 2017; 6: e006955. doi:10.1161/JAHA.117.006955

³⁷ Blecker S, Paul M, Taksler G, et al. Heart failure–associated hospitalizations in the United States. J Am Coll Cardiol 2013; 61: 1259–67. doi:10.1016/j.jacc.2012.12.038

³⁸ Varshney AS, Minhas AMK, Bhatt AS, et al. Contemporary Burden of Primary Versus Secondary Heart Failure Hospitalizations in the United States. Am J Cardiol 2021; 156: 140–2. doi:10.1016/j.amjcard.2021.06.036

³⁹ Lambrinou E, Kalogirou F, Lamnisos D, et al. Effectiveness of heart failure management programmes with nurse-led discharge planning in reducing re-admissions: a systematic review and meta-analysis. Int J Nurs Stud 2012; 49: 610–24. doi:10.1016/j.ijnurstu.2011.11.002

⁴⁰ Domus Medica. Betere diabeteszorg dankzij een nieuwe barometer, Available: https://www.domusmedica.be/actueel/betere-diabeteszorg-dankzij-een-nieuwe-barometer

⁴¹ RIZIV. Antibiotica: indicatoren voor het doelmatig voorschrijven van antibiotica door huisartsen, Available: https://www.riziv.fgov.be/nl/thema-s/kwaliteitszorg/indicatoren/antibiotica-indicatoren-voor-het-doelmatig-voorschrijven-van-antibiotica-door-huisartsen

⁴² Danhieux K, Buffel V, Remmen R, et al. Scale-up of a chronic care model-based programme for type 2 diabetes in Belgium: a mixed-methods study. BMC Health Serv Res 2023; 23: 141. doi:10.1186/s12913-023-09115-1

⁴³ Rogers EM. Diffusion of innovations. 5th edn. Simon and Schuster, 2003.

⁴⁴ van Royen FS, Venekamp RP, Bruijning-Verhagen PCJL, et al. Acute Respiratory Infections Fuel Cardiovascular Disease. J Am Coll Cardiol 2024; 84: 2468–70. doi:10.1016/j.jacc.2024.10.079