An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Rebecca A. Aslakson, MD, PhD, Sarina R. Isenberg, MA, Norah L. Crossnohere, MHS, Alison M. Conca-Cheng, BS, Madeleine Moore, Akshay Bhamidipati, Silvia Mora, Judith Miller, Carolyn Pastorini, Sarabdeep Singh, PhD, Sandy Swoboda, RN, MS, Timothy Pawlik, MD, Angelo Volandes, MPH, MD, John F.P. Bridges, MEc, PhD, Thomas J. Smith, MD, Matthew Weiss, MD, and Debra Roter, MPH, DrPH.
Author Information and AffiliationsDespite positive health outcomes associated with advance care planning (ACP), little research has investigated the impact of ACP in surgical populations. Our goal is to evaluate how an ACP intervention video affects the patient-centeredness and advance care planning content in patient–surgeon discussions of the risks and benefits of surgery. We hypothesized that patients who viewed the intervention would engage in more patient-centered communication and advance care planning with their surgeons as compared with patients who viewed a control video.
To improve advance care planning content and patient-centered communication for patients preparing for major surgery, particularly regarding the designation of a surrogate decision maker.
Randomized controlled comparative effectiveness trial of an ACP video with 2 study arms: intervention ACP video and control video; and 4 visits: baseline, presurgical consent, postoperative 1 week, and postoperative 1 month. Surgeons, patients, the principal investigator, and analysts were blinded to the randomization assignment.
Single, academic, inner city, tertiary care hospital.
Patients undergoing major cancer surgery who were recruited from 9 surgical oncology clinics.
In the intervention arm, patients viewed a patient preparedness, advance care planning video developed through extensive consultation with patients, surgeons, and other stakeholders. Patients randomized to the control arm viewed an informational video about the hospital surgical program.
The primary outcome was the patient-centeredness of patient–surgeon conversations during the presurgical consent visit as measured through the Roter Interaction Analysis System and content of ACP in the conversation. Secondary outcomes were patient Hospital Anxiety and Depression Scale score, patient goals of care, patient and surgeon satisfaction, video helpfulness, and medical decision maker designation.
We enrolled 92 patients (target enrollment of 90) over a 15-month period. No study-related harms were noted. Patient-centeredness was unchanged between the 2 study arms (incidence rate ratio = 1.06; CI, 0.87-1.3; P = .545) and there was a trend of increased ACP content in the intervention arm (23% intervention vs 10% control; P = .182). There were no differences between study arm groups in secondary outcomes other than that patients were more likely to find the intervention video “helpful” (P = .007) as compared with the control video. There was a statistically significant increase in mood symptoms Hospital Anxiety and Depression Scale (HADS) in both groups 1 week postoperative compared with those at enrollment (intervention P = 0.010; control P = .007). Patients in both groups most frequently noted cure-related goals of care.
Patients found the ACP video helpful and [well tolerated], but it did not significantly change the nature of the patient–surgeon communication or the content of ACP in the preoperative patient–surgeon conversation. Future studies could increase the “dose” of ACP by modifying video content and/or who presents that content before surgery and how.
This is a single-site study with a video initially conceptualized for a pancreatic cancer surgical population. The selected outcomes and study time frame may not be able to fully capture intervention effect. Surgeon-level factors could influence outcomes and we could not control for the potential effect of a patient's medical course on study outcomes.
In 2010, approximately 51 million surgeries were performed in the United States.1 Although most surgeries are performed successfully, patient morbidity and mortality persist,2-5 and some surgeries require postoperative life-sustaining treatments in an intensive care unit.6 While patients may be stratified for perioperative complications, it is difficult to impossible to predict which patients will die or suffer a major perioperative complication.3,5,7
Advance care planning (ACP) is a process by which individuals contemplate future health states, clarify and discuss their goals, and express goals-informed wishes for those health states—particularly if illness may render that person unable to make decisions for himself or herself in the future.8 Evidence supports that ACP discussions may decrease health care utilization, while increasing patient satisfaction, use of hospice and palliative care, and compliance with a patient's end-of-life wishes.9-13 For family members, ACP may also decrease anxiety, depression, and stress, while increasing satisfaction with the quality of care.9,14,15 ACP is appropriate throughout multiple stages of illness and in previous studies has not been associated with harm.16 The landmark 2014 Institute of Medicine report Dying in America advocates for increased ACP to explore patient wishes before they become acutely ill.17
Because patients undergoing major surgery are at risk of perioperative morbidity and mortality, it is likely beneficial for them to initiate ACP before surgery. A recent systematic review of palliative care interventions for surgical populations18 highlighted 5 studies that explored ACP interventions in surgical populations.19-23 These interventions involved further training or activation of surgical providers (eg, surgeons, anesthesiologists, nurses) to have an ACP conversation with patients before surgery and/or including a palliative care specialist specifically to have that conversation with the patient. These interventions found improved concordance and decreased decisional conflict between patients and surrogates about goals of care,19,20,22 showed improved documentation regarding power of attorney,21 and were deemed helpful by study participants20; none of these trials documented harms to patients or family members.
Verbal communication is the predominant modality for ACP between patients and providers24 and was the communication modality used in the above ACP interventions in surgical populations.19-23 Yet, multiple barriers can hinder optimal verbal communication in the patient–doctor relationship. Most important, verbal communication about ACP is inherently inconsistent and subjective; standardizing these conversations is challenging to impossible.25-29 Conversations may also inaccurately convey the burden and outcomes of medical interventions, particularly when the patient has no previous knowledge or experience of aggressive medical treatments (eg, intubation, artificial ventilation, artificial nutrition) or settings (eg, an intensive care unit).30 While ACP innately requires verbal communication between patients and providers, such communication can be facilitated or enhanced by educational tools, such as a video. Video ACP tools are considered more effective than verbal communication precisely because videos have inherently stable content and thus may be a more objective, simple-to-understand, and realistic modality to educate and activate patients about ACP.31,32 Thirteen randomized controlled trials in varying populations support that video-based ACP tools can empower patients and families to have ACP-related discussions,33-45 though none of these studies were completed in surgical populations.
This investigation builds on the paucity of research concerning video ACP tools in surgical populations.18 Our primary aim was to evaluate whether an ACP video developed for patients and families pursuing aggressive surgical treatment for cancer affects (1) the patient centeredness of and (2) the ACP content in the patient–surgeon conversation on risks and benefits of the surgery. We selected this primary aim based on 2 years of intense engagement with patients and family members, as well as other key stakeholders, including surgeons, anesthesiologists, surgical nurses, surgical intensive care unit nurses, palliative care clinicians, and health services researchers (Figure 1). We hypothesized that patients who viewed the intervention video would engage in more patient-centered communication with their surgeons than would patients who viewed the control video. We also hypothesized that the video would increase the content of ACP in the conversation.
Trial Timeline.
Our secondary aims addressed how the ACP intervention video may impact other related outcomes, such as the helpfulness of the video and longitudinal patient-reported anxiety, depression, and stated goals of care (Table 1).
Outcome Measurements Used in the Clinical Trial.
We involved diverse groups of stakeholders in the project at all stages, including design and conduct of research and dissemination of findings. The video design process involved extensive engagement with patients and families and key stakeholders, such as surgeons, palliative care clinicians, ACP experts, and surgical nurses; the process included in-depth interviews, focus groups, stakeholder summits, and a deidentified cross-sectional survey regarding potential video content.46-52 In all, more than 450 stakeholders were engaged in the development process.
From project inception, we identified the key stakeholder groups as patients and family members, surgeons and other surgical providers (eg, nurses, anesthesiologists), and ACP/palliative care clinicians and experts. At the stakeholder summits, members of each stakeholder group were carefully selected in a balanced fashion. While we aimed to engage as many surgeons as possible in this project, we were concerned that surgeons would outnumber other stakeholders. We ensured that in the makeup of our stakeholder summits, patient and family members were not outnumbered by any other clinician or research group.
Moreover, at key engagement points of the study, such as key informant interviews during the first year and editing of the advance care planning during the second year, we sought participants who represented the key stakeholder groups, to ensure equal representation across these groups.
We prospectively delineated the 3 key stakeholder groups and identified key stakeholder partners in each group through multiple methods. In brief, the first aim of our study was an environmental scan (including a systematic review, gray literature search, and key informant interviews) through which we identified many experts in the field. We identified partners for the specific stakeholder groups through the following means.
Both the systematic review and gray literature search identified stakeholders from this group for the key informant interviews. In these interviews, we also used snowball sampling to have respondents name other individuals who might have key insights on this topic. Through this sampling, we selected surgeon/surgical clinician partners who participated in our stakeholder summits and reviewed edits of the advance care planning video as it was developed.
Our lead patient/family stakeholder partner (Judi Miller) has been involved in the study from inception and continues to be closely involved. We identified other key patient/family partners (Mike Pardue, Carolyn Pastorini) through our lead surgeon partner (M.W.). For in-depth interviews regarding video content development, we also enrolled patient and family members from the outpatient clinic of our lead surgeon partner (M.W.) as well as through direct referral from other surgeon partners (ie, patients who proactively reported interest in this topic and their surgeon consequently referred them to us). For engagement regarding the storyboards for our video during development, we rented a booth at the 2014 Maryland State Fair and engaged with 359 self-identified patients/family members who approached our booth, viewed our storyboards, and wished to participate through a deidentified survey and general comments. For engagement regarding the edits of the video, we iteratively engaged with the Johns Hopkins Patient and Family Advisory Council (PFAC; 35 members).
Both the systematic review and gray literature search identified stakeholders from this group for the key informant interviews. In these interviews, we again used snow-ball sampling to have respondents name other individuals who might have key insights on this topic. Through this, we selected ACP/palliative care experts who participated in our stakeholder summits and reviewed edits of the advance care planning video as it was being developed.
Of note, we based this study on a foundation-funded study involving 2 years of qualitative research. In this study, we interviewed more than 30 long-stay patients, family members, surgeons, surgical intensive care unit (ICU) nurses, and surgical ICU physicians with a goal of better understanding the palliative care–related experience of long-stay surgical intensive care unit patients and their family members and whether their clinicians understood that experience. In brief, patients and family members repeatedly noted that they “expected everything but this”—they thought that their loved one would either “do well” after surgery and slowly but surely recover or he or she would “die on the table” in the operating room on the day of surgery. Patients and, particularly, family members noted that they were not prepared for a long postoperative surgical intensive care unit stay with the patient not actively dying but also not “getting better.” One family member said the following:
We chose Hopkins because we think it's the best. I think they've done everything they can. And basically what they're saying now, is that they've done everything…. I brought in a woman who was vital and now she can't speak…. It took 9 hours, they took half her insides out…. And now she's completely dependent … can't do anything anymore. She can't communicate. Of all the things we expected, this is not one of them. They went down the … list of all the problems you could have. We didn't think this was on the list.
Based on the detailed input from long-stay surgical ICU patients and their family members, our goal was to evaluate whether video-based advance care planning could help patients and family members preoperatively prepare for major surgery and to better know the patient's medical goals and wishes, particularly if postoperative complications should result in the patient having an unexpected and long postoperative recovery. We continued to engage patient and family stakeholders throughout the duration of the current study.
Throughout the funding period, we communicated and worked closely with our team patient and family co-investigators, Judi Miller, Mike Pardue, and Carolyn Pastorini. However, unfortunately, during the study period, both Pardue and Pastorini died of recurrent pancreatic cancer; thus, we finished the trial with only Judi Miller as a patient/family co-investigator.
Miller, Pardue, and/or Pastorini worked with us to develop the video content as well as to edit the final video and review study progress and results. We had monthly standing phone calls with each person, so that each could provide continuous assistance to the research team. Miller (along with the principal investigator, Rebecca Aslakson) has also presented project findings back to PCORI (late May 2014) and to the Johns Hopkins PFAC.
The study team performed 22 key informant interviews with surgeons, ACP/palliative care clinicians and experts, leaders of patient advocacy groups, and decision support tool content experts.47 The team asked key informants to discuss their thoughts about ACP and interventions to support ACP, particularly among surgical populations.
The study team rented a booth at the 2014 Maryland State Fair, an 11-day event held annually in Timonium, Maryland, and attended by approximately 400 000 people. State fairs, which typically have agricultural and livestock exhibitions as well as carnival and amusement activities, are gaining traction as research recruitment sites due to high attendance and attendee diversity.53,54 Individuals who indicated that they or a loved one had undergone major surgery were eligible to participate and were invited to review the storyboard.52 Over 10 days, 359 attendees of the 2014 Maryland State Fair evaluated the storyboards and 87% noted that they would be “very comfortable” or “comfortable” seeing the storyboard before major surgery, 89% considered the storyboards “very helpful” or “helpful,” and 89% said they would “definitely recommend” or “recommend” this story to others preparing for major surgery.
The study was presented and discussed by the PFAC in 2014, and a 7.5-minute refined prototype of the video was presented to the PFAC in June 2015. The council is an established group that meets monthly to identify patient and family needs and concerns, and advocates to integrate patient-centered care across the hospital. Of the 28 PFAC members present at this meeting, 15 were patients/family members, and 13 were faculty/hospital employees, none of whom had been involved in our previous engagements related to this project.
Over the course of the study, we conducted 3 stakeholder summits. The first, in November 2013, involved a diverse group of 14 stakeholders, including 2 patient/family members, 3 surgeons, 2 palliative care physicians, 1 decision support tool developer, 1 palliative care nurse, 1 surgical nurse, 1 patient-centered outcomes research expert, and 3 public health researchers. The second, in September 2014, brought together the same group of patients, family members, and other key stakeholders to view the video-recorded key informant interviews, select the final 8 interviews to be used in the video, and continue the discussion of the specific interview content and framing that should be selected for inclusion in the final video. Finally, our culminating engagement activity was presenting our clinical trial data to a stakeholder summit in February 2017. That summit involved a diverse group of stakeholders, including 2 patients, 3 family member/caregivers, 2 surgeons, 2 surgical intensive care unit nurses, 2 palliative care doctor-anesthesiologists, and 2 public health researchers. The stakeholders reviewed and further interpreted study findings and as well as discussed dissemination and next steps.
Key informants interviewed described the importance of initiating ACP preoperatively, despite potential challenges present in surgical settings. Stakeholders clearly specified the context of use to be among patients preparing for major cancer surgery and that the video should be presented to patients after a decision to pursue surgery but before the surgery itself.
In the intervention video development, participant stakeholders interacted closely with 16 decision support tools found through an environmental scan46 comprising a systematic review,48 a gray literature search, and interviews with key informants.47 The research team solicited stakeholder feedback on what content or design of each tool each they felt to be meaningful and/or relevant. Representing both user and organizational perspectives, the stakeholders discussed what they perceived the goals and content for the project decision support tool should be, as well as likely cultural barriers and/or constraints for operationalizing those goals and content with the decision support tool. Key content concepts the stakeholders identified included that the tool (1) use vignettes, (2) be framed in “how you want to live” rather than in a “death/dying” and “what do you want to die like” context, and (3) be specific to surgical populations. Key style concepts stakeholders identified included that the tool (1) be upbeat in tone, (2) involve both younger and older patients, and (3) involve patients of multiple races and ethnicities. Later in the video design process, patient and family stakeholders viewed a pilot version of the intervention video and suggested simplifying the message delivery while strengthening the ACP message dose. We incorporated both suggestions in the final version of the intervention video.
Patient engagement directly influenced the outcomes of our study. At the 2014 Maryland State Fair, the 359 participants who completed a deidentified survey also completed a prioritization task for potential study outcomes. Among the 12 potential outcomes presented, these self-identified patients and family members clearly noted that they felt the most important outcome for the study should be that “the video would enable me to have a meaningful discussion with my doctor about my goals and wishes.” Moreover, these patients and family members noted our intended primary outcome—mood symptoms—to be the least important. In the words of one of the participants, “whatever video you show me, I will be anxious before major surgery … and I have no expectation for that to be changed!”
Consequently, over a 5-month period in 2014-2015, we petitioned PCORI and ultimately changed the primary outcome of the clinical trial so that it could be directly responsive to this clear patient and family member input. We changed the primary outcome to measure whether the video would “enable a meaningful conversation with my doctor about goals and values.” Toward this goal, we brought the patient-medical care clinical communication researcher, Debra Roter, on to the study team, and, using her validated Roter Interactive Analysis System (RIAS)55 for evaluating patient–provider communication, we changed the primary outcome to the nature (“patient-centeredness”) and content (ACP content) of the provider–patient interaction. Thus, directly based on intense engagement with 359 patients and family members, we fundamentally changed the study and incorporated new team members so that our results could capture and measure that outcome identified by patients and family members.
Stakeholders remained engaged in the research process throughout the duration of the randomized control trial as well. When problems arose, such as low recruitment in surgical clinics, we consulted both clinician and patient/family member stakeholders, and their input resulted in changes in recruitment procedures, with a subsequent increase in enrollment.
Our patient/family co-investigators (J.M., M.P., and C.P.) were involved with study design and reviewed ongoing and final study progress and results. Moreover, 3 stakeholder summits (already described) engaged key stakeholders in issues directly affecting the study's rigor and quality.
The study–in its development, progress, and the results stages—was presented to key stakeholders through the 3 stakeholder summits. Engagement with these stakeholders directly impacted the study that we planned, conducted, and analyzed.
We involved patient/family and clinician stakeholders in interpreting study findings. During the February 2017 stakeholder summit, a diverse group of 20 different stakeholders were presented with and discussed study outcomes. They spent most of the day, including breakout groups, discussing and brainstorming the next steps for this work. Specifically, we discussed research and clinical practice initiatives that could stem from this project and findings. The patient/family member stakeholders strongly endorsed that preoperative ACP conversations should most likely not directly involve the surgeon. Thus, future studies/next steps propose involvement of a palliative care clinician before surgery to facilitate these discussions.
Of note, the study protocol has been published.56
The study was a 2-arm, randomized comparative effectiveness trial of an ACP video developed for patients undergoing major cancer surgery as compared with a control video that described the surgery department at the hospital but contained no ACP content.
Our study sample included patients undergoing major cancer surgery with 1 of 9 cancer surgeons at an academic tertiary care center. After meeting with patients in clinic, surgeons would determine their eligibility. If the surgeon deemed a patient potentially eligible and the patient was amenable, research staff met immediately with the patient to determine full eligibility, obtain written consent, and conduct the baseline visit activities.
Eligible patients were planning to undergo major surgery such that, due to the surgery itself and/or the patient's underlying medical conditions, the surgeon planned to postoperatively admit the patient to the surgical intensive care unit (SICU). Major surgery was defined as “surgery involving a risk to the life of the patient; specifically an operation upon an organ within the cranium, chest, abdomen, or pelvic cavity.”57 Study patients must have also been scheduled for nonemergent surgery so that they had at least a day to review the video. Potential study patients must also have met the following inclusion criteria: undergoing surgery with 1 of the participating surgeons, able to give informed consent, and able to speak and understand English (both study videos were in English only). We excluded patients if they were younger than 18 years old or had visual or hearing impairments that made them unable to view and/or hear the study videos.
We recruited patients out of 9 surgical oncology clinics. Of the 12 surgical oncologists who could have participated in this trial because their patients met enrollment criteria and their clinics were on site, we worked with 9 surgeons, who we chose because they had sufficient cancer patient populations and were willing to be in the trial. Of the 3 excluded surgeons, 2 had low-volume clinics and 1 was switching institutions partway through the study. In preparation for the study, surgeons described variations in their practice regarding presurgical visits and agreed on a single format to use uniformly for study patients; this format was composed of at least 2 visits with the surgeon before the actual surgery. All surgeons were comfortable with the ACP video and we showed them both the intervention and control videos before collecting data from their clinics. We also blinded participant surgeons as to which study video their patient was randomized to view.
With each study patient as a unit of randomization, we randomized immediately following enrollment. With only 9 participant surgeons, we did not have a sufficient number of surgeons to power randomization to be clustered by surgeon. We minimized contamination bias through blinding of the surgeons as to which video their patient was randomized.
We used a stratified approach to randomization because we hypothesized that individual differences in surgeon demeanor could affect primary and/or secondary outcomes. Thus, we stratified randomization by surgeon with a block size of 6.
Both videos were 6 minutes long.
Through extensive preliminary research and engagement with key stakeholders over 2 years,47-52,56,58 the study team developed a video-based ACP tool for patients pursuing aggressive surgical treatments for cancer. The video featured patients, companions, and medical professionals (2 surgeons, 1 anesthesiologist, 1 SICU nurse) discussing both the course of a typical surgical day—preoperative area, operating room, and SICU—as well as the importance of preoperative ACP—identifying a medical decision maker, discussing one's wishes with that decision maker, and communicating those wishes to the surgical team before the surgery. ACP content composed 37% of video time (Table 2). While the intervention video is not available in the public domain, example storyboards from the intervention video are depicted in Figure 2.
Intervention Video Content and Timing.
Example Storyboards From the Final Intervention Video.
The control video was an informational video about the Johns Hopkins surgery program, created by its Marketing Department independently from this study. The video, created for the hospital's 125th anniversary in 2014, catalogued the history and evolution of surgery at Johns Hopkins Medicine. It highlighted scientific developments and ongoing innovations in patient safety. The video is available at https://www.youtube.com/watch?v=Us33nu37qHw.
The study included 4 visits with each study patient: baseline visit (V1, not recorded), presurgical consent visit (V2, audio-recorded), postoperative 1-week visit (V3, not recorded), and postoperative 1-month visit (V4, not recorded; Figure 3).
Data Collection Flow Chart.
Once consented for the study, patients completed self-administered measures including a sociodemographic survey, as well as the Hospital Anxiety and Depression Scale (HADS),59 the Iowa Criteria Goals of Care survey,60 and a short survey documenting any prior ACP activities. We then randomized patients to either the intervention video or the control video. In the presence of study staff, patients immediately viewed the video to which they were randomized. This occurred either in a private clinic room or in the clinic waiting area via laptop and headphones. The study staff also provided the patient a web link to the video so that the patient could show the video to others in his or her family or view it again.
This visit occurred upon the patient's next appointment with the surgeon, often 1 or 2 days before the actual surgery. When the patient arrived in the surgeon's clinic waiting room, study staff greeted the patient, offered to show the patient the video again, and orally consented any companions accompanying the patient. Per direction from the Johns Hopkins Hospital IRB, oral consent consisted of a brief description of the study and the companion's agreement to be audio-recorded during the visit with the physician. Once staff escorted the patient to an examination room, the patient–surgeon visit was audio-recorded; the express purpose of the visit was for the patient and surgeon to discuss the risks and benefits of the surgery and for the patient to sign surgical consent. If patients had companions/family members present, their communication during the visit was also audio-recorded. This audio-recording was used for the primary outcome RIAS analysis (described in further detail below). Immediately after completing the audio-recording and leaving the examination room, surgeons completed a satisfaction survey. The patient completed a satisfaction survey, as well as the HADS survey, the Iowa Criteria Goals of Care survey, and a Helpfulness of the Video survey in the examination room before departing the visit. All outcome measures are described in further detail below.
Approximately a week after the surgery, a study staff member met with patients while they were still in the hospital. Patients completed the HADS survey and Iowa Criteria Goals of Care survey.
Approximately a month after surgery, a study staff member communicated with patients either in person during the patient's 1-month clinic visit with the surgeon or over the phone. The patient completed the HADS survey, the Iowa Criteria Goals of Care survey, and a short survey documenting any prior ACP activities.
A primary outcome was the surgeon–patient conversation as analyzed using the RIAS (see “Analytical and Statistical Approaches” section).
A number of content-specific markers of clinical proficiency related to communication about ACP were determined from a prior study29 and a review of initial recordings by study team members. These specific codes related to ACP and included medical decision maker designation (code of greatest interest to us), immediate death at time of surgery, long-term death after the surgery, severity/significance of the planned surgery, and goals for the surgery.
The secondary outcome of HADS uses 2 subscales to determine symptoms of anxiety and depression. Each subscale, consisting of 7 questions, results in a score ranging from 0, indicating no distress, to 21, indicating maximum distress; a score higher than 7 indicates clinically meaningful anxiety or depression.59 Overall HADS scores, encompassing both subscales, result in a total score ranging from 0 (no mood symptoms) to 42 (maximal mood symptoms).59 HADS is a patient-reported outcome.
The Iowa Goals of Care survey60 includes 7 non–mutually exclusive goals for why patients seek out medical care: “Cure my medical condition”; “Allow me to live longer”; “Improve my current health”; “Maintain my current health”; “Allow me to accomplish something particular in my life, a personal life goal”; “Make me comfortable”; and “Learn more about my disease or health.”
The survey itself has 2 questions. The first question asks patients to check which of the 7 goals relates to their surgery. The second asks patients to rank their top 3 goals. Goals of care data were collected at all 4 visits. The Iowa Goals of Care survey is a patient-reported outcome.
Patients completed a 3-item measure of their perceptions of the helpfulness of the video. Volandes et al used this measure in previous studies but did not report on the psychometric properties of the tool.31,33,39-41,43 This measure asks whether the patient was comfortable watching the video, whether the patient perceived the video to be helpful in preparing for surgery, and whether the patient would recommend the video to other patients. The Helpfulness of the Video survey is a patient-reported outcome.
After the audio-recorded conversation (V2), surgeons and patients each completed a short self-administered satisfaction questionnaire about the visit. The study team adapted measures developed and used by Roter et al in previous studies to address patient satisfaction with interpersonal and informational aspects of medical visits.61-64 In past studies, an 8-item patient satisfaction questionnaire had a Cronbach α of .89, and a clinician satisfaction questionnaire had a Cronbach α of .83.61-65 The satisfaction score, as the sum of the scores of 6 questions (all in Likert scale), ranges from 6 to 30, with a higher score indicating higher level of satisfaction. These satisfaction surveys involved a patient-reported outcome and a clinician-reported outcome.
At the baseline visit and postsurgical 1-month visit, patients completed a short survey documenting any prior ACP activities. The survey consisted of 2 questions regarding whether the patient had informally or formally designated a surrogate medical decision maker and, if so, how recently the patient had a conversation with that person about care preferences. The outcomes were an ordered categorical variable consisting of 4 possible answer options: (1) “No, I don't have a medical decision maker”; (2) “Yes, I have a medical decision maker, but we have not talked specifically talked about this [my care preferences]”; (3) “Yes, I have a medical decision maker, and our talk about this [care preferences] was over 6 months ago”; and (4) “Yes, I have a medical decision maker, and our talk about this [care preferences] was within the last 6 months.”
We made every effort to follow up with patients for the 3 subsequent data collection points in the study following the baseline visit. To ensure data collection at the presurgical consent visit, we coordinated with surgeons' scheduling staff and informed them of any enrolled patients. As an added measure, a member of the research staff also monitored the electronic schedules of all actively enrolled patients daily. In addition, one of our recruitment staff also gave patients a reminder call the night before their surgical consent appointment. For the postsurgical visits, surveys were conducted in hospital rooms, at follow-up visits with surgeons, or over the phone. The goal of these safeguards was to reduce attrition and missing data.
RIAS is a quantitative coding system for medical dialogue, which has demonstrated reliability and predictive validity for patient satisfaction, utilization, and adherence.55,66-70 This study used as a coding unit of analysis a complete thought that varied in length from a single word to a sentence. RIAS captured key proficiencies of content related to the study, including when a patient mentioned the designation of a medical decision maker or the study video. RIAS coders underwent rigorous training over 3 days covering all aspects of the RIAS coding process.71 RIAS coding has a reliability of >0.85 in most studies based on Pearson correlation coefficients.55 The study had 1 coder, who has demonstrated intercoder reliability on past RIAS studies. Audio-recordings were coded directly without transcription.
Because patients and family members noted wanting to have a “meaningful,” psychosocially focused conversation with their surgeon, we used RIAS to calculate the patient centeredness of the surgeon–patient interaction. Using RIAS, patient-centeredness can be calculated as a summary score, which previous studies have used with predictive and concurrent validity for a variety of patient and physician outcomes.73-80 The patient-centeredness summary score is a ratio of statements that reflect the psychosocial and socioemotional elements, rather than the biomedical and disease-focused elements, of the patient's lived illness experience. This score reflects the encounter as a whole, rather than an individual's dialogue. A value greater than 1 indicates a more patient-centered encounter emphasizing psychosocial and socio-emotional communication, whereas a value less than 1 indicates a less patient-centered and more biomedical encounter.
We based the sample size calculation on a measure of patient centeredness generated from the RIAS. The study was powered based on 2 previous studies that utilized RIAS29,72 (both powered = 0.8 and α = .05). This measure incorporates the verbal contributions of patients, surgeons, and companions. Steinwachs et al65 used this patient-centeredness variable as the primary outcome in a study testing the effectiveness of a 20-minute computer-based intervention to activate patients to address quality of care with their providers.65 The intervention group experienced visits with significantly higher levels of patient-centeredness than the control group, with an effect size of 0.6 (Cohen d).65
With a 0.6 effect size, the required sample size is 72 patients (36 per group) for a 1-tailed test of study hypotheses (power = 0.8 and α = .05). The study team determined that only a 1-tailed test was necessary given that we were testing whether the intervention improves patient-centered communication. Based on the previous study,65 we hypothesized that we would obtain recordings for 80% to 90% of recruited patients, with any discrepancies likely stemming from patient attrition, technology failure, and/or scheduling miscommunication. Accounting for an 80% recordings rate, the study team determined we would need to recruit 90 patients to obtain the desired number of 72 recordings.
The team set the overall level of statistical significance at P < .05. We performed all statistical analyses in Stata statistical software81 and reran the analysis in R statistical software to confirm results.82
The team calculated descriptive statistics to summarize patient characteristics and other baseline variables. We assessed comparability of the intervention arm and the control arm regarding preintervention sociodemographic and health status measures. For baseline demographic variables, we performed a 2-sample t test/Mann-Whitney test to further investigate any potential differences in 2 means or medians for continuous variables or used a Fisher exact test or chi-square test to investigate any difference in proportions for binary or categorical variables.
Further statistical analyses explored the association between intervention assignment and each of the outcomes. Based on the type of the data, the study team determined summary univariate (descriptive) statistics (mean, standard deviation, median, interquartile range, max, min, count, percentage) of all outcomes stratified by intervention assignment. We tested differences in outcomes between the 2 arms at each visit by 2-sample t test/Mann-Whitney test or Fisher exact test/chi-square test, based on the data types of the outcomes.
For the primary outcome and some of the secondary outcomes, we followed the descriptive statistical analyses with regression analyses, using generalized estimating equation models. To address the potential unmeasured influence of surgeon-level attributes on patient-level outcomes, we modeled the variable “surgeon” as a covariate. In many cases, the parameter of interest was the coefficient of the arm indicator, to be estimated as the intervention effect.
Due to the HADS scores being collected longitudinally as a continuous variable, we evaluated scores using generalized estimating equations. These models assessed the difference in HADS scores between the 2 arms at V2, V3, and V4. We incorporated 3 models and report on model 3 as our primary outcome. Model 1 included a term that accounted for both randomization assignment and visit. Model 2 included this same predictor, as well as surgeon, treated as a categorical variable, as a covariate. Model 3 includes the same predictor, as well as surgeon and the covariates of patient age, gender, education, and race as predictors. We ran a linear combination of estimators tests to evaluate any significant findings from the models.
The Johns Hopkins School of Medicine and the Sidney Kimmel Comprehensive Cancer Center IRBs reviewed and approved the study protocol. A data and safety monitoring board was convened and reviewed study results approximately halfway through the study period.
With a target enrollment goal of 90 patients, we enrolled a total of 92 patients in the study (see Figure 4 for the CONSORT diagram). The most common reasons for patients declining to participate included disinterest in participating in research, time restraints, or opting to have surgery at another hospital. In total, 74 patients completed the trial (see Figure 4 and the section Approach to Missing Data for details regarding patient attrition). We implemented the study protocol as planned. We made no significant changes to the approved research plan through the course of the clinical trial.
CONSORT Diagram.
We randomized 45 patients to the intervention video, and 47 to the control video (Table 3). There were no significant differences between the intervention and control groups on basic demographics, including age (P = .503), sex (P = .554), race (P = .756), and primary diagnosis (P = .749). The average age of participants was 60.4 years old, and the study population was 86% White, 11% African American, and 63% female. The most common primary diagnoses were pancreatic cancer (31.5%) and hepatobiliary/other gastrointestinal cancers (27.2%).
Baseline Characteristics of Study Patients.
There was no significant difference in the patient-centeredness of the visit comparing patients who saw the intervention video with those who saw the control video (incidence rate ratio = 1.06; CI, 0.87-1.30; P = .545; Table 4A). Similarly, looking at only visits in which a companion was present shows there was no significant difference comparing those who saw the intervention video with those who saw the control video (P = .487; Table 4A). The mean patient centeredness of visits in both study groups was greater than 1 (1.06-1.09), indicating conversations that slightly favored psychosocial and socioemotional, as compared with biomedical, communication.
Measured Patient Centeredness in the Presurgical Consent Visit (Study Primary Outcome).
Pearson chi-square Fisher exact tests revealed no significant differences in the mentions of any of the proficiencies between intervention and control groups (Table 4B). However, because 23% of conversations in the intervention group addressed advance care planning, compared with 10% of conversations in the control group, there was a non–statistically significant increase in the likelihood of discussing ACP after having viewed the intervention video (P = .182).
Measured ACP Content in the Presurgical Consent Visit (Study Primary Outcome).
HADS scores at visits before surgery remained stable (V1 and V2), peaked at the postsurgical 1-week visit (V3), and then dropped again at the postsurgical 1-month visit (Table 4C and Figure 5).
HADS Scores Across Study Arms Throughout the Study Period (Secondary Outcome).
HADS Score Variation Throughout the Clinical Trial Time Period.
In model 1, a generalized estimating equation that accounted for both randomization assignment and visit demonstrated that HADS score, and consequent anxiety and depression, was significantly greater for people at V3 in the control group than at baseline (coefficient = 3.364; P = .007). Evaluating a linear combination of estimators revealed no significant difference in HADS scores in V3 between intervention and control (P = .296). Based on these data there were no statistically significant differences in HADS scores between intervention and control in any time point in the study.
In model 2, when surgeon was considered as a covariate, the HADS score was significantly greater for people at V3 (1) in the control group than at baseline (coefficient = 3.155; P = .009), and (2) in the intervention group than at baseline (coefficient = 2.398; P = .049). When considering surgeon as a covariate, the difference in HADS scores in V3 between intervention and control group remained not significant (P = .604); thus, the data support that surgeon identity did not impact HADS score at V3.
Controlling for age, race, education, and gender in model 3, the HADS score was again significantly greater for people at V3 (1) in the control group than at baseline (coefficient = 3.05; P = .007), and (2) in the intervention group than at baseline (coefficient = 2.95; P = .010). When considering surgeon and other demographics as covariates, there was no significant difference in HADS scores in V3 between intervention and control group (P = .941); consequently, based on these data, HADS score did not statistically vary among the patients of any particular surgeon or within any demographic subpopulation.
Participants in both intervention and control groups at all 4 time points in the study most frequently selected the goal “Cure my medical condition” (Table 4D). The same 3 goals were most frequently selected across all time points and in both groups: “Cure my medical condition” (94% checked), “Allow me to live longer” (94% checked), and “Improve my current health” (85% checked; Figure 6).
Iowa Goals of Care Across Study Arms Throughout the Study Period (Secondary Outcome).
Iowa Goals of Care Goal Selection Variation Throughout the Clinical Trial Time Period.
To evaluate the importance of these different goals to patients, we examined the number of instances when a patient ranked the goal as 1 of their top 3 most important. Across visits, when asked to rank goals by importance, participants selected “Cure my medical condition” most frequently as a top 3 goal (in 268 instances), followed by “Allow me to live longer” (256 instances in the top 3) and “Improve my current health” (195 instances in the top 3).
For each of the questions probing the helpfulness of the video, we ran chi-square and Fisher exact tests to examine the 4-item categorical variable. When participants were asked if the video helped prepare them for surgery, more people in the intervention group, who saw the intervention video about ACP planning, reported it as “very helpful” or “somewhat helpful” than in the control group, who saw a video about the Johns Hopkins surgical program (P = .007; Table 4E).
Helpfulness of the Video Across Study Arms, Collected Immediately After the Audio-Recording at the Presurgical Consent Visit (Fisher Exact Test, P = .007; Secondary Outcome).
The distribution of ratings probing whether participants felt comfortable with the video was similar in the 2 study arms (P = .908; Table 4F), as was whether participants would recommend the video (P = .526). However, more people in the intervention group said would “definitely recommend it” (Table 4G).
Comfort With the Video Across Study Arms, Collected Immediately After the Audio-Recording at the Presurgical Consent Visit (Fisher Exact Test, P = .908; Secondary Outcome).
Recommendation of the Video to Others Across Study Arms, Collected Immediately After the Audio-Recording at the Presurgical Consent Visit (Fisher Exact Test, P = .526; Secondary Outcome).
Both patients and providers reported very high satisfaction scores after the audio-recorded presurgical consent visit (Table 4H).
Patient and Provider Satisfaction Scores Across Study Arms (Secondary Outcome).
At baseline, most patients in both intervention (80.0%) and control (76.6%) groups had designated a decision maker (Table 4I). Rates of medical decision maker designation increased in both groups between baseline and the postsurgical 1-month visit. We ran 2 multinomial logistic models to evaluate medical decision maker designation and conversation. The first included randomization as the predictor, and the second added surgeon as a covariate. Neither model demonstrated any significant differences between control and intervention groups in their category of medical decision maker designation and conversation (Table 4J); consequently, the data support the conclusion that neither the intervention nor the control video affected whether the patient designated a medical decision maker or had a conversation about advance care planning wishes.
Prevalence of Participants Who Acknowledge Having Named a Surrogate Decision Maker Across Study Arms Throughout the Study Period (Secondary Outcome).
Prevalence of Participants Who Acknowledge Having a Conversation With Their Surrogate Decision Maker Regarding Advance Care Planning Across Study Arms Throughout the Study Period (Secondary Outcome).
Due to the nature of the patient population of interest in this study, there were several reasons for attrition after viewing the video (see Figure 4). Ten patients did not maintain eligibility because they did not have surgery (and therefore did not have a surgical consent visit) within the study timeline or elected to undergo surgery with a surgeon who was not participating in our study. Additionally, 2 patients withdrew due to the emotional burden of their medical course and surgery. One patient was lost to attrition after the patient died before undergoing surgery. Four patients were lost to follow-up. There were no reported adverse events in the trial.
Of all 92 enrolled patients, a total of 18 patients (19.4%) did not complete the study, rendering the patients with missing data (Figure 4). To examine whether there were differences between the patients who completed and did not complete the study, we assessed bivariate analyses comparing study completion and demographic characteristics. We observed no differences between patients who completed and did not complete the study between any of the background characteristics captured throughout the trajectory of the study period, which included gender, race, ethnicity, education, age, having a previous ICU stay, having a previous surgery, which surgeon the patient was receiving surgery from, primary diagnosis, presence or absence of caregivers in the surgical consent visit, complication grade following surgery, and randomization assignment. Because we observed no differences between patients who completed and did not complete the study, we used all collected data for analysis in the present study.
Owing to several factors, including the need for emergent surgery, the decision to not undergo surgery, and the decision to undergo surgery at another hospital, not all enrolled patients had a presurgical consent visit during the study period. We included in the data analyses all preoperative consent visits recorded at least 1 day before surgery, including for patients who decided against surgery, for a total of 61 recordings.
From study inception, we planned to complete exploratory analyses for heterogeneity of treatment effects. This was essentially a negative trial. Even in subanalyses by gender, race/ethnicity, or surgeon, there were no statistically significant differences between study arms in either the primary outcomes or most of the secondary outcomes. We did see some differences in primary and secondary outcomes among different patient subgroups in our analysis of HADS score. Compared with White participants, Asian and African American participants scored significantly differently (coefficient = 9.42, P < .001; and coefficient = −2.45, P = .021, respectively), with Asian patients noting worse mood symptoms and African American patients noting better mood symptoms. When considering education, the HADS scores across levels of education were all different, although there was not a score directionality associated with more or less education. Men also scored significantly lower on HADS (indicating better mood symptoms) than women (coefficient = −3.34; P < .001).
Our study supports that an ACP video can be successfully integrated into preoperative care and that the video was thought “helpful” by participants.
However, the video did not change the nature of the surgeon–patient conversation (as measured through the RIAS patient-centeredness) or the content of ACP in that conversation. The ACP, compared with a control video, was not associated with changes in secondary outcomes, although participants were more likely to find the intervention video “very helpful” or “somewhat helpful” than “a little helpful” or “not helpful.” In the secondary outcomes, we found that patients have worse mood symptoms immediately following surgery (during the V3 visit), that participants were pursuing surgery primary with the goal of seeking a “cure” for their cancer, and that after surgery participants were more likely than before surgery to have named a surrogate decision maker and to have had a conversation with that person.
Our study results do not support that incorporation of the study video would change preoperative ACP. On review of study results, partner surgeons noted that they and many of their colleagues have an informal and highly individual mental “script” for how they consent patients; the data support the conclusion that the patients and family members who saw the intervention video do not seem to have changed that surgeon “script.” Yet, particularly because participants thought the video was helpful, future studies might explore how reframing the video (ie, presented by a palliative care clinician and/or as a starting point for a discussion) could change the impact of the video.
The decisional conflict is whether to incorporate ACP in the preoperative period. When we proposed this study in 2012, relatively little ACP research had been completed in surgical patient populations and this topic was considered highly novel (if not antithetical in some circles).83-88 Since then, ACP and palliative care as a topic and a specialty, respectively, have become more accepted and prevalent, even in surgical populations. A recent systematic review of the research about these topics presents a research agenda that specifically calls for more future research in these topics.18,89
We completed our study at an academic tertiary care center and there is good reason to expect study results would be generalizable to other cancer center surgical programs. Future studies would likely best be aimed to vary the ACP “dose” in the video by changing video content and/or how it is presented and discussed with patients and family members.
Particularly as the recent systematic review18 and research agenda89 concerning palliative care for surgical populations call for more ACP and palliative care studies in surgical populations, we found that a preoperative ACP/palliative care–focused intervention was safe. No incidents of harm occurred in the study and no patients in the intervention arm noted antipathy toward, or distress related to, the ACP content; if anything, patients felt the video with ACP content was “more helpful.” Thus, our study supports that patients do not have reservations against preoperative ACP or palliative care and, if anything, value it.
Our study has several limitations. First, the selected outcomes and time frame of the study may not be able to fully capture the effect of the intervention because the impact of the surgery and video may persist beyond the 1-month time frame of the study. Because other studies have shown benefit of ACP discussions as far as 12 months after hospitalization and patient death,90 we might hypothesize further benefits of the intervention to become apparent just before and after patient death, which is outside of the trial time frame.
Second, surgeon-level factors may have influenced study outcomes. All surgeons were privy to a general overview of the study and were given the opportunity to watch the intervention and control videos before agreeing to participate; therefore, the surgeons who ultimately decided to participate in the study may have been biased in their preexisting support for ACP. It is also possible that surgeons may have had their own unconscious selection biases when referring patients to the study. Of note, on poststudy review of the data, our surgeon partners note that many surgeons have an informal “script” that they follow when consenting patients for surgery, but that no standard exists other than to clearly communicate and answer questions regarding the risks of benefits of the surgery as detailed in the consent form.
Moreover, in preparation discussions for this study as well as in the study team's experience from a past study regarding communication for surgical patients, surgeons noted strong resistance to structured communication prompts.91 Knowing that patients were in this trial, regardless of which study arm they were randomized to, surgeons may also have altered their communication regarding ACP. This could not be avoided, as all surgeons needed to see both videos to be comfortable enrolling their patients in this trial. However, surgeons did not appear to alter their ACP-related communication style for study patients; the audiotapes showed low prevalence of ACP discussions between surgeons and patients, particularly in the control arm.
Third, although we enrolled 92 patients in the trial, only 61 had audio-recordings for the primary outcome. Of the 31 patients who lacked a recording, 10 did not pursue surgery either with one of the study surgeons, at the study institution, or for management of their cancer; 2 voluntarily withdrew from the study; and 1 died before surgery (Figure 4). The other 18 patients who lacked a recording were split between the control (8 patients) and intervention groups (10 patients); the lack of a recording was due to the patient or companion turning off the recorder (2 patients), technical difficulties with the recorder (2 patients), or the surgical team consenting the patient outside the protocol of the clinical trial—either on the day of surgery or without study team members or equipment available (14 patients).
Due to unpredictable changes in surgery scheduling and/or patient disease, the 14 patients consented outside the protocol of the clinical trial were taken to surgery at different times than initially planned; surgery was never delayed to accommodate our study team or the audio-recording of a consent conversation. Early in the clinical trial and in an attempt to capture these data despite unpredictably stochastic changes in surgical scheduling and/or patient illness, we audio-recorded 2 consents that were performed on the morning of surgery. However, on immediate review, it was clearly apparent that these recordings were fundamentally different from those obtained before the day of surgery. Within the study protocol, consent conversations occurred in the surgeon's outpatient clinic at least a day, if not multiple days, before the actual surgery. In contrast, on the day of surgery, surgical consent conversations occurred in preoperative area cubicles with the patient in a surgical gown, the patient's intravenous line for surgery in place, and with just minutes to go before the patient was taken back to the operating room. Perhaps not surprisingly and compared with consent conversations obtained per study protocol, the 2 consent conversations recorded on the day of surgery were markedly shorter in duration, with patients asking the surgeon minimal to no questions. While future studies could examine the differences in surgical consent obtained on the day of surgery compared with an earlier date, we did not consider the 2 equivalent. After detailed discussion with our entire study team, including our patient, surgeon, surgical nurse, and palliative care co-investigators (J.M., M.W., T.P., S.S., and T.S.), we specifically chose to analyze only conversations obtained per study protocol and before the actual day of surgery. Altogether, as the 31 patients who lacked an audio-recording were split nearly evenly between the control and intervention group, we did not expect these data to have likely changed study results.
A final limitation of the study is that it cannot control for the effect of a patient's medical course on study outcomes. Both presurgical factors such as diagnosis, and postsurgical factors such as surgical course or change in prognosis, might contribute to anxiety and depression, as well as to a patient's goals of care.
As discussed, future research in palliative care and ACP in surgical populations is clearly needed and being called for.48,89 This study shows the safety and feasibility of incorporating ACP into the surgical care of cancer patients. Future studies should focus on developing and testing palliative care–related outcomes that matter to surgical patients; developing and testing effectiveness of communication-based interventions with a goal of better aligning surgical treatments with patient goals and values as well as better communicating prognosis; and developing new ways to integrate palliative care into the care of surgical patients, specifically in the “routine management of seriously ill surgical patients” as well as through testing “early vs late palliative care” for surgical patients.
We successfully integrated an ACP video into the preoperative care of cancer patients pursuing major surgery. No documented harms were associated with the intervention. While patients thought the ACP video was more “helpful” than the control video and there was a trend for more ACP content in surgeon–patient conversations among patients randomized to the ACP video, the ACP video did not change the patient-centeredness of the patient–surgeon discussion nor significantly change the content of ACP discussed between patients and surgeons before surgery. Secondary outcomes did not support any difference between the study arms, although together they support that cancer surgery patients have worse mood symptoms immediately following surgery and that most were pursuing surgery with goals to “cure” cancer and to help them “live longer” and “improve their current health.”
This study shows that palliative care content, such as ACP, can be successfully incorporated into surgical care and that patients think it is “helpful.” Given surgeons' documented resistance to incorporating palliative care and related topics into presurgery discussions, this successful integration shows that further studies—perhaps with more concentrated “doses” of ACP or palliative care—are safe, acceptable to patients and family members, and needed.
Research reported in this report was [partially] funded through a Patient-Centered Outcomes Research Institute® (PCORI®) Award (#CD-12-11-4362 Further information available at: https://www.pcori.org/research-results/2013/does-advance-care-planning-video-help-patients-having-surgery-cancer-discuss
Aslakson RA, Isenberg SR, Crossnohere NL. et al. (2019). Does an Advance Care Planning Video Help Patients Having Surgery for Cancer Discuss Their Healthcare Preferences with Their Surgeons? Patient-Centered Outcomes Research Institute (PCORI). https://doi.org/10.25302/4.2019.CD.12114362
The [views, statements, opinions] presented in this report are solely the responsibility of the author(s) and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute® (PCORI®), its Board of Governors or Methodology Committee.
This book is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License which permits noncommercial use and distribution provided the original author(s) and source are credited. (See https://creativecommons.org/licenses/by-nc-nd/4.0/
Your browsing activity is empty.
Activity recording is turned off.
See more...
