Background:
The implantable cardioverter defibrillator (ICD) has been identified as a cornerstone in the treatment of life-threatening arrhythmias (1). Despite its proven efficacy in this regard many patients with an ICD will eventually die from a non-cardiac condition, for example cancer, dementia or renal failure. In such circumstances the ICD can effectively mean that patients are exchanging a “natural” death for a more protracted one. Given the current increase in the elderly population coupled with improved survival rates for patients with heart failure (2), disentangling when the increased susceptibility of being shocked at the end-of-life (3,4) outweighs the therapeutic benefit of an ICD remains crucial.
European and American expert guidelines published over the past 5 years have aimed at directing professionals on when and how to discuss ICD deactivation with patients and their families (5-9). In 2010, two expert statements (European Heart Rhythm Association (6) and Heart Rhythm Society (8)) specified professionals should discuss deactivation prior to ICD implantation with the information provided tailored to the patient’s disease progression and goals of care. A number of studies report patients consider pre-implantation (10-12) followed by end-of-life (13-15) as optimal times to discuss ICD deactivation. However, the limited evidence suggests it is rarely achieved-, none of the American patients in Goldstein et al’s study (16), only 13% (n=4) in the Canadian study by Strachan (17) and 3% (n=2) in the British study by Raphael et al. (10) were aware of deactivation as an option at the pre-implantation stage. As their disease progresses many patients are reluctant to deactivate their device for fear of losing that “minimal hope for some eventual curative option” (18). In a recent study by Thylen et al. (15) one third of patients stated they could not decide whether to deactivate their ICD if they were dying from cancer or another serious comorbidity.
Data suggests that many healthcare professionals are uncomfortable (19) and uncertain about initiation of a discussion related to deactivation of the ICD. Reasons offered include their lack of experience with such discussions (20), the “intrinsic nature of the ICD” (19,21) and the unpredictability of patients’ prognosis (22,23). Furthermore the discussion about ICD deactivation was often seen as not integral to routine practice (24) with diversification across countries in the use of deactivation protocols (25,26). To enhance compliance of practice to international expert guidelines, clinical factors which facilitate or inhibit the decision to discuss ICD deactivation require identification.
In the pursuit of a patient-centred approach to care, professionals need to be aware of each patient’s unique information needs and preferences (27,28), have the knowledge and ability to accurately determine prognosis (22) and communicate this effectively in order to maintain the balance between potentially frightening the patient unnecessarily and underplaying their condition (29). These demands often lead professionals to avoid (30,31) end-of-life discussions, creating a “conspiracy of silence” between themselves and the respective patient that ultimately stifles shared decision-making.
A number of studies have explored the perspectives of patients (14,32), professionals within a number of specialities (26,33,34) and next of kin (20,35) to some extent. The most pronounced diversity in opinions by patients (11,14,17) and professionals (24,33) is the acceptability of discussing deactivation prior to implantation. Additional findings included the complexity of the decision to deactivate the ICD (19,33,34) with further research and expert guidance (24) recommended to improve education and practice.
Aims:
The purpose of this case note review was to identify current practice and examine decisions made among patients with an ICD from the time of implantation to final documentation and demise. Specific aims were as follows:
· To determine the frequency of pre-implantation discussions about ICD deactivation in end-of-life ICD recipients.
· To evaluate the frequency of end-of-life discussions and describe characteristics associated with such discussions.
· To determine the frequency of ICD deactivation and ultimate manner of death in end-of-life ICD recipients.
· To identify the factors associated with the decision to deactivate an ICD at end-of-life.
Methods:
Design:
A retrospective case note review of all deceased patients with an ICD or cardiac resynchronisation therapy plus ICD (CRT-D) during a 12 month period (September 2012 and September 2013) within a regional hospital and community trust within Northern Ireland.
Sample:
Eligible patients were those who had their device implanted and attended the regional implantation centre within the trust. The sample was representative of all ICD patients in Northern Ireland as this is the only implantation centre. The sample population was identified electronically by a designated Cardiac Physiologist and respective case notes were retrieved by a trained trust employee.
Setting:
Approval was granted on the 6th June 2013 by the Standards, Quality and Audit Department of a regional hospital and community trust within Northern Ireland. Between January to December 2012 a total of 64 new and 37 replacement biventricular devices (CRT-D) in addition to 110 new and 47 replacement ICDs were implanted. No deactivation policy was in place to cover the documentation of a discussion prior to implantation or indicators warranting a discussion through-out a patient’s illness.
Data Collection:
A standardised protocol for data collection was developed after review of the literature (20, 36, 37) and under expert guidance of a cardiologist and experienced researcher. Data were collected by a highly trained cardiovascular nurse with heart failure expertise to promote accuracy and reliability. The following information was collected: demographics (i.e., age, gender), clinical characteristics (i.e., ischaemic aetiology of heart failure, New York Heart Association class (NYHA), diagnosis of malignancy), haemoglobin, sodium and NT-pro b-type natriuretic peptide (NT-pro BNP) levels, device related characteristics (i.e., type of device, indication, duration of implant, frequency of shocks), end-of-life discussion (i.e., frequency, professional involved, next of kin present, topics discussed, time from discussion to death, associating factors) and details of death ( i.e., device status, location of deactivation, type of death, location of death). A copy of the standardised protocol for data collection can be viewed in Appendix 1.
Given the relevance of aetiology of terminal illness (18), the frequency of ischaemic aetiology of heart failure and that of malignancy were both captured. Factors preceding death (38) were kept broad to allow a comprehensive examination of those which may have affected the deactivation decision (i.e., specialist palliative care referral, do-not-resuscitate order, hospice admission (39), advance directive or anticipatory care planning). Finally death was classified into one of four exclusive categories: sudden cardiac, non-sudden cardiac, sudden non-cardiac and non-sudden non-cardiac as per Goldstein, et al. (20) and based on published definitions (40).
As not all the information was available within the medical charts, written documentation was augmented with validated electronic health records, for example the Cardiovascular Information Management System (CVIS) software is a specialist database used by healthcare professionals to follow-up heart failure patients. Information on the cause and location of death was confirmed by communication between the trained cardiovascular nurse and all General Practitioners.
Analysis
Both descriptive statistics and correlative analysis were conducted using SPSS 19.0 (SPSS Inc, Chicago, IL, USA). Proportions among groups were compared using Chi-square. Independent samples T-test or Mann-Whitney U were used to compare means between relevant groups. Given the number of comparisons done to examine factors associated with the decision to deactivate the ICD at end-of-life, the p value was adjusted using the Bonferroni adjustment with the final p value for significance < 0.004.
Results:
A total of 59 patients were identified meeting the inclusion criteria. From these, 44 case notes were able to be retrieved and reviewed. Ten case notes could not be accessed due to logistical reasons, i.e. medical records could not be found and five due to medico-legal concerns i.e. internal hospital audit or post-mortem (Refer Figure 1).
Patient Characteristics
Patient characteristics are described in Table 1. The final sample consisted of case notes from 44 deceased patients with an ICD or CRT-D. The majority of patients were male (86%) and the mean age at time of death was 73 years. Most patients (86%: n=38) had an ischaemic aetiology of heart failure with less than a third of patients (29%: n=13) diagnosed with a malignancy prior to death. More patients were classified as heart failure NYHA III (34%: n=15) compared to patients who had heart failure NYHA IV symptoms (18%: n=8). Mean haemoglobin level was 11.5g/dl, mean sodium level 134.3mmol/l and mean NT-pro b-type natriuretic peptide level was 5323 ng/L. For both ICD and CRT-D devices the mean duration of implant was 50 months.
Aim 1: To determine the frequency of pre-implant discussions about ICD deactivation in end-of-life ICD recipients
There was no record of any patients having received pre-implantation information on deactivation.
Aim 2: To evaluate the frequency of end-of-life discussions and describe characteristics associated with this discussion.
Twenty-three patients (52%) had an end-of-life discussion, which was initiated most commonly by a cardiologist or general physician (Table 2). On one occasion the discussion was initiated by a heart failure nurse. Patients’ next of kin were present during fourteen of the documented discussions. Topics covered were the specific functions of the device, prognosis and possible future treatment options, for example intravenous medications. Of the 23 discussions, only 17 included ICD deactivation. In one discussion, only ICD deactivation was covered, whereas in 16 (36.4%) others, all three topics were covered. Details of the topics covered during discussion with one patient were not available. Deactivation of the device prior to death (n=16) occurred only in cases where an end-of-life discussion had taken place.
The median time from discussion until death was 7 days with more than half of the discussions (64%: n=14) associated with a subsequent do-not-resuscitate order. Ten (45%) of the discussions led to a referral to Specialist Palliative care.
There were no differences in occurrence versus non-occurrence of an end-of-life discussion based on patients’ gender, age, duration of device implantation, indication for device implantation, type of device, whether pre-implantation information was given, shock experience, aetiology of heart disease, diagnosis of cancer, patients’ NYHA class, sodium level, haemoglobin level or NT-pro b-type natriuretic peptide level.
Aim 3: To determine the frequency of ICD deactivation and ultimate manner of death in end-of-life ICD recipients.
Out of the 44 patients, sixteen (36.4%) had their ICD deactivated following an end-of-life discussion, with the majority carried out by a cardiac physiologist within a hospital environment (29.5%: n=13). Other locations where the ICD was deactivated included home (4.5%: n=2) and a hospice (2.3%: n=1). Refer to Table 3
Information on the cause of death was available in 42 patients, with information on the status of the device prior to death collected on 43 patients, 1 patient had their device removed. Of the 16 patients who had their device deactivated prior to death, seven experienced a non-sudden cardiac death, four a sudden cardiac death, four a non-sudden non-cardiac death and one patient a sudden non-cardiac death. A total of 27 patients had their device active when they died, and in two of these patients the cause of death was unspecified. In the remaining twenty-five patients, nine had a sudden cardiac death, six a sudden non-cardiac death, six a non-sudden non-cardiac death and four a non-sudden cardiac death (Refer Figure 1). There were no differences in cause of death based on whether the ICD was active or had been deactivated.
A total of 22 (50%) patients died in a hospital environment, 41% at home (n=18), 4.5% within a nursing home (n=2) and one patient died in a hospice. Information on one patient’s place of death was unavailable.
Aim 4: To identify the factors associated with the decision to deactivate an ICD at end-of-life.
Ten patients experienced a shock during the life of their ICD, of which five had at least 1, and as many as 5, shocks, while another four patients experienced multiple shocks (> 10). Among those who received multiple shocks, the mean length of time from last documented shock until death was six months compared to 38 months for patients who experienced less than five shocks during the time they had the device implanted.
Of the patients who had experienced a shock, 62.5% had an active ICD at death, while 93.7% of those who had their ICD deactivated had never had a shock (p = 0.003). This implies that patients who experienced a shock were less likely to have their device deactivated at the end-of-life compared to those who had not experienced a shock. Shorter time of ICD implantation (42 ± 40 months versus 87 ± 40 months) was associated with deactivation (p = 0.017), but after adjustment of the p value for multiple comparisons, this difference was no longer significant. There were no differences in the proportion of those who had the ICD deactivated compared to those in whom it remained active based on gender, age, aetiology, presence of cancer, NYHA classification, ICD indication, type of device, whether a pre-implantation discussion had taken place, or sodium, haemoglobin or NT-pro b-type natriuretic peptide level (Refer Table 4).
Discussion:
The key finding from this retrospective case note review was the lack of adherence to international clinical guidelines (6,8) in two main areas. Firstly there was no documented evidence that a discussion concerning ICD deactivation had been initiated or taken place between a professional and any of the 44 patients at the pre-implantation stage. End-of-life discussions were frequently initiated with patients when death was imminent as this review found the median time from discussion to death was only seven days. Secondly and perhaps most surprising, was patients who experienced a shock during device implantation were less likely to have their device deactivated at the end-of-life than those who never had a shock. These gaps in the translation of expert recommendations into clinical practice require further research exploration.
Decision to initiate the discussion:
The Heart Rhythm Society and the American Heart Association termed the discussion concerning ICD deactivation before device implantation or replacement, as the “good dialogue” (8,9). The current study found no documented evidence that such a discussion had occurred or patients were informed of the possibility of deactivation in the future. ICDs are implanted to treat life-threatening arrhythmias and prevent sudden death (1). For many patients, the ICD is portrayed at implantation as solely a “life-saving” device and this perception remains paramount in their minds throughout the remainder of their illness (17, 41). Informing patients of all risks and benefits (41) pre-implantation, including that of deactivation would facilitate a more comprehensive consent process and potentially enhance patients’ contribution to future decision-making.