National Trends in the Utilization of Cardiac Resynchronization Therapy with or without Implantable-Cardioverter Defibrillator

Abstract

Background

Candidates for cardiac resynchronization therapy (CRT) receive either a biventricular pacemaker (CRT-P) or a biventricular pacemaker with an implantable-cardioverter defibrillator (ICD; CRT-D). Optimal device selection remains challenging as the benefit of ICD therapy may not be uniform, particularly in patients at competing risk of non-sudden death.

Methods and Results

In this serial cross-sectional study using the National Inpatient Sample database, we identified 311,086 admissions associated with CRT implant between 2006-2012. CRT-D was the most common device-type (86.1%), including in patients ≥75 years old with 5 or more Elixhauser comorbidities (75.5%). Multivariate predictors of CRT-D implant included demographic, clinical, and geographic factors: prior ventricular arrhythmia (rate ratio [RR], 1.14; 95% CI, 1.13-1.14), ischemic heart disease (RR, 1.11; 95% CI, 1.10-1.11), male gender (RR, 1.10; 95% CI, 1.09-1.10), black race (RR, 1.06; 95% CI: 1.04-1.07), and Northeast geographic region (RR, 1.06; 95% CI, 1.04-1.09). There was significant inter-hospital variation in the use of CRT-D (10-90 percentile range, 72.9% to 98.0% CRT-D).

Conclusions

The majority of patients in this contemporary US cohort underwent implantation of CRT-D. Predictors of CRT-D implant included demographic, clinical, and geographic factors. In patient subgroups predicted to have an attenuated benefit from ICD therapy (older adults with multiple comorbidities), CRT-D remained the dominant device type. An improved understanding of the determinants of device selection may aid in decision-making and ultimately better align patient risk with device benefit at the time of CRT implantation.

Introduction

Cardiac resynchronization therapy (CRT) is an established therapy for heart failure patients with electrical dyssynchrony.^1-5 Resynchronization of the failing ventricle is associated with improvement in left ventricular ejection fraction, reduced risk of ventricular arrhythmia, improvement in heart failure symptoms, and improved survival.⁶ A significant majority of patients meeting guideline criteria for CRT additionally meet criteria for placement of an implantable cardioverter-defibrillator (ICD).^6-8 Randomized controlled trials have demonstrated reduced morbidity and mortality with both CRT pacemaker (CRT-P) and CRT with ICD therapy (CRT-D).⁶ Indeed, despite the rationale for ICD implantation at the time of CRT to reduce sudden death, there remains no direct comparison establishing the incremental efficacy of CRT-D compared to CRT-P alone.⁹

In addition, despite the established efficacy of ICD therapy, there is also increasing evidence that the benefit of ICD therapy may not be uniform.^{10, 11} Increasing comorbidities including age, peripheral vascular disease, diabetes, and chronic kidney disease may attenuate the survival benefit of ICD therapy.^{10, 12} Current guidelines discourage ICD implantation in patients expected to survive less than a year, although risk assessment at the time of CRT device selection remains practically challenging particularly for patients under-represented in clinical trials.¹³ An improved understanding of contemporary patterns of CRT device implantation (CRT-D vs. CRT-P), including predictors of implant type, may identify opportunities to align effective device therapies with patients most likely to benefit.

This study examines recent trends in CRT device implantation in the United States between 2006 and 2012. We utilized the National Inpatient Sample (NIS)¹⁴ to (1) determine the demographic and clinical characteristics of patients undergoing inpatient CRT, (2) compare characteristics of patients receiving CRT-D vs. CRT-P, with special attention to patient age and comorbidities at the time of implantation, and (3) assess temporal trends in device selection.

Methods

Data Source

The National Inpatient Sample (NIS) is the largest publicly available all-payer inpatient health care database in the United States, yielding national estimates of diagnoses, procedure utilization, and outcomes for hospital inpatient stays.¹⁴ The database contains a nationally representative sample from more than 7 million hospitalizations annually. Applying sample weights provided by NIS, the database projects estimates for more than 36 million hospitalizations annually. The NIS provides data on patient demographics, in-hospital clinical outcomes, hospital characteristics, and hospital charges. Federal hospitals are not included in the NIS. Quality control and validation of the NIS are performed by the Agency for Healthcare Research and Quality (AHRQ; Rockville, MD). The database was provided with de-identified patient information and thus was deemed exempt from institutional review by the Human Research Committee at Massachusetts General Hospital.

Study sample

We included all adults ≥18 years old who received a CRT device during a hospitalization between January 1, 2006 and December 31, 2012. CRT was defined by the International Classification of Diseases, 9^th Revision, Clinical Modification (ICD-9-CM) codes for CRT-P (00.50) and CRT-D (00.51). These two ICD-9-CM codes identify procedures where a total CRT system is implanted and do not identify CRT generator replacement.

Covariates and Other Exposures of Interest

Patient demographics included age, sex, race/ethnicity, primary insurance type (Medicare, Medicaid, Private, and other), and income quartile. Race/ethnicity was reclassified as non-Hispanic white, non-Hispanic black, other (Hispanic, Asian, and Native American), and missing. Other risk factors of interest included hospital type (rural, urban non-teaching, and urban teaching), type of admission (elective vs. acute), geographic region, and year of hospital discharge. We defined older adults as being age 75 or older.

Cardiac diagnoses of clinical relevance were identified by ICD-9-CM codes and included ischemic heart disease, non-ischemic cardiomyopathy, history of ventricular arrhythmia or cardiac arrest, atrial fibrillation, left bundle branch block, and complete atrio-ventricular (AV) block. All ICD-9-CM codes used in the data analysis are listed in Supplemental Table 1.

The Elixhauser comorbidity measure was originally developed using administrative data relying on the ICD-9-CM coding manual for 30 comorbidities.¹⁵ The comorbidities identified by the Elixhauser comorbidity index are significantly associated with in-hospital mortality and mortality beyond 30 days of hospitalization.^{16, 17} In this study, Elixhauser comorbidities were generated from ICD-9-CM diagnosis codes using the AHRQ Comorbidity Software.¹⁴ Of note, heart failure and arrhythmia were removed from the comorbidity score in order to be analysed independently, and hence, the total number of Elixhauser comorbidities analyzed was 28. The sum of Elixhauser comorbidities for each hospitalization was reclassified as 0, 1-2, 3-4, or ≥5. As a complement to the Elixhauser comorbidity score, we generated a heart failure comorbidity score comprised of seven comorbid conditions shown to be associated with attenuated ICD efficacy in patients with heart failure: ischemic heart disease, chronic kidney disease, diabetes, chronic pulmonary disease, atrial fibrillation, peripheral vascular disease, and tobacco use.^18-20 The sum of heart failure comorbidities for each hospitalization was reclassified as 0, 1-2, 3-4, or ≥5. As components of the Elixhauser and heart failure comorbidity score overlapped, they were analyzed separately.

Statistical Analyses

All analyses were performed using survey procedures in SAS version 9.4 (SAS Institute, Cary, NC) to account for the complex survey design of the NIS. Frequencies, proportions and 95% confidence intervals (CI) were calculated and weighted to reflect national estimates using inverse sampling weights provided by NIS. Chi-square tests were used to compare the demographic and clinical characteristics between patients who received different types of device (CRT-P vs. CRT-D). Additionally, chi-square tests were used to compare the percentage of CRT-D use between patients age <75 and age ≥75 stratified by different clinical factors. Multivariable logistic regression was used to identify predictors of CRT-D use vs. CRT-P use. Candidate variables for model inclusion were demographic and clinical factors (including the Elixhauser and heart failure comorbidity indices detailed above). There was no evidence of interaction between age and other predictors when the analysis was repeated stratified by age group (<75 and ≥75); therefore, the overall model was presented. We converted the odds ratios from the model and presented rate ratios in our analysis. To compare the change in the proportion of CRT-D implants over time, chi-square tests were used to compare the rate of CRT-D use from 2006 to 2012 between different patient groups.

Results

Study population

Between 2006 and 2012, 311,086 CRT devices were implanted in the inpatient setting in the United States. The majority of patients receiving CRT were older adults (72% were ≥65), male, and had the device implanted in urban teaching hospitals. Just over half of the procedures were performed during an acute hospital admission. At least one third of CRT recipients had three or more comorbidities reported in addition to heart failure. Of all CRT implants, a CRT-D system was implanted 86.1% of the time.

The demographic and clinical characteristics of patients receiving CRT, stratified by device type, are presented in Table 1. There were distinct differences between patients receiving CRT-D and CRT-P. Patients receiving CRT-D were more likely to have a history of ventricular arrhythmia (27.6% vs. 9.2%, p<0.001) and ischemic heart disease (67.7% vs. 53.1%, p<0.001) whereas patients receiving CRT-P were more likely to have atrial fibrillation (56.8% vs. 34.3%, p<0.001) or complete AV block (19.9% vs. 7.3%, p<0.001). The mean age of patients receiving CRT-P was higher compared to patients receiving CRT-D (75 vs. 69 years, p<0.001). CRT-P patients were more likely to have 3 or more of the 28 different comorbidities measured in the Elixhauser index compared to CRT-D patients (37.6% vs. 33.6%, p<0.001). The six most common comorbidities were hypertension, diabetes, chronic lung disease, chronic kidney disease, and anemia.

Table 1.

Characteristics of patients undergoing CRT implantation, stratified by device type.

	CRT (n=311,086)	CRT-D (n=267,901)	CRT-P (n=43,184)	P- value
Age category				<0.001
18-54	32,982 (10.6)	30,499 (11.4)	2,483 (5.7)
55-64	55,376 (17.8)	50,979 (19.0)	4,398 (10.2)
65-74	95,768 (30.8)	85,781 (32.0)	9,987 (23.1)
75-84	102,074 (32.8)	85,214 (31.8)	16,860 (39.0)
>85	24,886 (8.0)	15,429 (5.8)	9,457 (21.9)
Male	218,883 (70.4)	194,016 (72.4)	24,867 (57.6)	<0.001
Race				<0.001
White	196,786 (63.3)	168,439 (62.9)	28,348 (65.6)
Black	27,584 (8.9)	25,091 (9.4)	2,493 (5.8)
Other*	27,605 (8.9)	24,246 (9.1)	3,359 (7.8)
Unknown	59,111 (19.0)	50,126 (18.7)	8,985 (20.8)
Primary payer				<0.001
Medicare	225,462 (72.5)	190,614 (71.2)	34,848 (80.7)
Medicaid	14,861 (4.8)	13,677 (5.1)	1,184 (2.7)
Private	60,020 (19.3)	53,934 (20.1)	6,085 (14.1)
Other†	10,338 (3.3)	9,305 (3.5)	1,033 (2.4)
Low income quartile	80,865 (26.0)	70,656 (26.4)	10,208 (23.6)	<0.001
Region				<0.001
Northeast	60,734 (19.5)	54,276 (20.3)	6,458 (15.0)
Midwest	81,012 (26.0)	68,131 (25.4)	12,881 (29.8)
South	117,162 (37.7)	100,724 (37.6)	16,438 (38.1)
West	52,178 (16.8)	44,771 (16.7)	7,407 (17.2)
Hospital location/teaching				0.02
Rural	12,334 (4.0)	9,540 (3.6)	2,794 (6.5)
Urban, non-teaching	111,423 (35.8)	96,663 (36.1)	14,760 (34.2)
Urban, teaching	185,080 (59.5)	159,639 (59.6)	25,441 (58.9)
Elective admission	157,336 (50.8)	137,650 (51.4)	19,686 (45.6)	<0.001
Cardiac history
Ischemic heart disease	204,259 (65.7)	181,321 (67.7)	22,938 (53.1)	<0.001
Nonischemic cardiomyopathy	133,140 (42.8)	118,390 (44.2)	14,749 (34.2)	<0.001
Left bundle branch block	103,639 (33.3)	94,037 (35.1)	9,602 (22.2)	<0.001
Ventricular arrhythmia	77,932 (25.1)	73,945 (27.6)	3,987 (9.2)	<0.001
Atrial fibrillation	116,505 (37.5)	91,957 (34.3)	24,548 (56.8)	<0.001
Complete AV block	28,083 (10.5)	19,480 (7.3)	8,603 (19.9)	<0.001
Comorbidities
Diabetes	102,482 (32.9)	90,520 (33.8)	11,963 (27.7)	<0.001
Chronic kidney disease	64,714 (20.8)	55,289 (20.6)	9,425 (21.8)	0.04
Chronic lung disease	65,309 (21.0)	56,251 (21.0)	9,058 (21.0)	0.97
Anemia	33,404 (10.7)	27,310 (10.2)	6,093 (14.1)	<0.001
Hypertension	180,764 (58.1)	154,830 (57.8)	25,934 (60.1)	<0.001
Peripheral vascular disease	27,789 (8.9)	24,122 (9.0)	3,667 (8.5)	0.16
Tobacco use	60,481 (19.4)	53,654 (20.0)	6,826 (15.8)	<0.001
Elixhauser comorbidities‡ ≥3	106,242 (34.2)	89,996 (33.6)	16,246 (37.6)	<0.001
Heart failure comorbidities§ ≥3	106,558 (34.3)	92,117 (34.4)	14,441 (33.4)	0.15

^*Includes Hispanic, Asian or Pacific Islander, Native American, and other.

^†Includes self-pay, no charge, and other.

^‡Comprised of 28 comorbidities within the Elixhauser index (see Supplementary Table 1).¹⁴

^§Comprised of 7 comorbidities associated with mortality and reduced ICD efficacy among heart failure patients: ischemic heart disease, atrial fibrillation, diabetes, chronic lung disease, chronic kidney disease, peripheral vascular disease, and smoking.¹⁵

∥Percentages may not sum to 100 given missing data.

There was significant inter-hospital and geographic regional variation in CRT-D implantation (Figure 1, Supplementary Figure 1). As shown in Figure 1, there was marked variability in the proportional use of CRT-D across hospitals (10-90 percentile range: 72.9% to 98.0%). The variability in use was apparent even when stratifying hospitals by geographic region or teaching status (Supplemental Figure 1).

Figure 1. Hospital-level variation in the proportion of CRT-D implantation.

Shown is the proportion of CRT-D use at 512 hospitals with at least 10 admissions with CRT procedure between 2006 and 2011 (hospital identification was discontinued within the National Inpatient Sample in 2012).

Effect of age and comorbidity burden on CRT device selection

The percentage of patients receiving CRT-D, stratified by age (<75 vs. ≥75), are presented in Table 2. There was a significant reduction in CRT-D use among patients 75 years of age or older compared to younger patients (79.3% vs. 90.8%, p<0.001), although CRT-D was still implanted in the majority of patients. When comparing the percentage CRT-D use between older and younger patients within subgroups, the difference was least pronounced in patients with a history of ventricular arrhythmia (92.1% vs. 96.5%, p<0.001) and most pronounced in patients with an acute admission (76.6% vs. 90.5%, p<0.001) and in patients with five or more Elixhauser comorbidities (75.5% vs. 89.1%, p<0.001). When stratified by comorbidities known to attenuate the efficacy of ICD therapy in heart failure, there was no significant difference in the proportional use of CRT-D across strata of comorbidity burden.

Table 2.

Proportion of CRT patients receiving CRT-D compared to CRT-P device, stratified by age group

	% CRT-D
	Age 18-74 (n=157,258)	Age ≥75 (n=100,643)	Difference, % (95% CI)
Overall	90.8	79.3	11.6 (8.5 to 12.6)
Type of admission
Elective	91.2	81.7	9.3 (8.2 to 10.5)
Acute	90.5	76.6	13.7 (12.4 to 15.1)
Cardiac history
Ventricular arrhythmia	96.5	92.1	4.4 (3.5 to 5.2)
Ischemic heart disease	93.2	83.3	9.9 (9.0 to 10.9)
Atrial fibrillation	84.6	73.3	11.3 (10.1 to 12.6)
Complete AV block	72.7	65.9	6.8 (4.3 to 9.3)
Elixhauser comorbidities*
0	90.3	82.0	8.3 (6.8 to 9.7)
1-2	91.5	80.1	11.4 (10.3 to 12.6)
3-4	90.4	77.2	13.2 (11.7 to 14.8)
≥5	89.1	75.5	13.6 (11.2 to 16.0)
Heart failure comorbidities†
0	90.2	77.8	12.4 (10.3 to 14.6)
1-2	90.9	78.7	12.2 (11.0 to 13.5)
3-4	90.4	80.6	10.6 (9.4 to 11.8)
≥5	89.2	79.9	9.3 (6.3 to 12.3)

^*Measures 28 comorbidities of the Elixhauser index.¹⁴

^†Measures seven comorbidities associated with high mortality rate and reduced ICD efficacy among heart failure patients: ischemic heart disease, atrial fibrillation, diabetes, chronic lung disease, chronic kidney disease, peripheral vascular disease, and smoking.¹⁵

‡All comparisons between age 18-74 and age ≥75 had p<0.05.

Predictors of CRT-D

In the multivariable logistic regression model, significant predictors for CRT-D vs. CRT-P use included both clinical and non-clinical factors (Table 3). A history of ventricular arrhythmia and ischemic heart disease, black race, and treatment in the Northeast region were significantly associated with CRT-D use. Conversely, older age, calendar year of implant, and atrial fibrillation were significantly associated with CRT-P use. For example, patients with a history of ventricular arrhythmia had a 1.14 fold higher rate (95% CI, 1.13-1.14) of receiving CRT-D vs. CRT-P compared to those without such history, while patients age 85 or older had a 0.66 lower rate (95% CI, 0.62-0.70) of receiving CRT-D vs. CRT-P compared to patients age 18-54 years. The rate ratio estimates were qualitatively similar when the analysis was stratified by age group (<75 and ≥75).

Table 3.

Multivariable predictors of CRT-D implantation compared to CRT-P.

	CRT Procedures, N (% CRT-D)	CRT-D Predictors, RR (95% CI)
Calendar Year
2006	55,496 (89.0)	ref
2007	48,453 (88.2)	1.00 (0.98 - 1.02)
2008	46,282 (87.4)	1.00 (0.97 - 1.02)
2009	49,011 (86.7)	0.99 (0.96 - 1.01)
2010	40,688 (85.0)	0.97 (0.94 – 0.99)
2011	37,270 (82.1)	0.94 (0.91 – 0.97)
2012	33885 (81.5)	0.94 (0.91 - 0.96)
Age
18-54	32,982 (92.5)	ref
55-64	55,376 (92.1)	0.99 (0.98 - 1.00)
65-74	95,768 (89.6)	0.96 (0.95 - 0.98)
75-84	102,074 (83.5)	0.90 (0.88 - 0.92)
≥85	24,886 (62.0)	0.66 (0.62 - 0.70)
Male	218,883 (88.6)	1.10 (1.09 – 1.10)
Race
White	196,786 (85.6)	ref
Black	27,584 (91.0)	1.06 (1.04 - 1.07)
Other*	27,605 (87.8)	1.01 (0.99 - 1.03)
Primary payer
Private	60,020 (89.9)	Ref
Medicare	225,462 (84.5)	1.01 (1.00 -1.02)
Medicaid	14,861 (92.0)	1.02 (1.00 - 1.03)
Other†	10,338 (90.0)	0.99 (0.97 - 1.01)
Elective admission	157,336 (87.5)	1.04 (1.03 - 1.05)
Region
Midwest	81,012 (84.1)	ref
Northeast	60,734 (89.4)	1.06 (1.04 – 1.09)
South	117,162 (86.0)	1.01 (0.98 - 1.04)
West	52,178 (85.8)	1.01 (0.98 - 1.03)
Hospital location/teaching
Urban, teaching	185,080 (86.3)	ref
Urban, non-teaching	111,423 (86.8)	1.01 (0.99 - 1.03)
Rural	12,334 (77.4)	0.91 (0.77 - 1.00)
Ischemic heart disease	204,259 (88.8)	1.11 (1.10 - 1.11)
Ventricular arrhythmia	77,932 (94.9)	1.14 (1.13 – 1.14)
Left bundle branch block	103,639 (90.7)	1.08 (1.07 - 1.09)
Complete AV block	28,083 (69.4)	0.83 (0.81 - 0.85)
Atrial fibrillation	116,505 (78.9)	0.91 (0.90 - 0.92)
Elixhauser comorbidities
0	49,952 (87.1)	ref
1-2	154,891 (86.8)	1.00 (0.92 - 1.08)
3-4	81,745 (85.0)	0.98 (0.97 - 0.99)
≥5	24,497 (83.6)	0.98 (0.96 - 0.99)

^*Includes Hispanic, Asian or Pacific Islander, Native American, and other.

^†Includes self-pay, no charge, and other.

Temporal Trends of CRT Device Type

The absolute number of inpatient CRT-D devices implanted decreased over time from 49,363 in 2006 to 27,605 in 2012 while the number of CRT-P devices implanted remained relatively unchanged, with 6,133 in 2006 and 6,280 in 2012 (Figure 2); therefore, the percentage receiving CRT-D decreased over time. However, there was a bigger decrease in the proportion of CRT-D implantation in patients aged 75 years and older compared to those 18-74 years of age across the cohort period (83.1% to 73.1% in age ≥75 years vs. 92.9% to 87.3% in age 18-74 years, p<0.001). When stratified by the presence of comorbidities (≥3 vs. <3 Elixhauser comorbidities), CRT-D use decreased from 2006 to 2012 in both groups (86.7% to 79.8% in ≥ 3 comorbidities vs. 89.7% to 82.8% in < 3 comorbidities, p<0.001).

Figure 2. Distribution of inpatient CRT-D and CRT-P procedures between 2006 and 2012.

Shown are the absolute numbers of CRT-D and CRT-P implants at non-federal hospital in the United States between 2006 and 2012. There were a total of 311,086 inpatient CRT procedures during this time period.

Discussion

In this study, we examined contemporary implant trends of CRT with and without defibrillator therapy in the United States between 2006 and 2012. We demonstrate the following primary findings: (1) CRT-D was the most prevalent device type implanted in all patients, although there was significant inter-hospital and regional variation. (2) CRT-D implantation was highly prevalent (≥75%) in subgroups for whom ICD benefit may be attenuated (older adults with multiple comorbidities). (3) Predictors of CRT-D implantation included demographic (younger, male gender, black race), clinical (ischemic heart disease, history of ventricular arrhythmia, absence of atrial fibrillation) and non-clinical (geographic region) factors. (4) The proportion of CRT-D implants decreased between 2006 and 2012.

Strengths of this study include the magnitude of hospital encounters reflected, assessment of temporal trends, and focus on clinically relevant subgroups. The NIS represents the largest all-payer inpatient care database in the United States, reflecting approximately 260 million inpatient encounters from 2006 to 2012. This study cohort identified 311,086 hospitalizations with a CRT implantation. Patients receiving CRT-P were older, more likely to have atrial fibrillation, and less likely to have ischemic heart disease and ventricular arrhythmia. CRT-P patients were more likely to have multiple comorbidities compared to CRT-D when using the Elixhauser comorbidity index (37.6 vs. 33.6% with ≥3 vs. <3 Elixhauser comorbidities, p<0.001). Previous assessment of CRT device selection in the United States has been limited predominantly to single-center reports, particularly for clinically relevant subgroups such as the elderly.^{21, 22} For example, Kelli and colleagues identified a predominance of CRT-D implantation (90%) amongst a cohort of octogenarians (96 patients) at Emory University.²¹ Previous multi-center European cohorts assessing CRT implant patterns, similarly identified a predominance of CRT-D implantation (68-73%), though of lower magnitude compared to this study (86.1%).^{23, 24} Similar to our findings, patients receiving CRT-P were older, less likely male, less likely to have ischemic heart disease, and more likely to have atrial fibrillation.²⁴

CRT-D vs. CRT-P: Predictors and Temporal Trends

Leveraging the size and phenotyping available in this cohort, we identified several demographic, clinical, and geographic predictors of CRT-D implantation. We would propose the following hypotheses for some of the relationships identified. First, a majority of patients who meet criteria for CRT implantation additionally meet guideline criteria for ICD implant.²⁵ In keeping with this overlap, we found that a significant majority of patients undergoing CRT implantation in the US between 2006 and 2012 additionally underwent ICD implant. ICD selection in patients who are candidates for CRT can be broadly categorized into primary and secondary prevention, reflecting the established relationships between history of ventricular arrhythmia, symptomatic heart failure, and low left ventricular ejection fraction with risk of sudden cardiac death.¹¹ Concordant with guideline- and pathophysiology-based risk, we identified a history of ischemic heart disease and prior history of ventricular arrhythmia as significant predictors of CRT-D implantation.

Second, we found that advancing age and increasing comorbidity burden were significantly associated with a reduced likelihood of CRT-D implantation. This may reflect patient and physician integration of competing risk at the time of CRT implant in these ‘at-risk’ groups (older adults, multiple comorbidities). Despite the established efficacy of ICD therapy in improving survival in appropriately selected patients, there is an increasing recognition that ICD benefit may not be uniform across all patients.¹¹ ICD efficacy may be particularly attenuated in subgroups at risk for competing mechanisms of death (non-sudden and non-cardiac death).^{11, 26} Previous work in both primary and secondary prevention cohorts have identified comorbidities (age ≥75 years, symptomatic heart failure, atrial fibrillation, chronic kidney disease, diabetes mellitus, tobacco use) associated with competing modes of death.^{12, 18, 20, 26} For example, van Rees and colleagues utilized a comorbidity-based risk score in patients with ischemic heart disease to identify subgroup of patients with nearly 40% risk of dying without an appropriate ICD therapy.^{19, 26} Similarly, in pooled patient-level data from primary prevention ICD efficacy trials, the presence of >3 comorbidities nullified the survival benefit associated with ICD implant.²⁰ Despite the identified inverse relationship between comorbidity burden and CRT-D implant, we would nonetheless highlight that CRT-D implantation remained the most prevalent device type in these subgroups (e.g. prevalence of CRT-D use patients age ≥75 years with ≥5 comorbidities was 80%).

Third, male gender and black race were significant predictors of CRT-D therapy. These findings are consistent with secular trends in disparities identified in contemporary ICD registries.²⁷ Female gender is a known predictor of left ventricular reverse remodeling and normalization of LV function in CRT (i.e. super-response)²⁸ – both of which have been associated with a reduced risk of ventricular arrhythmia.²⁹ Whether gender and its relationship to LV reverse remodeling influences decision making at the time of CRT implant may warrant additional investigation. Fourth, we identified large inter-hospital variation in CRT implant type (10-90 percentile range of CRT-D implant, 72.9% to 98.0%). The range of CRT-D vs. CRT-P implantation identified in this study is concordant with the heterogeneity and geographic variation previously identified in European cohorts.^{24, 30-32} Others have reported on regional variation in ICD implantation in the US.^{33, 34} In this study, we identified a higher likelihood of CRT-D implantation in the Northeast region and a numeric trend to reduced CRT-D implant in rural hospitals although hospital-level variation persisted even within regional and hospital teaching-status strata. This geographic and hospital-level heterogeneity may reflect loco-regional differences in patient characteristics, procedural infrastructure, and/or diffusion of technology although a more comprehensive understanding of regional differences in device selection appears warranted.

Finally, we identified significant decrease in the absolute number as well as the proportion of CRT-D implantation across the study period (2006 vs. 2012). The absolute numeric trend in CRT-D implantation may reflect a general shift to more outpatient CRT implantation, though we would highlight that the absolute numbers of CRT-P inpatient implants was unchanged across the study period. To the extent that those undergoing inpatient CRT implantation have more comorbidities compared to patients undergoing outpatient implant,³⁵ we would speculate that the proportional decline in inpatient CRT-D may reflect the intersection of high-comorbidity patients and the expectation of reduced ICD benefit in this patient population.

Clinical Implications

In patients undergoing CRT, the selection of concomitant ICD therapy has several implication including both short- and long-term device complications,³⁶ the morbidity and clinical impact of inappropriate device therapy,³⁷ and cost-effectiveness.³⁸ Given the increasing incidence and prevalence of heart failure – expected to double in the US alone by 2030 – the population of patients for whom this decision making will be impactful will only increase with time.³⁹

While this study cannot directly address issues of CRT-D vs. CRT-P efficacy, there remain several practical and clinical implications of device selection. First, the risk-benefit of ICD inclusion at the time of CRT should integrate the significant rate of ICD lead malfunction (~15% over 3 years), the higher rate of generator change associated with battery depletion, and the associated complications at the time of device replacement.^{36, 40} Second, the risk-benefit of ICD therapy should integrate the morbidity and survival impact of inappropriate device therapies in the CRT population.³⁷ Third, given the significant all-cause mortality in heart failure patients despite ICD implantation (e.g. ~70% at 5 years in heart failure patients with LVEF 30-35%),⁴¹ we would also highlight the clinical and practical challenges related to defibrillator deactivation.^42,43-45 For example, in recent assessment of ICD therapies at the end of life, investigators reported ICD therapies in nearly one-third of patients in the last 24 hours of life despite more than half having a do-not-resuscitate order.⁴⁶ Finally, from a healthcare policy perspective, there are significant cost-effectiveness implications regarding the choice CRT-D vs. CRT-P. The incremental cost per life year gained was significantly higher for CRT-D compared to CRT-P as demonstrated recently in the CARE-HF study (e.g. for a 65-year old patient: €35,864 vs. €7,011).³⁸ Concordant with the expected higher cost of CRT-D implantation, we found that the average hospital charge in this cohort was $36,293 more for an elective admission with CRT-D implant compared to CRT-P (data not shown). The impact of these cost differences on comparative effectiveness of CRT-D vs. CRT-P remains an open question.⁴⁷

Limitations

Although this analysis represents the largest study of CRT device selection to date, there are several limitations. First, clinical data including left ventricular ejection fraction were not available and may have refined our understanding of device selection. Second, these data represent a cross-sectional assessment of device implantation. Longitudinal clinical outcomes may have offered a more comprehensive understanding of clinical outcomes of patients receiving CRT-D vs. CRT-P therapy. Third, patient characteristics and comorbidities were ascertained using ICD-9CM codes which have previously demonstrated high specificity and lower sensitivity in validation assessment.⁴⁸ This likely explains the low prevalence of certain characteristics (e.g. 33% prevalence of left bundle branch block). This bias towards underestimation of comorbidity burden would only highlight the prevalence of CRT-D use in this high-risk group. Fourth, the NIS dataset is limited to inpatient CRT procedures which may influence the generalizeability of our findings. As we have demonstrated previously, patients undergoing inpatient CRT implant have greater comorbidities and more advanced heart failure compared to those undergoing outpatient CRT.³⁵ Secular trends and proportional use of CRT-D may be different in patients undergoing outpatient CRT implant – although given the identified inverse relationship between comorbidity and CRT-D use, we would hypothesize even higher rates of CRT-D in the outpatient setting. In addition, those undergoing inpatient CRT implant likely have a higher prevalence of cardiac arrest compared to outpatient CRT and this may have influenced the selection of CRT-D vs. CRT-P. Finally, guidelines for CRT implantation were refined in 2012 (at the end of this study cohort), and included modified recommendations based on bundle branch block morphology, QRS duration, and (when appropriate) the percentage of anticipated ventricular pacing.⁸ These data do not reflect the influence of these updated guidelines on CRT device selection.

Conclusion

In this large dataset of patients undergoing CRT implant in US hospitals between 2006 and 2012, CRT-D implantation was significantly more common than CRT-P. In patient subgroups predicted to have an attenuated benefit from ICD (older adults with multiple comorbidities), CRT-D implant rates remained ≥75%. An improved understanding of clinical and non-clinical factors underlying CRT device selection may better align device benefit and patient risk.