In today's complex healthcare organizations, the contribution and commitment of nurses in achieving exemplary patient and organizational outcomes are of critical importance.1 The meaningful expressions of gratitude expressed by patients and their families via The DAISY Award nomination for extraordinary nurses reveal core behaviors that shape, reinforce, and sustain a culture of patient-centered care that supports these outcomes.2 We know that meaningful staff recognition has been shown to significantly and positively improve Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) scores,3 predict job satisfaction,4 and contribute to creating a worksite culture that increases employee satisfaction and retention.5 What is not known is how technology contributes to the identification and analysis of core themes and behaviors that can be used to intentionally support and sustain a patient-centered culture.
Patients and families from 136 healthcare organizations in the United States utilized an inpatient interactive patient care (IPC) technology platform to generate more than 52 000 narrative nominations for The DAISY Award, which honors nurses internationally for compassionate, extraordinary care in memory of J. Patrick Barnes. In 2016, a collaborative partnership was formed between an IPC company and The DAISY Foundation to identify themes in these narratives. In this institutional review board–approved qualitative descriptive study, artificial intelligence (AI) using natural language processing and machine learning techniques was applied to a convenience sample of recognitions harvested via IPC. Researchers identified various behavioral themes embedded in nursing care that patients and families found most meaningful, compared these actions to what has been previously described in the literature, and outlined implications for nursing practice and organizations.
Review of the Literature
Meaningful recognition acknowledges how a person's actions affect the life of another, is relevant to the recipient, and is equivalent to his/her contribution.1 The impact upon nurses receiving acknowledgement about the care they provide fosters honor, pride, reinvigoration for the profession, and increased satisfaction.6 Patients and families frequently wish to express gratitude for and acknowledge the compassionate care received through some form of meaningful recognition. An effective way to provide positive feedback about these experiences is through The DAISY Award, which afford patients and families the opportunity to bestow meaningful recognition on caregivers. Along with reinforcing the delivery of compassionate care, this recognition also reconnects nurses with the art of nursing, which motivates them to provide these experiences to others.7 DAISY Award nominations written in narrative or story form capture an individual's experience, preferences, and priorities and provide a deeper understanding of what is important to the patient and family when receiving healthcare.8 Currently, more than 3700 healthcare facilities in 21 countries in addition to the US have access to stories of care and compassion collected through DAISY nominations.2
As a key component of a healthy work environment, meaningful recognition utilizes feedback to describe what matters most to patients and their families.2 Identifying the impact of nursing practice behaviors upon the patient experience allows nurses to recognize the link between their actions and positive outcomes. This recognition and reflection help nurses reinforce the meaning of their work.9 Along with reinforcing the purpose of one's work, additional outcomes associated with meaningful recognition include elevated self-esteem, perceptions of organizational support, workgroup cohesion, embeddedness, nurse and job satisfaction, patient satisfaction, and decreased burnout and compassion fatigue.10-14
One previous study identified behavioral patterns associated with extraordinary nursing in patient, family, and colleague recognition narratives submitted through the DAISY recognition program.1 In this study, 2195 nominations were analyzed; 70% were handwritten, and 30% were collected electronically. Content analysis identified emergent behavioral themes that were then submitted to secondary computerized content analysis using Diction version 5, an analytic software based on the rules of linguistics. Behaviors associated with extraordinary nursing were identified. These included compassion and caring, professionalism, positive attitude, accomplishes more than expected, teaches others, helpful, calming/patient, connects with family, exalted by coworker, intentionally present, and others.1 The study did not differentiate between electronic and handwritten behaviors contained within the nominations.
Analysis of Narratives and the Emerging Use of AI
Capturing themes through the linguistic analysis of narratives provides insight and validation of what is valued in healthcare by patients, families, and individuals. Surveys are often used to collect feedback, which provide a quantitative means to collect information from an overall organizational perspective. The use of patient and family narratives as a form of feedback provides rich qualitative data, identifying what is most important to them, as consumers of care. An integrated approach evaluating the patient experience is best accomplished by using a mixed-methods approach, including surveys and narratives in order to quantify and qualify what is valued in the healthcare experience.8 Finding the means to gather and analyze this feedback is integral to defining a quality patient experience.
The use of AI for analysis of healthcare data is growing. AI is divided broadly into 3 stages of development: artificial narrow intelligence (ANI), artificial general intelligence, and artificial superintelligence.15 Machine learning is a subset of AI, which creates programs based on data as opposed to programming rules. The software draws from large sets of relevant data and continuously learns and refines the analytical capabilities as it discovers recurring patterns.15 In the next decade, ANI has the highest potential for use in healthcare, analyzing large data sets and discovering new correlations.15 The literature has reported little about the use of AI in the analysis of patient and family satisfaction. Natural language processing is the computational linguistics technique of extracting structured meaning from unstructured natural language and has been used to analyze healthcare consumer comments, such as those found in online review platforms such as “Yelp.”16 Evidence regarding widespread use of such analysis is limited, but provides a greater depth and breadth of topics when compared with HCAHPS results, the US standard for evaluating patients' experiences after hospitalization.16 The electronic collection and analysis of narratives using AI may be worthy of organizational action and investment, expediting and enriching data reporting to better understand patient expectations, experiences, and overall well-being.17
Design and Research Questions
This was a retrospective qualitative research study to address the paucity of research, with 2 questions: 1) “What are the core themes identified by patients and families in hospitals that utilize an IPC technology solution to collect nurse recognition nominations?” and 2) “Are these themes similar to or different than what has been previously described?” The purpose of this study was to analyze meaningful nurse recognition nominations collected from patients and families using IPC technology and analyzed using AI and machine learning techniques to identify core themes and behaviors. Researchers wanted to understand how these data compared with what has previously been described and implications for nursing practice and organizations.
A convenience sample of 3 organizations utilizing an IPC technology platform to record DAISY Award nominations was invited to participate in this study. Attempting to represent the larger sample of facilities using both the DAISY Award and IPC technology, the selection process also incorporated bed size into the sampling methodology. Two sites represented large facilities, and the third resembled smaller institutions. The DAISY Foundation contacted liaisons at each of the facilities assessing their willingness to participate in this study. Institutional review board approval was obtained, and each organization agreed to participate, review the protocol, take part in several calls with the research team, and sign a data use agreement.
From a total database of 52 711 DAISY recognition narratives in 2016, 1577 were selected for analysis. After the application of the exclusion criteria of eliminating comments that were fewer than 5 words, 971 comments were analyzed. In this analysis, the research team utilized a business associate's patented AI technology to analyze the narratives in order to identify core themes and behaviors for the concept of nurse recognition utilizing their proprietary natural language processing and machine learning techniques.
The AI technology utilized several natural language processing and machine learning techniques for sentiment classification, including Naïve Bayes approaches, Maximum Entropy, and Support Vector Machines.18 These data were subjected to natural language processing and sentiment analysis using corollary dictionaries based on the rules of linguistics as a foundation for the analysis. Each DAISY Award nomination contained numerous insights and was categorized into predetermined themes and behaviors related to patient experience measures that had been identified through AI and verified by expert linguists. Summaries of themes and behaviors were provided to the research team for review and analysis.
Three percent of the total DAISY nominations (1577) were collected via the IPC technology platform in 2016 from the hospitals participating in this study. Refer to Table 1 for the frequency and percentage of comments, themes, and behaviors identified. Participating hospitals ranged from 168 to 811 licensed beds. Table 1 illustrates the total number of nominations, the number of nominations used to conduct this analysis after the application of the exclusion criteria, and the number of themes and behaviors that emerged from the data. An average of 2.8 themes and behaviors per phrase were identified within nomination narratives. As expected in a recognition program for nurses, all comments about the nurse were positive.
Information identified via AI techniques was provided to the researchers for analysis. Analysis of nurse recognition comments fit into 5 core themes: 1) courtesy and respect; 2) skill and knowledge; 3) reliability/scheduling; 4) explanation; and 5) listening.
Courtesy and respect was the highest ranking core theme and accounted for 64% of the total behaviors reported. This core theme had 622 phrases that revealed 22 underlying behaviors. The top 5 behaviors mentioned were empathy/compassion (22%), helpfulness (16%), kindness (15%), attentiveness (7%), and emotional comfort (7%). These 5 behaviors accounted for more than half (67%) of the behaviors within this core theme. Table 2 provides the frequency and detailed list of the underlying behaviors within this theme.
The skill and knowledge theme included 98 phrases (10% of total analysis) with 7 underlying behaviors. The behaviors described about the nurse were professional, knowledgeable, skill, keeping track, competence, dedication, and being thorough. Refer to Table 3 for the frequency and detailed information of underlying behaviors in this theme.
Fifty-eight phrases (6% of the total analysis) mentioned reliability and scheduling in nurses' recognition nominations. The top behaviors noted were the nurses' pace of work and availability, meaning that the nurses were available when needed and were timely in their care (ie, not rushed, spent adequate time to conduct procedures, schedules were kept, etc; 43%). Refer to Table 4 for detailed information of the behaviors described.
Two other themes fit the data, explanation and listening. Overlap was noted in the explanation about care core theme with skill and knowledge and the listening core theme with courtesy and respect, but upon more detailed review, phrases within each core theme appeared specific enough to categorize them separately. Explanation about care included 52 phrases that described this core theme as the nurse answering questions, providing status updates, being informative, using understandable vocabulary to explain care, explaining diagnosis, and explaining conflicting information. The listening core theme had 31 phrases that described listening, being attentive and helpful, prompt, having patience, empathy, and compassion.
In addition to the 5 core themes, phrases included information referencing the overall patient experience, including the reputation of and loyalty to the organization, meal service, environmental cleanliness, medications, tests and treatments, and the handling of care by the organization. Supplemental Digital Content 1, http://links.lww.com/NNA/A8, provides examples of phrases for the top 5 core themes: courtesy and respect, skill and knowledge, reliability and scheduling, explanation, and listening.
Discussion and Implications for Practice
The results of this research demonstrate that IPC-generated narratives depicting extraordinary nursing most often contained behavioral descriptions associated with compassionate care. In addition to compassion, themes associated with professionalism, helpfulness, and kindness also mattered to patients and families. Applying these findings to the workplace, leaders can expand their focus on those nursing tasks aligned with patient satisfaction metrics to also include the behaviors that matter most to patients and families, such as warm and genuine concern for their suffering and recognizing their unique needs. Meaningful nurse recognition narratives are reflections of nursing practice, with the power to reconnect nurses to why they became a nurse. In sharing these narratives with nurses, they are reminded of the privilege of practicing in the sacred space of intentional, transpersonal caring where human connection occurs.19 Patients and families are keenly aware of and value their nurses' knowledge and skill. Therefore, communication between nurses and patients during shift handoff as well as other opportunities to communicate would benefit from including education, preparation, certifications, and managing up team members. Understanding the impact of the perceived pace of care and availability of nurses, both emotionally and physically, upon the patient experience can influence nursing workflows so that they are fully present in helping patients and families navigate their care.
The nominations used to support recognition of nurses for The DAISY Award serve multiple purposes. The 1st is using the nominations to acknowledge best practices in compassionate care delivered by the nurse. This acknowledgement communicates to the nurse their value in the organization and can increase nurse engagement and retention.7 Second, the public recognition of nurse behaviors communicates the organizational values to others. Third, the nominations for the award communicate what patients and family members value from their nursing caregivers. Fourth, these themes can be highlighted as part of educational offerings for current staff and incorporated into onboarding processes for new staff. Fifth, DAISY Award narratives provide qualitative insight into an organization's top box HCAHPS scores. Harnessing these qualitative insights from patients, families, and colleagues provides depth to metrics and can inform additional strategic priorities and initiatives in organizations.
As nurse leaders, we recognize the power of technology to collect and analyze meaningful nurse recognition to fully understand what matters most to patients and families in our organizations. We can use these data to develop tactics to intentionally cultivate nurse behaviors, resulting in the creation of a workforce culture that supports nurse job satisfaction, self-esteem, organizational commitment, teamwork, and retention.1
AI has the potential to capture the attributes and characteristics of nurses most important to human caring. This study demonstrates how the use of AI can advance the science and practice of nursing by: 1) accurately and consistently analyzing this qualitative data that reflect the sentiments associated with nursing's impact upon human caring and 2) helping identify from these qualitative data actionable priorities for care. We must also recognize the limitations of certain techniques, as it is critical to translate the data analyzed and themes identified through this technology to the professional practice environment and the implications for nursing. Developing strategies for incorporating insights gained from qualitative data through AI into the practice of nursing is the domain of nurse leaders. In reviewing the literature in this study, a variety of terms were used referencing the same intent. For instance, the use of themes versus categories versus insights versus concepts is often used interchangeably, which can confuse the end user as to what is being discussed. It is also necessary to establish a common vocabulary in the field of nursing to facilitate the discussion and application of AI-enabled discoveries for the practice of nursing. For example, this common vocabulary should include the distinctions between analytical constructs such as categories, themes, and insights. In addition, the research team often referred to the concept of nurse recognition and the parallels between the themes and behaviors identified, and the attributes and characteristics of the concept. Establishing such common vocabulary and knowledge will empower nurse leaders to actively engage in how AI output is translated for implications to nursing practice and how AI technology can be leveraged for the advancement of nursing in the future.
The breadth of these findings may have been limited by a convenience sample of different hospitals utilizing the same online technology to access The DAISY Award nominations. In turn, the organizational culture and geographic location of each facility may have influenced the focus and content of The DAISY Award nominations subsequently affecting the results. Although The DAISY Award nominations were from different individuals, 70% of the total number of nominations used in this study were from 1 organization, likely impacting the emerging themes.
Along with these sample limitations, the specific AI processes utilized to identify the themes involved the proprietary taxonomy algorithm from 1 specific vendor. Another sentiment analytic taxonomy and/or AI algorithm exploring this same data set may reveal different themes. The lack of previous research exploring the ability of AI methodology to assess positive feedback about nurses providing extraordinary care limited our ability to compare and contrast our findings with other studies that utilized this technology. As one of the 1st studies to utilize this methodology to explore the positive feedback associated with extraordinary nurses, future studies may find different themes describing the difference nurses make in the lives of those they serve.
Recommendations for Future Research
This study could be replicated utilizing a larger sample size and/or another methodology for AI-driven patient experience analytics. Future research utilizing a more balanced sample of patient and family feedback about their experiences related to the work of extraordinary nurses may yield additional insights. This study was conducted in both adult and pediatric hospitals; future research analyzing meaningful recognition comments for just adult or pediatric hospital or comparing the two may be of interest as well as narratives focusing on specific nursing specialties. Because the data are able to be quantified using AI techniques, others may be interested in looking at the correlation between specific nurse recognition themes and behaviors, and patient experience, nursing and/or organizational outcomes. This study used AI techniques to analyze narrative comments. A future study might compare findings between more traditional qualitative methods and AI techniques.
AI techniques for qualitative analysis of nursing recognition narratives collected through IPC reveal nurse themes and behaviors most meaningful to patients and families. Nurses can advance the science of AI and guide its evolution so that it most accurately represents positive sentiments associated with nursing's impact upon human caring.
The authors thank Christine Bish, MBA, MHA, BSN, RN, Carroll Hospital, a LifeBridge Health Center; Cynthia LaFond, PhD, RN, CCRN-K, The University of Chicago Medicine; and Marilyn Prier, MPH, RN, and Pam Barlow, The Children's Hospital of Alabama.