Decision‐making regarding primary prevention implantable cardioverter‐defibrillators among older adults

Abstract Most implantable cardioverter defibrillators (ICDs) are implanted for the purpose of primary prevention of sudden cardiac death among older patients with heart failure with reduced ejection fraction. Shared decision‐making prior to device implantation is guideline‐recommended and payer‐mandated. This article summarizes patient and provider attitudes toward device placement, device efficacy and effectiveness, potential periprocedural complications, long‐term events such as shocks, quality of life, costs, and shared decision‐making principles and recommendations. Most patients eligible for an ICD anticipate more than 10 years of survival. Physicians are less likely to offer an ICD to patients ≥80 years of age given a perceived lack of benefit. There is a dearth of data from randomized clinical trials addressing device efficacy among older patients; there is a need for more research in this area. However, currently available data support the use of ICDs irrespective of age provided life expectancy exceeds 1 year. Advanced age is independently associated with complications at the time of device placement but not the risk of device infection. The risk of inappropriate shock may be comparable or lower than that of younger patients. While quality of life is generally not adversely impacted by an ICD, a subset of patients experience post‐traumatic stress disorder. ICDs are cost‐effective from societal and health care sector perspectives; however, out‐of‐pocket costs vary according to insurance type and level. Shared decision‐making encounters may be incremental and iterative in nature. Providers are encouraged to partner with their patients, providing them counsel tailored to their values, preferences, and clinical presentation inclusive of age.

purposes of the current manuscript, "advanced" or "older" age is generally defined as chronologic age ≥65 years. Biologic age may nonetheless diverge from chronologic age; some individuals 85 years of age may have biology similar to that of individuals 65 years of age and vice versa. Given the number of Americans ≥65 years of age is expected to increase from 46 million in 2014 to 74 million in 2030, 4 the number of older patients eligible for ICDs may correspondingly increase over time. Nonetheless, older patients have historically been less likely to receive an ICD than their younger counterparts (odds ratio [OR] 0.89, 95% confidence interval [CI] 0.87-0.91 per 5-year increase in age). 5 Low rates of ICD use among older patients are likely multifactorial and in part related to the lower likelihood of clinicians counseling their patients about ICDs, which decreases with advancing patient age >80 years (OR 0.90, 95% CI 0.89-0.92, per 1-year increase ≥80 years). 6 Current professional guidelines indicate that clinicians should engage eligible patients in shared decision-making about ICDs. 1 Specifically, device effectiveness, safety, and potential complications should be discussed in light of patients' health goals, preferences, and values. Notably, advanced age is not an exclusion criterion for ICDs.
Indeed, the 2017 guideline provided the following recommendation regarding older patients that was based on the results of an independently conducted systematic review: "For older patients and those with significant comorbidities, who meet indications for a primary prevention ICD, an ICD is reasonable if meaningful survival of greater than 1 year is expected." 1 However, this recommendation was based on observational data, and more data on the efficacy and safety of ICDs, preferably from randomized clinical trials, are needed in older patients. In 2018, the Centers for Medicare and Medicaid Services began to require that patients participate in shared decision-making as a condition for reimbursement for primary prevention ICDs. 7 To assist clinicians preparing for guideline-indicated and payer-mandated shared decision-making with older patients considering a primary prevention ICD, the intent of this article is to summarize patient and provider attitudes toward device placement, device efficacy and effectiveness, periprocedural complications, long-term events such as shocks and generator replacement, as well as additional considerations such as quality of life and costs. We then provide direction on how to best guide older adults toward a value-concordant decision regarding implantation of primary prevention ICDs.

| PATIENT AND PROVIDER ATTITUDES TOWARD DEVICE IMPLANTATION
Several key features of patients' perspectives toward device placement have been described. Despite having symptomatic heart failure, most patients eligible for an ICD anticipate more than 10 years of survival. 8 Many older patients who go on to receive an ICD may express a clear desire to prolong their lives and the importance of ICDs in that effort. They may do so without considering competing health risks unrelated to sudden death. 9 Some patients prefer to focus on their current health status rather than its future trajectory. Asymptomatic patients with satisfaction regarding their quality of life may in fact choose against ICD implantation, noting they will reconsider it if their symptoms worsen in the future. 10 These patients should be educated about the fact that in the majority of patients, sudden cardiac death occurs in the absence of symptoms. Others may favor not placing an ICD given a negative body image or perceptions regarding potential lifestyle changes such as exercise or sex. 11 Importantly, patients' preferences regarding how involved they wish to be in the decision to receive an ICD span the full spectrum. While some wish to abdicate the decision entirely to their physicians, others desire to play a more active role. 12 Studies on physicians' attitudes toward device implantation are sparse. Available data indicate cardiologists are generally more aware of guideline indications for device implantation than primary care physicians, 13 leading to a potential gap in referral for device consideration for some eligible patients. In general, physicians are less likely to offer an ICD to patients ≥80 years of age 14 given a perceived lack of benefit. 13 The role of shared-decision-making especially in such patients could not be overemphasized.

| EFFICACY AND EFFECTIVENESS
Landmark clinical trials demonstrated ICD efficacy in the reduction of mortality among patients with LV systolic dysfunction. Although older age was not an exclusion criterion, older patients were underrepresented in these trials. Consequently, several meta-analyses of the randomized data were performed (Table 1). [15][16][17] In one meta-analysis, the reduction in mortality among patients older than 60 years of age was significant (hazard ratio (HR) 0.75, 95% CI 0.61-0.91) but less pronounced compared with younger patients (HR 0.65, 95% CI 0.50-0.83). 15 A second meta-analysis yielded similar findings among patients ≥65 years of age (HR 0.66, 95% CI 0.50-0.87). Among patients ≥75 years of age, the survival benefit remained but was attenuated (HR 0.73, 95% CI 0.51-0.97). 16 These analyses were performed using trial-level estimates, precluding the possibility of adjustment for differences in comorbidities and medical therapies as well as the evaluation of age in a continuous rather than categorical fashion. Thus, a patient-level analysis of the clinical trials was performed. In unadjusted analyses, ICD recipients were less likely to die than nonrecipients in all age groups. In adjusted analyses, point estimates indicated ICD efficacy persisted but became less pronounced with increasing age. The sample size of patients aged ≥75 years was limited (n = 390), leading to continued uncertainty regarding the survival benefit of ICDs in this age subgroup. 17 The Danish Study to Assess the Efficacy of ICDs in Patients with Non-Ischemic Systolic Heart Failure on Mortality (DANISH) showed the ICD was not efficacious among patients with nonischemic cardiomyopathy (HR 0.87, 95% CI 0.68-1.12). 18 However, when these data were combined with all randomized trials that included patients with nonischemic cardiomyopathy, there was a survival benefit with primary prevention ICDs (HR 0.75, 95% CI 0.61-0.93). 19 These disparate findings have been attributed to the large proportion of patients in the DANISH study receiving cardiac resynchronization therapy, the high use of effective medical therapy, and requiring an elevated pro-BNP level for inclusion in the trial. In this setting, a secondary analysis of the DANISH study demonstrated a linearly decreasing relationship between ICD and mortality with increasing age (HR 1.03, 95% CI 1.00-1.06). Among patients >70 years of age, a survival benefit of ICD placement was not observed (HR 1.05, 95% CI 0.68-1.62). 20 Potential reasons underlying this finding include the aforementioned aspects of the DANISH study as well as a limited sample size of older patients.
Randomized trials have shown that cardiac resynchronization therapy with an ICD is superior to an ICD alone in the reduction of mortality and heart failure exacerbations among eligible patients with left ventricular systolic dysfunction and a widened QRS complex. [21][22][23] Similar to the primary prevention ICD trials, older age was not an exclusion criterion. While meta-analyses of the trials do not report the association of age with device efficacy, subgroups by age were reported in all of the trials. These unadjusted analyses did not demonstrate a reduction in efficacy with increasing age. In a secondary analysis of MADIT-CRT, the risk of death or hospitalization or hospitalization with heart failure was reduced among patients ≥75 years of age (n = 331, HR 0.57, 95% CI 0.37-0.87) and 60 to 74 years of age (n = 941, HR 0.55, 0.41-0.72) but not among those <60 years of age (n = 548, HR 0.80, 95% CI 0.52-1.23). 24 These findings were driven by a reduction in rehospitalization more so than mortality. 24 Similar results were observed in the Comparison of Medical Therapy, Pacing, and Defibrillation in Heart Failure trial. 21 More data are needed on CRT-D vs CRT-P, especially in older patients. A pilot study funded by the National Institutes of Health is currently underway to compare these two therapies in patients 75 years of age and older. 25 If proven feasible, a larger trial with adequate statistical power to compare survival will follow.
Randomized trials often enroll patients with fewer comorbidities than those seen in general clinical practice. Moreover, they are typically conducted in highly monitored and controlled settings. Thus, whether the randomized data apply to real-world clinical practice has been an area of keen interest. Rigorous analyses comparing registry data to trial data have shown that patients eligible for ICD therapy based on trial criteria and yet who are seen in routine clinical practice have better survival with an ICD. Specifically, there was no difference in survival between ICD recipients in the registry and the trials whether patients were comparable to those enrolled in MADIT-II (HR 1.06, 95% CI 0.85-1.31) or SCD-HeFT (HR 1.16, 95% CI 0.97-1.38). 26 The median ages of MADIT-II-like and SCD-HeFT-like patients were 68 and 67 years, respectively. In a secondary analysis of patients 65 years and older, the findings were comparable to the primary results. These data generally support the use of ICDs in routine clinical care. However, there was a dearth of data on patients >80 years of age, limiting the degree to which the findings can be extrapolated to patients in this advanced age group.
Observational analyses of real-world datasets to understand the role of age in ICD real-world effectiveness have been undertaken. In a cohort of 965 patients with ischemic or nonischemic cardiomyopathy, ICD therapy was shown to be associated with a lower risk of death (0.69, 95% CI 0.50-0.96). This relationship was consistent after T A B L E 1 Meta-analyses of the influence of age on ICD efficacy ≥75 years (n = 269). 27 Similarly, in the American Heart Association Get With the Guidelines-Heart Failure registry linked with Medicare claims, ICD therapy was associated with a lower risk of death over 3 years after device placement up to 84 years of age. 28 In a metaanalysis of predominantly observational data, a survival advantage of ICD vs no ICD was seen among older patients (HR 0.75, 95% CI 0.67-0.83). 29 In the Swedish Heart Failure Registry, compared with ICD nonreceipt, ICD receipt was associated with a lower risk of death interaction P = .86). 31 In addition to age, medical comorbidities, which are more prevalent with increasing age, influence the likelihood of patient survival irrespective of ICD placement. Statistical models predicting short-and long-term survival related to ICD implantation using data from either trials 32,33 or observational datasets 34,35 have been developed. These models show that a number of comorbidities influence patient survival, including atrial fibrillation, chronic kidney disease, chronic obstructive pulmonary disease, and diabetes mellitus ( Table 2). Among these, chronic kidney disease has particularly been shown to have a big influence. 35 The risk of death after primary prevention ICD placement is in fact proportional to the severity of chronic kidney disease. 36 It is most pronounced among patients with end-stage renal disease on dialysis; a total of 22.5% of ICD recipients on dialysis die within 1 year of device implantation. 37 Notably, the acuity of illness at the time of device placement may play a role in overall device effectiveness; patients hospitalized with heart failure or other comorbidities may not derive as much survival benefit as those receiving the device in an elective setting. 38 In a cohort derived from the National Cardiovascular Data Registry's ICD Registry linked with Medicare claims, geriatric conditions such as frailty and dementia were present in more than 10% of patients ≥65 years of age receiving an ICD and were associated with more than twice the risk of death within 1 year of device implantation (22% among patients with frailty, 27% among patients with dementia, 12% in the overall cohort). 39 Multimorbid patients, including those with chronic obstructive pulmonary disease and diabetes in conjunction with frailty and/or dementia, had still worse 1-year death rates (dementia with frailty, 29%; frailty with chronic obstructive pulmonary disease, 25%; frailty with diabetes, 23%). 39 In part a consequence of the substantial comorbidity burden and the concomitant risk for nonsudden death, one in two patients receiving an ICD after the age of 65 either die or are admitted to a hospice facility within 5 years of device receipt. Factors associated with reduced time to hospice enrollment include advanced age, heart failure class, and an ejection fraction <20%. 40 These data suggest that there may be opportunities to integrate palliative care principles into the decision to implant an ICD and to seek specialized palliative consultation even if lifeprolonging therapies such as ICDs are desired. 40 While prior randomized and observational studies can inform current clinical care, several ongoing studies have the potential to shape our understanding of ICD efficacy and effectiveness moving forward.
To address our gap in knowledge regarding ICD efficacy in older patients, a multisite, randomized clinical trial comparing ICD implantation and optimal medical therapy to optimal medical therapy alone among patients 70 years of age and older is underway in the Department of Veterans Affairs health care system. 41  as noncardiovascular conditions such as chronic lung disease, cerebrovascular disease, diabetes mellitus, and kidney disease are also associated with poorer survival. In the absence of a comparison group, the risk reduction in mortality associated with device replacement is not yet known. 51  Their values and preferences are elicited. Clinicians in turn take on a role of advisor or partner rather than a paternalistic arbiter. 59 This theoretical framework encompasses and supersedes the legalities of informed consent by aligning medical therapies with patient preferences and values (Figure 1).

| ADDITIONAL CONSIDERATIONS
At the outset of shared decision-making, it may be helpful to characterize patients' desired level of engagement in the decisionmaking process. Some patients may prefer that physicians make the decision for them; others may wish to make the decision with little or no physician input. Our experience has been that most patients lie somewhere along the spectrum of these extremes. Among patients who decide to partner with their provider in the selection of treatment options, assessing the level of detail desired in the discussion is an additional important and at times iterative step. Select patients may wish to understand precise values and percentages. In such instances, the detailed discussion above may prove particularly useful.

| CONCLUSIONS
There is a paucity of data regarding ICD efficacy among older patients.
The balance of available data, whether they be pooled from randomized trials or derived from observational datasets, supports the use of ICDs irrespective of age provided life expectancy exceeds 1 year. Advanced age has been shown to be an independent risk factor associated with complications at the time of ICD implantation; however, advanced age per se is not associated with increased risk of device infection. Rather, its impact may be indirect via comorbidities that are prevalent with F I G U R E 1 The shared decisionmaking theoretical framework highlights key elements of shared decision-making: (a) involvement of at least the patient and provider, (b) meaningful sharing of information, and (c) consensus building and agreement on the treatment to implement