Chapter  106:  Cardiac Resynchronization Therapy for Congestive Heart Failure

Pharmacologic Management

Afterload reduction with the combination of isosorbide dinitrate and hydralazine was one of the first pharmaceutical combinations tested and used to treat heart failure. It achieved a relative risk reduction in mortality of approximately one third when compared to placebo.^{26, 27} While enalapril (an angiotensin-converting enzyme [ACE]) inhibitor was subsequently proven to be more effective than hydralazine/isosorbide dinitrate, the combination remains useful for management of those heart failure patients intolerant of ACE inhibitors or angiotensin-receptor blockers (ARB), and in those with advanced renal insufficiency.¹⁰

ACE inhibitors were the first medications to show a reduction in mortality in advanced congestive heart failure patients (those with NYHA Class IV symptoms), with a relative risk reduction for the enalapril group (versus placebo) of 40% at 6 months.²⁸ Since then, 33 other randomized placebo-controlled trials²⁹ of ACE inhibitors have confirmed substantial benefits (in the order of 25% to 35% reductions in mortality and 27% to 33% reductions in hospitalization rates) in patients with all classes of heart failure (even asymptomatic patients with reduced ejection fraction). ACE inhibitors are now the cornerstones of treatment for systolic left ventricular dysfunction with or without symptoms.³⁰

Beta-blockers are the second choice in the pharmacologic treatment of systolic dysfunction. Twenty-two randomized controlled trials have demonstrated consistent benefits in reducing mortality (relative risk reduction 35%) and morbidity (relative risk reduction 36% in hospitalization),³¹ and improving NYHA class and quality of life.³²

Angiotensin-receptor blockers have been demonstrated to reduce a combined endpoint of morbidity and mortality by 13% when added to ACE inhibitor therapy and have an indication as primary therapy for heart failure in ACE-inhibitor intolerant patients, either alone or with beta-blockers. The reduction in hospitalization alone is 27% when added to an ACE inhibitor. Caution is urged when patients are concomitantly taking an ACE-inhibitor and beta-blocker, as no additional benefit was seen in a group of patients on all three agents.^{33, 34}

Aldosterone blockade (with spironolactone or eplerenone) is associated with relative risk reductions of 15% to 30% in mortality for patients with NYHA Class III/IV symptoms and an ejection fraction <35%,³⁵ or patients with an ejection fraction <40% and a recent myocardial infarction.³⁶ These benefits were largely due to reduced rates of sudden cardiac death. Furthermore, both agents are associated with 27% to 35% relative risk reductions in heart failure hospitalizations.^{35, 36} Digoxin has also been shown to reduce hospitalization rates, but without significant impact on mortality in patients with CHF.³⁷

Non-Pharmacologic Management

Lifestyle modification is important in the management of symptomatic CHF patients. The importance of avoiding non-steroidal anti-inflammatory drugs has been highlighted in numerous studies where use of anti-inflammatory drugs has been associated with a two- to four-fold increase in hospitalizations for acute heart failure.^{38, 39} Non-compliance with sodium restriction has been identified as the cause of up to 22% of hospitalizations for CHF, highlighting the importance of dietary modification for these patients.³⁹ Finally, addressing concomitant risk factors for underlying coronary artery disease (dyslipidemia, tobacco use, hyperglycemia) is important in the secondary prevention of myocardial ischemia in CHF patients. Another non-pharmacologic strategy includes specialty clinics that provide a multidisciplinary disease management approach to this complex disease.⁴⁰ While a systematic review of 11 randomized trials showed that patients managed in specialized multidisciplinary heart failure clinics had better processes of care and outcomes,⁴¹ it is unclear whether these benefits can also be expected with “specialist” physicians operating outside the bounds of a multidisciplinary clinic. Exercise training can also result in substantial improvements in functional status for patients with CHF, but existing studies have not reported consistent results or have involved regimens that are not practical in an ambulatory care setting.^{42, 43}

Surgical Management

There are some surgical interventions for heart failure that may also improve survival when directed to the appropriate patients at risk. Left ventricular assist devices can improve short-term cardiac function and provide a bridge to cardiac transplantation for appropriately selected NYHA Class IV patients. Indeed, they have recently even been approved as destination therapy for patients with end-stage heart failure.⁴⁴ Valvular repair or replacement, revascularization of ischemic myocardium,⁴⁵ and other surgical approaches to heart failure have been proven or are under investigation to improve both survival and quality of life in selected patients. Finally, cardiac transplantation is effective. However, this is offered to fewer than 2500 individuals per year in the United States and is not a solution for the vast majority of heart failure patients.

Unfortunately, even with the best care using optimal combinations of these non-pharmacologic, pharmacologic, and surgical approaches, the mortality rate remains high and quality of life is usually poor for patients with CHF. As an example, in a specialized heart failure clinic where 86% of patients are on ACE inhibitors or ARB, 46% are on a beta-blocker, and 45% of patients seen since 1999 are on spironolactone, the mortality rates are still 27%, 43%, and 74% at 1, 2, and 5 years.⁴⁶

Clearly, there is a need for additional treatment strategies in CHF that can improve function, diminish symptoms, reduce hospitalizations, and/or increase survival. Recent studies have shown that cardiac resynchronization therapy (CRT) with biventricular pacemakers offers improvements in quality of life, NYHA class, and six-minute walk test results, but have yet to demonstrate a conclusive mortality benefit. Implantable cardioverter defibrillators (ICDs) do not improve functional outcomes, but do provide a substantial mortality benefit (through the prevention of sudden cardiac death) in patients with a history of ventricular arrhythmias or at high risk due to ischemic substrate and poor ejection fraction.⁴⁷

Literature Search

A medical librarian identified relevant databases and developed search strategies based on the following terms: biventricular pacing, biventricular pacer, biventricular stimulation, BiV, congestive heart failure, CHF, chronic heart failure, artificial cardiac pacing, heart diseases, chronic cardiac failure resynchronization therapy, dual-chamber pacing, cardiac resynchronization, Medtronic, InSync, ELA medical, randomized controlled trial, controlled clinical trial, meta-analysis, multi-center trial, safety, risk, adverse effects, side effects, harm, etiology, aetiology, contraindications, causation, causality, predict.

In addition, the investigators contacted the primary authors of key studies, sought Food and Drug Administration (FDA) reports, and reviewed the reference lists of all included articles. Additional unpublished data were also sought from the companies that make biventricular devices: Medtronic Inc., Guidant Corporation, and ELA Medical. The search was not limited by language or publication status and is considered up-to-date to June 2003.

Selection and Inclusion

Table 2. Inclusion/exclusion criteria for review on (more...)

Table 2. Inclusion/exclusion criteria for review on cardiac resynchronization therapy

Criterion	Efficacy review	Safety review
Study Design	Include: RCT (parallel or crossover) > 2 weeks duration.	Include: RCT (parallel or crossover) or non-RCT (e.g., registry data, prospective cohort, FDA document, etc.) > 2 weeks duration.
	Exclude: non-RCTs, acute physiological studies, studies not involving human subjects	Exclude: acute physiological studies, studies not involving human subjects
Participants	Include: symptomatic CHF (NYHA Class ≥ II), decreased LVEF, prolonged QRS. Must be receiving stable optimal drug therapy	Include: symptomatic CHF (NYHA Class ≥ II), decreased LVEF, prolonged QRS. Must be receiving stable optimal drug therapy
Intervention	Treatment with active CRT (also called BVP, multi-site pacing, dual chamber pacer) compared to either placebo pacing or uni-ventricular pacing or optimal drug therapy	Treatment with active CRT (also called BVP, multi-site pacing, dual chamber pacer). Comparison group not necessary
Outcome measures	Mortality (all-cause, CHF, sudden cardiac death, non-cardiac), CHF hospitalizations, 6MWT, NYHA class, QOL	Peri-implant mortality, successful implant rate, risks of /during implantation, risks following implantation

BVP = biventricular pacing; LVEF = left ventricular ejection fraction; NYHA = New York Heart Association; QOL = quality of life; RCT = randomized controlled trial; 6MWT = six-minute walk test.

Quality Assessments

Efficacy review. The methodological quality of RCTs was assessed independently by two reviewers (JE, CS) using two quality assessment methods. First, allocation concealment was assessed as adequate, inadequate, or unclear. Second, a five-point scoring system validated by Jadad⁵⁸ was used to assess randomization, double blinding, and reporting of withdrawals and dropouts (Table B-3). In addition, the funding source and whether authors reported use of intention-to-treat analysis were noted. Discrepancies were resolved through discussion between the two reviewers.

Safety review. Studies included in the safety review were assessed for quality independently by two reviewers (CS, LH) using a partially validated checklist developed by Downs and Black.⁵⁹ The checklist includes 28 questions evaluating five criteria (Table B-4). These five criteria are: Reporting (10 questions, total score 11), External validity (three questions, total score 3), Internal validity - bias (seven questions, total score 7), Internal validity - confounding (six questions, total score 6) and Power (two questions, total score 2). Decision rules regarding application of the tool were developed a priori through discussions with a cardiologist and a methodologist. Discrepancies in quality assessment were resolved through discussion.

Data Extraction

Data were extracted using standard forms (Tables B-5 and B-6) designed for either RCTs (efficacy and safety reviews) or cohort studies (safety review) and entered into an Excel spreadsheet. Data were extracted by one reviewer (JE, CS, NW, or LH) and checked for accuracy and completeness by a second (JE, NW, or CS). Extracted data included study characteristics, inclusion/exclusion criteria, baseline drug use, characteristics of participants, procedural data, and outcomes. Efficacy outcomes included mortality (all cause as well as CHF death, sudden cardiac death, non-cardiac death) and time to death; CHF hospitalizations, six-minute walk test, NYHA class, quality of life, peak oxygen consumption, left ventricular ejection fraction, mitral regurgitation jet area, ECG and echocardiogram results. Safety outcomes included risks during implantation (death, lead misplacement, device-related malfunctions, procedural complications, implant tools, heart function, and patient complaints), risks following implantation (mechanical malfunction, lead dislodgment, infection, pain), and successful implant rate.

Data Analysis

Efficacy review. The following imputations and manipulations were performed to form useable data. Standard errors were converted into standard deviations. Graph extraction was performed using Corel Draw 9.0 (Vector Capital, San Francisco, California) for time to death in the COMPANION trial.⁶⁰ Means were approximated by medians, and 95% empirical intervals were used to calculate approximate standard deviations for the MIRACLE trial.⁶¹ Change from baseline data was used wherever possible for continuous data (PATH CHF⁶² being the exception); however, since correlations between baseline and endpoint data were never reported, a correlation of 0.5 was assumed⁶³ to calculate the appropriate standard deviation for change from baseline data. Change from baseline and endpoint data were combined; both entities estimate differences between treatment groups. Efficacy results were extracted rather than intention-to-treat results when provided.

Numerical results were meta-analyzed primarily in Stata 7.0;⁶⁴ metagraphs were generated using S-plus 6.0.⁶⁵ For dichotomous results (e.g., CHF hospitalizations), the review included relative risks (RR) for each individual study as well as a pooled result among those studies that could be combined. Additionally, risk differences were considered where zero total events occurred in both groups. Increments of one or more across functional classification (e.g., from Class III to Class II) were considered a dichotomous outcome. For continuous variables, mean differences (e.g., six-minute walk test) were calculated for separate studies and the weighted mean difference (WMD) for the pooled estimate. Time-to-event data (i.e., death) were summarized by the log hazards ratio;⁶⁶ Kaplan Meier curves were generated. An individual patient dataset for this analysis was constructed using summary monthly mortality tables in the trial manuscripts. The Log Rank test assessed for treatment group differences across curves. All results were reported with 95% confidence intervals (CIs) where reasonable.

Due to the differences expected between studies (particularly in control group therapies), we decided a priori to combine results primarily using random effects model;⁶⁷ fixed effects models were considered in sensitivity analyses. Statistical heterogeneity was assessed by the Chi-square test; a conservative level of significance (p<0.10) was considered heterogeneous.⁶⁸ Also, heterogeneity was quantified and appropriated using the I-squared statistic.⁶⁹ Inclusion of studies with active control arms^{62, 70} was assessed in sensitivity analyses. Relevant direct subgroup comparisons were summarized, including effects of CRT in patients with more severe heart failure symptoms (NYHA Class III or IV). Implantable cardioverter defibrillators (ICDs) were considered in an indirect subgroup comparison using meta-regression.

Estimates of carryover effect were extracted from crossover designs. Only period one data were used for irreversible outcomes (i.e., death and CHF hospitalizations). Standard errors for crossover WMD were calculated according to Curtin.⁷¹

Safety review. Numerical results were meta-analyzed primarily in S-plus 6.0. Risks were simply pooled and all results were reported with 95% CIs. Statistical heterogeneity was assessed by the Chi-square test; p<0.10 was considered heterogeneous.⁶⁸ Also, heterogeneity was quantified and appropriated using the I-squared statistic.⁶⁹ Exclusion of NYHA class II data and studies with active control arms^{62, 70} was assessed in sensitivity analyses but is not reported here.

The possibility that reports may have been less judicious in reporting adverse events that did not occur was considered. Sensitivity analyses were performed where studies (randomized controlled or cohort) did not report a particular risk (e.g., death); zero adverse events were assumed for these studies. In addition, some implantation risks were reported by event and not by patient. This non-independence was small and would not be expected to affect results importantly.

Structure of the Decision Model

The primary analysis considered patients with NYHA III heart failure and prolonged QRS duration. This analysis considered the lifetime horizon, as recommended.⁷² A state-transition Markov model compared costs and outcomes of congestive heart failure treated by CRT and medical therapy vs. medical therapy alone. A cycle length of one month was used.

Figure 1. Decision model for medical therapy versus (more...)

Click on image to enlarge 

   Figure 1. Decision model for medical therapy versus cardiac resynchronization therapy for heart failure

Figure 1

Decision analyses were performed with DATA Pro software (TreeAge Software Inc., Williamstown, Massachusetts) and Excel 2000 software (Microsoft Inc., Redmond, Washington). Statistical analyses were performed with S-Plus (Insightful Inc., Seattle, Washington).

Input Data

Survival and Hospitalization. Baseline probabilities of adverse events, the probabilities of cardiovascular death, arrhythmic death, heart failure death, hospitalization for heart failure, and adverse effects associated with the therapy options, as well as estimates of life expectancy associated with each therapy, were derived from the meta-analyses described above. Nine trials were included in the efficacy analysis: Comparison of Medical Therapy, Pacing, and Defibrillation in Chronic Heart Failure (COMPANION),⁶⁰ CONTAK CD,⁷³ Garrigue et al.,⁷⁰ Multicenter InSync Randomized Clinical Evaluation (MIRACLE),⁶¹ Multicenter InSync Randomized Clinical Evaluation ICD (MIRACLE ICD),³⁶ Multisite Stimulation in Cardiomyopathies Sinus Rhythm (MUSTIC SR),⁷⁴ Multisite Stimulation in Cardiomyopathies Atrial Fibrillation (MUSTIC AF)⁷⁵, the Pacing Therapies for Congestive Heart Failure trial (PATH CHF)⁶² , and the RD-CHF Trial (Leclerq C, personal communication, December 2003).

The rate of events observed among patients randomized to medical therapy was annualized by using an exponential approximation.^{76, 77} Transition probabilities incorporated into the Markov Model were adjusted for the cycle length. Pooled relative risks were calculated by using the Dersimonian and Laird random effects model.⁶⁷

The base case analysis considered only the effect of CRT on all-cause mortality since it is difficult to subclassify causes of death in patients with cardiovascular disease. Cardiac and non-cardiac deaths were only considered in secondary analyses that accounted for the patient age at implantation (ie. differences in mortality due to unrelated causes).

Quality of Life. The health-related quality of life of patients with heart failure was estimated by eliciting utilities, since current standards suggest use of such outcome measures, and the relative cost-effectiveness of some cardiac therapies is sensitive to the magnitude of the difference between the utilities associated with either treatment.^{78, 79} Since the purpose of the decision analytic model was to consider an intervention in the context of resource allocation among different types of interventions, the use of a generic source of preferences was used as recommended.⁷⁸ A convenience sample (n=90) of members of the general public was recruited. Inclusion criteria were: fluent in English, provision of informed consent, age greater than 40, and no underlying cardiac disease. Consenting subjects estimated utilities for standardized health state descriptions by using the standard gamble technique.⁸⁰ Four health states were considered: NYHA functional class II, III, IV, and congestive heart failure severe enough to require hospitalization. Hypothetical scenarios describing what one would typically feel and experience if living with each of these health states were developed with input from an expert panel based on descriptors from the Health Utilities Index.⁸¹

Table 3. Input data for decision analysis on cardiac (more...)

Table 3. Input data for decision analysis on cardiac resynchronization in patients with NYHA Class III heart failure

Variable	Best Estimate	Lowest	Highest	Distribution	Threshold Value	References
Age of patient (yr.)	60	50	70	Normal	n/a	Assumed
Annual probability of cardiac death without CRT (%)	24.3	20.0	29.2	Beta	n/a	This systematic review
Relative risk of death with CRT	0.75	0.60	0.93	Normal	0.78	This systematic review
Annual probability of heart failure hospitalization without CRT (%)	56.0	47.6	66.2	Beta	n/a	This systematic review
Relative risk of heart failure hospitalization with CRT	0.68	0.41	1.12	Normal	0.85	This systematic review
Annual rate of cardiac death without CRT (%)	20.3	15.1	27.0	Beta	n/a	This systematic review
Relative risk of cardiac death with CRT	0.60	0.36	1.01	Normal	n/a	This systematic review
Relative risk of death due to unrelated causes with CRT	1.0	0.9	1.1	Triangular	n/a	Assumed
Annual rate of lead infection (%)	2.0	1.1	3.2	Beta	n/a	This systematic review
Annual rate of lead failure (%)	13.7	11.7	16.1	Beta	n/a	This systematic review
Annual rate of battery replacement (%)	10.8	8.7	13.4	Beta	n/a	This systematic review
Probability of death during insertion (%)	0.4	0.2	0.7	Beta	n/a	This systematic review
Probability of death during lead infection (%)	1.0	0	10	Beta	n/a	Assumed
Probability of death during lead failure (%)	1.0	0	10	Beta	7.3	Assumed
Probability of death during battery replacement (%)	1.0	0	10	Beta	8.6	Assumed
Utility of NYHA II heart failure	0.94	0.84	0.99	Triangular	n/a	Survey
Utility of NYHA III heart failure	0.84	0.71	0.98	Triangular	n/a	Survey
Utility of NYHA IV heart failure	0.74	0.58	0.91	Triangular	n/a	Survey
Utility of hospitalization for heart failure	0.57	0.48	0.80	Triangular	n/a	Survey
Relative Utility of heart failure with CRT	1.0	0.9	1.1	Triangular	n/a	Assumed
Duration of hospitalization for CRT implantation	5 days	0	1 mos.	Triangular	n/a	Assumed
Duration of hospitalization for lead failure	5 days	0	1 mos.	Triangular	n/a	Assumed
Duration of hospitalization for lead infection	5 days	0	1 mos.	Triangular	n/a	Assumed
Duration of hospitalization for battery replacement	5 days	0	1 mos.	Triangular	n/a	Assumed
Discount rate for future costs and effects	3%	0%	10%	Triangular	n/a	Gold78
Cost of CRT insertion *	$33,495	$16,747	$50,242	Triangular	$41,000	Survey
Cost of CRT, per mo.	$771	$385	$1,216	Triangular	n/a	Owens92
Cost of hospitalization for lead infection	$30,997	$15,499	$46,496	Triangular	n/a	Owens92
Cost of hospitalization for lead failure	$30,997	$15,499	$46,496	Triangular	n/a	Owens92
Cost of battery replacement	$28,835	$14,417	$43,252	Triangular	$38,000	Owens92
Cost of heart failure hospitalization	$15,427	$10,660	$20,193	Normal	n/a	Kaul82

CRT=cardiac resynchronization therapy; NYHA= New York Heart Association

*

Cost estimates in U.S. dollars

Assumptions. We made several assumptions about the costs and effects of CRT. First, unit costs of complications related to CRT are identical to those for implantable cardioverter defibrillators. Second, unit costs of heart failure therapy (ie. costs of out patient or in patient care) were otherwise identical between medical therapy and CRT. Third, the incidence of complications associated with CRT was constant over time from implantation. Since the duration of study followup in each trial was relatively short and the incidence of adverse effects observed in the randomized trials was higher than is generally accepted for implantable cardioverter defibrillator use (which may reflect relative inexperience with biventricular pacemakers),⁸³ the primary analysis assumed that the annual probability of each of lead infection, lead failure or mechanical malfunction was 3.0% (range 0% to 10%). We also assumed that any mechanical malfunction of the device required battery replacement with consequent costs and consequences. Fourth, uncomplicated CRT had identical health-related quality of life as that of medical therapy. Fifth, the utility for hospitalization was identical, regardless of the reason for hospitalization. The morbidity of hospitalization was incorporated into the model by assigning the utility for hospitalization to it's duration, and assigning the utility for the health state prior to hospitalization to the remainder of the cycle. Finally, age-specific mortality due to unrelated causes was based on life tables.⁸⁴

An intervention that was more effective and less costly than the other was considered to be “dominant” and hence preferred to the alternative.

Uncertainty and Variability Analyses

The analysis distinguished between parameter uncertainty (i.e., variation in costs and effects due to sampling and measurement error) and variability (i.e., heterogeneity in costs and effects between groups of patients with systematic differences in cost or effects). Uncertainty was assessed by using 10,000 probabilistic Monte Carlo simulations.^{85, 86} Empirical cost variables were assigned log-normal distributions. Empirical probability variables were assigned beta distributions.⁸⁵ Variables without a known distribution form (i.e., those with assumed values, or those with values based on a range of published reports) were assigned triangular distributions.⁸⁷ Since there is no absolute cost-effectiveness criterion,⁸⁸ the results of the Monte Carlo simulations were illustrated as a scatter plot of incremental effects (in quality-adjusted life years) versus incremental costs. In such a plot, the incremental cost-effectiveness ratio is represented by the slope of incremental costs to incremental effects. The uncertainty in costs and effects was also illustrated as a cost-effectiveness acceptability curve. An acceptability curve is a conditional probability plot showing the proportion of the observed incremental cost-effectiveness density that lies below a threshold ratio, which represents the monetary value of a unit of health gain. The plot is conditional on the threshold ratio, and therefore the decision maker can interpret the data in light of their threshold willingess to pay for the incremental health outcome.

Multiway sensitivity analyses were also performed. Since the effectiveness of both interventions may be correlated, we varied simultaneously the probability of arrhythmic death with medical therapy, and relative risk of arrhythmic death with CRT. Since the rate of adverse effects observed in the randomized trials was higher than is generally accepted for implantable cardioverter defibrillator use,⁸³ the incidence of device-related adverse effects observed in the trials was substituted into the model.

Figure 2. Markov Model of Resynchronization Therapy (more...)

Click on image to enlarge 

   Figure 2. Markov Model of Resynchronization Therapy of Heart Failure

Figure 2

Finally, subgroup analyses considered the incremental cost-effectiveness of CRT versus medical therapy alone for patients with NYHA II or NYHA IV heart failure by substituting the appropriate health-related quality of life weight.

Description of Included Studies

Nine randomized trials met the inclusion criteria for the efficacy review:

Abraham 2002: Multicenter InSync Randomized Clinical Evaluation (MIRACLE trial).⁶¹

Auricchio 2002: Pacing Therapies for Congestive Heart Failure (PATH CHF trial).⁶²

Bristow 2003: Comparison of Medical Therapy, Pacing, and Defibrillation in Chronic Heart Failure (COMPANION trial).⁶⁰

Cazeau 2001: Multisite Stimulation in Cardiomyopathies Sinus Rhythm (MUSTIC SR trial).⁷⁴

Garrigue et al.⁷⁰

Guidant Corporation: the CONTAK CD trial.⁷³

Medtronic 2001: Multicenter InSync Randomized Clinical Evaluation ICD (MIRACLE ICD).^{36, 93, 94}

Leclercq 2000: Multisite Stimulation in Cardiomyopathies Atrial Fibrillation (MUSTIC AF trial).⁹⁶

RD-CHF: presented at the 2003 European Society of Cardiology Meeting and personal communication from Dr. C. Leclerq November, 2003.

Table 5. Description of studies included in the efficacy (more...)

Table 5. Description of studies included in the efficacy review: CRT for CHF

				Number of Participants
Study name	Country	Design	Control arm	Total enrolled	Exclusions	Total randomized	Treatment*	Control*	Withdrawals	Device	Authors' primary outcome(s)	Other outcomes
		Duration								Method of implant
Abraham 2002 MIRACLE61	U.S. and Canada	RCT parallel	Pacer “Inactive”	571	4	453	228	225	Treatment 1	InSync model 8040 Medtronic	NYHA class; QOL, 6MWT	Peak O2 consumption; time on a treadmill; LVEF; LVEDD; severity of mitral regurgitation; duration of QRS interval; clinical composite response (improved, worsened, unchanged); death; worsening heart failure; number of days spent in hospital
		6 mo.							Control 8	Transvenous
Auricchio 2002 PATH-CHF62	Germany, Netherlands	RCT X-over	Uni-ventricular pacing (4 right, 36 left)	42	1	41	24	17	Treatment 2	Vigor / Discovery with LV lead Guidant	O2 uptake at peak exercise; O2 uptake at anaerobic threshold; 6MWT	NYHA; QOL
		1 mo.							Control 5	Transthoracic
Bristow 2003 COMPANION60	U.S. multicenter	RCT parallel 3 arms	OPT	-	-	1520	CRT=617	308	-	CONTAK TR Models 4510–4513, Model 1241 CONTAK CD Model 1823 Guidant.	CRT: all-cause mortality & hospitalization. CRT-ICD: all-cause mortality & hospitalization	Cardiac morbidity; ADRs, implant success; peak O2 uptake at exercise
		12 mo.					CRT-D=595			Transvenous
Cazeau 200174 MUSTIC-SR	Europe (15 sites)	RCT X-over	Pacer “Inactive”	67	3	58	29	29	Treatment 4	Chorum 7336 ELA Medical or InSync 8040 Medtronic	6MWT	QOL (main secondary outcome); peak O2 uptake; hospital admissions because of decompensated heart failure; patient's preference with regard to pacing (active vs. inactive) at the end of crossover; death
		3 mo.							Control 3	Transvenous
Garrigue 200270	France	RCT X-over	Left ventricular pacing	-	0	13	6	7	0	Model 2188 Medtronic	Clinical and hemodynamic variables	Hemodynamic changes during exercise
		2 mo.								Transvenous
Guidant 2002 CONTAK-CD73 FDA Report	U.S. (47 sites)	RCT	Pacer “Inactive”	581	15	490	245	245	Treatment 3	CONTAK-CD	All-cause mortality; CHF hospitalizations; ventricular tachyarrhythmia requiring device intervention	Peak O2 consumption; QOL; 6MWT, all adverse events
		Phase I period 1 of X-over 3 mo							Control 1	Transvenous and Transthoracic
		Phase II parallel 6 mo.
Leclercq 2002 MUSTIC-AF75	Europe (15 sites)	RCT X-over	Right ventricular pacing	64	10	43	25	18	Treatment 1	ELA medical, Medtronic	6MWT	Peak O2 consumption; QOL; hospital admissions for decompensated heart failure; mortality; patient's preferred period at end of crossover
		3 mo.							Control 2	Transvenous
									Both 2
Leclercq 2003 RD-CHF** Unpublished	France	RCT X-over	Right ventricular pacing	56	Unclear	44	22	22	Unclear	Transvenous	CHF hospitalization	N/a
		3 mo.
Medtronic 2001 MIRACLE-ICD36 FDA report	U.S. and Canada (53 sites)	RCT (post implant) parallel	Pacer “Inactive”	659	105	554	272	282	Treatment 7	Model 7272 InSync ICD, Attain LV leads models 4189, 2187, 2188.	NYHA; QOL; 6MWT;	ADRs; QRS; peak O2 uptake; echocardiographic indices; LV lead electrical performance; VT/VF therapy; CHF composite response; implant ventricular defibrillation criterion; ATP therapy; healthcare utilization; death
		6 mo.							Control 6	Transvenous

ADR = adverse reaction; CRT = cardiac resynchronization therapy; CRT ICD = CRT with implanted cardioverter defibrillator; FDA = Food and Drug Administration; LVEF = left ventricular ejection fraction; LVEDD = left ventricular end diastolic diameter; N/a = not available; NYHA = New York Heart Association class; QOL = quality of life; OPT = optimal pharmacological therapy; RCT = randomized control trial; 6MWT = 6-minute walk test; X-over = crossover study

*

the intervention that crossover studies received in the first period

**

detailed data not available for Leclercq 2003 (RD-CHF) at the time this report was prepared

t

excludes deaths and implant failures, where applicable

Note: While the published report for MIRACLE-ICD was used for patients with NYHA Class III or IV symptoms, the FDA report was used for NYHA Class II patients as they were not included in the published MIRACLE-ICD manuscript

Table 6. Baseline characteristics of patients in trials (more...)

Table 6. Baseline characteristics of patients in trials included in the efficacy review: CRT for CHF

					NYHA class				ECG			Physical exam			Other co-morbidities
					II	III	IV	Atrial fibrillation	PR interval	QRS interval	LVEDD	Systolic BP	Diastolic BP	HR
Trial name	Study Group	Males n (%)	Age mean SD	Ischemic %	%	%	%	n (%)	mean SD (msec)	mean SD (msec)	mean SD (mm)	mean SD	mean SD	mean SD	LBBB %	LVEF %
Abraham 200261 MIRACLE	CRT	155 (68)	64 +/- 11	50	-	90	110	0	-	167 +/- 21	70 +/- 10	114 +/- 18	69 +/- 10	73 +/- 13	-	22 +/- 6
	Control	153 (68)	65 +/- 11	58	-	91	9	0	-	165 +/- 20	69 +/- 10	115 +/- 18	68 +/- 10	75 +/- 13	-	22 +/- 6
Auricchio 2002 PATH-CHF62	CRT first	11 (46)	59 +/- 7	42	-	88	13	0	190 +/- 34	174 +/- 30	71 +/- 10	-	-	77 +/- 16	88	21 +/- 6
	Control first	10 (59)	60 +/- 5	6	-	82	18	0	207 +/- 30	178 +/- 34	75 +/- 13	-	-	80 +/- 13	100	20 +/- 7
	All	21 (50)	60 +/- 7	29	-	86	14	0	196 +/- 33	175 +/- 32	73 +/- 11	-	-	78 +/- 15	93	21 +/- 7
Bristow 200360 COMPANION	CRT	413 (67)	65	54	-	87	13	0	-	159	-	-	-	-	69	22
	Control	213 (69)	67	59	-	82	18	0	-	156	-	-	-	-	70	23
	CRT-ICD	399 (67)	66	55	-	86	14	0	-	159	-	-	-	-	73	23
Cazeau 200174 MUSTIC-SR	CRT first	19 (66)	64 +/- 11	-	-	100		0	-	172 +/- 22	-	-	-	-	-	-
	Control first	24 (83)	64 +/- 8	-	-	100		0	-	175 +/- 19	-	-	-	-	-	-
	All	43 (74)	64 +/- 9	37	-	100		0	215 +/- 43	174 +/- 20	73 +/- 10	-	-	-	-	23 +/- 7
Garrigue 200270	All	13 (100)	64 +/- 12	46	-	77	33	100	-	208+/- 15	-	-	-	-	62	25 +/- 8
Guidant 200273 CONTAK-CD FDA report	CRT II–IV	210 (85)	66 +/- 11	67	32	60	8	0	205 +/- 42	160 +/- 27	-	118 +/- 21	67 +/- 12	73 +/- 12	54	21 +/- 7
	Control II–IV	211 (83)	66 +/- 11	70	33	57	10	0	202 +/- 49	156 +/- 26	-	118 +/- 21	69 +/- 12	75 +/- 14	55	22 +/- 7
	CRT III/IV	90 (77)	66 +/- 11	65	17*	73	10	0	204 +/- 41	164 +/- 27	-	116 +/- 20	68 +/- 12	75 +/- 13	50	21 +/-6
	Control III/IV	86 (78)	66 +/- 11	71	10*	71	19	0	200 +/- 54	152 +/- 24	-	117 +/- 23	67 +/- 14	74 +/-15	54	21 +/- 6
Leclercq 200274 MUSTIC-AF	CRT first	21 (84)	65 +/- 9	-		100		25 (100)	-	209 +/- 21	70 +/- 9	-	-	75 +/- 6	-	23 +/- 7
	Control first	14 (78)	66 +/- 9	-		100		18 (100)	-	208 +/- 12	66 +/- 7	-	-	74 +/- 5	-	30 +/- 12
	All	35 (81)	65 +/- 8	43		100		43 (100)	-	209 +/- 18	68 +/- 8	-	-	74 +/- 5	-	26 +/- 10
Leclercq 2003 RD-CHF** Unpublished	N/a	N/a	73 +/- 8	N/a	0	N/a	N/a	23 (52)	N/a	N/a	N/a	N/a	N/a	N/a	N/a	25 +/-/9
Medtronic 200136 MIRACLE-ICD FDA report	CRT II–IV	231 (82)	66 +/- 12	38	32	60	8	16 (6)	-	166 +/- 23	71 +/- 9	113 +/- 19	67 +/- 11	73 +/- 13	0.7	21 +/- 7
	Control II–IV	217 (80)	66 +/- 11	32	38	55	7	12 (4)	-	163 +/- 22	70 +/- 9	113 +/- 17	68 +/- 12	73 +/- 13	0	21 +/- 7
	CRT III/IV	142 (76)	67 +/- 11	64		88	12	49 (26)	-	165 +/- 22	70 +/- 9	-	-	71 +/- 12	75	21 +/- 7
	Control III/IV	136 (77)	68 +/- 9	75		89	11	31 (18)	-	162 +/ -22	71 +/- 9	-	-	72 +/- 13	71	20 +/- 6

CRT = cardiac resynchronization therapy; FDA = Food and Drug Administration; HR = heart rate; LVEDD = left ventricular end diastolic diameter; LVEF = left ventricular ejection fraction; LBBB = left bundle branch block; N/a = not available; NYHA= New York Heart Association class

*

pre-implant assessment

**

detailed data not available for Leclercq 2003 (RD-CHF) at the time this report was prepared

Note: While the published report for MIRACLE-ICD was used for patients with NYHA Class III or IV symptoms, the FDA report was used for NYHA Class II patients as they were not included in the published MIRACLE-ICD manuscript

No trial specifically recruited patients based on the etiology of their heart failure, although patients with uncorrectable valvular disorders and hypertrophic or restrictive cardiomyopathy were excluded from all trials. In the three trials that evaluated CRT with ICDs, the majority of patients had ischemic etiology (~59%);^{36, 60, 73} in the other trials, ischemic etiology ranged from 37%⁷⁴ to 55%.⁶¹ Other comorbidities such as diabetes mellitus, renal failure, previous cardiac history, or a history of sudden cardiac death were reported in only one trial.⁹³

All trials (except PATH CHF, which did not specify an ejection fraction criteria)⁶² were limited to patients with an ejection fraction < 35% or < 40%. The mean ejection fractions were similar in all trials, and ranged from 21% to 30%. Six trials also specified a left ventricular end diastolic dimension: ≥ 55 mm in two trials^{36, 61} and ≥ 60 mm in the other four trials.^{60, 70, 74, 75} The reported mean left ventricular end diastolic dimension for the trials were similar (68 mm to 73 mm).

QRS width was a criteria for all trials, with three trials specifying ≥ 120 msec,^{60, 62, 73} two trials ≥ 130 msec,^{36, 61} one trial >140 msec,⁷⁰ one trial >150 msec,⁷⁴ one trial >180 msec, and one trial >200 msec.⁷⁵ Six of the nine trials had a mean QRS between 159 msec and 175 msec, with the MUSTIC AF trial having a mean QRS of 209 msec. Left bundle branch block was present in most patients (mean 64%; range 0 to100%).

Five trials were restricted to patients in normal sinus rhythm,^{60, 62, 73, 74} but two were restricted to patients with atrial fibrillation^{70, 75} In the eighth trial ~14% of patients had atrial fibrillation.⁹³ Overall, approximately 5% of randomized patients included in this meta-analysis had atrial fibrillation. Three trials reported PR intervals that ranged from 196 msec to 215 msec and two required a prolonged PR interval >150 msec for inclusion.^{60, 62} Three trials required a six-minute walk test result of less than 450 meters as an inclusion criterion.^{61, 74, 75} The physical exam findings at baseline were similar among trials, with systolic blood pressure (range 113 to 118 mmHg), diastolic blood pressure (range 67 to 69 mmHg), and heart rate (range 71 to 80 bpm) all similar to other trials in heart failure.⁹⁷

Table 7. Baseline drug therapy for trials included (more...)

Table 7. Baseline drug therapy for trials included in the efficacy review: CRT for CHF

		Medications
Trial name	Study Groups	ACE %	ARB %	BB %	Spironolactone %	Digoxin %	Nitrates %	Others Name %
Abraham 200261 MIRACLE	CRT	93*	-	62	-	78	-	Diuretics (94)
	Control	90*	-	55	-	79	-	Diuretics (93)
Auricchio 200262 PATH-CHF	CRT	96**	-	71	-	**	71	Amiodarone (29)
	Control	100**	-	65	-	**	71	Amiodarone (35)
	All	95**	-	67	-	**	69	Amiodarone (30)
Bristow 200360 COMPANION	CRT	89*	*	68	53	-	-	Diuretics (100); digoxin (100)
	Control	89*	*	66	55	-	-
	CRT-ICD	90*	*	68	55	-	-
Cazeau 200174 MUSTIC-SR	All	96*	*	28	22	48	-	Amiodarone (31), diuretics (100)
Garrigue 200270	All	100	-	100	-		-	Diuretics (100); amiodarone (100); CCB (100)
Guidant 200273 CONTAK-CD FDA report	CRT, II–IV	86*	*	48	-	69	-	Diuretics (88)
	Control, II–IV	89*	*	46	-	68	-	Diuretics (83)
	CRT, III/IV	81*	*	45	-	72	-	Diuretics (92)
	Control, III/IV	89*	*	40	-	68	-	Diuretics (86)
Leclercq 200275 MUSTIC-AF	All	100*	*	23	16	58	-	-
Leclercq 2003 RD-CHF*** Unpublished	N/a	N/a	N/a	N/a	N/a	N/a	N/a	N/a
Medtronic 200136 MIRACLE-ICD FDA report	CRT, II–IV	93	-	63	-	71	31	Diuretics (91); anti-depressant (16); CCB (6); antiarrhythmic (40); positive ionotrope (71); anti-coagulant (77).
	Control, II–IV	90	-	59	-	72	30	Diuretics (90); anti-depressant (17); CCB (6); anti-coagulant (79); antiarrhythmic (33); positive ionotrope (72)
	CRT, III/IV	92	-	63	-	76	36	Diuretics (93); anti-depressant (19); CCB (7); anti-coagulant(77); antiarrhythmic (42); positive ionotrope (76)
	Control, III/IV	88	32	58	-	73	33	Diuretics (94); anti-depressant (20); CCB (6); anti-coagulant (81); antiarrhythmic (32); positive ionotrope (73)

ACE= angiotensin converting enzyme inhibitors; ARB= angiotension-receptor blocks; BB= beta blocker; CCB= calcium channel blocker

*

receiving angiotensin-converting-enzyme inhibitors or angiotension-receptor blockers

**

receiving angiotensin-converting-enzyme inhibitors or digoxin

***

detailed data not available for Leclercq 2003 (RD-CHF) at the time this report was prepared

N/a =not available

Four of the trials employed a parallel study design.^{36, 60, 61, 73} One of these had planned a crossover period but changed protocol mid-study and excluded crossover data from the analysis;⁷³ five others completed a crossover design.^{62, 70, 74, 75, RD-CHF} The duration of treatment was 4 weeks in PATH CHF,⁶² 2 months in Garrigue et al.,⁷⁰ 3 months in RD-CHF, 3 months/phase in MUSTIC AF⁷⁵ and MUSTIC SR,⁷⁴ 6 months in MIRACLE⁶¹ and MIRACLE ICD,³⁶ and 12 months in the COMPANION⁶⁰ trial. Eight of the nine trials used a transvenous approach for placement of the epicardial leads (54 patients in CONTAK CD⁷³ required a transthoracic approach), while PATH CHF⁶² used a transthoracic approach.

COMPANION⁶⁰ was a three-arm, parallel-group trial that compared optimal pharmacological therapy (n=308), cardiac resynchronization therapy (n=617), and cardiac resynchronization therapy with cardioverter defibrillator (n=595) randomized in a 1:2:2 fashion before device implantation. The primary outcome was a composite of all-cause mortality and all-cause hospitalization (including emergency room visits or unscheduled office visits requiring >4 hours of intravenous vasoactive or inotropic drug).

CONTAK CD⁷³ was a two-part trial with an initial pilot crossover involving two 3-month phases (n=205 for first 3 month results). We were not able to obtain complete data from the second phase; hence this part of the trial was excluded from this report. For part two, they conducted a parallel design study with 6-month followup (n=151). The device used also had a cardioverter defibrillator. The primary outcome was a composite of all-cause mortality, heart failure hospitalizations, and ventricular arrhythmias requiring device intervention.

Crossover Trials. In MUSTIC SR,⁷⁴ 67 patients were enrolled and implantation attempted, followed by an 8- to 12-week observation phase; 58 patients were then randomized into a 3-month crossover of either CRT ‘on’ or ‘off’ (Phase 1: n=29, Phase 2: n=29). In MUSTIC AF,⁷⁵ 64 patients were enrolled and implantation attempted, followed by 8 to 12 weeks of observation; 43 were randomized into a 3-month crossover of CRT ‘on’ or ‘off’ (Phase 1: n=25, Phase 2: n=18). Both trials used the six-minute walk test as the primary outcome. Neither used a ‘washout’ period between phases and neither detected a carryover effect.

PATH CHF⁶² was a 4-week crossover study in which 42 patients were enrolled and implantation attempted; 41 patients were randomized to CRT 'on’ or ‘off’ in two phases with a 4-week washout period between the two (Phase 1: n=24, Phase 2: n=17). The primary endpoint was peak oxygen uptake on a maximal exercise test. This trial did detect a carryover effect.

Garrigue et al.⁷⁰ used a 2-month crossover design in which 13 patients had the device implanted (plus HIS-bundle ablation). The 13 patients were randomized to either CRT or left-univentricular pacing alone. The primary endpoint was peak endocardial acceleration time as measured by an implantable sensor. Patients were crossed over after 2 months without a washout phase. The study did not assess for a carryover effect in the analysis.

Methodological Quality of Included Studies

Table 8. Methodological quality of randomized trials (more...)

Table 8. Methodological quality of randomized trials included in the efficacy review

	Randomization		Double-blinding
Study	Stated	Method described	Stated	Method described	Description of withdrawals/drop-outs	Jadad Score	Allocation Concealment
Abraham 2002 MIRACLE61	Yes	Adequate	Yes	Adequate	Adequate	5	Clear
Auricchio 2000 PATH-CHF62	Yes	Unclear	Yes	Adequate	Adequate	4	Unclear
Bristow 2003 COMPANION60	Yes	Adequate	No	N/a	Unclear	3	Unclear
Cazeau 2001 MUSTIC-SR74	Yes	Adequate	No	N/a	Adequate	3	Unclear
Garrigue 200270	Yes	Unclear	No	N/a	Unclear	1	Unclear
Guidant 2002 CONTAK-CD73	Yes	Unclear	Yes	Adequate	Adequate	4	Unclear
Leclercq 2002 MUSTIC-AF75	Yes	Adequate	No	N/a	Adequate	3	Unclear
Leclercq 2003 RD-CHF* (unpublished)	Yes	N/a	No	N/a	N/a	1	N/a
Medtronic 2001 MIRACLE-ICD36	Yes	Unclear	Yes	Adequate	Adequate	4	Unclear

N/a = not available

*

detailed data not available for Leclercq 2003 (RD-CHF) at the time this report was prepared

All trials were described as randomized; however, the description of randomization detail varied. As far as blinding, two were open-label,^{60, 70} three were single-blind (i.e., the patient was blinded)^{60, 74, 75} and four were double-blind (patient and the outcome assessor was blinded).^{36, 61, 62, 73} In CONTAK CD,⁷³ MIRACLE,⁶¹ and MIRACLE ICD,³⁶ the independent events committee was blinded to which arm of the trial the patient was in; no information was available for COMPANION.⁶⁰ Notably, five trials randomized patients before implantation.^{60, 62, 73–75} An intention-to-treat statistical analysis was specified in all trials, and MIRACLE⁶¹ and PATH CHF⁶² performed an intention-to-treat analysis.^{61, 62} In the remaining trials, it was either unclear^{60, 74, 75} or multiple analyses were done for the FDA that included an intention-to-treat analysis of all patients involved in the trial program^{36, 73} Withdrawals and dropouts were clearly described in all trials except COMPANION.⁶⁰ Unscheduled crossovers occurred in zero to nine percent of the patients in these trials and were generally balanced between study arms. Withdrawals ranged from zero to 3% for the cardiac resynchronization group and from zero to 2.5% for the control groups. Allocation concealment was unclear for all but the MIRACLE⁶¹ trial.

Industry sponsored seven of the nine trials; two also received funding from government sources.^{74, 75} Guidant Corporation sponsored three,^{60, 62, 73} Medtronic Inc. sponsored four,^{36, 61, 74, 75} and ELA Recherche funded two.^{74, 75} Garrigue et al.⁷⁰ did not report funding sources.

Description of Included Studies

Table 9. Description of studies included in the safety (more...)

Table 9. Description of studies included in the safety review: CRT for CHF

				Number of Participants
Study name	Country	Design	Control arm	Total enrolled	Exclusionst	Total randomized	Treatment*	Control*	Withdrawals	Device	Authors' primary outcome(s)	Other outcomes
		Duration								Method of implant
Abraham 2002 MIRACLE61	US. & Canada	RCT parallel	Pacer “Inactive”	571	--	453	228	225	Treatment 1	InSync model 8040 Medtronic	NYHA class; QOL, 6MWT	Peak O2 consumption; time on a treadmill; LVEF; LVEDD; severity of mitral regurgitation; duration of QRS interval; clinical composite response (improved, worsened, unchanged); death; worsening heart failure; number of days spent in hospital
		6 mo.							Control 8	Transvenous
Auricchio 2002 PATH-CHF62	Germany & Netherlands	RCT cross-over	Uni-ventricular pacing (4 Rt., 36 Lt.)	42	1	41	24	17	Treatment 2	Vigor / Discovery with LV lead Guidant	O2 uptake at peak exercise; O2 uptake at anaerobic threshold; 6MWT	NYHA; QOL
		1 mo.							Control 5	Transthoracic
Bristow 200360 COMPANION	US. multicentre	RCT parallel 3 arms	OPT	-	-	1520	CRT=617	308	-	CONTAK TR Models 4510–4513, Model 1241 CONTAK CD Model 1823 Guidant.	CRT: all-cause mortality & hospitalization. CRT-ICD: all-cause mortality & hospitalization	Cardiac morbidity; ADRs, implant success; peak O2 uptake at exercise
		12 mo.					CRT-D= 595			Transvenous
Cazeau 200174 MUSTIC-SR	Europe (15 sites)	RCT cross-over	Pacer “Inactive”	67	3	58	29	29	Treatment 4	Chorum 7336 ELA Medical or InSync 8040 Medtronic	6MWT	QOL (main secondary outcome); peak O2 uptake; hospital admissions because of decompensated heart failure; patient's preference with regard to pacing (active vs. inactive) at the end of crossover; death
		3 mo.							Control 3	Transvenous
Cazeau 199699	France	Single-arm trial	None	8	0	-	6	-	0	Chorus TM 6234 Ela Medical, Medtronic Thera 7940 DR Transthoracic (7)	Potential hemodynamic benefit of CRT	Not stated
		2 mo.- 1 yr.								Transvenous (1)
Filho 200298	Brazil	RCT cross-over	Rt. uni-ventricular pacing	24	0	-	-	-	0	Not reported	NYHA	QRS; mortality; LVEF; hospitalization
		30, 90, 180 days								Transthoracic
Gras 2002100 INSYNC Italian Registry	Europe Canada	Single-arm trial	None	117	0	-	103	-	9	InSync Model 8040, Medtronic	Feasibility, safety and long term effects	NYHA, QRS duration, 6MWT, QOL
		up to 1 yr.								Transvenous
Guidant 200273 CONTAK-CD FDA Report	US. (47 sites)	RCT	Pacer “Inactive”	581	15	490	245	245	Treatment 3	CONTAK-CD	All-cause mortality; CHF hospitalizations; ventricular tachyarrhythmia requiring device intervention	Peak O2 consumption; QOL; 6MWT, all adverse events
		Phase I period 1 of cross-over 3 mo							Control 1	Transvenous and Transthoracic
		Phase II parallel 6 mo
Krahn 2002101	Canada	Single-arm trial	None	45	0	-	40	-	-	Medtronic InSync pacemaker or ICD or Guidant ContakTM or ICD	QOL; NYHA	Mortality, electrocardiographic measures; transplants
		1, 3, 6 mo. then q 6 mo.								Transvenous
Kuhlkamp 2002102	Germany	Single-arm trial	None	84	0	-	81	-	-	InSync Model 7272 Medtronic	Not stated	6MWT; QOL; NYHA; complications; death
		3 mo.								Transvenous
Leclercq 200096	France	Single-arm trial	None	37	0	34	49	-	-	Various models Medtronic	Mortality	NYHA;electrocardiographic measures; exercise tolerance.
		1, 3, 6 mo. then q 6 m								Transvenous
										Transthoracic
Leclercq 200275 MUSTIC-AF	Europe (15 sites)	RCT cross-over 3 mo.	Rt. ventricular pacing	64	10	43	25	18	Treatment 1	ELA medical, Medtronic	6MWT	Peak O2 consumption; QOL; hospital admissions for decompensated heart failure; mortality; patient's preferred period at end of crossover
									Control 2	Transvenous
									Both 2
Leclercq104 Unpublished	France	Single-arm trial	None	-	-	-	125	-	-	Not stated	Not stated	Survival, QRS duration and axis; NYHA; LVEF; exercise tolerance
		1, 3, 6 mo. Then q 6 mo.								Transvenous
Leclercq 2003 RD-CHF** Unpublished	France	RCT cross-over	Rt. ventricular pacing	56	N/a	44	22	22	N/a	Transvenous	CHF hospitalization	N/a
		3 mo.
Medtronic 200136 MIRACLE-ICD FDA Report,	US. & Canada (53 sites)	RCT parallel	Pacer “Inactive”	660	36	555	272	282	Treatment 7	Model 7272 InSync ICD	NYHA; QOL; 6MWT	Adverse events; QRS; peak O2 uptake; echocardiographic indices; LV lead electrical performance; VT/VF therapy; CHF composite response; implant ventricular defibrillation criterion; ATP therapy; healthcare utilization; death
		6 mo.							Control 6	Transvenous
Leon 2002106	US.	Single-arm trial	None	20	0	-	20	-	0	Revised RV pacing system to CRT Medtronic leads	Improved ventricular function	Success of procedure; NYHA; QOL; hospitalization
										Transvenous
Molhoek 2002103	Netherlands	Single-arm trial	None	40	0	-	40	-	0	Easytrack 4512–80, Contak TR or CD Guidant, or InSync III	Clinical benefit, long-term prognosis	NYHA; QOL; 6MWT; electrocardiogram; hospitalization; mortality
		3, 6 mo.								Transvenous

CRT-ICD = CRT with implanted cardioverter defibrillator; LVEDD = Left ventricular end diastolic diameter; LVEF = left ventricular ejection fraction; N/a = not available; NYHA = New York Heart Association; OPT = optimal pharmacological therapy; QOL = quality of life; RCT = randomized control trial; 6MWT = 6-minute walk test

*

the intervention that cross-over studies received in the first period

**

detailed data not available for Leclercq 2003 (RD-CHF) at the time this report was prepared

t

excludes deaths and implant failures, where applicable

Table 10. Baseline characteristics of patients in trials (more...)

Table 10. Baseline characteristics of patients in trials included in the efficacy review: CRT for CHF

					NYHA class				ECG			Physical exam			Other co-morbidities
					II	III	IV	Atrial fibrillation	PR interval	QRS interval	LVEDD	Systolic BP	Diastolic BP	HR
Trial name	Study Group	Males n (%)	Age mean SD	Ischemic %	%	%	%	n (%)	mean SD (msec)	mean SD (msec)	mean SD (mm)	mean SD	mean SD	mean SD	LBBB %	LVEF %
Abraham 200261 MIRACLE	CRT	155 (68)	64 +/- 11	50	-	90	10	0	-	167 +/- 21	70 +/- 10	114 +/- 18	69 +/- 10	73 +/- 13	-	22 +/- 6
	Control	153 (68)	65 +/- 11	58	-	91	9	0	-	165 +/- 20	69 +/- 10	115 +/- 18	68 +/- 10	75 +/- 13	-	22 +/- 6
Auricchio 200262 PATH-CHF	CRT first	11 (46)	59 +/- 7	42	-	88	13	0	190 +/- 34	174 +/- 30	71 +/- 10	-	-	77 +/- 16	88	21 +/- 6
	Control first	10 (59)	60 +/- 5	6	-	82	18	0	207 +/- 30	178 +/- 34	75 +/- 13	-	-	80 +/- 13	100	20 +/- 7
	All	21 (50)	60 +/- 7	29	-	86	14	0	196 +/- 33	175 +/- 32	73 +/- 11	-	-	78 +/- 15	93	21 +/- 7
Bristow 200360 COMPANION	CRT	413 (67)	65	54	-	87	13	0	-	159	-	-	-	-	69	22
	Control	213 (69)	67	59	-	82	18	0	-	156	-	-	-	-	70	23
	CRT-ICD	399 (67)	66	55	-	86	14	0	-	159	-	-	-	-	73	23
Cazeau 200174 MUSTIC-SR	CRT first	19 (66)	64 +/- 11	-	-	100	-	0	-	172 +/- 22	-	-	-	-	-	-
	Control first	24 (83)	64 +/- 8	-	-	100	-	0	-	175 +/- 19	-	-	-	-	-	-
	All	43 (74)	64 +/- 9	37	-	100	-	0	215 +/- 43	174 +/- 20	73 +/- 10	-	-	-	-	23 +/- 7
Cazeau 199699	All	7 (88)	66 +/- 5	50	-		100	3 (38)	200 +/- 20	200 +/- 35	-	-	-	-	25	22 +/- 8
Filho 200298	All	23 (96)	55 +/- 13	38	33	67	4	-	225 +/- 80	181 +/- 31	-	-	-	-	100	19 +/- 5.2
Gras 2002100 INSYNC Italian Registry	All	81 (77)	67 +/- 10	48	0	68	32	0	-	178 +/- 28	72 +/- 10	-	-	-	-	22 +/- 6
Guidant 200273 CONTAK-CD FDA report	CRT II–IV	210 (85)	66 +/- 11	67	32	60	8	0	205 +/- 42	160 +/- 27	-	118 +/- 21	67 +/- 12	73 +/- 12	54	21 +/- 7
	Control II–IV	211 (83)	66 +/- 11	70	33	57	10	0	202 +/- 49	156 +/- 26	-	118 +/- 21	69 +/- 12	75 +/- 14	55	22 +/- 7
	CRT III/IV	90 (77)	66 +/- 11	65	17*	73	10	0	204 +/- 41	164 +/- 27	-	116 +/- 20	68 +/- 12	75 +/- 13	50	21 +/-6
	Control III/IV	86 (78)	66 +/- 11	71	10*	71	19	0	200 +/- 54	152 +/- 24	-	117 +/- 23	67 +/- 14	74 +/-15	54	21 +/- 6
Krahn 2002101	All	37 (82)	65 +/- 10	69	6	76	18	15 (33)	240 +/- 63	166 +/- 20	-	-	-	-	98	19 +/- 5
Kuhlkamp 2002102	All	74 (91)	64 +/- 9	-	32	59	9	19 (23) PAFr 10 (12 AF	-	170 +/- 30	71 +/- 11	-	-	-	-	25 +/-7
Leclercq 200275 MUSTIC-AF	CRT first	21 (84)	65 +/- 9	-	-	100	-	25 (100)	-	209 +/- 21	70 +/- 9	-	-	75 +/- 6	-	23 +/- 7
	Control first	14 (78)	66 +/- 9	-	-	100	-	18 (100)	-	208 +/- 12	66 +/- 7	-	-	74 +/- 5	-	30 +/- 12
	All	35 (81)	65 +/- 8	43	-	100	-	43 (100)	-	209 +/- 18	68 +/- 8	-	-	74 +/- 5	-	26 +/- 10
Leclercq 2000107	All	34(92)	68 +/- 8	38	0	70	30	14 (28)	260 +/- 30	181 +/- 23	87 +/- 8	-	-	-	-	23 +/-5
Leclercq104 unpublished	All	110 (88)	68 +/- 9	38	-	68	32	36 (29)	248 +/- 56	183 +/- 29	74 +/- 8.5	-	-	-	-	21 +/- 6
Leclercq 2003 RD-CHF** Unpublished	-	N/a	73 +/- 8	N/a	0	N/a	N/a	23	N/a	N/a	N/a	N/a	N/a	N/a	N/a	25 +/- 9
Leon 2002106	All	17 (85)	70 +/- 11	55	0	60	40	20 (100)	-	213 +/- 40	68 +/- 8	-	-	-	-	22 +/- 7
Medtronic 200136 MIRACLE-ICD FDA report 2001	CRT II–IV	217 (80)	66 +/- 12	38	32	60	8	16 (6)	-	166 +/- 23	71 +/- 9	113 +/- 19	67 +/- 11	73 +/- 13	10	21 +/- 7
	Control II–IV	231 (82)	66 +/- 11	32	38	55	7	12 (4)	-	163 +/- 22	70 +/- 9	113 +/- 17	68 +/- 12	73 +/- 13	0	21 +/- 7
	CRT III/IV	142 (76)	67 +/- 11	64	-	88	12	49 (26)	-	165 +/- 22	70 +/- 9	-	-	71 +/- 12	75	21 +/- 7
	Control III/IV	136 (77)	68 +/- 9	75	-	89	11	31 (18)	-	162 +/ -22	71 +/- 9	-	-	72 +/- 13	71	20 +/- 6
Molhoek 2002103	All	31 (78)	64 +/- 10	48	mean = 3.3 +/- 0.5			4 (10)	-	range 120–240	-	-	-	-	100	24+/- 9

CRT = cardiac resynchronization therapy; FDA = Food and Drug administration; LBBB = left bundle branch block; LVEF = left ventricular ejection fraction; LVEDD = left ventricular end diastolic diameter; N/a = not available; NYHA = New York Heart association; SD = standard deviation

*

pre-implant assessment

**

detailed data not available for Leclercq 2003 (RD-CHF) at the time this report was prepared

Kuhlkamp¹⁰² required participants be survivors of a cardiac arrest and have sustained ventricular tachycardia; otherwise no studies specifically based inclusion on the etiology of their heart failure. Patients with correctable valvular disorders, hypertrophic or restrictive cardiomyopathy, unstable angina, or acute myocarditis were excluded. Four studies did not state exclusion criteria.^{99, 102, 103, 106} Ischemic etiology ranged from 29%⁶² to 69%.¹⁰¹ Other comorbidities such as diabetes mellitus or renal failure were reported in only one study.³⁶

Eleven studies limited inclusion to patients with an ejection fraction <35%, one restricted it to <30%,⁹⁸ one to <25%,⁶⁰ and 3 did not specify this criterion.^{62, 99, 101} The mean ejection fraction ranged from 19% to 26%. Nine studies specified a left ventricular end diastolic dimension inclusion criterion of ≥55 mm^{60, 61, 74, 75, 93, 96, 100, 102, 104} or ≥ 60 mm. The reported mean left ventricular end diastolic dimension for the trials ranged from 68 mm to 74 mm.

QRS width was a criterion for 13 studies, with six specifying ≥ 120 msec,^{60, 62, 73, 103, 107} four studies ≥ 130 msec,^{61, 93, 101, 102} two trials >150 msec,^{74, 100} and one > 200 msec.⁷⁵ Ten of the 16 studies reported a mean QRS between 158 msec and 180 msec, and six ranged from 180 msec to 213 msec.^{70, 75, 96, 99, 104, 106} In the nine studies that reported the presence of left bundle branch block, the mean was 68% (range 0 to 100%).

Table 11. Baseline drug therapy in included patients (more...)

Table 11. Baseline drug therapy in included patients in the safety review: CRT for CHF

		Medications
Trial name	Study Groups	ACE %	ARB %	BB %	Spironolactone %	Digoxin %	Nitrates %	Others Name %
Abraham 200261 MIRACLE	CRT	93*	-	62	-	78	-	Diuretics (94)
	Control	90*	-	55	-	79	-	Diuretics (93)
Auricchio 200262 PATH-CHF	CRT	96**	-	71	-	**	71	Amiodarone (29)
	Control	100**	-	65	-	**	71	Amiodarone (35)
	All	95**	-	67	-	**	69	Amiodarone (30)
Bristow 200360 COMPANION	CRT	89*	*	68	53	-	-	Diuretics (100); digoxin (100)
	Control	89*	*	66	55	-	-
	CRT-D	90*	*	68	55	-	-
Cazeau 200174 MUSTIC-SR	All	96*	*	28	22	48	-	Amiodarone (31), diuretics (100)
Cazeau 199699	All	100	-	-	-	-	-	All on maximal medical therapy; 4 on IV dobutamine &/or dopamine
Filho 200298	All	92	-	-	42	100	13**	Diuretics (100); carvedilol (29); amiodarone (50);
Gras 2002100 INSYNC Italian Registry	All	70	16	17	-	58	-	Amiodarone (50); diuretic (93); vasodilators (17); CCB (15) anticoagulants (34); IV inotropic support (2)
Guidant 200273 CONTAK-CD FDA report	CRT, II–IV	86*	*	48	-	69	-	Diuretics (88)
	Control, II–IV	89*	*	46	-	68	-	Diuretics (83)
	CRT, III/IV	81*	*	45	-	72	-	Diuretics (92)
	Control, III/IV	89*	*	40	-	68	-	Diuretics (86)
Krahn 2002101	All	84	-	56	-	-	-	Antiarrhythmic agents (47); diuretic (100)
Kuhlkamp 2002102	All	85	-	54	-	32	31	Diuretics (54); anticoagulants (56); antiarrhythmics (58)
Leclercq 200275 MUSTIC-AF	All	100*	*	23	16	58	-	-
Leclercq 200096	All	100	-	-	-	60	-	Diuretics (100), captopril (95 +/- 30mg/d)
Leclercq104 unpublished	All	98	-	34	-	60	-	Diuretics (100),
Leclercq 2003 RD-CHF*** Unpublished	N/a	N/a	N/a	N/a	N/a	N/a	N/a	N/a
Leon 2002106	All	90	-	25	-	60	-	Diuretics (95)
Medtronic 200136 MIRACLE-ICD FDA report	CRT, II–IV	93	-	63	-	71	31	Diuretics (91); anti-depressant (16); CCB (6); antiarrhythmic (40); positive ionotrope (71); anti-coagulant (77).
	Control, II–IV	90	-	59	-	72	30	Diuretics (90); anti-depressant (17); CCB (6); anti-coagulant (79); antiarrhythmic (33); positive ionotrope (72)
	CRT, III/IV	92	-	63	-	76	36	Diuretics (93); anti-depressant (19); CCB (7); anti-coagulant (77); antiarrhythmic (42); positive ionotrope (76)
	Control, III/IV	88	32	58	-	73	33	Diuretics (94); anti-depressant (20); CCB (6); anti-coagulant (81); antiarrhythmic (32); positive ionotrope (73)
Molhoek 2002103	All	Not stated	-	-	-	-	-	-

ACE= angiotensin-converting-enzyme inhibitors; ARB= angiotension-receptor blocker; BB= beta blocker; CCB= calcium channel blocker

*

ACE or ARB

**

receiving nitrates or hydralazine

***

detailed data not available for Leclercq 2003 (RD-CHF) at the time this report was prepared

Quality of Included Studies

Overall, the studies were rated as having ‘good’ quality on the Downs and Black⁵⁹ scoring system. Eight studies were described as randomized (three parallel trials,^{36, 95, 108} five crossover trials,^{62, 73, 74, 96, 98} and the remaining eight were prospective cohort studies.^{99, 96, 100–104, 106} Reporting was generally good with 8 of 16 rating 11/11, the rest ranging from 7 to 10. External validity assessment posed some problems because authors did not report the source population for patients or the proportion of eligible patients selected for inclusion, nor compare the distribution of main confounding factors with the source population. (For this review we defined the source population as those with symptomatic CHF.) Since this procedure can only be performed in specialist centers, we determined that all facilities were representative. Internal validity concerning assessment of bias ranged from 5/7 to 7/7; the lack of blinding was the main shortfall. Four studies were double-blind,^{36, 61, 62, 73} two were single-blind (i.e., the patient),^{74, 75} and the remaining trials were open-label. Internal validity assessments concerning confounding ranged between 1/6 to 6/6, with only two studies receiving 2 or less.^{96, 100} Many authors did not state the period of time over which patients were recruited, the source of patients, or how the randomization code was generated and concealed. Withdrawals and dropouts were clearly described in all studies except COMPANION.⁶⁰ Four studies included a power calculation,^{60, 61, 74, 75} and nine had sufficient sample sizes to determine a clinically important effect.^{36, 60–62, 73–75, 98, 104}

Table 12. Methodological quality assessments of included (more...)

Table 12. Methodological quality assessments of included studies: safety review

Study name	Sponsor	Downs and Black Quality Score59
		Reporting Maximum 11	External validity Maximum 3	Internal validity [bias] maximum 7	Internal validity [confounding] Maximum 6	Power Maximum 2	Overall Maximum 29
Abraham 2002 MIRACLE61	Medtronic Inc.	11	2	7	6	2	28
Auricchio 200262 PATH-CHF	Guidant Corporation	11	2	7	5	1	26
Bristow 200360 COMPANION	Guidant Corporation	8	1	5	4	2	20
Cazeau 200174 MUSTIC-SR	ELA Recherche, Medtronic, Swedish Heart and Lung Association; Swedish Medical Research Council	11	1	6	5	2	25
Cazeau 199699	technical support from ELA Recherche	9	1	5	4	0	19
Filho 200298	None indicated	7	1	5	4	1	18
Gras 2002100 INSYNC Italian Registry	None indicated	9	1	5	1	1	17
Guidant 200173 CONTAK-CD FDA report	Guidant Corporation	11	2	7	4	1	25
Krahn 2002101	Heart and Stroke Foundation of Ontario	10	3	5	4	0	22
Kuhlkamp 2002102	Medtronic Inc.	11	1	5	3	0	20
Leclercq 200275 MUSTIC-AF	ELA Recherche, Medtronic, European Society of Cardiology, Centre Hospitalier Universitaire de Rennes; Swedish Heart and Lung Association, Swedish Medical Research Council	11	2	6	5	2	26
Leclercq 200096	None indicated	10	2	5	2	0	19
Leclercq Unpublished104	None indicated	10	2	5	3	1	21
Leclercq 2003 RD-CHF* Unpublished	N/a	N/a	N/a	N/a	N/a	N/a	N/a
Leon 2002106	Medtronic provided fellowship support	11	3	5	3	0	22
Medtronic 200136 MIRACLE-ICD FDA report	Medtronic Inc	11	1	7	3	1	23
Molhoek 2002103	Medtronic Inc.	10	2	5	3	0	20

N/a = not available

*

detailed data not available for Leclercq 2003 (RD-CHF) at the time report prepared

Efficacy Review

The nine trials included 2755 patients with NYHA Class III or IV CHF, and 461 patients with NYHA Class II symptoms.

Figure 4. Meta-graph of all-cause mortality

   Figure 4. Meta-graph of all-cause mortality

Figure 4

Figure 5. Time to all-cause death

   Figure 5. Time to all-cause death

Note: The COMPANION Trial consisted of three arms. To avoid “double counting” in calculating summary estimates of treatment effect, we divided the 58 deaths in 308 control patients into 29 deaths in 154 controls in each comparison.

Figure 5

Figure 6. Meta-graph of death due to congestive heart (more...)

Click on image to enlarge 

   Figure 6. Meta-graph of death due to congestive heart failure

Figure 6

Figure 7. Meta-graph of sudden cardiac death

Click on image to enlarge 

   Figure 7. Meta-graph of sudden cardiac death

Figure 7

Figure 8. Meta-graph of non-cardiac death

Click on image to enlarge 

   Figure 8. Meta-graph of non-cardiac death

Figure 8

Morbidity and functional outcomes

Figure 9. Meta-graph of hospitalizations due to congestive (more...)

Click on image to enlarge 

   Figure 9. Meta-graph of hospitalizations due to congestive heart failure

Heart failure hospitalizations. Pooled results from the six trials that reported heart failure hospitalizations showed a relative risk of 0.68 (95% CI 0.41 to 1.12; Figure 9 end of chapter) in favor of CRT compared to control. This result was heterogeneous (p=0.01, I-squared=65%) and was also insignificant in the fixed effects analysis (RR 0.80, 95% CI 0.64 to 1.003). Restricting the analysis to patients with more advanced heart failure (those with NYHA Class III or IV symptoms) revealed significant reductions (RR 0.65, 95% CI 0.48 to 0.88; NNT=12) and was statistically homogeneous (p=0.31, I-squared=16%).

Figure 10. Meta-graph of six-minute walk test

Click on image to enlarge 

   Figure 10. Meta-graph of six-minute walk test

Six-minute walk test. CRT was associated with an improved six-minute walk test distance, with a weighted-mean-difference of 23 meters in favor of CRT (95% CI 9 m to 38 m, Figure 10 end of chapter). This improvement was similar in those patients with NYHA Class III or IV symptoms (WMD 26 m, 95% CI 11 m to 41 m). However, this result was heterogeneous (p=0.06, I-squared=50%) in part due to one trial⁷⁴ in which the control group worsened by 24 meters; the control groups in all other trials that reported baseline results showed an improvement. Although the data from the RD-CHF Trial were not available for pooling, the RD-CHF investigators reported statistically significant improvements in six-minute walk test distances with CRT (personal communication, Dr. Christophe Leclercq, November 2003).

Figure 11. Meta-graph of changes in NYHA status

Click on image to enlarge 

   Figure 11. Meta-graph of changes in NYHA status

New York Heart Association Functional Class. We were able to meta-analyze data for NYHA Class from three studies (although five studies reported baseline and endpoint NYHA Class). Combining these three studies showed improvements in NYHA Class in 57% of CRT patients and 34% of controls (CRT was associated with a 1.63 times increased chance of improving at least one NYHA Class (95% CI 1.05 to 2.52), Figure 11). This result was statistically heterogeneous (p=0.04, I-squared=70%). In patients with NYHA Class III or IV symptoms, the relative risk improved slightly to 1.68 (95% CI 1.25 to 2.27). The data from MIRACLE ICD⁶¹ was not reported in a format that permitted pooling with the other 3 trials; however, the baseline median NYHA Class for both groups was III and the endpoint median was II in the CRT group and III in the control group. This improvement in NYHA Class was significant (p=0.01) and favored CRT; the specific statistical test used was not reported. In contrast, PATH CHF⁶² (which could also not be combined with the other trials due to the manner in which the data were reported) did not find a significant difference (p=0.36; Wilcox on matched-pairs test), although both groups showed significant improvement from baseline. Moreover, while the data from the RD-CHF Trial could not be pooled with the other trials, the RD-CHF investigators also reported statistically significant improvements in NYHA Class with CRT (personal communication, Dr. Christophe Leclercq, November 2003).

Figure 12. Meta-graph of quality of life (based on (more...)

Click on image to enlarge 

   Figure 12. Meta-graph of quality of life (based on Minnesota Living with Heart Failure instrument)

Quality of life. Quality of life was measured by the Minnesota Living With Heart Failure Instrument^109–111 for six of the eight trials; pooled results showed a significant improvement in favor of CRT (weighted mean difference -5.5 points (95% CI -9 to -2 points; Figure 12). This result was highly statistically heterogeneous (p=0.008, I-squared=68%); however results were consistent in direction. Restricting the analysis to only those patients with NYHA Class III or IV symptoms increased the difference between the CRT and control groups (WMD -6.4 points, 95% CI -9.4 to -3.4 points), but the results remained significantly heterogeneous (p=0.07, I-squared=50%). These differences are clinically significant since the minimal clinically important difference for the Minnesota Living with Heart Failure Questionnaire has been established to be 5 points.^109–111 Further, although the use of a different scale prevented pooling with the other trials, the RD-CHF Investigators reported statistically significant improvements in quality of life with CRT (personal communication, Dr. Christophe Leclercq, November 2003).

Other outcomes. Peak oxygen consumption significantly improved in the CRT patients compared to the control patients (1.05 ml/kg/min versus 0.39 ml/kg/min; WMD 0.65 ml/kg/min, 95% CI 0.27 to 1.04 ml/kg/min). Ejection fraction also significantly improved in the CRT arm compared to the control arm (4.17% versus 0.82%; WMD 3.35%, 95% CI 1.22 to 5.48%). The QRS interval also showed greater improvement in the CRT group (-30 msec versus -2 msec for controls; WMD -28, 95% CI -47 msec to -9 msec). Other echocardiographic indices including mitral regurgitation jet area were not significantly different between groups; there were insufficient data that could be pooled for left ventricular end-diastolic or end-systolic diameters.

Sensitivity analyses. Many a priori subgroup and sensitivity analyses (including examining any interactions between the effects of CRT in patients with different etiologies of heart failure, or by ethnic background, gender, age, comorbidities, and baseline medication use) could not be performed due to the paucity of subgroup data in the trial reports and our inability to obtain individual patient-level data from each of the trial lists despite our requests.

Figure 13. Meta-graph of all-cause mortality by cardioverter (more...)

Click on image to enlarge 

   Figure 13. Meta-graph of all-cause mortality by cardioverter defibrillator vs non-cardioverter defibrillator

Post-hoc, we explored the impact of ICDs on the efficacy of CRT using meta-regression (a between-study non-randomized comparison). When the data were pooled for all-cause mortality from the two trials that included an ICD in both the experimental and control arms (i.e., CRT+ICD versus medical therapy +ICD),^{36, 73} the relative risk was 0.84 (95% CI 0.54 to 1.28). On the other hand, pooled data from the other five trials comparing CRT versus medical therapy demonstrated a relative risk of 0.78 (95% CI 0.56 to 1.08) with CRT (Figure 13).^{61, 62, 74, 75} This difference was not statistically significant (p=0.80), supporting the assertion that the benefits of CRT on all-cause mortality are not appreciably altered by addition of an ICD. Using the same meta-regression model, secondary outcomes (including heart failure hospitalizations, six-minute walk test, quality of life, and NYHA improvements) were not significantly different in patients with, or without an ICD, in addition to their biventricular pacemaker. Of note, while the data from COMPANION⁶⁰ could not be used in the meta-regression for ICDs (since none of the arms in COMPANION⁶⁰ consisted of ICD alone), the COMPANION⁶⁰ data does provide the only direct comparison between CRT plus ICD versus CRT alone. This analysis did approach statistical significance (p=0.07) in favor of the CRT plus ICD group for all-cause mortality; however, the reductions in heart failure hospitalizations were similar in CRT-treated patients with/without ICDs. Until the detailed data from the COMPANION⁶⁰ sub-analyses are made available, the most conservative conclusion that can be made at this stage is that the benefits of CRT are similar with or without ICDs.

Post-hoc we also explored the relationship between the baseline use of beta-blockers and/or digoxin with the impact of CRT on all-cause mortality. Both meta-regressions were non-significant (p=0.37 and p=0.31, respectively), suggesting that the benefits of CRT are not modified by use of these medications. However, and as expected, the linear trends showed improved survival with greater use of beta-blockers.

Publication bias could not be quantitatively measured since all of the smaller studies had shorter followup times, hence confounding results. In a post-hoc subgroup analysis examining published versus unpublished studies, there was no appreciable difference between effect estimates for all-cause mortality (p=0.64).

Fixed effects estimates did not substantially alter the results of our meta-analysis (although the confidence intervals were naturally narrower and, as a result, the analysis for progressive heart failure mortality (RR 0.59, 95% CI 0.35 to 0.98) reached statistical significance).

Safety Review

Table 13. Peri-implant risks for cardiac resynchronization (more...)

Table 13. Peri-implant risks for cardiac resynchronization therapy

Study	n/N	Simple Pool
		Risk % [95% CI]
Peri-implant deaths
MUSTIC-SR 200174	0/64	0
MUSTIC-AF 200275	0/59	0
PATH-CHF 200262	0/41	0
MIRACLE 200261	2/571	0.4 [0.1,1.4]
MIRACLE-ICD 200394	0/429	0
COMPANION unpub60	5/617	0.8 [0.3,2.0]
COMPANION-ICD unpub60	3/595	0.5 [0.1,1.6]
Cazeau 199699	1/7	14.3 [8.4,54.0]
Filho 200298	0/24	0
Leclercq unpub104	0/139	0
Total [N=10]	13/3113	0.4 [0.2,0.7]
Sensitivity Total [N=16]	13/3456	0.4 [0.2,0.7]
Implant successes
MUSTIC-SR 200174	59/64	92.2 [82.0,97.1]
MUSTIC-AF 200275	54/59	91.5 [80.6,96.8]
MIRACLE 200261	528/571	92.5 [89.9,96.8]
PATH-CHF 200262	41/41	100
MIRACLE-ICD 200394	379/429	88.3 [84.8,91.1]
COMPANION unpub60	538/617	87.2 [84.2,89.7]
COMPANION-ICD unpub60	540/595	90.8 [88.1,92.9]
CONTAK-CD unpub73	501/567	88.4 [85.4,90.8]
RD-CHF unpub#	45/56	80.4 [67.2,89.3]
Cazeau 199699	6/7	85.7 [42.0,99.2]
Gras 2002100	125/139	89.9 [83.4,94.2]
Krahn 2002101	40/45	88.9 [75.2,95.8]
Kuhlkamp 2002102	81/84	96.4 [89.2,99.1]
Leon 2002106	20/20	100
Filho 200298	24/24	100
Molhoek 2002103	40/40	100
Leclercq unpub104	125/139	89.9 [83.4,94.2]
Total [N=16]*	3124/3475	89.9 [88.8,90.9]
Related to left ventricular lead
MUSTIC-SR 200174	5/64	7.8 [2.9,18.0]
MUSTIC-AF 200275	5/59	8.5 [3.2,19.4]
MIRACLE-ICD 200336	31/429	7.2 [5.0,10.2]
CONTAK-CD unpub73	17/517	3.3 [2.0,5.3]
Gras 2002100	12/117	10.3 [5.6,17.6]
Krahn 2002101	6/45	13.3 [5.5,27.5]
Kuhlkamp 2002102	4/84	4.8 [1.5,12.4]
Filho 200298	0/24	0
Total [N=8]	80/1339	6.0 [4.7,7.2]
Sensitivity Total [N=14]**	80/2055	3.9 [3.1,4.8]
Related to the device and battery
PATH-CHF 200262	3/41	7.3 [1.9,21.0]
MIRACLE 200261	2/571	0.4 [0.06,1.4]
MIRACLE-ICD 200394	3/429	0.7 [0.2,2.2]
CONTAK-CD unpub73	7/517	1.4 [0.6,2.9]
Filho 200298	0/24	0
Total [N=5]	15/1582	0.9 [0.6,1.6]
Sensitivity Total [N=14]**	15/2055	0.7 [0.4,1.2]
Related to implant procedure and/or tools
MIRACLE 200261	35/571	6.1 [4.4,8.5]
MIRACLE-ICD 200394	19/429	4.4 [2.8,7.0]
CONTAK-CD unpub73	50/517	9.7 [7.3,12.6]
Gras 2002100	3/117	2.6 [0.7,7.9]
Krahn 2002101	4/45	8.9 [2.9,22.1]
Kuhlkamp 2002102	4/84	4.8 [1.5,12.4]
Leon 2002106	0/20	0
Filho 200298	0/24	0
Total [N=8]	115/1807	6.4 [5.3,7.6]
Sensitivity Total [N=14]**	115/2055	5.6 [4.7,6.7]
Related to heart function
MIRACLE 200261	1/571	0.2 [0.009,1.1]
MIRACLE-ICD 200394	11/429	2.6 [1.4,4.7]
Cazeau 199699	1/7	14.3 [8.4,54.0]
Filho 200298	0/24	0
Total [N=4]	13/1031	1.3 [0.7,2.2]
Sensitivity Total [N=14]**	13/2055	0.6 [0.4,1.1]

*

only safety outcome that includes data from RD-CHF

**

COMPANION omitted - full report not available; Leclercq 2002 omitted, - there was 10% or 14/139 failed implants but they did not specify the event

Table 14. Post-implantation risks of cardiac resynchronization (more...)

Table 14. Post-implantation risks of cardiac resynchronization therapy

Study	n/N	Simple Pool
		Risk % [95% CI]
Mechanical malfunction
MUSTIC-SR 200174	2/58	3.4 [0.6,13.0]
MUSTIC-AF 200275	2/54	3.7 [0.6,13.8]
CONTAK-CD unpub73	22/448	4.9 [3.2,7.5]
MIRACLE-ICD unpub36	25/364	6.9 [4.6,10.1]
Leclercq 200096	3/37	8.1 [2.1,23.0]
Gras 2002100	4/103	3.9 [1.3, 10.2]
Kuhlkamp 2002102	1/84	1.2 [0.1,7.4]
Filho 200298	3/24	12.5 [3.3,33.5]
Leclercq unpub104	25/125	20.0 [13.6,28.3]
Total [N=9]	87/1297	6.7 [5.4,8.2]
Sensitivity Total [N=15]*	87/1968	4.4 [3.6,5.4]
Lead dislodgement
MUSTIC-SR 200174	8/58	13.8 [6.6,25.9]
MUSTIC-AF 200275	5/54	9.3 [3.5,21.1]
MIRACLE 200261	30/524	5.7 [4.0,8.2]
CONTAK-CD unpub73	31/448	6.9 [4.8,9.8]
MIRACLE-ICD unpub36	46/364	12.6 [9.5,16.6]
Cazeau 199699	2/6	33.3 [6.0,75.9]
Leclercq 200096	2/37	5.4 [0.9,19.5]
Gras 2002100	10/103	9.7 [5.0,17.5]
Krahn 2002101	4/40	10.0 [3.3,24.6]
Kuhlkamp 2002102	7/84	8.3 [3.7,17.0]
Filho 200298	0/24	0
Molhoek 2002103	3/40	7.5 [2.0,21.5]
Leclercq unpub104	15/125	12.0 [7.1,19.3]
Total [N=13]	163/1907	8.5 [7.4,9.9]
Sensitivity Total [N=15]*	163/1968	8.3 [7.1,9.6]
Infection
MIRACLE 200261	7/524	1.3 [0.6,2.9]
MIRACLE-ICD unpub36	2/364	0.5 [0.1,2.2]
Gras 2002100	2/103	1.9 [0.3,7.5]
Kuhlkamp 2002102	2/84	2.4 [0.4,9.1]
Filho 200298	1/24	4.2 [0.2,23.1]
Leclercq unpub104	3/125	2.4 [0.6,7.4]
Total [N=6]	17/1224	1.4 [0.8,2.3]
Sensitivity Total [N=15]*	17/1968	0.9 [0.5,1.4]
Arrhythmias associated with CRT
MUSTIC-AF 200275	1/54	1.9 [0.1,11.2]
PATH-CHF 200262	4/41	9.8 [3.2,24.1]
MIRACLE-ICD unpub36	3/364	0.8 [0.2,2.6]
Filho 200298	0/24	0
Molhoek 2002103	1/40	2.5 [0.1,14.7]
Total [N=5]	9/523	1.7 [0.8,3.4]
Sensitivity Total [N=15]*	9/1968	0.5 [0.2,0.9]
Pain
Krahn 2002101	1/40	2.5 [0.1,14.7]
Filho 200298	3/24	12.5 [3.3,33.5]
Total [N=2]	4/64	6.3 [2.0,16.0]
Sensitivity Total [N=15]*	4/1968	0.2 [0.07,0.6]

*

COMPANION omitted - full report not available

Problems with implantation of the left-ventricular lead were encountered in 6.0% (95% CI 4.7% to 7.2%) of cases. Assuming that any studies failing to report any implantation problems had zero occurrences in sensitivity analysis led to a reduced estimate of 3.9% (95% CI 3.1% to 4.8%). The device or battery was problematic in 0.9% (95% CI 0.6% to 1.6%) of cases (0.7%, 95% CI 0.4% to 1.2% with sensitivity analysis), and the procedure or equipment used for the procedure was reported to be a problem in 6.4% (95% CI 5.3% to 7.6%) of cases attempted (reduced to 5.6%, 95% CI 4.7% to 6.7% in sensitivity analysis). Further detailed information on the type of equipment failure was not uniformly available, so pooling of results was not possible. Some of the specific problems in this category included lead fracture, loss of capture, inappropriate sensing, and extra-cardiac stimulation.

Effectiveness of Medical Therapy

Expanding on the results from our systematic review outlined above, annual event rates were calculated (median followup time was 22.8 weeks in these trials). The annual risk of death was mean 24.3% (95% CI 20.0% to 29.2%) and the annual risk of heart failure hospitalization was 56.0% (95% CI: 47.6% to 66.2%).