Advance Directives Are More Likely Among Seniors Asked About End-of-Life Care Preferences

Objectives  To estimate the proportion of seniors in a large health maintenance organization (HMO) who had been asked about their end-of-life care preferences (EOLCPs) by a clinician and who had completed an advance directive (AD). To examine the association of having had an EOLCP discussion and AD completion.

Subjects and Methods  A random sample of HMO members aged 65 years or older were asked to complete a mailed survey about health and health-related issues in 1996. Data provided by 5117 seniors (80% response rate) were used to estimate the prevalence of EOLCP and AD among seniors overall and in specific risk groups. Bivariate and multiple logistic regression models were used to identify predictors of AD completion, especially having been asked about EOLCP.

Results  One third of seniors reported having an AD on file with the HMO, but only 15% had talked with a clinician about EOLCP. Both having been asked about EOLCP and having an AD were positively associated with age, but not significantly associated with sex, race/ethnicity, marital status, or self-rated health status. Having been asked by a clinician about EOLCP was significantly associated with completion of an AD.

Conclusion  Clinicians can play an important role in increasing AD completion rates among seniors by bringing up the subject of EOLCPs.

SINCE the implementation of the Patient Self-Determination Act in December 1991, health maintenance organizations (HMOs) and other health care institutions have been required to inform all adult patients about advance directives (ADs) and hospitals have been required to ask all patients whether they have completed ADs. Presence of an AD or documentation in the medical record that patients have been advised about ADs has become a quality indicator for the Joint Commission on Accreditation of Healthcare Organizations (JCAHO). While public support for ADs is high, surveys conducted shortly after the implementation of the Patient Self-Determination Act in December 1991 have shown that a relatively small proportion of adults in the United States have an AD.^1-3

A 1993 random sample survey of adults enrolled in the Kaiser Permanente Medical Care Program's Northern California Region found that less than one fourth of seniors had completed and filed an AD (Durable Power of Attorney for Health Care, Living Will, or other written ADs) with the health plan.⁴ In response to this problem of a low AD completion and filing rate, in 1996 the Kaiser Permanente Medical Care Program in Northern California disseminated a guideline about ADs to its clinical staff. This guideline recommends that all seniors, irrespective of health status, and members of all ages with a chronic disease or life-threatening illness should be asked if they have an AD. If a patient in one of these groups does not have a written AD, the clinician is requested to inquire about and document the patient's end-of-life care preferences (EOLCPs) and a designated person who could make health care decisions if the patient was unable to do so, and should advise the patient to complete a legally binding AD.

As a baseline for this effort to increase the proportion of members who have had a discussion about EOLCPs with a health care professional and filed an AD with the health plan, we analyzed data from a 1996 survey of Northern California Kaiser members (1) to estimate the proportion of seniors who have filed an AD with the health plan and have been asked by a clinician about EOLCPs, and (2) to examine the relationship between having a clinician ask about EOLCPs and completion of an AD.

During spring 1996, a stratified random sample of Kaiser members aged 20 years or older were asked to complete a confidential mailed member health survey questionnaire, which had the stated main purposes of helping the HMO learn about members' needs for specific types of health-related services and to determine whether the HMO was providing quality medical care to the membership. This survey was approved by the HMO's institutional review board. After up to 4 contact attempts, 5365 (80%) of the 6706 members aged 65 years or older (hereafter, seniors) who were included in this sample returned the questionnaire. The core questionnaire elicited a wide range of information about sociodemographic and health-related characteristics of adult members. In a supplemental section, seniors were asked (1) "Has a Kaiser nurse or doctor ever asked you about your EOLCPs (such as resuscitation instructions or other major interventions if you are terminally ill or in a coma? (yes/no)"; (2) "Do you have a Durable Power of Attorney for Health Care [DPAHC], Living Will, or written Advance Directives regarding EOLCPs on file with Kaiser? [yes/no]."

For this study, we first examined the proportion of seniors who had completed an AD and the proportion who said that a physician or nurse asked them about their EOLCPs. We looked at AD completion and end-of-life care preference discussion rates for both seniors overall, and for seniors whom the AD guideline suggested should be especially targeted for AD completion, ie, older seniors, those with a life-threatening illness (in this study, survivors of a heart attack or stroke in the previous 12 months) or chronic potentially life-threatening problem (in this study, angina or other heart problem), those self-reporting poor health, and those who reported an overnight hospitalization or emergency department visit in the previous 12 months. We calculated prevalence ratios and 95% confidence intervals to compare AD completion rates among those who had a health care provider ask about EOLCPs with rates for those who did not recall being asked. We next used χ² tests to examine whether seniors' sociodemographic characteristics (age, sex, race/ethnicity, education, and marital status), perceived level of participation with the physician in making decisions about medical treatment and other aspects of health care (none or slight, moderate, or quite a bit), and whether the person reported having a usual or personal physician were significantly associated with having been asked about EOLCPs and having completed an AD. Finally, we used a multiple logistic regression model to test whether factors significant in bivariate tests would remain significant after controlling for potential confounders. The adjusted odds ratios derived from exponentiated coefficients from this multivariate model are not reported because they cannot meaningfully be interpreted as relative risks, given that AD completion (the outcome) is a non-rare event. However, the statistical test of the coefficients in the model are valid regardless of the rarity of the outcome.⁵

Table 1 shows selected sociodemographic, health, and health care-related characteristics of the 5117 seniors who provided data about AD completion status. Respondents were predominantly in their 70s, well educated, and in good health.

Rates of AD completion and end-of-life care preference discussions are shown in Table 2. One third of the seniors reported that they had completed an AD and had it on file with the health plan. However, only 15% reported that a physician or nurse had ever asked them about their EOLCPs. This same pattern was seen for those subgroups of members whom the HMO's AD guideline suggests should be especially encouraged to complete an AD. Among these subgroups, 36% to 40% reported having an AD, but only 14% to 35% reported that a physician or nurse had asked about EOLCPs.

Both having an AD and having had a clinician ask about EOLCPs were significantly associated with age. Approximately 26% of early seniors (aged 65-69 years), 35% of middle (aged 70-79 years), and 46% of older seniors (aged ≥80 years) reported having an AD and approximately 9%, 16%, and 27% of these age groups reported having been asked about EOLCPs. Completion of an AD was not significantly associated with sex, race/ethnicity, marital status, self-rated health status, or having a usual physician. Education was significantly associated with AD completion, although not in the expected direction.

Table 3 compares AD completion among members who reported that they had or had not been asked by a clinician about their EOLCPs, overall and by age, sex, and health-related characteristics. Overall, the prevalence ratios indicate that those who reported having had a physician or nurse ask about EOLCPs were nearly 3 times as likely to report having a written AD on file as those who had not been asked, with the exception of those in poor health.

In the multiple logistic regression model, the relationship between having been asked about EOLCPs and completing an AD remained statistically significant after controlling for age, sex, education, and extent of involvement in health care decision making (race/ethnicity, marital status, health status, and having a usual physician were not included in the model due to lack of significant association in bivariate analyses). Age, sex, education (lower completion among college graduates vs those with <12 years of education), and perceived level of involvement in decisions about one's health care also remained statistically significant independent variables.

The results of this study suggest that as of 1996, less than one third of seniors in this large group practice model HMO had completed and filed an AD, even though efforts to encourage members had been made through waiting room posters, articles about ADs in the health plan newsletter, and the personal efforts of some health plan staff. As has been found in previous studies, rates of AD completion were higher among seniors who were older and in poorer health than among those who were still relatively young and healthy.⁶ However, in contrast to other studies,^7-9 we found that college graduates were significantly less likely to have an AD on file than those with less education, even after controlling for age, sex, and having been asked about EOLCPs by a physician or nurse.

Less than 15% of the seniors overall, and not much higher percentages of seniors in the high-risk subgroups, reported that they had been asked about EOLCPs by a health plan physician or nurse. This rate is comparable with that for outpatients shown in studies conducted shortly after the Patient Self-Determination Act was implemented,³ suggesting that physicians are still not considering this a high priority issue.While one cannot interpret causality from a cross-sectional study, we found that seniors who were asked about EOLCPs were nearly 3 times more likely to have filed an AD than those who said they had never been asked. These results parallel findings of intervention studies that physician-initiated discussion about ADs significantly increases rates of AD completion.^10-14

A strength of this study is that the population surveyed is more representative of the general population than previous studies that have generally been conducted with recently hospitalized patients or those at high risk for hospitalization. Most of the seniors who responded were in good health, had not been recently hospitalized, and were not living in nursing homes. However, as with any study based on cross-sectional, unvalidated self-reported data, there are issues that affect the interpretation of the results.

First, since our survey question asked whether the individual had completed and filed an AD with the HMO, the percentages of seniors with an AD may actually be higher. For example, some individuals may have given a written or oral AD to a family member or friend, but not to the HMO. However, since these often unnotarized ADs may not be legally binding, or readily available in a medical emergency, our findings probably reflect the magnitude of intervention needed for HMOs to have ADs on file prior to hospitalization. It is also possible that the self-reported data overestimate AD completion in view of anecdotal evidence from clinicians that seniors often get confused between ADs and wills or estate planning documents. We tried to minimize that error through the specific wording of our question.

Because the results are based on analysis of cross-sectional data, we cannot interpret a causal relationship between having been asked by a health care professional about EOLCPs and completion and filing of an AD. Unfortunately, we do not know the sequence of the 2 events based on the way the questions were asked. Also, we cannot rule out the possibility of recall bias, ie, that people who had filed an AD may have been more likely to recall being asked about their preferences than those who who had not done so.

Physicians and nurses will probably need to be held more accountable for initiating discussion about ADs if a health plan wants to have ADs on file for most of its seniors and high-risk members. Morrison et al¹⁵ found that the strongest and most consistent barriers to physicians' discussion of ADs with patients is the erroneous belief that ADs are unnecessary for young, healthy patients, and the lack of knowledge about how to formulate ADs. However, they also found that time constraints during the oupatient visit and lack of comfort with issues relating to death and dying were significantly related to likelihood that physicians would initiate a discussion about ADs with their patients.¹⁵ This suggests that if physicians and nurses are to be held more accountable, they will need to be given guidelines about AD completion, as well as training in how to initiate discussion about EOLCPs. Group appointments may be an effective forum for physicians to educate their patients about end-of-life care treatment options and ADs because they would allow the physician to engage with several patients at one time and in a less stressful situation than might occur in the context of a one-on-one physician office visit. Most patients will probably require systematic follow-up to make sure that they take the time to complete and file an AD.^{1,7,9,12,16,17} While such system supports will require resources, they will probably more than pay for themselves not only by reducing the use of unwanted costly life-support procedures but also by reducing the pain and suffering of patients, their loved ones, and their health care teams as well.

Accepted for publication August 4, 1998.

Corresponding author: Nancy P. Gordon, ScD, Division of Research, Kaiser Permanente, 3505 Broadway Ave, Oakland, CA 94611 (e-mail: npg@dor.kaiser.org).