Elders' Environs and Their End-Of-Life Preferences

Article Outline

• Abstract
• Methods
• Results
  • Questions 1 and 2: Endotracheal intubation for acute pneumonia?
  • Question 3: Nursing home placement if required for progressive dementia?
  • Comparing Choices for Acute Illness versus Progressive Dementia
• Discussion
• Appendix 1. Elders' Choice Questionnaire
  • Demographic information
• References
• Copyright

Hypothesis

Elders' predilections regarding end-of-life interventions vary with their living environs.

Methods

Patients in 3 settings—assisted living/outpatient, skilled nursing facility (SNF), and acute hospitalization—were asked to complete a brief questionnaire.

Results

A total of 269 patients who averaged 80.0 ± (SD) 8.1 years, 44% male, 70% white were studied. Eighty-five patients were outpatient elderly, 101 were hospitalized for acute illnesses, and 83 were interviewed in SNFs. Outpatients (44/85; 52%) and acutely ill inpatients (40/101; 40%) were more likely than patients residing in SNFs (19/81; 23%) to choose comfort care only (P = .047) for acute pneumonia requiring endotracheal intubation (ETI). Overall, 32% changed their choice for ETI, opting for comfort care only if acute pneumonia was followed by disposition to an SNF. However, ambulatory and acutely ill elderly patients were 3 times as likely as SNF patients to change from aggressive to comfort care if the most likely outcome was disposition to an SNF (P < .001). In multivariate regression models, age (>80), gender, number of lost ADLs (>2), and self-described quality of life were not associated with choosing comfort care instead of ETI, whereas place of residence (SNF versus home) was independently associated with choosing ETI (odds ratio = 3.5; 95%CI = 1.9–6.4). Similarly, those already living in an SNF were more likely to opt for remaining there for advancing dementia (odds ratio = 7.7; 95%CI = 3.8–15.8). However, choices for ETI did not coincide with choosing an SNF for advancing dementia.

Conclusions

Elders residing in nursing homes were more likely than ambulatory patients to request invasive end-of-life care, a difference that was more pronounced when outcome required disposition to an SNF. These preferences were not dependent on patients' self-described disability or quality of life. This study suggests that qualitative outcomes matter to patients and their choices are associated with their place of residence.

Keywords: End-of-life, preferences, nursing home, skilled nursing facility, resuscitation, mechanical ventilation

A number of studies have examined elderly patients' end-of-life preferences over time,¹, ², ³, ⁴ suggesting that as patients age and acquire greater burden of illness, they are more likely to prefer invasive life-sustaining therapies. Previous studies have centered on the effects of patients' self-reported health status on their predilections. Fried and colleagues³ suggest that instability of patients' health partially accounts for observed instability of their end-of-life predilections in hers and previous studies. Because patients' living milieu is likely a marker of disease severity, we reasoned that patients at 3 levels of care, ie, outpatient, acute hospitalization, and nursing home, may differ in their preferences regarding life-sustaining therapies.

Back to Article Outline

Methods 

This study was conducted after review and approval by the Institutional Review Committee of Bridgeport Hospital, a 350-bed community teaching hospital affiliated with Yale University School of Medicine. Three groups of patients, aged older than 65 years, were studied: acutely ill hospitalized patients (admitted between October 1 and 31, 2008), ambulatory outpatients, and nursing home residents. Patients with a previous diagnosis of terminal illness, dementia, or who had acute delirium were excluded from the study.

Questionnaires were administered by study personnel reading from a script (see Appendix 1). All acute inpatients were interviewed in person between 24 and 48 hours of admission. Eighty-five ambulatory outpatients were included in the study in 3 types of settings: assisted living facilities, home patients, and follow-up visits to primary care providers. Questions were asked directly or over the phone. Eighty-three nursing home residents were included. The Confusion Assessment Method was used in every patient to assess for and exclude patients with delirium. Patients were also excluded if they could not understand the questions or if they had a terminal illness.

A uniform data abstraction sheet was used to collect responses which were tabulated using Excel and SPSS 17. Categorical variables were reported as percentages of total numbers of patients, and comparisons were made using chi-square tests. The frequencies of responses to end-of-life questions were compared based on demographic characteristics, self-described functional status (activities of daily living [ADLs], instrumental activities of daily living [IADLs], dependency, quality of life [QOL]) and place of residence. Logistic regression analyses were used to determine independent predictors of choosing comfort versus supportive (eg, endotracheal intubation, nursing home) care for questionnaire scenarios. Acutely hospitalized patients who resided in SNFs were excluded (because their disposition destination could not be home). Variables found to be significantly associated with choices in univariate analyses and those with biological plausibility were included in multiple logistic regression models. Analyses were performed using SPSS Statistics 16 (IBM Company, Chicago, IL) and EpiInfo version 3.4.1 software (Centers for Disease Control and Prevention, Atlanta, GA). A 2-tailed P value of less than .05 was considered statistically significant.

Back to Article Outline

Results 

A total of 269 patients were studied. They averaged 80.0 ± 8.1 years, 119 (44%) were male, and 70% were white. Eighty-five patients were outpatient elderly, 101 were hospitalized for acute illnesses, and 83 were interviewed in skilled nursing facilities (SNFs). Demographic characteristics were similar in the 3 groups (see Table 1).

Table 1. Demographic Characteristics of Respondents

Characteristics	Outpatient		Inpatient		Nursing Home
	n		n		n
	85	%	101	%	83	%
Gender
Male	37	44	42	42	40	48
Female	48	57	59	58	43	52
Age
Average	79.8		78		82.5
Median	80		78		82
Range	62–100		65-93		56–100
Race
White	54	63	67	66	67	81
Black	21	25	14	14	11	13
Asian	10	12	1	1	5	6
Other	0		19	19	0	0
Religion
Catholic	42	49	59	58	46	55
Protestant	12	14	13	13	11	13
Orthodox	9	11	2	2	2	2
Other Christian	5	6	20	20	12	14
Muslim	0	0	1	1	1	1
Jewish	1	1	3	3	4	5
Agnostic	10	12	1	1	2	2
Other	6	7	2	2	1	1
Comorbidities
Diabetes	28	30	35	34	23	28
CAD	26	31	35	35	18	22
CHF	8	10	19	19	15	18
CVA	17	20	8	8	16	19
COPD	12	14	25	25	18	22
CKD	5	6	11	11	2	2

CAD, coronary artery disease; CHF, congestive heart failure; CVA, stroke; COPD, chronic obstructive pulmonary disease; CKD, chronic kidney disease.

There was a strong correlation between patients' self-described ADLs/IADLs and QOL. For example, patients who described their QOL as fair or poor reported inability to complete an average of 1.6 ADLs and 2.5 IADLs as compared with an average of 1.4 impaired ADLs and 1.1 impaired IADLs in those who described their QOL as good or excellent (P < .001). SNF residents were more likely than ambulatory elderly patients to require assistance with more than 2 ADLs (36% versus 12%; P = .048) and IADLs (84% versus 52%; P < .001). Fifty-seven percent of SNF residents described themselves as “dependent” (versus 30% of outpatients; P < .001) and were more likely to report poor or fair quality of life (48% versus 32% of outpatients; P = .049).

Questions 1 and 2: Endotracheal intubation for acute pneumonia? 

Taken as a whole, women (51.7%) were more likely than men (47%; P = .02) to prefer only comfort care in the event of acute pneumonia requiring mechanical ventilation. No other demographic characteristic or functional measure (ie, ADL, IADL, QOL) was associated with choosing endotracheal intubation (ETI) versus comfort care only. Outpatients (44/85; 52%) and acutely ill inpatients (40/101; 40%) were more likely than patients residing in SNFs (19/81; 23%) to choose comfort care only (P = .047). If the outcome of ETI/hospitalization for acute pneumonia required disposition to an SNF, overall, 32% changed their choice for ETI, opting for comfort care only. Although acceptance of ETI declined for all groups, ambulatory and acutely ill elderly patients (60% and 54%, respectively) were 3 times as likely as SNF patients to choose comfort care only for acute pneumonia if the most likely outcome was disposition to a nursing home (P < .001). Among acutely ill hospitalized patients, 16 of 101 resided in an SNF before hospitalization and these patients were more likely than those residing at home before admission to choose ETI for acute pneumonia (54.1% versus 43.8%, P = .04). In multivariate regression models, age (>80), gender, number of lost ADLs (>2), and self-described quality of life were not associated with choosing comfort care instead of ETI, whereas place of residence (SNF versus home) was independently associated with choosing ETI (odds ratio [OR] = 3.5; 95% confidence interval [CI] = 1.9–6.4).

Question 3: Nursing home placement if required for progressive dementia? 

Taken as a whole, whites (62%) were more likely than non-whites (50%) to choose long-term SNF for progressive dementia. Outpatients (43/83; 52%) and acutely ill inpatients (51/100; 51%) were more likely than patients residing in an SNF (11/83; 12%) to choose only comfort care (P = .047). The 16 (of 101) patients who resided in an SNF before hospitalization for acute illness were more likely than those residing at home to choose only comfort care in the event of advancing dementia (68.8% versus 52.9%, P = .01). In multivariate regression models, age (>80) tended to be associated with choosing nursing home placement (P = .1), but otherwise only place of residence (home versus SNF) was independently associated with choosing a long-term SNF for advancing dementia (OR = 7.7; 95% CI = 3.8–15.8).

Comparing Choices for Acute Illness versus Progressive Dementia 

In general, choices for invasive care for acute illness (ie, ETI) did not coincide with the same patient's choices for an SNF for progressive dementia. Additionally, no index of function (ADL, IADL, QOL) was associated with choices. In both ambulatory elderly patients and those residing at home before hospitalization for acute illness, choosing ETI did not coincide with choosing an SNF for progressive dementia; 62.2% of patients who chose ETI for acute pneumonia opted for comfort care rather than prolonged SNF in the setting of advancing dementia and 53.8% of patients who preferred only comfort care for acute pneumonia/respiratory failure also chose comfort care rather than prolonged SNF.

The last part of the analysis focused on patients who answered questions 1 and 10 (ie, querying choice of ETI for pneumonia with insubstantial differences in wording) the same. A total of 124 ambulatory or acutely hospitalized and 50 nursing home residents were consistent in their choices. As in the overall cohort, nursing home patients were more likely to choose intensive care for acute pneumonia (74% versus 36.3%; P < .001) and hospitalization for acute illnesses if institutionalized for progressive dementia (88% versus 43.5%; P < .001). After adjusting for age, sex, place of residence, and self-described activity levels and quality of life, multiple logistic regression analyses showed that odds of patients residing in nursing homes choosing ETI for respiratory failure were more than 3 times those of outpatient and acutely hospitalized elderly (OR = 3.6, 95% CI = 2.0–6.6). And their odds of choosing hospitalization for acute illness after institutionalization for progressive dementia were nearly 7 times those of outpatients and acute-care inpatients (OR = 6.7, 95% CI = 3.4–13.4). Of patients in SNFs who were consistent in their preference for critical care and/or hospitalization for acute illness if progressively demented, 36% preferred acute hospitalization irrespective of their disabilities, 29% declined if they could no longer feed themselves or recognize loved ones, and 6% changed their choice if unable to toilet.

Back to Article Outline

Discussion 

This study suggests that patients' end-of-life predilections may vary with their living milieu. Elders who already resided in nursing homes were 60% more likely to prefer mechanical ventilation for acute pneumonia than ambulatory elderly patients. Ambulatory elderly patients were 3 times as likely to choose comfort care only for acute pneumonia if the most likely outcome was disposition to a nursing home. These data support our hypothesis that patients' living situation is associated with their willingness to accept invasive treatments and chronic institutionalization; those with the greatest levels of dependency were most likely to accept invasive therapies. This study also demonstrates a surprisingly high frequency of elders from all 3 environs (18%–60%) who would reject mechanical ventilation for pneumonia, and instead opt for comfort care only, if a nursing home is required for convalescence following hospitalization.

In 2536 outpatients followed over 2 years, Danis and colleagues¹ demonstrated that many more patients were likely to move from acceptance to rejection of cardiopulmonary resuscitation (CPR) over time (509/1272), than rejection to acceptance (21/261). In a study of 645 physicians followed over 6 years, those who experienced functional decline and depression had 4-fold excess odds of preferring “high-burden” treatments that included CPR, mechanical ventilation, and chemotherapy.² In a study of 189 outpatient elders with cancer, heart failure, or chronic obstructive pulmonary disease (COPD), participants were followed for up to 2 years and end-of-life preferences vacillated significantly. Fried and colleagues³ concluded that in this cohort, preferences tended to be more consistent with self-rated health and quality of life than with “core values.” In 1590 acutely hospitalized patients, Rosenfeld and colleagues⁴ reported stability of patients' preferences regarding CPR over shorter time horizons (73% preferred it initially and 70% 2 months later). A small longitudinal study by McParland and colleagues⁵ that included both ambulatory and acutely ill elderly patients, suggested patients' preferences for invasive therapies over longer time periods (12 and 24 months) were unstable, ie, roughly the same number who had opted out changed preferences to opt in and vice versa. Our study, although not longitudinal, is unique in that it compares preferences of outpatients, acute inpatients, and nursing home residents. However this is also a weakness of the study (see later in this article).

In an elegant series of articles, Lynn and colleagues⁶, ⁷, ⁸ examined end-of-life preferences in patients with cancer, congestive heart failure, and COPD. For all 3 terminal conditions, patients' preferences for only comfort care increased with time as level of dysfunction increased. We performed this study with the hypothesis that preferences for life-sustaining therapies and institutionalization might vary with the burden/severity of nonterminal illnesses as marked by their living environs. For example, death is more of an abstract notion for young, well patients—it is difficult to contemplate increasing burden of illness, and the gravity of end-of-life decisions are not as relevant. As patients age, acquire increased illnesses, and the possibility of nearer-term death increases, patients might opt increasingly to prolong their lives using measures that were not acceptable (because the situation was not fully comprehensible) when they were younger. As the burden of illness and dependence increase, patients may reach a point at which they no longer wish to prolong their lives and increasingly choose to forgo invasive life-sustaining (or death-prolonging) therapies.⁶, ⁷, ⁸ This may explain why our outpatients, unable to fathom life in an SNF, were more likely to choose comfort care instead. And why SNF patients, who are certainly more acclimated to institutionalization and greater dependency, opted for more aggressive therapies until they reach advanced debilitation. Whereas 77% of our nonterminal SNF patients wanted intubation for pneumonia, two thirds declined acute hospital care if they became unable to recognize family or eat (see Table 2). This idea is supported by the work of Lynn and colleagues⁶, ⁷, ⁸ and more recent work suggesting that elderly patients favor less aggressive management with increasing dysfunction.⁹ Nonetheless, our findings are unique in that residence remained independently associated with choices even after adjustment for function and quality of life. It may be surprising to some clinicians that, overall, institutionalized elderly would opt for more aggressive management than outpatient elderly. Our findings suggest that such patients' preferences should NOT be presumed; to respect their autonomy requires repetitive discussions (as conditions change) and informed consent.

Table 2. Responses of 3 Groups to the Questionnaire

ADL, activities of daily living; CPR, cardiopulmonary resuscitation; IADL, instrumental ADL; SNF, skilled nursing facility; QOL, quality of life.

This idea supports Maclean's suggestion¹⁰ that it may not be possible to forecast with sufficient accuracy decisions of our “future selves.” If true, advance directives (AD) may be helpful when a patient becomes suddenly unable to make health care decisions, but only if they are regularly reviewed to ensure they remain consistent with patients' current preferences.¹¹ ADs have been criticized because some caregivers may extrapolate intent of patients and withhold some treatments inappropriately, eg, patient intends only “no CPR,” which is extended to mean, no critical care. ADs could therefore misinform clinicians of a patient's wishes if inappropriately extrapolated to preclude care not specified in the AD and/or if the patient's preferences are no longer reflected in the AD.

Our study has a number of notable limitations. We did not perform longitudinal analysis of the same patients over the continuum of life and disease, which would have allowed for more robust comparisons (patients' answers compared with previous over time and increasing burden of illness). Although we ascertained IADLs, ADLs, and patients' self-reported wellness, we did not precisely quantify acute and chronic illness burden. Rather residence (home versus SNF) was considered a surrogate for burden of illness and dependency. Accordingly, differences of demography and/or underlying illnesses might also account for observed differences rather than severity of illness and dependency. It is also possible (or likely) that patients who reside in SNFs do not reflect views of elderly in general; some have selected institutionalization as opposed to end-of-life at home and our findings may represent some selection bias.¹² We did not exclude nursing home patients from our acute care inpatient sample, nor did we determine whether nursing home patients were permanent or temporary (eg, rehabilitation) residents. Arguments can be made to include or exclude overlapping groups. For example, nursing home patients who opt for acute hospital care could have different predilections than rehabilitation patients or permanent residents who refuse acute hospitalization. Irrespective, our results were consistent across questions and suggest that patients in nursing homes, whether long- or short-stay, were more inclined to find invasive life-sustaining therapies preferable, compared with outpatient elders. Second, our questionnaire was not validated before administration. We included 2 questions to test internal consistency of individuals' answers (to questions 1 and 10; see Appendix 1), intra-responder variability was poor. Although it is somewhat reassuring that the same relationships were observed in those patients who answered these nearly identical questions the same, our results should be interpreted cautiously until this hypothesis is retested, using more robust tools in larger groups of patients, preferably followed longitudinally during wellness, acute illness, and chronic institutionalization. Finally, although patients with evident dementia or delirium were not included, our study did not screen for patients with depression, which may significantly affect end-of-life predilections.², ¹³, ¹⁴

In conclusion, our study suggests that elders' end-of-life preferences may differ based on their living environs. Elders also considered both quantitative and qualitative information in their end-of-life decision making. Our findings support the notion that increasing dependency is associated with preferences for more invasive acute life-sustaining therapies and institutionalization.

Back to Article Outline

Appendix 1. Elders' Choice Questionnaire 

(read script/questions verbatim)

Script:

“We are conducting a study to examine the preferences of older adults about their medical care. None of your answers will impact your future treatments and your answers will remain strictly confidential if you choose to participate.”

“Are you willing to participate?” If yes, continue with questionnaire.Initials:

Demographic information 

(If Yes then skip the next question)

Back to Article Outline

References 

1. Danis M, Garrett J, Harris R, Patrick DL. Stability of choices about life-sustaining treatments. Ann Intern Med. 1994;120:567–573
   • View In Article
   • MEDLINE
2. Straton JB, Wang N, Meoni LA, et al. Physical functioning, depression, and preferences for treatment at the end of life: The Johns Hopkins precursors study. J Am Geriatr Soc. 2004;52:577–582
   • View In Article
   • MEDLINE
   • CrossRef
3. Fried TR, O'Leary J, Van Ness P, Fraenkel L. Inconsistency over time in the preferences of older persons with advanced illness for life-sustaining treatment. J Am Geriatr Soc. 2007;55:1007–1014
   • View In Article
   • CrossRef
4. Rosenfeld KE, Wenger MS, Phillips RS. Factors associated with change in resuscitation preference in seriously ill patients. Arch Intern Med. 1996;156:1558–1564
   • View In Article
   • MEDLINE
5. McParland E, Likourezous A, Chichin A, et al. Stability of preferences regarding life-sustaining treatment: A two-year prospective study of nursing home residents. Mt Sinai J Med. 2003;70:85–92
   • View In Article
   • MEDLINE
6. Levenson JW, McCarthy EP, Lynn J, et al. The last six months of life for patients with congestive heart failure. J Am Geriatr Soc. 2000;48:S101–S109
   • View In Article
   • MEDLINE
7. McCarthy EP, Phillips RS, Zhong Z, et al. Dying with cancer: Patients' function, symptoms, and care preferences as death approaches. J Am Geriatr Soc. 2000;48:S110–S121
   • View In Article
   • MEDLINE
8. Lynn J, Ely EW, Zhong Z, et al. Living and dying with chronic obstructive pulmonary disease. J Am Geriatr Soc. 2000;48:S91–S100
   • View In Article
   • MEDLINE
9. Fried TR, Van Ness PH, Byers AL, et al. Changes in preference for life-sustaining treatment among older persons with advanced illness. J Gen Intern Med. 2007;22:495–501
   • View In Article
   • CrossRef
10. Maclean AR. Advance directives, future selves and decision-making. Med Law Rev. 2006;14:291–320
    • View In Article
    • MEDLINE
    • CrossRef
11. Reilly RB, Teasdale TA, McCullough LB. Projecting patients' preferences from living wills: An invalid strategy for management of dementia with life-threatening illness. J Am Geriatr Soc. 1995;42:997–1003
    • View In Article
    • MEDLINE
12. Pritchard RS, Fisher ES, Teno JM, et al. Influence of patient preferences and local health system characteristics on the place of death. SUPPORT investigators. Study to understand prognosis and preferences for risks and outcomes of treatments. J Am Geriatr Soc. 1998;46:1242–1250
    • View In Article
    • MEDLINE
13. Wright AA, Zhang B, Ray A, et al. Associations between end-of-life discussions, patient mental health, medical care near death and caregiver bereavement adjustment. JAMA. 2008;300:1665–1673
    • View In Article
    • CrossRef
14. Menon AS, Campbell D, Ruskin P, Hebel JR. Depression, hopelessness, and the desire for life-saving treatments among elderly medically ill veterans. Am J Geriatr Psychiatry. 2000;8:333–342
    • View In Article
    • MEDLINE
    • CrossRef