Effects of Offering Advance Directives on Medical Treatments and Costs Lawrence J. Schneiderman, MD; Richard Kronick, PhD; Robert M. Kaplan, PhD; John P. Anderson, PhD; and Robert D. Langer, MD, MPH

• Objective: To examine the effects of advance directives on medical treatments and on patient satisfaction and well-being and to determine whether the enhancement of patient autonomy through advance directives provides a more ethically feasible approach to cost control than does the imposition of limits through rationing. • Design: Randomized, controlled trial. • Setting: University and Veterans Affairs medical center. • Patients: Two hundred and four patients with lifethreatening illnesses, 100 of whom died after enrollment in the study. • intervention: Patients randomly assigned to the experimental group were offered the California Durable Power of Attorney (a typical proxy-instruction directive), and patients assigned to the control group were not offered the advance directive. Hospital admissions were monitored to assure that a summary of the document was present in the active medical record at each hospitalization. • Measurements: Cognitive function, patient satisfaction, pyschological well-being, health locus of control, sense of coherence, health-related quality of life, receipt of medical treatments, and medical treatment charges. • Results: No significant differences were found between advance-directive and control groups regarding psychosocial variables, health outcome variables, and medical treatments or charges. Patients offered an advance directive had an average hospital stay of 40.8 days (95% CI, 32.2 to 49.4 days), compared with an average of 33.1 days (95% CI, 26.0 to 40.2 days) for controls. Patients offered an advance directive were charged an average of $19 502 (95% CI, $13 030 to $25 974) for medical treatments in the last month of life compared with $19 700 (95% CI, $13 704 to $25 696) for controls. • Conclusions: Despite claims that public demand for longer life accounts for rising medical costs, most surveys suggest that patients are calling for less, not more, of the expensive, high-technology treatment often used in terminal phases of illness. Executing the California Durable Power of Attorney for Health Care and having a summary copy placed in the patient's medical record had no significant positive or negative effect on a patient's well-being, health status, medical treatments, or medical treatment charges.

Annals of Internal Medicine. 1992;117:599-606. From the University of California, San Diego, California. For current author addresses, see end of text.

1 wo social movements over the past decade of health care in the United States have attracted the particular attention of medical ethicists and health policy planners. The first movement, in response to soaring costs, is the urgent call for rationing and resource allocation that is more specific and consistent than the haphazard socioeconomic forms now in place. The second movement is the rising support for patient autonomy, including mechanisms to maintain control of terminal treatment even after the loss of decision-making capacity. Nearly every state has legalized some form of advance directive (living will, durable power of attorney for health care) (1) and the U.S. Congress has passed the Patient Self-Determination Act (2). The Act requires that all hospitals and other institutions certified by Medicare or Medicaid inform their patients about the availability of advance directives and the patient's legal right to accept or refuse medical treatment. Surprisingly, the potentially synergistic connection between patient autonomy and cost containment has not been recognized. Callahan (3), in fact, sees these two entities on a collision course, citing the "powerful, unremitting public demand for better health and a longer life" as a major obstacle to limiting health care costs. However, although the constant rise in health care costs is a demonstrable fact, placing the blame on consumer demand is an unproven assumption. Surveys (4) and studies by us and others (5-9) (with the exception of one study of elderly patients with intensive-careunit experience who expressed a willingness to undergo intensive-care-unit treatment again [10]) suggest that patients are calling for less, not more, of the costly hightechnology treatment often used in terminal phases of illness. Notwithstanding the theoretically sound arguments supporting patient self-determination in medical decision-making (1, 11) and the widespread professional and public endorsement of advance directives (9, 11-15), nothing, to our knowledge, has been reported regarding the effect of these documents on medical practice and outcomes. Nevertheless, under the Patient Self-Determination Act, regulations specifying the kinds of processes that hospitals can establish to be in compliance with the new law have been imposed without any pretesting of whether they are likely to result in beneficial improvements in patient care and without a well-specified set of hypotheses linking the policy change to the desired behavior change. The purpose of our study was to evaluate the effects of advance directives on patient outcomes and health care costs. We report a prospective study involving 204 patients with life-threatening illnesses who were randomly assigned to an experimental group (offered the © 1992 American College of Physicians

Downloaded from by Tulane University user on 01/19/2019


California D u r a b l e P o w e r of A t t o r n e y for H e a l t h Care) (16) or a control g r o u p (not offered the a d v a n c e directive). Methods Participants and Setting The study was done at the University of California, San Diego (UCSD), Medical Center and the Veteran Affairs (VA) Medical Center. Institutional review boards at both institutions granted approval. All physicians in the participating specialty clinics (Table 1) were informed of the study and agreed to participate. Study subjects included 204 patients identified by their physicians as having life-threatening disease, defined by a 5-year life expectancy of no better than 50% as judged by the physician. Patients were contacted during sequential outpatient clinic visits or by mail from 6 October 1987 to 24 February 1989 (before passage of the Patient Self-Determination Act). Forty-five physicians (24 faculty, 21 fellows) were asked to identify all the patients under their care who would qualify for this study. All qualifying patients seen in the clinics during this period (n = 337) were contacted (see Table 1). Patients were then randomly assigned either to the advance-directive group and offered the opportunity to execute the California Durable Power of Attorney for Health Care or to the control group and not offered the advance directive. The study design incorporated a modified Zelen procedure in which randomization occurred before patients were offered the opportunity to participate in the study (17). The traditional design of randomly assigning subjects after obtaining consent was not desirable because making those in the control group aware of advance directives would have been an intervention in and of itself. If informed consent had been obtained before randomization, subjects in the control group might subsequently have behaved differently than other persons in the community who had not been informed of advance directives. In compensation for their participation, patients were offered $10 on the completion of each interview. Intervention Patients in the advance-directive group were given the California Durable Power of Attorney for Health Care in the format provided by the California Medical Association. This instrument, a combined proxy-instruction directive, was one of the first to be enacted (1984) and has served as a prototype for advance directives in other states. Before the study, the principal investigator provided an informational booklet and training sessions to physicians, enabling them to educate their patients and encourage participation. No further recruitment efforts were made after patients' first refusal, and patients were reassured that their treatment would not be compromised. In the AIDS (acquired immunodeficiency syndrome), cardiology, family medicine, oncology, and pulmonary clinics, physicianproviders informed the patients of the study and offered the advance directive in the office or the examining room. For patients treated in the neurology and renal clinics, information about the study and the offer of the document was made by mail. In the case of the patients treated in the renal clinic, this method was chosen to minimize cross-contamination, because most medical contacts occur in the dialysis setting where private conversations are difficult to carry out. In the case of patients treated in the neurology clinic, the infrequency of office visits led to our choice of contacting patients by mail. Because of the nature of the intervention, physicians could not be blinded to treatment conditions. However, they did not know the specific research hypotheses. The document provides space to designate surrogate decision makers, as well as a few lines to write in medical treatment instructions, and, in addition, offers three alternatives related to situations in which patients may or may not desire life-prolonging treatment. To maximize the effectiveness of the advance directive, hospital admissions were monitored so that hospitalized patients who had executed the California Durable Power of Attorney for Health Care had that information prom600

1 October 1992 • Annals of Internal Medicine

inently placed in the active medical record in the form of a "Notice to My Health Care Provider" summarizing the patient's designated proxy and the patient's treatment instructions. Without exception, we found such monitoring necessary because the " N o t i c e " was invariably filed with the inactive medical record and, therefore, not readily accessible to physicians during hospitalization. In addition, an orange sticker was placed on the front cover of the active medical record to alert health care providers that the chart contained a signed summary of the patient's executed California Durable Power of Attorney for Health Care. Survey Instruments After the initial contact, patients were interviewed at 3 months, 6 months, 1 year, and every 6 months thereafter. Several interview instruments were applied at each follow-up point. Cognitive Function At baseline and at each follow-up, all potential participants in the study were given the Mini-Mental Status Examination (18). This instrument is designed to identify cognitive dysfunction in the elderly. It has a high test-retest reliability (r = 0.89) and a high inter-rater reliability (r = 0.82). The Mini-Mental Status Examination has been used in various studies and has been shown to distinguish those known to be cognitively impaired from those with no dysfunction. Only patients with scores of 24 or greater were included in our study. Patient Satisfaction We used the scales described by DiMatteo and Hays to measure patient satisfaction (19). The scales include five items on general satisfaction with physicians; eight items on physician communication; nine items on physician behavior; and three items on physician technical competence. Psychological Weil-Being An 18-item version of the Dupuy general well-being schedule was used to assess the patient's psychological well-being. The measure has been validated by comparing it with clinical measures (20), and discriminate evidence of validity has been shown by comparing it with similar interview-based measures (21). Health Locus of Control A multidimensional health locus of control scale, derived from social learning theory, was used (22). This scale seeks to classify a person's belief that his or her health is or is not determined by his or her behavior. Persons who believe that their health outcomes are determined by their own behavior are classified as "internals." Persons who believe that their health is determined by external factors, such as chance, fate, luck, or powerful others, are classified as "externals." Alpha reliabilities for the multidimensional health locus of control scales range from 0.67 to 0.77. Various validity studies have been conducted (22). Sense of Coherence A 29-item sense of coherence scale developed by Antonovsky (23) was used to measure an aspect of personal control that is more general than that tapped by internalityexternality measures. Health-related Quality of Life The Quality of Well-Being Scale was used as a general health-related quality-of-life outcome. The Quality of WellBeing Scale combines preference-weighted measures of symptoms and functioning to provide a numerical point-in-time expression of well-being; the scale ranges from 0 for death to 1.0 for asymptomatic optimum functioning (24, 25). Validity of the measure has been reported (26), and the sensitivity and the predictive value of the measure have both been estimated (27). Empirical derivation of all preference weights has been reviewed (28, 29).

• Volume 117 • Number 7

Downloaded from by Tulane University user on 01/19/2019

Medical T r e a t m e n t s

Supervision and Follow-up

Variables extracted from the medical record included frequency of cardiopulmonary resuscitation (CPR); the presence or absence of do-not-resuscitate (DNR) orders; number of days receiving renal dialysis, artificial ventilation, and parenteral nutrition and hydration; total days in the intensive care unit; and total days in the hospital.

Continuous participation in the study was encouraged by the repeated contacts of interviewers and supplemented by periodic health-information letters from the principal investigator. For completeness, we obtained permission to review records of treatments provided at hospitals other than the UCSD and the VA medical centers. Patients who died in the hospital were identified through our concurrent monitoring of hospital admissions. Patients who died in other settings were so identified at the time of their semi-annual interview. Within 6 weeks after we became aware of a patient's death, the designated proxy or next of kin or other person familiar with the patient's last days was contacted and interviewed to obtain, in as much detail as possible, information on the patient's satisfaction, quality of well-being, and treatment decisions just before death. Complete data on medical care charges are currently available on most patients who died before January 1991 (minimum followup, 23 months from the date of entry into the study).

Medical T r e a t m e n t C h a r g e s Data on medical treatment charges were obtained from medical records, questionnaires completed by patients and families, and bills generated by the UCSD Medical Center. We obtained copies of all bills for inpatient and outpatient services for study patients served at the UCSD Medical Center; institutional charges for these patients were taken from these bills. The VA Medical Center does not charge patients based on the mix of services received. Medical records from the VA Medical Center were reviewed and each service that the patient received was coded. Charges for these services were imputed using data from the UCSD Medical Center. Thus, imputed charges for patients served at the VA Medical Center are an estimate of the amount the patients would have been charged had they been served at the UCSD Medical Center. These calculations were facilitated by a computer-based system that was developed and used for all record abstractions (30). During periodic interviews, patients and their families were asked about other health care they had received. For hospitalizations at institutions other than the UCSD or the VA medical centers, we reviewed the medical record and imputed charges applying the same technique used for the VA Medical Center. For services rendered by other providers (for example, skilled nursing facilities, hospices, or durable medical equipment suppliers), we contacted the vendor to obtain an estimate of the average charge for the type of service received by the patient. We decided to carry out this interim analysis when we had complete charge data on the 100 patients who had died. The first 100 patients with complete charge data included 50 experimental patients and 50 controls. Data on these patients are used in the analysis of the effects of being offered the California Durable Power of Attorney for Health Care on medical care charges.

Statistical Analysis Baseline data on the treatment and control groups were compared using Mests and chi-square tests; medical care charges for treatment and control groups were compared using Mests and, for confirmation, the Mann-Whitney U test; only Mest results are reported. Evaluation of differences between groups over time was assessed using repeated-measures analysis of variance.

Results Baseline C o m p a r i s o n s R e c r u i t m e n t w a s terminated w h e n agreement to participate in the s t u d y w a s obtained from 204 p a t i e n t s : O n e h u n d r e d and four of the 185 advance-directive group patients and 100 of the 152 control g r o u p patients (see Table 1). Patients c o n t a c t e d for the advance-directive g r o u p w e r e slightly less likely t h a n t h o s e c o n t a c t e d for the control g r o u p to b e e n t e r e d into the s t u d y

Table 1. Baseline Characteristics of the Advance-Directive and Control Groups* Variable

Advance-Directive Group

Patients contacted, n Patients entered into the study, n Patients not entered into the study, n Refused Executed a previous advance directive Died before first interview Failed the Mini-Mental Status Examination Patients from the UCSD Medical Center, n Patients from the VA Medical Center, n Specialty Clinics, «/«§ AIDS Cardiology Family Medicine Neurology Oncology Pulmonary Renal Average age, y Men, n Women, n Median income, $ Patients lost to follow-up at 12 months, n

Control Group

185 104

152 100

49 3 23 6 30 74

32 1 13 6 32 68

18/18 3/5 3/9 3/19 54/92 18/31 5/11 56.2 86 16 10 000-15 000 8

14/14 1/4 2/3 9/19 47/73 22/31 5/8 60.5 82 18 10 000-15 000 13

P Value

0.07 0.16t 0.2 > 0.2

< 0.05t > 0.2

* AIDS = acquired immunodeficiency syndrome; UCSD = University of California, San Diego; VA = Veterans Affairs. t /-statistic. t Chi-square test done for the percent of patients who refused among those who did not die before the first interview. § Patients entered in the study/patients v contacted for entry into the study.

1 October 1992 • Annals of Internal Medicine

Downloaded from by Tulane University user on 01/19/2019

• Volume 117 • Number 7


Table 2. Choices of Patients Who Executed an Advance Directive Advance-Directive Option

Patients Who Executed an Advance Directive (n = 69)*

Patients Offered an Advance Directive Who Died and for Whom Complete Cost Data Were Available (n = 32)t n(%)

Choice 1: I do not want my life prolonged and I do not want life-sustaining treatment to be provided or continued if the burdens of the treatment outweigh the expected benefits. I want my agent to consider the relief of suffering and the quality of life as well as the extent of the possible extension of my life in making decisions concerning life-sustaining treatment. Choice 2: I want my life to be prolonged and I want life-sustaining treatment to be provided unless I am in a coma that my doctors reasonably believe to be irreversible. Once my doctors have reasonably concluded I am in irreversible coma, I do not want life-sustaining treatment to be provided or continued. Choice 3: I want my life to be prolonged to the greatest extent possible without regard to my condition, the chances I have for recovery, or the cost of the procedures.

55 (80)

29 (91)

12 (17)




* Of 104 patients assigned to the advance-directive group, 69 executed the directive and 35 did not. t Of 50 patients offered the advance directive, 32 actually executed the directive.

(P = 0.07). Although a borderline difference in participation rates between the two groups was observed, no statistically significant differences were found in any subcategories of the reasons for nonparticipation. Patients contacted for the advance-directive group were slightly more likely to die before the first interview than were patients contacted for the control group, but the difference was not statistically significant (P > 0.2). The procedure for scheduling interviews was identical in the experimental and control groups, suggesting that the greater mortality among the advanced-directive group was probably a random event. Among patients who did not die before the first interview, a slightly larger percent of those in the advanced-directive group refused to participate relative to the control group, although, again, the difference was not statistically significant at conventional levels. Twelve patients completed an advanced directive but then refused to participate further in the study; it is possible that these "extra" refusals introduced slight noncomparabilities between experimental patients and control patients. However, except for a small difference in average age (P < 0.05), characteristics of the control and experimental patients were similar. Further, baseline measures of quality of well-being were identical in the two groups, suggesting no difference in severity of illness. In each group approximately 50% of the patients were recruited from an oncology clinic; 20%, from a pulmonary clinic; and 15%, from an AIDS clinic. Approximately 70% of participants were patients at the VA Medical Center; more than 80% of patients were male. At 12 months, 8 patients in the.advance-directive group and 13 patients in the control group were lost to followup. The 100 patients who died and on whom we have complete cost data were similar to the entire group of participants regarding age, diagnosis, sex, medical center, and median income.

it and returned the Notice to My Health Care Provider, which was inserted in the medical record (Table 2). None of the patients wrote in personal instructions; all patients, in addition to designating a proxy, responded to the offered instruction choices. Of the 69 patients completing an advance directive, 80% chose the least aggressive option offered—expressing a preference against treatment if "the burdens of the treatment outweigh the expected benefits." Twelve patients chose to undergo aggressive treatment unless they were in an "irreversible coma," and 2 patients preferred aggressive treatment regardless of the circumstances. The choices made by the 50 patients in the advance-directive arm of the study who died and on whom we have complete cost information were similar to the choices made by the entire advance-directive group. No statistically significant differences (at the 95% confidence level) were observed between those who signed an advance directive and those who did not regarding baseline quality of well-being, internal and external locus of control, age, diagnostic category, and length of time before death (data not shown). Length of Survival Length of time from entry into the study until death was nearly identical for patients offered an advance directive and for control patients (Figure 1). In each group, approximately one half of the patients died within 16 months of entering the study, and two thirds of the patients died within 30 months of entering the study. This analysis is based on all 183 patients who were entered into the study and not lost to follow-up; in all cases, we were able to evaluate patient status for at least 34 months. Satisfaction and Psychological Well-Being

Patient Preferences Of the 104 patients offered the California Durable Power of Attorney for Health Care, 69 (66%) executed 602

Psychological and health-status characteristics of patients in the advance-directive group and those in the group who chose to execute the advance directive (not

1 October 1992 • Annals of Internal Medicine • Volume 117 • Number 7

Downloaded from by Tulane University user on 01/19/2019

all patients in the advance-directive group actually executed the advance directive) were compared with those in the control group (all of whom did not execute an advance directive) at entry and over time. No significant intergroup differences were found in general well-being (P > 0.2), patient satisfaction (P > 0.2), internal (P = 0.2) and external (P = 0.2) locus of control and other subscales from multidimensional health locus of control scales, sense of coherence (P > 0.2), or quality of well-being at time of entry to the study (P > 0.2). Further, differences between groups did not emerge at any follow-up point, although the power of these tests was reduced because of the high mortality rate. Quality of well-being scale scores over time were nearly identical in the experimental and control groups (Figure 2). Patients who died were given the score of 0 for this analysis. No evidence was found that executing the California Durable Power of Attorney for Health Care had any beneficial or adverse effects on health status (P > 0.2). Both groups declined significantly in quality of well-being over time (P < 0.0001), but no evidence was found for a differential decline between the treatment groups (P > 0.2). Health Care Treatments No significant differences were found between control patients and patients offered an advance directive regarding the frequency of CPR events; percent of patients with DNR orders; the average duration of stay in the hospital, intensive care unit, or nursing home; the numbers of days receiving artificial ventilation or parenteral nutrition and hydration; the average charges for antibiotics or narcotics; the percent of patients who died at home; or the average length of time between entry into the study and death (Table 3). Among patients offered an advance directive, no significant differences regarding any of these outcome variables were found between patients who executed a directive and those who did not. Not only were no significant differences in treatments detected between patients offered or signing an advance directive and those in the control group, the direction of differences observed was, in many instances, contrary to the expected direction of the effect. Patients offered or signing an advance directive spent more days in the hospital than did patients in the control group; control patients were more likely than experimental patients to have had a DNR order and less likely to have undergone CPR; control patients received the same amount of narcotics (suggestive of comfort care) as did those who were offered an advance directive. However, differences were not statistically significant.

group, but the differences were not significant at the P = 0.05 level. Patients in both groups consumed substantial medical care 'resources while dying. Average charges were approximately $50 000 for the last 6 months of life and more than $80 000 for the period from entry into the study until death. Results of multivariate analyses, in which various patient baseline characteristics were included as independent variables, were similar to results of the bivariate analysis (data not shown). A comparison of average charges among patients who signed the advanced directive with those among patients who were offered the advance directive but did not sign showed no significant differences between groups (data not shown). Discussion Executing the California Durable Power of Attorney for Health Care and having a summary inserted in the medical record appeared to have no beneficial or harmful effect on patient well-being, health status, or satisfaction. Also, despite the fact that all patients in our study who executed the advance directive chose to limit life-sustaining treatments, no discernible reduction in treatments or charges was observed in this group when compared with the control group. Thus, our hope that executing the California Durable Power of Attorney for Health Care might provide a more ethical approach to reducing health care costs by enhancing autonomy was not realized. It is during the last month of life that one would expect an advance directive to exert the most effect by enabling a patient to avoid unwanted days in the intensive care unit or unwanted aggressive and expensive treatments (31, 32). We had hypothesized therefore that the resulting avoided charges would become apparent when those executing an advance directive were compared with patients who had not been able to signify their wishes during the last month of life. However, we

Medical Treatment Charges Average health care charges during the last month of life for advance-directive and control groups were virtually identical—an average of approximately $19 000 per patient in each group (Figure 3). Average charges for the 3 months before death, for the 6 months before death, and for the entire study* period were slightly higher in the advance-directive group than in the control

Figure 1. Survival by number of months from entry into the study. The analysis included 183 patients (8 patients in the advance-directive group and 13 in the control group were lost to follow-up). All 183 patients are accounted for at each fol-

1 October 1992 • Annals of Internal Medicine • Volume 117 • Number 7 Downloaded from by Tulane University user on 01/19/2019


not have resulted in a finding of a significant advancedirective effect in the hypothesized direction.

Physician Response to Instructions

Figure 2. Mean quality of well-being score for advance-directive and control groups at each follow-up for year 1. Bars represent 95% confidence intervals.

found no evidence of cost savings resulting from the execution of an advance directive for any of the several time periods studied. In fact, the direction of the differences we observed indicated greater costs for those patients executing an advance directive, although the differences were not statistically significant. To understand why executing the advance directive did not produce the hypothesized results, we considered the following four possibilities: The sample size was not big enough to allow for meaningful comparisons; physicians did not respond to the instructions in the advance directives when making treatment decisions; physicians caring for the control patients were just as likely as those caring for advance-directive patients to limit heroic treatments; circumstances rarely arise in which the presence of advance directives in medical records is a significant factor in treatment decisions because of patient incapacity. Sample Size If one were to examine charges for only the last month of life, a sample size of approximately 800 control patients and 800 patients offered a durable power of attorney would be necessary to have a 90% chance of detecting an experimental effect of 10%, that is, to detect a difference in charges of $1900 between the control group and the advance-directive group. Given the actual sample size, the 95% CI for the effect of advance directives on medical care charges in the last month of life ranges from a reduction of $8606 to an increase of $9002; these data cannot rule out the hypothesis that offering patients advance directives has a substantial effect on medical treatment charges. However, the lack of any difference between charges for the groups at 1 month before death and the suggestion of higher charges among advance-directive patients at 3 months, 6 months, 1 year, as well as for the entire study period, suggest that a larger sample size would 604

1 October 1992 • Annals of Internal Medicine

Studies by others show that physicians have poor insight into the wishes of their patients regarding lifesustaining treatments (33-36). These studies were done, however, in patients who had not declared their wishes in writing. Unpublished data we have collected indicate that even when patients do express their wishes in writing, physicians' insights into their patient's wishes are still poor, and in fact, appear to be influenced by what physicians would want for themselves. In our study, all the patients expressed a preference for limiting aggressive lifesaving treatments. Theoretically, that should have resulted in reduced numbers of procedures and reduced medical costs. Evidence suggests, however, that proxies are more likely to favor a life-sustaining treatment intervention than the patients themselves (36). Because physicians consulting a patient's advance directive would properly discuss treatment decisions with the patient's designated proxy for assistance in interpreting the patient's brief general instructions, it is possible, therefore, that patients might have had more aggressive treatments imposed on them than they would have chosen themselves. In our chart reviews, however, we found no evidence that patients, physicians, or proxies were ever involved in conflicts about treatment plans in violation of the patients' written preferences.

Withholding or Limiting Treatment At both the VA and UCSD medical centers, patients' preferences regarding DNR orders were routinely elicited by physicians before patients lost their decisionmaking capacity, regardless of whether patients had executed an advance directive. In these hospitals, such discussions are now required as a standard protocol when it is apparent that a patient is declining and unlikely to survive cardiac arrest or another life-threatening event. Among the 40 patients who died in the hospital, more than 90% had documentation in the chart of the discussion of code status at a time when the patient was still capable of participating in treatment decisions. This finding represents a marked improvement in physicians' recognition of patient autonomy over the past decade (33). In only three cases was there no confirmation of the preferences expressed in the advance directive, either in the patient's chart or by the person interviewed after the patient's death. However, in both of these institutions there exist no formal protocols or policy-guiding discussions about the many other potentially lifesaving treatments, such as use of the ventilator, artificial nutrition and hydration, admission to the intensive care unit, use of blood transfusions, antibiotic therapy, or chemotherapy. Thus, a vast area of decision-making remains on which a patient's advance directive might have influence if the patient were incapable of participating in the decision-making process.

• Volume 117 • Number 7

Downloaded from by Tulane University user on 01/19/2019

Usefulness of the Advance Directive At the initiation of our study, we shared the notion that many seriously ill patients would lose their decision-making capacity for a long enough period to make the advance directive useful. However, most of our patients retained their decision-making capacity through their declining state of health to allow anticipatory discussions to take place regarding treatment decisions in the terminal stages of disease. In only three patients did discussion of treatment decisions call on the advance directive. One patient, a 65-year-old white man with lymphoma, executed a Durable Power of Attorney for Health Care and chose option 1 ("quality of life"). He was listed as " n o code comfort care." The day before the patient died, however, he changed his mind and expressed his wish for a full code. A few hours before death, the patient was asked if he wanted CPR in the event that his heart stopped and the patient wrote " a little bit." Cardiopulmonary resuscitation was done but the patient died. A second patient, a 56-year-old white man with small-cell lung cancer, executed the California Durable Power of Attorney for Health Care and chose option 1 ("quality of life"). The patient was " n o code full c a r e " because the patient's Notice to My Health Care Provider was not in the active medical chart until the patient's wife produced a copy of the advance directive 6 days before death and the patient's code status was changed to " n o code comfort care." (To our knowledge, this is the only instance in which we failed to place the Notice in the active medical record during hospitalization.) The third patient, a 67-year old white man, had chronic obstructive pulmonary disease. He had executed the advance directive and chosen option 2 (no life-sustaining treatment in the event of irreversible coma). At admission, the patient was placed on " n o code full care," but after discussions with his physician he executed a new California Durable Power of Attorney for Health Care indicating option 1 ("quality of life") and was then placed on " n o code comfort care." Few of our patients underwent a long interval of

Figure 3. Medical care charges in dollars for advance-directive and control groups by length of time before death. Bars represent 95% confidence intervals.

survival with diminished capacity such as persistent vegetative state or dementia. The effect of the advance directive on treatment decisions might be quite different, of course, in patients with diseases associated with long periods of mental incapacity, such as AIDS (although, in the 14 patients in this study, discussion before death was so extensive it precluded dependence on the advance directive) and more particularly Alzheimer disease. Further, that some discussion of treatment options and patient preferences occurred does not mean that patients fully understood their prognosis or the likely consequences of alternative treatment decisions that might be made. Because our study was done at an academic center, we are unable to state how generalizable our findings are. Not only would patient populations with different categories of disease be likely to provide different findings, but physicians practicing in other medical centers

Table 3. Medical Treatment in the Advance-Directive and Control Groups* Variable

Mean hospital days/patient Mean ICU days/patient Nursing-home days/patient Artificial feeding days/patient Ventilator hours/patient Patients with DNR orders, % Patients with CPR events, % Antibiotic charges/patient, $ Narcotic charges/patient, $ Patients dying at home, % Mean days in study

Control Group in = 50)

± ± ± ± ± 80 14 3590 ± 826 ± 30 352 ± 33.1 3.1 5.8 0.5 6.8

3.6 1.0 1.8 0.4 5.9 614 251 30

AdvanceDirective Group (n = 50) 40.8 2.5 2.3 0.9 10.5

± ± ± ± ± 68 22 4485 ± 758 ± 30 343 ±

4.4 1.0 0.7 0.5 6.9 808 125 32

P Valuet

0.18 > 0.2 .07 > 0.2 > 0.2 0.2 > 0.2 > 0.2 > 0.2 > 0.2 > 0.2

Patients in the Advance-Directive Group Who Executed and Returned the Document (n = 32) 41.1 3.0 2.8 0.4 16.4

± 5.6 ± 1.3 ± 1.0 ± 0.4 ± 10.8 63 25 3954 ± 992 586 ± 91 31 371 ± 43

Patients in the Advance-Directive Group Who Did Not Execute and Return the Document (n - 18) 40.2 1.6 1.6 1.8

± 7.2 ± 1.4 ± 0.6 ± 1.2 0 78 17 5429 ± 1399 1063 ± 298 28 294 ± 47

P Value*

> > > > > > > >

0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.14 > 0.2 0.2

* Means are expressed ± SE. CPR = cardiopulmonary resuscitation; DNR = do not resuscitate; ICU = intensive care unit. t To compare advance-directive and control groups, the Fisher exact chi-square test was used for proportions and the /-test for continuous variables. t Comparison of patients who executed the advance directive with those who did not.

1 October 1992 • Annals of Internal Medicine • Volume 117 • Number 7

Downloaded from by Tulane University user on 01/19/2019


might also respond differently to their patients, with or without documented instructions. Physicians in community hospitals might well know their patients more intimately than fellows and faculty members in a university setting; however, they might also be more anxious about legal and economic consequences to which academic salaried physicians might consider themselves more immune. Finally, although we found no evidence that physicians ignore patient preferences, we are still dubious about the current state of the art regarding advance directives. Whether offering the patients a list of procedures or attempting to gain insights into their qualityof-life preferences (5, 37), we find it hard to believe that any piece of paper, however artfully designed, will suffice as a satisfactory means of doctor-patient communication (38). In our opinion, only if the document stimulates discussion and communication between patients and physicians will the current enthusiasm for advance directives be justified. Grant Support: By grant R18 HS05617 from the Agency of Health Care Policy Research. Requests for Reprints: Lawrence J. Schneiderman, MD, Community and Family Medicine, 0622, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0622. Current Author Addresses: Drs. Schneiderman, Kronick, Kaplan, Anderson, and Langer: Division of Health Care Services, University of California, San Diego School of Medicine, La Jolla, CA 92093-0622. References 1. Schneiderman LJ, Arras JD. Counseling patients to counsel physicians on future care in the event of patient incompetence. Ann Intern Med. 1985;102:693-8. 2. Omnibus Reconciliation Act 1990. Title IV. Section 4206. Congressional Record, October 26, 1990:12638. 3. Callahan D. What Kind of Life: The Limits of Medical Progress, New York: Simon and Schuster; 1990:80. 4. Taylor H. Withholding and withdrawal of life support from the critically ill. N Engl J Med. 1990;322:1891-2. 5. Schneiderman LJ, Pearlman RA, Kaplan RM, Anderson JP, Rosenberg EM. Relationship of general advance directive instructions to specific life-sustaining treatment preferences in patients with serious illness. Arch Intern Med. [In press.] 6. Shmerling RH, Bedell SE, Lilienfeld A, Delbanco TL. Discussing cardiopulmonary resuscitation: a study of elderly outpatients. J Gen Intern Med. 1988;3:317-21. 7. Everhart MA, Pearlman RA. Stability of patient preferences regarding life-sustaining treatments. Chest. 1990;97:159-64. 8. Danis M, Southerland LI, Garrett JM, Smith JL, Hielema F, Pickard CG, et al. A prospective study of advance directives for life-sustaining care. N Engl J Med. 1991;324:882-8. 9. Emanuel LL, Barry MJ, Stoeckle JD, Ettelson LM, Emanuel EJ. Advance directives for medical care—a case for greater use. N Engl J Med. 1991;324:889-95. 10. Danis M, Patrick DL, Southerland LI, Green ML. Patients' and families' preferences for medical intensive care. JAMA. 1988;260: 797-802. 11. Buchanan AE, Brock DW. Deciding for Others: The Ethics of Surrogate Decision Making. Cambridge: Cambridge University Press; 1989.


12. Current Opinions of the Council on Ethical and Judicial Affairs of the American Medical Association. Chicago: American Medical Association; 1989:2.20, 2.21. 13. Orentlicher D. Advance medical directives. JAMA. 1990;263:2365-7. 14. American Academy of Neurology. Resolution on legislation regarding durable power of attorney for health care. Minneapolis: American Academy of Neurology; 1989. 15. The Hastings Center. Guidelines on the termination of life-sustaining treatment and the care of the dying. Bloomington: Indiana University Press; 1987. 16. California Code. 1987 Supplement. 1985. Article 5, Sections 24302444. 17. Zelen M. A new design for randomized clinical trials. N Engl J Med. 1979;300:1242-5. 18. Folstein MF, Folstein S, McHugh PR. Mini-Mental State: a practical method from grading the cognitive state of patients for the clinician. J Psych Res. 1979;12:189-98. 19. DiMatteo MR, Hays R. The significance of patients' physician contact: a study of patient satisfaction in a family practice center. J Community Health. 1980;6:18-34. 20. Wan T, Liveratos B. Validating a general well-being index by clinical measures of health. Gerontologist. 1977;17:129-30. 21. Fazio AF. A concurrent validational study of the NCHS General Weil-Being Schedule. Vital Health Statistics, Series 2, 1977. 22. Wallston KA, Wallston BS, DeVellis R. Development of the multidimensional health locus of control scales. Health Education Monographs. 1976;16:160-70. 23. Antonovsky A. Health, Stress and Coping: New Perspectives on Mental and Physical Well-Being. San Francisco: Jossey-Bass; 1981. 24. Kaplan RM, Anderson JP. The general health policy modes: an integrated approach. In: Spilker B, ed. Quality of Life Assessments in Clinical Trials. New York: Raven, 1990:131-49. 25. Kaplan RM, Anderson JP. A general health policy model: update and applications. Health Serv Res. 1988;23:203-35. 26. Kaplan RM, Bush JW, Berry CC. Health status: types of validity for an index of well-being. Health Serv Res. 1976;11:478-507. 27. Anderson JP, Bush JW, Berry CC. Classifying function for health outcome and QOL evaluation: self versus interviewer modes. Med Care. 1986;24:454-69. 28. Kaplan RM, Anderson JP, Wu AW, Mathews WC, Kozin F, Orenstein D. The quality of well-being scale: applications in AIDS, cystic fibrosis, and arthritis. Med Care. 1989;27(Suppl 3):S27-S43. 29. Kaplan RM. Quality of Life Assessment. In: Karoly P, ed. Measurement Strategies in Health Psychology. New York: Wiley; 1985: 115-46. 30. Langer RD, Wark R, Schneiderman LJ, Kaplan RM, Anderson JP, Chen M. A research-oriented medical cost accounting system. Int J Biomed Comput. 1991;28:161-7. 31. Scitovsky AA. The high cost of dying: what do the data show? Milbank Memorial Fund Quarterly/Health and Society. 1984;62:591609. 32. Lubitz J, Prihoda R. The use and costs of Medicare services in the last two years of life. Health Care Financ Rev. 1984;5:117-31. 33. Bedell SE, Delbanco KL. Choices about cardiopulmonary resuscitation in the hospital: when do physicians talk with patients? N Engl J Med. 1984;310:1089-93. 34. Uhlmann RF, Pearlman RA, Cain KC. Understanding of elderly patients' resuscitation preferences by physicians and nurses. West J Med. 1989;150:705-7. 35. Uhlmann RF, Pearlman RA, Cain KC. Physicians' and spouses' predictions of elderly patients' resuscitation preferences. J Gerontol. 1988;43:M115-21. 36. Seckler AB, Meier DE, Mulvihill M, Paris BE. Substituted judgment: how accurate are proxy predictions? Ann Intern Med. 1991; 115: 92-8. 37. Emanuel LL, Emanuel EJ. The medical directive: a new comprehensive advance care document. JAMA. 1989;261:3288-93. 38. Brett AS. Limitations of listing specific medical interventions in advance directives. JAMA. 1991;266:825-8.

1 October 1992 • Annals of Internal Medicine • Volume 117 • Number 7

Downloaded from by Tulane University user on 01/19/2019

Effects of offering advance directives on medical treatments and costs.

To examine the effects of advance directives on medical treatments and on patient satisfaction and well-being and to determine whether the enhancement...
2MB Sizes 0 Downloads 0 Views