An exploration of the association between ischemic etiology and the likelihood of heart failure hospitalization following cardiac resynchronization therapy

Abstract

Background

Myocardial scarring resulting from cardiomyopathy with ischemic etiology may interfere with effective lead placement during implantation of multi‐lead cardioverter‐defibrillators for cardiac resynchronization therapy (CRT‐D). Extensive scarring is known to be associated with poorer physiological and survival outcomes in patients who undergo CRT‐D.

Hypothesis

Ischemic CRT‐D recipients respond as well as nonischemic recipients, using hospital admission for heart failure (HF) as a measure of response.

Methods

Patients who underwent CRT‐D between February 1, 2013, and February 1, 2014, were identified in an insurer's claims. Inclusion required 1 year of enrollment pre– and post–CRT‐D. The sample was divided into nonischemic and ischemic groups, and a subset of the ischemic group with a history of ST‐segment elevation myocardial infarction (STEMI) was identified. The likelihood of HF hospital admissions in the year before and after CRT‐D was computed for each group, as well as for the subset of patients with HF admissions prior to CRT‐D.

Results

A significant (P = 0.02) association was found between ischemic etiology and the post–CRT‐D HF admission likelihood. No association was found between history of STEMI vs nonischemic status and likelihood of post–CRT‐D HF admission. All groups had significantly lower risk of HF admissions after CRT‐D. None of the comparisons involving only patients with a HF hospitalization in the year prior to CRT‐D were significant.

Conclusions

Patients with nonischemic etiology were significantly less likely to experience a HF admission after CRT‐D, but the risk of HF admission improved significantly in all groups after CRT‐D.

1. INTRODUCTION

Cardiac resynchronization therapy with defibrillation (CRT‐D), compared with defibrillation alone, has been shown to reduce the risk of hospitalization or mortality; however, there is evidence that all subpopulations of patients do not benefit equally from CRT‐D.1, 2, 3 Patients who undergo CRT or CRT‐D are put at 0.3% risk of in‐hospital mortality, 0.9% risk of pneumothorax, 2.0% risk of coronary sinus complications, and 5.7% risk of experiencing lead dislodgment.4 Thus, it is important to weigh the benefits of CRT‐D for patients with specific etiologies, such as ischemic heart disease, against the potential complications that it may cause.

Although appropriate use criteria state that CRT‐D is appropriate for some patients with both ischemic and nonischemic cardiomyopathy, the relative benefits that both groups of patients receive from CRT‐D are ambiguous.5 Some prior research suggests that heavier scar burden accounts for a reduced likelihood of physiological improvement after CRT‐D among patients with ischemic cardiomyopathy, compared with those who have nonischemic cardiomyopathy. Patients with nonischemic cardiomyopathy have been shown to have a greater probability of survival free from transplant or mechanical support in the 5 years following CRT‐D, relative to patients with ischemic cardiomyopathy. Furthermore, ischemic patients with greater scar burden have been found to have worse clinical and functional outcomes after CRT‐D.6 In other research, avoiding the placement of leads over scar tissue in patients with ischemic cardiomyopathy resulted in improved pump function during CRT.7 Nevertheless, the Multicenter Automatic Defibrillator Implantation Trial with Cardiac Resynchronization Therapy (MADIT‐CRT) found that in comparison with defibrillation alone, the effect of CRT‐D on heart failure (HF) events and death did not significantly differ between patients with ischemic cardiomyopathy and patients with nonischemic cardiomyopathy.2, 8 Likewise, the Cardiac Resynchronization–Heart Failure (CARE‐HF) study found that the impact of CRT on symptoms, quality of life, morbidity, and mortality are similar for patients with and without ischemic cardiomyopathy.9

Building upon clinical efforts to understand the relationship between ischemia and the efficacy of CRT‐D, our study uses real‐world claims data to evaluate the association between ischemic etiology and post–CRT‐D HF admissions. Most prior studies of this relationship have relied upon tightly controlled clinical trials2, 8, 9 or have used physiological measures to predict response.10

2. METHODS

2.1. Study design

This retrospective cohort study examined the association between patients' diagnoses of ischemic vs nonischemic cardiomyopathy in the year prior to CRT‐D and the likelihood with which patients were hospitalized for HF in the year following CRT‐D. As patients with a history of ST‐segment elevation myocardial infarction (STEMI) are likely to have more intense scarring than other patients with ischemic cardiomyopathy, the outcomes experienced by the subset of patients with a history of STEMI were also compared with those experienced by patients with nonischemic cardiomyopathy.11 (An additional comparison of the ischemic patients without a history of STEMI with the nonischemic patients was not possible, because some of the patients without documentation of STEMI may have experienced a STEMI event earlier than within the period 12 months prior to CRT‐D that was examined for this study.) The comparisons were conducted both among all patients who had undergone CRT‐D as well as with a subpopulation of patients who had a HF admission prior to CRT‐D.

Hospitalization for HF in the 12 months prior to and following implantation was ascertained for each patient. As a history of HF‐related hospitalization has been shown to increase risk for subsequent HF‐related rehospitalization, there was an additional analysis of patients with a history of HF admission.12 HF hospitalization was identified by examining patients' claims history for claims with International Classification of Diseases, Ninth Revision (ICD‐9) code 428.0 (congestive heart failure, unspecified) as the primary diagnosis.

Institutional review board approval was obtained for this research through Schulman IRB. Investigations were conducted in accordance with the Declaration of Helsinki.

2.2. Sample selection

Medicare medical claims were extracted from the database of Humana Inc., a healthcare company that in 2013 provided medical benefit plans to 12.0 million individuals across the nation.13 Initially selected patients received CRT‐D between February 1, 2013, and February 1, 2014, were age ≥ 18 years at the time, and were eligible for research. Patients who received CRT‐D were identified by examining the Current Procedural Terminology (CPT) codes of claims for instances where patients received left ventricular pacing lead add‐on (33225) and insertion or replacement of defibrillator with leads (33249) on the same day. The pacing lead's presence indicates that the implantable cardioverter‐defibrillator was inserted for the purposes of CRT‐D, and was not being used solely to provide infrequent defibrillation shocks. The first instance of the co‐occurrence of these codes in the claims data was used as the implantation date. Exclusions were then made for patients who had either of the CRT‐D–related CPT codes on a claim in the 12 months prior to CRT‐D. Patients who were not continuously enrolled with their health plan from 12 months before CRT‐D to 12 months after CRT‐D were also excluded. Figure 1 shows the participant flow.

Figure 1.

Participant flow diagram. Abbreviations: CRT‐D, cardiac resynchronization therapy with defibrillation; HF, heart failure; STEMI, ST‐segment elevation myocardial infarction

2.3. Statistical analysis

Patients were classified as having had ischemic etiology of HF or a history of STEMI using diagnosis codes (see Supporting Information in the online version of this article). Patients without diagnoses or procedure codes qualifying them to be placed in the ischemic etiology or STEMI groups were classified into the nonischemic etiology group. Fisher exact tests were conducted to evaluate whether an association existed between ischemic vs nonischemic etiology and the likelihood of an HF admission in the year following implantation. The test was repeated to evaluate whether there was association between STEMI vs nonischemic etiology and post‐implantation HF admission. Both tests for an association between etiology and post‐implantation HF admission were then conducted in the subpopulation of patients who had experienced a HF admission in the year prior to CRT‐D, as these patients may have been more equally predisposed to hospitalization for HF. Likelihood of pre‐implantation and post‐implantation HF admission was compared within groups using McNemar tests.

3. RESULTS

A total of 1303 patients were included in the study (Figure 1). Of the 555 exclusions for lack of continuous enrollment during the year following CRT‐D, 4 were due to deaths and the others were due to disenrollment. A hospitalization for HF in the year prior to CRT‐D was experienced by 79.7% of those with nonischemic etiology, 64.3% of those with ischemic etiology, and 60.7% of those with a history of STEMI. Table 1 presents patient age and sex, organized by etiology and presence of pre–CRT‐D HF admissions.

Table 1.

Patient demographics by etiology and presence of pre–CRT‐D HF admissions

	N	Male Sex	Mean Age, y
Nonischemic	1084 (100)	797 (73.5)	72.6 ± 8.5
No HF admission pre–CRT‐D	864 (79.7)	643 (74.4)	72.7 ± 8.4
HF admission pre–CRT‐D	220 (20.3)	154 (70.0)	72.0 ± 7.8
Ischemic	219 (100)	162 (74.0)	72.0 ± 8.0
No HF admission pre–CRT‐D	141 (64.4)	108 (76.6)	71.3 ± 8.8
HF admission pre–CRT‐D	78 (35.6)	54 (69.2)	73.3 ± 6.8
STEMI	79 (100)	57 (72.2)	72.5 ± 8.3
No HF admission pre–CRT‐D	48 (60.8)	39 (81.3)	72.0 ± 7.1
HF admission pre–CRT‐D	31 (39.2)	18 (58.1)	73.3 ± 8.2

Abbreviations: CRT‐D, cardiac resynchronization therapy with defibrillation; HF, heart failure; SD, standard deviation; STEMI, ST‐segment elevation myocardial infarction.

Data are presented as n (%) or mean ± SD.

Figure 2 depicts the proportion of patients with pre–CRT‐D and post–CRT‐D HF admissions for each of the groups in the study sample, as well as the proportion of patients with post–CRT‐D HF admissions for the subgroups with a HF hospitalization in the year prior to CRT‐D. When pre–CRT‐D and post–CRT‐D HF hospitalization likelihood was compared within the 3 groups, each group was found to have a significantly lower risk of HF admission post–CRT‐D. Both pre‐ and post‐implantation HF hospitalizations were more common in the ischemic etiology group and STEMI subgroup than in the nonischemic etiology group. In the subpopulation with a HF admission in the year prior to CRT‐D, there were also consistently significant reductions in hospitalization following CRT‐D for each etiology subgroup.

Figure 2.

Pre/post comparison of HF hospitalizations within 1 year pre– and 1 year post–CRT‐D by etiology. P < 0.01 for all pre/post comparisons using McNemar test. Abbreviations: CRT‐D, cardiac resynchronization therapy with defibrillation; HF, heart failure; STEMI, ST‐segment elevation myocardial infarction

As expected, Table 2 shows that ischemic compared with nonischemic etiology was significantly associated with a greater risk of post–CRT‐D HF hospitalization. However, no statistically significant association was detected between etiology and post–CRF‐D HF hospitalization when only the STEMI subgroup was considered. Among the subpopulation with a HF hospitalization in the year prior to CRT‐D, there was no significant association between etiology and likelihood of hospitalization post–CRT‐D when the ischemic group was compared with the nonischemic group, or when the STEMI group was compared with the nonischemic group. Post hoc power analyses (α = 0.05) were conducted to assess whether the study was adequately powered for the comparisons with nonsignificant results. None of these comparisons had ≥80% power for detecting the observed differences.

Table 2.

Comparison of HF hospitalizations 1 year post–CRT‐D by ischemic etiology

	N	HF Hospitalization in 1 Year After CRT‐D	P Value	Post Hoc Power (α = 0.05), %
Total study sample
Nonischemic	1084	141 (13.0)	—	—
vs ischemic	219	42 (19.2)	0.02	65.7
vs STEMI	79	13 (16.5)	0.39	16.2
Subgroup of patients with ≥1 HF hospitalization in 1 year pre–CRT‐D
Nonischemic	220	47 (21.4)	—	—
vs ischemic	78	24 (30.8)	0.12	39.3
vs STEMI	31	5 (16.1)	0.64	8.2

Abbreviations: CRT‐D, cardiac resynchronization therapy with defibrillation; HF, heart failure; SD, standard deviation; STEMI, ST‐segment elevation myocardial infarction.

Data are presented as n (%) unless otherwise noted.

4. DISCUSSION

In the overall study group containing patients with and without an HF admission in the year prior to CRT‐D, results demonstrated an association between a proxy for myocardial scarring (ischemic etiology) and the likelihood of HF admission in the year following CRT‐D. Prior HF admissions were more common among patients with ischemic etiology (35.6%) than among those without ischemic etiology (20.3%). Due to their more frequent history of HF admission, patients with ischemic etiology may have been at greater risk of post‐implantation HF admission, and indeed had a higher likelihood of post‐implantation admission. Nevertheless, patients with ischemic etiology experienced a larger reduction in HF admissions than did patients with nonischemic etiology. In analyses restricted to patients who had been hospitalized in the year prior to CRT‐D and in analyses that considered only patients from the ischemic group who had a recorded diagnosis of STEMI, no association between etiology and post–CRT‐D hospitalization was detected.

Despite evidence that myocardial scarring is associated with a reduced likelihood of successful response to CRT‐D,6 findings from both the present study and the MADIT trial2, 8 suggest that the health benefits of CRT‐D do not significantly differ between patients with ischemic and nonischemic etiology. Furthermore, as the reduction in HF admissions likelihood was greater for the patients with ischemic etiology, the findings of this study do not suggest that they experience less benefit from CRT‐D.

4.1. Study limitations

A number of limitations need to be weighed. This study was conducted entirely using administrative claims data, which are subject to error. Some patients in the study sample may have had a prior STEMI that was not reflected in claims from the year prior to CRT‐D. Patients with old histories of STEMI may have been misclassified as having ischemic, non‐STEMI as a result, which might have detracted from the representativeness of the STEMI subgroup. Misclassification of ischemic/nonischemic etiology was less likely, as physicians persist in coding patients as having ischemic etiology, whereas STEMI is coded as a 1‐time event that is unlikely to be listed several months after it occurs. The limitations of using claims data prevented the analysis from including a control group of patients for whom CRT‐D had been considered but not performed for nonclinical reasons.

Furthermore, lack of statistical power may have prevented detection of an association between etiology and post–CRT‐D hospitalization in comparisons involving subgroups with a diagnosis of STEMI and/or HF hospitalization in the prior year. Additionally, this study only considered patients who survived for a year after CRT‐D and were continuously enrolled in their health plan over a 2‐year period. Although few patients died during the year following CRT‐D, the sample only represents people who were in good‐enough health to receive CRT and then survive a year after receiving it. Finally, restricting follow‐up to 1 year may have diminished the ability of the study to detect differences in HF admission likelihood post–CRT‐D, as long‐term differences were not measured.

5. CONCLUSION

Patients appear to experience a significantly lower likelihood of HF admissions in the year following CRT‐D than in the year prior, regardless of whether they have ischemic etiology or a history of STEMI. Ischemic etiology was found to be associated with a significantly higher likelihood of post–CRT‐D HF, which may reflect the fact that patients with ischemic etiology also had a higher likelihood of pre–CRT‐D HF hospitalization than did patients with nonischemic etiology. The change in risk of HF admissions pre–CRT‐D vs post–CRT‐D was more dramatic for patients with ischemic etiology than for patients with nonischemic etiology. Thus, a reduction in post–CRT‐D HF likelihood occurred regardless of etiology. This study demonstrates the utility of using claims‐based outcomes research to corroborate the findings from clinical trials. Further research is needed to definitively determine whether claims‐based diagnoses can be used to assess when CRT‐D is indicated. This is particularly true for analyses involving subgroups characterized by a diagnosis of STEMI, for which this study was very underpowered. As multipayer claims databases are built, it will become feasible to examine how larger populations of patients respond to CRT‐D, overcoming the power limitations impacting this study.

Powell AC, Rogstad TL, Deshmukh UU, Price SE, Simmons JD. An exploration of the association between ischemic etiology and the likelihood of heart failure hospitalization following cardiac resynchronization therapy. Clin Cardiol. 2017;40:1090–1094. 10.1002/clc.22779