A meta-analysis of randomized control trials of home-based secondary prevention programs for coronary artery disease

Abstract

Background

A variety of different types of secondary prevention programs for coronary heart disease (CHD) exist. Home-based programs have become more common and may be more accessible or preferable to some patients. This review compared the benefits and costs of home-based programs with usual care and cardiac rehabilitation.

Methods

A meta-analysis following a systematic search of 19 databases, existing reviews, and references was designed. Studies evaluated home-based interventions that addressed more than one main CHD risk factor using a randomized trial with a usual care or cardiac rehabilitation comparison group with data extractable for CHD patients only and reported in English as a full article or thesis.

Results

Thirty-nine articles reporting 36 trials were reviewed. Compared with usual care, home-based interventions significantly improved quality of life [weighted mean difference: 0.23; 95% confidence interval (95% CI): 0.02-0.45], systolic blood pressure (weighted mean difference: − 4.36mmHg; 95% CI: − 6.50 to − 2.22), smoking cessation (difference in proportion: 14%; 95% CI: 0.02-0.26), total cholesterol (standardized mean difference: − 0.33; 95% CI: − 0.57 to − 0.08), and depression (standardized mean difference: − 0.33; 95% CI: − 0.59 to − 0.07). Effect sizes were small to moderate and trials were of low-to-moderate quality. Comparisons with cardiac rehabilitation could not be made because of the small number of trials and high levels of heterogeneity.

Conclusion

Home-based secondary prevention programs for CHD are an effective and relatively low-cost complement to hospital-based cardiac rehabilitation and should be considered for stable patients less likely to access or adhere to hospital-based services. Eur J Cardiovasc Prev Rehabil 17:261-270 © 2010 The European Society of Cardiology

cardiac rehabilitation, disease management, health behavior, health promotion, health services, secondary prevention

Introduction

This systematic review examines the effects of homebased versus hospital-based secondary prevention programs for coronary heart disease (CHD) [1–39]. Programs provided in hospital settings are well established internationally and use multidisciplinary health care teams to address the main modifiable risk factors [40]. Such programs are beneficial [41], but patient access to them internationally is low (around 30%) [42], and is lower in groups with greater need for risk factor reduction [43]. Recent guidelines [44, 45] recommend that to address these inequalities alternative models to traditional hospital programs should be used more widely. A promising means of doing this is to provide interventions directly in patients’ homes.

Home-based secondary prevention interventions are formalized interventions for the secondary prevention of CHD with predominant or exclusive home-based components. These interventions can be provided in a range of ways including paper, face-to-face, electronic, or telephone-based methods. Although the effectiveness of secondary prevention programs is not dependent on placement in a hospital [36, 46, 47], more evidence is needed to identify the relative effectiveness and costs of home-based programs.

Methods

A multi-file search was undertaken to identify randomized control trials evaluating home-based interventions for patients with CHD. Nineteen indexing databases were searched using more than 100 search terms related to prevention, rehabilitation, and support services.

Only trials of secondary prevention programs for CHD with a predominant or exclusive home-based content were included. Other exclusion criteria were nonrandomized study or trial without a usual care or cardiac rehabilitation comparison group; intervention was not focused on secondary prevention (e.g. drug only intervention, heart failure management program, or intervention for other forms of atherosclerotic disease); program focused on a single risk factor (other than exercise) or was a predominantly hospital-based intervention (e.g. pre-discharge only, in-hospital, family or other medicine setting). Articles had to be published in English as a full article or thesis with original data that were extractable for CHD patients.

Study design filters were used only for databases that retrieved large numbers of records or had superior indexing, such as Medline and EMBASE. Study design filters were not used in databases that did not have adequate indexing or publication type limits. Two members of the team (A.M.C. and J.K.) independently searched the titles and abstracts of all citations against the a priori exclusion criteria to identify appropriate trials.

Data on outcomes for selected studies were extracted by A.M.C. and M.H. independently, and double checked by J.S. Definitions used for each outcome by the investigators were accepted. Cardiac rehabilitation interventions were hospital-based programs self-identified by authors of trials as constituting ‘cardiac rehabilitation programs provided in a centralized health or community setting'. After initial extraction, 26 original investigators were contacted by T. McC. by email (based on latest addresses as found in recent articles/Google searchers) to obtain data missing from the original articles.

Data analysis was performed using Review Manager (RevMan, version 4.2.8 for Windows. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2003). Differences in effects were examined for two comparisons (intervention vs. usual care, and intervention vs. cardiac rehabilitation). For dichotomous outcomes, that is, all-cause mortality, cardiovascular (CV) death, recurrent myocardial infarction (MI), morbidity CV event (excluding transient ischemic attack/stroke), studies were combined using risk ratios (RR) with corresponding 95% confidence intervals (95% CI), with the exception of smoking, and risk differences (absolute difference in the proportion of smoking cessation). For continuous outcomes, change from baseline data were used if available, otherwise endpoint data were used. Weighted mean difference with 95% CI or standardized mean difference (SMD) with 95% CI were calculated for continuous variables with the same or different scales, respectively. Standard deviations of the effect sizes were calculated exactly from available data whenever possible. In cases in which exact values could not be obtained, standard deviations were estimated using P values, ranges, or interquartile ranges. For standard deviations of change from baseline, when a measure of correlation was absent, a correlation estimate of 0.5 was assumed.

All pooling of data was done using the DerSimonian and Laird random effects model. Statistical heterogeneity was measured using I  ² statistic; based on this statistic, heterogeneity for each outcome was classified as negligible (I  ² = 0%), minimal (I  ² ≤ 20%), moderate (20% < I  ² < 50%), or substantial (I  ² ≥ 50%) [48]. Subgroup analyses were conducted based on the length of the follow-up period reported (short-term and long-term, defined as less than 12 months and equal to or greater than 12 months, respectively). If the studies were too heterogeneous to combine quantitatively, qualitative analysis was done. On account of the small sample size, there was not enough statistical power to examine for publication bias using the existing methods. Measurements of quality of life varied widely between trials. It was not therefore possible to pool quality of life measures from the trials. As an alternative, the SMDs of quality of life between trials were pooled to attain an overall difference in effect.

Results

From a preliminary screening of the titles and abstracts of 4986 citations, 69 were selected for review of full articles (Fig. 1). Additional search of the titles of articles included in previous reviews of secondary prevention programs or risk factors [41, 46, 49–55] yielded an additional 21 citations for full review.

Further screening by A.C. and M.H. identified 39 articles (reporting 36 unique trials) [1–39] meeting the eligibility criteria (Table 1). The number of articles exceeded the number of trials as one trial [36] reported cost data in a separate article [8], one trial reported both short-term and long-term outcomes separately [2, 30], and another trial reported different risk factors in different articles from the same intervention [33, 34].

Fig. 1

Search process. ∗Programs that were nonhome-based secondary prevention or cardiac rehabilitation programs that contained minimal follow-up support at home; ‡Some overlapping cost data. CAD, coronary artery disease; CHD, coronary heart disease; CHF, congestive heart failure; RCT, randomized controlled trial.

Descriptions of programs

The trials evaluated paper-based (n = 16), telephonebased (n = 12), home-visit (n = 5), or electronic (n = 2) interventions. One trial did not state its mode of delivery [19]. Twenty trials used ‘usual care’ comparison groups, nine used cardiac rehabilitation comparison groups, and seven used both. Twenty-six trials evaluated comprehensive interventions (i.e. focusing on a range of modifiable CV risk factors), whereas 10 were exerciseonly interventions.

Paper-based programs mostly used self-contained structured handbooks/manuals to impart risk factor information and guidance on changing physical activity, diet, and psychosocial well-being. Some of these interventions incorporated telephone backup from health professionals [8, 18, 20, 21, 28, 36, 37]. The most frequently researched proprietary intervention [4, 8, 18, 20, 22] was ‘The Heart Manual'. Telephone-based programs involved providing remote support from health professionals to patients, predominantly through regular telephone calls supplemented with ongoing telephone contact when needed. The majority of the interventions were selfidentified as being individualized [2, 11, 12, 16, 26, 30, 38]. Only a minority focused on exercise as opposed to comprehensive counseling on risk factor reduction [2, 26, 30]. Electronic-based programs were both webbased [1, 31] and involved comprehensive risk factor reduction support through a combination of email communication with health professionals, web-based content, and on-line discussion. Both trials incorporated remote monitoring; either of blood pressure and heart rate [1] or health behavior change [31]. Home-visit programs varied more widely; most involved home visits from nurses [24, 27, 29, 39], although one involved consultation with a physiotherapist [23]. These were comprehensive risk factor reduction programs provided after hospital discharge with two exceptions [23, 24].

The trials reviewed were generally of short duration: 27 had follow-up of less than 12 months. Relatively few measured mortality (all-cause), recurrent acute MI, or CV-related deaths. The methodological quality of the trials was moderate to low with Jadad scores ranging from 1 to 3 (Table 2). Most trials (24 of 36) did not clearly describe allocation concealment and eight did not adequately describe the services provided to the comparator arm.

Table 1

Characteristics of the interventions reviewed [1 − 39]

Type	Study	Size (n)	Population	Setting	Mean age (years)	Duration (months)	Content of intervention
Article/manual	Jolly et al. [18]	525	After AMI, coronary revascularization or CABG	UK	60.3	1.5, 2, 2.25, or 3	'Heart Manual’ introduced to patients on an individual basis in hospital or on a home visit. Risk factor counseling and support provided by telephone at 3 weeks, and three other visits took place at 6 weeks and 12 weeks. Contact could be made with rehabilitation nurse, if necessary
Article/manual	Dalal et al. [8]; Taylor et al. [36]	104	Hospitalized for AMI	UK	60.6	2-2.5	Patients given Heart Manual to use over 6 weeks. Cardiac rehabilitation nurse made one home visit in 1st week after discharge followed up by telephone calls over 6 weeks (typically one call in weeks 2, 3, 4, and 6)
Article/manual	Senuzun et al. [28]	60	Treated for CHD in a cardiology unit and eligible for cardiac rehabilitation	Turkey	53	3	Manual/Audio Visual-based aerobic exercise program with additional telephone support with health professional around goals, monitoring of progress and feedback; supplemented by one preliminary group counseling session; offered to literate patients only
Article/manual	Vale et al. [37]	792	With CABG, PCI, AMI, or unstable angina, coronary angiography with planned revascularization	Australia	58	6	Manual-based program in combination with individualized telephone calls at 2 weeks, one call every 6 weeks by coach. Used COACH program to provide support for risk factor reduction, incorporating process of continuous improvement to obtain results of risk factors, education regarding targets, monitoring of progress. Patients able to contact coach at any time by telephone
Article/manual	Lewin et al. [21]	142	From family medical practices prescribed nitrates over the previous year; no angina for over 1 year	UK	67	6	Manual: ‘Angina-Plan’ with audio-based tape for relaxation; some additional face-to-face counselling at start regarding misconceptions and follow-up telephone calls at 1, 4, 8, and 12 weeks to monitor progress
Article/manual	Clark et al. [7]	570	Women with cardiac disease (any condition involving the heart requiring regular treatment)	USA	72	12	Home workbook (PRIDE) supplemented with weekly community-based group meetings of women with health educator and peer leader over 4 weeks to learn self-regulation skills with physical activity as focus
Article/manual	Bell [4]	149	After AMI	UK	58	3	Participants used Heart Manual, and were contacted four times (2 weeks, 4 weeks, 7 weeks, and 18/19 weeks) post-AMI around progress, risk factor change and psychosocial well-being
Article/manual	Brosseau et al. [5]	80	At high risk after cardiac surgery	Canada	58.5	2	Patients received home aerobic training program, which was progressive and individualized. All participants were called twice in 1st week, once in 2nd week, and once every 2 weeks until 8th week
Article/manual	Linden [22]	41	Admitted to a CCU	UK	61.5	1.5	Heart Manual provided 2-3 days post-AMI; nurse follow-up and counseling after 1, 3, and 6 weeks given to both intervention and control
Article/manual	Brown et al. [6]	24	Males after elective CABG with no access to cardiac rehabilitation	Canada	55	3	Written, structured program for a progressive walking program. Patients kept a log of distance walked, duration and HR before and at the end of exercise. Logs were reviewed at 6 and 12 weeks
Article/manual	Heller et al. [17]	450	Age < 70 years with confirmed AMI admitted to hospital	Australia	59	6	Intervention included letter to GP (benefits of ASA/β-blockers) and three mail-out packages focusing on nutrition/exercise/smoking cessation; specific target for fat reduction with contract agreement; letters of encouragement with additional contracts regarding diet improvements
Article/manual	Sparks et al. [32]	52	Eligible for cardiac rehabilitation	USA	52	3	Individualized aerobic exercise program given as supplement to risk factor counseling as part of cardiac rehabilitation
Article/manual	Lewin et al. [20]	176	Aged ≤ 80 years hospitalized after AMI	UK	55	12	'Heart Manual’ including education, home-based exercise program, and audio tape-based stress management. Audiotape also provided to spouses; additional support offered to both groups 1, 3, and 6 weeks post-AMI through telephone or home visit
Article/manual	Taylor et al. [34]	210	Men aged ≤ 70 years hospitalized for AMI	USA	52	6	Individualized prescription for aerobic exercise from trained nurse
Article/manual	DeBusk et al. [9]	127	Men aged ≤ 70 years discharged with AMI	USA	53	6	Written home-based exercise program; containing information on benefits of exercise, exercises, recognizing worsening angina and changes in health; supplemented with initial specialist consultation and remove monitoring of intensity and health status
Article/manual	Miller et al. [25]	198	Males hospitalized for AMI	USA	52.9	2 or 5.5-6.5	Stationary cycling based on 70-85% of peak HR. ECG transmitted to nurse. Patients recorded exercise logs that were mailed to nurse every 2 weeks
Electronic	Zutz et al. [1]	15	On a waiting list for cardiac rehabilitation living within 60 km of site. No specific disease conditions mentioned	Canada	58	3	Intervention delivered through Web; included HR/BP monitoring Online intake forms (medical, risk factor, and lifestyle forms), one-on-one chat sessions at 12 weeks with RN case manager, exercise specialist, and dietician; weekly slide-based exercise presentations. Expert group chat sessions
Electronic	Southard et al. [31]	104	CHD or CHD and heart failure identified by hospital or primary care records	USA	62	6	Internet-based intervention with education modules, e-mail communication with case manager and dietician, optional on-line discussion group and entry of health behavior data to monitor self-progress
Telephone	Hanssen et al. [16]	288	Hospitalized for confirmed AMI	Norway	60	6	Individualized telephone support (weekly for first 4 weeks, then 6, 8, and 12 weeks; last call at 25th week) to support patient coping about lifestyle change and risk reduction; incorporating goal setting, emotional support, information on risk reduction and emotional support
Telephone	Mittag et al. [26]	343	Confirmed AMI, CABG or PCI eligible for CR	Germany	59	12	Individualized patient support by nurse-provided telephone calls provided monthly over 1 year supplemented with manual; both relating to risk factor reduction and cardiac ailments; based on cognitive theory
Telephone	Wu et al. [38]	54	Males after CABG surgery	Taiwan	61	3	Exercise program individualized based on stress test. Patients asked to keep exercise log. Program updated by office or telephone consultation every 2 weeks by rehabilitation nurses
Telephone	Arthur et al. [2]; Smith et al. [30]	242	After CABG surgery	Canada	63	6	Individualized exercise program based on ACSM guidelines and VO2max test. Patients asked to keep exercise log, and were telephoned every 2 weeks by exercise specialist
Telephone	Gallagher et al. [12]	196	Women hospitalized for: AMI, CABG, Angina, PCI	Australia	67	1.5	Individualized nurse-delivered counseling (1, 2, 3, and 6 weeks postdischarge) to promote self-management and psychosocial adjustment; included goal setting for risk factor reduction and benefits of rehabilitation; aimed to improve control, self-efficacy and increase knowledge
Telephone	Frasure-Smith et al. [11]	1376	Admitted to hospitals with AMI not related to coronary procedures	Canada	59	12	Individualized contact by experienced nurses monthly for 1 year; patients responded to questionnaire (anxiety, depression, impairment scale); if score low or readmitted, project RN provided additional home visits to evaluate patient's needs
Telephone	Taylor et al. [35]	585	Admitted to hospital aged ≤ 70 years with confirmed AMI	USA	57	6 min	Initial screening and risk factor reduction counseling in hospital, followed by provision of work book on range of risk factors, relaxation audiotape for stress management and supplementary nurse assessment and counseling by telephone
Telephone	Fletcher et al. [10]	88	Males with CAD aged ≤ 73 years with physical impairment	USA	62	6	Participants provided with wheelchair ramp and rollers and telephone ECG. Instructed to exercise 5days/week for a total of 100 min. Also received diet instruction
Telephone	Gortner and Jenkins [15]	156	Patients (married) who had undergone cardiac surgery (bypass and valve)	USA	58	1	Telephone follow-up for 4 weeks around recovery, risk factor reduction, physical activity coaching and spousal concerns; initial counseling around family coping and conflict resolution
Telephone	Beckie [3]	74	Scheduled for CABG surgery after hospital discharge	Canada	NR	1.5	Telephone education and support program after CABG involving discussion between specialist cardiac rehabilitation nurse and patient; four to six calls after 1 week postdischarge during the following 6 weeks
Telephone	Gortner et al. [14]	67	Patients (married) who had undergone cardiac surgery (bypass and valve)	USA	62	1	Telephone follow-up for 4 weeks around recovery, risk factor reduction, physical activity coaching and spousal concerns; initial counseling around family coping and conflict resolution
Telephone	Garding et al. [13]	51	Confirmed first time AMI admitted to hospital	Canada	NR	2	Telephone-base education done by two cardiac rehab nurses and one non-specialized nurse (mean =3 calls per patient) to address individual learning needs and education around risk factors
Home-visit	Sinclair et al. [29]	324	Discharged from hospital with AMI and aged ≥ 65 years	UK	73	2	At least two home visits (1–2 and 6–8 weeks postdischarge) from trained support staff nurse to encourage patients around compliance, support for risk factor reduction, advice on stress, exercise, smoking cessation and diet, increase social interaction. Visits supplemented by telephone support and manual
Home-visit	Robertson et al. [27]	68	Discharged from hospital with AMI	Canada	NR	1	Weekly visit for 4 weeks from critical care nurse providing support and education around risk factor reduction
Home-visit	Marchionni et al. [23]	270	Referred to CR by intensive care units after AMI and aged > 45 years	Italy	NR	14	Supervised instructions in hospital around exercise with individualized exercise prescription followed by weekly physical therapist visits to adjust exercise prescription
Home-visit	Young [39]	146	Elevated cardiac markers for AMI	Canada	68	2	Disease management program containing education, follow up and support for risk factor reduction received a minimum of 6 home care visits from cardiac nurse
Home visit	McHugh et al. [24]	98	On waiting list for CABG surgery	UK	62	8	Educational intervention of 1–2 h provided alternately by specialist nurse/community nurse team to reduce risk factors, improve knowledge and reduce distress using motivational techniques provided to patients waiting for CABG surgery; utilized guidelines
Other	Kugler et al. [19]	52	Males with their 1st AMI and entered the CR 7–14 weeks post-MI	USA	NR	2	Mode of delivery unclear. Both Intensive and less intensive home-program evaluated

ACSM, American College of Sports Medicine; AMI, acute myocardial infarction; ASA, Aspirin; BP, blood pressure; CABG, coronary artery bypass graft; CAD, Coronary artery disease; CCU, coronary care unit; CHD, coronary heart disease; CR, cardiac rehabilitation; GP, general practitioner; HR, heart rate; NR, not recorded; PCI, percutaneous coronary intervention; RN, registered nurse.

Although two trials were restricted to women only, the mean proportion of women in the trials was 24%. Twenty-three trials had no age-based exclusion criteria, 11 trials had upper age limit criteria (at a mean age of 73 years), and two trials had criteria intended to exclude patients who were not older adults [7, 23]. Most trials did not include any breakdown of the ethnic status of participants (71% of trials). In those that did record this, representation of participants categorized as being from ethnic minorities was 19% (± 18.9%, range: 3–58%).

Table 2

Methodological quality of trials

Study	Jaded score	Described as randomized	Method of randomization described and appropriate	Allocation concealment	Adequate description of usual care
Jolly et al. [18]	3	Yes	Yes	Adequate	Yes
Dalal et al. [8]; Taylor et al. [36]	3	Yes	Yes	Adequate	Yes
Senuzun et al. [28]	1	Yes	Unclear	Unclear	No
Vale et al. [37]	3	Yes	Yes	Adequate	Yes
Lewin et al. [21]	2	Yes	No	Unclear	Yes
Clark et al. [7]	3	Yes	Yes	Unclear	Yes
Bell [4]	3	Yes	Yes	Unclear	Yes
Brosseau [5]	2	Yes	No	Unclear	Yes
Linden [22]	2	Yes	No	Inadequate	Yes
Brown et al. [6]	1	Yes	No	Unclear	Yes
Heller et al. [17]	1	Yes	No	Unclear	No
Sparks et al. [32]	1	Yes	Unclear	Unclear	Yes
Lewin et al. [20]	2	Yes	Unclear	Adequate	Yes
Taylor et al. [34]	2	Yes	Unclear	Unclear	Yes
DeBusk et al. [9]	2	Yes	Unclear	Unclear	Yes
Miller et al. [25]	1	Yes	No	Unclear	Yes
Zutz et al. [1]	2	Yes	No	Unclear	No
Southard et al. [31]	3	Yes	Yes	Unclear	No
Hanssen et al. [16]	3	Yes	Yes	Unclear	Yes
Mittag et al. [26]	3	Yes	Yes	Unclear	Yes
Wu et al. [38]	1	Yes	No	Unclear	No
Arthur et al. [2]; Smith et al. [30]	2	Yes	No	Unclear	Yes
Gallagher et al. [12]	2	Yes	No	Unclear	Yes
Frasure Smith et al. [11]	3	Yes	Yes	Adequate	Yes
Taylor et al. [35]	2	Yes	Unclear	Unclear	No
Fletcher et al. [10]	2	Yes	No	Unclear	Yes
Gortner and Jenkins [15]	2	Yes	No	Unclear	Yes
Beckie [3]	1	Yes	Unclear	Unclear	Yes
Gortner et al. [14]	1	Yes	Unclear	Unclear	No
Garding et al. [13]	1	Yes	No	Unclear	No
Sinclair et al. [29]	3	Yes	Yes	Adequate	Yes
Robertston et al. [27]	3	Yes	Yes	Adequate	Yes
Marchionni et al. [23]	1	Yes	No	Adequate	Yes
Young [39]	3	Yes	Yes	Adequate	Yes
McHugh et al. [24]	2	Yes	Unclear	Unclear	Yes
Kugler et al. [19]	2	Yes	No	Unclear	No

The effectiveness of interventions was compared against usual care and/or provider-based cardiac rehabilitation programs. Usual care was defined as normal health care and/or risk factor management at the time the trial was undertaken without supplementary secondary prevention intervention. Cardiac rehabilitation was defined as dedicated secondary prevention programs provided by health professionals in an acute (hospital) or community care provider setting.

All-cause mortality

Home-based intervention compared with usual care

Home-based intervention [4, 10, 11, 37] (four trials, 2510 patients) did not improve mortality over usual care (RR: 1.22, 95% CI: 0.83-1.80). There was no evidence of significant statistical heterogeneity between the trials (P = 0.85, I  ² = 0%). However, there was only 17% power to detect a 20% relative risk reduction (RRR), as this analysis is based on only 100 deaths in 2510 patients.

Home-based intervention compared with cardiac rehabilitation

The relative effect on all-cause mortality was similar between home-based and cardiac rehabilitation programs [4, 8, 18, 29, 30, 39] (six trials, RR: 1.08, 95% CI: 0.73-1.60) and there was no evidence of statistical heterogeneity between trials (P = 0.80, I  ² = 0%). However, there was only 18% power to detect a 20% RRR, as this analysis is based on only 97 deaths in 1548 patients.

Cardiovascular events

Home-based intervention compared with usual care

Home-based intervention [11, 17, 22, 25, 31] (five trials, 2078 patients) reduced risk of CV events (excluding stroke, transient ischemic attack, and heart failure) by 9%; however, this did not reach statistical significance (RR: 0.91, 95% CI: 0.78-1.05). There was no evidence of significant statistical heterogeneity between the trials (P < 0.47, I  ² = 0%). However, there was only a 73% power to detect a 20% RRR, as this analysis is based on only 447 deaths in 2078 patients.

Home-based intervention compared with cardiac rehabilitation

Although the pooled data suggested that there were no significant differences with home-based interventions [18, 25, 39] (three trials, 778 patients) compared with cardiac rehabilitation in reducing CV events (RR: 0.90, 95% CI: 0.33-2.43), there was substantial statistical heterogeneity (P < 0.0001, I  ² = 89.7%). One trial [39] reported a statistically significant benefit in the homebased arm, but it differed from the two negative trials as it involved six home visits to patients from a cardiac nurse and was supplemented by referral to specialist care. This specialist care and more intensive support may have contributed to its favorable results.

Quality of life

Only one trial identified a significant improvement in quality of life [2]; this trial compared a home-based program with cardiac rehabilitation.

Home-based intervention compared with usual care

The summative weighted mean difference in quality of life for the five trials reporting quality of life [4, 16, 20, 24, 31] (644 patients) was 0.23 (95% CI: 0.02-0.45), suggesting that quality of life was improved in those patients exposed to the home-based intervention. Significant improvements were particularly evident in short-term effects (three trials, 391 patients, follow-up <12 months, SMD: 0.37, 95% CI: 0.17-0.57), but not long-term effects (two trials 253 patients, ≥ 12 months, SMD: 0, − 2.25 to 0.25).

Home-based intervention compared with cardiac rehabilitation

The summative mean difference for the five trials reporting quality of life [2, 4, 7, 8, 30] (1070 patients) was 0.13 (95% CI: − 0.03 to 0.30) using data for shortterm and long-term outcomes (< 12 months and ≥ 12 months). This represents a nonsignificant quality-of-life improvement. Effects were significant in the short term (two trials, 346 patients, SMD: 0.28, 95% CI: 0.06-0.50), but this difference did not persist in the long term (SMD: 0.06, 95% CI: − 0.13 to 0.25).

Table 3

Effects on risk factors in the interventions reviewed (compared with usual care)

Risk factor	Home versus usual care	Heterogeneity	Possible explanation of heterogeneity
Systolic blood pressure (rest)a	− 4.36mmHg (n = 1095; 95% CI: − 6.50 to − 2.22) [1, 4–6, 10, 24, 28, 37]	P = 0.53 I  2 = 0%	NA
Smoking cessationa	DP: 14% (n = 1660; 95% CI: 0.02-0.26) [4, 16, 17, 22, 24, 37]	P < 0.000001 I  2 = 84.0%	Intervention intensity and length
Total cholesterola (≤ 12 months)	SMD: −0.33 (n = 995; 95% CI: − 0.57 to − 0.08) five trials [1, 10, 24, 28, 37]	P = 0.17 I  2 = 37.4%	Intervention intensity
BMI	WMD: − 0.46 (n = 1063; 0.66 to − 0.29) [1, 4, 24, 28, 31]	P = 0.51 I  2 = 0%	NA
Depressiona	SMD: −0.33 (n = 2332; 95% CI: − 0.59 to − 0.07) [4, 11, 20, 21, 37]	P < 0.0001 I  2 = 85.7%	Population and measurement differences
METS achieved on testing	WMD: 0.96 METS (n = 197; 95% CI: − 0.24 to 2.16) [1, 5, 25, 28]	P = 0.0009 I  2 = 81.9%	Program length and exercise type

CI, confidence interval; DP, difference in proportion; METS, metabolic equivalents; NA, not applicable; SMD, standardized mean difference; WMD, weighted mean difference. ^aSignificant improvement in favor of home-based programs.

Table 4

Effects on risk factors in the interventions reviewed (compared with cardiac rehabilitation)

Risk factor	Home versus CR	Heterogeneity	Possible explanation of heterogeneity
Systolic blood pressure (rest)	− 4.43mmHg (n = 860; 95% CI: −10.23 to 1.38) [4, 8, 18, 26]	P = 0.01 I  2 = 72.4%	Intervention type and length
Smoking cessation	DP: 5% (n = 942; 95% CI: − 0.04 to 0.14) [4, 8, 18, 26]	P = 0.05 I  2 = 61.4%	Unclear
Total cholesterol	SMD: − 0.11 (n = 966; 95% CI: − 0.11 to 0.32) [4, 8, 18, 26]	P = 0.07 I  2 = 58.2%	Findings
BMI	WMD: − 0.30 (n = 800; 95% CI: − 0.73 to 0.13) [4, 8, 18, 30]	P = 0.40 I  2 = 0%	NA
Depression	SMD: 0.06 (n = 1576; 95% CI: − 0.15 to 0.04) [2, 4, 7, 8, 18, 26]	P = 0.75 I  2 = 0%	NA
WHR	WMD: −0.02 (n = 446; 95% CI: −0.04 to 0.00) [2, 4, 30]	P = 0.11 I  2 = 55%	Unclear
METS	WMD: 0.23 (n = 596; 95% CI: − 0.34 to 0.81) [2, 8, 25, 30]	P = 0.01 I  2 = 72.9%	Findings

CI, confidence interval; CR, cardiac rehabilitation; DP, difference in proprotion; METS, metabolic equivalents; NA, not applicable; SMD, standardized mean difference; WHR, waist hip ratio; WMD, weighted mean difference.

Table 5

Costs for interventions (where provided)

Study	Cost saving (Y/N)	Data-led conclusion (Y/N)	Cost per patient of home intervention (inflationary adjustment)	Amount of costs saved (total cost of reduced hospitalizations minus cost of home program/per patient)
Jolly et al. [18]	N	Y	$330 ($336)	NR
DeBusk et al. [9]	Y	NR	$328 ($592)	NR
Lewin et al. [20]	Y	N	$50 ($68)	NR
Southard et al. [31]	Y	Y	$453 ($495)	$965 per patient
Mittag et al. [26]	Y	N	NR	NR
Taylor et al. [33]	Neutral	Y	NR	NR
Robertson et al. [27]	Y	Y	$312 ($356)	$178.62 per patient
Marchionni et al. [23]	Y	Y	$179 ($196)	NR

NR, not recorded; Y/N, yes/no.

Coronary heart disease risk factors

Significant benefits of home-based programs over usual care were evident in resting systolic blood pressure, cholesterol levels, smoking cessation rates, and depression scores (Table 3). The effects on other risk factors were positive, but nonsignificant. No differences in the effect on risk factors were evident compared with cardiac rehabilitation (Table 4).

Costs

Eight trials provided costs (Table 5) [9, 18, 20, 23, 26, 27, 31, 36]; six reported that interventions were ‘cost saving’ [9, 20, 23, 26, 27, 31], though only three of these studies presented data to support this conclusion [23, 27, 31]. The costs of home-based interventions were modest and averaged around US$300 for each patient adjusting for inflation.

Discussion

Home-based interventions for the secondary prevention of CHD showed a number of benefits on the quality of life and atherosclerotic risk factors compared with usual care. Individually, reductions in risk factors were moderate, but, in combination, would lead to clinically important reductions in risk [56, 57].

Home-based programs showed a 14% superior smoking cessation rate than usual care. Smoking cessation in patients with CV disease lowers mortality risk by around 46% in men and women [58]. The home-based interventions also reduced total cholesterol. Although differences in recording in the trials prevented the synthesis of these data beyond the calculation of standardized mean change, changes of 1mmol/l total cholesterol in patients with CHD represent a high relative risk reduction and are associated with the reduced risk of cardiac death and nonfatal MI by 18–28% [59]. The 4mmHg reduction in systolic blood pressure found in home-based programs should reduce subsequent CV events by up to 20% [59]. As around half of all patients with CHD are hypertensive [60], these reductions are clinically important. Similarly, reductions in depression were small, but significant because depression after MI is common and doubles the risk of adverse events [61].

On account of the small number of trials and their heterogeneous nature there was inadequate trial data to establish whether home-based secondary prevention programs are as effective as traditional hospital-based programs. Future trials should be carried out to determine this.

This review is limited by the small number and relatively short-term follow-up of the trials, and low mortality and morbidity rates. Thus, although we showed beneficial changes in risk factors, larger studies with longer followup are required to confirm the improvements in the major clinical endpoints because it takes around 2 years for mortality benefits from secondary prevention programs to emerge [41]. In addition, participants in the trials were younger and, compared with clinical populations, were likely to be more motivated and have fewer comorbidities [41]. Furthermore, despite contact with original trial authors, we were unable to comprehensively define the care provided to the control arms. As with most systematic reviews, we confined our search to published trials in English. Although it was not feasible to statistically evaluate the possibility of publication bias because of the relatively small number of trials, published studies may be a subset of all the trials conducted on this topic. Issues remained with regard to the comparability of trials, and statistical heterogeneity was evident in many of the syntheses performed. Heterogeneity in meta-analyses can be caused by sampling error, known or unknown differences in intervention, and population or methodological differences between the trials [48, 62, 63]. For example, this review included interventions that were defined as either ‘secondary prevention’ or ‘cardiac rehabilitation’ and/or differed in potentially salient characteristics such as mode of provision, duration, and health care personnel involved. However, because of the small number of trials, it was not possible to examine the effects of these factors or to explain the differences in outcomes. Future trials should attempt to examine the influence of such factors. These trials should be adequately described to allow for rigorous subanalyses or sensitivity analysis.

Although most patients with CHD do not achieve lifestyle, risk factor, and therapeutic targets for secondary prevention [64], the low likelihood of harm associated with secondary prevention interventions and the relatively low costs involved, home-based secondary prevention programs are a viable and effective method of secondary prevention that may circumvent some of the common problems associated with patient access for some populations. Given the marked reductions in participation associated with increasing distance of residency from a cardiac rehabilitation centre and the inability to compare home-based with hospital-based programs, home-based programs offer a promise for individuals less likely to access hospital-based cardiac rehabilitation. Until further high-quality evidence accrues, the relative effectiveness of home-based programs over hospital-based programs remains uncertain. Patients who are frail, clinically complex, or with multiple comorbidities that are often absent from trials of secondary prevention interventions, should continue to receive close supervision [41].

Acknowledgements

The authors are grateful to the trial authors who provided additional data or clarifications: Dr Wendy Young, Dr Heather Arthur, Dr Kate Jolly, Dr H. Dalal and Dr Rod Taylor, Dr Bob Lewin, Dr Noreen Clark, Dr Ann Brown, Dr C Barr Taylor, Dr Nancy Miller, Dr Nancy Frasure-Smith, Dr Louise Jenkins, Dr Theresa Beckie, Dr Grace Lindsay (for McHugh), Kim Robertson and Dr Tove Hanssen and Dr Robert DeBusk. They also thank Lisa Hartling, Dr Brian Rowe, and Dr Terry Klassen from the Evidence-Based Practice Center, University of Alberta for their methodological support.

The study was funded by the Public Health Agency of Canada, Ottawa, Ontario. A.M.C. and F.A. Mc. A are supported by career awards from the Alberta Heritage Foundation for Medical Research, Edmonton, Alberta. A.M.C. and M.H. are supported by career awards from the Canadian Institutes of Health Research, Ottawa, Ontario. F.A. McA is joint holder of the Merck-Frosst/Aventis Chair in Patient Health Management.

Competing interests: the authors have no personal conflicts. The data reported do not necessarily represent or reflect the perspective of any funding body.