|Home | About | Journals | Submit | Contact Us | Français|
Advances in heart failure (HF) treatments have prolonged survival but more patients die from HF than from any type of cancer. Little is known about the current practice in end-of-life (EOL) care in HF.
Two EOL cohorts (HF and cancer) were identified using Medicare data linked with pharmacy and cancer registry data. We assessed use of hospice, opiates, and acute care services (hospitalizations, emergency room [ER] visits, intensive care unit [ICU] admissions, and death in acute care). Time trends and predictors of use were assessed using multivariate regression including demographics and cardiovascular and non-cardiovasuclar comorbidities.
Among 5,836 HF patients with median age 85, 77% female and 4% black, 20% were referred to hospice compared to 51% of 7,565 cancer patients. A modest rise in hospice use over time was parallel in the two groups. Twenty two percent of HF patients filled opiate prescriptions during 60 days prior to death compared to 46% of cancer patients. Use of acute care services in the 30 days prior to death was higher for HF (64% vs. 39% for ER visits, 60% vs. 45% for hospitalizations, and 19% vs. 7% for ICU admission). More HF patients died during their hospitalization than cancer patients (39% vs. 21%).
Patients dying of HF were less likely to be supported by hospice and opiates but more likely to die in hospitals than patients with cancer. Our study suggests that opportunities may exist to improve hospice and opiate use in HF.
Advances in understanding the pathophysiology of heart failure (HF) introduced many effective therapies, which have prolonged survival in trial and community populations1. As more patients survive into late-stage HF, they carry a longer burden of dyspnea, pain, and other end-of-life symptoms prior to death. Studies have shown that 20% to 78% of HF patients report pain and/or severe dyspnea2-7 which may be more severe closer to death.2. As for pain and dyspnea in late-stage diseases, opiates can be effective for HF8 as well as for cancer. Hospice has been associated with better patient satisfaction9, better quality of death 10, and less emotional stress in patients and families11. While more patients die from HF than from any type of cancer12, quality of care improvement for HF has focused largely on education and initiation of life-saving treatments.
We previously reported that the number of HF hospitalization is an independent predictor of prognosis in HF patients13 and the median survival decreased progressively with repeated hospitalizations (median survival <1.5 years after two hospitalizations and < 1 year after three hospitalizations). These patients may be particularly appropriate for studying the use of hospice and opiates.
The purpose of the current study is to 1) describe the use of hospice, opiates, and acute care services at the end-of-life in HF patients with repeated hospitalizations and 2) compare HF and cancer patients with respect to their use of these medications and services.
We linked Medicare files (Part A, Part B, long-term care, hospice, and enrollment files), the state cancer registry, and pharmaceutical dispensing data from the state-sponsored pharmacy assistance program in Pennsylvania (PA). Pennsylvania Pharmaceutical Assistance Contract for the Elderly (PACE) is the largest state prescription benefits program for the low to middle income elderly. PACE has no deductibles or maximum annual benefit and charges a modest co-payment ($6 generic, $9 brand-name). The 2007 income ceiling for eligibility was $27,676 for a married couple. These data sources provided basic demographic information, death date and causes of death from death certificate as well as coded diagnostic, procedural, and pharmacy dispensing information with high accuracy14. The Institutional Review Boards of the Brigham and Women’s Hospital approved this study, and data-use agreements were established. All potentially traceable personal identifiers were removed from the data prior to analyses to protect patients’ privacy.
To define a cohort of end-of-life HF patients, we first required patients to have two or more HF hospitalizations during 1997-2004 in PA. The HF hospitalization was defined as an admission with HF as the primary discharge diagnosis (ICD9-CM of 428.xx), which has previously been shown to have a positive predictive value of 94% for HF using the Framingham criteria.15 We further required that 1) patients died from cardiac diseases16 using information in the death certificate and filled 1 ≥ prescriptions for loop diuretics during 365 days prior to death. All patients were active participants in their insurance programs. We also identified a subset of the HF cohort whose cause of death was specifically HF to conduct the sensitivity analyses.
The end-of-life cancer cohort consisted of patients diagnosed with one of the 4 most frequent solid malignant tumors (colorectal, lung, breast, and prostate). Patients were identified from the state cancer registry as having a diagnosis between 1/1/1997 and 12/31/2004. These patients also had been active participants in their insurance programs. We further required that the patients died during the study period with cancer as the primary cause of death.
From the enrollment files for Medicare and PACE, we obtained patient demographic information such as date of birth, gender, race (Black vs. White and other), and income. Patients’ comorbid conditions were assessed during the 1 year prior to death using diagnosis codes from in-patient and out-patient files. The measured comorbidities included the patient’s history of myocardial infarction (MI) 17, prior HF 15, other ischemic heart disease, cerebrovascular disease17, peripheral artery disease17, atrial fibrillation, hypertension, diabetes, chronic kidney diseases (CKD) 18, chronic pulmonary diseases, gastrointestinal bleeding19, cancer, arthritis, depression and dementia.. We also assessed the cardiovascular prescription drug use (See Table 1)
We assessed the date of the first hospice enrollment after the date of the index HF hospitalization or cancer diagnosis. As both HF and cancer patients at the end-of-life may suffer from considerable pain and dyspnea 4-5, 20-21 that may require use of opiates , we also assessed whether these patients received any short- or long-acting opiate within 60 days prior to death. To ensure the complete capture of pharmacy claims, opiate use was only evaluated only in a subset of patients who reside outside the nursing home or hospice in the last 60 days prior to death.
To describe the intensity of acute medical management at the end-of-life, we assessed whether patients had 1) any emergency room (ER) visit within 30 days prior to death, 2) any acute care hospitalization within 30 days prior to death, 3) any admission to intensive care unit (ICU) within the 30 days prior to death, or 4) died in an acute care setting.
Trends in the use of hospice, opiates, ER visits, ICU admissions, acute care hospitalizations, and deaths in an acute care setting were plotted for yearly intervals from 1997-2004 for each cohort. To assess major predictors of comfort care, we constructed multivariate modified Poisson regression models 25, which allowed direct estimation of risk ratios (RR) for binary outcomes with the corresponding 95% confidence intervals (95% confidence interval [CI]) even with frequent outcomes. SAS for Windows software (release 9.2) was used for all statistical analyses (The SAS Institute, Cary, NC)
No extramural funding was used to support this work.
We identified a total of 45,559 patients who had at least one HF admission between 1/1/1997 and 12/31/2004 and had been active participants in their insurance programs for >1 year prior to the index HF admission in our base population. Of those, 5,836 had >2 HF hospitalizations, died from cardiac disease during the study period, were on loop diuretics during 365 days prior to death, and thereby entered the EOL HF cohort. Among the 15,231 patients linked to the state cancer registry with active participation in their insurance programs for at least 1 year prior to diagnosis, 7,565 died of cancer, and thereby entered the EOL cancer cohort. Table 1 compares the characteristics of the study patients. The mean age of HF patients who died of cardiac causes was 85 years at the time of death and 77% were female. Several comorbid conditions were quite common: 22% had a previous MI, 91% had coronary artery disease, 44% had cerebrovascular disease, 53% had diabetes, 61% had chronic pulmonary disease, 55% had chronic kidney disease, 29% had depression, and 33% had dementia. Similar characteristics were observed in the subgroup of these patients who died of HF (N=1,350) were similar. The cancer patients were relatively younger with fewer cardiovascular comorbidities, CKD, depression, or dementia but with similar prevalence of chronic pulmonary disease.
Hospice enrollment was observed in 51% of cancer patients compared to 20% of HF patients (Table 2). Figure 1 shows that the modest rise in hospice use over time is parallel across the two groups. In the multivariate analysis of HF patients, younger age, male gender, income, and dialysis treatment were associated with 2% per year (95% CI: 1- 3%), 25% (95% CI: 14- 35%), 2% per $1,000 increase (95%CI: 0.2 - 3%) and 65% (95% CI: 44- 78%) reduction in the enrollment in hospice, respectively.
Figure 2 shows the use of opiates over time in the subset of patients not admitted to a hospice or nursing home prior to death (N= 1,795 for HF cohort and 1,888 for cancer cohort). In HF, 22% of patients filled opiate prescriptions during the 60 days prior to death. The fraction was lower for those who eventually died in an acute-care-hospital (19%) compared to those who died outside of non-acute care settings (27%) (Table 2). Opiate use was 23% higher in patients with cancer prior to death (46%). For the multivariate analysis in HF patients, opiate dispensing was less likely to occur in male patients (risk ratio [RR] =0.70; 95% CI: 0.56-0.87), Black patients (RR=0.56; 95% CI: 0.33-0.97), patients with dementia (RR=0.72; 95% CI: 0.55-0.94), or patients with cerebrovascular disease (RR=0.82; 95% CI: 0.68-0.99). Those with chronic arthritis (RR=1.43; 95% CI: 1.19-1.71) were more likely to dispense opiates. Even after adjusting for these factors, there was no trend for increased opiate use over time (p=0.13). Finally, we conducted sensitivity analyses by changing the length of the assessment window for opiate use to 30 days or 365 days. The frequency of opiate dispensing decreased to15% for HF patients and 33% for cancer patients using the 30-day window and increased to 40% for HF patients and 63% in cancer patients using the 365-day window). However, the difference between HF and cancer patients and the time trend for opiate dispensing remained similar to those in the primary analysis using 60-day window.
Use of acute care services within the 30 days prior to death was higher in HF patients than that in cancer patients. (60% vs. 39% for ER visits, 64% vs. 45% for hospitalizations, and 19% vs. 7% for ICU admissions). There was no trend over time in the frequency of ER visits, acute care hospitalizations, or ICU admissions during the 30 days prior to death in either cohort. More HF patients died in acute care hospitals than cancer patients (39% vs. 21%). However, we observed a modest decrease over time in the frequencies of deaths in acute care hospitals, which appeared parallel in HF and cancer patients (Figure 3).
We repeated all analyses in the subset of HF patients whose death was specifically attributed to HF (N=1,350) which yielded no substantial changes from the main analyses.
Using population-based health care utilization data for the elderly, we characterized use of hospice, opiates, and acute care services in HF patients at end-of-life and compared it to that in cancer patients. Overall, use of opiates and hospice was not high and use of acute care services was high in both HF and cancer patients. Over time, hospice use has increased slowly in both patient groups, but the contribution of hospice remained lower in HF. Also, more HF patients died in acute care settings. Various components of effective end-of-life care may be provided in non-hospice settings that are not easily captured. However, we were able to assess opiate use in HF patients in non-hospice settings, which was low compared to that in cancer patients and remained low over the duration of study period. To our knowledge, this is the first population-based study described trends in the use of hospice, opiates, and acute care services in elderly HF patients near death and compared them to those in cancer patients.
Prior studies found that hospice utilization was lower in patients who died of heart disease compared to that in patients who died of cancer26-27. Connor et al. found that 12% of patients died of heart disease used hospice compared to 65% in patients died of cancer26 using 100% Medicare hospice file in 2002. White race, female gender, and advanced age were shown to be associated with higher utilization of hospice 26, 28. In our study, the enrollment to hospice in 2002 was 26% in HF patients, much higher than the national figure. This higher rate can be explained by the demographic characteristics of our study population. Our study population is dominated by older, White, and/or female patients Geographic variation in the use of hospice has been reported26-27 and PA is one of the states with relatively lower use of hospice compared to other states 26 However, it is possible that there is a within-state variation contributing to the higher use in our patients since our study patients are a subset of Medicare beneficiaries in PA.
Patients with advanced HF experience significant pain 2-7, dyspnea2, 5-7, and diminished quality of life 29. While the specific causes of pain are unclear, 20 - 70% of patients with advanced HF report pain in a population based study 4 or in community dwelling representative samples 4-5. No evidence exists on how to manage pain and symptomatic dispend at the end-of-life in HF patients 30. Opiates are considered “essential” medications for palliative care by the World Health Organization31. Opiates have been used to alleviate pain in cancer patients and can be effective in managing pain and dispend in patients with end stage HF8. Hospice and other palliative care services can address various aspects of suffering at the end of life including physical, emotional, psychological and spiritual issues and provides caregiver support32. Although patients with HF and cancer have similar needs for palliative care as assessed by symptom burden, depression, and spiritual well-being21, we found that approximately 20% of HF patients were enrolled in hospice or dispensed opiates compared to approximately 50% of cancer patients.
Although our data do not provide insights on reasons and appropriateness of hospice, opiates, or acute care service use in end-of-life HF patients, they shed light on the current status of end-of-life care for HF patients. Several factors could have potentially contributed to the low use of hospice and opiates and high use of acute care services prior to death in our HF patients. First, physicians may be unsure about or underestimate the high mortality after multiple HF hospitalizations and feel reluctant to discuss limited prognosis with HF patients. According to a recent survey for physicians taking care of HF patients, only 16% of the physicians stated that they could predict death within 6 months well 33. Thomas et al. showed that physicians overestimated the survival in 68% of HF patients vs. 11% of cancer patients34. Secondly, physicians may not discuss with patients about their end-of-life preferences. While physicians’ discussion of hospice was independently associated with a large increase in the likelihood of hospice use (odds ratio = 5.3; 95% CI: 2.3-13) 34, only 2.5%35 or 7%34 of HF patients had a discussion about resuscitation preferences or hospice. A recent study showed that patients often have a change in their preferences in survival during or following a recent hospitalization. However, their preferences tend to remain stable during the next 6 months36 where many HF patients with advanced symptoms are more interested in feeling well than in living longer 36.
In addition, patients and families may be unwilling to hear about their limited life expectancy. However, in a survey conducted in elderly patients with cancer, HF, or chronic obstructive pulmonary disease and their caregivers37, 55% of HF patients who had no discussion of their life expectancy with physicians reported that they wanted one. Also, 75% of the caregivers reported that they preferred to have a discussion on life expectancy37. Finally, unwelcome connotations of “hospice” and “narcotics’ in non-cancer settings and limited availability of palliative care services for HF38 may have contributed to the lower utilization in HF patients.
Several limitations should be noted in interpreting our results. First, our data are limited to low to middle income Medicare beneficiaries who participated in the pharmacy benefit program. The findings in our study may not be generalizable to younger or wealthier patients. However, approximately a half of the states have similar pharmacy assistance programs providing services to patients similar to ours in socioeconomic status. Second, our data lack detailed clinical information and the study population may have included those who were hospitalized for HF but did not die of HF. However, sensitivity analyses using more stringent definition of end-of-life HF cohort requiring 2 HF hospitalizations +loop diuretic use + death due to HF) or less stringent definition requiring (2 HF hospitalizations + cardiac death) yielded very similar results. Also, this potential misclassification of including patients who might have not died of the disease applies to the cancer cohort and therefore is likely to be non-differential. Third, we could not assess whether patients were consulted to palliative care service as no specific procedure code was available to identify palliative care referrals. Finally, we could not assess opiate use after our patients entered a nursing home or enrolled in hospice because the dispensing information is incompletely recorded during their stay in a nursing home or hospice . Opiate use can be much higher in HF patients enrolled in hospice as it has been reported that 74% of HF patients in hospice were prescribed the medication6.
We found that HF patients were less likely to be supported by a hospice or opiates prior to death and more likely to die in acute care hospitals than patients dying of cancer. More detailed studies are needed to understand the appropriateness and reasons of the low hospice and opiate use. Opportunities may exist to improve the end-of-life care in HF patients by recognizing limited prognosis in patients with repeated HF hospitalizations, communicating about the prognosis and the role of hospice and opiates, and improving availability of palliative care services in HF patients.
Financial support: Supported by grant AG18833
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of interest: None
The abstract of the study was presented at American Heart Association Annual Scientific Sessions, New Orleans, LA in November 2008.