aWe calculated patient participation rate (46%) as the number of patients consented and enrolled divided by the total number of patients invited to participate.
bThe number of clusters decreased from 41 (number of clinician clusters) to 35 (number of clusters for analyzable patients) because some clinician clusters either enrolled no patients or their patients’ data were not able to be analyzed.
cUnlike the primary outcomes analysis reported elsewhere,34 these secondary documentation outcomes were not dependent on return of patient surveys, so the denominator includes all patients who died within 24 months of enrollment.
eTable 1. Baseline Characteristics of Clinicians with At Least One Deceased Patient
eTable 2. Baseline Characteristics of Deceased Patients
Data Sharing Statement
Customize your JAMA Network experience by selecting one or more topics from the list below.
Paladino J, Bernacki R, Neville BA, et al. Evaluating an Intervention to Improve Communication Between Oncology Clinicians and Patients With Life-Limiting Cancer: A Cluster Randomized Clinical Trial of the Serious Illness Care Program. JAMA Oncol. 2019;5(6):801–809. doi:10.1001/jamaoncol.2019.0292
Does a communication quality-improvement intervention improve the occurrence, timing, quality, and electronic medical record accessibility of documented conversations about values, goals, and preferences (ie, serious illness conversations) between oncology clinicians and patients with advanced cancer?
In this cluster randomized clinical trial of 91 clinicians and 278 patients, the intervention resulted in more, earlier, and better serious illness conversations with clinicians for patients with advanced cancer, in addition to more accessible documentation of patient goals, in the face of life-limiting illness.
A communication quality-improvement intervention that provides clinical tools, clinician training, and system changes can bring about meaningful improvement in timely, comprehensive serious illness conversations between patients and primary oncology clinicians.
Earlier clinician-patient conversations about patients’ values, goals, and preferences in serious illness (ie, serious illness conversations) are associated with better outcomes but occur inconsistently in cancer care.
To evaluate the efficacy of a communication quality-improvement intervention in improving the occurrence, timing, quality, and accessibility of documented serious illness conversations between oncology clinicians and patients with advanced cancer.
Design, Setting, Participants
This cluster randomized clinical trial in outpatient oncology was conducted at the Dana-Farber Cancer Institute and included physicians, advanced-practice clinicians, and patients with cancer who were at high risk of death.
Main Outcomes and Measures
The primary outcomes (goal-concordant care and peacefulness at the end of life) are published elsewhere. Secondary outcomes are reported herein, including (1) documentation of at least 1 serious illness conversation before death, (2) timing of the initial conversation before death, (3) quality of conversations, and (4) their accessibility in the electronic medical record (EMR).
We enrolled 91 clinicians (48 intervention, 43 control) and 278 patients (134 intervention, 144 control). Of enrolled patients, 58% died during the study (n=161); mean age was 62.3 years (95% CI, 58.9-65.6 years); 55% were women (n=88). These patients were cared for by 76 of the 91 enrolled clinicians (37 intervention, 39 control); years in practice, 11.5 (95% CI, 9.2-13.8); 57% female (n=43). Medical record review after patients’ death demonstrated that a significantly higher proportion of intervention patients had a documented discussion compared with controls (96% vs 79%, P = .005) and intervention conversations occurred a median of 2.4 months earlier (median, 143 days vs 71 days, P < .001). Conversation documentation for intervention patients was significantly more comprehensive and patient centered, with a greater focus on values or goals (89% vs 44%, P < .001), prognosis or illness understanding (91% vs 48%, P < .001), and life-sustaining treatment preferences (63% vs 32%, P = .004). Documentation about end-of-life care planning did not differ between arms (80% intervention vs 68% control, P = .08). Significantly more intervention patients had documentation that was accessible in the EMR (61% vs 11%, P < .001).
Conclusions and Relevance
This communication quality-improvement intervention resulted in more, earlier, better, and more accessible serious illness conversations documented in the EMR. To our knowledge, this is the first such study to demonstrate improvement in all 4 of these outcomes.
ClinicalTrials.gov identifier: NCT01786811
Absent, late, and poor-quality clinician-patient communication in serious illness contributes to avoidable suffering for patients and families and overly aggressive, lower-quality care that may not align with patient goals.1-5
Although up to 87% of patients with terminal cancer have documentation of a discussion with clinicians about their goals by the end of their lives, these discussions begin, on average, 1 month before death, and the majority take place in acute care settings with clinicians who are not the treating oncologist.6 In broader seriously ill populations, the proportion of patients or surrogates who report having such conversations varies significantly (18%-64%) but is generally much lower than rates of documented discussions.1,6,7 In an increasingly fragmented health care system dependent on the electronic medical record (EMR) for communication among clinicians, documentation of these discussions is often inadequate, inconsistent, and inaccessible.8,9 Furthermore, researchers and national consensus bodies recommend making these conversations part of routine care for seriously ill patients and broadening the focus of these discussions from discrete decisions about life-prolonging procedures to conversations that address patients’ broader values, goals, and informed preferences, or serious illness conversations.10-14 There is a need for more patients to have serious illness conversations with clinicians; there is also a need for earlier initiation of these conversations, better discussions that address more comprehensive and patient-centered domains, and more accessible conversations that are appropriately documented in the electronic medical record.
Preparing clinicians who care for seriously ill patients to have timely, high-quality serious illness conversations requires addressing numerous barriers: inadequate clinician training in communication-skills, clinician time constraints, patient anxiety, clinician uncertainties about appropriate conversation timing or patients’ readiness to discuss these issues, ambiguous responsibility among multiple clinicians for holding a conversation, and inadequate systems to support clinicians in eliciting and documenting patient goals.10,15-20 Numerous interventions have attempted to improve these conversations and associated patient outcomes, and results have been mixed.21-29 While many interventions focus exclusively on clinician training and education, there is insufficient evidence that training alone leads to improvements in clinician conversation practices or patient-level outcomes.30,31 Therefore, we need comprehensive system-level solutions that address the entire range of barriers to support integration of serious illness communication into clinical practice.
We designed a communication quality-improvement intervention, the Serious Illness Care Program (SICP), to improve serious illness conversations. The intervention was extensively pretested with clinicians and patients32; each intervention component addressed 1 or more of the barriers to these discussions (Table 1).33 We tested the effect of the intervention with oncology clinicians and their patients in a cluster randomized clinical trial at Dana-Farber Cancer Institute. The study’s primary outcomes (goal-concordant care and peacefulness at the end of life) and secondary patient outcomes (anxiety, depression, therapeutic alliance, and survival) are reported elsewhere.34 The goal of this analysis was to evaluate the effect of the intervention on secondary conversation outcomes, including the occurrence, timing, quality, and documentation accessibility of oncology clinician–led serious illness conversations. We hypothesized that the results of this intervention would be (1) that a higher proportion of oncology patients would have at least 1 documented conversation before death, (2) that the conversations would start earlier in the illness course, and that conversations would be (3) more comprehensive and (4) more accessible in the EHR.
The study was conducted at the Dana-Farber Cancer Institute (DFCI), a large National Cancer Institute–designated cancer center, and 2 affiliated satellite clinics. It was approved by the DFCI institutional review board and registered, and all participants provided written informed consent. The trial protocol is available in Supplement 1.
The communication quality-improvement intervention included clinical tools, clinician training, and system changes (Table 1).32 The primary clinician tool was a structured communication guide called the Serious Illness Conversation Guide (SICG). Patient tools included a preconversation letter given to the patient at study enrollment, which introduced the SICG, and a “Family Guide,” outlining an approach for continuing the conversation with their family after the patient-clinician discussion. The clinician training included a 2.5-hour, skills-based training session on the SICG led by palliative care faculty. System changes included the following systematic components: (1) clinicians were asked the “surprise question,” “would you be surprised if this patient died in the next year?”33; (2) clinicians were sent email reminders and given the SICG by study staff the day before an outpatient visit; (3) an accessible, structured, EMR documentation template mirroring the SICG was provided, and clinicians were trained on its use; and (4) in-person, email-based, or telephonic clinician coaching on the SICG was provided by palliative care faculty. The quality-improvement elements of these intervention components are described in more detail elsewhere.32
Control clinicians provided usual care and did not receive any intervention components, except for regular identification of eligible patients using the surprise question. The SICG template was available in an EMR advance-care planning module. Although all clinicians at DFCI were made aware of the module prior to the study, control clinicians were not trained to use the SICG template. Control patients received no supporting documents.
All clinicians received a $150 restaurant gift card for participation. Patients and caregivers did not receive compensation for participation.
We tested the SICP in a cluster randomized clinical trial from September 2012 to June 2016. Clinicians were randomized in clusters at enrollment to intervention or control arms, as elsewhere reported (Figure 1).34 Clusters were units of clinicians within a disease center (eg, breast oncology), typically including 1 nurse practitioner or physician’s assistant and 2 or 3 physicians, although the numbers varied. Patients, but not clinicians, were blinded to study arm.
All physicians, nurse practitioners, and physician’s assistants from the selected sites were eligible to participate in the study; clinicians from 10 disease centers and 2 satellites enrolled. Both intervention and control clinicians identified patients (18 years or older, receiving ongoing oncology care at the center) for eligibility using the surprise question: “Would you be surprised if this patient died in the next year?”33 Patients for whom the response was “no, I would not be surprised” were eligible for the study. We excluded patients who were not English speaking, had cognitive impairment, or were unable to identify a caregiver to participate.32 For this analysis, patient participants who died within 2 years of study enrollment were eligible for inclusion.
We powered the study based on its primary outcomes, goal-concordant care and peacefulness at the end of life, which are reported elsewhere.34 At baseline, we collected clinician and patient demographic characteristics for both study arms.
We obtained the secondary outcomes reported herein from EMR review for patients who had died, including (1) proportion of patients with at least 1 documented serious illness conversation before death; (2) median timing of the first documented serious illness conversation before death; (3) quality of documented serious illness conversations; and (4) proportion of patients with documentation of a serious illness conversation in an accessible EMR location.
For patients who died during the study period, study staff blinded to study arm conducted a retrospective record review of ambulatory oncology EMR records (incorporating all progress notes and a structured module designed to capture all advance care planning documentation, including the SICG template). A multidisciplinary team, with expertise in palliative care and psychiatry, iteratively developed and refined through pilot testing a thematic codebook to facilitate record abstraction and text coding. Using this codebook, a trained, blinded researcher abstracted all instances in the EMR of serious illness conversations that addressed at least 1 of 4 domains: (1) values or goals; (2) prognosis or illness understanding; (3) end-of-life care planning; or (4) life-sustaining treatment preferences. A second trained, blinded researcher independently abstracted and double-coded 10% of patient records to verify consistency across coders, with 98% agreement on all abstracted elements. For each patient, the reviewer recorded absence or presence of at least 1 serious illness conversation; date of each conversation; retrieval location (progress note vs structured module); and content of each conversation for coding.
To measure the occurrence of serious illness communication, we noted the presence or absence of at least 1 documented serious illness conversation before death for each patient. To measure the timing, we calculated the time (in days) from the initial conversation to death for patients who had at least 1 serious illness conversation. To measure the quality of serious illness conversations, we coded all abstracted notes for each patient for the presence or absence of each of the 4 conversation domains and reported these data in 2 ways: (1) mean number of domains per patient; and (2) presence or absence of each of the 4 domains per patient. We also recorded the total number of notes per patient. To measure accessibility, we recorded the presence or absence of serious illness conversation documentation in the EMR structured advance care planning module (vs progress note) for patients who had at least 1 conversation. Accessibility was calculated for the subset of patients who died before June 1, 2015, because there was an institutionwide EMR change after that date, and the new EMR did not have a working module until after the study period ended.
We performed statistical analyses with SAS software, version 9.4 (SAS Institute Inc). We used proportions to describe results for categorical variables and means or medians for continuous variables. We used intention-to-treat for all analyses and accounted for clustering of patients within clinician teams.
When comparing baseline clinician and patient characteristics across arms, we used robust generalized estimating equations (GEE)35 Wald tests clustering by clinician team, with arm as a dichotomous covariate, and linear and logistic link functions for continuous and dichotomous characteristics, respectively.
We analyzed differences between arms in the proportion of patients who had at least 1 documented serious illness conversation before death using robust GEE Wald tests to evaluate occurrence of conversations. We used a log-rank test to compare the distribution of time from randomization until initial documented conversation between intervention and control arms. For quality of documented serious illness conversations, we coded the content into thematic domains using NVivo, version 10.0 (QSR International). Between arms, we then compared the mean number of conversations and domains per patient using robust GEE t tests. We also compared the proportion of patients with documentation within each of the 4 domains between arms using robust GEE Wald tests.36 These 4 outcomes were considered equally important, and we used the Holm-Bonferroni method to adjust for multiple testing and to ensure an overall type I error of 5% when comparing the 4 outcomes across arms. For accessibility, we analyzed differences between arms in the proportion of patients who had documentation in the structured EMR module using robust GEE χ2 tests.
We enrolled and randomized 91 oncology clinicians; 76 oncology clinicians had at least 1 enrolled patient who died during the study period and were included in this analysis. Of 278 patients enrolled, 161 died during the study period and were included in this analysis. There were no significant differences in baseline characteristics of clinicians (eTable 1 in Supplement 2) or patients (eTable 2 in Supplement 2) across study arms.
We conducted EMR reviews of all 161 patients who died during the study period, and all absolute numbers as well as the following percentages are reported in Table 2: Significantly more intervention patients than controls (96% vs 79%, P = .005) had at least 1 documented serious illness conversation before death. The timing of the first conversation was significantly earlier in the intervention group (median 143 days before death; interquartile range [IQR], 71-325 days) than in the control group (median 71 days; IQR, 33-166 days) (P < .001). Significantly more intervention patients had a documented discussion about values or goals (89% vs 44%, P < .001), prognosis or illness understanding (91% vs 48%, P < .001), and life-sustaining treatment preferences (63% vs 32%, P = .004). Documentation of discussions about end-of-life care planning (80% intervention vs 68% control, P = .08) did not differ significantly between arms. Conversations for intervention patients also included significantly more domains on average (mean 3.2 domains vs 1.9 domains, P < .001). There were significantly more conversations per patient in the intervention group than in the control group (mean 3.1 conversations per patient vs 2.1 conversations per patient, P = .02). For the subset of patients who died before June 1, 2015 (when there was an institutionwide EMR change), and had at least 1 documented serious illness conversation (n = 111; 57 intervention and 54 control), significantly more intervention patients than controls had a discussion documented in the accessible structured EMR module (61% vs 11%, P < .001).
This cluster randomized clinical trial demonstrates that a communication quality-improvement intervention resulted in significant improvements in 4 key conversation indicators, including more, earlier, better, and more accessible oncology clinician–led serious illness conversations. The SICP resulted in nearly universal access to such conversations, with 96% of intervention patients having at least 1 documented conversation before death. Conversations also occurred a median of 2.4 months earlier in the intervention arm than in the control arm, and the documentation of the conversations with intervention patients was more accessible in the EMR and contained significantly more comprehensive and broader information about patients’ values, goals, and preferences. While several interventions have been shown to increase goals-of-care communication and documentation or patient-reported quality,22-24,26,29 to our knowledge, no other study has demonstrated an effect on all 4 of these conversation indicators.
The significant increase in conversations in the intervention arm suggests that successfully integrating these conversations into a typical oncology practice is feasible. A structured conversation guide and matching EMR documentation tool, reminder process, and clinician training and coaching appear to have addressed known barriers to serious illness communication, including clinicians’ lack of communication training, time constraints, and cumbersome (or absent) systems for having and documenting conversations.17,18 Improving the documentation of conversations with a structured EMR template also provides easier access to this critical information, an issue increasingly recognized as essential for patient safety.9,37 If clinicians, especially those clinicians who are unfamiliar with the patient and facing an emergency, cannot easily retrieve documentation of patient goals, values, and preferences, patients may be at increased risk of receiving unwanted care.
The earlier initiation of conversations in the intervention arm marks an important shift from the current practice of beginning serious illness discussions late in the illness course and usually in acute care settings.6 By prospectively identifying patients and reminding clinicians at the point of care, we stimulated earlier discussions by reducing uncertainties about the appropriate time to start these conversations.16 Earlier conversations that provide perspective on the patient’s disease trajectory and prompt discussion of values and goals in the outpatient setting, in advance of crises, benefit patients by allowing them to make earlier decisions that support their goals and to prepare themselves and their families for the end of life.3,38,39 The significant increase in the mean number of documented discussions per patient as a result of the intervention suggests that initiating serious illness communication earlier may prompt more discussions over time.
Rather than only discussing discrete decisions about life-prolonging treatments (eg, cardiopulmonary resuscitation), the conversations driven by the SICP focused on sharing clinical realities and eliciting patients’ broader values and goals,40-43 critical competencies that are increasingly recommended by experts and national consensus bodies10-14,44,45 and are central to the widely supported goal of shared decision making.13,14,46,47 Additionally, patients in the intervention arm discussed significantly more conversation domains on average than patients in the control arm, suggesting that the intervention stimulated more comprehensive conversations. The use of a structured communication tool that provided a patient-centered, psychologically informed script or guide,10,26,29 accompanied by brief training and supportive coaching, appeared to engage clinicians in a different approach to these difficult conversations, which led to richer and more complete elicitation and documentation of patients’ goals and preferences. Although the intervention did not affect the primary outcomes of goal-concordant care and peacefulness, the improvement in the secondary outcomes of anxiety and depression34 suggests that these values-based discussions may have immediate mental-health benefits for patients with serious illness.
In addition, this study demonstrates the potential system-level value in well-designed, comprehensive changes—including infrastructure improvements. Notably, the use of a structured documentation template in the EMR allows for system-level tracking of critical conversation processes, such as frequency, timing, setting, and key elements addressed, which facilitates local quality-improvement efforts and measurement of communication outcomes.12,29,45 In addition, when tested in the high-risk primary care context, the SICP intervention has demonstrated significant improvements in the occurrence, quality, and accessibility of documented serious illness conversations for patients before they died,29 supporting its potential replicability in different populations and clinical settings. Though the multicomponent nature of the intervention prevents a more detailed assessment of the association between individual components and outcomes, we believe that successful implementation of SICP depends on the synergy of its various components. Since each of these intervention components addresses a systems barrier or promotes a best practice in serious illness communication,10 the effect of serious illness conversations across the illness trajectory is likely the greatest when these components work together. See Figure 2 for an illustration of how all 4 goals—more, earlier, better, and more accessible conversations—can optimize the effect of serious illness communication on quality of care.
This study has several limitations. In this analysis, we examined documented and not actual discussions. Documentation does not capture important clinician communication behaviors, including what words are spoken and how clinicians respond to emotion. The use of documentation to measure and monitor conversations will need to be coupled with other efforts to assess quality of communication, including patient or family experience surveys and the use of audio recording.12 In this study, we examined oncology notes in the outpatient setting and did not examine inpatient notes, which may have included documentation of goals and values. The differences between arms may also have been attenuated by having all clinicians answer the surprise question and having all patients complete surveys, activities which may have prompted conversations in the control arm. The study’s generalizability is limited because it was conducted at a single oncology institution with a relatively homogenous white patient population. Finally, the long-term sustainability of the program is unknown.
In conclusion, our study shows that this institutionwide intervention, which trained and supported oncology clinicians in improving their conversation practices and integrating those conversations into routine outpatient practice, resulted in broad access to more, earlier, and better serious illness conversations for patients with advanced cancer as well as more accessible documentation of these conversations in the EMR. This intervention provides a comprehensive and potentially scalable model for enhancing communication about patient values, goals, and preferences across a system.
Accepted for Publication: January 20, 2019.
Corresponding Author: Joanna Paladino, MD, Serious Illness Care Program, Ariadne Labs, 401 Park Dr, Third Floor E, Boston, MA 02215 (email@example.com).
Published Online: March 14, 2019. doi:10.1001/jamaoncol.2019.0292
Author Contributions: Drs Paladino and Bernacki contributed equally as co–first authors. Drs Paladino, Bernacki, and Block had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Paladino, Bernacki, Lamas, Gass, Lipsitz, Gawande, Block.
Acquisition, analysis, or interpretation of data: Paladino, Bernacki, Neville, Kavanagh, Miranda, Palmor, Lakin, Desai, Sanders, Gass, Henrich, Lipsitz, Fromme, Block.
Drafting of the manuscript: Paladino, Gass, Lipsitz.
Critical revision of the manuscript for important intellectual content: Paladino, Bernacki, Neville, Kavanagh, Miranda, Palmor, Lakin, Desai, Lamas, Sanders, Henrich, Lipsitz, Fromme, Gawande, Block.
Statistical analysis: Neville, Gass, Lipsitz.
Obtained funding: Bernacki, Gawande, Block.
Administrative, technical, or material support: Paladino, Bernacki, Kavanagh, Lakin, Desai, Lamas, Gawande, Block.
Study supervision: Paladino, Bernacki, Henrich, Fromme, Gawande, Block.
Conflict of Interest Disclosures: Dr Block receives compensation as Palliative Care Editor from Up to Date. Dr Fromme was paid to coauthor an article on workforce training in communication skills by the Gordon and Betty Moore Foundation. Dr Gawande reports compensation from produced writing and other media on health care including the subject of serious illness care and checklists during the conduct of this study. Dr Gawande is employed as the CEO of the new nonprofit-seeking healthcare venture parented by Amazon, Berkshire Hathaway, and JP Morgan Chase. No other disclosures are reported.
Funding/Support: Research reported in this publication was supported by the Branta Foundation, Charina Endowment Fund, Margaret T. Morris Foundation, Richard A. Cantor Fund, Partners HealthCare, and the John A. Hartford Foundation.
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.
Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the Branta Foundation, Charina Endowment Fund, Margaret T. Morris Foundation, Richard A. Cantor Fund, Partners HealthCare, and the John A. Hartford Foundation.
Meeting Presentation: This article was presented at the American Academy of Hospice and Palliative Medicine 2019 Annual Assembly; March 14, 2019; Orlando, FL.
Additional Contributions: We thank Ami Karlage, BA, for her contributions in critical revision of the manuscript content, and we thank the patients and clinicians who participated in this study. All clinicians received a $150 restaurant gift card for participation; otherwise, no persons received compensation for their contributions beyond the normal course of their employment.
Data Sharing Statement: See Supplement 3.
Create a personal account or sign in to: