Abstract
This study uses Medicare claims data to identify and assess disparities in the use of advance care planning among adults 65 years or older who died between 2000 and 2015 and received intensive care during the last 30 days of life.
Approximately 30% of adults older than 65 years are treated in an intensive care unit (ICU) during the last month of life.1 Advance care planning (ACP)—a process that involves documenting wishes in an advance directive, appointing a surrogate decision maker, and having conversations about values, goals, and preferences2—can give such persons more control over their care.3 By contrast, those without ACP risk receiving unwanted, high-intensity, lower-quality care.4 This can lead to individual harms, including pain and suffering, and family harms, including psychosocial and financial distress.5 We therefore examined ACP completion among older adults treated in an ICU during their last month of life to determine the prevalence and factors associated with no ACP.
Methods
Study participants were derived from the Health and Retirement Study (HRS), a population-based cohort of older Americans with linked Medicare claims. Participants in the HRS are interviewed every 2 years, and, on dying, researchers contact next of kin to discuss the circumstances of death, decision-making, and other factors.6
We included HRS participants who were 65 years or older; were enrolled in fee-for-service Medicare; died between January 1, 2000, and December 31, 2015; and were discharged from an ICU during the last 30 days of life. We identified ICU admissions by using Medicare Provider Analysis and Review Research Data Assistance Center codes 200 to 202, 204, 206, 208 to 209, and 214 per previous methods.1 The HRS was approved by the institutional review board at the University of Michigan, Ann Arbor, and participants and their proxies verbally consented to participate. Our study was approved by the institutional review board at the University of California, San Francisco.
The primary outcome—no ACP—was defined from HRS surveys as providing no written instructions (ie, advance directive), no legally designated surrogate, and no discussions about treatment preferences (all reported by next of kin). We used multivariable logistic regression to assess whether demographic characteristics, net worth, year of death, chronic conditions, functional status, or prognostic awareness (death expected by next of kin) were associated with no ACP. We present estimated probabilities of no ACP for various subgroups and performed linear regression to assess time trends. Analyses were performed in Stata, version 15.1 (StataCorp) and SAS, version 9 (SAS Institute Inc) using HRS-provided survey weights. We present raw numbers and weighted percentages of respondents.
Results
Of 1730 participants (mean [SD] age at death, 81 [8.4] years; age range, 65.1-111.2 years; 954 [55.1%] women; 803 [46.4%] married; and 197 [10.3%] nursing home residents), 997 (57.6%) died during the index hospitalization and most (1296 [82.1%]) were non-Hispanic white (consistent with the US population older than 65 years). Next of kin completing after-death interviews were the participants’ adult children (887 [50.9%]), spouses (538 [31.3%]), siblings (56 [3.3%]), and other family members (75 [4.5%]) as well as individuals other than family (174 [10.1%]).
Between 2000 and 2015, 469 respondents (28.8%) had completed all 3 ACP components, 864 (50.4%) had completed 1 or 2 components, and 397 (20.7%) had completed no components. The unadjusted prevalence of no ACP decreased by 1.4% annually, from 31.3% in 2000 to 11.1% in 2015 (P < .001; Figure) owing to increases in each component of ACP. In multivariable analyses, minority race/ethnicity (eg, estimated prevalence for African American race/ethnicity vs non-Hispanic white, 36.3%; 95% CI, 29.2%-43.4% vs 15.7%; 95% CI, 13.4%-18.0%; P < .001) and lower net worth (for lowest [<$7000] vs highest [>$300 000] quartile of net worth, 25.7%; 95% CI, 20.5%-30.9% vs 13.0%; 95% CI, 9.2%-16.8%; P < .001) were significantly associated with no ACP. Older age, high school completion, nonmarried status, disability, and female sex were associated with higher rates of ACP (eg, for participants ≥85 vs 65-74 years, estimated prevalence was 15.9%; 95% CI, 12.5%-19.4% vs 22.1%; 95% CI, 17.6%-26.6%; P = .03). Chronic conditions and expectedness of death were not associated with ACP (Table).
Figure. Advance Care Planning (ACP) Completion Among Adults 65 Years or Older Who Were Treated in the Intensive Care Unit (ICU) During the Last Month of Life, 2000-2015.
Percentage of study participants treated in an ICU during the last month of life who had completed each component of ACP (sums to >100% because some participants had completed multiple components), those who had completed all 3 ACP components, and those who had completed no ACP components. Deaths are grouped into 2-year bins. We used survey weights provided by the Health and Retirement Study to account for the complex survey design. The rate of no ACP decreased by 1.4% per year during the study period (P < .001), while the percentage of decedents appointing a surrogate (1.5% increase per year; P = .01), having discussions about treatment (1.3% increase per year; P = .005), providing written instructions (1.1% increase per year; P = .01), or completing all 3 ACP components (0.9% increase per year; P = .049) all increased at similar rates.
Table. Estimated Prevalence of No Advance Care Planning for Various Subgroupsa.
| Characteristic | Participants, No. (%)b | Unadjusted Prevalence of No ACP, %c | Estimated Prevalence of No ACP (95% CI), % | P Valued | Global P Valuee |
|---|---|---|---|---|---|
| Age at death, y | |||||
| 65-74 | 459 (27.2) | 25.7 | 22.1 (17.6-26.6) | .08 | |
| 75-84 | 663 (38.6) | 20.4 | 18.1 (14.6-21.6) | .14 | |
| ≥85 | 608 (34.2) | 17.1 | 15.9 (12.5-19.4) | .03 | |
| Sex | |||||
| Male | 776 (44.9) | 22.0 | 20.9 (17.5-24.3) | ||
| Female | 954 (55.1) | 19.7 | 16.4 (13.4-19.3) | .05 | |
| Race/ethnicity | |||||
| Non-Hispanic white | 1296 (82.1) | 15.6 | 15.7 (13.4-18.0) | <.001 | |
| African American | 271 (10.5) | 45.3 | 36.3 (29.2-43.4) | <.001 | |
| Hispanic | 129 (5.5) | 45.0 | 33.1 (23.0-43.2) | <.001 | |
| Otherf | 34 (1.9) | 36.5 | 29.9 (12.6-47.1) | .05 | |
| Net worth | |||||
| Lowest quartile (<$7K) | 429 (21.9) | 31.8 | 25.7 (20.5-30.9) | <.001 | |
| Second quartile ($7K-$90K) | 434 (24.0) | 27.9 | 23.9 (18.4-29.0) | .56 | |
| Third quartile ($90K-$300K) | 434 (27.1) | 14.6 | 14.5 (10.8-18.1) | .001 | |
| Highest quartile (>$300K) | 433(27.0) | 11.6 | 13.0 (9.2-16.8) | <.001 | |
| Year of death | |||||
| 2000-2001 | 211 (11.0) | 31.3 | 25.4 (18.4-32.4) | .003 | |
| 2002-2003 | 252 (14.0) | 28.3 | 24.7 (18.7-30.7) | .88 | |
| 2004-2005 | 278 (15.2) | 25.7 | 21.9 (16.3-27.4) | .43 | |
| 2006-2007 | 222 (13.0) | 19.9 | 17.2 (11.2-23.1) | .08 | |
| 2008-2009 | 251 (14.5) | 16.2 | 14.2 (8.9-19.6) | .01 | |
| 2010-2011 | 244 (15.0) | 13.0 | 13.2 (8.7-17.7) | .003 | |
| 2012-2013 | 159 (10.2) | 17.1 | 17.3 (9.7-24.9) | .14 | |
| 2014-2015 | 113 (7.1) | 11.1 | 10.0 (3.9-16.1) | .005 | |
| Marital status | |||||
| Unmarried/unpartnered | 927 (53.6) | 19.3 | 14.6 (11.7-17.6) | ||
| Married/partnered | 803 (46.4) | 22.4 | 22.9 (18.8-26.9) | .003 | |
| Educational attainment | |||||
| Completed high school | 1053 (64.0) | 14.9 | 15.8 (13.2-18.4) | ||
| <High school | 677 (36.0) | 31.1 | 23.6 (19.1-28.0) | .002 | |
| Admitted from SNF | |||||
| No | 1523 (89.4) | 20.9 | 18.7 (16.3-21.1) | ||
| Yes | 197 (10.3) | 19.5 | 14.4 (8.9-19.9) | .19 | |
| Dementiag | |||||
| No | 1273 (76.2) | 19.2 | 17.4 (14.9-20.0) | ||
| Possible or probable | 436 (23.0) | 25.6 | 21.0 (15.9-26.2) | .21 | |
| Heart diseaseh | |||||
| No | 789 (45.5) | 21.9 | 19.4 (16.2-22.7) | ||
| Yes | 934 (54.0) | 19.8 | 17.4 (14.4-20.4) | .34 | |
| Chronic lung disease | |||||
| No | 1332 (77.0) | 20.4 | 17.8 (15.3-20.3) | ||
| Yes | 394 (22.8) | 21.7 | 19.7 (14.9-24.6) | .47 | |
| Cancer | |||||
| No | 1332 (76.7) | 21.3 | 18.2 (15.6-20.8) | ||
| Yes | 391 (22.9) | 18.5 | 18.3 (13.8-22.9) | .95 | |
| Stroke | |||||
| No | 1296 (75.3) | 21.3 | 19.3 (16.6-22.0) | ||
| Yes | 430 (24.5) | 19.0 | 15.2 (11.5-19.0) | .09 | |
| Count of chronic conditionsi | |||||
| 0 | 309 (19.1) | 19.6 | 17.9 (13.1-22.6) | .79 | |
| 1 | 616 (35.2) | 22.4 | 19.8 (16.1-23.5) | .53 | |
| 2 | 514 (28.9) | 20.3 | 17.3 (13.4-21.1) | .84 | |
| ≥3 | 291 (16.8) | 19.3 | 17.9 (12.8-22.9) | 1.0 | |
| Functional status | .02 | ||||
| Independent for all ADLs | 1177 (69.6) | 21.2 | 19.9 (17.1-22.7) | ||
| Dependent for ≥1 ADL | 553 (30.5) | 19.6 | 14.6 (11.1-18.1) | ||
| Expected deathj | |||||
| Yes | 905 (51.5) | 19.4 | 17.0 (14.1-19.5) | ||
| No | 778 (45.0) | 22.0 | 20.0 (16.6-23.4) | .17 | |
| Time to deathj | |||||
| <1 wk | 607 (34.9) | 18.9 | 16.2 (12.8-19.5) | ||
| ≥1 wk | 1092 (63.1) | 21.5 | 19.1 (16.3-22.0) | .18 |
Abbreviations: ACP, advance care planning; ADLs, activities of daily living; K, thousand; SNF, skilled nursing facility.
Estimated prevalences were calculated using the postestimation margins command following multivariable logistic regression analysis, wherein “no ACP” is a function of the predictor variable adjusted for age, sex, race/ethnicity, educational attainment, and net worth. In the case of age, sex, race/ethnicity, educational attainment, and net worth, each predictor is adjusted for the other 4 variables.
Percentages may not add to 100% owing to rounding, missingness, and the application of Health and Retirement Study–provided survey weights. Proportion missing was less than 5% for all variables.
Estimated using Health and Retirement Study survey weights to account for the complex survey design.
Compares results for the specific subgroup with the reference group on the basis of the multivariable model, with Health and Retirement Study survey weights applied.
Expresses significance across the subgroup categories.
Other included non-Hispanic American Indian, Asian, or something else.
Dementia determined by Glymour probability, with scores of 0.5 or higher considered indicative of dementia.
Heart disease, chronic lung disease, cancer, and stroke were considered present if reported in any preceding Health and Retirement Study core interview wave.
Count of heart disease, chronic lung disease, cancer, stroke, and dementia.
As reported by next of kin in exit interview.
Discussion
In this study of adults 65 years or older who received intensive care during the last 30 days of life, we found gains in ACP across all populations between 2000 and 2015. However, our findings suggest that ACP remains less common in vulnerable populations and that many people with chronic illness lack ACP. The observed increases in ACP are important because critically ill older adults are at imminent risk of receiving goal-discordant care owing to their high acuity of illness, the availability of life support technologies, and the need for urgent or emergent action. Study limitations included the use of next of kin report to determine whether ACP occurred and the focus on individuals admitted to the ICU, which may have selected for those without ACP.
References
- 1.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: 10.1001/jama.2018.8981 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Sudore RL, Lum HD, You JJ, et al. . Defining advance care planning for adults: A consensus definition from a multidisciplinary Delphi panel. J Pain Symptom Manage. 2017;53(5):821-832.e1. doi: 10.1016/j.jpainsymman.2016.12.331 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Silveira MJ, Kim SY, Langa KM. Advance directives and outcomes of surrogate decision making before death. N Engl J Med. 2010;362(13):1211-1218. doi: 10.1056/NEJMsa0907901 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bischoff KE, Sudore R, Miao Y, Boscardin WJ, Smith AK. Advance care planning and the quality of end-of-life care in older adults. J Am Geriatr Soc. 2013;61(2):209-214. doi: 10.1111/jgs.12105 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Khandelwal N, Hough CL, Downey L, et al. . Prevalence, risk factors, and outcomes of financial stress in survivors of critical illness. Crit Care Med. 2018;46(6):e530-e539. doi: 10.1097/CCM.0000000000003076 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.The Health and Retirement Study: an introduction. https://hrs.isr.umich.edu/sites/default/files/Intro-to-HRS_0.pdf. Accessed April 3, 2018.