Author + information
- Received January 20, 2004
- Revision received April 8, 2004
- Accepted May 11, 2004
- Published online August 18, 2004.
- Finlay A. McAlister, MD, MSc, FRCPC*,* (, )
- Simon Stewart, PhD, FESC, FAHA†,
- Stefania Ferrua, MD‡ and
- John J.J.V. McMurray, MD, FESC, FACC§
- ↵*Reprint requests and correspondence:
Dr. Finlay A. McAlister, 2E3.24 WMC, University of Alberta Hospital, 8440 112 Street, Edmonton, Alberta, Canada T6G 2R7
Objectives The aim of this study was to determine whether multidisciplinary strategies improve outcomes for heart failure (HF) patients.
Background Because the prognosis of HF remains poor despite pharmacotherapy, there is increasing interest in alternative models of care delivery for these patients.
Methods Randomized trials of multidisciplinary management programs in HF were identified by searching electronic databases and bibliographies and via contact with experts.
Results Twenty-nine trials (5,039 patients) were identified but were not pooled, because of considerable heterogeneity. A priori, we divided the interventions into homogeneous groups that were suitable for pooling. Strategies that incorporated follow-up by a specialized multidisciplinary team (either in a clinic or a non-clinic setting) reduced mortality (risk ratio [RR] 0.75, 95% confidence interval [CI] 0.59 to 0.96), HF hospitalizations (RR 0.74, 95% CI 0.63 to 0.87), and all-cause hospitalizations (RR 0.81, 95% CI 0.71 to 0.92). Programs that focused on enhancing patient self-care activities reduced HF hospitalizations (RR 0.66, 95% CI 0.52 to 0.83) and all-cause hospitalizations (RR 0.73, 95% CI 0.57 to 0.93) but had no effect on mortality (RR 1.14, 95% CI 0.67 to 1.94). Strategies that employed telephone contact and advised patients to attend their primary care physician in the event of deterioration reduced HF hospitalizations (RR 0.75, 95% CI 0.57 to 0.99) but not mortality (RR 0.91, 95% CI 0.67 to 1.29) or all-cause hospitalizations (RR 0.98, 95% CI 0.80 to 1.20). In 15 of 18 trials that evaluated cost, multidisciplinary strategies were cost-saving.
Conclusions Multidisciplinary strategies for the management of patients with HF reduce HF hospitalizations. Those programs that involve specialized follow-up by a multidisciplinary team also reduce mortality and all-cause hospitalizations.
Heart failure (HF) is the most common discharge diagnosis in elderly patients, accounts for almost a quarter of all cardiovascular hospitalizations, and consumes 1% to 2% of total health care expenditures (1). Although a number of pharmacologic treatments have been shown to improve outcomes in patients with HF (2), the prognosis of these patients remains poor (3). Thus, there is a need for other approaches to management. Although the vast majority of HF research has focused on drug or electrical therapies, programs involving multidisciplinary teams are increasingly touted as a potential strategy for further improving outcomes in HF patients (2). Although some of the purported improvements may arise from better application of the evidence into practice, these multidisciplinary strategies may also better address the complex interplay between medical, psychosocial, and behavioral factors facing patients with HF and their caregivers.
Most initial studies of multidisciplinary interventions were non-randomized, “before and after” case series, raising concerns about their interpretation (4). However, in an earlier systematic review restricted to the 11 randomized trials published up until 1999, we demonstrated that management strategies for patients with HF that involved specialized follow-up by a multidisciplinary team reduced hospitalizations and appeared to be cost-saving (5). These conclusions were reinforced in a recently published systematic review incorporating another seven trials (for a total of 18 trials) (6). However, their effectiveness in improving other clinical outcomes (particularly mortality) was indeterminate in both earlier reviews owing to limited numbers of events. Importantly, the number of studies was also insufficient to allow comparison of the relative benefits of the different types of interventions. In the past five years, several more trials of multidisciplinary strategies for the management of HF have been published.
Thus, we updated our earlier systematic review to incorporate this new evidence and to investigate which types of programs are most efficacious.
Identifying relevant studies
We searched for randomized trials in Medline 1966 to 2003, Embase 1980 to 2003, Cinahl 1982 to 2003, Sigle 1980 to 2003, AMED (Allied and Complementary Medicine) 1985 to 2003, the Cochrane Controlled Trial Registry, and the Cochrane Effective Practice and Organization of Care Study Registry. Language restrictions were not applied. The following textword terms and medical subjectheadings were used: “case management (exp), comprehensive health care (exp), disease management (exp), health services research (exp), home care services (exp), clinical protocols (exp), patient care planning (exp), quality of health care (exp), nurse led clinics, special clinics” and “heart failure, congestive (exp).” Bibliographies of identified studies were hand-searched. We also contacted content experts and authors of the primary studies to identify any studies missed by the electronic searches.
Study selection and data abstraction
F.M. and J.M. independently reviewed the results of the search strategy and selected all studies reporting the impact of outpatient-based multidisciplinary management strategies on mortality or hospitalization rates in patients with HF. Studies were excluded if they: appeared in abstract form only, did not report either of the outcomes of interest, employed inpatient interventions only, did not utilize a multidisciplinary approach, or if they enrolled patients with multiple diseases and data for patients with HF were not reported separately or could not be determined after contact with the authors.
F.M. and S.S. independently assigned each reported intervention to one of four a priori defined groups: 1) a. multidisciplinary HF clinic; b. multidisciplinary team providing specialized follow-up but not in a hospital or practice-based clinic; 2) telephone follow-up or telemonitoring and enhanced communication with primary care physician (including advice to deteriorating patients to see their regular physician); or 3) educational programs designed to enhance patient self-care activities. It should be noted that patient education was a key component of allfour types of interventions (Table 1).These groupings were reviewed and approved at an ad hoc meeting of HF disease management investigators, including six of the authors of primary studies included in our systematic review (personal communication, Simon Stewart, February 26, 2004). All outcome data were extracted by F.M. and J.M. independently.
Outcomes from each study were assigned according to the intention-to-treat principle, and we accepted the definitions for each outcome used by the investigators in the primary studies. Original investigators were contacted to clarify the published data: authors for 14 of the 20 studies contacted provided further data. We defined “all-cause hospitalization rate” and “heart failure hospitalization rate” as the number of patients in each trial arm who were hospitalized at least once (thus, each patient could only contribute one event to these analyses). We defined “total hospitalizations” and “total heart failure hospitalizations” as the total number of hospitalizations in each trial arm—thus, these end points incorporate multiple re-admissions from the same patient.
Analyses were performed using the Meta-Analyst 0.998 software (J. Lau, New England Medical Center, Boston, Massachusetts). As the outcomes of interest were relatively common, we calculated risk ratios. Intention-to-treat analyses were done, and the DerSimonian and Laird random effects model was employed for all analyses. Because of the differences expected between studies (particularly in the usual care arms), a priori we decided to examine for statistical heterogeneity using Cochran's Q test. If the heterogeneity p value was <0.20, we elected not to pool the data but to examine outcomes in the a priori defined intervention groups separately (but only if there was no substantial heterogeneity within the subgroups). Sensitivity analyses, defined a priori, were conducted to look at the effects of duration of intervention, quality of trial (using the Jadad scale) (7), length of follow-up, and year of study completion on the summary risk ratios. We calculated adjusted indirect comparisons to compare different types of interventions according to the method of Song et al. (8).
Study selection and evaluation
Ofthe 853 citations identified in our search, 76 were eligible for inclusion in this systematic review (Fig. 1,QUOROM flow chart). However, 38 of these studies were excluded as follows: 21 were not randomized, 5 did not report the underlying diagnoses in the enrolled patients (and such data could not be obtained even after contacting authors), 4 did not include the outcomes of interest, 2 evaluated telephone follow-up in lieu of clinic visits rather than a multidisciplinary strategy, 2 did not include a “usual care” control arm, 2 were protocols for ongoing trials, 1 tested an inpatient-based intervention only, and 1 tested patient self-management alone. Of the 29 randomized trials included in this systematic review, 7 reported results in more than one publication (only the main publication from each trial is referenced here) (9–15).
Disagreement among the two reviewers regarding eligibility of the studies occurred on seven occasions for a kappa value of 0.77, and disagreement on type of intervention occurred on four occasions for a kappa value of 0.86. All disagreements were resolved by consensus.
Studies included in the systematic review
Quantitative data synthesis
Two of the 22 trials reporting mortality end points found a significant difference between the intervention and control patients during the pre-specified study periods (Table 2);a third group of investigators (11) subsequently reported a significant difference after 18 months of follow-up (2 deaths in the intervention arm and 9 in the control arm, p = 0.02) (38). The summary risk ratio (RR)for all 22 trials (3,781 patients) was 0.83 (95% confidence interval [CI] 0.70 to 0.99); however, although heterogeneity testing was not significant (p = 0.15), there was a strong enough trend that we thought the data should not be pooled across all trials. Rather, we examined results within the a priori defined intervention subgroups. Multidisciplinary teams providing specialized follow-up were associated with significant reductions in mortality (RR 0.75 [95% CI 0.59 to 0.96], number needed to treat [NNT] 17), with similar results whether that follow-up was done in HF clinics (RR 0.66, 95% CI 0.42 to 1.05) or not (RR 0.81, 95% CI 0.65 to 1.01). On the other hand, no mortality reductions were seen with those strategies that were based on telephone follow-up with instructions to see regular physician if deteriorating (RR 0.91, 95% CI 0.65 to 1.29) or those programs that emphasized enhanced patient self-care activities (RR 1.14, 95% CI 0.67 to 1.94). The adjusted indirect comparison (8) confirmed that mortality reductions were greater with multidisciplinary teams providing specialized follow-up (whether in HF clinics or non-clinic settings) than programs that relied on enhanced self-care activities alone (adjusted RR 0.66, 95% CI 0.46 to 0.94) or programs that employed regular telephone contact and follow-up with primary care practitioners (adjusted RR 0.82, 95% CI 0.54 to 1.25).
All-cause hospitalization rate
Whereas three trials reported a statistically significant reduction in hospitalizations for any reason in patients randomized to the multidisciplinary management strategies, 18 failed to detect any significant difference and 1 reported excess hospitalization rates in the intervention arm (Table 2). Although for the 23 trials (4,313 patients) the summary risk ratio of 0.84 (95% CI 0.75 to 0.93) is consistent with a beneficial impact from the interventions, there was significant heterogeneity in the results (p < 0.01).
This statistical heterogeneity again highlights the differences between the intervention categories noted earlier. Those trials evaluating the impact of frequent telephone contact with advice to those patients exhibiting deterioration to see their regular physician demonstrated no impact on all-cause hospitalization rates (RR 0.98, 95% CI 0.80 to 1.20). This is in contradistinction to trials evaluating multidisciplinary teams providing specialized follow-up (RR 0.81 [95% CI 0.71 to 0.92] and NNT 10, with similar results for follow-up in clinic (RR 0.76, 95% CI 0.58 to 1.01) or non-clinic (RR 0.81, 95% CI 0.72 to 0.91) settings, or programs that emphasized enhanced patient self-care activities (RR 0.73, 95% CI 0.57 to 0.93). The adjusted indirect comparison (8) confirmed that, compared with programs that employed regular telephone contact and follow-up with primary care practitioners, all-cause hospitalizations were affected more by programs that emphasized enhanced patient self-care activities (adjusted RR 0.75, 95% CI 0.63 to 0.88) or multidisciplinary teams providing specialized follow-up (adjusted RR 0.81, 95% CI 0.63 to 1.04) compared with telephone contact and follow-up with primary care physician.
HF hospitalization rate
Six of the 19 trials reporting this end point found statistically significant reductions in the need for at least one HF hospitalization with multidisciplinary management strategies, and the pooled effect estimate suggested a substantial impact (RR 0.73, 95% CI 0.66 to 0.82, p = 0.36 for heterogeneity, NNT = 11 to prevent 1 patient from being hospitalized for HF) on this outcome (Table 2). Results were similar for specialized follow-up by a multidisciplinary team (RR 0.74, 95% CI 0.63 to 0.87) either in an HF clinic (RR 0.76, 95% CI 0.58 to 0.099) or in a non-clinic setting (RR 0.72, 95% CI 0.59 to 0.87), programs employing telephone follow-up (RR 0.75, 95% CI 0.57 to 0.99), or programs emphasizing enhanced patient self-care activities (RR 0.66, 95% CI 0.52 to 0.83). Given the similarity in the summary estimates, adjusted indirect comparisons were not performed.
Total number of hospitalizations
Twenty-one trials reported total number of hospitalizations by treatment arm; in 11 of these trials, the intervention arm had significantly fewer total hospitalizations. Pooling the data from all 21 trials revealed a marked reduction in total number of hospitalizations (RR 0.70, 95% CI 0.62 to 0.80), with no appreciable differences between multidisciplinary HF clinics (RR 0.61, 95% CI 0.39 to 0.94), specialized follow-up by multidisciplinary team (RR 0.68, 95% CI 0.56 to 0.83), telephone follow-up (RR 0.77, 95% CI 0.63 to 0.95), or enhanced patient education and self-care activities (RR 0.68, 95% CI 0.54 to 0.87).
Total HF hospitalizations
The frequency of multiple re-admissions for HF was particularly affected by the multidisciplinary strategies, such that total HF hospitalizations were markedly reduced (20 trials, RR 0.57, 95% CI 0.49 to 0.67). There were no appreciable differences between types of interventions: RR 0.67 (95% CI 0.53 to 0.83) for multidisciplinary HF clinics, RR 0.47 (95% CI 0.36 to 0.61) for specialized follow-up by multidisciplinary team, RR 0.65 (95% CI 0.49 to 0.85) for telephone follow-up, and RR 0.60 (95% CI 0.41 to 0.88) for enhanced patient education and self-care activities.
Other end points
Six (9,16,18,20,29,37) of the 10 trials (10,13,14,27) that assessed the use of proven efficacious medications demonstrated significantly higher prescribing rates (or dosing) in those patients randomized to the multidisciplinary strategies. Five (18,23,24,31,35) of the six (36) trials that examined patient compliance reported significantly higher adherence rates to medications and dietary/fluid restrictions in patients exposed to the multidisciplinary interventions. Nine (9,14,18,19,23,25,32,33,37) of the 18 trials (10,12,16,17,28,30,31) that assessed patient quality of life or functional status demonstrated significantly better outcomes in the intervention arms (another two trials [13,26]reported a nonsignificant trend toward better quality of life scores). Finally, although none of these trials included formal cost-effectiveness analyses, 15 (9,11–13,15,16,19,22,23,25,28,33,34,36,37) of the 18 trials describing the costs of the intervention reported that it was cost-saving (the other three trials [18,26,35]reported their interventions to be cost-neutral).
Analyses did not reveal any significant effects of year of publication, duration of intervention, Jadad score, sample size, or length of follow-up on our results.
In summary, pooling the data from the 29 randomized trials of multidisciplinary management strategies for patients with HF reveals that these programs are associated with a 27% reduction in HF hospitalization rates (NNT = 11) and a 43% reduction in total number of HF hospitalizations. Those strategies that incorporate specialized follow-up by a multidisciplinary team or in a multidisciplinary HF clinic also reduce all-cause mortality by approximately one-quarter (NNT = 17) and all-cause hospitalizations by one-fifth (NNT = 10). The beneficial effects of these multidisciplinary strategies compare favorably with established drug treatments for HF (for example, angiotensin-converting enzyme inhibitors reduce mortality by 20% [NNT = 19] and HF hospitalization rates by 33% [NNT = 16] in similar patient populations) (39) and suggest that the current enthusiasm for these programs is not misplaced. Further, although none of these trials incorporated formal cost-effectiveness analyses, 15 of the 18 trials reported that their multidisciplinary interventions were cost-saving (the other three trials reported cost neutrality), and recent analyses in the U.K. and U.S. have shown that when applied on a national basis, multidisciplinary management strategies have the potential to deliver substantial cost savings (6,40).
In our earlier systematic review (5), we were unable to draw any conclusions about which elements of these multidisciplinary strategies were most efficacious in patients with HF. Incorporating the trials published over the past five years, enhanced patient self-care, follow-up monitoring by specially trained staff, and access to specialized HF clinics appear to be the most efficacious approaches. Although telephone follow-up programs that advise deteriorating patients to see their regular physician do reduce HF hospitalizations, they have no impact on mortality or all-cause hospitalizations. Of note, while this conclusion is based on the 10 trials (1,897 patients) of telephone support programs published thus far, it is also consistent with the preliminary report from the DIALTrial Investigators (1,518 patients from 51 centers) presented at the 2002 American Heart Association Meeting (41).
Whereas the specific elements to be included in any multidisciplinary disease management program should take into account those features unique to each health region (including prevalence, local barriers to optimal care, and local resources), we believe that three elements are crucial to the success of these programs. First, specially trained HF nurses should be key components of any intervention. Second, efforts should be made to educate patients and their caregivers about HF, precipitating factors, and the need for compliance with medication and dietary advice (particularly because most HF re-admissions are caused by factors that patients can be taught to recognize and avoid) (42). Finally, there must be ready access to clinicians trained in HF.
There are a few important caveats with this systematic review. First and foremost, all but two of these trials tested programs in high-risk HF patients (i.e., those who had been recently discharged from hospital), and all programs were administered by specially trained staff from academic health centers in urban areas. Whether the same benefits could be obtained in lower risk patients or if the programs were delivered by staff without special training is uncertain, but seems unlikely. Second, these trials were relatively short in duration, raising questions about their long-term effectiveness. However, the median survival of patients with HF after hospital discharge is only 18 to 24 months (3,43). Thus, interventions that demonstrate substantial benefits so quickly in HF are particularly welcome. In addition, preliminary evidence suggests that the benefits from at least one of these programs were maintained over a further five years of follow-up (44). Although concerns may be raised that attendance at a specialized HF clinic may improve the provision of effective HF therapies at the expense of other needed medications (45), we chose all-cause mortality as our primary outcome and all-cause hospitalizations as one of our secondary outcomes for this very reason. Further, our evaluation of the incremental benefits of the various components of each intervention is hampered by the lack of direct comparisons. For example, a recent trial reported that daily nurse-supervised telemonitoring of HF patients was more efficacious and less costly than home visits by nurses who were not specially trained in HF (46). However, adjusted indirect comparisons such as those we present in this systematic review have been shown to provide reasonable estimates of the relative efficacy of different interventions when the control arms are similar (8). Finally, few of these trials reported on prescribing rates for angiotensin-converting enzyme inhibitors or beta-blockers, preventing us from examining whether the benefits of these programs are merely attributable to improved prescribing patterns or patient compliance, or whether the benefits are additive to optimal medical therapy.
In conclusion, a wide variety of multidisciplinary strategies to manage patients with HF reduce HF hospitalizations. Strategies that incorporate follow-up monitoring by specially trained staff and/or access to specialized HF clinics also reduce mortality and all-cause hospitalizations. Indeed, the benefits and cost-effectiveness of these programs compare favorably with established drug treatments for HF. The question is no longer whether these programs work, but rather which of these programs work best. Our systematic review suggests that patient education to enhance self-care, follow-up monitoring by specially trained staff, and access to specialized HF clinics are the most efficacious approaches. These conclusions are based on adjusted indirect comparisons, and direct comparisons of different types of interventions (and/or different intensities) in head-to-head trials should be the next frontier for research in this field. The challenge for policymakers is to incorporate this evidence into future health care planning to deal with the current epidemic of HF patients being discharged from acute care institutions throughout North America.
The authors thank the following study investigators who kindly provided further data from their studies: Drs. Jaarsma, Krumholz, Laramee, Harrison, Jerant, Riegel, Stromberg, Stewart (two studies), Naylor (two studies), Gattis, Trochu, and Tsuyuki.
Dr. McAlister is a Population Health Investigator of the Alberta Heritage Foundation for Medical Research and a New Investigator of the Canadian Institutes of Health Research; Dr. Stewart is supported by the National Health and Medical Research Council of Australia and the National Heart Foundation of Australia.
- Abbreviations and acronyms
- confidence interval
- heart failure
- number needed to treat
- risk ratio
- Received January 20, 2004.
- Revision received April 8, 2004.
- Accepted May 11, 2004.
- American College of Cardiology Foundation
- MacIntyre K.,
- Capewell S.,
- Stewart S,
- et al.
- Song F.,
- Altman D.G.,
- Glenny A.M.,
- Deeks J.J.
- Doughty R.N.,
- Wright S.P.,
- Pearl A,
- et al.
- Cline C.M.J.,
- Israelsson B.Y.A.,
- Willenheimer R.B.,
- Broms K.,
- Erhardt L.R.
- Ekman I.,
- Andersson B.,
- Ehnfors M.,
- Matejka G.,
- Persson B.,
- Fagerberg B.
- Capomolla S.,
- Febo O.,
- Ceresa M,
- et al.
- Stromberg A.,
- Martensson J.,
- Fridlund B,
- et al.
- Hanchett E.,
- Torrens P.R.
- Serxner S.,
- Miyaji M.,
- Jeffords J.
- Jaarsma T.,
- Halfens R.,
- Huijer Abu-Saad H,
- et al.
- Pugh L.C.,
- Havens D.S.,
- Xie S.,
- Robinson J.M.,
- Blaha C.
- Blue L.,
- Lang E.,
- McMurray J.J.V,
- et al.
- Krumholz H.M.,
- Amatruda J.,
- Smith G.L,
- et al.
- ↵Tsuyuki RT, Fradette M, Johnson JA, et al. Multicentre disease management program for hospitalized patients with heart failure: the Review of Education on ACE inhibitors in Congestive heart failure Treatment (REACT) Study. J Card Fail 2004. In Press.
- Trochu JN, Baleynaud S, Mialet G, et al. Efficacy of a multidisciplinary management of chronic heart failure patients: one year results of a multicentre randomized trial in French medical practice. Eur Heart J 2004. In Press.
- Stewart S.,
- Blue L.,
- Walker A.,
- Morrison C.,
- McMurray J.J.
- Stewart S.,
- Horowitz J.D.