Which Patients Need Palliative Care Most? Challenges of Rationing in Medicine's Newest Specialty

Rationing Scarce Palliative Care Consultations

First, we considered prioritizing consults according to their potential clinical impact. If consultations for some indications (e.g., discussions with LK's family regarding extubation) are more likely to have clinical impact than others (e.g., symptom management for MG), it would be reasonable to allocate scarce palliative care resources to those problems for which impact is likely to be greatest. Unfortunately, for the patients above and many others, it is not clear which patients are most likely to benefit. Although the clinical impact of palliative care has been defined for certain very specific populations, ⁴ this evidence base is not well developed enough to allow informed decisions about resource allocation across heterogeneous populations. In addition, even if it were known that palliative care consults are most effective in clarifying goals in one population, evidence might also support symptom management for a second population.

Second, we considered selecting those consults most likely to have favorable financial impact for the hospital. Here, too, there is limited evidence to guide decisions, but at least teams could make educated guesses about which consults are most likely to reduce costs. For instance, it is reasonable to assume that the consult regarding LK's ventilator withdrawal is more likely to result in cost reduction than the consult for MG's symptom management and support. In addition, measurement of financial impact is arguably easier than measurement of clinical outcomes, because it relies on a limited set of universally defined and collected variables (e.g., length of stay, readmissions, direct costs). Our team could monitor impact and readjust consult criteria to optimize impact on these financial metrics.

Despite the inherent appeal of using financial impact as a strategy to prioritize consults, there are, of course, ethical problems with this approach. It would be highly unusual for any consultative service to select consults based on measures that are relevant to the hospital but not necessarily to patients or health care providers. Such an approach might be appropriate for basic infrastructure services that are paid for by a hospital such as utilization review. But a consultation service that can be requested by health care providers is a medical service, and physicians who request a consult expect that clinical criteria will be used to prioritize consults. Moreover, financial explanations are likely to offer little comfort to those patients and families who cannot access palliative care services.

Either of these first two approaches might be more palatable to referring providers if instituted for a well-defined population (e.g., ICU patients) with transparent reasoning (e.g., reduction in length of stay). However, population-based limits would constrain the informal education that accompanies consultation to the nurses and physicians on a few units.

Third, we considered a purely random strategy, in which consults would be declined using a predetermined formula. This might include a fixed refusal rate (e.g., every fifth consult) that could be adjusted based on the team's workload. Such a strategy would ensure equal access for all patients and providers. However, we rejected this approach because it is insensitive to patient needs and the suspected impact of a consultation.

In the end, we chose a strategy that caps the service when consult volume or census becomes unmanageable. After the cap, new consults are declined and a “curbside” telephone consultation is offered. Requesting physicians can try again the following day if clinical need persists. This “first come, first served” strategy is explicitly tied to capacity, making it easier to explain refusals to requesting providers. It also offers flexibility in relation to the team's workload, ensuring that the maximum possible number of consults will be seen.

This strategy also has weaknesses. Like the previous strategy, it is insensitive to patient needs and potential impact. Providers can increase the likelihood of a consult by calling early in the morning or calling back on multiple days. However, this ability to “game the system” has benefits too. In effect, it decentralizes rationing decisions, placing those decisions in the hands of requesting providers. If MG's oncologist was more convinced that her patient needed palliative care than the resident caring for LK, she would call as early as possible (or call repeatedly until the consult was accepted). This rationing system is analogous to the burdensome prior authorizations that insurance companies require before costly tests. The amount of effort that a provider will expend to secure prior authorization (or a scarce palliative care consult) is directly proportionate to that provider's perception of likely benefit.

Conclusions

Under our model, MG and LK are equally likely to be seen, unless one patient's physician is more aggressive in placing consults. Consults with lower clinical and financial impact may be accepted early in the day, while higher-impact consults may be turned down later. Ultimately, we reassured ourselves that primary teams should be able to provide primary palliative care (with telephone guidance) and call back the following day. But there are occasions when a palliative care provider surreptitiously sees an urgent consult after the cap.

There are no easy answers for the increasing gap between supply and demand for subspecialty palliative care. Widespread palliative care education will take decades to implement. ⁵ In the meantime, palliative care teams wrestle with the unanswerable question of “which patient” needs palliative care most.