Racial Differences in Physicians’ Shared Decision-Making Behaviors during Intensive Care Unit Family Meetings

To the Editor:

Shared decision making (SDM) facilitates collaborative treatment decisions based on patient and family values, goals, and preferences (1). In intensive care unit (ICU) settings, effective SDM results in goal-concordant treatment decisions, improved psychological outcomes among families and clinicians, and shorter ICU lengths of stay (1–7). Despite evidence of racial disparities in clinician communication and ICU outcomes, it is unknown if ICU clinicians equitably engage racially minoritized families in SDM (8–10). Thus, our objective was to compare ICU physicians’ SDM behaviors in audio-recorded meetings with Black versus White families.

Methods

We conducted a mixed-methods study using qualitative data from a randomized trial of a decision aid about prolonged mechanical ventilation (⩾10 d) (11). The trial was conducted in 13 medical and surgical ICUs at 5 hospitals in North Carolina, Pennsylvania, and Washington. Families randomized to the intervention viewed a decision aid about prolonged mechanical ventilation outcomes, then ICU physicians held unscripted, audio-recorded meetings with families from both trial arms. From 146 meetings, we purposively sampled 129 that had complete recordings, were in English, and included a primary surrogate decision maker (hereafter referred to as “family member”) who self-identified as Black or White.

We used summative content analysis to quantify SDM behaviors (12). An a priori coding schema was adapted from a validated SDM framework for ICU physicians (Table E1 in the data supplement) (5, 13). A research team consisting of an ICU nurse (H.Y.), a palliative care physician (J.E.M.), a communication studies scientist (M.C.H.), and an ICU physician (D.C.A.), all with qualitative or health equity research expertise, met biweekly throughout the analysis. Four audio-recorded family meetings were analyzed by all team members to ensure consistency in code application through discussion, then the remainder were coded independently (14). Fourteen codes were summed to calculate an SDM score for each meeting.

Results

The sample included meetings with 22 (17.1%) Black and 107 (82.9%) White family members. Family members had a median age of 53.0 years (interquartile range [IQR], 44.0, 62.0 yr) and were predominantly female (75.2%). They endorsed high health literacy (median, 5.0; IQR, 4.0, 6.0), hope (96.9%), and social support (96.1%). Compared with White families, Black families were more commonly randomized to the intervention (59.1% vs. 45.8%) and less commonly interacted with racially concordant physicians (0% vs. 80.4%). Patients’ age and race were similar to those of their family members, and 26.4% died in the hospital (Table 1). Ninety-six physicians participated in one to four family meetings each. They had a median age of 35.0 years (IQR, 32.0, 42.0 yr), and most were White (75.0%), male (64.6%), and preferred to engage families as equal partners in decision making (80.4%) (Table 2).

Table 1.

Family member and patient demographics

	White Family Member	Black Family Member	Total
Number	107	22	129
Family member characteristics
Age, yr, median (IQR)	54.0 (45.0, 62.0)	48.5 (39.0, 56.0)	53.0 (44.0, 62.0)
Female, %	75.7	72.7	75.2
Relationship to patient, %
Partner	53.3	27.3	48.8
Parent	20.6	31.8	22.5
Child	14.0	27.3	16.3
Other	12.2	13.6	12.4
Randomized to intervention, %	45.8	59.1	48.1
Health literacy, median (IQR)*	5.0 (4.0, 6.0)	5.0 (3.0, 7.0)	5.0 (4.0, 6.0)
Endorses hope for future, %†	96.3	100	96.9
Endorses social support, %‡	98.1	86.4	96.1
Family–physician sex concordance, %
Female–female	25.2	18.2	24.0
Male–male	16.8	18.2	17.0
Discordant	58.0	63.6	58.9
Family–physician race concordance, %	80.4	0	66.7
Patient characteristics
Age, yr, median (IQR)	54.0 (41.0, 67.0)	48.0 (35.0, 59.0)	53.0 (40.0, 64.0)
Female, %	32.7	40.9	34.1
Race, %
White	97.2	0	80.6
Black	0.9	100	17.8
Other	1.9	0	1.6
Charlson comorbidity index score (29), median (IQR)	3.0 (1.0, 6.0)	3.0 (1.0,5.0)	3.0 (1.0, 6.0)
APACHE II score (30) at enrollment, median (IQR)	23.0 (18.0, 28.0)	23.5 (17.0, 31.0)	23.0 (18.0, 28.0)
In-hospital mortality, %	26.2	27.3	26.4

Definition of abbreviations: APACHE = Acute Physiology and Chronic Health Evaluation; IQR = interquartile range.

^*Ranging from 3 to 15 in order of decreasing health literacy.

^† The statement assessing hope was, “I look forward to the future with hope.”

^‡ The statement assessing social support was, “I feel that I have someone I can turn to for advice about making important decisions (like medical decisions).”

Table 2.

Physician demographics

Demographic Characteristic	Data
Number	96
Age, yr, median (IQR)	35.0 (32.0, 42.0)
Female, %	35.4
Race, %
White	75.0
Black	4.2
Asian	10.4
Other	10.4
Self-reported decision-making style, %
Leads decision making	10.9
Engages patient or family as equal partner	80.4
Allows patient or family to make decision	8.7

Definition of abbreviation: IQR = interquartile range.

Among 14 components of SDM, physicians addressed a median of 6.5 (IQR, 5.0, 8.0) with Black families and 6.0 (IQR, 4.0, 9.0) with White families. They most often discussed informational components, including discussing prognosis or the purpose of the visit (95.3% and 72.9%, respectively), in meetings with families of both races. Physicians less often engaged in preference-sensitive components, including eliciting preferences or linking decisions to preferences (51.9% and 43.4%, respectively), although no meaningful racial differences were evident. Regardless of family race, physicians least frequently broached difficult topics, including death or spirituality (17.1% and 7.8%, respectively). The only meaningful racial difference was that physicians provided less validation about medical decisions and family roles to Black than to White families (27.3% vs. 49.5%) (Figure 1).

Figure 1.

Frequency of shared decision-making components used by intensive care unit physicians in meetings with Black versus White family members, by order of frequency (percentage) among all family members.

Discussion

In this mixed-methods study, ICU physicians appeared to approach SDM similarly with Black and White families. However, they less frequently validated Black families’ medical decisions or involvement in care. Although ICU physicians commonly discussed informational components of SDM, they engaged in preference-sensitive communication in half of the meetings and discussed difficult topics in less than one-fourth of meetings.

That physicians addressed most components of SDM equally with Black and White families is encouraging. In particular, the high frequency of informational components, such as discussing prognosis, in both groups suggests that racially minoritized families have equal access to medical information about their critically ill loved ones. For other components of SDM, future studies should interrogate whether equality signals equity. For example, if Black families have greater spiritual needs than White families, an equitable approach may require more frequent discussion of spirituality with Black families (15).

We confirmed several findings from other studies. First, discussion of preferences did not occur in half of meetings, despite the need to make highly value-laden decisions (16–19). Second, physicians infrequently discussed death or spirituality, although we know that families want clinicians to do so (20, 21).

A notable finding was that physicians provided validation less commonly to Black than to White families. This is the only component of our coding schema that assessed an emotional rather than cognitive feature of SDM. Emotional support is increasingly recognized as necessary to promote high-quality decision making and psychological well-being among decision makers (11, 22, 23). Furthermore, validating families’ roles and decisions is critical to establishing trustworthiness, which is important for all families but is especially salient for those who are minoritized (24). Disparate empathic communication should be a focus of future study and may represent a novel mechanism of racial disparities in ICU outcomes (25). The effect of family–physician race concordance on communication should also be evaluated, although this may be challenging because of limited critical care workforce diversity (26, 27).

This study has several strengths, including a rich, multicenter qualitative database and an interdisciplinary analytic team. There are also limitations. First, the Hawthorne effect is possible (28). However, family meetings were unscripted, and racial bias was not a focus of the original trial; therefore, physicians are unlikely to have altered their approach to SDM as relevant to racially minoritized families. Second, in this hypothesis-generating study, we identified the presence of SDM components, but not how physicians communicated these to families (e.g., how prognosis was framed). Additional inductive qualitative analyses are needed. Third, we could not analyze nonverbal communication (13).

In conclusion, although ICU physicians appeared to approach SDM similarly with Black and White families, we identified a racial difference in their validation of Black families’ decisions and roles. We also confirmed opportunities to improve preference-sensitive communication and discussion of difficult topics with all families.