Unadjusted health-related quality-of-life (HRQL) scores over time on the subscales of the Minnesota Living With Heart Failure Questionnaire (MLHFQ) and the 12-Item Medical Outcomes Short-Form Health Survey (SF-12) for patients randomized to treatment with an implantable cardioverter defibrillator (ICD) and standard medical therapy (STD). Directions of the scales have been standardized so that the vertical axis represents increasing HRQL for all subscales. A, Physical scale of the MLHFQ. B, Emotional scale of the MLHFQ. C, Physical component score (PCS) of the SF-12. D, Mental component score (MCS) of the SF-12. The upper range of the y-axis for the SF-12 has been reduced from the traditional 100 points to 65 points, as more than 95% of all recorded values fell within a narrow range near the midpoint of the scale. Confidence intervals for HRQL scores obtained later in the study were wider owing to the declining numbers of patients providing HRQL data and the increased heterogeneity of those scores. Error bars represent 95% confidence intervals.
The 12-Item Medical Outcomes Short-Form Health Survey (SF-12) scores adjusted by time in trial by treatment group. The modeled changes in health-related quality-of-life (HRQL) on the mental component score (MCS) and the physical component score (PCS) of the SF-12 throughout the trial are shown. The y-axis represents the mean SF-12 score, with the full scale ranging from 0 to 100. Higher scores represent better HRQL.
Minnesota Living With Heart Failure Questionnaire (MLHFQ) scores adjusted by time in trial by treatment group. A, The modeled changes in health-related quality-of-life (HRQL) on the emotional scale of the MLHFQ are shown. The y-axis represents the 50th through 100th percentiles of the 0- to 25-point emotional scale. B, The modeled changes in HRQL on the physical scale of the MLHFQ are shown. The y-axis represents the 50th through 100th percentiles for the 0- to 40-point physical scale. Decreasing scores represent a better HRQL for both measures. ICD indicates implantable cardioverter defibrillator; STD, standard medical therapy.
Impact of shocks on health-related quality-of-life (HRQL): mean change and percentage of overall score. The direction of the scales has been normalized so that positive values represent a lower HRQL. The mean HRQL differed significantly compared with preshock values for the emotional scale of the MLHFQ and the mental component scale (MCS) of the 12-Item Medical Outcomes Short-Form Health Survey (SF-12) (asterisks, P = .04 for comparison with preshock HRQL). However, these changes represent a small component of each total score (< 2.0% for each). Error bars represent 95% confidence intervals. Emotional indicates the emotional score on the Minnesota Living With Heart Failure Questionnaire (MLHFQ); Physical, the physical score on the MLHFQ; and PCS, the physical component score on the SF-12.
Passman R, Subacius H, Ruo B, Schaechter A, Howard A, Sears SF, Kadish A. Implantable Cardioverter Defibrillators and Quality of LifeResults From the Defibrillators in Nonischemic Cardiomyopathy Treatment Evaluation Study. Arch Intern Med. 2007;167(20):2226-2232. doi:10.1001/archinte.167.20.2226
The Defibrillators in Nonischemic Cardiomyopathy Treatment Evaluation study demonstrated that implantable cardioverter defibrillators (ICDs) significantly reduce the risk of sudden cardiac death in patients with nonischemic cardiomyopathy and an ejection fraction of 35% or less, with no statistically significant decrease in overall mortality. The impact of ICD placement and shock on health-related quality of life (HRQL) in this population is unknown.
The 12-Item Medical Outcomes Short-Form Health Survey and the Minnesota Living with Heart Failure Questionnaire were administered to 458 patients with nonischemic cardiomyopathy, an ejection fraction of 35% or less, and either nonsustained ventricular tachycardia or 10 or more premature ventricular depolarizations per hour at baseline, 1 month after randomization, and every 3 months thereafter throughout the trial. The subjects were randomized to an ICD or standard medical therapy. Outcomes were compared using hierarchical linear regression.
Overall, there were no significant differences in HRQL throughout the trial between patients randomized to an ICD or standard medical therapy. However, in patients with 1 or more ICD shocks, HRQL declined 0.5 ± 0.2 (mean ± SD) points per shock on the emotional scale of the Minnesota Living with Heart Failure Questionnaire (P = .04) and 1.0 ± 0.5 points per shock on the mental component score of the 12-Item Medical Outcomes Short-Form Health Survey (P = .04).
Overall, HRQL was not affected by ICD implantation in patients in the Defibrillators in Nonischemic Cardiomyopathy Treatment Evaluation study. Implantable cardioverter defibrillator shock was associated with a reduction in some measures of HRQL, but the effects were unlikely to result in a clinically observable alteration until 5 or more shocks were experienced.
Improving both the duration and the quality of life remains the major goal of medical therapy. Studies have conclusively demonstrated that implantable cardioverter defibrillators (ICDs) reduce mortality when they are used to prevent sudden cardiac death in high-risk individuals.1,2 The benefit of ICDs, however, could potentially be offset by detrimental changes in health-related quality of life (HRQL) that result from device implantation or from shocks aimed at terminating life-threatening arrhythmias. Also, as with any preventive intervention, more individuals will receive an ICD than will derive a survival benefit from its presence. Therefore, the impact of ICD implantation and shocks on HRQL warrants consideration and assessment.
Health-related quality of life is defined as the subjective perceptions of an individual regarding his or her health-related physical, psychological, and social functioning and well-being.3 Previous studies of HRQL in ICD recipients have yielded inconsistent results. However, nearly all these patients had suffered a myocardial infarction or had already manifested life-threatening ventricular arrhythmias, and neither of these conditions is necessary under current implant guidelines. Also, to our knowledge, no data are available regarding HRQL in patients with nonischemic cardiomyopathy.
The Defibrillators in Non-Ischemic Cardiomyopathy Treatment Evaluation (DEFINITE) study demonstrated that ICDs significantly reduce the risk of sudden cardiac death and result in a trend toward decreased overall mortality in patients with symptomatic systolic dysfunction and ventricular arrhythmias that are not caused by coronary artery disease.4 The purpose of the present study was to assess the impact of ICD implantation and shocks on HRQL of patients in the DEFINITE study.
The DEFINITE study was a randomized trial comparing ICDs and standard medical therapy in patients with (1) a left ventricular ejection fraction of 35% or less that was not caused by coronary artery disease, (2) a history of symptomatic heart failure, and (3) either nonsustained ventricular tachycardia or 10 or more premature ventricular depolarizations per hour. A total of 458 patients were enrolled in the study; 229 in each arm. Informed consent was obtained from all patients, and the study was approved by the institutional review boards at all sites. Implantation of the ICD was associated with a 35% reduction in all-cause mortality (P = .08) and an 80% reduction in sudden death as a result of arrhythmia (P = .006).4
The HRQL was measured using version 1 of the 12-Item Medical Outcomes Study Short-Form Health Survey (SF-12) and the Minnesota Living with Heart Failure Questionnaire (MLHFQ). The HRQL questionnaires were self-administered at baseline, 1 and 3 months after randomization, and every 3 months thereafter up to 63 months. The results of the SF-12 are expressed as a physical component score (PCS) and a mental component score (MCS).5 Scores for each component range from 0 to 100, with lower scores reflecting a poorer HRQL. A recall period of 30 days was used for the purposes of this study.
The MLHFQ was designed to measure the effects of heart failure and its treatments on an individual's HRQL. The MLHFQ asks the individuals to indicate, using a scale ranging from 0 to 5, how much each of 21 symptoms prevented them from living as they desired during the past 30 days. Total score may be derived by summing the responses. The MLHFQ has 2 subscales. The physical scale is composed of 8 questions, with scores ranging from 0 to 40. The emotional scale is composed of 5 questions, with scores ranging from 0 to 25 points. Higher scores reflect a poorer HRQL. For the purposes of the analyses, the subscales of the MLHFQ were used instead of a total score because of (1) a high proportion of missing responses on questions pertaining to work and sex and (2) the reported high correlation between total score and each subscale.6
The ICD shocks were assessed at each follow-up visit or when indicated by symptoms. The shocks were classified by the events committee based on associated symptoms and standard criteria from intracardiac electrogram analysis. All shocks were considered in the analysis because of the hypothesis that shocks may be similarly perceived by the patient and because of the insufficient individual numbers of appropriate and inappropriate shocks.
Comparisons of baseline demographic and clinical variables between the 2 treatment groups were made using χ2 tests for dichotomous variables or t tests for continuous variables. Two-stage mixed-effects hierarchical linear regression was used to model changes in HRQL during the trial.7 In the first stage, a within-patient model representing patients' HRQL changes during the trial was formulated. In the second stage, a between-patient model representing the differences among patients in their HRQL change rates and baseline status was fitted. For patients who did not provide complete data, missing months were treated following a full information restricted maximum likelihood estimation approach.8 The analyses controlled for baseline differences and predetermined characteristics that have a potential association with HRQL, including sex, age, New York Heart Association class, race, ejection fraction, duration of congestive heart failure, and history of atrial fibrillation. The covariates were entered into and removed from the model in a stepwise fashion at the group level, with α = .05 and α = .10 as criteria for entry and removal, respectively.
Changes in HRQL throughout the study were best captured using a post hoc 3-parameter patient-level model that included (1) baseline HRQL at the time of randomization, (2) initial rate of change in HRQL from baseline through the second follow-up visit at approximately 3 months after randomization, and (3) long-term HRQL change rate after the second follow-up visit. Baseline HRQL, initial rate of change, and long-term rate of change in HRQL were used as outcomes in the assessment of ICD effect.
The impact of shocks was also assessed. The average deviation of HRQL between all postshock visits from the preshock trajectory was modeled at the patient level. A predictor corresponding to the number of shocks received before each visit was added to the previously described model. A 2-tailed significance level of .05 was used to judge statistical significance of all analyses. The sample allowed for a 90% power to detect an effect size of 0.25, or half the accepted clinically relevant threshold of ± 0.5 SD of change. All data were analyzed using intention-to-treat principles. Statistical packages included HLM 6.0 (Lincolnwood, Illinois), SPSS 13.0 (Chicago, Illinois), and SAS 9.0 (Cary, North Carolina).
Of the 458 patients enrolled in the DEFINITE trial, 5 did not provide any HRQL data and were therefore excluded from these analyses. Of the remaining 453 patients, 227 were randomized to an ICD and 226 to standard therapy. There were 4042 patient visits, with HRQL data reported from 9.4 ± 5.1 (mean ± SD) visits per patient. The HRQL information was provided for every follow-up visit by 145 patients; data from 1 or 2 visits were missing on 130 patients; and the remaining 178 patients had missing HRQL data from more than 2 visits. No relationship existed between HRQL and varying lengths of follow-up or dropping out of the study. There were no significant differences in age, sex, or New York Heart Association class between patients with complete HRQL data and patients with incomplete HRQL data. Patients without missing data were more likely to be white and have better ejection fractions and less likely to have diabetes than those with missing data (P < .05 for all). Those with complete data were more likely to report a better baseline HRQL. There were no interactions between data completeness and treatment group (P = .20). Baseline characteristics by treatment allocation are presented in the Table. The mean ± SD age of the patients was 58 ± 13 years; 71% of the patients were male; and the mean ± SD ejection fraction was 21% ± 6%. Fewer patients assigned to the ICD arm had a more than 1-year history of heart failure, but the groups were otherwise similar in terms of other characteristics, including HRQL scores.
Unadjusted MLHFQ and SF-12 scores by treatment group were evaluated. None of these pairwise comparisons reached significance, indicating no detectable difference in HRQL between the ICD and the standard therapy groups during this period (Figure 1). In the time-adjusted regression models, there were overall significant improvements in the PCS and MCS of the SF-12 and the emotional and physical scales of the MLHFQ within the entire cohort between the time of randomization and the time of the second follow-up visit (approximately 3 months, P < .001 for all comparisons). After the second follow-up visit, however, scores on the physical scale of the MLHFQ and the MCS of the SF-12 showed a slow but significant decline in HRQL toward baseline values (P ≤ .001 for both). There was no significant change in the PCS scale of the SF-12 (P = .12) or in the emotional scale of the MLHFQ (P = .59) from the time of the second follow-up visit to the end of entire observation period.
Changes in HRQL by study arm throughout the course of the trial were examined. For the SF-12 (Figure 2), the MCS scores improved significantly from enrollment to the time of the second follow-up visit (approximately 3 months) for ICD patients only (+1.0 ± 0.2 points per month, P < .001), with a nonsignificant trend toward improvement in the patients who were receiving standard therapy (+ 0.3 ± 0.2 points per month, P = .07; P = .01 for interaction). After this initial improvement, both standard therapy and ICD groups experienced a slow but significant decline in MCS scores toward baseline values (−1.8 ± 0.3 points per year for standard therapy, P = .01; −0.7 ± 0.3 points per year for ICD, P < .01), with no significant difference in long-term MCS scores between the randomized groups (P = .89). The PCS scores on the SF-12 showed significant short-term (ie, approximately 3 months) improvements for both standard therapy and ICD groups (+1.0 ± 0.2 and +1.2 ± 0.2 points per month, respectively, P < .001 for both). After these initial improvements, there was no significant change in PCS scores in either group (−0.4 ± 0.3 points per year, P = .20 for standard therapy; −0.2 ± 0.2 points per year for ICD, P = .40) and no significant difference in long-term PCS scores between the groups. Overall, these results demonstrate no significant differences in long-term HRQL between randomized groups on either subscale of the SF-12.
For the MLHFQ (Figure 3), patients randomized to standard therapy or to ICD had a significant improvement in the emotional and physical scale scores on the MLHFQ from the time of enrollment until the time of the second follow-up visit (ie, approximately 3 months; emotional scale: −0.7 ± 0.1 points per month for standard therapy; −0.8 ± 0.1 points per month for ICD; physical scale: −1.6 ± 0.2 points per month for standard therapy; −1.5 ± 0.2 points per month for ICD; P ≤ .01 for all). After the initial improvement, scores on the emotional scale remained stable in both groups (P > . 5), while scores for the physical scale decreased equally toward baseline values (+0.7 ± 0.3 points per year for standard therapy; +0.6 ± 0.2 points per year for ICD, P ≤ .01 for both). Overall, these results demonstrate no significant differences in long-term HRQL between randomized groups on either subscale of the MLHFQ.
Potential interactions of HRQL and patient variables were evaluated to identify particular subgroups of patients who experienced a decline in HRQL after device implantation. No differences in HRQL were observed after stratification by age, sex, New York Heart Association class, race, duration of congestive heart failure, ejection fraction, or history of atrial fibrillation (P > .05 for all). Death during the study period was not associated with premorbid changes in HRQL, nor did it interact with randomized group. In sum, these results imply that clinical variables cannot be used to identify patients who are likely to exhibit a decline in HRQL after ICD implantation.
The HRQL data were available after 171 shocks in 77 patients. Twenty-seven patients received 1 shock; 19 received 2 shocks; and 25 received more than 2 shocks. No patients experienced clusters of multiple shocks. The median time between shock and subsequent HRQL assessment was 32 days (range, 1-170 days).
Pairwise comparisons of HRQL scores between shocked and nonshocked patients with ICDs did not demonstrate any significant differences between the 2 groups (P > .05 for all). However, analyses that compared intraindividual preshock and postshock HRQL revealed that, overall, shocks were associated with a reduction in the scores on the mental and emotional components of the HRQL, with a decline of 0.5 ± 0.2 points per shock on the emotional scale of the MLHFQ (P = .04) and 1.0 ± 0.5 points per shock on the MCS of the SF-12 (P = .04). These changes, while statistically significant, represented only small changes in overall score (Figure 4). There was no evidence to support a negative impact of shock on either of the physical components of HRQL (+0.7 ± 0.4 points per shock on the physical component of the MLHFQ, P = .08; +0.3 ± 0.3 points per shock on the PCS of the SF-12, P = .23). There was no significant difference in the impact of shock on HRQL between assessments performed 30 days or less after shock or more than 30 days after shock.
Interactions were evaluated between baseline characteristics of the patients in the ICD group and changes in HRQL after ICD shock. Female sex was associated with postshock decline in the scores on the physical components of both scales (P < .01 for MLHFQ; P = .05 for SF-12). Male sex, older age, duration of congestive heart failure, history of atrial fibrillation, and white race were all associated with a differential impact of shock on only 1 of the 4 scales used in the analyses.
The major finding of our study is that in patients with nonischemic dilated cardiomyopathy and no previous sustained ventricular tachyarrhythmias, ICD implantation is not associated with any significant long-term alteration in HRQL compared with medical therapies alone. However, the delivery of ICD shocks was associated with a statistically significant decrease in the scores on the emotional and mental scales of the HRQL measures used in this study. This impact of shocks is unlikely to reach a clinically observable alteration in HRQL until 5 or more shocks are experienced.
The baseline scores of HRQL in the DEFINITE cohort compare favorably with those in other heart failure populations.9 Interestingly, the largest changes in HRQL observed in this study occurred in both groups in the first months after randomization. While this improvement may reflect the favorable effects of being closely followed up in a clinical trial, an alternative explanation may be that patients were enrolled at a nadir in their HRQL.10
While some previous studies have suggested that ICD implantation may be associated with a decline in physical functioning, mood disturbances, and anxiety related to shocks, others have observed either no effect or an actual improvement in HRQL.11- 15 Three randomized studies of ICDs that also evaluated HRQL yielded inconsistent results. The Coronary Artery Bypass Graft Patch Trial found that ICD implantation was associated with poorer HRQL in ICD recipients compared with controls, while the Antiarrhythmics Versus Implantable Defibrillator study found no discernible difference between the ICD- and amiodarone-treated groups.16,17 In contrast, the Canadian Implantable Defibrillator Study trial demonstrated that ICD recipients experienced superior HRQL in terms of psychological distress, psychological well-being, energy level, physical mobility, sleep disturbances, and lifestyle impairment.18 Major differences in patient populations, HRQL measures, and treatment protocols for the control arms of these studies likely account for some, if not all, of the observed disparities. More recently, the Sudden Cardiac Death in Heart Failure investigators published the cost-effectiveness data from that trial. Although no formal HRQL analysis was presented, patients randomized to the ICD group had no difference in utilities using a time trade-off instrument compared with the amiodarone- or placebo-treated patients.19
While individual studies have presented conflicting results, broader reviews have concluded that treatment with the ICD is equal to or better than treatment with antiarrhythmic medications on most HRQL indicators. A meta-analysis concluded that virtually all of the burden associated with the ICD is attributable to the presence of ventricular tachyarrhythmias and not to the device itself and suggests that some of the smaller individual studies may have overestimated the negative psychological effects of the implant.20
The experience of shock remains a distinguishing characteristic of ICD recipients over and above the device implantation itself. Nonetheless, the changes in the scores on the mental and emotional scales seen with a single shock in our study were small and unlikely to reach the threshold for a clinically observable change despite the observed statistical significance. The psychological distress due to shock may be more likely to occur after the additive effects of multiple shocks. As in the current study, trial data indicate that even 1 shock is associated with reduced HRQL.17 Other studies, however, have suggested that 5 or more shocks define the threshold for a decreased HRQL outcome.16,18,21 Of note, the women in our study reported a greater decrease in the physical scores on both HRQL scales after a shock. While sex may play a role in adjustment difficulties after ICD implantation, the sex-specific reaction to defibrillation reported in our study is novel and requires further evaluation.22 These results suggest that enhanced algorithms aimed at reducing inappropriate shocks may improve HRQL and support the potential benefits of programming interventions designed to terminate more stable ventricular tachyarrhythmias first with antitachycardia pacing.23
The present study provides a unique analysis of the impact of shocks. Whereas previous studies have compared HRQL between those patients who did and did not receive ICD shocks, such an approach cannot discern cause from effect. As evidence supports the contention that psychological disturbances may increase the risk for arrhythmic events, the more appropriate comparison performed in our study used average preshock and postshock HRQL levels within each individual patient, thereby resulting in a better estimation of the impact of shocks themselves. Furthermore, as changes in clinical variables such as heart failure status may affect both HRQL and propensity for arrhythmias, the current methodology is better suited to analyze such potential confounders.
Our study has certain limitations. The instruments used in this study have been designed to measure a patient's general health and the impact of heart failure symptoms on HRQL. These scales do not have specific questions pertaining to ICD implantation or to the impact of shocks and may be particularly insensitive in capturing the anxiety related to ICD discharges. Furthermore, these scales have not been validated in the ICD population and therefore may be insensitive to changes that result from device implantation or shocks. However, because no validated measures of HRQL specifically designed for ICD recipients were available at the inception of the DEFINITE study, and because similar global and disease-specific measures of HRQL have been used in this population in previous studies, the use of these scales in the present study appears appropriate until better measures become available.24,25 Also, caution must be exercised in correlating statistically significant findings with clinically recognizable differences in HRQL, as the small population-based alterations observed in this study may not be clinically relevant in a single individual. Finally, our data were derived from individuals who met criteria for the DEFINITE trial and may not pertain to other populations for which ICD implantation is indicated.
Correspondence: Rod Passman, MD, MSCE, Cardiology Division, Northwestern Memorial Hospital, 201 E Huron, Suite 10-240, Chicago, IL 60611 (email@example.com).
Accepted for Publication: June 29, 2007.
Author Contributions: Dr Passman had full access to all of the data and takes full responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Subacius, Schaechter, and Kadish. Acquisition of data: Schaechter, Howard, and Kadish. Analysis and interpretation of data: Passman, Subacius, Ruo, Sears, and Kadish. Drafting of the manuscript: Passman, Subacius, Ruo, Howard, and Sears. Critical revision of the manuscript for important intellectual content: Passman, Subacius, Ruo, Schaechter, Sears, and Kadish. Statistical analysis: Subacius. Obtained funding: Kadish. Administrative, technical, and material support: Schaechter and Howard. Study supervision: Schaechter and Kadish.
Financial Disclosure: Dr Sears has acted as a consultant for Medtronic and Boston Scientific and has received honoraria from Medtronic, Boston Scientific, and St Jude Medical and research grants from Medtronic and St Jude Medical.
Funding/Support: This study was funded by St Jude Medical. Dr Passman is funded by grant K23 HL068814-01 from the National Institutes of Health.
Role of the Sponsor: St Jude Medical had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.
Previous Presentations: Abstracts were presented at meetings of the American Heart Association Scientific Sessions; November 9, 2004; New Orleans, Louisiana; and Heart Rhythm Society Scientific Sessions; May 6, 2005; New Orleans.