Prognostic Conversations Between Parents and Physicians in the Pediatric Intensive Care Unit

The study took place from September 2019 through August 2020 at a 40-bed quaternary care PICU in Chicago, Illinois. A detailed description of our methods and surveys was published previously.³  The hospital’s institutional review board approved the study with a waiver of signed consent.

Study participants included parents of PICU patients less than 18 years old who demonstrated multiorgan dysfunction syndrome according to the Goldstein criteria within 24 hours of PICU admission; participating parents also completed an initial survey 5 to 10 days after PICU admission.¹⁷  Parent refers to a biological parent, foster parent, or legal guardian with decision-making authority. One parent per patient was invited to participate. Exclusion criteria for parents were inability to communicate in English or Spanish, participation during a previous PICU stay, indication from a member of the health care team that the parent’s participation would be inappropriate, child who died in the PICU, and involvement of the Department of Child and Family Services. We also included the discharging PICU attending physicians caring for the patients whose parents completed surveys.

We invited parents to complete a survey at the time of their child’s discharge from the PICU. Surveys (developed as described previously) focused on conversations between health care team members and parents about the patient’s prognosis related to PICU mortality and post-PICU physical, cognitive, and psychological morbidities.³  Surveys also included demographic information and asked whether the parents had conversations with health care team members about their own post-PICU psychological morbidity. For each topic, we asked parents about the occurrence of prognostic conversations, what was discussed, with whom the discussion occurred, and whether they wanted more information about their child’s prognosis or their own outcomes. Parents who indicated that they talked with the health care team about the possibility that their child might die were also asked if they wanted more information about their child’s possible mortality. Physician surveys, which were sent by email to PICU attending physicians within 24 hours of receiving a parent survey, included parallel questions related to the same topics. Study surveys (in English and Spanish) were created and administered by using Research Electronic Data Capture (REDCap), a secure web-based application designed to support data capture for research studies.¹⁸  Patient demographic and clinical information was obtained from the electronic medical record.

We have previously published a full description of our statistical analysis.³  In brief, we used descriptive statistics to present parent and physician information about reported prognostic conversations for each topic, demographics, clinical information, and whether parents wanted more prognostic information. We used an adapted κ statistic to evaluate concordance between parent and physician reports about the occurrence of prognostic conversations for each topic and any prognostic outcome.

During the study period, we screened 812 patients; 137 met the inclusion criteria. Of 110 parents approached for the initial surveys conducted 5 to 10 days after PICU admission, 101 submitted analyzable surveys (initial survey response rate: 92% of parents approached, 74% of parents who met the inclusion criteria) and 49 completed a survey at PICU discharge, a 49% retention rate (see Figure). Complete results of the initial surveys have been published previously.³  Table 1 shows patient and parent demographics for the initial and discharge surveys. The discharge surveys indicated that most patients were admitted with a respiratory-related diagnosis, had 2-system organ dysfunction, and were transferred to another service in the hospital upon PICU discharge. Parents who completed discharge surveys were most commonly White, Hispanic or Latinx; educated through the high school or college level; and most commonly had an income of less than $50 000. Notable demographic differences between initial survey and discharge survey demographics were a decrease in the percentage of patients with a new do-not-attempt-to-resuscitate order during the PICU stay (from 12% to 4%) and an increase in the percentage of parents who identified as Hispanic or Latinx (from 34% to 69%).

Twenty PICU attending physicians completed surveys for the 49 patients (some physicians completed surveys for more than 1 patient). Eleven physicians (55%) were female; 7 (35%) had completed fellowship training less than 5 years earlier; 6 (30%) had completed fellowship training 5 to 10 years earlier; and 7 (35%) had completed fellowship training more than 10 years earlier.

At PICU discharge, 30 parent surveys (61%) and 29 physician surveys (59%) reported any prognostic conversations. Parent and physician reports of prognostic conversation occurrence for each topic are shown in Table 2. More than 50% of parents reported not having prognostic conversations related to each of the topics. According to parents, most conversations that did occur were regarding post-PICU physical morbidities. Greater than 70% of physician surveys indicated that the physician did not know if conversations related to parent or patient psychological morbidity occurred. More attending physicians than parents reported not knowing about prognostic conversation occurrence.

Concordance between parents and physicians about reported prognostic conversation occurrence related to any topic was slight (κ = 0.19; 95% CI, −0.16 to 0.46). For each outcome, parent and physician concordance was as follows: mortality, 0.15 (95% CI, 0.15-0.43); physical morbidities, 0.06 (95% CI, −0.21 to 0.28); cognitive morbidities, 0.13 (95% CI, −0.13 to 0.32); patient psychological morbidities, −0.09 (95% CI, −0.23 to 0.13); and parent psychological morbidities, −0.1 (95% CI, −0.3 to 0.13).

Table 3 shows how parent responses regarding prognostic conversation occurrence changed from the initial survey to the discharge survey. Depending on the topic, 5% to 28% of parents did not recall prognostic conversations that they had previously reported. Of parents who reported no prognostic conversations about patient post-PICU cognitive and psychological morbidities on the initial surveys, 60% and 75% of parents, respectively, noted the occurrence of conversations on these topics on discharge surveys.

The Supplemental Table (available online only at www.ajcconline.org) shows the specific topics discussed during prognostic conversations. This table includes free-text responses from parents who selected “other” and wrote a response.

Table 4 shows data from parent surveys indicating which health care professionals initiated prognostic conversations. Parents reported receiving most prognostic information from physicians or nurse practitioners; however, parents noted that bedside nurses were more often involved than physicians or nurse practitioners in conversations pertaining to psychological morbidities for the patient and parent. Social workers were the professionals most frequently involved in conversations about parent psychological morbidities.

Twenty-six parents (53%) requested more information about any topic. Table 5 shows parents’ responses about wanting more information according to whether they reported that a prognostic conversation occurred. Of the parents who indicated that prognostic conversations about each outcome topic did not occur, 63% reported wanting more information about potential physical morbidities; 46%, cognitive morbidities; and 32%, patient psychological morbidities. For topics that were discussed, 18% to 31% of parents still wanted more prognostic information.

Emerging data about high rates of PICS support the need for pediatric intensivists to engage in prognostic conversations with families.^1,3  This prospective study describes reports about prognostic communication from parents of PICU patients and PICU attending physicians at the time of PICU discharge. We found that only 61% of parents and 59% of physician surveys indicated that prognostic conversations occurred during the PICU stay for patients at high risk of developing PICS. The topic for which parents and physicians reported the most prognostic conversations was patient post-PICU physical morbidities (45% and 47%, respectively). Parents least frequently reported discussions related to mortality (18%) and cognitive morbidities (20%). Physicians least frequently recalled discussions about patient and parent psychological morbidities (14% and 22% of surveys, respectively). Parents noted that most prognostic conversations included physicians and nurse practitioners; however, parents reported that bedside nurses often discussed patient psychological morbidities, and bedside nurses and social workers often discussed parent psychological morbidities. Fifty-three percent of parents wanted more prognostic information. Our results highlight opportunities to improve PICU prognostic conversations.

Fewer parents reported prognostic conversations on discharge surveys (61%) than on initial surveys (88%), as reported previously.³  Although part of this difference may reflect sampling bias, some of this difference reflects a change in parent report. Depending on the topic, up to 28% of parents who had reported a prognostic conversation occurrence earlier in the PICU stay indicated at discharge that either those conversations had not occurred or that they did not remember them. Additionally, some parents who had received prognostic information wanted more information. Experts describe the importance of repeating information over multiple conversations, particularly when talking with patients or families in stressful situations. Providing information in multiple formats helps meet the diverse needs of parents with varied learning styles.^19-21  These findings underscore the need to repeat information. Currently, no protocol exists to guide prognostic conversations with families; future work should include creating and evaluating such a communication protocol. These data also highlight the importance of revisiting previous conversations with parents if relevant for the patient or family or at least asking parents if they would like more information about a relevant topic.

Some parents who reported prognostic conversations at the time of discharge did not report these conversations on the initial surveys. This result may reflect parents underreporting conversations 5 to 10 days after admission (further reinforcing the need to repeat information) or parents reporting conversations that occurred later during the PICU stay. No rules or recommendations guide when such conversations should occur. In many cases, clinicians may wait to have prognostic conversations because they do not anticipate serious morbidities or they do not have enough information to provide meaningful predictions. However, previous data show that parents want more prognostic information, even early in a PICU stay.³  The parents’ expressed desire for more prognostic information suggests that research is needed to better understand when clinicians should communicate information about prognosis to parents.

Parents’ desire for more prognostic information is reflected in our finding that 53% of parents requested more information about some topic. Parents requesting more information is a ubiquitous finding in studies of pediatric patients regarding prognosis discussions and difficult conversations, highlighting that clear communication helps alleviate parental anxiety.^22-27  We posit multiple explanations for why clinicians may not engage in prognostic conversations: prognostic uncertainty, discomfort in communicating prognosis, inadequate training in prognostic communication, and perception that the patient’s clinical trajectory does not warrant such discussions. Knowledge about the prevalence of PICS is mostly related to physical morbidities^3,9  and is still emerging, making it challenging for clinicians to prognosticate. This situation may explain why parents most commonly reported they had prognostic conversations about patients’ physical morbidities. As more information about PICS becomes available and enhances clinicians’ ability to prognosticate, understanding barriers to clinician engagement in prognostic conversations with parents will become increasingly important.

Other key findings included the following: (1) most surveys of attending physicians indicated that they did not know whether prognostic conversations occurred, (2) concordance between parent and physician surveys was poor, and (3) bedside nurses and social workers were more likely than other medical professionals to discuss psychological morbidities. These findings suggest opportunities and a need to improve team communication. Medical professionals in the PICU are a multidisciplinary group of clinicians who change frequently, introducing challenges to team communication.^3,28  At the institution where this study took place, attending physicians and fellows rotate at least once per week, and advanced practice nurses and bedside nurses rotate more frequently. Information sharing among clinicians from different disciplines is also not consistent; for example, social workers or bedside nurses may not convey all information about their communication with families to other team members and vice versa. Medical professionals in the PICU need better approaches for sharing information among the multidisciplinary team about prognostic conversations with parents. Documentation in the medical record can support this goal, but relying on the medical record for multidisciplinary communication may not be the most efficient and effective approach. Another approach is to introduce a program like pediatric ethics and communication excellence rounds, which are weekly meetings to identify and share goals of care for children with prolonged PICU stays. A study of pediatric ethics and communication excellence rounds at a single center showed improved perceptions of team communication, decreased moral distress among clinicians, and shorter stays.²⁹ 

Results of this work should be viewed in the context of its strengths and limitations. One strength is that we collected information at PICU discharge, reflecting prognostic conversations that occurred throughout the PICU stay. However, because information was collected at discharge, our data may be subject to recall bias because the average PICU stay was almost 3 weeks. Additionally, we did not evaluate parents’ sense of preparedness for post-ICU morbidities. In future studies, researchers should evaluate family preparedness at and after discharge to compare expectations with reality. We administered the survey only to each patient’s attending physician at the time of discharge and thus cannot comment on other health care providers’ knowledge about prognostic conversations. Future studies should address prognostic conversations at the time they occurred from the perspectives of clinicians, families, and the multidisciplinary team. Our data are also subject to response bias because only 49% of parents who completed an initial survey 5 to 10 days after PICU admission also completed the PICU discharge survey. Reasons for attrition may include hospital COVID-19 restrictions limiting researchers’ ability to engage with families, survey length, repeated survey administration, or inadequate incentive. Responses to our question about whether parents wanted more prognostic information could be subject to confirmation bias; however, the varied responses to each topic potentially suggest otherwise. Future studies should retrospectively evaluate the accuracy of prognostic discussions and whether families wished they had received information differently after the patient was discharged from the PICU. Our sample size was small and included parents and physicians at only 1 hospital, potentially impacting generalizability. However, the group of parents who participated was diverse with respect to race, ethnicity, education, and income. We also did not clarify where families received information (daily rounds, family conferences, or another source).

At discharge, approximately 60% of parent and physician surveys indicated that prognostic conversations about any topic occurred, with only slight concordance. Most parents recalled conversations about post-PICU physical morbidities. Fifty-three percent of parents requested more prognostic information about some topic. The findings of this study suggest that recommendations for improving prognostic conversations should include how information is provided (and the need to repeat information), when information is provided, and how to enhance team communication about conversations with patients and their families.