The impact of implantable cardioverter-defibrillator implantation on health-related quality of life in the DANISH trial

Abstract

Introduction

The goal of treating heart failure (HF) is to reduce symptoms, prevent hospitalization, and improve survival, and thus to improve both quantity and quality of life (QoL).¹ Heart failure is associated with extensively impaired health-related quality of life (HRQoL) compared with other chronic diseases with significant impact on all dimensions of HRQoL.^2–5 However, most trials of new treatments for HF have used composite endpoints such as HF hospitalization or cardiovascular (CV) death as their primary outcome, while less, little or no attention has been paid to effects on symptoms or HRQoL. From the patient perspective, well-being may be just as important as survival and some patients are willing to trade life expectancy for a better HRQoL.⁶ Thus, patients, physicians, and societies, such as the European Society of Cardiology and the American Heart Association, have recently emphasized the importance of evaluating patient-reported outcomes when assessing the efficacy of therapeutic interventions in CV medicine.^7,8

Prophylactic implantation of an implantable cardioverter-defibrillator (ICD) is a Class 1 recommendation for symptomatic HF patients with reduced left ventricular ejection fraction (LVEF <35%).^9,10 However, the evidence of benefits gained from ICD implantation is primarily based on patients with ischaemic HF, where an association of ICD implantation with reductions in rate of sudden cardiac death (SCD) and total mortality has been demonstrated.^11,12 The Danish Study to Assess the Efficacy of ICD in Patients with Nonischemic Systolic Heart Failure on Mortality (DANISH) trial demonstrated that ICD implantation in patients with systolic HF not due to ischaemic causes was not associated with overall improved survival.¹³

The effect of ICD implantation on HRQoL has not been consistent in previous trials.¹⁴ Thus, it is uncertain whether HRQoL perceptions are primarily altered by negative factors such as the implantation procedure, anxiety related to possible shock therapy, and increased disease-awareness or whether this is counterbalanced by positive factors such as reassurance due to presumed prevention of SCD or enabling return to an autonomous life.^14,15

In this study, we assessed the impact of prophylactic ICD implantation on the pre-specified secondary endpoint HRQoL in the DANISH trial.¹³

Methods

Study design and subjects

In brief, the DANISH trial was a primary preventive, randomized, unblinded, multicentre study that assessed the efficacy of prophylactic ICD implantation in patients with symptomatic non-ischaemic systolic HF.¹⁶ The 1116 patients in the DANISH trial were recruited from the five ICD implanting centres in Denmark from 7 February 2008 to 30 June 2014, and randomized to ICD or control group. Criteria for participation were New York Heart Association (NYHA) functional Class II or III, [or IV if cardiac resynchronization therapy (CRT) was planned], LVEF ≤35%, and increased level (>200 pg per millilitre) of N-terminal pro-brain natriuretic peptide (NT-proBNP). Patients were required to receive optimal medical therapy in accordance with guidelines for their HF if tolerated, including angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers, beta-blockers and an aldosterone antagonist. The randomization (1:1) was performed by a web-based system and if randomized to ICD, implantation was conducted within 4 weeks from randomization. Patients randomized to the ICD group with clinical indication for a CRT system received a CRT defibrillator (CRT-D) device, whereas patients randomized to the control group with clinical indication for CRT received a CRT pacemaker (CRT-P) system. The DANISH trial showed an age interaction on mortality for ICD vs. control, and therefore, we also included subgroup analyses of HRQoL to investigate if this interaction was evident within HRQoL as well.

Patients were enrolled only after providing informed consent. The DANISH trial is registered at ClinicalTrials.gov with the identifier NCT00542945 and the study was approved by the regional scientific ethics committee for the capital region (H-D-2007-0101).

Quality of life measures

Minnesota Living with Heart Failure Questionnaire (MLHFQ) was prespecified as the principal HRQoL measure.¹⁶ The questionnaire has good psychometric properties¹⁷ and is one of the most widely used instruments to assess HRQoL in HF patients globally with high validity, sensitivity, and reliability,¹⁸ especially when used in clinical trials with at least 6 months follow-up.¹⁹

MLHFQ is a 21-item self-administered, disease-specific instrument rated on a six-point Likert scale (0–5) representing the degree of impact of HF on HRQoL and was completed at the hospitals. The MLHFQ questionnaire asks for the facets of the disease that prevented the patients from living as desired with a low score indicating desired HRQoL. The instrument is divided into two subscales, emotional (five domain items) related to psychological distress and physical (eight domain items). The last eight items of the total 21 are included for the overall score (0–105).²⁰ A clinically relevant change is considered to be approximately 5 points.²¹

Studies suggest that the use of a third ‘social’ subscale may provide further refinement of the questionnaire and it is generally accepted that HRQoL is a multifactorial concept consisting of three domains that affect one’s disease—psychological, physical, and social functioning.^5,22 In addition to the original two, emotional and physical, Munyombwe’s social factor²² has been suggested and validated and scored highest among proposed social factors with a high reliability.²⁰ It is scored from the 21 items of the MLHFQ (item 8, 9, 10, 14, 15, and 16).

Statistical analysis

Baseline characteristics were presented as median [interquartile range (IQR)] for continuous variables and frequency (percentages) for categorical variables. Differences between ICD and control group were assessed by t-test, Mann–Whitney U test, and χ² test as appropriate.

The primary objective was to assess changes in HRQoL from randomization to 8 months-visit by MLHFQ in the ICD group vs. control group. We also assessed changes in HRQoL at study visits 14, 20, 26, and 32 months after randomization. For the primary analysis, changes in MLHFQ scores were assessed using a repeated measures mixed effects model with the baseline score as a covariate and CRT, centre and time as factors, and a treatment-by-time interaction.

Missing HRQoL data were imputed with the mean method^23,24 if data were semi-incomplete but with more than half of the 21 MLHFQ questions answered and patients were excluded if less than half were completed (the half-rule).^25,26 Furthermore, MLHFQ overall scores were assessed with and without imputation for missing data due to death. In these analyses, all visits after the date of death were given a MLHFQ score of 105 (worst possible).

Clinically relevant MLHFQ changes were subdivided into improved (MLHFQ decrease ≥5), stable (MLHFQ change between +5 and −5), and worsened (MLHFQ increase ≥5) scores. Furthermore, the impact of ICD implantation on HRQoL was examined in relevant subgroups.

In analyses on the impact of shocks on HRQOL, we used MLHFQ measurements 12 months pre-shock to 4 months post-shock. All statistical analyses were conducted using SAS 9.4 (Cary, NC, USA).

Results

Patient characteristics

Among the 1116 patients enrolled in the DANISH study, 935 HRQoL questionnaires (84%) were collected at randomization (Supplementary material online, S1). The majority of these patients were reassessed at visits during follow-up (Figure 1). The median age at randomization was 63 years (IQR 56–71), 28% were females, the mean LVEF was 24% and 55% were in NYHA class II.

Figure 1

Number of patients with available HRQoL data throughout follow-up. The centred percentages are percentage of baseline. Due to the study design of the DANISH trial with continuous inclusion the percentages do not represent percentages of the eligible patients (some patients only had a inclusion period of 2 years and could not experience visits >24 months). HRQoL, health related quality of life; ICD, implantable cardioverter defibrillator; mths, months.

Open in new tabDownload slide

HRQoL respondents compared with HRQoL non-respondents had shorter median duration of HF [16 months (IQR 8–60) vs. 40 months (IQR 11–90), P < 0.0001] and fewer had pre-existing or planned CRT [526 (56%) vs. 119 (66%), P = 0.02] (Supplementary material online, S2). However, the baseline characteristics of the HRQoL respondents were well balanced with no significant differences between the ICD group and control group (Table 1).

Follow-up and events

The median follow-up was 67.5 months (IQR 48–84). The follow-up period questionnaire completion rate, compared with baseline, was 61% (566/935) at 8 months for the primary analysis, 47% (441/935) at 32 months and 16% (148/935) at 68 months (Figure 1). Of the 935 patients, there was a cumulative response rate of 86% as total of 802 patients completed the questionnaire at least once within 32 months. The primary outcome in DANISH, death from any cause, occurred in 97 (21%) patients in the ICD group vs. 99 (21%) in the control group among those included in the HRQoL analysis (22% vs. 23% in primary DANISH trial). Cardiovascular death occurred in 61 (13%) in the ICD group and in 72 (16%) in the control group.

Termination of ventricular tachycardia by anti-tachycardia pacing occurred in 78 patients (17%) in the ICD group. Among patients in the ICD group, 48 ICD recipients (10%) received appropriate shocks for ventricular fibrillation or rapid ventricular tachycardia. Additionally, 29 patients (6%) received inappropriate shocks. In addition to the ICD, several patients received CRT, with 56% in the control group and 57% in the ICD group.

Health-related quality of life

The overall mean MLHFQ scores at randomization were similar in the ICD vs. control group (37.3 vs. 38.6, n = 935, P = 0.37). From randomization to 8 months, for the patients with 8 months Follow-up visit assessment (n = 566), both groups experienced a significant improvement in MLHFQ: in the ICD group (n = 274) from 35.5 to 26.6 (P < 0.0001) and in the control group (n = 292) from 37.3 to 29.9 (P < 0.0001) (Figure 2A and B).

Figure 2

HRQoL during follow-up in patients randomized to ICD vs. controls, with (A) and without (B) imputation for death. The P-value is calculated by a repeated measures mixed effects model with the baseline score as a covariate and CRT, centre and time as factors, and a treatment-by-time interaction. CRT, cardiac resynchronization therapy; HRQoL, health related quality of life; ICD, implantable cardioverter defibrillator.

Open in new tabDownload slide

At 8 months, changes in MLHFQ overall scores were similar between the ICD group and the control group with a least square mean change of −7.0 vs. −4.2 (n = 586, P = 0.13) in the mixed effects model imputed for death. Results were similar without imputation for death [−8.9 vs. −6.9, (n = 566, P = 0.20)] (Figure 2). At 8 months, a clinically relevant improvement (decrease ≥5 points) was seen in 151 patients (55%) of those assigned to ICD therapy compared with 148 patients (51%) in the control group (P = 0.25). The change in scores only insignificantly favoured patients in ICD group (Supplementary material online, S4).

After 32 months, the change in MLHFQ overall score was 0.8 in the ICD group vs. 1.3 in control group (n = 523, P = 0.82). Results were similar without imputation for death (−6.7 vs. −6.5, n = 441, P = 0.92). At 32 months, 126 patients (58%) in the ICD group and 123 patients (55%) in the control group experienced a clinically relevant improvement (P = 0.58). Complete case analysis, excluding patients with incomplete HRQoL assessments and excluding imputation for death, yielded similar results.

Subgroup analyses of health-related quality of life

Results were generally similar in subgroup analyses of clinically relevant improvement in the ICD group vs. control group by 8 months (Figure 3). Patients with an age ≤70 years had a worse MLHFQ score at randomization compared with those >70 years (39 vs. 34, P = 0.008), however, the clinically relevant improvement was similar between the two age groups (P = 0.85). Duration of HF was the only factor with a nominally significant p value for interaction of 0.01 not accounting for multiple testing, as patients with a HF duration of <18 months improved in the ICD group only (P = 0.01).

Figure 3

Rate of clinical relevant improvement in HRQoL within 8 months in prespecified subgroups. CI, confidence interval; CRT, cardiac resynchronization therapy; GFR, glomerular filtration rate; HRQoL, health related quality of life; ICD, implantable cardioverter defibrillator; LV, left ventricular; NT-proBNP, N-terminal pro-brain natriuretic peptide; NYHA, New York Heart Association.

Open in new tabDownload slide

Regarding social factors patients that were married had an insignificantly better HRQoL at baseline compared with those not married (37 vs. 39, P = 0.18) but after 8 months there was no difference in clinical relevant improvement (52% vs. 53%, P = 0.73). Highest level of completed education did not have a significant impact on HRQoL at baseline. The worst baseline scores were for patients that had completed high school and the best for patients with a higher education above 4 years (44 vs. 36, P = 0.64). After 8 months, 39% of patients that had completed high school had a clinical relevant improvement compared with 59% in patients that had completed a higher education above 4 years, however, the difference was insignificant when compared across all education levels (P = 0.40). Patients with a QRS duration >120 ms had a higher degree of clinical relevant improvement compared with those <120 ms (58% vs. 40%, P < 0.0001).

Subdomains of the MLHFQ were all insignificant measured as change from randomization to 8 months by ICD vs. control [emotional (P = 0.81), physical (P = 0.08), and social (P = 0.11)] (Supplementary material online, S3 and S4). Assessing the impact of CRT, we found a significant clinically relevant improvement within 8 months when comparing CRT-D (n = 160) vs. ICD alone (n = 116) (63% vs. 44%, P = 0.0023) but not between CRT-D (n = 160) and CRT-P (n = 155) (63% vs. 59%, P = 0.57) (Supplementary material online, S3).

Patients experiencing shocks (appropriate and inappropriate)

An appropriate shock was associated with worsening of patients’ HRQoL with five points (IQR −4 to 37, P = 0.09) but only 12 of these patients (25% of the 48 patients who received an appropriate shock) fulfilled the criteria of having measured HRQoL maximum 12 months before and no more than 4 months after. In patients experiencing inappropriate shocks, there was a median improvement of −1 (IQR −10 to 5, P = 0.61) in overall MLHFQ score in the 10 patients fulfilling the criteria (34% of the 29 patients who received an inappropriate shock). The number of patients with multiple shocks was not sufficient to describe the impact of this phenomenon on subsequent HRQoL. Without application of the above mentioned time criteria, appropriate shocks was associated with a median worsening in HRQoL of 4 (IQR −7 to 24, n = 26, P = 0.15), with a median post-shock time to assessment of 124 days (IQR 56–236). For inappropriate shocks, we found a median improvement of −1 (IQR −13 to 6, n = 24, P = 0.76) without time criteria with a median post-shock assessment of 140 days (IQR 72–268).

Discussion

The major finding of the present analysis of this prespecified secondary endpoint of the DANISH trial was that ICD implantation in patients with non-ischaemic systolic HF did not have a significant impact on HRQoL. The HRQoL of patients in both the ICD group and the control group improved significantly from randomization to 8 months and remained stable during follow-up. The outcome of death from any cause was equal in the two groups and therefore imputation of 105 in MLHFQ overall score in the subsequent visits after death affected the two groups equally with similar and insignificant intergroup score changes (Figure 2A and B).

Interestingly, the largest change in HRQoL occurred within the first months in both groups, which has also been found by others.²⁷ The high MLHFQ scores at randomization (Figure 2) might be caused by time of first questionnaire assessment, which was after study enrolment. It is well-established that HRQoL perceptions can be affected in clinical studies by more intense management and extra attention in general, especially at baseline.⁴ This possibly affected the magnitude and direction of change over time but since this was a pattern in both groups the impact was limited.

Implantable cardioverter-defibrillator implantation could be hypothesized to have a positive as well as a negative impact on HRQoL. Procedure complications, device malfunctioning, hospital admittance and check-ups, the sense of life-dependency of the ICD and the inserted ICD constantly reminding patients of their disease or even anticipating or recalling painful and unpredictable appropriate or inappropriate shocks, might lead to psychological distress.^28,29 Considering these potential negative factors it is reassuring that we found no significant difference in HRQoL between the ICD group and the control group. The improvement seen in HRQoL after ICD implantation that counterbalances distress might be explained by the ability of cardiac devices to reduce health concerns and symptoms. Patients may feel reassured and gain peace of mind through the security afforded by the device and its protection against SCD, enabling patients to return to autonomous lives.¹⁴

Previous trials that have evaluated the effect of ICD therapy on HRQoL have shown mixed results.¹⁴ Among primary and secondary prevention studies the AMIOVIRT (Amiodarone vs. Implantable Cardioverter-Defibrillator),³⁰ MADIT II (Second Multicenter Automatic Defibrillator Implantation Trial),³¹ AVID (Antiarrhythmics vs. Implantable Defibrillators Trial),³² SCD-HeFT (Sudden Cardiac Death in Heart Failure Trial),³³ and DEFINITE (Defibrillators in Non-Ischaemic Cardiomyopathy Treatment Evaluation)³⁴ trials did not find any evidence of worsened HRQoL in HF patients with ICD compared with medical therapy. In contrast, CIDS (Canadian Implantable Defibrillator Study)³⁵ reported clear HRQoL benefits in the ICD group, while the CABG-Patch (The Coronary Artery Bypass Graft Patch Trial)³⁶ found that psychological well-being was significantly reduced in patients receiving ICD.¹⁵ A substudy on QoL of the INTRINSIC-RV trial (Inhibition of Unnecessary RV Pacing with AV Search Hysteresis in ICDs) found a significant improvement on all SF-36 subscales from baseline through follow-up but since they did not include a non-ICD group it is not comparable to our study.³⁷ The heterogeneity of results might relate to small cohorts (e.g. only 51 ICD patients were included in the AMIORVIT study³⁰) usage of generic HRQoL questionnaires, differences in study design, inclusion and exclusion criteria, and differences in the use of medical therapy, CRT, and other adjunctive therapies. Some of the initial studies were performed more than 20 years ago and ICD technologies have improved considerably since then. Additionally, earlier studies have not focused solely on non-ischaemic HF.

In the DANISH trial, the effect of an ICD was age-dependent with a cut-off at >70 years³⁸ and a similar age-dependency could be hypothesized for HRQoL. Younger patients had worse HRQoL at randomization, however, comparing HRQoL change from randomization to 8 months with a cut-off of >70 years, there was no inter-group difference.

Treatment with CRT-D compared with CRT-P was not associated with a significant improvement in QoL. In contrast, and consistent with previous studies, we found that CRT-D therapy significantly improved HRQoL compared with those with ICD alone, probably by reducing HF symptoms (Supplementary material online, S3b).³⁹ It is noteworthy that patients receiving CRT had higher NYHA class, which was found to be associated with worse MLHFQ overall scores at randomization. The significant improvement in HRQoL among patients with CRT was mainly due to the high MLHFQ score at randomization as the non-CRT group and the CRT group had equal MLHFQ overall scores after 8 months. The maximum benefit of a CRT-P or CRT-D on HRQoL may therefore be seen in patients with NYHA Class III–IV, as they have worsened HRQoL before device implantation.³⁹

In other subgroup analyses, we only found a significant interaction on HRQoL by patients with HF duration <18 months (Figure 3). It may be that ICD implantation provides a sense of security for patients, and thereby increased HRQoL particularly in patients with a more recent diagnosis of HF.

Implantable cardioverter-defibrillator shocks have been reported to compromise psychological well-being,⁴⁰ while others find no association.³⁷ The results are discordant and might be due to heterogeneity of designs and methods.⁴¹

In our study, there was no significant change in psychological distress the months after shock therapy when compared with the psychological state during the 12 months period before for patients receiving an appropriate or inappropriate shock. Without time criteria but still with reasonable timing of HRQoL assessment, we found the same insignificant pattern. Other studies, however, have suggested that the threshold for developing a significant decrease in HRQoL is five or more shocks and that the effects of a shock are more likely detectable in the first month following a shock.^33,34 However, Whang et al.⁴² found that worsened HF is an independent risk factor for ventricular arrhythmias that require shock therapy. We found that worsened HF is associated with low HRQoL and therefore, the association between HRQoL and shock might be mediated by this relation.

Limitations

This study has several limitations. Firstly, from the 1116 patients randomized in DANISH, 16% were excluded due to missing HRQoL data (Supplementary material online, S1). HRQoL respondents vs. non-respondents generally had a shorter duration of HF and this could potentially introduce a bias and limits the generalizability of the results. The patients included may have had a more positive attitude towards HRQoL assessment and more likely to give positive evaluation of ICD therapy. However, this could also be a chance finding considering multiple testing and the significance level of the interaction and noticeably the median duration of HF was similar between included patients in ICD group and control group minimizing the risk of bias (Table 1). Secondly, we did not have complete information on HRQoL during follow-up and therefore used imputation for semi-incomplete questionnaires. As suggested,⁴³ we did not use last value carried forward as this method is not recommended for variables that change over time as HRQoL. Instead we utilized mean method for semi-incomplete questionnaires and worst-case scenario after death. Notably, complete case analysis yielded similar results. Thirdly, the baseline MLHFQ was assessed after study information, which potentially increased disease attention leading to a higher MLHFQ baseline score. Finally, the dropout rates during follow-up may limit the generalizability of the findings in this study, as the change in HRQoL after 8 months was not based on the entire baseline population. However, results were similar at each study visit and the cumulative response rate was 86% [802 of the 935 patients completed the questionnaire at least once during follow-up (Supplementary material online, S5)].

Major strengths of this study are the large and diverse population of HF, the long follow-up period, that patients were on contemporary optimized medical therapy, that HRQoL was assessed by the disease-specific MLHFQ and the inclusion of a third domain, Munyombwe’s social factor. Another strength of this study is that modern ICD devices with improved algorithms were implanted with the benefit of few inappropriate shocks.¹⁵

Conclusion

In conclusion, prophylactic ICD implantation in patients with non-ischaemic HF was not associated with a negative impact on HRQoL. These findings do not support that effects on HRQoL should impact on the recommendations for ICD implantation.

Acknowledgements

This study was a prespecified substudy of DANISH, which was supported by unrestricted grants from Medtronic, St. Jude Medical, Trygfonden and the Danish Heart Foundation.

Funding

This study was supported by the Independent Research Fund Denmark [7090-00032B].

Conflict of interest: S.P. reports personal fees from Abbott (St. Jude Medical) outside the submitted work. F.G. reports grants and personal fees from Orion Pharma, personal fees from Abbott, other from Corvia, personal fees from Novartis, personal fees from Bayer, personal fees from CARMAT, outside the submitted work. J.H.S. reports personal fees from Medtronic, personal fees from Biotronic, personal fees from AstraZeneca, personal fees from Boehringer Ingelheim, personal fees from Bayer, grants from Gilead , grants from St. Jude Medical, outside the submitted work. J.C.N. reports grants from Novo Nordisk Foundation, outside the submitted work. L.K. reports personal fees from Novartis, outside the submitted work. All remaining authors have declared no conflicts of interest.

References