Abstract
Context:
Cognitively impaired older adults frequently need surrogate decision-making near the end-of-life. It is unknown whether differences in the surrogate’s relationship to the decedent are associated with different end-of-life treatment choices.
Objectives:
To describe differences in end-of-life care for community dwelling, cognitively impaired older adults when children and spouses are involved in decision-making.
Methods:
Retrospective observational study.
Results:
Among 742 community-dwelling adults with cognitive impairment (mild cognitive impairment or dementia) prior to death, children participated in end-of-life decisions for 615 patients (83%) and spouses participated in decisions for 258 patients (35%), with both children and spouses participating for 131 patients (18%). When controlling for demographic characteristics, decedents with only a spouse decision-maker were less likely to undergo a life-sustaining treatment than decedents with only children decision-makers (p < 0.05). There was no difference in the probability of in-hospital death or burdensome transfers across facilities across decedent-decision-maker relationships. Differences in rates of life-sustaining treatment were greater when we restricted to decedents with dementia.
Conclusion:
Decedents with cognitive impairment or dementia were less likely to receive life-sustaining treatments when spouses versus children were involved with end-of-life treatment decisions but were no less likely to experience other measures of potentially burdensome end-of-life care.
Introduction
More than half of older Americans are diagnosed with mild cognitive impairment or dementia near the end of life1,2. Cognitive impairment is a significant risk factor for hospitalizations and emergency room visits3–5. Some 78% of community-dwelling and 65% of nursing home-based dementia patients are hospitalized at least once in the last year of life, with the majority of stays following an emergency department visit5. Patients are often unable to participate in medical decision-making during these episodes and require assistance from surrogate decision-makers.
Though there is little evidence that life-sustaining treatments such as feeding tube placement and cardiopulmonary resuscitation improve either the quality or length of life of persons with dementia, these services are frequently provided6–10. Life-sustaining treatments may be more likely to benefit patients with mild cognitive impairment and treatment guidelines for these patients weight patients’ previously expressed preferences more heavily11. In prospective studies, surrogate decision-makers have a moderate understanding of their loved one’s treatment preferences but report more difficulty choosing not to approve treatment even if they believe the patient might prefer palliative care12,13. A small scenario study of dementia patients and their spouses found that spouses selected more aggressive treatments than their partners preferred14. Additionally, decision makers were found to place lower importance on patient preferences as their time as caregivers increased15.
Advance care planning is frequently recommended as a strategy to reduce potentially unwanted and burdensome end-of-life care. A prior study found that written advance directives were linked to less aggressive end-of-life care for community-dwelling older adults with dementia16, with other research finding that advance directives improve concordance between end-of-life treatments and the decedent’s wishes, especially when the patient expressed a preference for less-aggressive care13,17–23. However, less is known about how the decedent’s relationship to the surrogate decision-maker may influence treatment outcomes for patients with mild cognitive impairment or dementia. We tested whether decision-maker relationships were systematically associated with differences in end-of-life care for community dwelling patients with all types of cognitive impairment and the subset with dementia, for whom life-sustaining treatments are not believed to meaningfully extend length or quality of life8,24.
Methods
Data and Methods
We performed a retrospective, observational cohort study to compare end-of-life treatment for patients with cognitive impairment when children and spouses were involved in treatment decisions using the nationally representative Health and Retirement Study linked to Medicare claims.
We identified decedents and surrogate decision-makers using Health and Retirement Study (HRS) Core and Exit Interviews. The HRS, a longitudinal study of older Americans, interviews households every two years and, after the death of a respondent, conducts post-mortem exit interviews with a proxy informant to collect information on end-of-life care and other developments since the final interview25,26. HRS respondents designate a next-of-kin proxy prior to death who will be able to complete the exit interview on their behalf.
From these exit interviews, we identified decedents whose proxies reported that they required a treatment decision near the end of life and had a spouse or descendant (referred to as “children” but including small numbers of children-in-law and grandchildren) participate in the decision. Exit interview respondents report whether “any decisions have to be made about the care and treatment of [the decedent] during the final days of [her/his] life” and, if decisions were required, provide a list of all participants. 94.4 percent of proxies who reported that decisions needed to be made were involved in the EOL decision-making process.
We restricted our sample to respondents who died from 2002 to 2015 who were 65 years and older at death and had linked Medicare claims data available. We focused on the subset of decedents who were cognitively impaired as of their last HRS core interview, lived in the community prior to death, and required assistance with end-of-life decision-making. We omitted nursing home decedents in order to isolate the role of family decision-makers from nursing home institutional practices and staff who may have played a part in EOL decision-making.
Of the 7,410 HRS decedents with an Exit Interview and complete MedPAR claims, 742 respondents met our sample criteria (see Sample Flowchart, Figure 1). We restricted our sample to decedents who required an end-of-life treatment decision (N = 3,252 respondents), had a spouse or child participating in said decision (N = 2,939), were cognitively impaired with or without dementia (N = 1,851), and were living in the community (N = 742) at the time of death.
Figure 1:
Sample Enrollment Flowchart
Outcomes and Measures
Using linked Medicare Provider Analysis and Review (MedPAR) claims from the year of death, we collected three previously validated measures of potentially burdensome end-of-life care16,27,28. We identified in-hospital deaths from MedPAR discharge codes. Burdensome transfers included any transfers during a patient’s last three days, three or more hospitalizations in the last 90 days of life, and more than one hospitalization for dehydration, pneumonia, sepsis, or a urinary tract infection in the last 120 days of life27. Life-sustaining treatments included receipt of CPR, gastrostomy, feeding tube insertion, tracheostomy, dialysis, or mechanical ventilation in the last 6 months of life28.
The majority of Medicare beneficiaries receive care through Fee-for-Service Medicare, generating claims for all covered services. Some Medicare beneficiaries opt into Medicare Advantage (MA) managed care plans instead. We omitted respondents in MA plans who died prior to 2009, when MA hospitalization data was first reported in MedPAR. We included MA enrollees from 2009 or later and controlled for MA status in regression models.
Exit interview proxies also report whether the decedent had a written advance directive (AD) requesting limited care, which have been associated with more limitedlower treatment intensity in previous studies16. Because surrogates frequently rely on advance directives when making treatment decisions, we compared the rate of advance directive completion between decedents with spouse vs child decision-makers. We focused on advance directives requesting limited care, since ADs requesting “all care possible” are extremely rare (N = 11, 1.5%).
Decedent Cognitive Status
We identified decedent cognitive status using an algorithm designed to assess cognition in the HRS16,29,30. We used decedent responses to the Telephone Interview for Cognitive Status (TICS) and proxy assessments of cognitive ability from the last core interview to classify cognitive function as cognitively normal, cognitive impairment with no dementia (CIND), and dementia. On average, these assessments were completed 479 days before death, suggesting that cognitive impairment and/or dementia was already present during the last six months of life and was therefore present when surrogates were involved with the reported treatment decisions. Decedents who were cognitively normal during their last core interview but may have been temporarily impaired by delirium during hospitalization or whose cognition may have rapidly declined from normal at the last assessment were not included in our sample in order to focus on the group whose cognition was impaired prior to treatment would remain impaired after the acute medical episode.
Statistical Analysis
We compared the probability of an in-hospital death, burdensome transfer, or life-sustaining treatment when spouses or children participated in treatment decisions. We studied differences between decedents with only children, children and the spouse, or only the spouse participating in decisions. We estimated multivariate logistic regressions for each outcome, controlling for demographic (decedent age, sex, race/ethnicity, education, veteran status, household wealth quartile), family (married, number of children, children living within 10 miles), MA enrollment, years with dementia, and comorbid conditions that may also influence clinical decision-making and prognosis (cancer, lung or heart condition, diabetes, stroke, or a mental health condition such as depression or anxiety). All models were estimated using Stata 15MP, with p-values of 0.05 considered statistically significant. Based on our regression models, we report average probabilities that a decedent in our sample would experience each of the EOL outcomes with child or spouse involvement. Our study was exempt from review by the Johns Hopkins School of Public Health IRB.
Surrogate decision-makers may not be aware of cognitive impairment when patients are mildly cognitively impaired, and these patients may also participate more actively in treatment decisions. Additionally, there is more concern in the literature about life-sustaining treatments being potentially burdensome to patients with dementia without increasing length of life8,24. Given this, we also studied the relationship between surrogate relationships and end-of-life treatment intensity in the subset of decedents with dementia prior to death (N = 335) based on the HRS cognitive assessment.
We undertook several specification checks to ensure that our results were not sensitive to methodological decisions. We used HRS sampling weights to test whether our estimates were robust to re-weighting based on population representativeness. Because end-of-life treatment intensity varies geographically31, we also confirmed that results were not sensitive to controlling for regional practice patterns using decedents’ Hospital Referral Region (HRR) quartile of EOL Medicare spending. Finally, because comorbid health conditions may result in more frequent treatments, transfers, and hospitalizations near the end of life, we excluded controls for comorbidities (cancer, lung or heart condition, diabetes, stroke, or psychiatric problems) to test whether results were driven by differences in underlying health.
Results
Sample Characteristics
Our sample of 742 community-dwelling decedents with cognitive impairment and/or dementia was primarily White (N = 577, 77.8%; Table 1), female (N = 390, 52.6%), and had less than a high school education (N = 377, 50.8%). Decedents were 82.7 years old (sd = 8.7 years) at death, on average, and were cognitively impaired for 7.7 years prior to death (sd = 5.1 years). More than one-third of decedents (287, 38.7%) died in the hospital, while 194 (26.1%) experienced a burdensome transfer and 233 (31.4%) experienced a life-sustaining treatment near the end of life.
Table 1:
Characteristics of Community-Dwelling Decedents with Cognitive Impairment or Dementia with End-of-Life Treatment Decisions (n = 742)a,b,c,d,e
| All Decedents (n = 742) | Only Children Participated (n = 484) | Children & Spouse Participated (n = 131) | Only Spouse Participated (n = 127) | |
|---|---|---|---|---|
|
| ||||
| No. (%) | No. (%) | No. (%) | No. (%) | |
| Died in Hospital | 287 (38.7) | 175 (36.2) | 58 (44.3) | 54 (42.5) |
| Burdensome Transfer | 194 (26.1) | 116 (24.0) | 38 (29.0) | 40 (31.5) |
| Life-Sustaining Treatment | 233 (31.4) | 142 (29.3) | 50 (38.2) | 41 (32.3) |
|
| ||||
| No. (%) | No. (%) | No. (%) | No. (%) | |
| Female | 390 (52.6) | 330 (68.2) | 29 (22.1) | 31 (24.4) |
| Veteran | 204 (27.5) | 86 (17.8) | 67 (51.1) | 51 (40.2) |
| Hispanic | 97 (13.1) | 64 (13.2) | 17 (13.0) | 16 (12.6) |
| Black | 140 (18.9) | 97 (20.0) | 23 (17.6) | 20 (15.7) |
| Other Race | 25 (3.4) | 16 (3.3) | **f | **f |
| < High School | 377 (50.8) | 252 (52.1) | 60 (45.8) | 65 (51.2) |
| High School | 186 (25.1) | 127 (26.2) | 34 (26.0) | 25 (19.7) |
| Some College | 109 (14.7) | 68 (14.0) | 20 (15.3) | 21 (16.5) |
| College-Plus | 70 (9.4) | 37 (7.6) | 17 (13.0) | 16 (12.6) |
| Married | 340 (45.8) | 87 (18.0) | -- | -- |
| Children w/in 10 Miles | 144 (19.4) | 74 (15.3) | 39 (29.8) | 31 (24.4) |
| Medicare Advantage | 164 (22.1) | 106 (21.9) | 36 (2 7.5) | 22 (17.3) |
| Cancer | 202 (27.2) | 123 (25.4) | 38 (29.0) | 41 (32.3) |
| Lung Condition | 148 (19.9) | 91 (18.8) | 26 (19.8) | 31 (24.4) |
| Diabetes | 244 (32.9) | 146 (30.2) | 49 (37.4) | 49 (38.6) |
| Heart Condition | 375 (50.5) | 235 (48.6) | 74 (56.5) | 66 (52.0) |
| Stroke | 214 (28.8) | 142 (29.3) | 37 (28.2) | 35 (27.6) |
| Mental Health Problems | 159 (21.4) | 104 (21.5) | 19 (14.5) | 36 (28.3) |
| Written Adv. Directive | 327 (44.1) | 210 (43.4) | 63 (48.1) | 54 (42.5) |
|
| ||||
| Mean (sd) | Mean (sd) | Mean (sd) | Mean (sd) | |
| Mean age, yr | 82.7 (8.7) | 84.9 (8.2) | 80.0 (7.9) | 77.3 (7.7) |
| Living Children | 3.6 (2.2) | 3.6 (2.1) | 3.6 (2.1) | 3.7 (2.5) |
| Yrs w/ CI or Dementia | 7.7 (5.1) | 8.2 (5.0) | 6.9 (5.4) | 6.7 (5.0) |
Community-dwelling HRS respondents who were 65 or older with CIND/dementia prior to death, with proxy-reported need for treatment decision at the end of life, and had a spouse or child reported as a treatment decision participant. Family member participation was identified from post-mortem HRS Exit Interviews.
Caregivers were spouses or children identified as both the decedent’s primary caregiver and a participant in EOL treatment decisions.
Data sources: Health and Retirement Study (HRS) core and Exit Interviews (1992–2016) and linked Medicare claims (2001–2015).
CI = cognitive impairment.
Adv. Directive = Advance Directive. Only includes ADs requesting limited care.
Our data use agreement requires masking cells containing small samples to avoid identifying individual study members.
Children and spouses were frequent participants in end-of-life (EOL) treatment decisions. Children participated in 615 (82.9%) decisions (more than 97% of decedents had at least one living child). Children were the only decision-makers for 484 (65.2%) decedents and made decisions jointly with spouses in 131 (17.7%) cases. Spouses participated in 258 decisions (127 as the sole decision-maker), 34.8% of all decisions and 74.4% of decisions for decedents married at their final report.
Prior to adjusting for decedent characteristics, 36.2 percent (N = 175) of decedents with only child decision-makers died in the hospital, compared to 44.3 percent (N = 58) for decedents with children and spouses and 42.5 percent (N = 54) for those with just spouse decision-makers (Table 1). Rates of burdensome transfers were 24.0 percent for decedents with only children (N = 116), 29.0 percent for decedents with children and spouses (N = 38), and 31.5 percent (N = 40) for decedents with only spouses participating. Burdensome treatments occurred for 29.3 percent (N = 142) of decedents with only child decision-makers, compared to 38.2 percent (N = 50) of decedents with children and spouse decision-makers and 32.3 percent (N = 41) of decedents with only spouse decision-makers.
Less than one-fourth of decedents with a spouse participating in decisions were female, compared to more than two-thirds of decedents with only child decision-makers. Relative to decedents with only child decision-makers, those with a spouse involved were younger on average by 4.9 years (children and spouse involved) and 7.6 years (only spouse involved) and were more likely to have at least a college degree. Race and rates of comorbidities were similar for decedents with different decision-maker teams.
Provision of a written advance directive requesting limited end-of-life care was similar for decedents with children or the spouse involved in decisions. 327 (44.1%) decedents had an AD requesting limited care, with 210 (43.4%) decedents with only child decision-makers, 63 (48.1%) decedents with spouse and child decision-makers, and 54 (42.5%) decedents with only spouse decision-makers providing written instructions requesting limited care.
Decision-Maker Relationships and EOL Care Outcomes
After adjusting for demographic characteristics and comorbid conditions, decedents were similarly likely to die in the hospital when only children participated in treatment decisions (38.6 percent; 95% CI 33.4–43.9) (Figure 2; Table 2), children and the spouse participated (39.4 percent; 95% CI 29.4–49.4) or the spouse participated without children (38.1 percent; 95% CI 28.3–48.0). Decedents were similarly likely to undergo a burdensome transfer with child decision-makers only (25.9 percent; 95% CI 21.1–30.7), child and spouse decision-makers (25.7 percent; 95% CI 17.1–34.3), or a spouse decision-maker only (27.4; 95% CI 18.7–36.2). In contrast, decedents with only a spouse participant were significantly less likely to receive life-sustaining treatments compared to those with only children making decisions (spouse only: 23.6 percent, 95% CI 16.1–31.0; children only: 34.6 percent, 95% CI 29.5–39.8; p<0.05). Decedents with both spouses and children participating were not significantly less likely to receive life-sustaining treatments than those with child decision-makers only (29.4 percent; 95% CI = 21.0–37.8).
Figure 2: End-of-Life Treatment Outcomes when Children and Spouses are Involved with Decision-Making for Patients with Cognitive Impairment or Dementiaa.
a Predicted probabilities (percentages, 95% CIs) from logistic regressions, adjusting for decedent demographics, family characteristics, whether enrolled in Medicare Advantage (MA), and comorbid health conditions. * p<0.05.
Table 2:
Percentage of Decedents Receiving Potentially Burdensome End-of-Life Care by Relationship to Decision-maker(s) for Patients with Cognitive Impairment or Dementiaa,b,c
| Died in Hospital | Burdensome Transfer | Life-Sustaining Treatment | N | |
|---|---|---|---|---|
|
| ||||
| Only Children Participated | 38.6 (33.4 – 43.9) |
25.9 (21.1 – 30.7) |
34.6 (29.5 – 39.8) |
484 |
| Children & Spouse Participated | 39.4 (29.4 – 49.4) |
25.7 (17.1 – 34.3) |
29.4 (21.0 – 37.8) |
131 |
| Only Spouse Participated | 38.1 (28.3 – 48.0) |
27.4 (18.7 – 36.2) |
23.6*
(16.1 – 31.0) |
127 |
Predicted probabilities (percentages, 95% CIs) from logistic regressions, adjusting for decedent demographics, family characteristics, whether enrolled in Medicare Advantage (MA), and comorbid health conditions.
p<0.05.
Spouses and children were identified as treatment decision participants from HRS Exit Interviews.
Data sources: Health and Retirement Study (HRS) core and Exit Interviews (1992–2016) and linked Medicare claims (2001–2015).
Table 3 and Figure 3 shows adjusted likelihoods of aggressive end-of-life care for decedents with dementia only, excluding decedents with CIND. Decision-maker relationship to the decedent was not associated with the likelihood of dying in the hospital or experiencing a burdensome transfer, but decedents with only spouse decision-makers were less-likely to experience potentially burdensome life-sustaining treatments compared to decedents with only child decision-makers (spouse only: 15.0 percent, 95% CI 6.7–23.2; children only: 33.5 percent, 95% CI 26.3–40.7; p<0.01). This difference is larger in magnitude than the difference found among all decedents with CIND or dementia. Decedents with spouse and child decision-makers were also less likely to receive life-sustaining treatments (16.2 percent, CI 6.8–25.5, p<0.05) than when only children were involved.
Table 3:
Percentage of Decedents Receiving Potentially Burdensome End-of-Life Care by Relationship to Decision-maker(s) for Patients with Dementiaa,b,c
| Died in Hospital | Burdensome Transfer | Life-Sustaining Treatment | N | |
|---|---|---|---|---|
|
| ||||
| Dementia Only | ||||
| Only Children Participated | 37.9 (30.1 – 45.6) |
25.5 (18.5 – 32.6) |
33.5 (26.3 – 40.7) |
224 |
| Children & Spouse Participated | 29.7 (15.3 – 44.0) |
19.5 (8.3 – 30.8) |
16.2*
(6.8 – 25.5) |
52 |
| Only Spouse Participated | 34.2 (20.0 – 48.3) |
27.4 (14.7 – 40.0) |
15.0** (6.7 – 23.2) |
59 |
Predicted probabilities (percentages, 95% CIs) from logistic regressions, adjusting for decedent demographics, family characteristics, whether enrolled in Medicare Advantage (MA), and comorbid health conditions.
p<0.05.
Spouses and children were identified as treatment decision participants from HRS Exit Interviews.
Data sources: Health and Retirement Study (HRS) core and Exit Interviews (1992–2016) and linked Medicare claims (2001–2015).
Figure 3: End-of-Life Treatment Outcomes when Children and Spouses are Involved with Decision-Making for Patients with Dementiaa.
a Predicted probabilities (percentages, 95% CIs) from logistic regressions, adjusting for decedent demographics, family characteristics, whether enrolled in Medicare Advantage (MA), and comorbid health conditions. * p<0.05, ** p<0.01.
We weighted our sample to more closely represent the US population using HRS sampling weights. Our sample of 742 decedents represented 2,363,399 adult Americans (see Table A1). Estimates were very similar, with adjusted means and confidence intervals within 1–2 percentage-points of those presented in Table 2. Once again, we did not observe significant differences by decision-maker team in likelihood of dying in the hospital or undergoing a burdensome transfer, but decedents with only spouse decision-makers were less likely to receive life-sustaining treatment compared to decedents with only child decision-makers.
To account for the possibility that regional differences in end-of-life treatment practices were driving results, we added an additional control for the decedent’s Hospital Referral Region quartile ranking of end-of-life spending intensity. Adjusted means and confidence intervals were similar to the main results, with spouse decision-makers associated with less-frequent receipt of life-sustaining treatment but not with differences in the likelihood of in-hospital death or any burdensome transfer.
We also tested the sensitivity of our dementia-only estimates (Table A2). Results were robust to adding controls for regional end-of-life spending intensity and removing controls for comorbidities. When using HRS sampling weights (weighted N = 1,072,170 American adults), the difference in likelihood of life-sustaining treatment between decedents with only children vs only a spouse decision-maker was comparable in magnitude but no longer statistically significant (spouse decision-maker: 17.0 percent, 95% CI 7.3 – 26.8; children decision-makers: 31.2 percent, 95% CI 23.0 – 39.3).
Discussion
In this paper, we examined whether end-of-life decisions for community-dwelling decedents with cognitive impairment varied across types of decision-makers. We examined three important indicators of end-of-care intensity; namely, death in the hospital, burdensome transfer, and life-sustaining treatment. After adjusting for patient characteristics such as age and medical conditions, we found that decedents were less likely to receive life-sustaining treatment prior to death when a spouse was the sole decision-maker rather than the decedent’s children, but there was no association between spouse or child involvement and in-hospital death or burdensome transfers. The lower rate of life-sustaining treatments with only a spouse decision-maker was larger in magnitude among decedents with dementia, for whom these treatments are often perceived as more burdensome, compared to rates among patients with all forms of cognitive impairment. While children were more likely to be involved in decisions where life-sustaining treatments were provided, even in settings where it was potentially inappropriate, we did not find consistent evidence that child and spouse involvement lead to different end-of-life decisions when considering all of our outcomes together.
We observed a statistically significant difference in the rate of life-sustaining treatments for patients with only spouse and only child surrogates (11 percentage-points, p<0.05). The end-of-life treatments we study are often painful for decedents, stressful for family members, and expensive for providers8,32–34. There are several potential explanations for the higher rates of life-sustaining treatments observed when children are the sole decision-makers. Since our sample is community-dwelling, spouses were generally living together, increasing the likelihood that the surrogate decision-maker spouse was available to provide medical history or treatment preference information at the time of hospital admission. Acutely ill patients may be given life-sustaining treatments before cognitive impairment or dementia is even recognized if the patient comes to the hospital alone or while the medical team is waiting for a child surrogate to arrive. Since we found no differences in rates of burdensome transfer or in-hospital death, processes which typically unfold over a longer time horizon, we cannot rule out similar decision-making styles among children and spouses. As hospitals and health systems improve electronic medical records and policy efforts like POLST and MOLST forms allow families and providers better access to patients’ needs and wishes, we may see further convergence between child and spousal decisions over time.
Alternatively, spouses and children may have different views about whether life-sustaining treatments are beneficial or consistent with the patients’ wishes. This creates a need for patients to prepare their surrogate decision-makers with information about the type of care they would prefer to receive in various hypothetical situations while still cognitively normal. The only information we have about patients’ treatment preferences, whether they wrote an advance directive indicating a desire to limit treatments in certain situations, was similar across surrogate relationships, suggesting that the differences in treatments may not reflect differences in patient preferences across groups. However, we are unable to determine whether children or spouses are more likely to make choices that are consistent with patient preferences, especially for patients with mild cognitive impairment who may experience greater benefit from life-sustaining treatments and are more able to accurately express treatment preferences11,
While our study represents one of the largest assessments of decedents and their surrogates, results should be interpreted in light of a few limitations. We used reports from proxy informants to identify decedents who required end-of-life treatment decisions. Although virtually all proxy respondents were involved in EOL decision-making, they may not be aware of all treatment decisions made or may have difficulties with accurate recall. However, prior work has found that decedents with dementia whose proxies reported a need for treatment decisions frequently experienced health events such as hospital admissions via ambulance or the emergency department35. Furthermore, we focused on patients who remained in the community, who have been found to be at higher risk of potentially burdensome end-of-life care16 and are also a minority of patients with dementia in our sample (742 of 1,851, 40 percent), so these results may not generalize to the experiences of nursing home patients. Since we required study participants to exhibit cognitive impairment on an assessment taken at their last core interview, we likely excluded some decedents who were cognitively impaired and required surrogate decision-making during their final days. Since questions about surrogate participation are only asked if the proxy reports that decisions were necessary, we are unable to include any decedents whose proxies were unaware of decision-making. While our unweighted sample included only 742 decedents, they represent more than 1 million older Americans and results were similar when we applied sample weights. Our findings are descriptive associations and do not have a causal interpretation.
To our knowledge, this is the first work to directly describe associations between patient-surrogate relationships and end-of-life treatments identified in medical claims. We used the largest available nationally representative sample with standardized assessments of cognitive function and linked claims data. The vast majority of HRS exit interviews were completed by next-of-kin who had participated in the treatment decisions; their reports are believed to accurately reflect care at the time of death17,31,36.
In conclusion, in a nationally representative sample of community-dwelling older adults who died with cognitive impairment or dementia, we observed elevated rates of life-sustaining treatments among decedents with children rather than the spouse acting as surrogate decision-makers. However, we did not find a systematic association between patients’ relationship to their surrogate decision-makers and receipt of high-intensity care, measured by in-hospital deaths and burdensome transfers in addition to treatments, near the end of life. Patients choosing healthcare proxies and providers counseling families should not necessarily expect biased decision-making based on these factors, though children may require additional information about the potential benefits and harms of life-sustaining treatments.
Supplementary Material
Key Message:
End-of-life treatment decisions did not differ systematically for community-dwelling, cognitively impaired decedents when children versus spouses were surrogate decision-makers, although life-sustaining treatments were more common when children were solely responsible for decision-making.
Disclosures and Acknowledgements
Ethics approval and consent to participate: Because this study used existing de-identified secondary data from deceased respondents, it was considered exempt by the Johns Hopkins Bloomberg School of Public Health’s Institutional Review Board.
Availability of data and materials: We used data from the Health and Retirement Study linked to Medicare claims. These data can be obtained through a data use agreement with the Health and Retirement Study and the Centers for Medicare and Medicaid Services as described at https://hrs.isr.umich.edu/data-products/restricted-data/cms-data.
Funding: This study was funded by the National Institute on Aging (R01AG059205).
Footnotes
Conflicts of interest: No competing interests are declared.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Plassman BL, Langa KM, Fisher GG, et al. Prevalence of Cognitive Impairment without Dementia in the United States. Annals of Internal Medicine. 2008;148(6):427–434. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Plassman BL, Langa KM, Fisher GG, et al. Prevalence of dementia in the United States: the aging, demographics, and memory study. Neuroepidemiology. 2007;29(1–2):125–132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Bynum JP, Rabins PV, Weller W, Niefeld M, Anderson GF, Wu AW. The relationship between a dementia diagnosis, chronic illness, medicare expenditures, and hospital use. J Am Geriatr Soc. 2004;52(2):187–194. [DOI] [PubMed] [Google Scholar]
- 4.Lopez OL. The growing burden of Alzheimer's disease. Am J Manag Care. 2011;17 Suppl 13:S339–345. [PubMed] [Google Scholar]
- 5.Feng Z, Coots LA, Kaganova Y, Wiener JM. Hospital And ED Use Among Medicare Beneficiaries With Dementia Varies By Setting And Proximity To Death. Health Affairs. 2014;33(4):683–690. [DOI] [PubMed] [Google Scholar]
- 6.Gozalo P, Teno JM, Mitchell SL, et al. End-of-Life Transitions among Nursing Home Residents with Cognitive Issues. New England Journal of Medicine. 2011;365(13):1212–1221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Mitchell SL, Morris JN, Park PS, Fries BE. Terminal care for persons with advanced dementia in the nursing home and home care settings. Journal of palliative medicine. 2004;7(6):808–816. [DOI] [PubMed] [Google Scholar]
- 8.Mitchell SL, Teno JM, Kiely DK, et al. The Clinical Course of Advanced Dementia. New England Journal of Medicine. 2009;361(16):1529–1538. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Teno JM, Gozalo PL, Mitchell SL, et al. Does feeding tube insertion and its timing improve survival? J Am Geriatr Soc. 2012;60(10):1918–1921. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Teno JM, Mitchell SL, Skinner J, et al. Churning: the association between health care transitions and feeding tube insertion for nursing home residents with advanced cognitive impairment. Journal of palliative medicine. 2009;12(4):359–362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Levine DA, Galecki AT, Plassman BL, et al. The Association Between Mild Cognitive Impairment Diagnosis and Patient Treatment Preferences: a Survey of Older Adults. Journal of general internal medicine. 2021:1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Rabins PV, Hicks KL, Black BS. Medical Decisions Made by Surrogates for Persons with Advanced Dementia within Weeks or Months of Death. AJOB primary research. 2011;2(4):61–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Black BS, Fogarty LA, Phillips H, et al. Surrogate Decision Makers' Understanding of Dementia Patients' Prior Wishes for End-of-Life Care. Journal of Aging and Health. 2009;21(4):627–650. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ayalon L, Bachner YG, Dwolatzky T, Heinik J. Preferences for end-of-life treatment: concordance between older adults with dementia or mild cognitive impairment and their spouses. International Psychogeriatrics. 2012;24(11):1798–1804. [DOI] [PubMed] [Google Scholar]
- 15.Reamy AM, Kim K, Zarit SH, Whitlatch CJ. Values and preferences of individuals with dementia: perceptions of family caregivers over time. The Gerontologist. 2013;53(2):293–302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Nicholas LH, Bynum JPW, Iwashyna TJ, Weir DR, Langa KM. Advance Directives And Nursing Home Stays Associated With Less Aggressive End-Of-Life Care For Patients With Severe Dementia. Health Affairs. 2014;33(4):667–674. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Silveira MJ, Kim SYH, Langa KM. Advance directives and outcomes of surrogate decision making before death. New England Journal of Medicine. 2010;362(13):1211–1218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bischoff KE, Sudore R, Miao Y, Boscardin WJ, SmithA K. Advance care planning and the quality of end- of- life care in older adults. Journal of the American Geriatrics Society. 2013;61(2):209–214. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Hickman SE, Nelson CA, Moss AH, Tolle SW, Perrin NA, Hammes BJ. The consistency between treatments provided to nursing facility residents and orders on the physician orders for life- sustaining treatment form. Journal of the American Geriatrics Society. 2011;59(11):2091–2099. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Hickman SE, Hammes BJ, Moss AH, Tolle SW. Hope for the future: achieving the original intent of advance directives. The Hastings Center Report. 2005;35(6):S26–S30. [DOI] [PubMed] [Google Scholar]
- 21.Tamura MK, Montez-Rath ME, Hall YN, Katz R, O’Hare AM. Advance directives and end-of-life care among nursing home residents receiving maintenance dialysis. Clinical Journal of the American Society of Nephrology. 2017;12(3):435–442. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Tolle SW, Tilden VP, Nelson CA, Dunn PM. A prospective study of the efficacy of the physician order form for life- sustaining treatment. Journal of the American Geriatrics Society. 1998;46(9):1097–1102. [DOI] [PubMed] [Google Scholar]
- 23.Detering KM, Hancock AD, Reade MC, Silvester W. The impact of advance care planning on end of life care in elderly patients: randomised controlled trial. Bmj. 2010;340. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Mitchell SL. A 93-year-old man with advanced dementia and eating problems. JAMA: The Journal of the American Medical Association. 2007;298(21):2527–2536. [DOI] [PubMed] [Google Scholar]
- 25.Juster FT, Suzman R. An Overview of the Health and Retirement Study. The Journal of Human Resources. 1995;30:S7–S56. [Google Scholar]
- 26.Sonnega A, Faul JD, Ofstedal MB, Langa KM, Phillips JWR, Weir DR. Cohort Profile: the Health and Retirement Study (HRS). International Journal of Epidemiology. 2014;43(2):576–585. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Teno JM, Gozalo P, Trivedi AN, et al. Site of death, place of care, and health care transitions among US Medicare beneficiaries, 2000–2015. Jama. 2018;320(3):264–271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Barnato AE, Farrell MH, Chang C-CH, Lave JR, Roberts MS, Angus DC. Development and validation of hospital “end-of-life” treatment intensity measures. Medical care. 2009;47(10):1098. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Crimmins EM, Kim JK, Langa KM, Weir DR. Assessment of cognition using surveys and neuropsychological assessment: the Health and Retirement Study and the Aging, Demographics, and Memory Study. Journals of Gerontology Series B: Psychological Sciences and Social Sciences. 2011;66(suppl_1):i162–i171. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Langa KM, Plassman BL, Wallace RB, et al. The Aging, Demographics, and Memory Study: study design and methods. Neuroepidemiology. 2005;25(4):181–191. [DOI] [PubMed] [Google Scholar]
- 31.Nicholas LH, Langa KM, Iwashyna TJ, Weir DR. Regional variation in the association between advance directives and end-of-life Medicare expenditures. Jama. 2011;306(13):1447–1453. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Teno JM, Mitchell SL, Kuo SK, et al. Decision-making and outcomes of feeding tube insertion: a five- state study. Journal of the American Geriatrics Society. 2011;59(5):881–886. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Teno JM, Freedman VA, Kasper JD, Gozalo P, Mor V. Is care for the dying improving in the United States? Journal of palliative medicine. 2015;18(8):662–666. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Hurd MD, Martorell P, Delavande A, Mullen KJ, Langa KM. Monetary Costs of Dementia in the United States. New England Journal of Medicine. 2013;368(14):1326–1334. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Nicholas LH, Halpern SD, Weir DR, et al. Surrogate Decision-Making for Patients with Severe Dementia Near the End-of-Life. Working Paper. 2021.
- 36.Kelley AS, Ettner SL, Morrison RS, Du Q, Wenger NS, Sarkisian CA. Determinants of Medical Expenditures in the Last 6 Months of Life. Annals of Internal Medicine. 2011;154(4):235–242. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.