Billed Advance Care Planning Conversations by a Growing Palliative Care Team: Insights from a Five-Year Review

Key Message

Along with palliative care team expansion over five years came a brisk uptrend in billed advance care planning (ACP) conversations. Findings suggested possible differences in ACP for certain patient groups. Future efforts could focus on overcoming disparities in access and shedding light on ACP’s ability to fulfill its intended aims.

Introduction

Advance care planning (ACP) is designed to align medical care with people’s values, goals, and preferences. ¹ The process intends to play a key role in upholding goal-concordant care, particularly toward the end of life. ACP encompasses such interventions as sharing disease- and care-related information with patients, caregivers, and other health professionals; initial and periodic conversations regarding patients’ goals; and completion of advance directive forms (e.g., living will, health care proxy, and Do Not Resuscitate order). Considerable research efforts have been dedicated to determine ACP’s efficacy. While some studies demonstrate the benefits of ACP, including increased documentation of end-of-life preferences, increased utilization of hospice care, and less aggressive care near death,^2–5 newer analysis, and expert opinion assert that ACP has either not achieved or underdelivered on its intended outcomes.^6–8

In 2015, the Center for Medicare and Medicaid Services (CMS) introduced two Current Procedural Terminology (CPT) codes allowing billing for time spent on ACP, defined as a “service between a Medicare physician (or other qualified health care professional) and a patient and/or family member(s), and/or surrogate, to discuss the patient’s healthcare wishes.” ⁹ Clinicians can use “99497” for the first 30 minutes of discussion and “99498” for each additional 30 minutes. The codes can be used in various care settings, at various stages of health, and multiple times per year. By design, the ACP codes reflect the amount of time spent on related elements and do not require any specific content.

Clinicians in Cleveland Clinic’s Department of Palliative and Supportive Care (DPSC) care for people with serious illness within the framework of a large health care system. Interprofessional palliative care (PC) clinicians see patients at all stages of illness in clinic, community, and hospital settings. From 2017 to 2021, DPSC expanded from the institution’s main academic campus, increasing its clinical workforce and launching inpatient PC consultation services at regional hospitals ranging from less than 100 to over 400 beds in capacity.

It is not yet clear how PC teams are incorporating the ACP codes into practice. This study aims to describe how a PC team within a large health care system integrated ACP billing into practice, including who received ACP conversations, where they were received, and how the codes were utilized.

Methods

All ACP conversations billed by DPSC clinicians from January 1, 2017 to December 31, 2021 were extracted from the institution’s central billing database for analysis. Information on unique patients was obtained from electronic health records. The study was approved by Cleveland Clinic’s Institutional Review Board.

Demographic information, dates of ACP conversations, and locations of service (i.e., hospitals, clinics, nursing facilities, and private homes) were recorded. The primary diagnosis was recorded using the tenth revision of the International Classification of Diseases (ICD-10) code designated for the clinical encounter where ACP occurred. A clinical encounter could indicate a hospital admission, an outpatient appointment, or a visit by a clinician to a patient’s private home or nursing facility. Multiple billed ACP conversations could take place over the course of a single clinical encounter, usually in the hospital setting. Hospitalized patients received care at Cleveland Clinic’s main academic campus or at any of its five regional hospitals. Patients’ primary address zip code at the time of ACP conversations and whether they resided within community outreach zones (COZ) were recorded. Cleveland Clinic has outlined COZ within the Greater Cleveland, Ohio area, which are zip code clusters known for health care underutilization. Information on PC clinicians was obtained from DPSC staffing records.

Patient-related information and data on billed ACP conversations were analyzed using descriptive statistics and summarized using frequencies and percentages. Of note, a single patient could have multiple clinical encounters, multiple billed ACP conversations, different medical insurance types, and different primary address zip codes during the five-year study period. To describe the trend in billed conversations using 2017 as baseline, the percentage change was calculated using the following formula: ([total ACP in year other than 2017 – total ACP in 2017]/total ACP in 2017) × 100. Similarly, the percentage change in PC clinicians was calculated by: ([total clinicians in year other than 2017 – total clinicians in 2017]/total clinicians in 2017) × 100. Statistical analysis was carried out using R version 4.2 (R Foundation, Vienna, Austria).

Results

A total of 17,091 billed ACP conversations with 10,307 unique patients were analyzed. As seen in Table 1, the majority of patients were female (52%), White (74%), and non-Hispanic (96%). Most patients at the time of conversations lived in-state (98%); 32% of whom resided in COZ. Most patients either had more than one medical insurance type (46%) or Medicare alone (40%). As shown in Table 2, conversations mostly occurred in the hospital setting (81%), in regional hospitals (86%), and on regular nursing floors (71%). The top five ICD-10 codes associated with ACP conversations included R codes (symptoms, signs, and abnormal clinical and lab findings), circulatory system, respiratory system, neoplasms, and Z codes (factors influencing health status and contact with health services).

A single 30-minute code (99497) was used in 13,910 (81%) billed conversations. Among the patients involved in those conversations, 52% (7262) were female, 74% (10241) were White, 96% (13396) were non-Hispanic, and 32% (4390) lived within COZ. Looking at the 3181 (19%) conversations, which used an additional 30-minute code (99498), 50% (1605) were female, 74% (2345) were White, 96% (3048) were non-Hispanic, and 30% (960) lived within COZ. Similar to all billed ACP conversations, the top ICD codes associated with conversations that used a single 30-minute code and those that used an additional 30-minute code were R code, circulatory system, respiratory system, neoplasms, and Z code (see Supplementary Table S1).

The total number of billed ACP conversations increased from 381 in 2017 to 5718 in 2021, whereas the total number of PC clinicians increased from 27 in 2017 to 41 in 2021 (see Table 3 and Fig. 1). Using 2017 as a baseline, Figure 2 shows that as the percentage of clinicians increased, the percentage of conversations also increased proportionally. Of note, as the number of hospital-based conversations steadily increased, those in clinics, private homes, and nursing facilities dropped in 2020 and 2021. In those two years, conversations completed utilizing telehealth (i.e., via telephone or videoconference/virtual visit) rose.

OPEN IN VIEWER

FIG. 1.

Billed advance care planning (ACP) conversations and palliative care (PC) clinicians over five years (logarithmic scale).

OPEN IN VIEWER

FIG. 2.

Correlation between percentage increase in palliative care clinicians and percentage increase in billed advance care planning conversations over five years.

Discussion

Looking at five years of billed ACP conversations, we saw that the majority of patients were female, White, and non-Hispanic. About a third of in-state patients lived in COZ. We observed a rise in ACP conversations year-over-year corresponding with the growth of our PC team. Most of the conversations took place in the hospital and stayed within a 30-minute time frame. ACP outside of the hospital setting decreased in 2020 and 2021. We surmise that this finding resulted from constraints imposed by the COVID-19 pandemic and coincided with the uptrend in telehealth conversations during those years.

ACP, in both terminology and practice, is a complex process influenced by motivation to initiate the conversation on the part of patients, families, and clinicians, by patients’ health trajectory, and by disparities in age, education, income, and race.^10–12 Findings of billed conversations mostly involved patients identified as female, White, and non-Hispanic, and a mere third of in-state patients resided in areas of health care underutilization, corresponded with results from prior studies that suggest the need to improve access for patients from underrepresented or marginalized communities.

Alongside the disparity in accessing ACP comes the equally vital issue of its efficacy. Using an extensive literature analysis, one study identified the underlying goals of ACP, which were respecting individual patient autonomy, improving quality of care, strengthening relationships, preparing for end-of-life, and reducing overtreatment. ¹³ To understand whether ACP is truly accomplishing these goals, we must realize the intrinsic complexity in predetermining one’s end-of-life care wishes and seeing them carried out by the designated decision-maker and medical team. Experts pointed out that successful ACP relies on multiple parts needing to fit seamlessly: patients can articulate their values and goals and how these might apply in hypothetical future scenarios, clinicians can elicit these values and goals, preferences are documented, directives or surrogates are available when patients lose decisional capacity, surrogates will decide the way patients would, clinicians will read prior documents and integrate patient preferences into conversations with surrogates, previously expressed wishes will be honored, and health care systems will commit resources and care delivery to support goal-concordant care. ¹⁴ As specialty PC clinicians, we affirm the potential value for ACP in advancing the aim of goal-concordant care and view this as a core part of any PC skillset. We also assert the importance of distinguishing “narrow” ACP conversations (such as determining a health care proxy) from broader “serious illness conversations” conducted by PC specialists to align medically appropriate care with each patient’s unique preferences. Differentiation of these different services is important, both for effective clinical practice and for future investigation of the impact of these conversations. Certainly, the utilization of CMS’s ACP billing codes contributes to the sustainability of these conversations.

As a final point, the volume of billed ACP conversations across our large health care system increased in alignment with the DPSC’s geographic expansion and staffing growth. While this trend seemingly demonstrates an increase in ACP in our system, this growth may be the result of increased specialty PC resources and increased awareness and use of ACP codes. It is important to note that billed conversations by non-DPSC clinicians were not included in this analysis. Future studies about ACP should assess which PC and non-PC clinicians are currently doing ACP conversations, as well as which clinicians may be best suited, and responsible, for conducting future ACP conversations. This knowledge may help design more impactful interventions while maximizing clinicians’ “top of license” skillsets.

The study was limited by its observational, retrospective, and single-center design. Future studies could include a control group to more closely examine factors related to efficacy and access disparity. Furthermore, the ambiguity and inconsistency around the content of conversations measured by ACP billing codes impact one’s ability to draw correlates and interpret outcomes. It is not difficult to imagine how two encounters, both billed as “30-minute ACP conversations,” can reflect different services. Future ACP studies should clearly define the specific ACP conversations involved.

Our findings depicted the integration of ACP billing into the practice of a large PC team over a five-year period. The brisk rise in billed ACP conversations was likely due to increased awareness and use of CPT codes, as well as expanding PC resources. We also observed possible differences in access to ACP for certain groups of patients. Future efforts—instituting policy changes, pursuing rigorous research, or developing alternative approaches—could focus on overcoming disparities in access and ascertaining ACP’s ability to fulfill its intended aims.