|Home | About | Journals | Submit | Contact Us | Français|
Heart failure chiefly affects the elderly, with frequent emergency admissions. Telemonitoring can identify worsening heart failure but previous randomized trials have enrolled selected patient populations. The Home-HF study examined the impact of home telemonitoring on typical heart failure patients discharged from three acute hospitals in North West London, UK.
Patients hospitalized with heart failure were randomized to telemonitoring or usual specialist care. Primary outcome measures were days alive and out of hospital. Secondary outcome measures were number and duration of heart failure hospitalizations, clinic visits, and quality of life. We recruited 182 patients. There was no difference in the primary outcome measure in the two groups, but there were significantly fewer unplanned hospitalizations for heart failure decompensation, and a reduction in clinic and emergency room visits in the telemonitoring group. There was no statistically significant difference in the mean direct health service costs.
Home telemonitoring in a typical elderly population of heart failure patients produces a similar outcome to ‘usual’ specialist care, but reduces clinic and emergency room visits and unplanned heart failure rehospitalizations at little additional cost. This method of disease monitoring may allow specialist services to increase the number of patients under their care.
Heart failure is one of the most common reasons for admission to hospital, and after discharge patients have a high risk of readmission with worsening symptoms.1 Much effort has been expended to reduce the risk of repeat hospitalization, particularly since this accounts for 60–70% of the health care expenditure on heart failure.2
Heart failure disease management programmes can reduce readmission rates and mortality.3,4 Although the interventions vary, a multi-disciplinary team approach with specialist heart failure nurses playing a pivotal role in monitoring signs and symptoms, optimizing medication, providing education, co-ordinating care, and facilitating self-care is now recommended in international guidelines.5,6
Replicating these multi-disciplinary interventions in routine practice is challenging, as many elderly patients are not able to attend frequent clinic visits, there are a limited number of specialist nurses, and a limit on the number of patients with whom they can maintain frequent contact.7 Importantly, in the UK only 20% of heart failure patients discharged from hospital have any form of specialist follow-up.8
Telehealth has the potential to deliver specialist services closer to patients’ homes. While this innovative technology shows promise much of the evidence for its use in heart failure management comes from observational studies of relatively young patients with advanced heart failure.9–14 Furthermore, the two largest randomized studies to report in full excluded patients with an ejection fraction above 3915 and 35%,16 despite this group of patients being common and having a risk of readmission close to those with systolic heart failure.17,18
Perhaps not surprisingly, given the variation in patients included and the type of technology, there is considerable heterogeneity in the impact of telemonitoring on patient outcome. A recent overview suggests that there is little impact on total hospitalization, but a reduction in mortality and a possible reduction in heart failure hospitalizations.19
The HOME-HF Study was designed to determine if home telemonitoring of the signs and symptoms of typical heart failure patients recently discharged from hospital could reduce the risk of all-cause re-hospitalization, when compared with usual specialist care (UC), and at what cost.
Home-HF was a multi-centre randomized controlled trial involving three acute hospitals in urban, multi-ethnic North West London. All patients admitted to hospital with a diagnosis of heart failure as defined by the European Society of Cardiology criteria6 (either a new diagnosis or an acute decompensation of chronic heart failure) were screened for eligibility. Patients were recruited between July 2006 and August 2007.
All patients aged over 18 years with a primary diagnosis of heart failure and deemed fit for discharge by their medical team, had a home telephone line, lived within the catchment population of the recruiting hospitals, and were in New York Heart Association class II–IV at the time of discharge, were eligible for inclusion in the study. Patients were not recruited if they had cognitive impairment sufficient to interfere with their use of the telemonitoring equipment.
After informed consent was obtained patients were randomized in a 1:1 ratio to receive either usual follow-up care available at each hospital from the cardiology service or daily home telemonitoring of signs and symptoms (the intervention group).
An initial home visit was made to patients in both arms of the study by the study nurse. At this visit patients received advice on self-monitoring of heart failure. In addition, those in the telemonitoring group (TM) had the equipment (Honeywell HomMed™) installed and were taught how to use it.
All patients received UC for that centre. Each of the three sites provided a heart failure service that included at least one cardiologist or a physician with a specialist interest in heart failure, and at least one heart failure specialist nurse. Each hospital provided regular clinic review by the heart failure team and telephone support was available during normal working hours, Monday to Friday. The clinic review provided life-style advice and optimization of heart failure medication. The frequency of clinical follow-up was at the discretion of the heart failure team.
In addition to UC, patients in the intervention group had telemonitoring equipment installed in their home to monitor signs and symptoms indicative of worsening heart failure. The equipment included an electronic weighing scale, automated blood pressure cuff, pulse oximeter, and a control box which was connected to the domestic phone line. Each morning the patient followed the verbal instructions from the monitor to weigh themselves and use the equipment to record their blood pressure, heart rate, and oxygen saturation (Figure 1). They then answered four questions related to symptoms indicative of heart failure decompensation (breathlessness, orthopnoea, dizziness, ankle swelling) by pressing buttons marked ‘yes’ or ‘no’ in response to the questions from the monitor. All the readings were encrypted and then transmitted via a dedicated telephone line to the base station in each of the hospitals.
The transmitted data were reviewed on a daily basis (Monday to Friday) by a heart failure nurse. Any variation of these vital signs from predefined parameters triggered an alert suggesting clinical deterioration and resulted in a telephone call for further patient assessment. This then led to one of four possible responses: life-style advice, advice regarding medication, recommendation to contact primary care, or early review in secondary care.
All patients were followed-up for study purposes for a maximum period of 6 months, at which point the telemonitoring equipment was removed from the homes of the intervention group. UC continued thereafter for both groups.
After fully informed consent was obtained from an eligible patient, third party randomization was undertaken using computer generated random blocks of random size, with patients randomized in a 1:1 ratio to UC or TM. Randomization was stratified by hospital site. Both the recruiting nurse and the patient were therefore unaware of the group allocation until after consent was obtained.
Home-HF was designed to assess if heart failure patients using home telemonitoring following discharge from hospital had a reduced risk of all-cause re-hospitalization when compared with UC. A cost analysis was incorporated into the study.
The primary study outcome was the number and duration of all non-elective hospitalizations. To adjust for the impact of death, the number of days alive and out of hospital was used as the primary outcome measure. An admission was defined as an overnight stay in hospital. We chose the clinically meaningful outcome of all-cause hospitalization to enable us to determine the clinical effectiveness of telemonitoring upon more of the patient ‘journey’ and to undertake a cost effectiveness analysis.
The secondary outcome measures were: the number and duration of heart failure related admissions (primary cause of admission as recorded by the attending physician); and the health-related quality of life measured by self-administered postal questionnaires—the disease-specific Minnesota Living with Heart Failure questionnaire (MLwHF)20 and the generic Euroqol (EQ 5D);21 and direct health service costs. Anxiety and depression were assessed using the Hospital Anxiety and Depression Score (HAD).22
Demographic data, medication history, and quality of life data were collected at the time of randomization. Patients were also given health diaries to take home, with instruction to record details of health care utilization made during the study follow-up. Details of hospital admissions, hospital visits, and telephone contacts with the heart failure nurses were also collected using hospital records, supplemented by the patient diaries.
At 3 and 6 months after randomization, quality of life questionnaires were posted to patients and a stamped addressed envelope was provided. They received one reminder if no response was obtained within 2 weeks. Primary care physicians were also contacted at these time points for details of the medication history and primary care contacts made over the follow-up period.
There were insufficient data to estimate an appropriate sample size for days alive and out of hospital. However, we assumed a risk of all-cause rehospitalization of 50% with UC and 36% in the telemonitoring arm over a 6 month period. A 30% relative reduction in the risk of rehospitalization with telemonitoring was considered clinically meaningful. For the study to be 80% powered at the 5% level of significance to detect this difference, we required 300 patients to be randomized and complete follow-up.
The intention-to-treat principle was applied. Baseline characteristics were expressed as mean and standard deviation or median and interquartile range, and proportion. The number of days alive and out of hospital were calculated and compared using a Poisson regression model, adjusted for time spent in the study. Time to first event was plotted as Kaplan–Meier curves, with comparison between the groups using Cox proportional regression analysis. A P-value of <0.05 was accepted as statistically significant. All analysis was undertaken by researchers blinded to the allocation group using SPSS™ version 15 software. We calculated the costs of home telemonitoring vs. UC over the 6 month period of the study; because of the time horizon discounting was unnecessary. We took a National Health Service perspective. The analysis was undertaken at the individual (patient) level. The costs included in the analysis were telemonitoring equipment cost, hospital readmission costs, drug costs, primary care visit costs, secondary care visit costs, and hospital transport costs. Costs for each component were computed by multiplying the volume of resources used in each component by their unit costs. Resource use data were obtained directly from the trial. The unit costs were obtained from published national sources.23,24
Quality-of-life data were compared for all patients. The problem of missing data was handled by comparing the results of analyses using multiple imputation, and lowest, highest, and mean value carried forward methods.25 This did not affect the outcome of the analyses.
The investigation conforms with the principles outlined in the Declaration of Helsinki. The study was approved by the medical Ethics Committee at each of the participating hospitals and all patients gave written informed consent to participate in the study. The trial is registered at www.clinicaltrials.gov (NCT00312884).
Between June 2006 and August 2007, 456 patients were assessed for eligibility and 182 (40%) patients consented to inclusion in the study, and were randomized in a 1:1 ratio to TM or UC (Figure 2). Where patients cited a reason for declining participation the most common was that they were overwhelmed with their health problems. Four patients withdrew from the study after initial consent. No patient was lost from follow-up during the 6 month study period.
After 13 months, recruitment was stopped due to funding arrangements.
Compliance with daily telemonitoring was excellent, with 95% of patients using the system daily for >90% of the time.
The mean time from randomization to installation of the telemonitoring equipment was 4 days. Three patients died in the TM arm before the equipment had been installed in their home, and 1 patient died in the UC arm before the initial home visit.
The baseline characteristics of the two groups were similar (Table 1). The mean age of patients was 72 years (SD 12; range 33–93), 82 (45%) were aged 75 years or over, and 66% were male. The study population was multi-ethnic, with 20% of South Asian background. At the time of randomization 66 of 168 (39%) patients with a measureable ejection fraction (EF), had preserved left ventricular systolic function (EF ≥ 40%). Coronary artery disease was the single most common aetiology (55%), and co-morbidity was common, with hypertension in 62%, diabetes mellitus in 36%, and chronic airways disease in 9%. Using a cut-off value of ≥8 on the HAD, 51 (28%) were at least mildly depressed, and 59 (32%) at least mildly anxious. For 45% of patients, the diagnosis of heart failure had been made for the first time during the index admission to hospital. Forty three (24%) patients had had at least one heart failure hospitalization in the year preceding randomization.
Patients in both groups were well treated at the time of randomization, with no statistical difference between the two groups: 102 (56%) were receiving a beta-blocker, 127 (70%) an ACE-inhibitor, 36 (20%) an angiotensin receptor blocker, and 74 (41%) an aldosterone antagonist. One hundred and sixty-nine patients (93%) were on a loop diuretic at the time of randomization.
On 166 occasions [in 54/84 (64%) patients who were telemonitored] the telemonitoring data and subsequent nurse telephone assessment identified clinical deterioration. On 72 (43%) of these occasions the patient was offered advice regarding their diuretic therapy, 32 (19%) lead to early outpatient clinic assessment or attendance in the emergency room, and 14 (8%) led to a recommendation to contact primary care. On the remaining 48 (29%) occasions the reinforcement of self-care information (including medication reminders) was the main focus of the intervention.
During the 6 months of follow-up there was no difference in the median number of days alive and out of hospital in the two groups: 180 days (IQR: 165–180) in UC and 178 days (IQR: 90–180) in the TM group (P = 0.30). Fifty-six (31%) patients had at least one non-elective hospital admission (all-cause acute medical or surgical admission); 23 patients in the UC and 33 in the TM group.
The median number of days spent in hospital (for those hospitalized) was also very similar in both groups, with a median of 17 (IQR 6–25) in the TM group and 13 (IQR 8–34) in UC (P = 0.99).
Twenty-seven patients (15%) were hospitalized with worsening heart failure during the follow-up period: 10 patients (16 hospitalizations) in UC and 17 patients (22 hospitalizations) in the TM group. There was no difference in the time to first heart failure-related hospital admission in the two groups (P = 0.11) (data not shown). Table 2 lists the primary and main secondary outcomes.
There were significantly more emergency heart failure admissions in the UC group compared with the TM, [UC: 13/16 (81%) vs. TM: 8/22 (36%), P = 0.01].
We noted a reduction in the number of secondary care outpatient visits with telemonitoring (733 with UC vs. 622 with TM) and a reduction in emergency room visits (32 UC vs. 20 TM). The number of primary care visits was very similar in both groups (403 UC vs. 421 TM).
There was no change in overall health-related quality-of-life as measured through the EQ5D over the 6 month follow-up period, although quality-of-life measured through the disease specific Minnesota living with heart failure questionnaire improved slightly (data not shown). There was no significant difference between the groups (P = 0.5 for EQ 5D and P = 0.6 for MLwHF).
If mean direct NHS costs are considered, the incremental cost per patient for telemonitoring is statistically non-significantly higher by £1600 per patient (P = 0.20), with a mean direct NHS cost for a telemonitored patient of £4610 (SD £7377), and £3006 (SD £3847) for usual care. The cost data were highly skewed: if the above comparison was performed using median (rather than mean) costs, the total median direct NHS costs per patient over the 6 month study period were £1688 (IQR: £878–£6305) for the telemonitoring arm, and £1498 (IQR: £751–£4053) for usual care.
Our study shows that simple home telemonitoring of a well-treated elderly heart failure population leads to outcomes similar to that provided by usual care from a secondary-care-based heart failure service. This outcome was achieved at a modest incremental cost. In the group being telemonitored there were less outpatient and emergency room visits, and when heart failure worsened to the extent that hospitalization was required for restabilization, such an admission was much more likely to occur in a planned manner. We have not shown that telemonitoring reduces all cause admissions, at least over a 6 month period after discharge from hospital.
We have also shown that telemonitoring was acceptable to, and highly useable, by this elderly multi-ethnic cohort of patients. Only one patient asked for the equipment to be removed, and 95% of patients used the monitoring equipment on at least 90% of the days they were in the study. This is consistent with other studies that report good patient satisfaction and usage of telemonitoring.15 In contrast to an earlier study of telephone monitoring in mixed ethnic populations,26 we noted no difference in the daily use of the telemonitoring regardless of age, gender, or ethnicity.
Although we failed to show an impact on our primary outcome, we have confirmed that home telemonitoring can allow early detection of worsening symptoms, and permit better scheduling of hospitalization. Planned rehospitalization is likely to be much less stressful for a patient than an emergency readmission, which is often out-of-hours and not necessarily under the care of the heart failure team.
Our study has extended the evidence base for telemonitoring for heart failure, in that we included patients with preserved systolic function. Previous randomized studies have only included patients with severe systolic dysfunction,15,16 despite heart failure with preserved systolic function being common, particularly in the elderly, and associated with a rehospitalization risk close to that of systolic heart failure.17,18 Also, the average age of patients in our study was considerably higher than earlier studies, where the average age is typically in the 60s rather than the 70s. Our population is therefore much more representative of the typical patient who is hospitalized with heart failure.
This study was not adequately powered for robust subgroup analysis. Exploratory analyses did not identify any specific subgroup that obtained a significantly different outcome in terms of hospitalization.
Other randomized studies have also failed to show a reduction in hospitalization with home telemonitoring,15,16 although they report a trend towards shorter admission durations. They do report a reduction in mortality, albeit in younger patients with poorer ventricular systolic function and more advanced symptoms. Similar to previous studies, we failed to show an impact on quality of life by 6 months.16
While our study was designed to use telemonitoring to detect heart failure decompensation it also led to the identification of other disease processes such as respiratory disease, pleural effusion, and arrhythmia. In addition, the ready communication route between patient and health care professional may have resulted in patients more easily reporting other symptoms that necessitated in-patient management. This indicates the complexity of caring for the more elderly heart failure patient with multiple co-morbidities and is being studied as part of a qualitative sub-study due to report later this year.
Home telemonitoring also empowers patients to be actively involved in monitoring and managing their condition, shifting the locus of control towards the patient in their home.
It is possible that a larger randomized trial may have detected a clinically significant reduction in hospitalizations with telemonitoring. Our planned sample size was not achievable during the recruitment phase of the study, despite 40% of eligible patients being enrolled from three large hospitals. However, our study still had 80% power to detect a reduction in all-cause hospitalization from 50 to 34% (at the 5% level of significance). The risk of rehospitalization observed was lower than expected at just over 30% of all patients within 6 months of randomization.
A longer duration of follow-up may have demonstrated greater benefit. However, two other randomized studies15,16 have also failed to detect a reduction in hospitalization. Those studies did report a reduction in mortality, but our study was not designed or powered to detect a difference in mortality.
The high quality of UC in this study, and a nurse home visit in both arms, may have reduced the impact of home telemonitoring. This is in keeping with results of other studies of multidisciplinary interventions, with more recent studies suggesting lower impact as the quality of hospital care improves, particularly in centres interested in heart failure management.27
Home telemonitoring of typical elderly patients recently hospitalized with heart failure does not reduce hospitalization, but does lead to a similar outcome to that of UC in UK hospitals with an interest in heart failure. Outpatient clinic and emergency room visits are reduced, and rehospitalization for stabilization of decompensation can be scheduled in a much more controlled manner. The incremental cost of using telemonitoring is small. Such an approach could enable a higher proportion of patients with heart failure to be monitored by specialist services. The technology is acceptable to patients and compliance is high. It is likely that our study will add further weight to the trend for technology to assist in the monitoring of heart failure, with less need for the patient and health care professional to interact face-to-face at an expensive outpatient clinic or home visit.
O.D., J.R., M.R.C.: design of study, conduct of study, analysis, and writing and editing manuscript. C.C., S.W.D., S.D.R.: principal investigators at study sites, editing manuscript. S.M.: design, analysis, and editing of manuscript. M.R.: statistician involved in design and analysis. M.R.C.: senior author underwriting study.
Honeywell HomMed provided the funds to perform this study and the telemonitoring equipment used, via a research contract agreement with Imperial College London. The design, conduct, analysis, and manuscript have not been influenced by Honeywell HomMed or any of its employees.
The authors thank the study nurses for co-ordinating the activities of the trial at each site (Katie Burke, Jessica Priestley, Patricia Taraborreli) and the heart failure specialist nurses for their support (Barbara Byrne, Sarah McDonagh, Helen Penston).
Conflict of interest: none declared.