Oncology Fellow–Led Quality Improvement Project to Improve Rates of Palliative Care Utilization in Patients With Advanced Cancer

Abstract

PURPOSE:

ASCO guidelines recommend palliative care (PC) referral for patients with advanced or metastatic cancer. Despite this, implementation has considerable hurdles. First-year oncology fellows at our institution identified low rates of PC utilization in their longitudinal clinic as a metric needing improvement.

METHODS:

A fellow-led multidisciplinary team aimed to increase PC utilization for patients with advanced cancer followed in he first-year fellows’ clinic from a baseline of 11.5% (5 of 43 patients, July to December of 2018) to 30% over a 6-month period. Utilization was defined as evaluation in the outpatient PC clinic hosted in the cancer center. The team identified the following barriers to referral: orders difficult to find in the electronic medical record (EMR), multiple consulting mechanisms (EMR, by phone, or in person), EMR request not activating formal consult, no centralized scheduler to contact or confirm appointment, and poor awareness of team structure. Plan-Do-Study-Act (PDSA) cycles were implemented based on identified opportunities. Data were obtained from the EMR.

RESULTS:

The first PDSA cycle included focus groups with stakeholders, standardizing referral process via single order set, identifying a single scheduler with bidirectional communication, and disseminating process changes. PDSA cycles were implemented from January to June of 2019. Rates of PC use increased from 11.5% before the intervention to 48.4% (48 of 99 patients) after the intervention.

CONCLUSION:

A multidisciplinary approach and classic quality improvement methodology improved PC use in patients with advanced cancer. The pilot succeeded given the small number of fellows, buy-in from stakeholders, and institutional and leadership support. Straightforward EMR interventions and ancillary staff use are effective in addressing underreferrals.

INTRODUCTION

The benefits of outpatient palliative care (PC) in advanced cancer are supported by randomized trials.^1-8 PC has been shown to improve quality of life for patients and families by focusing on pain, symptoms, and functional limitations; sharing information about prognosis; assessing understanding of illness; clarifying goals of care; and providing support for the psychosocial and spiritual ailments of serious illness.⁹

Although PC can be used for any stage of cancer and is recommended early in the disease course,⁹ referrals are often delayed or not made at all.^10,11 This is unfortunate given the evidence that PC referral improves quality-related outcomes, including physical and emotional symptoms for both patients and caregivers, end-of-life acute care utilization, hospice use, intensive care unit stay, and costs of care.^12-15 Although a recommended list of indications for PC referral are included in ASCO guidelines,⁹ rates of outpatient specialty PC use are as low as 2.0%.^16-20 At Johns Hopkins (Baltimore, MD), despite having a well-established PC clinical service, the referral rate for patients with advanced or metastatic cancer is 25%.

Barriers to PC referral for patients with advanced cancer remain unclear. As oncologist’ knowledge of guidelines and understanding of PC improve,^21-23 external system-level factors may limit guideline adherence.²⁴ One possible factor is the colocation of oncology and PC clinics.^25,26 At Johns Hopkins, PC and oncology share space, but patients are not evaluated in joint visits due to challenges in scheduling. Medical oncology fellows have rotated through specialty solid tumor divisions and noted limited use of PC services for their own patients. We hypothesized poor usability of the electronic medical record (EMR) as a potential cause.²⁷ To our knowledge, no study has investigated EMR design and its impact on ASCO PC guideline adherence. Therefore, we formed a multidisciplinary quality improvement (QI) team to investigate and develop sustainable interventions to increase appropriate PC utilization in a first-year fellow longitudinal cancer clinic.

METHODS

A multidisciplinary team consisting of attending physicians (PC and oncology, including program leadership), oncology fellows, PC team members, information technology analysts, and clinic schedulers was recruited. The study did not require approval by the institutional review board.

Oncology fellows have a half-day of continuity clinic every week where they staff patients they follow longitudinally as the primary oncologist. This project was focused on first-year fellows because they had not yet established their practice patterns and would be more receptive to change, substantial deficiencies in PC educational domains have been noted in medical oncology fellowships,²⁸ and intent to pilot before implementation across the cancer center.

Aim Statement

The primary outcome measure was the proportion of patients with advanced cancer who were referred to PC by first-year medical oncology fellows. We aimed to increase the rate of PC utilization among patients with advanced cancer followed by first-year fellows at Johns Hopkins from a baseline of 11.5% to 30% over 6 months.

Plan

We obtained baseline rates of PC utilization between July 2018 and December 2018. We reviewed the schedule of each first-year oncology fellow to identify patients eligible for PC, defined as newly diagnosed patients with advanced solid cancers (stage IV solid tumor or stage III lung or pancreatic cancer) and a prognosis of ≤ 2 years.²⁹ Patients with hematologic malignancies were excluded because they are not followed in fellow clinic. Baseline data were obtained from 43 eligible patients. Data were analyzed and figures created using QIMacros software (KnowWare International, Denver, CO).

The multidisciplinary team met with stakeholders to identify reasons for low PC consultation. A process map of requesting outpatient PC consultation (Fig 1) and a fishbone diagram for low consultation rates (Fig 2) were created. It emerged that the EMR caused multiple layers of inefficiencies. There were multiple nonstandardized pathways to request a consult, without a set point person for scheduling.

Fig 1.

Flowchart depicting the process of ordering palliative care consultation before the intervention.

Fig 2.

Fishbone diagram for low consultation rates to palliative care for patients with advanced solid cancers. EMR, electronic medical record.

The core multidisciplinary team reviewed the data and voted on the top barriers to PC utilization (20 total votes, 4 members with 5 votes each; multiple votes on same cause allowed). A Pareto chart was created and identified the following factors: fellows felt they were doing a good enough job managing symptoms and distress; lack of a standardized way of consulting PC; and no formal way to communicate with the longitudinal PC team before and after a consult (Appendix Fig A1).

Do

The “do” interventions of the Plan-Do-Study-Act (PDSA) cycle were implemented from December 2018 to January 2019, with postintervention data collection from January to June 2019. Two interventions were applied initially—EMR changes and culture change and education.

EMR changes.

The process of requesting a PC consult in the EMR was standardized with input from frontline users, the PC team, and actual schedulers. A single point person who managed scheduling was identified to avoid confusion. Urgency of need for consultation and other data could be communicated to the scheduler directly via the new order. This was communicated to all first-year fellows.

Culture change and education.

Starting in December 2018, reminders for fellows regarding low rates of PC utilization were held at weekly QI sessions (5 minutes before QI conference, held six times), and two educational sessions were dedicated to this topic. Baseline data for the entire class were presented to the group, and ASCO guidelines were reviewed.⁹ Key studies reporting improved survival and outcomes with early involvement of PC were presented.^9,29-37 Fellows discussed some themes including the role of oncologists in primary and secondary PC, oncologists overestimating their ability, and patients being hesitant to mention everything to oncologists. A national leader in medical oncology and palliative medicine (T.J.S.) mediated the educational sessions. He clarified the breadth and scope of PC as an important step to optimize PC referral and use of services. These sessions were purposefully left unstructured and open-ended to allow a healthy discussion. We anticipated that this would allow true culture change and not be a unidirectional flow of information.

Study

From January 1 to June 30, 2019, we collected data on PC utilization rates. The fellows, PC team, and scheduler were interviewed to gather feedback and comments. Control charts were used to interpret the types of variation, process shifts, and trends (Appendix Fig A2).

Act

After completion of the first data collection period, the team reconvened to determine future directions. We identified the need for standardizing communication between the primary oncologist and PC both before and after the consultation. In addition, based on interest expressed by the remaining fellows and faculty, we decided to expand the program to the entire cancer center starting July 1, 2019. Data collection for this phase is ongoing.

RESULTS

In the preintervention group, 43 patients (mean age, 67.5 years; 46.5% women; 74.5% white) met eligibility criteria for referral to PC. Of these, only 5 patients (11.5%) were seen by PC. The average number of days from referral to appointment was 21 days. Baseline characteristics of patients in the preintervention and intervention groups are listed in Table 1. There were no differences in age distribution (P = .42), sex (P = .38), race or ethnicity (P = .32), or cancer type (P = .45) between the two groups.

TABLE 1.

Baseline Characteristics of Patients Meeting Criteria for Palliative Care Consultation in Medical Oncology Fellow Clinic

Overall, 48 (48.4%) of 99 patients meeting referral criteria were seen in the postintervention period. A greater number of patients were eligible as fellows continued to pick up new patients throughout their first year of training. The average number of days from referral to being seen was 10 days. All 48 referral orders met criteria based on ASCO guidelines.⁹ Reasons for referral are included in Appendix Table A1. No adverse balance measures by PC providers were reported during the intervention period.

To plan for future studies, participating fellows were surveyed (one time in March 2019; 100% response rate) to assess reasons for nonreferral. The most frequently reported reasons included inconvenience for patients to see another provider (88%); persistent association of PC with death, hospice, and/or end-of-life care (63%); and belief that palliation is the role of the medical oncologist and referral should be when symptoms cannot be managed (50%; Appendix Fig A3).

DISCUSSION

Although prior work has examined how oncologists’ attitudes about PC can impede referral,^38-40 our study is the first, to our knowledge, that implicates the EMR and the process of obtaining formal consultation as barriers to PC uptake for patients with advanced cancer. Difficulties in obtaining consultation may adversely affect communication and collaboration.⁴¹ Our PDSA intervention more than doubled our PC referral rate, given improvements in workflow and care coordination.⁴² Before our intervention, barriers to PC referral included multiple consulting mechanisms, difficulty finding the appropriate order in the EMR, and lack of clarity around who schedules patients into PC clinic.

International experts in PC and oncology have listed indications for PC referral based on needs or time-based criteria,⁴³ but adoption and implementation have been slow.^17,44-48 As oncologists’ knowledge of guidelines and understanding of PC improve,⁴⁹ external system-level factors may limit adherence to guidelines.²⁴ Previous efforts to increase PC referral rates include smart charts with electronic reminders when patients meet referral criteria,⁵⁰ note changes that alert providers of sick patients,⁵¹ and automated consult triggers.²⁹ None of these studies focused on medical oncology fellows. Because long-term practice is based on institutional culture and training patterns and PC research is rarely inclusive of medical oncology fellows,^28,52-54 we targeted our intervention on these trainees.

Our intervention was novel, fellow led, and evidence based. It brought together faculty and resources from medical oncology and palliative medicine and leaders in quality. Our data have implications for other cancer centers looking to enhance PC services use and have a high likelihood for scalability. These efforts do have limitations, however. Although we noted an increase in PC referral rates, the follow-up period was short and may not be reflective of long-term practice. In addition, multiple interventions were applied simultaneously, and it is not possible to know which intervention effectively increased PC referral. Other potential confounders, such as growing familiarity with the PC team and changing attitudes and beliefs of referring fellows, likely contributed. Patient outcomes were not assessed in the current study because ample research demonstrates that early PC referral is associated with favorable quality indicators.^1,5,14,55

We learned several valuable lessons from this study. First, achieving buy-in for the intervention from the departments of medical oncology and PC, hospital leadership, frontline providers, and administrators was critical to the success of the intervention. First-year medical oncology fellows and frontline PC providers were shown the rationale for the intervention and given a forum to voice concerns. PC leadership understood the potential consequence of extra consultations if the intervention was successful and allotted to increase staffing if needed. Interestingly, patients were seen in a more timely fashion despite an increase in number of referrals, without changes in the number of available providers.

Second, after sharing our data with hospital leadership, the order set was made available to all physicians at our institution and was heavily used. Before our intervention, 60% of patients with cancer seen in the PC clinic were referred via the EMR. The remaining 40% of patients were referred by e-mail, phone call, and/or a secure messaging system. We reviewed the charts of 50 consecutive new patients referred to the PC clinic after the availability of the new order set. Almost all (> 90%) of new patient consults to PC from medical oncology were referred this way. This change further benefits our cancer center by allowing better tracking of referral patterns over time. This year, our team will investigate balance measures by interviewing the PC providers and scheduler. We are interested in whether delays occurred for patients without a cancer diagnosis and if increased demand strained availability or morale of the PC team.^56,57

In conclusion, using PDSA cycles, we identified contributing factors for poor referral to PC among first-year medical oncology fellows. We increased consultation without implementing triggers, which has been the major limiting factor for generalizability in this space. We will continue to monitor the long-term effects of our intervention and measure sustainability. Despite the improvements noted, the proportion of overall referrals to PC in our cohort remains low. Therefore, although streamlining the referral process is associated with positive change, it is not a stand-alone solution. Future study will integrate these changes across the cancer center in combination with behavioral economic interventions, such as use of peer comparisons.⁵⁸

APPENDIX

Fig A1.

Pareto chart demonstrating barriers to palliative care utilization. A small stakeholder meeting of 4 core members, with 5 votes each, tallying to a total of 20 votes. Members could vote on the same factor more than once. Factors were identified using 3 rounds of iterative voting.

Fig A2.

Run chart demonstrating rate of patients with advanced solid tumors in medical oncology fellow clinic referred to palliative care. CL, confidence limit; LCL, lower confidence limit; PDSA, Plan-Do-Study-Act; UCL, upper confidence limit.

Fig A3.

Fellow responses to reasons for not referring patients with advanced cancer to palliative care. Ten fellows responded to the survey, each allowed to vote up to 3 times. Choices were prechosen by our team, with a write-in option available.

TABLE A1.

Reasons for Palliative Care Referral as Documented by Medical Oncology Fellows After the Intervention

AUTHOR CONTRIBUTIONS

Conception and design: Ramy Sedhom, Arjun Gupta, Mirat Shah, Marcus Messmer, Joseph Murray, Ilene Browner, Thomas J. Smith

Administrative support: Thomas J. Smith

Provision of study material or patients: Thomas J. Smith

Collection and assembly of data: Ramy Sedhom, Arjun Gupta, Melinda Hsu, Thomas J. Smith, Kristen Marrone

Data analysis and interpretation: Ramy Sedhom, Arjun Gupta, Joseph Murray, Thomas J. Smith, Kristen Marrone

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Oncology Fellow–Led Quality Improvement Project to Improve Rates of Palliative Care Utilization in Patients With Advanced Cancer

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/op/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).

Marcus Messmer

Stock and Other Ownership Interests: Schrodinger

Thomas J. Smith

Employment: UpToDate

Patents, Royalties, Other Intellectual Property: Royalties from Oxford Textbook of Cancer Communication, co-editor

Open Payments Link: https://openpaymentsdata.cms.gov/physician/202382/general-payments

Kristen Marrone

Honoraria: Takeda, AstraZeneca, Compugen

Consulting or Advisory Role: AstraZeneca, Compugen

Research Funding: Bristol-Myers Squibb (Inst), AstraZeneca (Inst)

Travel, Accommodations, Expenses: Compugen

No other potential conflicts of interest were reported.