Author + information
- Received October 29, 2009
- Revision received March 5, 2010
- Accepted March 19, 2010
- Published online July 27, 2010.
- Daniel Ferrante, MD, MSc⁎ (, )
- Sergio Varini, MD,
- Alejandro Macchia, MD,
- Saúl Soifer, MD,
- Raul Badra, MD,
- Daniel Nul, MD,
- Hugo Grancelli, MD,
- Hernán Doval, MD,
- GESICA Investigators
- ↵⁎Reprint requests and correspondence:
Dr. Daniel Ferrante, GESICA Foundation, Avenida Rivadavia 2358, 1 Piso, Departamento 4, Buenos Aires 1034, Argentina
Objectives The purpose of this study was to assess the rate of death and hospitalization for heart failure (HF) 1 and 3 years after a randomized trial of telephone intervention aimed to improve education and compliance in stable patients with HF ended.
Background The long-term effects of HF programs are not well known.
Methods In all, 1,518 patients with HF were randomized into the DIAL (Randomized Trial of Phone Intervention in Chronic Heart Failure). After completion of the trial, patients were followed up to 3 years to assess major outcomes. Compliance with diet, weight control, and treatment was evaluated. The effect of the intervention on mortality and HF hospitalizations was assessed using relative risk (RR), relative risk reduction, and Cox proportional hazards model for adjusting by potential confounders.
Results The rate of death or hospitalization for HF was lower in the intervention group (37.2% vs. 42.6%, RR: 0.81, 95% confidence interval [CI]: 0.69 to 0.96; p = 0.013) 1 and 3 years (55.7% vs. 57.5%, RR: 0.88, 95% CI: 0.77 to 1.00; p = 0.05) after the intervention ended. This benefit was mainly caused by a reduction in admission for HF (28.5% vs. 35.1% after 3 years, RR: 0.72, 95% CI: 0.60 to 0.87; p = 0.0004). Patients who showed improvement in 1 or more of 3 key compliance indicators (diet, weight control, and medication) had lower risks of events.
Conclusions The benefit observed during the intervention period persisted and was sustained 1 and 3 years after the intervention ended. This effect may be explained by the impact of the educational intervention on patients' behavior and habits.
Heart failure (HF) remains a serious health problem because of the sustained increase in prevalence, incidence, and high mortality as well as the poor quality of life that this condition confers on patients (1–5). The decrease in the quality of life is mainly related to the frequent hospital admissions. Furthermore, hospitalizations impose the most relevant personal, social, and economic burdens related to this syndrome (6). The majority of hospital admissions have preventable causes such as diet and treatment noncompliance, inappropriate social support, or delayed medical consultation for symptoms of HF progression (7–9).
Different intervention programs based on comprehensive care and regular follow-up by a multidisciplinary team have recently shown admission and cost reduction (10–17). However, most of this evidence comes from small, single-center, nonrandomized studies that use complex strategies for high-risk populations (18,19). A recent review highlighted the poor quality of the evidence that most of these programs had led so far (20).
Different questions arise regarding the most appropriate strategies to implement these types of programs. Moreover, the required intensity, duration, and long-term effects of the intervention are still unknown. Previous studies have not conducted long follow-ups to assess these potential benefits over time.
In this context, DIAL (Randomized Trial of Phone Intervention in Chronic Heart Failure) (21,22) offers a unique opportunity to study both the long-term effectiveness of the interventions and the potential mechanisms by which these interventions provide favorable results.
The design and results of the DIAL trial have been reported elsewhere (21,22). Briefly, the study was a randomized, controlled, open-label, multicenter trial that compared a centralized regular phone intervention versus usual care in outpatients with stable, chronic HF. Patients were recruited from 51 participating centers in Argentina.
The protocol was approved by internal review boards of each participating institution. All patients provide a signed informed consent at inclusion. For those patients who did not agree to receive the intervention, we used the intention-to-treat analysis (Zelen criterion) (23).
Patients assigned to the intervention group received an explanatory booklet at randomization and were followed up with a telephone intervention by specialized nurses. The objectives of the intervention were to improve diet and treatment compliance, to promote exercise, to regularly monitor symptoms, weight, and edema, and to promote early visits if signs of clinical deterioration were detected.
Nurses could adjust diuretic dose and suggest unscheduled visits to the attending cardiologist. Patients were initially called every 14 days, and after the fourth call, the frequency could be adjusted according to the severity of each case and patient compliance. More severe and less compliant patients received more calls according to pre-established criteria (22). Nurses were not allowed to up-titrate medication. They were only allowed to adjust short-term changes in diuretics, under supervision by the attending cardiologist (22).
Subjects in the control group continued treatment with their cardiologist in the same manner as the intervention group, except for the phone calls and the explanatory booklet.
The primary end point of the study was the rate of all-cause mortality or hospital admissions for HF 1 year after randomization.
Hospital admission for HF was defined as the presence of congestive HF at admission and the need of specific treatment for at least 24 h.
All events were assigned by a committee blinded to randomization status.
The DIAL study extended follow-up
The objective of this extended study was to determine the rate of death or hospitalization for HF, and the occurrence of each single component of this composite end point, 1 and 3 years after the intervention ended (e.g., 2 and 4 years after randomization). To assess the potential decrease in effectiveness of the intervention after being discontinued, because the main trial was stopped, and interaction of the time and intervention status, we used Cox regression analysis to test for time dependency. Additionally, we assessed whether patients with early improved compliance with diet, weight control, and medication showed greater long-term benefits.
Characteristics of patients with and without intervention are reported as percentages and mean ± SD and compared with the Pearson chi-square test and ttest for categorical and continuous variables, respectively.
All analyses were performed by intention to treat. The primary end point was assessed using the log-rank test and depicted using Kaplan-Meier curves. The effect of the intervention is reported using relative risk (RR) and relative risk reduction (RRR), and tested using log-rank tests, considering person-time incidence rates. These analyses were done both at the end of the trial and at 1 and 3 years after the intervention ended.
To assess a potential time-dependent effect of the intervention, particularly a progressive decrease of effectiveness across time, a Cox proportional hazards model was used including the intervention as a time-dependent variable. Results are expressed as hazard ratio (HR) and 95% confidence interval (CI) (24,25).
To assess the impact of adherence improvement, the change in adherence was determined from the first visit and from the first telephone call until the first 45 days of intervention. Patients were categorized as having improved adherence in 3 areas: diet, weight control, and medication. Intervention patients were classified as having improved in no indicator, 1 indicator, 2 indicators, or 3 indicators. The adherence indicators were derived from the questionnaire applied at every telephone call to the patients. At each call, nurses addressed if the patient complied always, sometimes, or never since the last call in prescribed medications, diet, and daily weighting.
The log-rank test was used to test the null hypothesis of no difference between groups, and Kaplan-Meier curves were constructed. These variables were included in a Cox regression model to assess the impact of adherence in the primary end point and HF admissions 3 years after follow-up, adjusting for potential confounders. In all analyses, a p value <0.05 was considered statistically significant.
Baseline characteristics of patients
In all, 1,518 patients were randomized between June 2000 and November 2001. Of these, 760 patients were assigned to the intervention group and 758 to the control group.
Overall, from the 760 who were offered the intervention, 10 patients refused the intervention, although they agreed to be contacted for follow-up. The study ended on August 1, 2002, with the confirmation of at least 400 primary end points.
Both groups were similar at baseline (Table 1)(22). Mean age was 65 years; 71% were male and 29% were female; 93% were on a regimen of angiotensin-converting enzyme inhibitors or angiotensin-II receptor blockers, and 62% were receiving beta-blocker therapy. Most patients were in functional class II or III, and approximately 80% showed left ventricular systolic dysfunction.
The mean follow-up duration at the end of the main study was 16 months, ranging from 1 to 27 months. Overall follow-up was 57 months from the beginning of the trial (16 months during the trial and 41 months after the trial), ranging from 1 to 84 months.
Effect of the intervention during the trial
At the end of the study, the primary end point (combined rate of death or hospitalization for HF) occurred in 200 (26.3%) patients in the intervention group and in 235 (31%) in the control group (RRR: 20%, 95% CI: 3% to 34%, p = 0.026). Kaplan-Meier survival curves started to separate before 3 months of follow-up. At day 180 and at 1 year, the RR was 0.63 (95% CI: 0.47 to 0.85, p = 0.0016) and 0.75 (95% CI: 0.60 to 0.93, p = 0.009), respectively.
The effect of the intervention on the primary end point was driven by a reduction in the rate of hospitalizations for HF. A total of 128 (16.8%) patients randomly assigned to the intervention group and 169 (22.3%) randomly assigned to the control group had an admission for HF (RRR: 29%, 95% CI: 9% to 44%, p = 0.005) with no significant effect on mortality (RRR: 5%, 95% CI: −27% to 23%, p = 0.69).
At the end of the trial, significantly more patients in the intervention group compared with the control group were taking beta-blockers (59.2% and 51.6%, p = 0.003), spironolactone (27.2% and 22.6%, p = 0.03), digoxin (33.4% and 28.6%, p = 0.04), and furosemide (77.3% and 70.5%, p = 0.007), and a similar trend was observed in the use of angiotensin-converting enzyme inhibitors (78.3% and 75.8%, p = 0.24). However, prescription patterns retrieved by physicians were similar between groups (data not shown).
Patients randomly allocated to the intervention group reported higher scores in the Minnesota Living with Heart Failure Questionnaire (scores from 0 to 105: 0 indicates no changes in quality of life, and 105 indicates a maximum alteration), both in the global scores (30.6 vs. 35.0, p = 0.001) and in the physical domain (11.2 vs. 12.8, p = 0.007) and emotional domain (6.7 vs. 7.9, p = 0.002). The median of ambulatory visits per year in both groups was 4.
Post-trial effects of intervention
The clinical benefit of the intervention remained 1 and 3 years after the trial period. One year after the intervention ended, there was a 19% reduction in the incidence of the primary end point (p = 0.013) and at 3 years, a 12% significant reduction (p = 0.05) (Table 2).This effect was caused by a reduction in the incidence of hospital admissions for HF: 174 HF admissions (22.9%) versus 220 admissions (29.0%; RR: 0.73, 95% CI: 0.60 to 0.90, p = 0.002) occurred 1 year after, and 217 HF admissions (28.9%) versus 266 admissions (35.1%; RR: 0.72, 95% CI: 0.60 to 0.87, p = 0.0004) occurred 3 years after for the intervention and the control group, respectively (Table 2, Fig. 1).All-cause mortality was the same between patients randomly assigned to the intervention and patients assigned to the control group at 1 year (RR: 0.94, 95% CI: 0.77 to 1.16, p = 0.59) and 3 years (RR: 1.02, 95% CI: 0.87 to 1.20, p = 0.73), respectively (Table 2, Fig. 1).
In the Cox time-dependent regression model, with the inclusion of the intervention as a time-dependent variable, no significant interaction was observed at follow-up. This finding shows a persistence of the effect of similar magnitude (HR) during the entire follow-up (Fig. 1). Also, proportionality hazards were maintained between groups, suggesting a persistent effect in time.
Of 760 patients in the intervention group, 69 (9.1%) did not improve compliance during the first 45 days, 296 (38.9%) improved only in 1 indicator (diet, weight control, or medication), 277 (36.4%) improved 2 indicators, and 118 (15.5%) improved in all 3 indicators. Overall, 82.8% improved in medication compliance, 40.7% improved in diet compliance, and 34.9% improved in daily weight control.
Considering the primary end point after 3 years of follow-up, the cumulative incidence in the control group was 57.5% (n = 436), 65.2% (n = 45) in the intervention groups with no improvement in compliance, 57.4% (n = 170) in the intervention group with improvement in 1 indicator, 52.3% (n = 145) in the intervention group with improvement in 2 indicators, and 53.4% (n = 63) in the intervention group with improvement in all 3 compliance indicators (log-rank test p = 0.041). Differences were more significant when admission for HF was considered as an end point: control group, 35.1%; intervention without improvement, 33.3%; with 1 indicator, 33.1%; with 2 indicators, 27.1%; and with all 3 indicators, 17.8% (log-rank test p = 0.0009) (Fig. 2).These differences persisted after being adjusted by other potential confounders in the Cox regression analyses (sex, age, New York Heart Association functional class, diabetes mellitus, previous admissions, chronic obstructive pulmonary disease, systolic dysfunction): the HR for the intervention group categorized by compliance versus the control group for HF admission according to compliance improvement were as follows: no improvement, HR: 0.91 (95% CI: 0.59 to 1.40, p = 0.67); 1 indicator improvement, HR: 0.87 (95% CI: 0.69 to 1.10, p = 0.25); 2 indicators improvement, HR: 0.71 (95% CI: 0.55 to 0.92, p = 0.012); and 3 indicators improvement, HR: 0.44 (95% CI: 0.28 to 0.69, p < 0.001).
At the end of follow-up, functional capacity was better in the intervention group (8.07% deteriorated in their functional class, 64.44% remained without changes, and 27.48% improved) than in the control group (39.31% with deterioration, 39.31% without changes, and 21.38% with improvement; p < 0.0001).
There was no significant change in blood pressure between groups during the trial period and follow-up.
Despite a number of clinical trials showed a short-term favorable effect of management programs for patients with HF (18,19), as already recognized in systematic reviews (20), most of these trials were small, conducted in selected centers, and had many methodological weaknesses. Additionally, no trial provided information about the effectiveness of these interventions on a long-term follow-up period.
To our knowledge, the DIAL trial is the first randomized multicenter study that assessed the effect of a simple, pragmatic, telephone intervention in a large population with HF that also has a long-term follow-up. This is particularly important as the benefit of follow-up programs has only been observed during their implementation.
The main result of this extension of the DIAL trial is that this simple, nurse-based telephone intervention was associated with a clear clinical benefit for patients with chronic HF 1 and 3 years after the intervention stopped. This education and monitoring based strategy was particularly effective in reducing the incidence of hospital admissions for HF. The benefit and the magnitude of benefit observed during the intervention remained in the midterm after the conclusion of the telephone follow-up with a similar effect over time.
Why did we observe a sustained effect after the conclusion of the intervention? The mainstay of the telephone intervention were both patient monitoring and education strategies aimed at improving treatment and diet compliance and self-monitoring of HF signs and symptoms, allowing early consultation.
This benefit may be related to a change of patients' habits and behaviors, as that is suggested by the improved patient compliance rate at the end of the study. Also, the intervention was associated with a quality of life improvement, not only in the disease-related physical aspects but also in the emotional factors that could have contributed to increased compliance. Improved compliance during the early stages of the intervention was related to a greater benefit up to 3 years after termination of the trial. The beneficial mechanism of the intervention can be partially explained by this change of patients' habits, mainly influenced by the nurses' interventions focused on education, self-management, and monitoring.
Furthermore, the analysis of events after the conclusion of the intervention shows that the benefit is maintained in the intervention group. This effect persistence shows that behavioral changes secondary to education can be the main mechanism of action, explaining the benefits of this intervention and that this methodology empowers the patients.
The lower event rate observed in this trial as compared with other studies that tested similar interventions may be explained by the inclusion criteria of the trial. The DIAL trial randomized stable outpatients with optimized therapy at enrollment. Other trials included post-discharge patients without optimal treatment, and part of their benefit might be related to an improvement in drug prescription in a higher risk population. Differences observed in medication use were explained by increased adherence, because prescription patterns were similar between groups. A higher use of beta-blockers and spironolactone could have explained part of the benefit observed, but this increase was not related to an intensified treatment but was due to a higher compliance. That highlights even more the findings of the DIAL trial, putting into perspective that this simple and pragmatic intervention was able to reduce HF admissions on top of the best medical treatment available.
Another important aspect that should be considered is the external validity of these results. This was a simple, low-cost intervention, $90 per patient per year, easily performed with different types of patients and centers. Moreover, the benefit observed in this study could be even higher among patients with HF treated by generalists. Optimal compliance was low in these patients (15%), and more efforts should be allocated to effective strategies to improve adherence.
One of the limitations of the study could be a potential biased classification of events due to an open trial design. To avoid this, an event committee blindly classified each event before the analysis. Also because of the open design, head physicians could have intensified care for patients assigned to the intervention. However, this was unlikely because drug prescription was similar for both groups during the whole study, and we found no differences in the number of outpatient visits (22).
The clinical evolution of optimally treated patients with HF can improve with the implementation of follow-up programs based on simple and low-cost interventions. This type of education and monitoring programs may provide sustained clinical benefits, having a strong impact in patients' behavior. Patients who change their behavior early during the intervention are at lower risk of events. In the light of the available evidence, the implementation of these programs could become a standard of care for patients with HF.
This trial was not submitted originally to an open database, but the main trial was originally published in the British Medical Journalin 2005.
- Abbreviations and Acronyms
- confidence interval
- heart failure
- hazard ratio
- relative risk
- relative risk reduction
- Received October 29, 2009.
- Revision received March 5, 2010.
- Accepted March 19, 2010.
- American College of Cardiology Foundation
- Cohen-Solal A.,
- Desnos M.,
- Delahaye F.,
- Emeriau J.P.,
- Hanania G.,
- Myocardiopathy and Heart Failure Working Group of the French Society of Cardiology,
- National College of General Hospital Cardiologists,
- French Geriatrics Society
- Capomolla S.,
- Febo O.,
- Ceresa M.,
- et al.
- Cline C.M.,
- Israelsson B.Y.,
- Willenheimer R.B.,
- Broms K.,
- Erhardt L.R.
- Jaarsma T.,
- Halfens R.,
- Huijer Abu-Saad H.,
- et al.
- Stewart S.,
- Blue L.,
- Walker A.,
- Morrison C.,
- McMurray J.J.
- Clark R.A.,
- Inglis S.C.,
- McAlister F.A.,
- Cleland J.G.,
- Stewart S.
- Grancelli H.O.,
- Ferrante D.C.
- Clark A.M.,
- Savard L.A.,
- Thompson D.R.
- GESICA Investigators
- Cox D.R.
- Stewart S.,
- Horowitz J.D.