Physicians Rarely Elicit Critically Ill Patients’ Previously Expressed Treatment Preferences in Intensive Care Units

To the Editor:

Advance care planning is essential to improving end-of-life care (1). The Centers for Medicare and Medicaid Services has announced reimbursement for discussing advance care planning. One important element of advance care planning is patients discussing their treatment preferences with their surrogates. Conversations about previously expressed treatment preferences are important both because they may provide dispositive guidance for clinicians about what treatments the patient prefers and also because they may help surrogates make better in-the-moment judgments when the prior conversations did not address the specific clinical situation at hand.

The extent to which patients’ previously expressed treatment preferences are discussed during family meetings in intensive care units (ICUs) is unknown. We sought to determine the frequency with which ICU physicians and surrogates discuss patients’ previously expressed treatment preferences and how often those prior preferences provide dispositive guidance for goals-of-care decisions.

Methods

We conducted a prospective cohort study in six U.S. medical centers as previously described (2). We analyzed family meetings for patients with acute respiratory distress syndrome (ARDS). Eligible patients were aged 18 years or older, lacked decision-making capacity, required mechanical ventilation due to ARDS (3), and had a greater than or equal to 50% risk of in-hospital mortality or severe long-term functional impairment. We enrolled all willing, self-identified surrogate decision makers.

Surrogates completed a questionnaire on ICU Day 3 that included the question, “Have you ever discussed with your loved one the treatments he/she would want (or would not want) if he/she were too sick to speak for him/herself?” We audio-recorded clinician–family meetings that took place on ICU Day 5 to determine whether those treatment preferences were discussed.

Qualitative analysis

We coded the audio-recorded family meetings and summed the number of conferences that had any mention of patients’ previously expressed preferences using the modified grounded theory approach and constant comparative methods (4). We developed consensus through investigator meetings (J.C., L.P.S., N.C.E., R.M.A., and D.B.W.) and the literature (5–7). We defined the prior preferences as dispositive when they directly addressed the patient’s clinical circumstance and provided information that could be applied without additional inference or extrapolation. Three investigators (J.C., L.P.S., and N.C.E.) coded all transcripts. Coders had good interrater reliability (κ = 0.83) and were blinded to each other’s coding. We used ATLAS.ti software (Berlin, Germany) for qualitative data management.

Results

Participants and family meetings

Two hundred seventy-five surrogates for 245 patients consented to participate. Table 1 contains participant characteristics. The majority of surrogates were the patient’s spouse (22%) or child (36%). Most patients were full code at the time of enrollment (89%). In-hospital mortality was 44%. The mean (±SD) duration of the family meeting was 24 (± 14) minutes. The mean numbers of participants in meetings were 1.8 clinicians and 2.6 family members.

Table 1.

Characteristics of Study Participants

Patient Characteristic	Surrogates (n = 448)*			Patients (n = 245)†
	Did Not Discuss AD with Patient (n = 192)	Discussed AD with Patient (n = 256)	P Value‡	Did Not Discuss AD with Any Surrogates (n = 63)	Discussed AD with ≥1 Surrogate (n = 182)	P Value‡
Female	110 (57)	182 (71)	0.003	26 (41)	87 (48)	0.383
Age, yr, mean (SD)	48 (14)	51 (14)	0.022	52 (19)	61 (15)	0.002
White	145 (76)	204 (80)	0.303	51 (81)	146 (80)	0.999
Full code at enrollment	—	—	—	58 (92)	161 (88)	0.487
Relationship to patient	—	—	<0.001	—	—	—
Spouse	17 (9)	80 (31)		—	—	—
Child	66 (34)	96 (38)		—	—	—
Parent	34 (18)	21 (8)		—	—	—
Sibling	33 (17)	29 (11)		—	—	—
Other	42 (22)	30 (12)		—	—	—
In-hospital mortality	—	—		30 (48)	77 (42)	0.466

Definition of abbreviation: AD = advanced directives.

Data are presented as n (%) unless otherwise indicated.

^*Total of 456 surrogates attended physician–family conferences. Five surrogates did not complete baseline questionnaire and three surrogates did not answer question about discussing advanced directives with the patient. Therefore, n = 448.

^†Total of 249 patients had physician–family conferences. Four patients did not have surrogates answer the advanced directive question. Therefore, n = 245.

^‡Comparison tests used: t test and Fisher’s exact test.

Findings

Among the 245 patients, 182 (74%) had surrogates who had previously discussed treatment preferences with the patient. Patients’ prior treatment preferences were discussed in only 19% of conferences (35 of 182). Figure 1 illustrates the prevalence of discussions about patients’ treatment preferences. Clinicians first broached the topic in 33 of these 35 conferences. Physicians most often asked about prior direct conversations (17 of 33, 52%).

Figure 1.

Discussion about previously expressed treatment preferences during intensive care unit family meetings.

In only 1 out of 245 conferences (0.4%) did surrogates provide information about the patient’s previously expressed preferences that was dispositive for the clinical decisions at hand. In this conference, the physician asked, “If he survives to leave the hospital, is he going to wake up enough to be off the ventilator? Have you talked about those kinds of issues before?” The surrogate responded, “He was always like any chance of survival, he’d want to be kept going.” This led to a decision for tracheostomy.

In the remaining conferences in which patients’ treatment preferences were discussed, in 22 of 35 (63%), the patient’s prior preferences did not directly apply to the current clinical condition. For example, one surrogate of a patient with an uncertain prognosis said, “We know that her wishes…she doesn’t want to be intubated if there’s no chance of recovery.” In 12 of 34 (34%) cases, the prior treatment preferences discussed during the family conference did not address the life-sustaining treatments under consideration.

Discussion

We found that, although most patients had some prior conversation with their surrogates about treatment preferences, those preferences were rarely elicited by physicians during goals-of-care conversations in ICUs. When discussed, they rarely provided dispositive guidance regarding goals of care.

Our findings confirm and extend the results of a smaller study reporting no discussion of patients’ prior preferences in roughly 60% of cases (8). Given that patients often have discussed their treatment preferences with their surrogates, interventions are needed to help clinicians effectively explore patients’ preferences and use this information to make patient-centered decisions.

Our finding that information about treatment preferences is rarely dispositive extends prior work reporting that only 3% of directives contain instructions that direct care in the patient’s current clinical condition (9). In aggregate, these studies suggest that it may not be realistic to expect patients to articulate treatment preferences in advance of illness in a way that completely removes the need for surrogates to exercise judgment at the moment of decision-making. The policy implication of these findings is that advance care planning should be directed at preparing surrogates for in-the-moment decision-making rather than trying to articulate an exhaustive list of treatment preferences (10). In addition, it may be valuable for clinicians to reframe their expectations about how patients’ previously expressed preferences will contribute to goals-of-care conversations in ICUs by moving away from expecting the information to be dispositive and toward using the information as a starting point for conversations about the patients’ values.

Our study’s limitations include that we ascertained the presence of prior advance care planning and discussion of treatment preferences through examining family meetings, thereby potentially underestimating prevalence. In addition, we recorded a single meeting per patient and cannot exclude that preferences were discussed in other encounters.

In conclusion, we found that surrogates often had prior conversations about patients’ treatment preferences; however, these preferences were rarely discussed in ICU family meetings. These findings support developing novel strategies to help clinicians and surrogates incorporate patients’ values and preferences into decisions about goals of care in ICUs.