Palliative Care Referrals in Cardiac Disease

We conducted a retrospective single-institution cohort study in a quaternary pediatric institution as a contemporary follow-up and expansion to an earlier study.⁸  The institutional review board approved the protocol (additional details are in the Supplemental Information). Data from 2 separate, 3-year epochs were collected: January 1, 2007, to December 31, 2009 (early epoch), and July 1, 2015, to June 30, 2018 (late epoch). AHD was defined as a diagnosis of major cardiac disease (cardiomyopathy or congenital heart disease) necessitating hospitalization.^8,15  Administrative databases were interrogated for admissions, subspecialty PPC referrals, and all inpatient AHD deaths, including on floors or in ICUs. Out-of-institution death notifications were reviewed for available data but excluded from patient-level statistical analyses, in which inpatient deaths at our center were reported. Additional exclusion criteria included age >21 years and deaths unrelated to cardiac disease. Electronic medical records were reviewed to expand the existing database.⁸  A standardized data extraction process was employed to obtain demographic, clinical, and PPC consultation domains from discharge summaries, clinical and death notes, family meeting records, operation reports, medication charts, and flow sheets electronically and by manual review (Supplemental Information).

Measures of lifetime disease burden included CICU and hospital admissions, cardiac-surgical procedures, and technological dependence (tracheostomy and/or gastrostomy). Intensity of medical interventions and clinical status were characterized in the 24 hours before death, capturing mechanical ventilation; inotropic, sedative, analgesic, and paralytic infusions; level of consciousness; and enteral feeding. Need for mechanical circulatory support (extracorporeal membrane oxygenation [ECMO] or ventricular assist devices [VADs]) and cardiopulmonary resuscitation (CPR) events were reported in the 24 hours and 7 days before death. Organ failure was categorized on the basis of the Pediatric Logistic Organ Dysfunction-2 score.³¹  Cause of death was based on main organ system involvement. Modes of death were defined as follows: (1) discontinuation of life-sustaining therapy involving active cessation of disease-directed care; (2) nonescalation, or limitations to therapy in patients dying while receiving ongoing cardiorespiratory support; (3) maximal active resuscitation; or (4) brain death meeting established criteria.^21,32  Subspecialty PPC involvement; timing of PPC consultation; psychiatry, child life, ethics, pastoral care, and social work service input; parent and/or relative presence at EOL; ACP; and resuscitation status documentation were identified. Hospital charges for technical services (excluding professional bills from providers) for the day of death, the last 7 days of admission, and the total hospital admission were obtained from the Epic billing system.

Categorical variables are summarized by using numbers and percentages, and continuous variables are summarized by using medians and interquartile ranges (IQRs). Two epochs (2007–2009 and 2015–2018) were compared to evaluate temporal changes in patterns of EOL practices. In addition, we compared patients who died in the CICU with and without subspecialty PPC consultation. Comparisons were performed by using Fisher’s exact test for categorical variables, and the Wilcoxon rank test for continuous variables. Associations are reported as risk ratios (RRs) with 95% confidence intervals (CIs). A stratified analysis was performed to evaluate relationships separately for patients with high-risk diagnoses for PPC referral (pulmonary vein stenosis [PVS] and pulmonary hypertension). Hospital charges for each financial year were in US dollars and were adjusted for inflation to the 2018 fiscal year by using the Centers for Medicare and Medicaid Services National Health Expenditure Accounts data set.³³  All analyses were performed in Stata version 15 (Stata Corp, College Station, TX) and GraphPad Prism (version 8; GraphPad Software, San Diego, CA). P values <.05 were considered significant.

There were 110 inpatient deaths in the early epoch and 99 in the late epoch (unadjusted hospital mortality rates: 3.2% and 2.5%, respectively) (Fig 1). PPC referrals were more likely in the late epoch (45 [1.3%] early-epoch admissions compared with 146 (3.6%) admissions in the late epoch; RR 2.7 [95% CI 2.0–3.8]; P < .001). Overall, 76 (36%) inpatients who died had PPC consultations, with a greater proportion referred in the late epoch (19 [17%] vs 57 [58%]; RR 3.3 [95% CI 2.1–5.2]; P < .001). The 54 out-of-institution deaths included 18 (33.3%) children referred to PPC before institution discharge (5 [20%] early-epoch and 13 [45%] late-epoch out-of-institution deaths), and 9 (17%) children had ACP documentation (Supplemental Information).

In Table 1, we describe the study population overall and by epoch (also see Supplemental Information). Single ventricle circulation was the most common pathophysiology, with pulmonary hypertension and PVS diagnoses more frequent in the late epoch. No difference was observed in cardiac-surgical procedures, with 22% of patients having no surgical intervention before death (P = .31). The late epoch had a greater burden of disease and technological dependence. Three-quarters of the entire cohort had organ failure, with similar rates between epochs. The majority of deaths occurred in the CICU and followed discontinuation of life-sustaining therapy across both epochs, with parents more than twice as likely to be present at death in the late epoch (RR 2.3 [95% CI 1.8–2.9]).

Similar rates of medical interventions were observed in the 24 hours before death between epochs, with most patients receiving mechanical ventilation or inotropic and sedative infusions (Table 2). Fewer CPR events were observed in the late epoch. Frequency of referral to interdisciplinary services and resources differed, with greater involvement of subspecialty PPC, pastoral care, social services, and child life services in the late epoch. Indications for PPC referral included goals of care and ACP (58 [76%]), longitudinal support (53 [70%]), symptom management (28 [37%]), complex decision-making (26 [34%]), and hospice referral or care coordination (2 [3%]). In the late epoch, there were higher rates of ACP (RR 2.5 [95% CI 1.7–3.5]) and resuscitation status documentation (RR 2.9 [95% CI 1.8–4.1]) as well as earlier PPC referral (median time from initial consultation to death: 69 vs 21 days; Table 2). ACP meetings occurred a median 1 day before death (IQR 0–7.5). Adjusted for inflation, total charges for the hospital admission and day of death were similar, whereas 7-day charges were lower in the late epoch (Supplemental Information).

There were 97 and 89 CICU deaths in the early and late epochs, respectively. PPC consultations from the CICU were more likely in the late epoch (RR 4.5 [95% CI 2.5–7.8]; P < .001) (Table 3). Patient characteristics compared by PPC involvement are presented in Table 3, with PPC referral more likely in children who died of PVS or pulmonary hypertension or while technologically dependent. Before death, patients with PPC were more likely to be awake and be receiving enteral feeds and less likely to be supported on mechanical ventilation, inotropes, ECMO, or VADs. In the week before death, children without PPC were almost 3 times more likely to receive CPR or defibrillation (RR 2.8 [95% CI 1.5–5.3]) and had greater rates of ECMO or VAD use (RR 1.6 [95% CI 1.0–2.7]). A similar CICU length of stay was observed between cohorts (PPC, median 27 days [IQR 8–50 days], versus no PPC, median 17 days [IQR 4–47 days]; P = .12). PPC involvement was associated with twice the ACP meeting frequency (RR 2.2 [95% CI 1.6–3.0]) and higher rates of resuscitation status documentation (RR 3.6 [95% CI 2.3–5.7]). In patients without PPC involvement, median hospital charges were 61% higher on the day of death (median $13 301 [IQR $7342–$23 488] versus median $8243 [IQR $2632–$19 261]; P = .02) and 20% higher in the 7 days before (median $151 357 [IQR $105 073–$211 905] versus median $125 723 [IQR $73 737–$176 703]; P = .02), whereas total admission costs were comparable (Fig 2, Supplemental Information). Results were similar in an analysis stratified by high-risk diagnostic category in which patients with PVS and pulmonary hypertension were excluded (n = 56; Supplemental Information).

Most inpatient pediatric AHD deaths in our quaternary pediatric cardiac program occur in the CICU, two-thirds of which follow discontinuation of life-sustaining therapy. Over the past 10 years, involvement of subspecialty PPC and interdisciplinary psychosocial services has dramatically increased, and PPC referrals occur earlier in the disease course. The burden of disease and intensity of EOL care remains high in children with AHD. Importantly, patients who received PPC consultation experienced substantially less invasive medical interventions at EOL, with more frequent ACP meetings and documentation of resuscitation status as well as lower hospital charges at EOL.

Specific details on medical course and interventions at EOL in patients with AHD have been examined in few previous studies.⁸  Our findings of less intensive invasive interventions in patients receiving PPC builds on findings of a contemporary cohort in which children with AHD receiving concurrent PPC were more likely to die at home with care focused on comfort.³⁴  Similar benefits from subspecialty PPC have been shown in pediatric cancer populations, with less intensive treatment and fewer ICU deaths.^35–38  However, a paucity of high-quality evidence precludes specific referral recommendations.¹³  Parents of children who died of AHD reported late recognition that their child would not survive, and poor correlation between parental and clinician perspectives has been observed.^25,39  We found PPC involvement to be associated with less invasive measures that limit parent-child interactions. Three-quarters of consultations were related to goals of care discussions and ACP, which may imply that goal-concordant EOL care is more comfort focused for many families.¹⁰  These findings may reflect enhanced communication skills of interdisciplinary PPC staff in cultivating prognostic awareness and engaging in goals of care discussions, or alternatively, they may reflect provider tendency to engage the PPC team, with parent acceptance of this support over time.⁴⁰  Although hospital charge findings may suggest that overall admission characteristics were similar and that PPC influence specifically relates to EOL care, substantial variability exists in hospital charges, and because of the nature of continuous variables, modest absolute median cost differences are statistically significant with our sample size. However, when multiplied by the number of patients at each large institution and across the country, this difference may become more clinically relevant, with potential implications for families, society, and intensive care resources.

Although in absolute terms, rates of PPC and psychosocial support referrals in pediatric AHD are low, they have increased substantially at this institution over the last 10 years and occur earlier in the illness course. Our 4% PPC referral rate for all cardiac admissions, representing approximately half of late-epoch CICU deaths, compares to contemporary data.³⁴  Although this may represent progress and aligns with international recommendations for concurrent care, there remains immense room for improvement to incorporate PPC for patients with AHD.^5,28,41  Referrals for PPC in children with AHD are dramatically below proportions reported in other diseases, such as complex chronic conditions and malignancy.⁷  By comparison, three-quarters of oncology patients had PPC involvement >30 days before death, with a median time from consultation to death of 107 days.^7,36,42  We confirm that children with AHD represent a particularly challenging cohort with emerging high-risk diagnoses and significant disease burden and suffering. Patient factors differed according to PPC involvement, and goals of care delineation was the most common documented referral reason. These referral patterns require further exploration because factors influencing PPC consultation are likely multifactorial, including patient, parent and/or family, provider, institutional, and regional influences. The primary therapeutic goal for most patients with AHD is surgical palliation, correction, or transplant; however, this is evolving. For some conditions, such as PVS, symptom control may be achieved with repeated catheterizations and tracheostomy. This underscores the importance of high-quality communication skills to support decision-making that is based on patient and family preferences.¹⁴ 

Deaths in children with AHD remain highly medicalized. ICU deaths, CPR rates, and invasive support far exceed other high–mortality risk populations. Only 20% of pediatric AHD deaths occurred outside the hospital, equivalent to findings in other studies.^{34–36,42–44}  This contrasts with findings in pediatric oncology or complex chronic conditions, in which less than half of inpatient deaths occur in the ICU and fewer patients receive intensive therapies at EOL.^35,36,42,44  There are several possible explanations for this observed disparity. First, advances in perioperative care have resulted in extremely low mortality rates, making it difficult to predict which patients may do well with surgical or intensive interventions.^15,21,45,46  Cardiology providers report low confidence in their ability to prognosticate life expectancy.¹¹  This prognostic uncertainty contributing to late recognition of EOL may be even more pertinent in centers with a strong bias toward research and innovation. Second, the clinical condition of patients with AHD is unpredictable. Early data from our institution revealed that 68% of PPC consultations occurred during ICU admission, while patients were receiving highly medicalized care.¹⁰  De-escalation and/or discontinuation of interventions is more challenging compared with noninitiation, despite ethical equivalence.⁴⁷  Third, the mean age at death in our cohort was 7 months, significantly younger than that in other diseases. Moreover, for certain pathologies, invasive procedures can enhance quality of life. High rates of inpatient death may reflect difficulties in transition of care to home, especially for young infants assisted with technologies, due to lack of home hospice services.^21,43,48  Alternatively, for some families, identified goals of care are consistent with invasive surgical interventions, even with a low probability of survival. Despite these differences, the mode of death mirrors that of other critical care populations, with most deaths occurring after limitations or discontinuation of life-sustaining therapy and 10% to 20% of deaths occurring despite active resuscitation.^21,32,34,46 

This study has limitations relating to the single-center retrospective design and the nature of the surgical program that influence generalizability. Shifts in hospital culture regarding integration of PPC into the care of patients with AHD may explain differences between epochs, especially ongoing educational efforts to improve awareness and concurrent growth in other supportive services, such as social work, child life, and chaplaincy. Differences in documentation may be explained by electronic medical record changes between epochs or improved documentation by subspecialty PPC. Furthermore, our measures of medical intensity failed to capture patient, staff, and family experiences. Although PPC offers psychosocial support that may improve the patient and family illness experience, their perceptions of quality of life and EOL were not measured relative to intensive therapies used.¹²  Additionally, subspecialty PPC visit frequency was not captured, precluding evaluation of a dose response. Recorded out-of-institution deaths are likely an underestimate because many patients were national and international referrals, and limited follow-up was documented in the electronic medical record. Moreover, among out-of-institution deaths, we were unable to differentiate patients who died at home with hospice support from those who died of unexpected or unrelated causes. Finally, although we attempted to adjust for patient factors influencing PPC referral patterns in the sensitivity analysis and captured documented reasons for PPC involvement, we were unable to explore patient, family, and team triggers or barriers to PPC consultation.

Our data support value in integration of subspecialty PPC into the lives of hospitalized patients with AHD and their families, with benefits relating to greater ACP and less intensive interventions at EOL. Increasing referrals over time suggests that this is translating to clinical practice, with a shift in prioritization at departmental and institutional levels. Further exploration of what ways and of how subspecialty PPC augments the care of children with AHD is needed to determine if the mechanism relates to increasing prognostic awareness and goal-concordant care, noting challenges associated with the absence of outcome measures to assess this and inherent limitations and biases with follow-up of bereaved families.⁴⁹  There is potential utility in benchmarking outcomes beyond mortality with metrics to measure quality of EOL care needed. Our findings emphasize that research priorities for children with AHD should include determining if certain PPC components can be extracted to better integrate specific palliative care principles thereby reducing EOL suffering; and if improved primary palliative care, or novel integration models, can similarly lead to less intensive EOL interventions. Although subspecialty PPC consultation teams are increasingly available, they reach only a small proportion of patients and families, and demand will likely exceed supply.⁵⁰  Our findings justify the development of innovative solutions to more widely incorporate palliative care principles into the lives of patients with AHD. There are several models of care delivery that can achieve this outcome.^3,4,14,51  Primary palliative care remains a core competency of ICU clinicians; however, this requires dedicated clinical time and training for skill acquisition, optimization, and maintenance. Mixed or novel models of integration are emerging to overcome these barriers and challenges to PPC integration.^4,14,51,52  Identifying relevant components of PPC could guide focused teaching for interdisciplinary cardiac teams.

The intersection of PPC and pediatric cardiology is an evolving field, representing an opportunity to improve quality of care provided and EOL experiences of children with AHD and their families.¹⁵  Our study demonstrates that children with AHD may benefit from greater integration of palliative care into their medical care. Intensity of medical care at EOL remains high in this population; however, increasing involvement of PPC and greater ACP were associated with decreased invasive therapies at EOL. Further studies are needed to explore mechanisms of these findings and to explore what subspecialty PPC adds for children with AHD. Our data inform current practice, with implications to guide training and resource allocation and facilitate future interventions to improve quality of EOL care for pediatric AHD.

We thank Emma Jones, Derek Mathieu, and Jessica Serino-Cipoletta.

ACP

advanced care planning

AHD

advanced heart disease

CI

confidence interval

CICU

cardiac ICU

CPR

cardiopulmonary resuscitation

ECMO

extracorporeal membrane oxygenation

EOL

end of life

IQR

interquartile range

PPC

pediatric palliative care

PVS

pulmonary vein stenosis

RR

risk ratio

VAD

ventricular assist device