Abstract
Background/Objective:
Patients with a severe acute brain injury admitted to the intensive care unit often have a poor neurological prognosis. In these situations, a clinician is responsible for conducting a goals of care conversation with the patient’s surrogate decision makers. The diversity in thought and background of surrogate decision makers can present challenges during these conversations. For this reason, our study aimed to identify predictive characteristics of US surrogate decision makers’ favoring life sustaining treatment (LST) over comfort measures only (CMO) for patients with severe acute brain injury.
Methods:
We analyzed data from a cross-sectional survey study that had recruited 1588 subjects from an online probability-based US population sample. 792 subjects had randomly received a hypothetical scenario regarding a relative intubated with severe acute brain injury with a prognosis of severe disability but with the potential to regain some consciousness. 796 subjects had been randomized to a similar scenario in which the relative was projected to remain vegetative. For each scenario, we conducted univariate analyses and binary logistic regressions to determine predictors of LST selection among available respondent characteristics.
Results:
15.0% of subjects selected LST for the severe disability scenario compared to 11.4% for the vegetative state scenario (p=0.07), with those selecting LST in both groups expressing less decisional certainty. For the severe disability scenario, independent predictors of LST included having less than a high school education (adjusted OR=2.87, 95% CI=1.23-6.76), concern regarding prognostic accuracy (7.64, 3.61-16.15), and concern regarding the cost of care (4.07, 1.80-9.18). For the vegetative scenario, predictors included the youngest age group (30-44 years, 3.33, 1.02-10.86), male gender (3.26, 1.75-6.06), English as a second language (2.94, 1.09-7.89), Evangelical Protestant (3.72, 1.28-10.84) and Catholic (4.01, 1.72-9.36) affiliations, and low income (<$25K).
Conclusion:
Several demographic and decisional characteristics of US surrogate decision makers predict LST selection for patients with severe brain injury with varying degrees of poor prognosis. Surrogates concerned about the cost of medical care may nevertheless be inclined to select LST, albeit with high levels of decisional uncertainty, for patients projected to have severe disabilities.
Keywords: Intensive Care Units, Decision Making, Palliative Care, Family, Choice Behavior, Brain Injuries
INTRODUCTION
Many patients admitted to intensive care units (ICUs) with severe acute brain injury (SABI, e.g., ischemic stroke, intracerebral hemorrhage, traumatic brain injury, subarachnoid hemorrhage, hypoxic-ischemic encephalopathy, etc.) are projected to have poor neurologic prognoses despite aggressive management (1,2). Surrogate decision makers work with medical teams to exercise substituted judgment on behalf of these patients to make decisions regarding the continuation of life-sustaining treatment (LST), including tracheostomy and/or gastrostomy tube placement, versus transitioning to comfort measures only (CMO) (3–5). Despite how common these decisions are for neurologic patients in ICUs, studies examining characteristics of surrogates that predict their choices in these situations have been limited (6–9). There have been urgent calls in the neurocritical care community for additional research on factors that may influence surrogates to choose LST or CMO (10,11), with hopes that such understanding may facilitate improvements in ICU family support.
A recently conducted general US population survey characterized surrogates for hypothetical, intubated older adult SABI patients with varying degrees of poor prognosis into distinct groups based on their top concerns regarding goals-of-care decision making (12). The initial analysis of this data assigned respondents into “decisional groups,” each with different top concerns. For respondents who were given a prognostic scenario of a patient projected to have severe disability—but with the possibility of regaining limited consciousness—four distinct, separate decisional groups emerged with relative concerns regarding (1) patient’s advanced age, (2) family agreement on a decision, (3) prognostic accuracy, and (4) cost of care. For a different cohort of respondents who were given a scenario of a patient projected to remain in a vegetative state, the identified groups included the four listed above, as well as a separate, fifth group relatively concerned about religious beliefs. In addition to characterizing participants into decisional groups, the survey also asked participants to indicate whether they would personally select LST versus CMO for the hypothetical patients and their level of decisional certainty.
The purpose of this follow-up analysis was to discover independent predictors of LST selection amongst potential SABI surrogate decision makers in a variety of prognostic scenarios. Both traditional demographic and socioeconomic predictors, as well as the decisional group memberships as described above in the previous study, were included as potential predictors.
METHODS
This study is a secondary analysis of a dataset from a previously published online cross-sectional survey, conducted from February 18-29th in 2016 amongst the US population age 30 years or older (12). The study was approved by the Yale Human Investigation Committee.
Dataset description
A summary of relevant details regarding how the dataset was obtained and initially analyzed is provided below (12). For survey sampling, 1588 adults thirty years or older were recruited from GfK (now Ipsos) KnowledgePanel. KnowledgePanel is a probability-based online panel designed to cover 97% of US households by utilizing address-based sampling and providing panel members who do not have computers or internet service with both (13). The study had a 44.6% response rate amongst the KnowledgePanel members contacted for participation.
Study participants were randomized to one of the two survey versions. Both surveys asked a participant to consider an 85-year-old, previously healthy, close family member admitted to an ICU following a severe intracerebral hemorrhage who remains intubated and poorly responsive despite two weeks of care without limitations. Although the two surveys included very similar language regarding the hypothetical patient’s injury, each presented a different prognostic outcome. One version stated that the patient would most likely remain unresponsive, with a small but unlikely chance that the patient could gradually recover the ability to speak, eat, recognize others, and/or travel in a wheelchair with time (the “severe disability” survey, SD, Supplement 1). The other version stated a less common but more severe scenario that the patient would remain bedbound and unresponsive long-term with certainty (the “vegetative state” survey, VS, Supplement 1). For both surveys, participants were then informed that they would be responsible for guiding the medical team to pursue tracheostomy and gastrostomy placement (LST) or CMO for the patient.
Following the clinical scenario, participants for both surveys answered a series of Best-Worst Scaling questions designed to elicit their top decision-making concerns regarding the goals-of-care decision (14,15). Participants were also asked whether they would ultimately favor LST or CMO for the patient, certainty in their decision, and demographic information.
Latent class analysis was performed on the Best-Worst Scaling data from the two surveys separately, with data undergoing iterative proportional fitting to weight and further match the sample geodemographically to the US population age 30 and older (16). For participants taking the SD survey, this analysis yielded four distinct decisional groups: (1) those relatively concerned with their loved one being too old to be kept alive and severely disabled; (2) those relatively concerned with family agreement on goals-of-care; (3) those relatively concerned that the doctor’s prognosis may be incorrect; and (4) those relatively concerned with paying for long-term care. For participants taking the VS survey, the latent class analysis yielded five distinct decisional groups: four of the groups were similar to those for the SD survey, with a fifth group relatively concerned about acting according to religious beliefs. The result of this original analysis allowed each survey participant in the dataset to be assigned as a member of a decisional group that best fits their survey responses.
Sample
For the analysis described in this manuscript, all survey participants in the original dataset described above were included.
Variables
For each survey participant, information on age, gender, race, education, household size, household internet access, region of residence, marital status, employment status, and annual income was obtained directly from KnowledgePanel records. All other demographic variables for participants were obtained directly from the survey. Each participant’s decisional group assignment from the original study was considered as a predictor variable, in addition to the demographic data listed above. Participants who had been randomized to the SD survey were labeled as a member of one of the four decisional groups resulting from the original survey analysis, while those randomized to the VS survey were each labeled as a member of one of the five groups that that survey analysis revealed.
Age data was available via KnowledgePanel both as a continuous variable and as categorical age groups. Household size and years in the US were reported as continuous variables. All other variables were categorical.
Outcome
The primary outcome of this study was participant selection of LST versus CMO for the randomized hypothetical scenario. A secondary outcome was the participant’s self-reported certainty in their selection measured as a categorical variable with participants choosing “Very Certain,” “Certain,” “Uncertain,” or “Very Uncertain”.
Statistical analysis
Data from the SD survey and the VS survey were analyzed separately. Weights included from the original analyses were incorporated into all analyses. For each survey, univariate associations of demographic characteristics and decisional group assignments with LST versus CMO selection were first tabulated, using standard methods for continuous and categorical variables. Variables with p-values 0.1 or lower in the univariate analyses were then included in subsequent binary logistic regressions exploring independent predictors of LST selection for each of the two prognostic scenarios. Only those surveys with complete data were included in the binary regressions. All analyses were conducted using SAS software v14 (SAS Institute Inc., Cary, NC).
With regards to sample size, this dataset was powered for the previously published initial latent class analyses, and the post-hoc analyses in this manuscript were exploratory.
RESULTS
Survey randomization
Table 1 summarizes demographic information for the SD and VS survey participants. There were no demographic differences between the participants randomized to either survey. The samples for both survey versions reflected the geodemographic distribution of the US population aged 30 years or older after weighting.
Table 1.
Comparison of participants randomized to the severe disability versus vegetative state surveys
| Severe disability | Vegetative state | P | |
|---|---|---|---|
| n | 792 | 796 | |
| Data provided by KnowledgePanel | |||
| Age, mean (SD), years | 53.6 (14.3) | 53.3 (14.2) | 0.73 |
| Age groups, n (%) | |||
| 30-44 | 160 (31.8) | 149 (30.55) | 0.87 |
| 45-59 | 272 (33.9) | 272 (35.1) | |
| 60+ | 360 (34.3) | 375 (34.3) | |
| Gender, n (%), male | 369 (47.7) | 385 (47.7) | 1.00 |
| Race, n (%) | |||
| White, non-Hispanic | 593 (67.5) | 605 (67.6) | 1.00 |
| Other | 199 (32.5) | 191 (32.4) | |
| Education, n (%) | |||
| Less than high school | 72 (11.9) | 64 (11.7) | 1.00 |
| High school | 242 (29.9) | 269 (30.0) | |
| Some college | 227 (26.1) | 200 (26.1) | |
| Bachelor’s degree or higher | 251 (32.1) | 263 (32.2) | |
| Household size, mean (SD), persons | 2.61 (1.4) | 2.59 (1.6) | 0.79 |
| Household head, n (%), yes | 698 (85.9) | 719 (87.2) | 0.54 |
| Housing type, n (%) | |||
| One-family house | 623 (75.2) | 626 (74.8) | 0.98 |
| Apartment building | 132 (19.8) | 134 (19.9) | |
| Other: mobile home, boat, etc. | 37 (5.1) | 36 (5.3) | |
| Household internet access, n (%), yes | 621 (76.7) | 657 (76.6) | 0.98 |
| Region of residence, n (%) | |||
| Northeast | 147 (18.1) | 147 (18.1) | 1.00 |
| Midwest | 189 (21.5) | 199 (21.5) | |
| South | 294 (37.2) | 280 (37.3) | |
| West | 162 (23.2) | 170 (23.1) | |
| Marital status, n (%) | |||
| Married or living with partner | 507 (64.4) | 517 (63.0) | 0.83 |
| Widowed | 58 (5.9) | 54 (5.3) | |
| Divorced or separated | 123 (14.1) | 119 (15.5) | |
| Never married | 104 (15.5) | 106 (16.2) | |
| Employment status, n (%) | |||
| Working | 399 (55.9) | 385 (56.0) | 0.29 |
| Not working – retired | 243 (23.6) | 261 (24.0) | |
| Not working – disabled | 55 (5.9) | 66 (8.0) | |
| Not working – other | 95 (14.5) | 84 (12.1) | |
| Income, n (%) | |||
| Under $24,999 | 139 (17.5) | 135 (17.6) | |
| $25,000-$74,999 | 315 (38.2) | 328 (38.4) | 0.86 |
| $75,000-$149,999 | 279 (35.9) | 262 (34.4) | |
| $150,000 or more | 59 (7.8) | 71 (9.2) | |
| Data obtained by survey | |||
| Years in the US, n | 783.5 | 785.8 | |
| Years, mean (SD) | 51.5 (16.5) | 51.2 (16.2) | 0.77 |
| First language, n | 789.5 | 793.2 | |
| Non-English, n (%) | 36 (7.0) | 40 (7.6) | 0.63 |
| Religion, n | 786.7 | 785.8 | |
| Evangelical Protestant, n (%) | 100 (12.5) | 98 (11.1) | 0.71 |
| Catholic | 178 (22.6) | 197 (26.0) | |
| Other Christian | 282 (33.5) | 278 (32.9) | |
| Other faith | 75 (10.2) | 71 (9.7) | |
| Not affiliated with religion | 153 (21.0) | 146 (20.4) | |
| Frequency of religious services, n | 788 | 788 | |
| Once a year or less, n (%) | 351 (45.5) | 343 (44.7) | 0.48 |
| A few times a year | 140 (18.5) | 123 (16.0) | |
| A few times a month | 69 (8.3) | 73 (10.1) | |
| Once a week or more | 228 (27.7) | 249 (29.2) | |
| Prior experience caring for disabled, n | 790 | 793 | |
| Yes, n (%) | 209 (23.3) | 220 (25.0) | 0.47 |
| Previously made similar decision, n | 785 | 786 | |
| Yes, n (%) | 289 (31.8) | 290 (33.3) | 0.55 |
| Outcome of prior similar decision, n | 286 | 288 | |
| LST, n(%) | 35 (14.9) | 26 (11.4) | 0.32 |
| Goals-of-care decision in survey | |||
| Decision for patient in survey, n | 785 | 793 | |
| LST, n(%) | 104 (15.0) | 81 (11.4) | 0.07 |
| Certainty of decision, n | 791 | 796 | |
| Very certain, n (%) | 389 (45.7) | 418 (48.1) | 0.72 |
| Certain | 273 (37.8) | 267 (36.3) | |
| Uncertain/Very uncertain | 129 (16.5) | 111 (15.6) |
LST = selection of life-sustaining treatment by survey participant. SD = standard deviation.
All counts in the table are raw survey numbers, while all percentages factor in the weights assigned to each participant.
All means are weighted means.
For “Data obtained by survey” and “Goals-of-care decision in survey,” the n listed next to each variable heading are the number of participants responding to each question.
Summary of survey outcomes
Table 1 also summarizes LST versus CMO selection amongst the SD and VS survey participants. A higher percentage of participants presented with the SD scenario selected LST (104 of 785 participants answering the survey question, weighted percentage = 15.0%) compared to participants presented with the VS scenario (81/793, 11.4%, p=0.07).
For the SD survey, a higher percentage of participants selecting LST reported being “Uncertain” or “Very uncertain” with their decision (55/104, 51.3%) compared to the percentage amongst participants selecting CMO (71/681, 10.1%, p<0.0001). This difference was also observed amongst responses to the VS survey (40/81, 46.0% vs. 69/712, 11.4%, p<0.0001).
Univariate analysis of severe disability survey
Table 2 lists univariate comparisons of participant characteristics between those choosing LST versus CMO for the SD survey. Participants choosing LST tended to be less likely to identify as non-Hispanic white (55.1% vs. 69.7%, p=0.01), less educated (e.g., less than high school education, 24.2% vs. 9.7%, p=0.001), and less likely to have internet access (67.8% vs. 78.3%, p=0.03).
Table 2.
Univariate comparison of participants selecting Life Sustaining Therapy (LST) versus Comfort Measures Only (CMO) for the severe disability survey (n=785)
| LST | CMO | p | |
|---|---|---|---|
| n (%) | 104 (15.0) | 681 (85.0) | |
| Data provided by KnowledgePanel | |||
| Age, mean (SD), years | 53.4 (14.2) | 53.6 (14.3) | 0.16 |
| Age groups, n (%) | |||
| 30-44 | 29 (43.2) | 131 (30.0) | 0.05 |
| 45-59 | 36 (30.5) | 233 (34.5) | |
| 60+ | 39 (27.1) | 317 (35.4) | |
| Gender, n (%), male | 53 (52.5) | 311 (46.6) | 0.30 |
| Race, n (%) | |||
| White, non-Hispanic | 64 (55.1) | 525 (69.7) | 0.01 |
| Other | 40 (44.9) | 156 (30.3) | |
| Education, n (%) | |||
| Less than high school | 18 (24.2) | 53 (9.7) | 0.001 |
| High school | 31 (29.3) | 209 (30.0) | |
| Some college | 29 (23.6) | 195 (26.5) | |
| Bachelor’s degree or higher | 26 (22.8) | 224 (33.8) | |
| Household size, mean (SD), persons | 2.59 (1.56) | 2.61 (1.43) | 0.06 |
| Household internet access, n (%), yes | 70 (67.8) | 547 (78.3) | 0.03 |
| Region of residence, n (%) | |||
| Northeast | 18 (14.4) | 128 (18.8) | 0.15 |
| Midwest | 19 (17.8) | 168 (22.0) | |
| South | 48 (48.3) | 243 (35.3) | |
| West | 19 (19.5) | 142 (23.9) | |
| Marital status, n (%) | |||
| Married or living with partner | 65 (65.3) | 438 (64.4) | 0.54 |
| Widowed | 5 (4.2) | 52 (6.1) | |
| Divorced or separated | 16 (11.9) | 106 (14.5) | |
| Never married | 18 (19.5) | 85 (14.8) | |
| Employment status, n (%) | |||
| Working | 59 (60.5) | 337 (55.2) | 0.15 |
| Not working – retired | 23 (16.5) | 218 (25.0) | |
| Not working – disabled | 5 (4.0) | 50 (6.3) | |
| Not working – other | 17 (19.0) | 76 (13.6) | |
| Income, n (%) | |||
| Under $24,999 | 22 (22.9) | 116 (16.6) | 0.19 |
| $25,000-$74,999 | 44 (42.4) | 266 (37.7) | |
| $75,000-$149,999 | 31 (29.7) | 247 (37.2) | |
| $150,000 or more | 7 (5.1) | 52 (8.4) | |
| Data obtained by survey | |||
| Years in the US, n | 104 | 675 | |
| Years, mean (SD) | 51.2 (16.2) | 51.5 (16.5) | 0.07 |
| First language, n | 104 | 680 | |
| Non-English, n (%) | 7 (11.0) | 29 (6.3) | 0.20 |
| Religion, n | 103 | 679 | |
| Evangelical Protestant, n (%) | 20 (16.8) | 79 (11.9) | 0.07 |
| Catholic | 26 (25.4) | 151 (22.2) | |
| Other Christian | 41 (39.1) | 238 (32.4) | |
| Other faith | 7 (8.9) | 67 (10.5) | |
| Not affiliated with religion | 9 (9.8) | 144 (23.1) | |
| Frequency of religious services, n | 104 | 678 | |
| Once a year or less, n (%) | 35 (33.7) | 315 (47.7) | 0.12 |
| A few times a year | 19 (21.7) | 120 (17.9) | |
| A few times a month | 11 (9.8) | 57 (8.0) | |
| Once a week or more | 39 (34.8) | 186 (26.3) | |
| Prior experience caring for disabled, n | 103 | 681 | |
| Yes, n (%) | 28 (26.5) | 180 (22.9) | 0.52 |
| Previously made similar decision, n | 104 | 675 | |
| Yes, n (%) | 29 (26.3) | 260 (32.9) | 0.20 |
| Outcome of prior similar decision, n | 29 | 257 | |
| LST, n (%) | 17 (67.7) | 18 (7.4) | <0.0001 |
| Decisional group, n | 104 | 681 | |
| Concern re: patient age, n (%) | 15 (12.0) | 256 (38.6) | <0.0001 |
| Concern re: family agreement | 17 (16.6) | 198 (16.9) | |
| Concern re: prognostic accuracy | 49 (43.2) | 119 (16.9) | |
| Concern re: cost of care | 23 (28.2) | 108 (16.6) |
LST = life-sustaining treatment. CMO = comfort measures only. SD = standard deviation. Outcome of prior similar decision = for participants indicating that they had previously made a similar decision, selection of LST vs. CMO for that prior decision.
All counts in the table are raw survey numbers, while all percentages factor in the weights assigned to each participant. All means are weighted means.
For “Data obtained by survey,” the n listed next to each variable heading are the number of participants responding to each question/exercise.
The distribution of survey participants’ membership in decisional groups (i.e., concern regarding patient age, family agreement, prognostic accuracy, and cost of care) differed significantly between those selecting LST versus CMO (p<0.0001), with 43.2% of participants selecting LST being in the “prognostic accuracy” concern group versus 16.9% of those selecting CMO. Of note, 28.2% of participants selecting LST were in the “cost of care” concern group, versus 16.6% of those selecting CMO.
Binary logistic regression of severe disability survey
Table 3 contains the results of a binary logistic regression model conducted to identify independent predictors of LST selection among SD survey participants. Having less than a high school education was predictive of LST selection in comparison to having a bachelor’s degree or higher (adjusted odds ratio = 2.87, 95% confidence interval = 1.23-6.76). In contrast, the confidence intervals for the odds ratios for race and household internet access did not meet thresholds for significance.
Table 3.
Binary logistic regression for predictors of Life Sustaining Therapy (LST) selection among the severe disability survey participants
| Odds Ratio | 95% Confidence Interval | |
|---|---|---|
| Age groups | ||
| 60+ | ||
| 45-59 | 1.16 | 0.60 – 2.22 |
| 30-44 | 2.07 | 0.70 – 6.11 |
| Race | ||
| Other | ||
| White, non-Hispanic | 0.67 | 0.39 – 1.15 |
| Education | ||
| Bachelor’s degree or higher | ||
| Some college | 1.13 | 0.57 – 2.25 |
| High school | 1.23 | 0.58 – 2.20 |
| Less than high school | 2.88 | 1.23 – 6.76 |
| Household size | 1.10 | 0.91 – 1.33 |
| Household internet access | ||
| No | 1.75 | 0.94 – 3.25 |
| Years in US | 1.00 | 0.96 – 1.03 |
| Religion, OR | ||
| Not affiliated with religion | ||
| Evangelical Protestant | 2.00 | 0.73 – 5.48 |
| Catholic | 1.91 | 0.72 – 5.05 |
| Other Christian | 1.84 | 0.76 – 4.50 |
| Other faith | 1.14 | 0.39 – 3.33 |
| Decisional group | ||
| Concern re: patient age | ||
| Concern re: family agreement | 1.57 | 0.67 – 3.68 |
| Concern re: prognostic accuracy | 7.64 | 3.61 – 16.15 |
| Concern re: cost of care | 4.07 | 1.80 – 9.18 |
LST = life-sustaining therapy. OR = odds ratio. CI = confidence interval.
All percentages are weighted percentages.
Demographic variables in Table 2 that had p≤0.10 were included in this analysis.
Affiliation with the “prognostic accuracy” decisional concern group was a strong predictor of LST selection (7.64, 3.61-16.15), compared with affiliation with the reference group most concerned that the patient in the hypothetical scenario was too old to live with significant disability. Affiliation with the “cost of care” decisional concern group was also predictive of LST selection in comparison with the same reference group (4.07, 1.80-9.18).
Univariate analysis of vegetative state survey
Table 4 lists univariate comparisons of participant characteristics between those choosing LST versus CMO for the VS survey. The group of respondents selecting LST for this clinical scenario was younger (mean age 49.3 years, standard deviation 16.3; vs 53.9 years, 16.0, p=0.01), had a higher percentage of males (68.9% vs. 45.2%, p<0.0001) and non-native English speakers (15.7% vs. 7.0%, p=0.04), and had lived a lower mean number of years in the US (46.2 years, 20.8 vs. 51.9, 18.31, p=0.01). Those selecting LST also differed from those selecting CMO with respect to employment status (e.g., working, 61.1% vs. 55.1%, p<0.03) and income (e.g., under $25K, 28.9% vs. 15.8%).
Table 4.
Univariate comparison of participants selecting Life Sustaining Therapy (LST) versus Comfort Measures Only (CMO) for the vegetative state survey (n=793)
| LST | CMO | P | |
|---|---|---|---|
| n (%) | 81 (11.4) | 712 (88.6) | |
| Data provided by KnowledgePanel | |||
| Age, mean (SD), years | 49.31 (16.3) | 53.93 (16.0) | 0.01 |
| Age groups, n (%) | |||
| 30-44 | 24 (44.4) | 124 (28.6) | 0.01 |
| 45-59 | 32 (34.4) | 240 (35.5) | |
| 60+ | 25 (22.2) | 348 (35.9) | |
| Gender, n (%), male | 51 (68.9) | 333 (45.2) | <0.0001 |
| Race, n (%) | |||
| White, non-Hispanic | 54 (60.0) | 549 (68.7) | 0.17 |
| Other | 27 (40.0) | 163 (31.3) | |
| Education, n (%) | |||
| Less than high school | 9 (14.4) | 54 (11.3) | 0.67 |
| High school | 34 (33.3) | 235 (29.6) | |
| Some college | 18 (25.6) | 181 (26.2) | |
| Bachelor’s degree or higher | 20 (26.7) | 242 (32.8) | |
| Household size, mean (SD), persons | 2.49 (1.6) | 2.60 (2.2) | 0.56 |
| Household internet access, n (%), yes | 62 (74.4) | 593 (77.1) | 0.60 |
| Region of residence, n (%) | |||
| Northeast | 19 (22.2) | 127 (17.5) | 0.12 |
| Midwest | 24 (25.6) | 175 (21.2) | |
| South | 30 (41.1) | 249 (36.8) | |
| West | 8 (11.1) | 161 (24.6) | |
| Marital status, n (%) | |||
| Married or living with partner | 48 (57.8) | 468 (63.8) | 0.05 |
| Widowed | 5 (4.4) | 49 (5.4) | |
| Divorced or separated | 10 (11.1) | 108 (16.1) | |
| Never married | 18 (26.7) | 87 (14.7) | |
| Employment status, n (%) | |||
| Working | 43 (61.1) | 340 (55.1) | 0.03 |
| Not working – retired | 19 (17.8) | 241 (24.8) | |
| Not working – disabled | 13 (14.4) | 53 (7.3) | |
| Not working – other | 6 (6.7) | 78 (12.8) | |
| Income, n (%) | |||
| Under $24,999 | 24 (28.9) | 109 (15.8) | 0.03 |
| $25,000-$74,999 | 34 (40.0) | 294 (38.5) | |
| $75,000-$149,999 | 18 (24.4) | 243 (36.2) | |
| $150,000 or more | 5 (6.7) | 66 (9.5) | |
| Data obtained by survey | |||
| Years in the US, n | 78 | 706 | |
| Years, mean (SD) | 46.24 (20.8) | 51.93 (18.3) | 0.01 |
| First language, n | 79 | 711 | |
| Non-English, n (%) | 9 (15.7) | 31 (7.0) | 0.04 |
| Religion, n | 80 | 708 | |
| Evangelical Protestant, n (%) | 13 (14.6) | 85 (10.6) | 0.07 |
| Catholic | 29 (37.1) | 167 (24.5) | |
| Other Christian | 20 (25.8) | 258 (34.0) | |
| Other faith | 9 (11.2) | 62 (9.5) | |
| Not affiliated with religion | 9 (12.4) | 136 (21.5) | |
| Frequency of religious services, n | 79 | 707 | |
| Once a year or less, n (%) | 28 (36.4) | 313 (45.6) | 0.58 |
| A few times a year | 15 (18.2) | 108 (15.7) | |
| A few times a month | 8 (11.4) | 65 (10.1) | |
| Once a week or more | 28 (34.1) | 221 (28.6) | |
| Prior experience caring for disabled, n | 80 | 710 | |
| Yes, n (%) | 21 (23.6) | 198 (25.1) | 0.74 |
| Previously made similar decision, n | 79 | 704 | |
| Yes, n (%) | 24 (31.8) | 266 (33.2) | 0.74 |
| Outcome of prior similar decision, n | 24 | 264 | |
| LST, n(%) | 11 (53.6) | 15 (6.0) | <0.0001 |
| Decisional group, n | 81 | 712 | |
| Concern re: prognostic accuracy, n (%) | 37 (44.4) | 189 (27.1) | 0.01 |
| Concern re: family agreement | 9 (11.1) | 167 (22.8) | |
| Concern re: patient age | 9 (13.3) | 158 (21.1) | |
| Concern re: religion | 12 (13.3) | 125 (16.7) | |
| Concern re: cost of care | 14 (18.9) | 73 (12.1) |
LST = life-sustaining treatment. CMO = comfort measures only. SD = standard deviation. Outcome of prior similar decision = for participants indicating that they had previously made a similar decision, selection of LST vs. CMO for that prior decision.
All counts in the table are raw survey numbers, while all percentages factor in the weights assigned to each participant.
All means are weighted means.
For “Data obtained by survey,” the n listed next to each variable heading are the number of participants responding to each question/exercise.
The distribution of survey participants’ membership in decisional groups (i.e., concern regarding patient age, family agreement, prognostic accuracy, religious beliefs, and cost of care) also differed significantly between those selecting LST vs. CMO (p=0.01). Of note, 44.4% of participants selecting LST were affiliated with the “prognostic accuracy” concern group, versus 27.1% of those selecting CMO. Furthermore, 18.9% of participants selecting LST were affiliated with the “cost of care” concern group, versus 12.1% of those selecting CMO.
Binary logistic regression of vegetative state survey
Table 5 contains the results of a binary logistic regression model conducted to identify independent predictors of LST selection among VS survey participants. Covariates that were predictive of LST selection were the youngest age group (ages 30-44, 3.33, 1.02-10.86, compared to ages 60+), male gender (3.27, 1.76-6.06), English as a second language (2.94, 1.10-7.89), and Evangelical (3.72, 1.28-10.84) and Catholic (4.01, 1.72-9.36) Christian denominations (in comparison to the “Not affiliated with religion” level). Of note, reporting an annual income of under $25K was predictive of LST selection, in comparison to all higher income levels.
Table 5.
Binary logistic regression for predictors of Life Sustaining Therapy (LST) selection among the vegetative state survey participants
| Odds Ratio | 95% Confidence Interval | |
|---|---|---|
| Age groups | ||
| 60+ | ||
| 45-59 | 1.86 | 0.78 – 4.39 |
| 30-44 | 3.33 | 1.02 – 10.86 |
| Gender | ||
| Male | 3.27 | 1.76 – 6.06 |
| Marital status | ||
| Married or living with partner | ||
| Widowed | 0.85 | 0.23 – 3.12 |
| Divorced or separated | 0.69 | 0.29 – 1.65 |
| Never married | 1.33 | 0.64 – 2.79 |
| Employment status, | ||
| Working | ||
| Not working – retired | 1.36 | 0.58 – 3.17 |
| Not working – disabled | 1.52 | 0.59 – 3.88 |
| Not working – other | 0.34 | 0.11 – 1.01 |
| Income | ||
| Under $24,999 | ||
| $25,000-$74,999 | 0.41 | 0.18 – 0.96 |
| $75,000-$149,999 | 0.26 | 0.10 – 0.66 |
| $150,000 or more | 0.24 | 0.07 – 0.80 |
| Years in US | 1.00 | 0.97 – 1.03 |
| English Language, | ||
| Non-English | 2.94 | 1.10 – 7.89 |
| Religion | ||
| Not affiliated with religion | ||
| Evangelical Protestant | 3.72 | 1.28 – 10.84 |
| Catholic | 4.01 | 1.72 – 9.36 |
| Other Christian | 2.63 | 1.00 – 6.90 |
| Other faith | 2.67 | 0.80 – 8.93 |
| Decisional group | ||
| Concern re: patient age | ||
| Concern re: prognostic accuracy | 2.35 | 0.94 – 5.93 |
| Concern re: family agreement | 0.77 | 0.25 – 2.38 |
| Concern re: religion | 0.95 | 0.33 – 2.76 |
| Concern re: cost of care | 1.09 | 0.35 – 3.339 |
LST = life-sustaining therapy. OR = odds ratio. CI = confidence interval.
Demographic variables in Table 4 that had p≤0.10 were included in this analysis.
Overall, affiliations with the 5 decisional groups were not significant predictors of LST selection in the VS survey multivariate analysis, although affiliation with the “prognostic accuracy” decisional concern group trended towards being predictive of LST (adjusted OR 2.35, 0.94-5.93).
DISCUSSION
When asked to make a goals-of-care decision on behalf of a hypothetical intubated older adult relative with an SABI projected to have severe disability, respondents were more inclined to select tracheostomy and feeding tube placement if they had less than a high school education or belonged in the decisional groups concerned with prognostic accuracy or cost of care. When asked to make a similar decision for a relative projected to remain in a vegetative state, respondents were more inclined towards selecting LST if they were young, were male, spoke English as a second language, were Evangelical Christian or Catholic, or had a low annual income. Overall, respondents selecting LST for either prognostic scenario were more uncertain regarding their decisions compared to those selecting CMO, a finding likely reflective of these respondents having higher levels of conflict amongst their top concerns in decision making.
These findings, focused on surrogate-level decisions for critically ill brain-injured patients, complement prior research on patient-level demographic predictors of high-intensity care at the end-of-life. Prior studies examining patient-level variables have reported that male gender, religious affiliation, and lower levels of education are all associated with a patient receiving increased intensity of end-of-life care or expressing a personal preference for such care (17–20). Of note, prior studies have repeatedly described non-white individuals as less likely to prefer or experience care limitations in scenarios of general severe illness (17,18,21–25) and specifically SABI (26–30). While the exact reasons for these observations have been debated (25,31–33), this study suggests that in the US they may be related to socioeconomic, educational, or religious factors among surrogates that are stronger predictors of LST selection than race/ethnicity itself.
The fact that this dataset includes not only traditional demographics but also information about top decisional concerns distinguishing groups of surrogates from one another, provides several additional insights, particularly in the common situation where a brain injured patient could likely survive an ICU stay but be left with severe functional disability. By far, the strongest predictor of LST selection in this scenario is a surrogate worrying that the neurologic prognosis given by the medical team is inaccurate, a finding that complements prior qualitative and mixed-methods studies (9,34–36). However, the finding that the group of surrogates most concerned with the cost of continued care was paradoxically predisposed towards selecting LST highlights the complexity of surrogate goals-of-care decisions in brain injury. Of note, the “cost of care” decisional group in the original SD survey study exhibited difficulty with prioritizing distinct concerns during the original Best-Worst Scaling exercise and tended to report all concerns of relatively similar priority (12). This observation allows for speculation that the relative prioritization of the cost of care amongst members of this group, compared to the other decisional groups, may in part be a byproduct of general uncertainty regarding decisional priorities, and that their relative inclination to select LST is a manifestation of such.
This study has limitations (12). With a 44.6% completion rate among invited survey respondents, the original study sample may be subjected to non-response sampling bias. The risk of this bias was mitigated by the use of a well-established probability sample of the US population that was further matched to a US geodemographic sample by sample weighting.
Another limitation of the study is the use of surveys presenting hypothetical scenarios. As described in the original survey publication, the surveys underwent extensive content revisions and cognitive testing among ICU surrogates, ICU decision making experts, and lay members of the New Haven community to ensure both their readability and experiential realism. Surveys introducing hypothetical scenarios have been extensively used to elicit important insights into ICU surrogate decision making (6,37–39). However, the surveys do rely on the reader’s understanding of described situations without the ability to ask clarification questions. The patient-physician interaction plays a key role in goals-of-care conversations. It is possible the lack of conversation between the participant and a provider in a face-to-face discussion with concordant review of pertinent images such as MRI or CT scans contributed to the uncertainty reported by some participants. Finally, the hypothetical scenarios used in the survey refers to a healthy older adult with no comorbidities and may limit the generalizability of results in situations where patients are younger and/or medical comorbidities severe enough to influence prognosis are present.
CONCLUSION
In the United States, several demographic and decisional characteristics of surrogate decision makers predict their selection of life-sustaining therapy (LST) for patients with severe brain injury with varying degrees of poor prognosis, including concern regarding prognostic accuracy, low educational level, religious affiliation, and income. Surrogates concerned about the cost of medical care may actually be inclined to select LST, albeit with prominent levels of decisional uncertainty, for patients projected to have severe disability. Recognition of surrogates’ expressed or silent concerns about impending financial toxicity (40,41) and their complicated relationship with goals-of-care decision making preferences may be an important first step for clinicians and researchers in steering many of these difficult shared decisions towards patient-centeredness.
Future directions based on the results of this project may include developing educational programs for clinicians centered on assessing which latent class analysis group a decision maker most closely aligns with and conducting goals of care discussions with the inclinations of that group in mind. Such programs could help prepare clinicians to navigate situations where surrogates who initially seem to be requesting LST nevertheless express strong concern about the cost of long-term care, as these surrogates may be at increased risk for future financial toxicity and/or decisional regret. Extending time allotted for family meetings and allowing such surrogates more time for decision making may help them work through initial their uncertainty, which they may or may not choose to openly express.
Supplementary Material
ACKNOWLEDGEMENTS
None.
Funding and Disclosure of Potential Conflict of Interests (COI)
Dr. Hwang reports other funding from American Brain Foundation, grants from the Neurocritical Care Society, and grants from the Apple Pickers Foundation during the conduct of the study. Dr. Fraenkel reports grants from the Rheumatology Research Foundation and grants from the National Institute of Arthritis and Musculoskeletal and Skin Disease during the conduct of the study (AR060231-06). Dr. White reports grants from the National Heart, Lung, and Blood Institute during the conduct of the study (K24 mentoring award HL148314). Dr. Garg has nothing to disclose. Ms. Soto has nothing to disclose. Dr. Knies has nothing to disclose. Dr. Stanislav has nothing to disclose. Ms. Schalk has nothing to disclose. Dr. Hammer has nothing to disclose. Dr. Holloway has nothing to disclose. Dr. Sheth has nothing to disclose
Footnotes
Publisher's Disclaimer: This Author Accepted Manuscript is a PDF file of an unedited peer-reviewed manuscript that has been accepted for publication but has not been copyedited or corrected. The official version of record that is published in the journal is kept up to date and so may therefore differ from this version.
MANUSCRIPT DETAILS
This manuscript complies with all author instructions. All authorship requirements have been met and the final manuscript was approved by all authors. This manuscript has not been published elsewhere and is not under consideration by another journal. All reporting was conducted as described in the STROBE reporting checklist.
This study was approved by the Yale Human Investigation Committee (protocols #1406014207 and #1505015893). Because this is a minimal risk study, participants indicated consent by completing the survey after being presented with an introductory page that included all of the essential components of informed consent. Informed consent was obtained from all individual participants included in the study. All research procedures were performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.
REFERENCES
- 1.Creutzfeldt CJ, Engelberg RA, Healey L, et al. Palliative Care Needs in the Neuro-ICU. Crit Care Med [Internet] 2015;43(8): 1677–84. Available from: https://www.ncbi.nlm.nih.gov/pubmed/25867905 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Schaefer. Neuro-Palliative Care. Pflege [Internet] 2012;25(2):144–6. Available from: 10.1007/978-3-319-93215-6 [DOI] [Google Scholar]
- 3.Cai X, Robinson J, Muehlschlegel S, et al. Patient Preferences and Surrogate Decision Making in Neuroscience Intensive Care Units. Neurocrit Care [Internet] 2015;23(1):131—41. Available from: https://www.ncbi.nlm.nih.gov/pubmed/25990137 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kon AA, Davidson JE, Morrison W, et al. Shared Decision Making in ICUs: An American College of Critical Care Medicine and American Thoracic Society Policy Statement. Crit Care Med [Internet] 2016;44(1):188–201. Available from: https://www.ncbi.nlm.nih.gov/pubmed/26509317 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Khan MW, Muehlschlegel S. Shared Decision Making in Neurocritical Care. Neurosurg Clin N Am [Internet] 2018;29(2):315–21. Available from: https://linkinghub.elsevier.com/retrieve/pii/S1042368017301535 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Cook DJ. Determinants in Canadian Health Care Workers of the Decision to Withdraw Life Support From the Critically Ill. JAMA J Am Med Assoc [Internet] 1995;273(9):703. Available from: https://www.ncbi.nlm.nih.gov/pubmed/7853627 [PubMed] [Google Scholar]
- 7.Daly BJ, Douglas SL, O’Toole E, et al. Complexity Analysis of Decision-Making in the Critically Ill. J Intensive Care Med [Internet] 2018;33(10):557–66. Available from: https://www.ncbi.nlm.nih.gov/pubmed/27872409 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Sharman M, Meert KL, Sarnaik AP. What influences parents??? decisions to limit or withdraw life support?*. Pediatr Crit Care Med [Internet] 2005;6(5):513–8. Available from: https://www.ncbi.nlm.nih.gov/pubmed/16148808 [DOI] [PubMed] [Google Scholar]
- 9.Quinn T, Moskowitz J, Khan MW, et al. What Families Need and Physicians Deliver: Contrasting Communication Preferences Between Surrogate Decision-Makers and Physicians During Outcome Prognostication in Critically Ill TBI Patients. Neurocrit Care [Internet] 2017;27(2): 154–62. Available from: 10.1007/s12028-017-0427-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Turgeon AF, Dorrance K, Archambault P, et al. Factors influencing decisions by critical care physicians to withdraw life-sustaining treatments in critically ill adult patients with severe traumatic brain injury. Can Med Assoc J [Internet] 2019;191(24):E652–63. Available from: https://www.ncbi.nlm.nih.gov/pubmed/31209132 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Wendler D, Rid A. Systematic Review: The Effect on Surrogates of Making Treatment Decisions for Others. Ann Intern Med [Internet] 2011;154(5):336. Available from: https://www.ncbi.nlm.nih.gov/pubmed/21357911 [DOI] [PubMed] [Google Scholar]
- 12.Hwang DY, Knies AK, Mampre D, et al. Concerns of surrogate decision makers for acute brain injury patients: a US population survey. Neurology 2020; [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.IPSOS. 10 Differentiators of the IPSOS KnowledgePanel® [Internet]. 2020. [cited 2020 Jul 1];1—2. Available from: https://www.ipsos.com/sites/default/files/20-04-55_10diff_v4.pdf
- 14.Flynn TN, Marley AAJ. Best-worst scaling: Theory and methods [Internet]. In: Handbook of Choice Modelling. Edward Elgar Publishing; 2014. p. 178–201. Available from: https://econpapers.repec.org/RePEc:elg:eechap:14820_8 [Google Scholar]
- 15.Flynn TN, Louviere JJ, Peters TJ, Coast J. Best-worst scaling: What it can do for health care research and how to do it. J Health Econ [Internet] 2007;26(1):171–89. Available from: 10.1016/j.jhealeco.2006.04.002 [DOI] [PubMed] [Google Scholar]
- 16.McCutcheon A Latent Class Analysis [Internet]. First Edit. Thousand Oaks, California: 1987. Available from: https://methods.sagepub.com/book/latent-class-analysis [Google Scholar]
- 17.Bardach N, Zhao S, Pantilat S, Johnston SC. Adjustment for do-not-resuscitate orders reverses the apparent in-hospital mortality advantage for minorities. Am J Med [Internet] 2005;118(4):400–8. Available from: https://linkinghub.elsevier.com/retrieve/pii/S0002934305000094 [DOI] [PubMed] [Google Scholar]
- 18.Garrett JM, Harris RP, Norburn JK, Patrick DL, Danis M. Life-sustaining treatments during terminal illness. J Gen Intern Med [Internet] 1993;8(7):361—8. Available from: 10.1007/BF02600073 [DOI] [PubMed] [Google Scholar]
- 19.Stream S, Nolan A, Kwon S, Constable C. Factors associated with combined do-not-resuscitate and do-not-intubate orders: A retrospective chart review at an urban tertiary care center. Resuscitation [Internet] 2018;130:1–5. Available from: https://linkinghub.elsevier.com/retrieve/pii/S0300957218302922 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Qureshi AI, Adil MM, Suri MFK. Rate of Use and Determinants of Withdrawal of Care Among Patients with Subarachnoid Hemorrhage in the United States. World Neurosurg [Internet] 2014;82(5):e579–84. Available from: https://linkinghub.elsevier.com/retrieve/pii/S1878875014006664 [DOI] [PubMed] [Google Scholar]
- 21.Blackhall LJ, Frank G, Murphy ST, Michel V, Palmer JM, Azen SP. Ethnicity and attitudes towards life sustaining technology. Soc Sci Med [Internet] 1999;48(12):1779–89. Available from: https://linkinghub.elsevier.com/retrieve/pii/S0277953699000775 [DOI] [PubMed] [Google Scholar]
- 22.Cardenas-Turanzas M, Gaeta S, Ashoori A, Price KJ, Nates JL. Demographic and Clinical Determinants of Having Do Not Resuscitate Orders in the Intensive Care Unit of a Comprehensive Cancer Center. J Palliat Med [Internet] 2011; 14(1):45—50. Available from: 10.1089/jpm.2010.0165 [DOI] [PubMed] [Google Scholar]
- 23.Chen Y, Criss SD, Watson TR, et al. Cost and Utilization of Lung Cancer End-of-Life Care Among Racial-Ethnic Minority Groups in the United States. Oncologist [Internet] 2020;25(1). Available from: 10.1634/theoncologist.2019-0303 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Hopp FP, Duffy SA. Racial Variations in End-of-Life Care. J Am Geriatr Soc [Internet] 2000;48(6):658–63. Available from: 10.1111/j.1532-5415.2000.tb04724.x [DOI] [PubMed] [Google Scholar]
- 25.Johnson RW, Newby LK, Granger CB, et al. Differences in Level of Care at the End of Life According to Race. Am J Crit Care [Internet] 2010;19(4):335—43. Available from: 10.4037/ajcc2010161 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ormseth CH, Falcone GJ, Jasak SD, et al. Minority Patients are Less Likely to Undergo Withdrawal of Care After Spontaneous Intracerebral Hemorrhage. Neurocrit Care [Internet] 2018;29(3):419—25. Available from: 10.1007/sl2028-018-0554-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Cooper Z, Rivara FP, Wang J, MacKenzie EJ, Jurkovich GJ. Racial Disparities in Intensity of Care at the End-of-Life: Are Trauma Patients the Same as the Rest? J Health Care Poor Underserved [Internet] 2012;23(2):857–74. Available from: http://muse.jhu.edu/content/crossref/journals/journal_of_health_care_for_the_poor_and_underserved/v023/23.2.cooper.html [DOI] [PubMed] [Google Scholar]
- 28.Zahuranec DB, Brown DL, Lisabeth LD, et al. Ethnic differences in do-not-resuscitate orders after intracerebral hemorrhage. Crit Care Med [Internet] 2009;37(10):2807—11. Available from: http://journals.lww.com/00003246-200910000-00019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Zurasky JA, Aiyagari V, Zazulia AR, Shackelford A, Diringer MN. Early mortality following spontaneous intracerebral hemorrhage. Neurology [Internet] 2005;64(4):725–7. Available from: 10.1212/01.WNL.0000152045.56837.58 [DOI] [PubMed] [Google Scholar]
- 30.Murthy SB, Moradiya Y, Hanley DF, Ziai WC. Palliative Care Utilization in Nontraumatic Intracerebral Hemorrhage in the United States*. Crit Care Med [Internet] 2016;44(3):575–82. Available from: http://journals.lww.com/00003246-201603000-00015 [DOI] [PubMed] [Google Scholar]
- 31.Brown CE, Engelberg RA, Sharma R, et al. Race/Ethnicity, Socioeconomic Status, and Healthcare Intensity at the End of Life. J Palliat Med [Internet] 2018;21(9):1308–16. Available from: 10.1089/jpm.2018.0011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Rhodes RL, Batchelor K, Lee SC, Halm EA. Barriers to End-of-Life Care for African Americans From the Providers’ Perspective. Am J Hosp Palliat Med [Internet] 2015;32(2):137–43. Available from: 10.1177/1049909113507127 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Yancu CN, Farmer DF, Leahman D. Barriers to hospice use and Palliative care services use by African American adults. Am J Hosp Palliat Med 2010; [DOI] [PubMed] [Google Scholar]
- 34.Boyd EA, Lo B, Evans LR, et al. “It’s not just what the doctor tells me:” Factors that influence surrogate decision-makers’ perceptions of prognosis*. Crit Care Med [Internet] 2010:38(5): 1270—5. Available from: http://journals.lww.com/00003246-201005000-00005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Meeker MA, Jezewski MA. Metasynthesis: withdrawing life-sustaining treatments: the experience of family decision-makers. J Clin Nurs [Internet] 2009;18(2):163—73. Available from: https://www.ncbi.nlm.nih.gov/pubmed/19120746 [DOI] [PubMed] [Google Scholar]
- 36.White DB, Carson S, Anderson W, et al. A Multicenter Study of the Causes and Consequences of Optimistic Expectations About Prognosis by Surrogate Decision-Makers in ICUs*. Crit Care Med [Internet] 2019;47(9):1184—93. Available from: http://journals.lww.com/00003246-201909000-00003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Sjokvist P, Berggren L, Cook DJ. Attitudes of Swedish physicians and nurses towards the use of life-sustaining treatment. Acta Anaesthesiol Scand 1999; [DOI] [PubMed] [Google Scholar]
- 38.Zier LS. Surrogate Decision Makers’ Interpretation of Prognostic Information. Ann Intern Med [Internet] 2012; 156(5):360. Available from: 10.7326/0003-4819-156-5-201203060-00008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Chapman AR, Litton E, Chamberlain J, Ho KM. The effect of prognostic data presentation format on perceived risk among surrogate decision makers of critically ill patients: A randomized comparative trial. J Crit Care 2015; [DOI] [PubMed] [Google Scholar]
- 40.Khandelwal N, Hough CL, Downey L, et al. Prevalence, Risk Factors, and Outcomes of Financial Stress in Survivors of Critical Illness. Crit Care Med [Internet] 2018;46(6):e530–9. Available from: https://pubmed.ncbi.nlm.nih.gov/29505422 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Hauschildt KE, Seigworth C, Kamphuis LA, et al. Financial Toxicity After Acute Respiratory Distress Syndrome: A National Qualitative Cohort Study*. Crit Care Med [Internet] 2020:48(8):1103—10. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.