Physician Allocation of Medicare Resources for Patients with Advanced Cancer

Introduction

In 2011, approximately 15% of the federal budget was spent on Medicare, with a significant portion of this going toward end-of-life care. ¹ More than a quarter of the Medicare budget is spent on beneficiaries in their last year of life, and 40% of that is spent in their last 30 days. ² Current projections suggest that Medicare and end-of-life spending will increase and potentially threaten the viability of this important program.^1,3

Despite considerable interest in end-of-life care, the literature is sparse and divergent on patient preference versus physician preferences regarding what services should be covered by a health insurance plan for persons facing the end of life. One study looked at patient and physician preferences for different aspects of care such as resuscitation or ventilation within the context of a terminal illness. It found that patient and provider preferences differ when viewed as a group, but found a correlation between individual providers and their patients. ⁴ Steinhauser and colleagues demonstrated that patients and physicians have similar views on what is important at the end of life. ⁵ Several other studies have shown that factors such as race, religiosity, and current health status affect end-of-life treatment preferences, but these studies looked at specific end-of-life therapies and not at comprehensive health coverage.^6–11 Earle and colleagues showed that there is a trend toward more aggressive and more expensive cancer treatment at the end of life, but they did not assess whether this approach matches with patient or provider views. ¹²

To our knowledge, no one has studied how patients and physicians would craft a Medicare coverage plan for patients with advanced cancer. This information is important for several reasons. First, it is vital that limited Medicare resources are employed in an efficient manner—both for patients and physicians. Second, when physicians have end-of-life discussions with patients, it is important for physicians to understand both their own biases and those of their patients. Better understanding of these biases may lead to more effective discussions. Third, physicians are a powerful advocacy force for their patients, and these advocacy efforts could be better informed by data on the values of their patients and how those values may differ from their own.

Methods

“Priorities for Medicare Advanced Cancer Care” is a study that is currently being conducted through the Duke Sanford School of Public Policy, in conjunction with the Duke Cancer Care Research Program and the Duke Cancer Institute. This study is using a modified version of a validated tool called Choosing Healthplans All Together (CHAT) ¹³ to assess how cancer patients and their caregivers would craft a new Medicare plan for patients with advanced cancer. Advanced cancer was defined as a cancer that is not generally considered curable and is expected to shorten a person's life. All patients with a cancer diagnosis are eligible, along with their caregivers.

The CHAT tool is a four-round, individual and group exercise in which participants allocate a limited pool of resource units among well-defined benefit categories (see Table 1 and online Supplementary material at http://www.liebertpub.com) in order to create a new Medicare plan. CHAT has 15 different benefit categories, each with between one and four different levels of coverage available. Detailed descriptions of each benefit category are provided. In order to fully cover each category at the highest level, 88 resource units would be required. Participants are allowed to choose only 50 units, requiring prioritization of coverage allocations. In rounds one and four, the participants allocate their resources individually, and in rounds two and three, group allocations are made.

Given the significant time commitment and group participation needed for the traditional CHAT exercise, a single-round, online version of the modified CHAT tool was developed for use with physicians (see Figure 1). Institutional review board approval was obtained from the Duke University School of Medicine. Three separate emails were sent to all members of the American Academy of Otolaryngology – Head & Neck Surgery requesting participation in this exercise. The goal was for these otolaryngology, head and neck surgery (OHNS) physicians to create a Medicare plan for patients with advanced cancer. A link and password to access the online game were provided in the email.

OPEN IN VIEWER

FIG. 1.

Screenshot from online CHAT tool. Numbers refer to resource units.

Basic demographic information was collected, but no personal identifying information was obtained. Participants then completed the CHAT exercise. Data entered into the online game were captured in a password-protected database. A unique identifier for each invited participant's email address was recorded in a separate password-protected database. This identifier was used to ensure that duplicate email requests were not sent and to ensure that there were no duplicate responses. The unique identifier was not associated with the demographic or game data to preserve anonymity.

OHNS physician data were compared with cancer patient and caregiver data obtained from the ongoing study “Priorities for Medicare Advanced Cancer Care.” Data for patients and caregivers were taken from round four of their exercise. Taking into account the obvious familiarity that physicians would have with the subtleties of coverage in different benefit categories and with medical terminology, we assumed that the allocations of physicians in a single-round exercise would be roughly equivalent to patient and caregiver individual allocations in round four.

Chi-square tests were used to compare the two groups' allocations, and analysis of variance (ANOVA) was used to investigate the association of demographics factors with allocation in each of the benefit categories by physicians and patients/caregivers. P-values for statistical significance were less than 0.05.

Results

In total, 767 OHNS physicians were recruited and completed the game. Responses were obtained from 146 cancer patients and 114 caregivers. Patient and caregiver demographic data are listed alongside physician demographic data in Table 2. Physician, patient, and caregiver allocation data are listed in Table 3.

There were no significant differences between patient and caregiver allocations in any of the benefit categories (see Table 4), so these two groups were combined for purposes of analysis. A comparison of physician and patient/caregiver allocations was performed, and statistically significant differences were shown in resource allocation for each of the 15 categories except Home Care. Physicians elected more coverage in the Advice, Emotional Care, Palliative Care, and Treatment for Cancer benefit categories. Patients and their caregivers elected more coverage in the Cash, Complementary Care, Cosmetic Care, Dental and Vision, Drug Coverage, Home Improvement, House Calls, Nursing Facility, Other Medical Care, and Primary Care benefit categories.

Association of demographic variables with resource allocation was analyzed and summarized in Table 5. When respondents were stratified by age, older physicians were more likely to allocate resources to Home Care (p=0.0026), House Calls (p=0.0007), and Other Medical Care (p=0.0483). Younger physicians were more likely to allocate resources to Primary Care (p=0.0113) and Treatment for Cancer (p=0.0006). Patients and caregivers showed no significant trends when stratified by age.

Frequency of cancer care was significantly associated with different allocation levels in several categories (see Table 5). Otolaryngologists who care for cancer patients more frequently are more likely to allocate resources to Drugs (p=0.0264), House Calls (p=0.0129), and Treatment for Cancer (p=0.0050). When comparing otolaryngologists who frequently care for cancer patients (i.e., those who selected that they care for cancer patients often or that it is the primary thing that they do) to the cancer patients and their caregivers, significant differences in preferences remained in every category but Home Care (see Table 6).

Discussion

This is the first study to look at how physicians would craft a Medicare plan for patients with advanced cancer, and to compare these allocations to those made by cancer patients and their caregivers. Our data indicate that otolaryngology physicians value Advice, Emotional Care, Palliative Care, and Treatment for Cancer more than cancer patients and their caregivers. Cancer patients and their caregivers, on the other hand, value Cash, Complementary Care, Cosmetic Care, Dental and Vision, Drug Coverage, Home Improvement, House Calls, Nursing Facility, Other Medical Care, and Primary Care more than their physicians.

Our objective was not to emerge from this study with a suggested distribution of coverage in a Medicare end-of-life health plan. Rather, our study highlights considerations crucial to future conversations on a resource distribution model based on value-driven prioritizations from patient, caregiver, and provider perspectives. We discuss generalizations and observations based on the prioritized allocations from each group in our study to serve as a conversation starter.

Otolaryngologists value Treatment for Cancer more than patients and caregivers (see Figure 2). We were not surprised by this result, as we expected physicians to highly value the services they provide. The fact that physicians also value Palliative Care more highly than patients and caregivers, however, was initially surprising. We expected that both groups would choose either an aggressive course of cancer treatment with little palliative care, or the opposite. Thus, we surmised that if patients and caregivers chose less Treatment for Cancer than the physicians, they would choose more Palliative Care. Instead, the patients valued both categories less than the physicians.

OPEN IN VIEWER

FIG. 2.

Comparison of resource allocations.

It is possible that there is an incomplete understanding of palliative care by patients contributing to this result, or that palliative care has a stigma that cancer patients in the midst of treatment are uncomfortable embracing. Alternatively, patients and caregivers may value other categories much more. Thus, patients and caregivers are more likely to shunt resources away from high-cost categories like Treatment for Cancer and Palliative Care in favor of categories like Cash and Home Improvements that may have a greater impact on their quality of life (see Figure 2).

Our data suggest that physicians value advice and emotional care more than the cancer patients and caregivers (see Figure 2). This finding may suggest that physicians underestimate their patients' ability to cope with and handle the difficult choices associated with advanced cancer. Alternatively, it could suggest that patients underestimate the difficulty of the treatment course for their cancer. Certainly the fact that many of these patients did not have advanced cancer limits our ability to answer this question, but it does suggest again that there is a divergence in viewpoint between physicians and patients.

The data also showed older physicians are more likely to elect higher levels of coverage in Home Care, House Calls, and Other Medical Care; while younger physicians elected higher levels of coverage in Treatment for Cancer and Primary Care. One would expect younger participants to more aggressively pursue life-prolonging care. Older participants may have more experience with seriously ill family members or serious illness themselves, and so it is not unexpected that they would value categories aimed at easing the day-to-day difficulties of disease or, with Other Medical Care, treat comorbid medical conditions. Otolaryngologists who treat cancer patients more frequently are more likely to elect for more Treatment for Cancer, again not surprising, as one would expect them to highly value their own services. This data is summarized in Table 5.

This study is both significant and timely. Given the enormity of this country's investment in Medicare and end-of-life care,^1–3 it is essential to understand physician and patient preferences if scarce medical resources are to be efficiently allocated. Current reform proposals have appropriately focused on cost savings,^14–16 but no one has examined how current coverage provisions square with patient and provider preferences. Covering benefit categories that patients and physicians value and not covering those categories that are not valued, even if it does not lead to cost savings, will increase patient and physician satisfaction and decrease waste.

These data are also useful for physicians when having end-of-life discussions with patients. Physicians may have a tendency to focus these discussions on areas that are more directly related to their profession, such as the decision between palliative care and treatment for cancer. Patients and their caregivers, on the other hand, may have unmet concerns regarding day-to-day living with a cancer diagnosis (e.g., cash for covering out-of-pocket expenses).

Research suggests that these end-of-life conversations, if they happen at all, occur late in the course of treatment. ¹⁷ A better understanding of physician and patient biases may lead to physicians initiating these conversations earlier and more often. Recognition of the day-to-day concerns of patients in end-of-life discussions may also lead to a greater level of trust and more effective counseling. Moreover, some have suggested that encouraging these discussions earlier may decrease end-of-life costs through advance directives or by more awareness and utilization of hospice care. ¹⁵

These findings are significant for physicians who engage in advocacy both for themselves and for their patients. Physicians are a powerful and respected lobbying force. Part of this respect is certainly based on the assumption that physicians are attuned to the values and concerns of their patients. This study may call this assumption into question, but, if approached correctly, would provide a springboard for a fuller understanding of these values and concerns. Certainly physician advocacy efforts are more effective with a strong understanding of the values and biases of physicians and patients.

This study has several limitations. First, physician participants were limited to otolaryngologists only, perhaps limiting its generalizability. However, otolaryngology is a broad field, encompassing physicians treating general complaints with minimal contact with cancer patients, and physicians who specialize only in the treatment of head and neck cancer. It also includes both medical and surgical management of conditions and diseases. Certainly a broader assessment of physician preferences for resource allocation would be useful. Second, the patient data were not limited to those with a diagnosis of advanced cancer, but instead included patients with any cancer stage and prognosis. This group may not adequately represent what patients with an advanced cancer diagnosis would want. Third, there may be a selection bias. The physicians who respond, for example, may be those who are more interested or involved in cancer treatment, making them more likely to elect for more cancer treatment coverage. Lastly, the benefit category definitions used in the CHAT model may not be adequate in explaining or describing to participants what services are truly rendered. This may result in some misunderstanding among participants, especially the patients and caregivers, regarding what each category includes.

Conclusion

This study is the first to examine how physicians would allocate a limited pool of resources to create a Medicare plan for patients with advanced cancer. We demonstrated that there are significant differences in how physicians and cancer patients and their caregivers allocate resources. This suggests that even among physicians in a specialty that frequently cares for cancer patients, there is a significant divide between their values and those of their patients. These findings are not only important for understanding how different groups would allocate limited medical resources for end-of-life care, but also significant for physicians having end-of-life care discussions with their patients and for those physicians who engage in advocacy efforts.

Author Disclosure Statement

No competing financial interests exist.