Previous analyses have shown that there is lower mortality with cardiac resynchronization therapy defibrillators (CRT-D) in patients with left bundle branch block (LBBB) but demonstrated mixed results in patients without LBBB. We evaluated the comparative effectiveness of CRT-D versus standard implantable defibrillators (ICDs) separately in patients with LBBB and right bundle branch block (RBBB) using Medicare claims data. Medicare records from CRT-D and ICD recipients from 2002 to 2009 that were followed up for up to 48 months were analyzed. We used propensity scores to match patients with ICD to those with CRT-D. In LBBB, 1:1 matching with replacement resulted in 54,218 patients with CRT-D and 20,763 with ICD, and in RBBB, 1:1 matching resulted in 7,298 patients with CRT-D and 7,298 with ICD. In LBBB, CRT-D had a 12% lower risk of heart failure hospitalization or death (hazard ratio [HR] 0.88, 95% confidence interval 0.86 to 0.90) and 5% lower death risk (HR 0.95, 0.92 to 0.97) compared with ICD. In RBBB, CRT-D had a 15% higher risk of heart failure hospitalization or death (HR 1.15, 1.10 to 1.20) and 13% higher death risk (HR 1.13, 1.07 to 1.18). Sensitivity analysis revealed that accounting for covariates not captured in the Medicare database may lead to increased benefit with CRT-D in LBBB and no difference in RBBB. In conclusion, in a large Medicare population, CRT-D was associated with lower mortality in LBBB but higher mortality in RBBB. The absence of certain covariates, in particular those that determine treatment selection, may affect the results of comparative effectiveness studies using claims data.
Cardiac resynchronization therapy (CRT), either alone or in combination with an implantable cardioverter defibrillator (ICD, CRT-D), is an increasingly used therapy for heart failure and has been shown to reduce heart failure symptoms, heart failure hospitalizations, and mortality while improving quality of life. Current professional society guidelines for CRT give a class I indication to patients with left bundle branch block (LBBB) and a QRS duration ≥150 ms, whereas patients without LBBB receive either a class IIa or class IIb recommendation. Recent analyses from pre-market clinical trials and post-market registry data demonstrated that there, indeed, is a significantly lower mortality with CRT-D in patients with LBBB but that there is no reduced mortality in patients with solely right bundle branch block (RBBB). Although registry data are very useful for answering questions that cannot be answered with pre-market data alone, in some situations, registry data may not be available. In these cases, claims data may then be useful to assess long-term outcomes in large real-world populations. We evaluated the comparative effectiveness of CRT-D versus ICD separately in patients with LBBB and RBBB using Medicare claims data.
Methods
This study was approved by the US Food and Drug Administration (USFDA) Research in Human Subjects Committee and the Centers for Medicare and Medicaid Services (CMS). It included all Medicare patients who received a primary prevention CRT-D ( International Classification of Diseases, Ninth Revision, Clinical Modification [ ICD-9-CM ] procedure code “00.51”) or ICD device ( ICD-9-CM procedure code “37.94”) from July 1, 2002, to September 30, 2009, who were also continuously enrolled in Medicare Part A (inpatient hospital coverage) and B (outpatient medical coverage) ≥12 months before implantation. The following patients were excluded from further analysis: patients with an ICD-9 code for ventricular fibrillation, ventricular flutter, cardiac arrest, or sudden cardiac arrest as part of secondary prevention and patients with end-stage renal disease or hypertrophic cardiomyopathy ( Supplementary Methods ).
The presence of preexisting co-morbidities and other covariates ( Table 1 ) was assessed through ICD-9 codes in a 12-month look-back window, whereas demographics were determined using the Medicare enrollment database. To account for other competing factors for death, we also included the Charlson co-morbidity score. The Charlson score is a score to predict 10-year mortality based on whether a patient has certain health conditions ; further explanation can be found in the Supplementary Methods .
| Matched Left Bundle Branch Block | Matched Right Bundle Branch Block | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| CRT-D | ICD | SMD | CRT-D | ICD | SMD | |||||
| N | % | N | % | N | % | N | % | |||
| Variable | 54,218 | 20,763 | 7,298 | 7,298 | ||||||
| Male | 36,357 | 67% | 13,877 | 67% | 0.01 | 6,218 | 85% | 6,217 | 85% | 0.00 |
| Age (years) | ||||||||||
| 0-64 | 5,625 | 10% | 2,181 | 11% | 0.00 | 758 | 10% | 771 | 11% | 0.01 |
| 65-69 | 8,625 | 16% | 3,348 | 16% | 0.01 | 1,123 | 15% | 1,129 | 16% | 0.00 |
| 70-74 | 12,319 | 23% | 4,733 | 23% | 0.00 | 1,634 | 22% | 1,628 | 22% | 0.00 |
| 75-79 | 13,915 | 26% | 5,259 | 25% | 0.01 | 1,901 | 26% | 1,872 | 26% | 0.01 |
| 80-84 | 10,065 | 19% | 3,846 | 19% | 0.00 | 1,401 | 19% | 1,436 | 20% | 0.01 |
| 85+ | 3,669 | 7% | 1,393 | 7% | 0.00 | 481 | 7% | 462 | 6% | 0.01 |
| Black | 4,418 | 8% | 1,755 | 9% | 0.01 | 591 | 8% | 594 | 8% | 0.00 |
| Other | 1,616 | 3% | 620 | 3% | 0.00 | 250 | 3% | 246 | 3% | 0.00 |
| White | 48,184 | 89% | 18,386 | 89% | 0.01 | 6,457 | 89% | 6,458 | 89% | 0.00 |
| Charlson Score 0 | 639 | 1% | 212 | 1% | 0.02 | 70 | 1% | 63 | 1% | 0.00 |
| Charlson Score 1 | 5,931 | 11% | 2,306 | 11% | 0.01 | 537 | 7% | 517 | 7% | 0.01 |
| Charlson Score 2 | 9,649 | 18% | 3,756 | 18% | 0.01 | 1,044 | 14% | 1,025 | 14% | 0.01 |
| Charlson Score 3 | 10,625 | 20% | 4,027 | 19% | 0.01 | 1,397 | 19% | 1,410 | 19% | 0.00 |
| Charlson Score 4+ | 27,374 | 51% | 10,461 | 50% | 0.00 | 4,250 | 58% | 4,283 | 59% | 0.01 |
| Diabetes Mellitus | 25,482 | 47% | 9,745 | 47% | 0.00 | 3,776 | 52% | 3,791 | 52% | 0.00 |
| Mitral or Aortic Valve Disorder | 31,642 | 58% | 12,047 | 58% | 0.01 | 4,217 | 58% | 4,181 | 57% | 0.01 |
| Hypertension | 47,407 | 87% | 18,144 | 87% | 0.00 | 6,540 | 90% | 6,523 | 89% | 0.01 |
| Myocardial Infarction | 21,299 | 39% | 8,171 | 39% | 0.00 | 3,871 | 53% | 3,875 | 53% | 0.00 |
| Coronary Heart Disease | 32,445 | 60% | 12,449 | 60% | 0.00 | 4,979 | 68% | 4,962 | 68% | 0.00 |
| Tricuspid or Pulmonary Valve Disorder | 8,734 | 16% | 3,408 | 16% | 0.01 | 1,273 | 17% | 1,273 | 17% | 0.00 |
| Atrial Fibrillation | 24,630 | 45% | 9,299 | 45% | 0.01 | 3,691 | 51% | 3,664 | 50% | 0.01 |
| Prior Heart Failure Hospitalization | 15,826 | 29% | 6,024 | 29% | 0.00 | 2,013 | 28% | 1,934 | 27% | 0.02 |
| Prior Stroke | 2,724 | 5% | 1,063 | 5% | 0.00 | 462 | 6% | 462 | 6% | 0.00 |
| Peripheral Vascular disease | 11,965 | 22% | 4,652 | 22% | 0.01 | 2,000 | 27% | 2,000 | 27% | 0.00 |
| Ventricular Tachycardia | 19,236 | 36% | 7,348 | 35% | 0.00 | 3,310 | 45% | 3,283 | 45% | 0.01 |
| Percutaneous Coronary Intervention | 1,888 | 4% | 686 | 3% | 0.01 | 416 | 6% | 411 | 6% | 0.00 |
| Coronary Bypass | 5,047 | 9% | 1,921 | 9% | 0.00 | 971 | 13% | 975 | 13% | 0.00 |
To further specify the cohort to only include incident patients with CRT-D and ICD with LBBB or RBBB and a primary prevention implant, we used ICD-9 procedure codes and current procedural terminology codes used for maintenance and follow-up of patients with CRT-D and ICD devices (patients without an ICD-9 code for LBBB or RBBB were excluded). Patients were excluded if they either had a maintenance code or a cohort-defining event (CRT-D or ICD implantation code) in the 12 months before implantation, indicating a previous device implant. All codes used for cohort determination and the assessment of preexisting comorbidities can be found in the Supplementary Appendix .
The defined end points for this study were (1) heart failure hospitalization or death and (2) all-cause mortality. All-cause mortality was determined using the Medicare Master Beneficiary Summary File from CMS, which documents the date of death for all enrolled patients assessed from the Social Security Administration. Heart failure hospitalization was defined as having an ICD-9 code for heart failure as the primary diagnosis on an inpatient claim. Patients were censored if they did not reach the end point after a maximum of 48 months of follow-up or if they were no longer continuously enrolled in Medicare Part B.
Propensity score matching was performed separately in patients with LBBB and RBBB to reduce potential treatment selection bias and differences in baseline characteristics between patients with CRT-D and ICD. A multivariable logistic regression model including all variables listed in Table 1 was used to calculate the propensity score. Patients with LBBB ICD were 1:1 matched (with replacement) to those with LBBB CRT-D, and patients with RBBB ICD were 1:1 matched to those with RBBB CRT-D using a 0.1 caliper width. Standardized mean differences (SMD) were calculated to ensure that there were no significant differences between patients with CRT-D and ICD. Treatment groups were considered balanced when the standardized mean difference was <0.10 as this has been taken to indicate a negligible difference in the mean or prevalence of a covariate between treatment groups. Kaplan-Meier curves and multivariable Cox proportional hazards models were generated to compare the effect of CRT-D to ICD in patients with both LBBB and RBBB. We also performed sensitivity analyses for both end points to investigate the potential influence of unmeasured variables because, contrary to registry databases, important cardiac confounders, such as New York Heart Association (NYHA) heart failure class, ejection fraction, and QRS duration, are not available in Medicare claims data ( Supplementary Appendix ). Statistical analyses were performed using SAS Statistical Software (version 9.3; SAS Institute, Cary, North Carolina) and STATA (version 11; StataCorp, College Station, Texas). All reported 95% confidence intervals (CIs) and p values are 2 sided.
Results
After propensity score matching, the cohort consisted of 54,218 patients with CRT-D and 20,763 with ICD in LBBB and 7,298 patients with CRT-D and 7,298 with ICD in RBBB. There were no significant differences between CRT-D and ICD treatment groups in LBBB or RBBB, indicating that they were properly balanced ( Table 1 ). However, there were some differences between the matched LBBB and RBBB populations. Patients with RBBB more often were men, had diabetes, previous myocardial infarction, coronary heart disease, atrial fibrillation, peripheral vascular disease, ventricular tachycardia, a higher Charlson score, a higher percutaneous coronary intervention, and coronary bypass procedure rate compared with those with LBBB. Follow-up data were available for a median of 32 months in LBBB and 24 months in RBBB.
The rate of heart failure hospitalization or death was lower in patients with LBBB compared with those with RBBB (51.2% vs 58.6%). In LBBB, unadjusted heart failure hospitalization or death was lower for CRT-D than ICD (50.1% vs 54.0%), whereas this relation was reversed in patients with RBBB (60.8% vs 56.4%) ( Figure 1 ).
Multivariable-adjusted Cox proportional hazards models were determined separately in patients with LBBB and RBBB. In LBBB, patients with CRT-D had a 12% lower risk for heart failure hospitalization or death (hazard ratio [HR] 0.88, 95% CI 0.86 to 0.90) compared with ICD, whereas in RBBB, patients with CRT-D had a 15% higher risk (HR = 1.15, 1.10 to 1.20) than those with ICD ( Table 2 ).
| Covariates | Heart Failure Hospitalization or Death | All-Cause Mortality | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Left Bundle Branch Block | Right Bundle Branch Block | Left Bundle Branch Block | Right Bundle Branch Block | |||||||||
| Hazard Ratio | 95% CI | P-Value | Hazard Ratio | 95% CI | P-Value | Hazard Ratio | 95% CI | P-Value | Hazard Ratio | 95% CI | P-Value | |
| CRT-D | 0.88 | 0.86-0.90 | <0.001 | 1.15 | 1.10-1.20 | <0.001 | 0.95 | 0.92-0.97 | <0.001 | 1.13 | 1.07-1.18 | <0.001 |
| ICD | Ref | Ref | Ref | Ref | ||||||||
| Male | 1.11 | 1.08-1.13 | <0.001 | 0.91 | 0.86-0.97 | 0.002 | 1.27 | 1.24-1.31 | <0.001 | 1.02 | 0.95-1.09 | 0.59 |
| Age (years) | ||||||||||||
| 0-64 | 1.21 | 1.16-1.26 | <0.001 | 1.11 | 1.02-1.22 | 0.017 | 1.07 | 1.01-1.13 | 0.020 | 0.99 | 0.88-1.10 | 0.81 |
| 65-69 | Ref | Ref | Ref | Ref | ||||||||
| 70-74 | 1.04 | 1.01-1.08 | 0.019 | 0.995 | 0.92-1.07 | 0.89 | 1.11 | 1.06-1.16 | <0.001 | 1.06 | 0.97-1.16 | 0.22 |
| 75-79 | 1.17 | 1.13-1.21 | <0.001 | 1.12 | 1.04-1.20 | 0.002 | 1.35 | 1.30-1.41 | <0.001 | 1.28 | 1.18-1.40 | <0.001 |
| 80-84 | 1.35 | 1.30-1.40 | <0.001 | 1.32 | 1.23-1.43 | <0.001 | 1.65 | 1.58-1.72 | <0.001 | 1.54 | 1.41-1.68 | <0.001 |
| 85+ | 1.64 | 1.57-1.71 | <0.001 | 1.54 | 1.40-1.70 | <0.001 | 2.18 | 2.07-2.30 | <0.001 | 2.01 | 1.80-2.24 | <0.001 |
| Black | 1.34 | 1.29-1.39 | <0.001 | 1.35 | 1.25-1.45 | <0.001 | 1.12 | 1.07-1.17 | <0.001 | 1.05 | 0.95-1.15 | 0.34 |
| Other | 1.18 | 1.12-1.25 | <0.001 | 1.01 | 0.90-1.14 | 0.87 | 1.05 | 0.98-1.13 | 0.17 | 0.92 | 0.80-1.06 | 0.26 |
| White | Ref | Ref | Ref | Ref | ||||||||
| Charlson Score 0 | 0.79 | 0.68-0.91 | <0.001 | 0.69 | 0.47-0.999 | 0.050 | 0.70 | 0.57-0.85 | <0.001 | 0.76 | 0.48-1.20 | 0.24 |
| Charlson Score 1 | Ref | Ref | Ref | Ref | ||||||||
| Charlson Score 2 | 1.21 | 1.15-1.27 | <0.001 | 1.39 | 1.23-1.57 | <0.001 | 1.27 | 1.19-1.35 | <0.001 | 1.39 | 1.20-1.62 | <0.001 |
| Charlson Score 3 | 1.33 | 1.27-1.40 | <0.001 | 1.61 | 1.43-1.81 | <0.001 | 1.46 | 1.37-1.54 | <0.001 | 1.71 | 1.48-1.98 | <0.001 |
| Charlson Score 4+ | 1.77 | 1.69-1.85 | <0.001 | 2.20 | 1.96-2.47 | <0.001 | 2.06 | 1.94-2.18 | <0.001 | 2.56 | 2.22-2.95 | <0.001 |
| Diabetes Mellitus | 1.14 | 1.11-1.16 | <0.001 | 1.09 | 1.04-1.14 | <0.001 | 1.10 | 1.07-1.13 | <0.001 | 1.02 | 0.97-1.08 | 0.50 |
| Mitral or Aortic Valve Disorder | 1.07 | 1.05-1.10 | <0.001 | 1.11 | 1.06-1.17 | <0.001 | 1.10 | 1.07-1.13 | <0.001 | 1.12 | 1.06-1.19 | <0.001 |
| Hypertension | 0.97 | 0.94-1.01 | 0.13 | 0.96 | 0.89-1.04 | 0.31 | 0.86 | 0.83-0.89 | <0.001 | 0.91 | 0.83-0.999 | 0.029 |
| Myocardial Infarction | 1.06 | 1.04-1.08 | <0.001 | 0.96 | 0.92-1.01 | 0.12 | 1.01 | 0.99-1.04 | 0.42 | 0.94 | 0.89-0.990 | 0.017 |
| Coronary Heart Disease | 1.17 | 1.14-1.19 | <0.001 | 1.10 | 1.05-1.16 | <0.001 | 1.18 | 1.15-1.21 | <0.001 | 1.07 | 1.01-1.13 | 0.017 |
| Tricuspid or Pulmonary Valve Disorder | 1.05 | 1.02-1.08 | <0.001 | 1.02 | 0.97-1.08 | 0.41 | 1.06 | 1.03-1.10 | 0.001 | 1.00 | 0.94-1.07 | 0.94 |
| Atrial Fibrillation | 1.29 | 1.27-1.32 | <0.001 | 1.24 | 1.18-1.29 | <0.001 | 1.29 | 1.26-1.32 | <0.001 | 1.24 | 1.18-1.31 | <0.001 |
| Prior Heart Failure Hospitalization | 1.69 | 1.65-1.72 | <0.001 | 1.87 | 1.78-1.96 | <0.001 | 1.53 | 1.49-1.57 | <0.001 | 1.65 | 1.57-1.75 | <0.001 |
| Stroke | 1.06 | 1.02-1.11 | 0.007 | 1.03 | 0.94-1.11 | 0.56 | 1.10 | 1.04-1.15 | <0.001 | 1.03 | 0.94-1.13 | 0.55 |
| Peripheral Vascular disease | 1.16 | 1.13-1.19 | <0.001 | 1.12 | 1.07-1.18 | <0.001 | 1.19 | 1.16-1.22 | <0.001 | 1.18 | 1.12-1.24 | <0.001 |
| Ventricular Tachycardia | 1.21 | 1.18-1.23 | <0.001 | 1.09 | 1.04-1.14 | <0.001 | 1.22 | 1.19-1.25 | <0.001 | 1.12 | 1.06-1.18 | <0.001 |
| Percutaneous Coronary Intervention | 0.65 | 0.62-0.69 | <0.001 | 0.68 | 0.62-0.75 | <0.001 | 0.57 | 0.53-0.62 | <0.001 | 0.58 | 0.51-0.65 | <0.001 |
| Coronary Bypass | 0.93 | 0.89-0.96 | <0.001 | 0.96 | 0.90-1.02 | 0.15 | 0.84 | 0.80-0.87 | <0.001 | 0.85 | 0.79-0.92 | <0.001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
