Missed Opportunities and Health Disparities for Advance Care Planning Before Elective Surgery in Older Adults

Key Points

Question  What patient factors are associated with perioperative advance care planning (ACP) documentation among older surgical patients (65 years and older) at risk of postoperative complications?

Findings  In this cohort study, perioperative ACP rates were low among all groups; however, rates were lower among men and patients who recorded a non-English preferred language. Patients with Medicare were more likely to have ACP.

Meaning  The findings in this study indicate that older surgical patients tended not to have ACP documentation, a practice that should be encouraged in this population; moreover, when addressing ACP for surgical patients, particular attention should be paid to overcoming language-related disparities.

Importance  Advance care planning (ACP) prepares patients and caregivers for medical decision-making, yet it is underused in the perioperative surgical setting, particularly among older adults undergoing high-risk procedures who are at risk for postoperative complications. It is unknown what patient factors are associated with perioperative ACP documentation among older surgical patients.

Objective  To assess ACP documentation among high-risk patients 65 years and older undergoing elective surgery.

Design, Setting, and Participants  In this observational cohort study including 3671 patients 65 years and older undergoing elective surgery at a tertiary academic center in California, electronic health record data were linked to the National Surgical Quality Improvement Project outcomes data and the California statewide death registry. The study was conducted from January 1 to December 31, 2019. Data were analyzed from January to May 2022.

Exposures  Elective surgery requiring an inpatient admission.

Main Outcomes and Measures  ACP documentation, defined as a discussion regarding goals of care documented in an ACP note, an advance directive, or a physician order for life-sustaining treatment (POLST) form, within 90 days before elective surgery requiring inpatient admission. Multivariate regression was performed to identify factors associated with missing ACP.

Results  Among 3671 patients (median [IQR] age 72 [65-94] years; 1784 [48.6%] female; 401 [10.9%] Asian, 155 [4.2%] Black, 284 [7.7%] Latino/Latina, 2647 [72.1%] White, and 184 [5.0%] of other races or ethnicities, including American Indian or Alaska Native, Native Hawaiian or Pacific Islander, multiple races or ethnicities, other, and unknown or declined to respond, combined owing to small numbers), 539 (14.7%) had ACP documentation in the 90-day presurgery window. Of these 539, 448 (83.1%) had advance directives, and 60 (11.1%) had POLST forms. The 30-day and 1-year mortality were 0.7% (n = 27) and 6.6% (n = 244), respectively. Missing ACP was significantly associated with male sex (adjusted odds ratio [aOR], 1.39; 95% CI, 1.14-1.69) and having a non-English preferred language (aOR, 1.78; 95% CI, 1.18-2.79). Medicare insurance was significantly associated with having ACP (aOR for missing ACP, 0.63; 95% CI, 0.40-0.95).

Conclusions and Relevance  In this study, perioperative ACP was uncommon, particularly in men, individuals with a non-English preferred language, and those without Medicare insurance coverage. The perioperative setting may represent a missed opportunity for ACP for older surgical patients. When addressing ACP for surgical patients, particular attention should be paid to overcoming language-related disparities.

The goal of advance care planning (ACP) is to prepare patients and their surrogate decision-makers for medical decision making to help ensure patients receive medical care aligned with their goals and preferences.^1,2 Benefits of effective ACP discussions include increased patient satisfaction, reduction of goal-discordant care, fewer rehospitalizations, and increased use of hospice care when appropriate.^3-5

Nearly 20 million older adults undergo major elective surgical procedures annually, yet very few complete ACP.⁶ This gap is a missed opportunity to ensure optimal patient-aligned medical decisions and communications. Serious conditions leading to major surgery and any ensuing disease progression or complications can be associated with worsening health status after surgery, particularly in older adults. These deficits range from anticipated short-term physical and cognitive impairment to unanticipated major and long-term impairment or even death.⁷ In both preoperative and postoperative phases, patients and surrogate decision-makers must make complex decisions, but many are unprepared to articulate their individual care goals. As a result, older adults may undergo surgical and postsurgical treatments inconsistent with their values and goals.^8-10

Although ACP is broadly endorsed as an important quality metric in the field of surgery, real-world successes are lacking. Even in clinical trial environments, efforts to integrate ACP into surgical care have repeatedly failed.^11,12 National initiatives and surgical programs have endorsed ACP, including jointly developed guidelines from the American College of Surgeons and the American Geriatric Society.¹³ Similarly, the Centers for Medicare & Medicaid Services Innovation Center Programs, through their Bundled Payment Care Initiative Advanced Model, include an ACP measure as one component of the composite quality score used to calculate incentive payments. Despite strong evidence and financial incentives, ACP use in surgery patients remains low.^1,14 Even individuals at high risk who are undergoing major surgery have limited awareness of ACP and engagement in its completion.

Therefore, as a first step toward developing a pragmatic framework for improving ACP in surgery patients, we aimed to describe the prevalence of preoperative ACP as well as patient- and procedure-level characteristics associated with ACP completion in a large cohort of patients 65 years and older undergoing elective surgery requiring inpatient admission. To our knowledge, this analysis represents the largest cohort of older adult surgical patients with real-world data in whom rates and predictors of ACP have been evaluated.

This was a retrospective cohort study at a single tertiary academic medical center conducted using data from the electronic health record (EHR), the American College of Surgeons National Surgical Quality Improvement Program (NSQIP), and the California state death registry. The study was approved by the University of California, San Francisco, institutional review board.

All patients 65 years and older who underwent elective surgery with a planned inpatient overnight stay of 1 or more nights through the general surgery, surgical oncology, cardiac, vascular, plastic, abdominal transplant, orthopedic, neurosurgery, urology, and otolaryngology services at the University of California San Francisco Health hospitals were included. Patients having outpatient, unplanned, or emergency surgery (as designated by the operating room log), American Society of Anesthesiologist classification (ASA) V or VI, those with an E suffix (eg, IIIE), or those with a tracheostomy were excluded from analysis. To better understand ACP practices in abdominal transplant patients, a separate analysis was conducted.

Only patients undergoing elective surgery with an inpatient stay in 2019 were included because surgical services were disrupted in 2020 in response to the COVID-19 pandemic and at least 1 year of follow-up was required for the mortality data to be valid.

Patient demographic characteristics and procedure-related information were obtained from the University of California, San Francisco, medical record. Derived variables developed by University of California San Francisco Health Informatics were used, in which patients self-identify race and ethnicity. Race and ethnicity were included because prior data from nonsurgical populations have shown differing ACP documentation rates among minoritized populations and we wanted to understand if such disparities exist among surgical patients. Categories for race were determined by University of California San Francisco Health and included American Indian/Alaska Native, African American/Black, Native Hawaiian/Other Pacific Islander, Asian, White, decline to answer, unknown, and other, with a space for write-in responses. Patients were asked to identify themselves as Hispanic/Latino or not (or to decline responding), identify as many race categories as appropriate, and select as many ethnicity categories (n = 34) as appropriate. To account for patient comorbidities, ASA classification and the Charlson Comorbidity Index were included. Preferred language was extracted from the EHR. To evaluate associations between social determinants of health and ACP completion, the neighborhood deprivation index was linked by zip code and a binary distinction was made between patients in the bottom quartile of the neighborhood deprivation index in this cohort vs those in other quartiles.

The primary outcome of interest was preoperative ACP documentation as identified in the EHR. This documentation included any of the following: documented ACP discussions in a defined ACP note or text bookended by ACP markers in the EHR, previously completed advanced directives scanned into the EHR, or physician orders for life-sustaining treatment (POLST) scanned into the EHR. The University of California, San Francisco, uses a preoperative anesthesia appointment in which the standard protocol includes instructing all patients to bring any existing advance directive or POLST documents on the day of surgery. The preoperative holding area scans documents brought on the day of surgery into the EHR. Other types of ACP, including free text documentation in clinical notes, may exist but were not readily identifiable and were therefore not included. Prior analyses indicated most elective surgery at the University of California, San Francisco, is scheduled within 3 months of initial consultation.¹⁵ Therefore, to focus on ACP in preparation for major surgery requiring an inpatient admission, only documentation entered in the 90-day window prior to surgery was included because older ACP was considered potentially outdated for the surgical indication.

We further evaluated ACP rates in the subset of patients included in the NSQIP registry. University of California San Francisco Health targets colectomy, proctectomy, pancreas, liver, cystectomy, and ventral hernia, representing approximately 1400 patients per year. NSQIP complication data was not available for all patients, such as cardiac and orthopedic surgery patients. In this subset, we conducted a separate analysis of those who incurred NSQIP-defined serious morbidity (ie, cardiac arrest, myocardial infarction, pneumonia, progressive kidney insufficiency, acute kidney failure, return to the operating room, deep incisional surgical site infection, organ/space surgical site infection, septic shock, unplanned intubation, urinary tract infection, or wound disruption), including those who died within 30 days or 1 year of surgery. We also evaluated the association of ACP with hospitalization-related data, including length of stay, postoperative intensive care unit admission, ventilator days (as defined by the US Centers for Disease Control National Healthcare Surveillance Network), and discharge disposition.

Descriptive statistics and multivariate logistic regression were performed using all available covariates. Analysis was conducted with R version 4.1.2 (R Foundation), considering P values less than .05 to be statistically significant.

Minimal data were missing from the EHR. However, the completeness of mortality data could not be confirmed beyond data available in the California death registry, which was treated as the criterion standard. Some patients may have died in other locations or circumstances that were not captured by the registry, resulting in an overall underestimate of mortality.

A total of 3671 patients 65 years and older underwent elective surgery requiring an inpatient admission in 2019 and were included in the analysis (Table 1). Of this cohort, 1784 patients (48.6%) were female, and the median (IQR) age at the time of surgery was 72 (65-94) years. According to self-reported race data, 401 patients (10.9%) were Asian, 155 (4.2%) were Black, 284 (7.7%) were Latino/Latina, 2647 (72.1%) were White, and 184 (5.0%) were of other races or ethnicities, including American Indian or Alaska Native, Native Hawaiian or Pacific Islander, multiple races or ethnicities, other, and unknown or declined to respond, combined owing to small numbers. Most patients listed English as their preferred language (3329 [90.7%]) and were insured through Medicare (3274 [89.2%]). Most patients underwent orthopedic procedures (1027 [28.0%]), followed by neurosurgical (727 [19.8%]), urologic (487 [13.3%]), general (638 [17.4%]), and otolaryngologic operations (258 [7.0%]). Abdominal transplant, plastic, cardiac, and vascular surgery services each accounted for about 200 patients. Overall, the population was high risk, with a median ASA of II and 1751 individuals (47.7%) having a Charlson Comorbidity Index score of 3 or greater.

Within 90 days prior to surgery, 539 patients (14.7%) had ACP documentation entered in the EHR. Of these, most of the documentation was advance directives (448 [83.1%]) or POLST forms (60 [11.1%]). Sixty patients (11.1%) had ACP notes with text bookended by functions in the EHR completed by a clinician prior to surgery (Figure 1). We qualitatively analyzed the content of most ACP notes that had been created using an EHR-specific note template. Nearly all of these notes (30 of 33 [90.1%]) were written by either a primary care professional or a social worker, likely as part of routine health management. None of the notes were written by a surgical professional. Five of the notes were written during an inpatient encounter, but most (28) were written on an outpatient basis. Most advance directives and POLST forms (376 of 508 [74.0%]) were uploaded to the EHR on the day of surgery, and it is unknown if they were created with surgery in mind or were preexisting.

Although the overall ACP rates were higher in patients who had more complicated hospitalizations, most patients who were admitted to the ICU (523 [81.5%]) or discharged to a facility (629 [82.4%]) did not have ACP completed prior to surgery (Table 2).

A multivariate logistic regression model for risk of missing preoperative ACP was developed using available preoperative data; ASA class was omitted for collinearity with Charlson Comorbidity Index (Table 3). Missing preoperative ACP was significantly associated with male sex (adjusted odds ratio [aOR], 1.39; 95% CI, 1.14-1.69; P < .001) and having a non-English preferred language (aOR, 1.78; 95% CI, 1.18-2.79; P = .009). Medicare insurance (aOR for missing ACP, 0.63; 95% CI, 0.40-0.95; P = .03) was significantly associated with having preoperative ACP.

For patients 65 years and older undergoing elective surgery requiring inpatient admission, both 90-day and 1-year mortality rates (2.3%; n = 83 and 6.6%; n = 244, respectively) were higher than the 30-day mortality rates (0.7%; n = 27) (Figure 2). Patients with ACP had lower 30-day mortality (0.7%) and higher 1-year (10.0%) mortality compared with those patients who lacked ACP (0.9% at 30 days and 6.1% at 1 year). In multivariate analysis of this subgroup, there were no significant predictors of missing ACP. Notably, of 1247 patients 75 years and older undergoing surgery, 1057 (84.7%) did not have ACP completed.

In the overall cohort, 326 patients (8.9%) had NSQIP-defined serious morbidity. Of these, 48 (14.7% of those with serious morbidity and 1.3% of the total sample) had completed preoperative ACP. When we stratified the cohort by actual length of stay, we found that among 1875 patients with an actual length of stay less than 3 days, 1-year mortality was 3.2% (n = 60) and 237 patients (12.6%) had preoperative ACP within 90 days vs 860 patients with an actual length of stay more than 5 days, who had a 1-year mortality of 14.7% (n = 126) and an ACP documentation rate of 17.4% (n = 150). Multivariate regression identified no significant predictors of missed ACP in this subgroup. We included the subset analysis of patients with NSQIP-defined severe morbidities to highlight that, in general, ACP adoption was not stratified by surgical factors.

Patients undergoing abdominal transplant surgery (elective and nonelective) had the third highest rate of preoperative ACP (11 of 69 [15.9%]) compared with other surgery services, with the highest rates among patients undergoing cardiac (17 of 97 [17.5%]) and vascular (41 of 246 [16.7%]) procedures. In univariate analysis, White patients, those with commercial insurance, and those with a Charlson Comorbidity Index score less than 3 were significantly more likely to have preoperative ACP, while those from more socioeconomically distressed neighborhoods were less likely to have preoperative ACP. In multivariate regression, Asian race (aOR, 2.88; 95% CI, 1.44-6.31; P = .005), Latino/Latina ethnicity (aOR, 2.46; 95% CI, 1.24-5.25; P = .01), and Charlson Comorbidity Index score of 3 or greater (aOR, 2.11; 95% CI, 1.19-3.72; P = .01) were significantly associated with missing ACP.

The 14.7% rate of ACP documentation in our study was low compared to previously published rates in other surgical populations, ranging from 22%¹⁶ to 38%,¹⁷ and compared to a general medical population (37%¹⁸), but is likely representative of other academic health systems with a large and complex referral network. Most of the ACP documentation in this study consisted of advance directives uploaded on the day of surgery, suggesting that much of the ACP in the EHR was not elicited by conversations with the surgical care team. The policy at the University of California, San Francisco, of instructing patients to bring any existing advance directive or POLST to be uploaded on the day of surgery likely accounts for the surge of ACP on the day of surgery. Therefore, there may be a missed opportunity to engage patients in ACP specific to their more acute surgical condition and treatment. Although rates of ACP documentation were low across all groups, male patients and those with a non-English preferred language were more likely to lack perioperative ACP.

The surgical team’s unique knowledge and experience should facilitate the discussion of ACP for older and higher-risk patients. Well-documented ACP conversations may guide postoperative care, and thoughtful discussions may prompt patients and their surrogates to consider their preferences for the future even if no postoperative issues arise. While the benefits of ACP have recently been the subject of vigorous debate,¹⁹ robust data continue to show that ACP is associated with decreased anxiety and depression, posttraumatic stress disorder, and complicated grief.²⁰ Concerns about the alleged shortcomings of ACP hinge on a static definition of ACP, whereas many practitioners endorse ACP as an iterative process, which prepares patients and their family and friends for decision-making in the future.

As we strive to provide equitable surgical care, it is important to acknowledge that patients’ preferred language and sex were associated with missing ACP. Patients with a non-English preferred language had a higher rate of missing ACP compared with patients for whom English was listed as the preferred language. These findings held true in patients undergoing abdominal transplant as well, an area with significant resource commitment to preoperative evaluation.²¹ It is likely the existing transplant workup, while thorough from a social and functional standpoint, does not typically assess or record patient goals and preferences. Further study is needed to understand and develop structured approaches to addressing inequities in ACP discussions and documentation. Although complex cultural differences in expectations for care planning may impact differential rates of ACP among racial and ethnic minority groups, prior work with Latino/Latina patients has shown that, although rates of ACP documentation were lower, patients were equally likely to want to discuss their quality of life and end-of-life care wishes with a health care professional.²² To work toward equity, we believe specifically targeted efforts to increase access to perioperative, language-concordant, culturally sensitive ACP need to be developed in a pragmatic but rigorous manner.

Another finding in our data were that patients with more comorbidities and higher ASA class had similar rates of ACP completion prior to surgery as their healthier counterparts. These data suggest that surgeons may rely on patient engagement with ACP rather than initiating ACP discussions with patients who are at greater risk of complications because of greater underlying disease burden. Our finding that rates of ACP documentation were similar among patients with and without serious postoperative morbidity also suggests there are many high-risk elective surgical patients who are not engaging in preoperative ACP. These results highlight that surgical teams may be missing opportunities to engage in ACP with patients who may be especially vulnerable to complications. ACP efforts in surgical patients may focus on those at greatest risk of mortality, or ACP can be routinely implemented to conduct a population-level intervention. We emphasize routine ACP, including identification of surrogates, preparation, and goal-concordant decision-making, because these discussions can benefit any patient with chronic or serious illness, including older patients undergoing a major elective operation. Even if patients are not currently in situations requiring ACP, many will encounter changes in health status within 1 year of surgery that may necessitate ACP discussions, and most will at some point in the future.

Our study also shows that patients with Medicare insurance were more likely than those with non-Medicare insurance to have ACP documentation in the EHR preoperatively. This may be related to the introduction of Medicare advance care planning Current Procedural Terminology codes to reimburse clinicians making ACP part of the Medicare annual wellness visit.²³ Further work is required to understand how these codes are used outside the context of primary care.²³

We also reported a 1-year mortality rate of 6.6%, which is similar to other data on 1-year all-cause mortality among post–elective surgery patients.^24,25 In line with other reports, the 30-day mortality of 0.7% was low but then increased over the subsequent postoperative year. A greater proportion of patients who died within the first 90 days lacked ACP documentation compared to those who died after 90 days. While those who died within the first 90 days may not be captured in typical 30-day mortality metrics, these are likely patients for whom surgery was a major turning point in their health trajectory, if not the proximate cause of mortality. These patients are likely to have benefitted from ACP discussions, including their goals and preferences, how the planned procedure might align with those goals and preferences, and identification of their surrogate decision-maker. The surgical indication should have prompted such verbal discussions with the surgical team, but there is no evidence any of these important conversations were recorded in an identifiable manner in the EHR. We believe that surgery is a key time during which patients reflect on their condition and goals in general and therefore can serve as a natural prompt for ACP.

Our study focused on ACP prior to surgery, which is just one component of a comprehensive approach to goal-concordant care. ACP is dynamic and should be integrated into routine postoperative surgical care and reconsidered upon transfer to a higher level of care or the onset of serious postsurgical morbidity. Ultimately, developing a surgical culture where ACP is considered dynamic is essential, but work is needed to identify how and when ACP documentation is used and updated by medical teams.

As we look to the future, we need to develop pragmatic, scalable approaches to integrating ACP into the presurgical setting, including strengthening transdisciplinary partnerships within surgical teams. Efforts to date have mostly focused on increasing patient activation to complete ACP documentation, but considerable barriers remain, including varying levels of familiarity and comfort conducting ACP conversations among the surgical team, lack of dedicated time during the presurgical care episode for these often-delicate conversations, and lack of appropriate patient-facing ACP tools to help patients and caregivers make complex decisions about their surgical treatment. PREPARE for Your Care,²⁶ an interactive ACP tool, is an example of a scalable approach that has been effective at increasing ACP completion in primary care,²⁷ yet further study is needed to establish its acceptability and efficacy in surgery patients.

This study has limitations. First, this is a retrospective cohort study at a single tertiary academic medical center, which limits generalizability. Second, while perhaps the most important metric regarding ACP is how it affects clinical trajectory and experience, our data did not include note content, so we were not able to speculate about the impact of the presence (or absence) of ACP documentation. Third, ACP documentation may exist elsewhere in the EHR that was not captured in our analysis, so rates reported may be underestimates of the prevalence of ACP.

In this study, adults 65 years and older who underwent elective surgery requiring an inpatient admission incurred significant 30-day morbidity and 1 year mortality, yet had low rates of ACP. Our study demonstrated inequities in perioperative ACP for male patients and those with a non-English preferred language. Future research in surgical ACP is needed to address efficacy and scalability for these essential components of care.

Accepted for Publication: May 30, 2022.

Published Online: August 24, 2022. doi:10.1001/jamasurg.2022.3687

Corresponding Author: Elizabeth Wick, MD, 513 Parnassus Ave, HSW 1601, San Francisco, CA 94143 (elizabeth.wick@ucsf.edu).

Author Contributions: Dr Wick had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Colley and Lin served as co–first authors, each with equal contribution to the manuscript.

Concept and design: Colley, Lin, Pierce, Sudore, Finlayson.

Acquisition, analysis, or interpretation of data: Colley, Lin, Pierce, Wick.

Drafting of the manuscript: Colley, Lin, Sudore, Wick.

Critical revision of the manuscript for important intellectual content: Colley, Lin, Pierce, Sudore, Finlayson.

Statistical analysis: Colley, Lin, Pierce.

Obtained funding: Colley.

Administrative, technical, or material support: Pierce, Wick.

Supervision: Lin, Pierce, Sudore, Wick.

Conflict of Interest Disclosures: Dr Lin reported grants from National Institutes of Health outside the submitted work. Dr Finlayson reported being the founder of Ooney, Inc, outside the submitted work. No other disclosures were reported.

Funding/Support: This study was supported by National Institutes of Health grants 5T32CA251070-02 (to Dr Colley) and AHRQ R01HS024532 (to Dr Wick) and National Institute on Aging grants R01AG058616 and R01AG067507 (to Dr Finlayson) and K24AG054415 (to Dr Sudore).

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Meeting Presentation: This article was presented at the Pacific Coast Surgical Association Meeting; February 13, 2022; Maui, Hawaii.

Additional Contributions: We thank Pamela Derish, MA, University of California, San Francisco, for her expert help in reviewing the manuscript for clarity and diction. No compensation was provided.