Integration of Person-Centered Narratives Into the Electronic Health Record

Abstract

Background:

Incorporating a patient’s personal narrative into the electronic health record is an opportunity to more fully integrate the patient’s values and beliefs into care, thus creating opportunities to deliver high-quality/high-value, person-centered care.

Objectives:

The aim of the study was to present a study protocol of a narrative intervention to (a) compare the effects of the narrative intervention to usual care on primary outcome of person’s (patient) perceptions of quality of communication, (b) compare the effects of the narrative intervention on secondary outcomes of biopsychosocial well-being, and (c) examine the feasibility and acceptability of the narrative intervention from the perspective of both persons: the patient and the acute care bedside nurse.

Methods:

A randomized control trial is being conducted with a targeted enrollment of 80 patient participants and 80 nurse participants. The patient participants include individuals who are admitted to the acute care hospital for either heart failure or end-stage renal disease. An acute care beside nurse who has cared for the patient participant is also enrolled. Through a 1:1 random allocation scheme, stratified by illness, we will enroll 40 in the narrative intervention group and 40 in the usual care group. Patient participants will be assessed for patient-reported outcomes of patient’s perception of quality of communication and biopsychosocial well-being.

Results:

The study began in October 2019; 53 potential patient participants have been approached, 21 have enrolled, and 20 have completed the data collection process.

Discussion:

The testing and integration of a person-centered narrative into the electronic health record is a novel approach to provide opportunities for improvement in communication between patients and nurses. The results from this study will provide important preliminary knowledge to inform future randomized clinical trials of narrative interventions leading to advances in how to best provide high-value, high-quality, person-centered care for persons living with serious illness.

The Institute for Healthcare Improvement (IHI) defines “person- and family-centered care” as “putting the patient and the family at the heart of every decision and empowering them to be genuine partners in their care” (IHI, n. d.). Since 2015, the IHI encourages clinicians to ask the individual receiving healthcare a simple question: “What matters to you?” instead of “What’s the matter?” This reframing of the clinician–person interaction orients the care toward the whole person and sheds a different light on a plan of care that opens the door for opportunities to involve the whole person (IHI, n.d.). A whole-person palliative care approach improves communication, leading to better quality of life for patients (Ferrell et al., 2018); yet, discordant care continues in part because of knowledge gaps about a person’s psychological, social, and spiritual needs (Evans & Ume, 2012). Person-centered narrative interventions can fill these knowledge gaps, yielding an increased understanding of patients’ psychological, social, and spiritual needs (Akard et al., 2013; Bingley et al., 2008; Crogan et al., 2008; Hall et al., 2011; Stanley & Hurst, 2011; Thomas et al., 2009; Wise et al., 2009) to help clinicians develop tailored palliative care interventions (Leininger & McFarland, 2002; Purnell, 2012).

The use of narrative is an effective way for patients to communicate their cultural values and beliefs (Akard et al., 2013; Chochinov, 2002; Crogan et al., 2008; Hall et al., 2011; Leininger & McFarland, 2002; Purnell, 2012; Thomas et al., 2009; Wise et al., 2009). However, patients do not always have the opportunity to discuss these values/beliefs with their clinicians. The electronic health record (EHR) is one of the primary modes of communicating information about patients to all clinicians. Therefore, incorporating a patient’s narrative into the EHR is an opportunity to more fully incorporate the patient’s cultural values and beliefs into their care. The use of illness narratives, coupled with the literature of narrative methodologies (Bingley et al., 2008; Charon, 2008; Coats et al., 2020; Kleinman, 1988; Riessman, 2008; Stanley & Hurst, 2011; Thomas et al., 2009), led to the development of our biobehavioral narrative intervention that integrates illness narratives into the EHR.

Coats et al. (2020) conducted a feasibility study to define/refine the person-centered narrative intervention that is integrated into the patient’s EHR for patients living with serious illness. Results of the feasibility study indicated that the intervention was feasible, acceptable, and beneficial for both hospitalized patients (n = 20) and their acute care bedside nurses (n = 18; Coats et al., 2020). Eighteen of 20 patient participants confirmed that they would be willing to participate in a similar type of intervention in the future. All 18 nurse participants described having positive experiences. Nurses described the intervention as “eye opening,” giving them a “different perspective about what their patients are going through emotionally and physically” and allowed them to “connect” to their patients.

Person-centered narrative interventions have been shown to increase understanding of patient’s psychological, social, and spiritual needs (Akard et al., 2013; Wise et al., 2009) and provide opportunity for patients and nurses to connect (Coats et al., 2020). Therefore, we hypothesize that the implementation of a person-centered narrative integrated into the EHR will result in enhanced quality of communication and improved biopsychosocial well-being.

The next step was to conduct a randomized study that aims to (a) compare the effects of the narrative intervention to usual care on primary outcome of person’s (patient) perceptions of quality of communication (Au et al., 2012; Curtis et al., 2004, 2018; Engelberg et al., 2006; Janssen et al., 2011); (b) compare the effects of the narrative intervention on secondary outcomes of the patients’ biopsychosocial well-being: physical function, anxiety, depression, fatigue, sleep, ability to participate in social roles/activities, pain interference and intensity (PROMIS-29 profile; Craig et al., 2014; Hays et al., 2018; Katz et al., 2017), and psychosocial illness experience (PROMIS Positive/Negative Psychosocial Impact; Bajaj et al., 2011; Biesecker et al., 2013; Carle et al., 2015; Salsman et al., 2012); and (c) examine the implementation of the narrative intervention from the perspective of both persons: the patient and the acute care bedside nurse. In this article, we present the study design and describe the study protocol for the person-centered narrative intervention.

METHODS

The design for this study is a randomized control trial. The study was approved by the local institutional review board and was a registered clinical trial prior to enrollment of participants. Upon completion of the study, results will be reported in accordance with the Consolidated Standards for Reporting Trials guidelines (Schulz et al., 2010) and clinical trials guidelines (National Institutes of Health, 2017).

Participants and Setting

Study participants (n = 160) include both patients (n = 80) and nurses (n = 80). Patients are those admitted to the acute care hospital in two serious illness groups: heart failure and end-stage renal disease, where palliative care focuses on improving quality of life for patients and families from the diagnosis of any serious illness and throughout the dynamic trajectory of that illness to end-of-life (World Health Organization, n.d.). The operational definition of serious illness for this study includes the following:

1. New York Heart Class III or IV heart failure; and/or

2. dialysis-dependent renal failure (Coats et al., 2020).

The disease criteria were chosen to identify groups of patients with a median survival of about 2 years (McMurray & Pfeffer, 2005; Steinhauser et al., 2006). Patients also must be of ages 18 years or older, able to read English, and capable of giving informed consent. Nurse inclusion criteria include those who are able to confirm verbally they were involved in the care of the corresponding enrolled patient. Participants are recruited in three units (cardiac medical, cardiac intensive care unit, and dialysis) at an inpatient academic hospital.

Study Procedures

Figure 1 depicts study flow. To enroll patient participants, research personnel complete weekly recruitment rounds, during which the charge nurse of each unit provides names of potentially eligible patients. Research personnel complete pre-screens of these patients for disease-specific study eligibility. Nurses of eligible patients are consulted to ensure it is an appropriate time to enter the patient’s room. Research personnel then meet with eligible patients to describe the study and provide the consent form and recruitment flyer for review. When a patient agrees to participate, consent is obtained (see Informed Consent, Supplemental Digital Content, http://links.lww.com/NRES/A360).

FIGURE 1.

Study recruitment and procedure flow chart.

After consent is obtained, the participant completes the demographic survey and Time 1 measures immediately or the next day, depending on their preference and/or time constraints. Once these measures have been completed, randomization occurs, and the patient is notified to which group they have been randomized. Randomization to the narrative intervention or usual care occurs at the patient level. Although intervention bias is a concern, the ability to blind in a biobehavioral intervention is not feasible.

The random allocation sequence is computer-generated, using random block sizes, an allocation ratio of 1:1, and stratification by illness group: heart failure or end-stage renal disease. Prior to the start of the study, the statistician generated the randomization scheme and uploaded it into the Research Electronic Data Capture (REDCap) database. The REDCap server executes the randomization based on this integrated scheme, to which all other research team members are blinded prior to randomization. After a patient completes Time 1 measures, research personnel enter the patient’s diagnosis and clicks the “Randomize” button, which prompts REDCap to assign the patient to a study group. The participant’s group assignment is permanent and not changeable within the participant record or in the audit log. Enrollment of nurse participants occurs in parallel with the process described above. Once a patient provides consent, research personnel notify the patient’s nurse of the patient’s enrollment and provides a description of what nurse participation would entail. If the nurse agrees to enroll, consent is obtained.

Person-Centered Narrative Intervention

The intervention in this study is a cocreated narrative whereby patients reveal their illness narrative with a research team member through an open-ended, audio-recorded interview about their illness (Table 1). During the interview, patients are prompted to convey their narrative through probing questions or statements, such as “tell me about your illness” and “tell me how your illness has affected your emotions, your relationships, and your spirituality.” These probing questions have been field-tested during prior studies (Coats et al., 2017, 2020). For patients assigned to the intervention group, the narrative collection can be conducted immediately after Time 1 or the next day. The audio recording and interview field notes are used to write the cocreated narrative. Content is organized thematically and chronologically to adhere to a traditional narrative format with the criteria that the patient’s narrative is (a) written in the patient’s first-person voice, (b) nonjudgmental, (c) captures the patient’s voice, (d) accurately reflects the content of the interview, and (e) is nondiagnostic (not labeling; Riessman, 2008). The goal is to include much of what the patient discussed in the interview—while curating it to a length that can be read in 5 minutes or less—in order to facilitate engagement of busy hospital staff. The cocreated patient narrative is then returned to the patient within 48 hours. At this time, the research personnel conduct a “member-check” (Riessman, 2008) with the patient where the written narrative is either read by or read to the patient. The patient is encouraged to suggest any needed changes, additions, or deletions to the written narrative. Once any suggested changes have been made and the patient approves the final version of the written narrative, research personnel upload the narrative to the “patient story” tab in the patient’s EHR (Figure 2). At this time, research personnel alert the patient’s nurse that the narrative is in the chart and provides education on its location.

Table 1.

Narrative Intervention Guide

Conversational Questions: May I call you? What is your illness? Conversational probes throughout intervention: For example Tell me more about that Anything else?

Open-ended Questions: Tell me what it has been like to have ______ (illness). Tell me how your illness has impacted you emotionally, your feelings? Tell me how your illness has impacted your relationships with family, friends, and others? Tell me how your illness has impacted your spirituality? Your faith, beliefs, your values, or your thoughts about a higher power?

Closing Question: Is there anything else we have not talked about that you would like to tell me?

FIGURE 2.

Electronic health record: Patient story. y.o. = years old; MRN = medical record number; Dx = diagnosis; NKDA = no known drug allergies.

Data Collection and Measures

The intervention group will be compared to a usual care group of participants: hospitalized patients who meet the same eligibility criteria as the narrative intervention group. Table 2 provides a description of all study measures.

Table 2.

Study Measures

Measures	Domains	Items
Demographics and general questionnaires	Age, gender, marital status, race/ethnicity, level of education, income level, patients’ perception of their general health (excellent, very good, good, fair, poor).	8
Quality of Communication (QOC)	Assesses patients’ perceptions of the quality of communication with clinicians across two constructs: general communication and end-of-life communication (Engelberg et al., 2006) Used for testing effects of quality of communication across other studies implementing communication interventions (Au et al., 2012; J. Curtis et al., 2004; J. R. Curtis et al., 2018; Janssen et al., 2011)	19
PROMIS-29 v2.0	Assesses patient’s self-reported physical, mental and social health and well-being (Craig et al., 2014) Used for testing the following health domains: Physical function, anxiety, depression, fatigue, sleep, ability to participate in social roles/activities, pain interference and intensity and provide ability to compare scores across different conditions (Craig et al., 2014; Hays et al., 2018; Katz et al., 2017)	29
PROMIS Positive Psychosocial Impact	Assesses for positive psychosocial outcomes of the illness that occurs as a result of confrontation with one’s mortality and one’s ability to adapt to the illness (Bajaj et al., 2011; Salsman et al., 2012) Used for testing effects of any illness (not disease specific) such as meaning/purpose in life, coping, self-concept, greater life appreciation, social connection, and personal resources (Biesecker et al., 2013; Carle et al., 2015)	8
PROMIS Negative Psychosocial Impact	Assesses for negative psychosocial outcomes of the illness that occurs as a result of confrontation with one’s mortality, distress that is distinct from general anxiety and depression (Bajaj et al., 2011; Salsman et al., 2012) Used to test effects of distress and other concerns of illness (not disease specific) such as lack of meaning and purpose in life and social isolation, and lack of coping (Biesecker et al., 2013; Carle et al., 2015)	8

Timing of Data Collection

Patient measures are assessed at three time points:

1. baseline (Time 1);

2. 24–48 hours after Time 1 (Time 2); and

3. 24–48 hours after Time 2 (Time 3; Table 3).

Table 3.

Timing of Patient Participant Data Collection

Patient Participant	Time 1	Narrative upload	Time 2	Time 3
When?	Baseline	4–24 hours after Baseline	24–48 hours post baseline	24–48 hours after Time 2
Who?	Intervention/Control	Intervention group only	Intervention/Control	Intervention/Control

The study duration for measure collection was chosen for three reasons:

1. concern of attrition common in research conducted with persons with serious illness (Jordhøy et al., 1999),

2. ability to compare intervention and control group differences on measures during a single hospitalization, and

3. having corresponding time frames with collection of the measures for both intervention and control group.

Time 2 and Time 3 are necessary to examine the effects immediately after the narrative upload and 2 days after the narrative upload.

Following the Time 3 outcome assessments, patients are scheduled within 1 week to collect exit interviews for gaining their feedback about enrollment in the study. This interview is audio-recorded and can be conducted in person or by phone. The patient is asked five or six questions, depending on group assignment, all pertaining to their experience in the study (Exit Interview Guides, Supplemental Digital Content, http://links.lww.com/NRES/A358).

Nurse exit interviews are audio-recorded and can occur in person or by phone. The nurse is asked 7 or 10 questions, depending on the corresponding patient’s group assignment, all pertaining to their experience in the study (Exit Interview Guides, Supplemental Digital Content, http://links.lww.com/NRES/A358). In addition, to evaluate usability of the person-centered narrative intervention, the System Usability Scale measure (Bangor et al., 2008) is administered to nurses whose corresponding patients were assigned to the intervention group (System Usability Survey, Supplemental Digital Content, http://links.lww.com/NRES/A358).

As a token of appreciation for their participation, patient participants are given a $25 Visa gift card after completion of Time 2 and then a $25 gift card after completion of the exit interview. Nurse participants are given a $25 gift card after completion of the survey and exit interview.

Data Management and Analysis

Participants complete all demographic and outcomes measures on an iPad, via the secure HIPAA-compliant REDCap mobile app (Harris et al., 2009). If the patient is unable to or desires not to use the iPad, the research personnel reads each measured item and corresponding answer options and completes the survey according to the patient’s responses. Updated data are transmitted to the REDCap database located on the secure, encrypted network. Data are reviewed for completeness and accuracy after each data upload. Raw data and corresponding code can be exported to standard statistical software packages.

Statistical Analysis Plan

Analyses will be performed using SAS software Version 9.4 (SAS Institute, Inc., 2018). Data will be screened for errors and outliers; appropriate transformations will be made for nonnormal outcomes. Baseline demographics will be summarized using standard descriptive statistics and baseline characteristics between intervention, and control patients will be compared using independent-samples t tests (continuous measures) and chi-square tests (categorical measures). Statistical adjustment will be used in outcome analyses in the event of pretreatment differences between conditions. The pattern of participant dropout will be examined to ensure a reasonable equal distribution of participants lost to follow-up between groups and based on baseline characteristics. Maximum likelihood techniques will include participants with incomplete data without need for imputation. In addition, we will conduct sensitivity analyses to examine the effect of departures from key assumptions made in the main analysis and to help determine effects of missing data.

Using an intent-to-treat approach, primary and secondary outcomes will be analyzed with linear mixed models using SAS Proc Mixed (SAS Institute, Inc., 2018) to compare the two groups (narrative intervention vs. control) over the assessment period. Separate models will be estimated for each of the primary and secondary outcomes, though relationships among outcomes will also be assessed. Mixed models will specify intervention group and time as fixed effects and subject as a random effect. Pairwise comparisons testing group effects at each time point are proposed to disentangle the effects of intervention process (Time 2) versus change in patient’s perspective on the nurse’s quality of communication (Time 3). Specifically, Time 2 occurs after narrative upload, but prior to affecting change in quality of communication. Time 3 occurs after the nurse has had time to view the narrative and therefore informs the patient’s perspective on the nurse’s quality of communication.

Sample Size and Power Analysis

Sample size estimates were generated to determine power to detect a time by treatment interaction within the proposed repeated-measures design. Assuming a two-sided alpha of .05, a sample size of 80 patients will provide 80% power to detect an effect size associated with treatment effects on changes over time as small as f = 0.14 (Cohen’s d = 0.28). This represents a small, clinically relevant effect but, more importantly, will provide stable measures of outcome means and standard deviations over time for use in the planned larger efficacy trial. Modern missing data techniques (Enders, 2010) will be used if patients drop out before all follow-up data are obtained. To minimize concerns for attrition, the time between baseline and follow-up data collection is small. Nonetheless, if patients are lost to follow-up, even a high level of attrition (e.g., 20%) will allow for detection of a small effect size of f = 0.16.

RESULTS

Since October 2019, 53 potential patient participants have been approached, 21 have enrolled, and 20 have completed the data collection process. Of the 32 patients who declined, the most common reasons they gave included feeling tired, believing they would discharge soon, or not wanting to discuss information about themselves. Sixteen nurse participants were screened, and 15 were enrolled. At this time, we have identified several challenges inherent in research conducted in inpatient acute care settings. Patients are often out for medical procedures, receiving clinical care at the bedside, or occasionally report feeling too tired or ill to complete an extra activity. To overcome this, research personnel is flexible with scheduling study procedures, including having the option to conduct exit interviews by phone if the patient has been discharged. In addition, close communication with the patients’ bedside nurses has been necessary to coordinate windows of time for study procedures.

DISCUSSION

The testing and integration of a person-centered narrative into the EHR is a novel approach to provide improvement in communication between patients and clinicians. It remains critical to develop and test narrative interventions in an acute care setting to provide opportunities for the delivery of person-centered care. The narrative intervention has been shown to be feasible, usable, and acceptable (Coats et al., 2020). Although there is strong literature evidence for the use of narrative to improve psychological, social, and existential well-being, there is a lack of implementation of EHR-integrated narrative interventions.

Conclusion

The results from this study will provide important preliminary knowledge to inform future large-scale (i.e., multisite, additional patient populations, and larger sample size), randomized clinical trials of narrative interventions leading to advances in how to best provide high-value, high-quality, person-centered care for persons living with a serious illness.