Clinical

Timing and targeting key to advance care planning discussions

Last Updated: 2014-02-05 10:52:26 -0500 (Reuters Health)

By Megan Brooks

NEW YORK (Reuters Health) - Timing and patient selection are crucial factors in effective advance care planning, say the authors of a Special Communication in JAMA Internal Medicine this week.

"Advance care planning is taught in undergraduate medical training and is familiar to practicing clinicians. However, the issues raised in our paper about timing and selection are not widely appreciated except for widespread unhappiness among clinicians about advance care planning being neglected, even for patients who obviously should be considering end-of-life decisions," J. Andrew Billings of Harvard Medical School Center for Palliative Care in Boston said in email to Reuters Health.

Billings and co-author Rachelle Bernacki say timing of advance care planning discussions is important because completing an advance directive (AD) too far from or too close to the time of death can lead to end-of-life decisions that do not optimally reflect the patient's values, goals, and preferences.

It's a bit like the adventures of Goldilocks who finds the soup too hot, too cold and then, just right. Decisions about limiting life-sustaining treatment can occur too early, too late, or just the right time, they say.

Patient selection is also important; "a poorly chosen target patient population that is unlikely to need an AD in the near future may lead to patients making unrealistic, hypothetical choices, while assessing preferences in the emergency department or hospital in the face of a calamity is notoriously inadequate," Billings and Bernacki write.

It all boils down to effective prognostication, they say. "Unfortunately, prognostication is difficult and empirical data do not offer the precision that clinicians desire." But some tools are available that can help, they say.

The "Surprise Question" has proven helpful in identifying patients at high risk of dying soon, at least among advanced cancer and dialysis patients, they say. Physicians simply ask themselves, "Would I be surprised if this patient died in the next year?" Studies show the hazard ratio of death in one year for patients with cancer whose clinician answered "no" to this question is 7.9. The HR for a "no" answer to the surprise question for those undergoing dialysis is 3.6.

Life expectancy tables can provide rough estimates of prognosis but they are "surprisingly unhelpful" for prognostication, Billings and Bernacki caution. Functional status, on the other hand, is a "consistent predictor of mortality in older adults."

There are also various disease-specific prognostic models including the Seattle Heart Failure model, the model for end-stage liver disease (MELD), and indices for renal failure. "However single disease models are problematic because many patients with chronic illness have multiple comorbidities," the authors say.

They consider this prognostic toolbox, which incorporates multiple comorbidities, to be helpful: http://www.eprognosis.org.

Billings and Bernacki conclude: "Guided by the Surprise Question, general awareness of patients' risk of dying soon or the use of prognostic indices, high-risk patients with identified poor functional status and significant comorbidities might readily define a target group in clinical practice - those at high risk for death in the foreseeable future and for whom the time is 'just right' to discuss and document choices about end-of-life care."

In the end, "the responsibility for initiating discussions about advance care planning belongs with clinicians," Dr. Billings told Reuters Health.

SOURCE: http://bit.ly/1bqTXYN

JAMA Intern Med 2014.