Printer Friendly

The accuracy of surrogate decisions in intensive care scenarios.


Critically ill patients are often unable to make decisions about life-sustaining treatments and surrogate decision-makers are relied upon. However, it is unclear how accurately the surrogates' decisions reflect patients' intentions and expectations.

We interviewed 36 pairs of patients and their appointed surrogate decision-makers about their decisions regarding nine treatments in each of three scenarios. The scenarios were persistent vegetative state, coma with likely neurological damage and chronic disease with dementia. The patients were interviewed 24 hours after they had undergone elective surgery under general anaesthesia. The surrogates were interviewed separately by the same interviewer.

There was poor agreement between decisions made by the patients and their surrogates. The patients' and surrogates' summary scores (median (interquartile range) [range]) for treatments were 0 (0-4) [0-9] vs 8 (0-9) [0-9] for the vegetative state scenario, 3 (0-9) [0-9] vs 9 (0-9) [0-9] for the coma scenario and 3 (0-9) [0-9] vs 9 (4-9) [0-9] for the chronic disease scenario. The significantly higher surrogate scores suggest that the surrogates' decisions would have resulted in the patients having far more treatment than the patients would have wanted.

In our participants, there was poor agreement between the decisions made by surrogates and patients. Further study is needed on measures such as facilitated discussions, advance directives and the difficulties that surrogates face, in order to improve the accuracy of surrogates' decisions and respect of patients' autonomy.

Key Words: surrogate decisions, advance directives


Critically ill patients are often unable to make decisions about their treatment. In Singapore, family members are often approached to act as surrogate decision-makers (referred to as surrogates). The surrogates should base their decisions on what the patients themselves would have decided had they been able to do so. This can be difficult to achieve, especially if there has been no prior discussion between the surrogates and patients and the needs of the patients are not known. Surrogates may also have difficulty distinguishing what they want for the patients and what they think the patients themselves would want and may project their own needs on the patients (1).

Previous work on surrogate decisions focused on patients suffering from chronic illnesses. This study aimed to compare the decisions on life-sustaining treatments between patients and their appointed surrogates in relatively healthy elective surgical patients.


We obtained approval from the Institutional Review Board and written consent from all patients and their appointed surrogates for this study. We recruited patients who had general anaesthesia for elective surgery. These patients would have experienced a period of loss of control and unconsciousness similar to that in the intensive care unit. Both the surrogates and patients would also have experienced a period of uncertainty just prior to the interview.

The interviews were conducted at least 24 hours after surgery, to allow the patients to completely recover from general anaesthesia. We interviewed the patients and surrogates separately and advised them not to discuss their decisions with each other until both participants in every pair had been interviewed. We interviewed most patients and their designated surrogates sequentially on the same day. To avoid worsening any preoperative anxiety, we did not do preoperative interviews.

We standardised the interviews and obtained the patients' and surrogates' decisions on nine treatment options in each of three scenarios. The scenarios were: persistent vegetative state with no hope of waking up, coma with likely severe neurological damage and chronic disease complicated by dementia. The treatment options were: care in an intensive care unit, tracheal intubation and ventilation, cardiac support with inotropic drugs, cardiopulmonary resuscitation, enteral feeding, parenteral feeding, haemodialysis, surgery and transfusion of blood products. All the treatments were explained in detail with the aid of photographs. Their decisions were recorded as 'yes' or 'no' for all treatments.

In a subset of surrogates in the later stages of this study, we enquired about their decisions on the treatments for themselves if they were the patients in the imaginary scenarios. This was to evaluate the effect of projection of the surrogates' own preferences on what they thought the patients would want.

We also asked the participants if they knew of and understood advance directives or living wills and power of attorney, and whether the patients had previously discussed these instruments with their surrogates. We also obtained feedback on whether this study had caused the participants anxiety, whether they had found it useful in provoking thinking about critical illness scenarios and whether they thought there was a need for discussion about end-of-life medical decisions.


We used SPSS version 13.0 for all analyses. We used the McNemar test for paired data to assess agreement between the patient and surrogate for each treatment option. We compiled a summary score for each patient and for each surrogate in each of the three scenarios. A score of one was given for each treatment that a participant decided 'yes' and zero when the participant decided 'no'. The minimum summary score for a scenario was 0 and the maximum summary score was 9. We compared the summary scores of the patients and surrogates for each scenario with the Wilcoxon signed rank test. In the subset of participants where the surrogates had also made decisions for themselves in the scenarios, we compared the summary scores using the Friedman two-way ANOVA tests for several related samples.

Our primary outcome measure was the summary score for the persistent vegetative state scenario. We estimated that the patients would have a median score of 3, regarded a difference of 2 as clinically important and estimated the standard deviation to be 2. Thirty-four patients and 34 surrogates would be needed for 80% power with an alpha value of 0.05 in unpaired analysis. We estimated that 36 pairs would provide at least 80% power in paired analysis.


We interviewed a total of 36 pairs of patients and surrogates. The participants' characteristics and the relationship between the patients and surrogates are detailed in Table 1.

The participants' decisions in the vegetative state scenario are presented in Table 2. The patients and surrogates agreed in 176 of 324 (54.3%) decisions. The participants' decisions for the coma scenario are in Table 3. The patients and surrogates agreed in 125 of 324 (38.5%) decisions. The participants' decisions for the chronic disease with dementia scenario are in Table 4. The patients and surrogates agreed in 192 of 324 (59.2%) decisions.

The summary scores for each of the three scenarios are in Table 5. The summary scores were significantly higher for the surrogates' decisions in all three scenarios, indicating that the surrogates decided on far more treatment than the patients would have wanted. The summary scores for the 12 surrogates who were asked about treatment decisions for the surrogates themselves are in Table 6. None of the Friedman tests were statistically significant and we did not carry out post hoc tests. In the persistent vegetative state and chronic disease with dementia scenarios, the surrogates tended to decide on less treatment for themselves compared to what they decided for the patients.

Eighteen (50%) of the patients and 16 (44.4%) of the surrogates had some knowledge of advance directives and only five pairs (13.9%) had previously discussed advance directives. Eighteen (50%) patients and 19 (52.8%) surrogates had some knowledge about power of attorney and only three pairs (8.3%) had previously discussed this instrument. Twelve (33%) of the patients and 10 (27.8%) of the surrogates felt some anxiety after the interviews. Twenty-seven (75%) of the patients and 28 (77.8%) of the surrogates felt that participating in this study had been useful in provoking thinking about their decisions in critical illness scenarios. Twenty-three (63.9%) of the patients and 27 (75%) of the surrogates thought that more discussion about end-of-life medical decisions was needed between individuals and their surrogate decision-makers. Only nine (25%) of the patients compared to 23 (63.9%) of the surrogates reported that knowing the costs of treatment would alter their decisions.


Our study showed poor agreement between patients and their surrogates in decision-making, with the proportion of agreement little better than that due to chance alone. The surrogates' decisions would have resulted in the patients having far more treatment than the patients would have wanted in all three scenarios. Second, in the subset of surrogates who were asked about treatment decisions for themselves, the surrogates wanted less treatment for themselves compared to what they chose for the patients. Third, our participants had a low level of knowledge of advance directives and power of attorney. Even if they were aware of these instruments, few had discussed these with their surrogates and family members.

One limitation of this study is that despite detailed explanations, the participants had variable understanding of the treatments and this may have influenced their decisions. However, we felt that this reflected realistic clinical practice and most of the treatments discussed would require expedient decisions. Second, our study included only relatively healthy participants, none of whom had serious or terminal illnesses and who were relatively young compared to the general ICU population. This was because we felt that discussions of these treatments were best done when the subjects were healthy and there was no pressure to make decisions. Third, we had not assessed if patients and their surrogates had previously discussed life-sustaining treatments. This may be more likely if there had been ongoing serious illnesses or previous life-threatening illnesses. We are unable to evaluate if previous discussions and experiences of illnesses influence the accuracy of surrogate decisions. Fourth, it is not realistic for the participants to make all the decisions at one interview as decisions on life-sustaining treatments occur over a period of time in clinical practice (2).

Our findings agree with earlier studies in other populations that surrogates are unable to predict patients' treatment wishes (1,3). Such differences may be attributed to lack of previous discussion and knowledge of the patients' wishes (3,4,5). A recent large meta-analysis found that not only was the accuracy of surrogate decisions poor, prior discussion did not improve accuracy (6). The majority of studies in this meta-analysis found no trend in surrogates under-treating or over-treating the patients. In our study, there was a definite tendency for surrogates to choose to treat in the belief that the patients would have wanted treatment. Despite patients and their families being at least partly responsible for the costs of health care in Singapore and surrogates being more likely to be influenced by treatment costs in their decisions, the surrogates still chose for the patients to have most treatments.

There did not appear to be projection of the surrogates' decisions for themselves on their decisions for the patients. The surrogates decided on less treatment for themselves if they were ill in similar scenarios compared to what they decided for the patients. Although surrogates were advised to make decisions based on what they thought the patients would want, it is difficult to exclude the influence of societal norms of family and filial responsibilities and of surrogates wanting to give the patients every possible chance of recovery. Even when patients have expressed their wishes in directives and discussions, some surrogates may still try to project their values when making decisions for the patients (7).

Preventing conflict between patients, surrogates and caregivers may be as important as deciding on the appropriate level of care and specific treatments. In a multi-centre study, social workers actively screened for patients at high risk of conflicts in their care (8). They then tailored interventions such as family conferences, ethics consults, social service consults and pastoral care to reduce potential conflict. Although the interventions did not improve the already high satisfaction with ICU care, they did clarify patients' and surrogates' preferences. This increased the likelihood of patients and surrogates deciding to forgo resuscitation, choosing palliative care and also choosing aggressive care. Appointing surrogate decision-makers and delegating power of attorney may also reduce other conflicts, such as when multiple relatives influence decisions.

That surrogates decided on more treatment than the patients would have wanted may compound any breach of patient autonomy. Despite our efforts to clarify the questions, the surrogates may still have been unsure if they were making decisions according to what they thought the patients would want, rather than what the surrogates wanted for the patients. These dilemmas are difficult to avoid in clinical practice and measures to reduce disagreements are needed.

Education of the public and facilitated discussions about advance directives may help to improve surrogates' decisions. In Singapore, the Ministry of Health has recently re-embarked on educating both the public and the medical community about advance medical directives. These advance medical directives are limited to situations where terminally ill patients are unconscious or incapable of expressing rational judgement and where death would be imminent regardless of the application of extraordinary life-sustaining treatments (9). In situations similar to this study's imaginary scenarios, advance medical directives may not apply as death is not imminent and surrogates decisions will still be needed. Hence discussion between patients and their family members should be encouraged before the need for such care arises. Previous work showed that advance medical directives had limited benefit or did not improve the accuracy of decisions by surrogates (10,11). One study found that advanced directives could improve the accuracy of decisions by emergency and critical care doctors who were less familiar with the elderly patients, whereas there was no improvement in accuracy for family member surrogates and primary care doctors (12).

In conclusion, there was poor agreement between surrogates' and patients' decisions in this study. While discussion of life-sustaining treatments can cause anxiety, most participants felt that there was a need for such discussion. We should explore if a combination of facilitated discussions and advance directives can improve the patient-surrogate relationship and understanding of the patients' wishes. Work is needed on understanding the difficulties that surrogates face, to improve the accuracy of surrogate decisions and better preserve and honour patients' autonomy.

Accepted for publication on August 25, 2006.


(1.) Fagerhn A, Ditto PH, Danks JH, Houts Rm, Smucker WD. Projection in surrogate decisions about life-sustaining medical treatments. Health Psychol 2001; 20:166-175.

(2.) Faber-Langendoen K, Bartels DM. Process of forgoing life-sustaining treatment in a university hospital: an empirical study. Crit Care Med 1992; 20:570-577.

(3.) Seckler AB, Meier DE, Mulvihill M, Paris BE. Substituted judgment: how accurate are proxy predictions? Ann Intern Med 1991; 115:92-98.

(4.) Hare J, Pratt C, Nelson C. Agreement between patients and their self-selected surrogates on difficult medical decisions. Arch Intern Med 1992; 152:1049-1054.

(5.) Layde PM, Beam CA, Broste SK, Connors AF Jr, Desbiens N, Lynn J et al. Surrogates' predictions of seriously ill patients' resuscitation preferences. Arch Fam Med 1995; 4:518-524.

(6.) Shalowitz DI, Garrett-Mayer E, Wendler D. The accuracy of surrogate decision makers: a systematic review. Arch Intern Med 2006; 166:493-497.

(7.) Bramstedt KA. Questioning the decision-making capacity of surrogates. Intern Med J 2003; 33:257-259.

(8.) Burns JP, Mello MM, Studdert DM, Puopolo AL, Truog RD, Brennan TA. Results of a clinical trial on care improvement for the critically ill. Crit Care Med 2003; 31:2107-2017.

(9.) Ministry of Health Singapore. Advanced medical directive. 2006.

(10.) Ditto PH, Danks JH, Smucker WD, Bookwala J, Coppola KM, Dresser R et al. Advance directives as acts of communication: a randomized controlled trial. Arch Intern Med 2001; 161:421-430.

(11.) None listed. A controlled trial to improve care for seriously ill hospitalised patients. The study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT). The SUPPORT Principal I. JAMA 1995; 274:1591-1598.

(12.) Coppola KM, Ditto PH, Danks JH, Smucker WD. Accuracy of primary care and hospital-based physicians' predictions of elderly outpatients' treatment preferences with and without advance directives. Arch Intern Med 2001; 161:431-440.

L. L. M. LI *, K. Y. P. CHEONG ([dagger]), L. K. YAW ([double dagger]), E. H. C. LIU ([section]) Department of Anaesthesia, Yong Loo Lin School of Medicine, National University of Singapore, Singapore

* B.Sc. Med.Sci., M.B., Ch.B., Foundation Doctor Year 1, Department of Critical Care Medicine, Dr Gray's Hospital, Elgin, United Kingdom.

([double]) M.R.C.G.P., Family Physician, Department of Medicine, National University Hospital.

([double dagger]) M.B., B.S., Registrar, Department of Emergency Medicine, Royal Perth Hospital, Perth, Western Australia, Australia.

([section]) M.Phil., F.R.C.A., Assistant Professor.

Address for reprints: Dr E. Liu, Department of Anaesthesia, National University of Singapore, 5 Lower Kent Ridge Road, 119074, Singapore.
Characteristics of the participants

 Patients Surrogates

Gender: male/ female 20/16 13/23
Age (median (range)) 38(18-67) 45(21-73)
Ethnic group (%)
Chinese 28 (77.8%) 28 (77.8%)
Malay 2 (5.6%) 2 (5.6%)
Indian 2 (5.6%) 2 (5.6%)
Sri Lankan 1 (2.8%) 1 (2.8%)
Pakistanis 1 (2.8%) 1 (2.8%)
Other 2 (5.6%) 2 (5.6%)
Spouse -spouse 20 (55.6%)
Parent - Child 2 (5.6%)
Child - Parent 10 (27.8%)
Sibling - sibling 2 (5.6%)
Other 2 (5.6%)

Patient and surrogate decisions in persistent vegetative state

Treatments Patients Surrogates No. of P
 pairs value
 Yes No Yes No

Care in ICU 8 28 22 14 18 0.01

Tracheal intubation
and ventilation 9 27 19 17 20 0.02

Cardiac support
with inotropes 8 28 20 16 20 0.004

resuscitation 9 27 17 19 22 0.057

Enteral feeding 10 26 20 16 18 0.031

Parenteral feeding 11 25 19 17 20 0.077

Dialysis 8 28 21 15 19 0.002

Surgery 10 26 20 16 20 0.021

Blood product
transfusion 11 25 20 16 19 0.049

Number of patient surrogate pairs=36. P values are for McNemar

Patient and surrogate decisions in coma with likely neurological
damage scenario

Treatments Patients Surrogates No. of P
 pairs value
 Yes No Yes No

Care in ICU 15 21 29 7 14 0.004

Tracheal intubation
and ventilation 15 21 29 7 14 0.004

Cardiac support
with inotropes 15 21 28 8 15 0.007

resuscitation 15 21 28 8 17 0.004

Enteral feeding 16 20 31 5 13 0.003

Parenteral feeding 16 20 30 6 12 0.007

Dialysis 15 21 30 6 13 0.003

Surgery 13 23 30 6 13 <0.001

Blood product
transfusion 17 19 31 5 14 0.004

Number of patient surrogate pairs=36. P values are for McNemar

Patient and surrogate decisions in chronic disease with
dementia scenario

Treatments Patients Surrogates No. of P
 pairs value
 Yes No Yes No

Care in ICU 19 17 27 9 22 0.057

Tracheal intubation
and ventilation 17 19 27 9 22 0.013

Cardiac support
with inotropes 17 19 27 9 22 0.013

resuscitation 17 19 26 10 21 0.035

Enteral feeding 18 18 28 8 20 0.021

Parenteral feeding 18 18 28 8 20 0.021

Dialysis 16 20 27 9 19 0.013

Surgery 17 19 26 10 23 0.022

Blood product
transfusion 18 18 27 9 23 0.022

Number of patient surrogate pairs=36. P values are for McNemar

Summary scores for patients and surrogates for three scenarios

 Patient Surrogate P value

Persistent 0 (0-4) [0-9] 8 (0-9) [0-9] 0.012
vegetative state

Coma with 3 (0-9) [0-9] 9 (0-9) [0-9] 0.005

Chronic disease 3 (0-9) [0-9] 9 (4-9) [0-9] 0.012
with dementia

Scores are median (interquartile range) [range]. P values are for
Wilcoxon signed rank tests.

Summary scores for patients' decisions, surrogates' decisions for
themsehw and surrogates' decisions for patients, in three

 Patients' Surrogate Surrogate P value
 decision decision decision
 for self for patient

Persistent 0(0-6) 3(0-9) 8(1-9) 0.054
vegetative state [0 - 9] [0-9] [0-9]

Coma with 2(0-9) 8(0-9) 9(6-9) 0.241
neurological [0-9] [0-9] [0-9]

Chronic disease 2(0-9) 0(0-9) 9(2-9) 0.135
with dementia [0-9] [0-9] [0-9]

Number of data sets=12. Scores are median (interquartile range)
[range]. P values are for Friedman tests for several related
COPYRIGHT 2007 Australian Society of Anaesthetists
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2007 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Li, L.L.M.; Cheong, K.Y.P.; Yaw, L.K.; Liu, E.H.C.
Publication:Anaesthesia and Intensive Care
Article Type:Clinical report
Geographic Code:9SING
Date:Feb 1, 2007
Previous Article:Trends in intra-aortic balloon counterpulsation: comparison of a 669 record Australian dataset with the multinational Benchmark Counterpulsation...
Next Article:Audit of use of size 3 Yeescope in a paediatric anaesthesia practice.

Related Articles
Sensitive operators.
Decision support can enhance patient care: use of research-based guidelines will complement addiction professionals' decision-making.
Evaluating stat testing options in an academic health center: therapeutic turnaround time and staff satisfaction.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters