Author + information
- Received March 19, 2012
- Revision received May 29, 2012
- Accepted June 5, 2012
- Published online October 2, 2012.
- Simon Stewart, PhD⁎,⁎ (, )
- Melinda J. Carrington, PhD⁎,
- Thomas H. Marwick, MD, PhD†,
- Patricia M. Davidson, PhD‡,
- Peter Macdonald, PhD§,
- John D. Horowitz, PhD∥,
- Henry Krum, PhD¶,
- Phillip J. Newton, PhD‡,
- Christopher Reid, PhD¶,
- Yih Kai Chan, PhD⁎ and
- Paul A. Scuffham, PhD#
- ↵⁎Reprint requests and correspondence:
Dr. Simon Stewart, Preventative Health, Baker IDI Heart and Diabetes Institute, P.O. Box 6492, St. Kilda Road Central, Melbourne, Victoria 8008, Australia
Objectives The goal of this study was to make a head-to-head comparison of 2 common forms of multidisciplinary chronic heart failure (CHF) management.
Background Although direct patient contact appears to be best in delivering CHF management overall, the precise form to optimize health outcomes is less clear.
Methods This prospective, multicenter randomized controlled trial with blinded endpoint adjudication comprised 280 hospitalized CHF patients (73% male, age 71 ± 14 years, and 73% with left ventricular ejection fraction ≤45%) randomized to home-based intervention (HBI) or specialized CHF clinic–based intervention (CBI). The primary endpoint was all-cause, unplanned hospitalization or death during 12- to 18-month follow-up. Secondary endpoints included type/duration of hospitalization and healthcare costs.
Results The primary endpoint occurred in 102 of 143 (71%) HBI versus 104 of 137 (76%) CBI patients (adjusted hazard ratio [HR]: 0.97 [95% confidence interval (CI): 0.73 to 1.30], p = 0.861): 96 (67.1%) HBI versus 95 (69.3%) CBI patients had an unplanned hospitalization (p = 0.887), and 31 (21.7%) versus 38 (27.7%) died (p = 0.252). The median duration of each unplanned hospitalization was significantly less in the HBI group (4.0 [interquartile range (IQR): 2.0 to 7.0] days vs. 6.0 [IQR: 3.5 to 13] days; p = 0.004). Overall, 75% of all hospitalization was attributable to 64 (22.9%) patients, of whom 43 (67%) were CBI patients (adjusted odds ratio: 2.55 [95% CI: 1.37 to 4.73], p = 0.003). HBI was associated with significantly fewer days of all-cause hospitalization (−35%; p = 0.003) and from cardiovascular causes (−37%; p = 0.025) but not for CHF (−24%; p = 0.218). Consequently, healthcare costs ($AU3.93 vs. $AU5.53 million) were significantly less for the HBI group (median: $AU34 [IQR: 13 to 81] per day vs. $AU52 [17 to 140] per day; p = 0.030).
Conclusions HBI was not superior to CBI in reducing all-cause death or hospitalization. However, HBI was associated with significantly lower healthcare costs, attributable to fewer days of hospitalization. (Which Heart failure Intervention is most Cost-effective & consumer friendly in reducing Hospital care [WHICH?]; ACTRN12607000069459)
The burden of chronic heart failure (CHF), characterized by costly hospitalization and poor survival, remains high (1,2). Randomized trials in mid to late 1990s (3–6), confirmed by meta-analyses (7), support the role of multidisciplinary CHF management programs (CHF-MPs) to apply gold-standard strategies, reduce recurrent hospital stay, and prolong survival (7). These programs, built on the patient-centered traditions of the chronic care model developed by Wagner and colleagues (8), can best be described as a system of coordinated healthcare interventions specifically designed to optimize the management of individuals with CHF and where patient self-care plays a vital role (9). As the evidence evolves, 2 issues have prompted substantial debate: the need for direct (as against remote) patient monitoring (10) and the relationship between interventional components and patient outcomes (11). Although strong arguments support direct patient contact (4), the precise form of contact associated with optimal effects is less clear. The WHICH? (Which Heart Failure Intervention Is Most Cost-Effective & Consumer Friendly in Reducing Hospital Care) trial investigators envisaged there might be important differences in the impact and acceptability (from a patient perspective) of the 2 most popular forms of CHF-MP (12). Both apply the same components of multidisciplinary care but via 2 very different settings: the first in the patient's own home and the second via a specialist outpatient clinic. We hypothesized that in typically older patients with CHF in whom there are multiple factors contributing to poor outcomes, a home-based approach would be more effective in optimizing health outcomes due to a better overall understanding of the patient and their environment. The WHICH? trial, therefore, tested the hypothesis that compared to an equivalent clinic-based program of management, a nurse-led, post-discharge, multidisciplinary management program for CHF patients age ≥45 years involving post-discharge home visits, will be superior in reducing the composite primary endpoint of unplanned (all-cause) readmission or death during 12 to 18 months follow-up. We further hypothesized that a home-based approach would also be more cost effective in reducing days of hospitalization (12).
A detailed description of the rationale and design of the WHICH? trial, a multicenter randomized controlled study, has been published previously (12). It conformed to the principles outlined in the Declaration of Helsinki and the CONSORT (Consolidated Standards for Reporting Trials) guidelines for pragmatic studies (13). All participants provided written informed consent. Undertaken in 3 Australian tertiary referral hospitals, all centers had pre-existing CHF-MPs modified for the purpose of the trial and guided by cardiologists experienced in CHF management (T.H.M., P.M., and J.D.H.).
A blinded endpoint committee (S.S., P.M., and C.R.) adjudicated on the type (elective vs. unplanned) and cause of all hospitalizations. The primary endpoint was unplanned hospitalization or mortality (both all-cause) during 12 to 18 months of follow-up (study commenced June 2008, with census of outcome data March 31, 2011). As pre-specified in the index report (12), these composite events were also examined separately as: 1) a proportion of affected individuals; and 2) in the case of hospitalization, rate of hospitalization and duration of hospitalization. We also examined event-free survival as expressed as “days alive out of hospital.” Pharmacological therapy (including prescribed doses of angiotensin-converting enzyme [ACE] inhibitors and beta blockers) at baseline and 12-month follow-up were also examined. Similarly, health-related quality-of-life data were collected using generic (EQ-5D ) and CHF-specific instruments (Minnesota Living With Heart Failure Questionnaire [MLWHFQ] ) at 12 months to examine changes from baseline (collected during the index admission). As also pre-specified, we prospectively collected health utilization data to perform an economic analysis of healthcare costs.
All patients admitted to participating centers were screened for study eligibility according to the following criteria: 1) age ≥18 years; 2) discharged to home with a diagnosis of CHF as confirmed by a cardiologist; 3) persistent moderate-to-severe symptoms (New York Heart Association [NYHA] functional class II to IV); and 4) a recent history of ≥1 admissions for acute heart failure. Those excluded lived outside a 30-km radius of the hospital, had a terminal condition, were non-English speaking, and/or were unable to provide informed consent. As shown in Figure 1 (study CONSORT flow chart ), of a total of 2,832 cardiac inpatients screened for eligibility, 298 (11%) were randomized into the study. However, 18 patients died or immediately withdrew consent from the study when allocated to a particular intervention before discharge from the index hospitalization and were excluded from subsequent analyses.
Block randomization (computer-based at independent site: Baker IDI) for each center, with stratification for the presence/absence of left ventricular systolic dysfunction, as determined by a left ventricular ejection fraction (LVEF) of ≤45%, was applied on a 1:1 basis to group assign patients.
Detailed demographic and clinical data were collected at baseline (see Table 1 for indicative profiling in a standardized manner by trained personnel). All surviving patients were subject to clinical follow-up at 6 months (brief telephone call), 12 months, and a final follow-up for up to 18 months (pre-scheduled home or clinic visit).
The key components and principles of post-discharge management of CHF, either delivered as an outreach, home-based intervention (HBI) or via a clinic-based intervention (CBI) coordinated via a specialist CHF outpatient clinic, according to contemporary guidelines, have been extensively described (12). The Australian healthcare system provides universal health care for the population, with only minimal costs (capped for those with chronic disease) for hospital treatment, pharmacotherapy, and community care (including family physicians). The study was designed to standardize the quality of management (often supported by the same cardiologists and family care physicians) but to vary how it was delivered, with explicit acknowledgement that patient interaction with the healthcare team would be modulated according to the mode of management.
Briefly, HBI patients were scheduled to receive a home visit by a trained CHF nurse within 7 to 14 days of hospital discharge. This comprised a structured and detailed assessment of the patient's clinical stability, application of gold-standard pharmacological and nonpharmacological management, and any factors likely to positively or negatively impact future health outcomes. Specific activities incorporated into this visit included: 1) a detailed clinical assessment that included assessment of their cognitive status; 2) comprehensive assessment of the patient's home environment; 3) counseling of family members/care givers in the home; 4) detailed assessment of the patient's social support and coping skills; 5) review of the patient's use of current and past medication; 6) assessment of the patient's food supply (including supplementary salt intake) and fluid consumption; 7) identification of key equipment (i.e., weight scales); 8) determination of the patient's level of mobility and ease of access to local health services; 9) referral to a community pharmacist for a comprehensive “home medicines review” (16); and 10) close liaison with the patient's family physician. Subsequently, a report was sent to the patient's family physician and cardiologist, and planned management (including telephone follow-up, referral to other healthcare professionals, and additional home visits) was arranged. Regardless of initial assessment, those discharged to home following an unplanned hospitalization were subject to re-evaluation of the relative success/failure of management by the CHF nurse. Similarly, CBI patients were scheduled to attend a post-discharge visit to the nurse-led specialist CHF clinic where they had immediate access to a multidisciplinary team including an experienced cardiologist and pharmacist. The same principles of assessment and follow-up as per HBI were applied. The key differences being that for the CBI group, management was primarily directed through the specialist CHF clinic on an outpatient basis, and the patient did not receive a comprehensive home visit (see the preceding text). No restrictions on access to other healthcare services were applied.
We estimated this study would have 80% power (2-sided alpha of 0.05) to detect a 15% absolute difference in the primary endpoint in addition to a 15% variation in the rate of all-cause hospital stay (in days) with 280 randomized patients; these thresholds were considered to be of clinical significance and would generate statistical differences in healthcare costs. All baseline and outcome data were analyzed using SPSS for Windows version 19.0 (SPSS, Inc., Chicago, Illinois) on an intention-to-treat basis. Continuous data are presented as a mean ± SD or median plus interquartile range (IQR). Categorical data are presented as a percentage. Univariate comparisons of baseline data involved chi-square analyses (with calculation of odds ratios [ORs] and 95% confidence intervals [CIs]) for categorical data, the Mann-Whitney U test for non-normally distributed continuous data (including the rate of hospital stay and healthcare costs), and the Student t test for normally distributed continuous data. All-cause mortality and event-free survival data were initially analyzed using Kaplan-Meier survival curves. Days alive out of hospital were calculated as days of survival free from unplanned and all-cause hospitalization. Healthcare costs are calculated per patient per day. All costs are expressed in 2009/2010 Australian dollars (AU$1.00 ≈ US$1.00). To initially examine the independent impact of group allocation on: 1) presence in the upper quartile group for most days of re-hospitalization; 2) event-free survival; and 3) all-cause mortality, entry model multiple logistic regression and Cox proportional hazards models were constructed. All models included the variables listed in Table 1 (including age, sex, cardiac function, and comorbidity) in addition to site of recruitment (17). Backward, step-wise models (entry at univariate level of p < 0.1 and retention of variables at adjusted level p < 0.05) were then performed to determine the independent correlates of these outcomes. Detailed methods of the health economics analysis (including components of health expenditure—see Online Table 1) are provided in the Online Appendix.
Table 1 shows the baseline demographic and clinical profile of this typically older study cohort (N = 280) with high levels of concurrent disease. There were more men (73%) than women. On average, men were younger (71 ± 13 years vs. 73 ± 14 years) and had more left ventricular systolic dysfunction (LVEF: 34.4 ± 13.4% vs. 42.9 ± 15.8%; p < 0.001). Overall, 254 (91%) patients were prescribed an ACE inhibitor/angiotensin receptor blocker or a beta-blocker, and most were also prescribed a loop diuretic. Despite gold-standard therapy, 15% were in NYHA functional class IV. Overall, the groups were fairly well matched for all baseline parameters, with the notable exception that HBI patients were, on average, 3 years younger (p = 0.046) and tended to have less comorbidity as measured by the Charlson Index of Comorbidity.
More HBI patients (140 [98%]) received a home visit than CBI patients (124 [91%]) who attended the specialist CHF clinic (OR: 1.08, 95% CI: 1.02 to 1.15; p = 0.008). HBI patients also tended to have more home visits than CBI patient appointments to the CHF clinic (3.4 ± 2.5 visits vs. 2.9 ± 2.5 visits; p = 0.061), although CBI patients had, on average, 1 more cardiologic outpatient visit overall. At the first visit or appointment, most patients were assessed as NYHA functional class II or III (218 [83%]), 180 (68%) had orthopnea, 76 (29%) peripheral edema, 45 (17%) a raised jugular venous pressure, and 32 (12%) had basal crackles. As expected, subsequent referrals to external multidisciplinary support were significantly greater for HBI than CBI patients (p < 0.05 for all comparisons), including referrals to a community-based pharmacist (39% vs. 18%), an exercise program (39% vs. 17%), and a dietician (16% vs. 10%). The same proportions of patients (11%) were referred for urgent medical review to address clinical instability. Similarly, 68% and 71% of HBI and CBI patients had planned telephone follow-up. Follow-up home visits resulted in 10% of individuals requiring immediate or emergency medical management due to clinical instability. The equivalent proportion in the CBI group (following a follow-up clinic visit) was 14%. Overall, both groups visited their family physician a similar number of times (average 21 to 23 visits).
Subsequent telephone contact occurred in 122 (85%) and 113 (83%) of HBI and CBI patients, respectively; 43% received 1 to 2 calls, and the remainder up to 19 calls. Of the 1,063 telephone calls for patient management, 584 (55%) were routine clinical follow-up, 283 (27%) involved active management (e.g., titrating pharmacotherapy), 111 (10%) were appointment reminders, and 85 (8%) were patient initiated. Overall, HBI patients received more calls (4.2 ± 3.8 vs. 3.3 ± 2.9; p = 0.002) and were exposed to more telephone management (39.9 ± 41.0 min vs. 27.1 ± 31.8 min; p = 0.01).
A total of 96 HBI (67%) patients and 85 CBI (62%) patients were clinically assessed between 12 and 18 months. More HBI patients received gold-standard therapy, but in lower doses. For example, more HBI patients were prescribed an ACE inhibitor and/or angiotensin receptor blocker (96% vs. 81%; p = 0.002, OR: 1.18, 95% CI: 1.06 to 1.32), but CBI patients who were prescribed perindopril (the most commonly prescribed ACE inhibitor) received higher daily doses (8.1 ± 3.0 mg vs. 5.2 ± 3.1 mg; p = 0.001).
During study follow-up (mean of 17.4 ± 1.4 months for surviving patients and 14.7 ± 5.3 months overall), unplanned hospitalization or death occurred in 102 of 143 HBI patients (71%) compared with 104 of 137 CBI patients (76%) patients (adjusted hazard ratio: 0.97, 95% CI: 0.73 to 1.30; p = 0.861). Online Figure 1 shows the adjusted event-free survival curves for both groups.
At 12 months, 56 patients (20%) died (comprising 18% of HBI and 22% of CBI patients). Figure 2 compares the adjusted survival curves for the 2 groups on the basis of 31 (22%) versus 38 (28%) deaths in the HBI and CBI groups, respectively, during total follow-up (adjusted hazard of risk: 0.75, 95% CI: 0.45 to 1.23; p = 0.252). Of all 69 deaths, 7 (10%) were sudden cardiac events.
The study cohort accumulated 474 and 615 unplanned and all-cause readmissions. This included 323 unplanned readmissions (68% of that category) primarily related to cardiovascular disease (including CHF). Overall, 96 (67%) HBI patients compared with 95 (69%) CBI patients had ≥1 unplanned hospitalizations, with 60 (62%) HBI versus 52 (55%) CBI patients with multiple hospitalizations (p = 0.194). The rate (per 100 days of follow-up/patient) of unplanned and total hospitalization was similar in the HBI group compared with the CBI group, respectively. This trend was also seen with respect to all cardiovascular and CHF-specific hospitalizations (Fig. 3).
Days of hospitalization
A total of 3,744 unplanned and 4,430 days of recurrent unplanned hospitalization were accumulated during study follow-up. Average length of hospital stay for all-cause, unplanned hospitalization was significantly lower (p = 0.004) in the HBI group (median: 4.0, IQR: 2.0 to 7.0 days vs. 6.0, IQR: 3.5 to 13 days). A similar, but nonsignificant, trend (p = 0.674) was observed in relation to elective hospitalization (median: 2.0, IQR: 1.0 to 5.0 vs. 6.0, IQR: 1.0 to 14). Figure 3 compares the rate of hospitalization and days of hospitalization for the 2 groups. As such, after adjusting for survival and follow-up, HBI patients accumulated fewer days of unplanned (668 days [31%] less; p = 0.059) and total hospitalization (938 days [34%] less; p = 0.003) than CBI patients. Group differences extended to cardiovascular-related days of hospitalization (2,628 days representing 59% of all hospitalization) and those directly attributable to CHF (1,845 days representing 70% of cardiovascular-related hospitalization). Overall, HBI patients accumulated significantly fewer days of cardiovascular-related hospitalization (588 days [37%] less; p = 0.025) but not CHF-related hospitalization (257 days [24%] less; p = 0.218).
Post-hoc analyses showed that the difference between groups in days of hospitalization was largely attributable to the 64 patients (22.9%) who accumulated ≥25 days of hospitalization during study follow-up (the upper quartile threshold for accumulated days of hospitalization) and who accounted for 75% of such activity (Online Fig. 2). Overall, 43 of these 64 patients (67%) belonged to the CBI group. They were significantly older (77 ± 10 years vs. 70 ± 14 years; p < 0.001), had more preserved left ventricular function (LVEF: 40 ± 13% vs. 35 ± 15%; p = 0.08), and had a longer index admission (11.1 ± 8.7 days vs. 8.1 ± 7.4 days; p = 0.03) than the rest. On an adjusted basis, the only independent predictors of prolonged hospital stay was advancing age (OR: 1.06, 95% CI: 1.03 to 1.09 per year; p < 0.001) and allocation to CBI (OR: 2.55, 95% CI: 1.37 to 4.73; p = 0.003).
Days alive out of hospital
There was a significant difference between groups regarding actual days out of hospital (from unplanned hospitalization) alive (452 ± 158 days vs. 418 ± 173 days; p = 0.019) and all days out of hospital (elective plus unplanned hospitalization) alive (451 ± 158 days vs. 414 ± 172 days; p = 0.009) in favor of HBI. The latter equated to 85% (11,869 days lost) versus 80% (14,248 days lost) of maximal survival without hospitalization.
Health-related quality of life
Baseline EQ-5D (0.76 ± 0.18 vs. 0.77 ± 0.18; p = 0.921) and MLWHFQ scores (48.8 ± 21.9 vs. 46.0 ± 20.5; p = 0.279) were similar for HBI and CBI patients; best and worst health states being 1.0 and 0.0, respectively. Reflecting similar trends at 6 months, among survivors with repeat scores at 12 months (87 HBI and 84 CBI patients), the mean changes in EQ-5D scores were not significantly different (−0.136 ± 0.363 vs. −0.183 ± 0.350, respectively; p = 0.305). Alternatively, MLWHFQ scores demonstrated improved quality of life in both HBI and CBI patients (median change: −10, IQR: −30 to 1 vs. median change: −12, IQR: −31 to 11, respectively; p = 0.445). As there were no differences between groups in EQ-5D scores or quality-adjusted life-years, a cost-minimization analysis was appropriate.
Table 2 summarizes the components of healthcare costs per day for the 2 groups (AU$1.00 ≈ US$1.00). Hospital costs accounted for 89% and 93% of total costs in the HBI and CBI groups, respectively, whereas the cost of applying the specific HBI and CBI programs per patient were similar AU$1,813 ± AU$220 and AU$1,829 ± AU$174 (representing 9.0% and 4.2% of total healthcare costs). Overall, predominantly due to fewer days of hospitalization, total healthcare costs were around 30% lower in the HBI group (AU$3.93 million vs. AU$5.53 million; p = 0.03 for median costs per day).
To our knowledge, this represents the first direct comparison of an outreach versus specialist clinic–based approach to post-discharge CHF management. We compared these commonly applied forms of CHF-MP in order to determine whether there were additional benefits derived from an outreach CHF-MP in typically old and fragile patients discharged to home following acute hospitalization. Overall, there was minimal difference with respect to the primary endpoint of (all-cause) death or unplanned readmission during 12 to 18 months of follow-up. Although, HBI had a better survival profile (6% absolute difference), this did not reach statistical significance. However, there were important differences with respect to the cost impact of HBI. Reflecting shorter episodes of hospitalization (the most costly component of CHF management ), HBI patients accumulated significantly fewer days of total all-cause (35% days fewer) and cardiovascular-related hospitalization (37% days fewer), with a consistent (borderline) trend with respect to less unplanned hospitalization (30% days fewer). HBI was also associated with a nonsignificant reduction in CHF stay. Combined, this resulted in significantly more prolonged days out of hospital alive in favor of HBI. Consequently, total healthcare costs were nearly one-third less within the HBI group. The combination of greater uptake of HBI overall, a more favorable profile free from hospitalization and death, and more favorable cost dynamics (largely mediated through a greater reduction in non-CHF–related hospitalization) are of clinical and public health significance. Although the primary endpoint was not statistically different, therefore, these data now provide a potentially important delineation between the 2 most utilized forms of face-to-face CHF-MP.
A number of aspects about the validity and wider applicability of the trial require consideration. First, similar to many contemporary CHF-MP trials, the standard of gold-standard pharmacotherapy was high. This is perhaps reflected in the relatively low mortality rates in high-risk patients (with high comorbidity) countered by high levels of morbidity (40% had multiple unplanned readmissions) and recurrent hospital stay. The WHICH? trial cohort is similar in age and sex profile to the COACH (Coordinating Study Evaluating Outcomes of Advising and Counseling in Heart Failure) study reported by Jaarsma et al. (19), but with higher levels of concurrent coronary artery disease, atrial fibrillation, and stroke. Although a similar proportion of patients were re-admitted for CHF (around 40%), our cohort had greater levels of recurrent, all-cause hospital stay. Despite this, case fatality was comparable, if not lower, than that reported by the COACH investigators (19) and in a recent trial of a brain natriuretic peptide–guided CHF-MP versus a standard CHF-MP and usual care (20). Nevertheless, case fatality and morbidity rates remain high, and efforts to enhance the efficacy of CHF-MPs, either through biomarker-guided therapy (20) or supplementary remote monitoring techniques, (21) remain a priority. The Australian healthcare system (while reflecting a hybrid version of universal healthcare and a user-pay system) provided a strong platform to compare 2 different forms of CHF-MP. With the exception of access to the CHF clinic (CBI group) and home visits (HBI), patients in both groups had access to the same healthcare professionals (including cardiologists and family physicians). This environment enabled us to standardize the level and qualities of care provided, and observe critical differences in patient journeys according to the mode of CHF-MP applied. Despite potentially important baseline differences in patient profile (particularly more advanced age and comorbidity in the CBI group), we were able to demonstrate that the critical difference between groups with respect to “high-cost” users of health care (two-thirds of whom were CBI patients) was independently influenced by group allocation.
Potential benefits of HBI
It is important to emphasize that the primary endpoint of unplanned hospitalization or death was not statistically different between groups and that this was, perhaps, an optimistic composite endpoint in that it relied on delaying and even preventing events. On an adjusted basis, there was little difference between the 2 forms of CHF-MP, most probably reflecting high standards of initial post-discharge management. The mechanism(s) underlying the apparent longer-term benefits of HBI with respect to a significant reduction in “high-cost” patients with more prolonged hospitalization, as always, are difficult to elucidate. Consistent with the overall benefits associated with the application of HBI in chronic disease management (including reduced healthcare costs ), there are a number of factors that may explain our findings. First, as indicated by the frequency and duration of patient contacts, the HBI group appeared to be more engaged and more likely, therefore, to benefit from the expert management on offer. This resulted in a greater number of referrals. By its very nature (i.e., visiting patients and significant others in their own homes), HBI has the potential to provide a more accurate assessment of a patient's overall clinical and psychosocial status and, critically, their ability to self-care. This in turn would facilitate more specific and tailored management (including pharmacotherapy) for a cohort of patients with complex needs, and therefore, lead to greater clinical stability (as reflected in the fewer days of hospitalization). The more “generic” approach to chronic disease management, as applied by the multidisciplinary HBI team, would be expected to have a broader impact on non-CHF–related health outcomes. Observed differences in all-cause and cardiovascular-related hospital stay in favor of HBI support this supposition. It is certainly feasible that patients simply prefer and respond better to a health intervention that appears to be more flexible to their needs: our prospectively planned analyses of health delivery preferences in this cohort (22) will further elucidate potential differences from a consumer perspective. Finally, we cannot discount the possibility that HBI patients were discharged earlier on the premise that extended hospital days could be replaced by increased home surveillance (despite not being a “hospital in the home” program).
As discussed in our methods report (12), there are a number of limitations, including the confounding effect of patient (and potentially healthcare team) bias in favor of a particular CHF-MP. Wherever possible, we minimized contamination between groups, and all endpoints were subject to blinded adjudication. Type II error cannot be ruled out with respect to our ability to examine a definitive difference with respect to all-cause mortality. As discussed, it was impractical to fully explore mechanisms of effect, and we cannot fully discount the possibility that observed differences were due to important differences in the quality of care, nor can we fully discount the influence of observed baseline differences in the age and clinical profile of the 2 groups. Moreover, it is important to consider the potential cost dynamics of more prolonged survival and exposed risk to greater morbidity in survivors (23).
In this prospective, multicenter, randomized trial of multidisciplinary CHF management delivered via an HBI versus CBI approach, we found no difference in unplanned (all-cause) hospitalization or death during 12- to 18-month follow-up. However, HBI was associated with a significant reduction in the duration of recurrent hospitalization and more prolonged survival free from hospitalization that resulted in an important delineation between the 2 most commonly applied forms of CHF-MP. Ultimately, this contributed to an approximate one-third reduction in healthcare costs in favor of the outreach HBI program. The fact that we observed the greatest differences with respect to all-cause and cardiovascular-related hospitalization in this typically older cohort of CHF patients suggests that the benefits of a more generic hospital transition program involving home visits outweigh a more CHF-specific focus.
The authors thank all the cardiac nurses, healthcare professionals, and patients who participated; and Alicia Calderone and the staff at Baker IDI who contributed to data management.
For detailed methods and supplemental table and figures, please see the online version of this article.
The WHICH? trial (number 418967) and Drs. Stewart, Carrington, Reid, and Scuffham are supported by the National Health and Medical Research Council of Australia. The authors have stated that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- chronic heart failure clinic–based intervention
- chronic heart failure
- chronic heart failure management programs
- confidence interval
- home-based intervention
- hazard ratio
- interquartile range
- left ventricular ejection fraction
- New York Heart Association
- odds ratio
- Received March 19, 2012.
- Revision received May 29, 2012.
- Accepted June 5, 2012.
- American College of Cardiology Foundation
- Stewart S.,
- Ekman I.,
- Ekman T.,
- Oden A.,
- Rosengren A.
- Cline C.M.,
- Israelsson B.Y.,
- Willenheimer R.B.,
- Broms K.,
- Erhardt L.R.
- Blue L.,
- Lang E.,
- McMurray J.J.,
- et al.
- McAlister F.A.,
- Stewart S.,
- Ferrua S.,
- McMurray J.J.
- Coleman K.,
- Austin B.T.,
- Brach C.,
- Wagner E.H.
- Inglis S.C.,
- Clark R.A.,
- McAlister F.A.,
- Stewart S.,
- Cleland J.G.F.
- Clark A.M.,
- Savard L.A.,
- Thompson D.R.
- Zwarenstein M.,
- Treweek S.,
- Gagnier J.J.,
- et al.
- Stewart S.,
- Blue L.,
- Walker A.,
- Morrison C.,
- McMurray J.J.
- Jaarsma T.,
- van der Wal M.H.,
- Lesman-Leegte I.,
- et al.
- Berger R.,
- Moertl D.,
- Peter S.,
- et al.
- Gaikwad R.,
- Warren J.
- Inglis S.C.,
- Pearson S.,
- Treen S.,
- Gallasch T.,
- Horowitz J.D.,
- Stewart S.