Introduction

Heart failure (HF) remains a major public health burden across the globe. It is increasing in prevalence and associated with poor clinical outcomes and high healthcare costs [1]. Angiotensin-converting enzyme (ACE) inhibitors, angiotensin receptor blockers (ARBs), β-blockers (BBs) and mineralocorticoid receptor antagonists (MRAs) comprise the cornerstone of contemporary pharmacological treatment for HF patients with reduced ejection fraction (HFREF) [2, 3]. These disease-modifying therapies slow or retard progression of HF by attenuating the deleterious effect of neurohormonal stimulation [2].

Optimal use of ACE inhibitors, ARBs, BBs and MRAs in patients with chronic HF reduces mortality and morbidity in contemporary clinical practice [4, 5]. Despite this, data from large observational studies suggest that these drugs are under-prescribed in chronic HF patients [6]. However, the full extent and impact of under-prescribing of evidence-based pharmacological therapies among patients with HFREF in contemporary practice is unclear. It is possible that further improvement in outcomes will be achieved by closing any ‘treatment gap’ [7].

Few studies have examined the prescribing patterns among HFREF patients in the ‘real-world’ clinical setting [8, 9]. In addition, the association between prescribing patterns and patient demographics or comorbidities in this population remains unclear. Over the last two decades, concerted efforts to improve patient care through a series of educational and quality improvement programs have been reported [10–12]. The effectiveness of these interventions in improving the optimal use of evidence-based pharmacological therapies deserves further attention. Along these lines, we performed a systematic review to assess the inter-relationships between baseline clinical characteristics, prescribing patterns and treatment outcomes among chronic HFREF patients enrolled in contemporary HF registries and population-based surveys. We also aimed to quantify the ‘treatment gap’ among HFREF patients.

Our review focused on clinical registries and population-based surveys as compared to other study designs (including clinical trials) because they recruit a broader spectrum of HF patients and hence better reflect ‘real-world’ settings [13]. In addition, registries contain data that are well-defined, usually collected close to the time of commencement of treatment and involve systematic follow-up of patients. For these reasons, registries are considered a reliable source for assessing the quality of patient care and treatment outcomes in clinical practice [14].

Methods

Search methods

We performed this systematic review according to the recommendations in the PRISMA (Preferred Reporting Items for Systematic Review and Meta-analyses) statement [15, 16]. Two reviewers (KLC and IH) systematically searched the PubMed, MEDLINE (Medical Literature Analysis and Retrieval System Online) via Ovid, EMBASE (Excerpta Medical Database), CENTRAL (Cochrane Central Register of Controlled Trials) and CINAHL (Cumulative Index to Nursing and Allied Health Literature) databases for studies published in the years 2000–2015, and which reported prescribing patterns of ACE inhibitors, ARBs, BBs and MRAs among patients with HFREF. We also manually searched the reference lists of relevant review articles, systematic reviews, treatment guidelines, conference proceedings and online trial registries for relevant articles.

Search strategies included both Medical Subject Heading (MeSH) and text word search terms for systolic heart failure; left ventricular dysfunction; registry; survey; systolic heart failu*; chronic heart failu*; myocard* failu*; cardia* failu*; outcom*; registr*; survey*; left ventricular systolic dysfunction; heart failure with reduced ejection fraction; HFREF; reduced ejection fraction; guidelines adherence; guideline-directed; guideline-driven; prescrib* rates; prescrib* pattern; prescription pattern; treatment outcom*; population-based; population based; community-based; community based; evidence based treatment; under-prescrib*; under-utiliz*; optim* treatment. Abstracts were exported into Endnote X7.

Inclusion/exclusion criteria

Studies were eligible for inclusion if they were based on clinical registries or population-based surveys that recruited more than 200 stable patients with HFREF (defined as LVEF ≤ 40 % and measured by echocardiogram, nuclear multiple-gated acquisition scan, contrast ventriculogram or magnetic resonance imaging scan) and were not using intravenous (IV) diuretics or inotropes, and reported on prescription of ACE inhibitors, ARBs, BBs and MRAs. Studies which included stable patients with both HFREF and heart failure with preserved ejection fraction (HFPEF) were included if they reported specific data for patients with HFREF. We restricted our search to studies published in the English language.

Selection process

KLC and IH checked all titles and abstracts for studies that potentially met the inclusion criteria. Subsequently, both reviewers independently reviewed and extracted data from eligible full text articles. Discrepancies were resolved by consensus with a third reviewer (AT). We also measured the inter-rater reliability by calculating the level of agreement between the reviewers on the inclusion of the eligible full text articles.

Data extraction

Information was collected on data collection period, number of study participants, baseline clinical characteristics, medical history, prescription of ACE inhibitors, ARBs, BBs, MRAs and treatment outcomes. Selected studies were categorised as registry-based, survey-based, or a composite of the two. Sample size-weighted means and standard deviations (SDs) of all variables of interest were calculated. We compared the clinical characteristics, medical history and prescribing patterns according to age, sex and comorbidities. Where data existed, we examined the association between ‘guideline adherence index’ (GAI) and treatment outcomes. GAI is the proportion of patients prescribed the indicated drug and is commonly used to measure quality of care [17].

Definition of comorbidities

Comorbidities were defined as any concomitant disease in HFREF patients based on medical documentation. Data for the following conditions were extracted for our analysis: hypertension, ischemic heart disease, previous myocardial infarction (MI), atrial fibrillation (AF), stroke, diabetes mellitus (DM), chronic kidney disease (CKD) and chronic obstructive pulmonary disease (COPD)/asthma.

Assessment of risk of bias in included studies

KLC and IH independently assessed the risk of bias of included reports using the Risk of Bias Tool for Non-randomised Studies (RoBANS). Similar to the Cochrane Collaboration’s risk of bias tool (for randomised controlled trials), RoBANS is a validated tool which assesses selection, performance, detection, attrition and reporting biases and has been used in other systematic reviews [18–20].

Assessment of treatment gap

A small proportion of patients did not have documented New York Heart Association (NYHA) classification in some of the studies. Hence, the cumulative percentage of the sample size-weighted means of NYHA Class I–IV was less than 100 %. By assuming that those with and without documented NYHA classification were proportionate in terms of disease severity, we estimated the adjusted means of NYHA Class I–IV for the overall population. The adjusted means of NYHA classification were used to estimate the proportion of patients who were eligible for ACE inhibitors/ARBs, BBs and MRAs as recommended by the American College of Cardiology Foundation/American Heart Association and European Society of Cardiology chronic HF guidelines [2, 21–27]. The IMPROVE HF study, which characterised over 15,000 patients recruited from 167 settings of different practice levels, had previously reported that the overall prevalence of contraindication/intolerance to ACE inhibitor/ARBs, BBs and MRAs were 7, 7 and 18 %, respectively [28]. Taken together, we estimated the ‘treatment gap’ by measuring the proportion of patients who had an indication and no contraindication or limiting side effect but were not prescribed the recommended treatments.

The studies were grouped into US or Europe based. The analysis was repeated when the studies were re-grouped based on study designs; i.e., registry-based, survey-based and composite. We excluded McKee’s study [29] as clinicians at the time used Scottish Intercollegiate Guideline Network guidelines on systolic heart failure (number 35) published in 1999. Data from studies which only examined a specific subgroup of HFREF patients were excluded stepwise to evaluate the rigor of the estimated treatment gap.

Assessment of prescribed doses

The beneficial effects of the evidence-based therapies have previously been reported in patients who received ≥50 % of the guideline-recommended target doses [30]. Where data existed, an arbitrary cutoff point ≥50 % of the target doses of each medication was also used in our analysis to describe the prescribed doses in the individual studies.

Results

Search results

The initial search identified a total of 855 articles, of which 399 were duplicates. Of the remaining 456, 335 did not meet the pre-specified inclusion criteria from review of their titles and/or abstracts. We reviewed the full text of 121 articles and excluded a further 100. Reasons for exclusion are listed in Fig. 1. Two additional articles were identified through references. This resulted in 23 reports [5, 8, 9, 28, 29, 31–48], with data from eight registries and ten surveys being selected for the analysis (Fig. 1). Four of the reports were based on Registry to Improve the Use of Evidence-Based Heart Failure Therapies in the Outpatient Setting (IMPROVE HF), three from the EuroHeart Failure Survey (EHFS), and two from Impact-Reco Programme I and II. The inter-rater agreement for included articles was 87 %. The risk of bias in individual studies is presented in ‘Appendix 1.’

Fig. 1
figure 1

Study selection process

Full size image

Study characteristics

Data were collected between 1994 and 2012, and the number of patients per study ranged from 252 to 45,392. A total of 83,605 patients with HFREF were included. Characteristics of the individual studies, categorised as registry-based and survey-based, are presented in ‘Appendices 2 and 3.’ ‘Appendices 4 and 5’ summarise the characteristics of the US-based and Europe-based studies and the recommendations of the use of ACE inhibitors, ARBs, BBs and MRAs in published chronic HF practice guidelines. Six studies (three registries and three surveys) originated from the US and 12 from Europe (five registries and seven surveys). The study by Hebert et al. [39] did not provide baseline characteristics for the overall study population. Hence, data from this study were only included for the analysis that compared prescribing patterns between patients with and without CKD. Two of the registries (IMPROVE HF [9, 28, 35, 40] and National Cardiovascular Data Registry Implantable Cardioverter-Defibrillator Registry (NCDR-ICD) [44]) used LVEF ≤35 % to define systolic dysfunction. The rest used LVEF ≤40 %.

Baseline clinical characteristics and prescribing patterns

Tables 1 summarises clinical characteristics and prescribing rates in the selected studies. Overall, patients had a mean (SD) age of 68.7 (2.4) years, 70.1 % (2.9 %) were male and mean (SD) LVEF was 28.1 % (3.8 %). The mean (SD) prescribing rates were 79.8 % (13.8 %) for ACE inhibitors/ARBs, 81.4 % (12.7 %) for BBs and 36.4 % (10.2 %) for MRAs.

Table 1 Comparison of clinical characteristics in observational studies
Full size table

The Cardiovascular Research Network (CVRN) [38] and NCDR-ICD [44] registries only recruited newly diagnosed HFREF patients and patients who had undergone cardiac resynchronisation therapy, respectively. The Shah survey [46] recruited only HFREF patients with diabetes mellitus. When data from these three studies were excluded (n = 49,707), the mean (SD) prescribing rates of ACE inhibitors/ARBs, BBs and MRAs were 84.0 % (5.8 %), 76.7 % (15.0 %) and 38.0 % (9.3 %), respectively.

Treatment gap

The registries were predominantly US-based (92 %), while the surveys were mostly European (90 %) (Table 2). The estimated treatment gaps in the overall population were 13.1 % for ACE inhibitors/ARBs, 3.9 % for BBs and 16.8 %s for MRAs. In addition, the estimated treatment gaps in the registries (16.8 % for ACE inhibitors/ARBs, 0.5 % for BBs and 19.4 % for MRAs) were different to the surveys (5.2 % for ACE inhibitors/ARBs, 20.1 % for BBs and −7.9 % for MRAs). When each of the included studies was compared with the relevant practice guidelines, MRAs may have been prescribed for indications other than HF in some of the patients in the EHFS [42], FUTURE [32] and Scrutinio’s [45] surveys. Assuming that all NYHA Class III/IV patients were indicated for MRAs as recommended by the European guidelines 2001 and 2005, the prescribing rates of MRAs in these three studies exceeded the proportion of patients who were eligible for the treatment by 4, 4, and 9.5 %, respectively.

Table 2 Estimated treatment gap
Full size table

Prescribed doses

There were only four studies (all from Europe) which evaluated the prescribed doses compared with guideline-recommended target doses (Table 3). More than 70 % of patients were prescribed with ≥50 % of the guideline-recommended target doses of ACE inhibitors. In addition, only half of patients were prescribed with ≥50 % of the guideline-recommended target doses of BBs. When ACE inhibitors/ARBs were prescribed together with a BB, 53 % of the patients were treated with ≥50 % of the guideline-recommended target doses [32]. Prescribed doses of MRAs were not assessed in a standardised manner. All patients who were treated with spironolactone were prescribed with at least 12.5 mg/day in the IMPACT-RECO I studies [33]. A total of 83 % of patients prescribed with MRAs were treated with ≥50 % of guideline-recommended target doses in the EHFS survey [8]. In the IMPROVE HF registry, 73 % of the patients who were eligible for MRAs were treated at or above target doses [28].

Table 3 Prescribed doses
Full size table

Prescribing patterns according to demographics and comorbidities

Prescribing rates according to age, gender and comorbidities are presented in Tables 4 and 5. Table 6 summarises the independent predictors derived from multivariable analysis explaining the prescription of drugs. Patients who were elderly, female or had renal failure were less likely to be prescribed ACE inhibitors/ARBs and MRAs. In addition, increasing age, being female and the presence of asthma/COPD were strong predictors for the under-prescribing of BBs.

Table 4 Prescribing rates according to demographics and non-cardiovascular comorbidities
Full size table
Table 5 Baseline characteristics and prescribing patterns according to presence of chronic kidney disease
Full size table
Table 6 Independent predictors of prescription of pharmacological treatments for HFREF
Full size table

Treatment outcomes

From the selected studies, only two registries evaluated the impact of optimal prescribing of ACE inhibitors/ARBs, BBs and MRAs on outcomes. Prescribing these drugs according to guidelines was associated with lower mortality risk in multivariable analyses. In the University Hospital HEidelberg, the Klinikum LUdwigshafen and the TKH MAnnheim (HELUMA) registry, where surviving patients were followed for an average of 38 months (from 23 to 56 months), mortality risk was reduced by 27 % (adjusted HR 0.73, 95 % CI 0.57–0.92) when ACE inhibitors/ARBs, BBs and MRAs were prescribed according to 2005 European guidelines [36]. Similarly, in the Austrian Heart Failure Registry, in which patients were followed for an average of 2.8 years (1.6–4.4 years), mortality risk was reduced by 45 % (adjusted HR 0.55, 95 % CI 0.34–0.90) when ACE inhibitors/ARBs and BBs were prescribed at ≥50 % of the recommended target doses recommended in 2008 European guidelines [5].

Discussion

Our results highlight several pressing issues regarding the medical management of HFREF patients. Prescribing rates of evidence-based drugs appear to vary according to age, sex and comorbidities, and the treatment gap lies between 4 and 17 %. Nearly half of the HFREF patients were treated with ≤50 % of the target doses of BBs. Information about the true impact of following evidence-based strategy in treating HFREF patients were limited as we only found two studies reporting such data. Nevertheless, the importance of closing the treatment gap is highlighted by the finding that optimisation of treatment according to guidelines was associated with reduced mortality.

Prescribing rates reported in other studies which included patients with LVEF > 40 % [4, 49, 50] were lower compared with those included in our review. In a large international survey conducted in 15 European countries, the Improvement Programme in Evaluation and Management of Heart Failure initiative (IMPROVEMENT-HF) reported that the prescribing rates of ACE inhibitors, BBs and combination of ACE inhibitors and BBs were 60, 34 and 20 %, respectively. The doses prescribed were about 50 % of those recommended in European guidelines [49]. Several years later, the Medical Management of Chronic Heart Failure in Europe and its related costs (MAHLER) study were conducted as a multicenter observational study in six European countries. In that study, the prescribing rates of ACE inhibitors, ARBs, BBs and spironolactone were 69, 17.6, 53 and 28 %, respectively. GAI was high for ACE inhibitors (85.4 %) but lower for BBs (58 %) and spironolactone (36 %) [4]. Both studies did not have any pre-specified ejection fraction. In Japan, the Chronic Heart Failure Analysis and Registry in the Tohoku District 2 (CHART-2) registry demonstrated that the prescribing rates of ACE inhibitors/ARBs and BBs were 72.3 and 49 %, respectively [50]. In that study, LVEF < 50 % was used as the cutoff point to define HFREF.

We found older age, female gender and presence of comorbidities to be independent predictors of under-prescribing, which is accord with the results of previous studies [11, 51, 52]. The IMPROVEMENT-HF survey also found that prescribing of guideline-recommended treatment varied significantly between countries [52].

The American College of Cardiology Foundation/American Heart Association and European Society of Cardiology have made concerted efforts in implementing performance measures, quality improvement and educational programs, public reporting of hospital-level performance data and provision of easy-to-access brief summary of updated practise guidelines to clinicians [2, 3]. However, the number of patients dying or being readmitted in the year after hospital discharge remains high [53]. Our results raise important questions concerning the effectiveness of various improvement programs that have been implemented in the US and Europe over the last decade. There is an urgent need to consolidate efforts and possibly re-examine the effectiveness of current approaches in closing the treatment gap at all levels. Current approaches to overcome barriers to optimal utilisation of evidence-based pharmacological treatments are usually segregated (either prescriber-focused or patient-focused) [54–57] and lack cohesion. It is noteworthy that barriers such as health literacy, polypharmacy, adherence to medication, perceived susceptibility to adverse events, clinical inertia and treatment costs are closely intertwined and not exclusive. Hence, there must be synergy between various improvement programs and a focus should be on transitions of care between hospitals and the community. Another issue worth considering is the potential mismatch between what medications clinicians think their patients are taking and what their patients are actually taking.

Greater efforts to optimise the use of ACE inhibitors, ARBs, BBs and MRAs are warranted. Clinicians’ concerns about adverse effects and the often lengthy time needed for up-titration are recognised as major obstacles in reaching target doses in stable HF patients, rather than real intolerance to a drug [58].

One-way forward is greater use of registries for pharmacovigilance. The proposed INTERnational Congestive Heart Failure (INTER-CHF) registry may be an ideal platform to benchmark the quality of care provided by institutions or countries [59]. Secondly, there could be more collaborative initiatives among authoritative bodies; such as those between the American Heart Association and American Diabetes Association [60, 61], to promote practice improvement. Thirdly, organised programs which incorporate an integrated multidisciplinary approach to patient care should be implemented. For example, the involvement of nurses and pharmacists in HF management programs resulted in improved prescribing at target doses, medication adherence and quality of life, while reducing rates of readmissions, medication errors, mortality and costs [62, 63]. Fourthly, incorporation of electronic triggers in established electronic prescribing systems may further improve the adherence to contemporary practice guidelines. Finally, individually tailored measures to improve self-care strategies may be useful in improving medication adherence. When appropriate, innovative patient support measures via telemonitoring, reminders and educational letters circulated through handheld communicating devices and social media may be useful.

Study limitations

There are limitations to our study that warrant discussion. Our data were gathered only from the US and Europe. We did not find any relevant publication from other parts of the world. Other discrepancies in the study designs and data collection methods of the studies included in our analysis may have also affected our results. The definition of comorbidities may differ between studies and could have changed with time. In addition, temporal changes in the uptake of ACE inhibitors, ARBs, BBs and MRAs, in tandem with changes made to the site-specific formularies and practice guidelines from the year 2000–2015 may have confounded our findings. Our findings on prescribed doses and impact of guideline-directed treatment on outcomes should be read with care as the data are limited and selection bias may be present in a registry data.

Contraindication/intolerance rates reported by Heywood et al. [28] were used because individual patient data were not available, and we acknowledge that these may not have been representative. However, the study population was at least large and included patients with various mortality risk levels.

Finally, we also acknowledge that in clinical practice, withholding or withdrawing treatment or using low doses may be necessary in patients who develop hypotension, hyperkalaemia, decreased renal function or other side effects. Such instances do not represent suboptimal treatment.

Conclusion

Our results suggest that the use of evidence-based medications, namely ACE inhibitors, ARBs, BBs and MRAs, remains suboptimal among HFREF patients. In view of the complexity in managing HFREF patients, it is crucial for clinicians, policy makers and other healthcare stakeholders to consolidate efforts and re-examine the effectiveness of current strategies in closing the treatment gap at the systems as well as individual levels.