The prevalence of heart failure (HF) is increasing, reaching > 10% in individuals older than 70 years.1–3

It is the main cause of hospitalization in persons older than 65 years.1 Heart failure has a considerable impact on health care systems and comprises about 2.5% of overall health care expenditure,1 mainly due to admissions. However, the cost attributable to informal care, typically provided by women in Spain, represents the largest part of the overall health care cost (59.1%-69.8%).4 The disease has a marked impact on the quality of life of patients and their caregivers.4,5

Patients with HF are complex and of advanced age. Their considerable number of comorbidities and readmissions affect their clinical treatment and prognosis.1,2

Many of the admissions are considered avoidable.1,2 Because numerous meta-analyses6–23 and 1 metareview24 have shown that HF management programs significantly reduce the number of readmissions, these approaches are recommended in European clinical practice guidelines (I-A).2

Nonetheless, these meta-analyses reviewed a limited number of studies, which were, moreover, heterogeneous in terms of populations studied, their characteristics and usual care, geographical area, and health care system. The extraordinarily complex interventions are frequently scarcely described. Thus, it is difficult to evaluate which characteristics and clinical contexts favor program success and could be used to guide the organization of the different health care systems when setting priorities. Some meta-analyses have studied certain characteristics indispensable for success14,16 in a limited number of studies. Numerous meta-analyses11,14–16,19–24 and recent articles25–27 have mentioned the need to explore all of these elements in greater depth.

The objectives of this systematic review were the following: a) to update our understanding of the effectiveness of HF management programs not using remote monitoring while accurately describing the type of patient, the organization and contents of the intervention, and their ability to reduce mortality and readmissions, and b) to identify the determinants of program success.

The present study of 66 RCT performed in 18 countries and including 13 553 patients is the largest study of this type so far. Previous comparable meta-analyses published between 2004 and 2012 (19 in total) evaluated between 6 and 35 RCT. Our study included 16 trials published between 2005 and 2013 not included in the previous meta-analyses (supplementary material59,71,72,76,77,82,83,86–91,93–95).

This meta-analysis confirmed the reduction in all-cause and HF admissions and showed that HF programs reduce mortality.

Our results on mortality reduction agree with those of 6 previous meta-analyses6,11,13,14,17,21; another 6 studies9,10,15,18,19,23 showed favorable nonsignificant tendencies and another8 showed neutral results, although the trials included in this last analysis were performed before 2001. One meta-analysis7 showed a reduction in the combination of mortality and readmissions, another 3 studies12,16,20 did not analyze pooled mortality and found a significant reduction in 3, 4, and 1 studies, respectively, and 1 study22 did not evaluate mortality. The metareview by Savard et al.24 did not combine meta-analyses but found significant reductions in 6 of 13 meta-analyses and nonsignificant but favorable tendencies in the remainder.

The reduction in all-cause readmissions concurs with the results of 13 meta-analyses.6–9,11–15,18,21–23 Another meta-analysis10 (which only included 6 trials from 1998 to 2003) showed a nonsignificant tendency toward a reduction, and another20 only found significant differences in reduced admissions in 2 of 8 studies due to their heterogeneity. Four meta-analyses12,16,19,20 did not provide pooled data, with 1 of these16 showing a reduction in the combination of readmissions and mortality in 8 management programs and, finally, 1 meta-analysis17 did not measure all-cause readmissions. In the metareview by Savard et al.,24 10 of the 13 meta-analyses found reductions in all-cause readmissions.

Our HF readmission results agree with those of the 9 meta-analyses evaluating this variable6,7,11,13,15,17,22 and with controlled trials evaluating the impact of pharmacist collaborative care in patients with HF by Koshman et al.18 The case management programs reported by Taylor et al.19 showed tendencies toward a reduction. Koshman et al.18 showed that pharmacist-directed care (without collaborative care) failed to reduce admissions. In the metareview by Savard et al.,24 the 9 meta-analyses measuring changes in HF-related hospitalization found significantly fewer HF readmissions.

Regarding the determinants of readmission and/or mortality reductions produced by the programs, our meta-analysis indicates that study performance in 2002 or later improved the morbidity and mortality results, as in the meta-analysis by Göhler et al.14 The improved results of programs from 2002 onward were associated with greater baseline use of ACEIs and beta-blockers: greater baseline use of ACEIs was a significant determinant of the results, a finding previously reported by Phillips et al.10 and Göhler et al.,14 whereas greater use of beta-blockers was a nonsignificant determinant of the results. Programs performed in the United States also showed worse results than those conducted in other countries. American programs are associated with predominantly home-based care—possibly due to fewer HF clinics or units—a lower proportion of specialist HF cardiologists and nurses, a higher frequency of case managers, and lower percentage of drug optimization, as well as added coordination difficulties due to US health care being delivered almost exclusively by private sector providers.

Most patient characteristics were not significantly associated with the results, as expected, given that the studies comprised RCT with usual care and comparable intervention groups and as previously reported for NYHA and LVEF16 and HF with ischemic etiology or diabetes mellitus.14 In contrast, Göhler et al.14 linked a lower NYHA class to a greater mortality reduction. Similar to these authors, we found an inverse relationship between age and mortality,14 contradicting the analysis by Yu et al.,16 which reported that demographic factors had no effect.

As mentioned in previous meta-analyses, the care team is the main organizational characteristic associated with significantly reduced morbidity and mortality, specifically, the number of members6,13,14 and the specialization of the multidisciplinary team, and the inclusion of specialist HF cardiologists and nurses.6,10,11,13,14,16 Our study also adds dieticians and social workers, although the inclusion of the latter was only nearly significant. A cardiologist not specialized in HF and pharmacist nonsignificantly reduced the risk. Our meta-analysis agrees with the observation of Yu et al.16 that the incorporation of the primary care physician to the specialized HF cardiologist and nurse team could be beneficial and, like the present authors, Whellan et al.12 questioned whether the participation of primary care physicians in these programs could be effective without the participation of specialized HF cardiologist and nurse team.

Programs with home- and clinic-based follow-up significantly reduced admissions for HF, unlike those with telephone care alone.

Our study agrees with previous studies identifying a duration > 6 months as a factor with a significant effect on readmission reduction,14,17,21 although an intervention < 3 months was also significant (this period typically shows more admissions, which may explain its stronger influence). These findings differ from those of Roccaforte et al.,11 who found an association with studies lasting 3-6 months. Similar to Holland et al.,13 we conclude that the study quality does not affect the results, in contrast to Roccaforte et al.,11 who contend that the higher the study quality, the greater the mortality reduction.

Regarding the intervention, we agree with other meta-analyses10,16,21 in finding that the number of intervention components evaluated in our study significantly reduced readmissions or mortality.

The elements of the intervention significantly reducing the risk of mortality and/or readmissions in our meta-analysis were identified in previous studies: flexible diuretic regimen,16 early treatment of deterioration,6,12,16 psychosocial aspects, and care organization and coordination.16

Other elements of the intervention evaluated in this meta-analysis reduced the risk of morbidity and mortality to a greater or lesser extent, although nonsignificantly: clinical and analytical monitoring, easy access to care, intravenous diuretic administration, and drug optimization and titration.16

Although patient education and self-care are considered positive elements of the intervention in numerous meta-analyses,6,15–17,23 our meta-analysis showed that the reduction was generally nonsignificant, despite reducing the risk of morbidity and mortality. However, some education and self-care components were essential for a significant risk reduction: self-monitoring of signs and symptoms and early care-seeking, keeping a self-care diary, understanding of treatment and exercise, psychosocial aspects, and care coordination, as well as education and self-care evaluation and caregiver inclusion. In-person and protocol-driven education was nearly significant. Other education and self-care elements reduced the risk but nonsignificantly, such as understanding of diet and fluid intake and inappropriate drugs, the importance of adherence, the discharge plan, weight control, and use of pillboxes.

The clinical implications of this study are as follows: a) specific programs should be universally introduced for patients with HF, at least for those who have had a hospital admission, because there is sufficient evidence for the implementation of successful programs, which can also be cost-effective by reducing hospital admissions and relieving the social burden of informal care; b) these programs should be implemented in hospital-based HF units, community-based HF clinics, and in the home setting; c) the program design should include at least the factors shown in this meta-analysis to significantly reduce admissions and mortality and those reducing the risk, which are converted into management quality criteria, and d) before the widespread introduction of programs, they should be contextualized, adapted, and tested in each health care system, and their efficacy and manner of implementation should be studied, in conjunction with continual reevaluation with quality indicators.

New high-quality primary studies are required to explore in greater depth the key factors for the success of the programs and their intervention and organizational characteristics through a description of the characteristics of the patients, health care system, and social and geographic context of the programs. Their application should be studied in subgroups with preserved LVEF, women, patients with psychiatric disorders or important comorbidities, and the manner and content of the participation of the primary care physician in the specialized HF cardiologist and nurse teams. Programs including telemonitoring should be studied with the same degree of precision mentioned and their cost-effectiveness should be contrasted with those of programs omitting such methods.

The present meta-analysis corroborates in December 2014 the reduction in all-cause and HF admissions of most previous meta-analyses but includes many more RCT (66 in total) and is more up-to-date, in addition to including RCT from Spain. Our work also consolidates the reduced mortality reported in some of these meta-analyses and identifies the determinants of readmission and mortality results.

None declared.