Narrative Medicine in Pediatric anti‑NMDA Receptor Encephalitis: A Mixed‑methods Evaluation of Triadic Outcomes Among Children, Parents, and Healthcare Providers Authors: Xiaolin Zhou, Xuezhen Chen, Pinggan Li, Xiangyang Luo, Xiaomei Zeng Published: 2026 ABSTRACT BackgroundPediatric anti-NMDAR encephalitis can cause severe disability and profound caregiver distress. Narrative medicine has been proposed as a humanistic intervention, but its impact across the complete caregiving triad of children, parents, and healthcare providers has not been systematically evaluated. This exploratory study aimed to assess the multidimensional impact of narrative medicine interventions in this setting.MethodsThis retrospective mixed-methods study included five children with severe anti-NMDAR encephalitis (admission mRS ≥4), their mothers, two physicians, and six nurses. Narrative medicine interventions included daily bedside narrative attention, parallel chart writing, family conferences, nurse-led narrative communication, and weekly sharing sessions. Quantitative outcomes included child mRS, parental anxiety (SAS), depression (SDS), caregiver burden (ZBI), and provider empathy (IRI), burnout (NBS), and meaningful work (WAMI). Qualitative data from parallel charts, reflective writings, and family interviews were analyzed thematically.ResultsAll five children showed favourable neurological improvement (median mRS: 5→1 at 6 months; p=0.039); three returned to school. Parental outcomes showed substantial improvements: SAS (42.8→23.4, p<0.001, d=9.7), SDS (54.8→27.4, p<0.001, d=6.5), and ZBI (59.6→29.6, p<0.001, d=3.4). Nurses (n=6) showed increased empathy (IRI: 63.7→78.0, p=0.003, d=1.9) and reduced burnout (NBS: 73.0→56.3, p=0.002, d=2.1); physicians showed parallel improvements. Thematic analysis identified three potential mechanisms: clinical narrative signals, a window of narrative intervention (weeks 2–4), and bidirectional healing.ConclusionIn this small, exploratory retrospective case series, narrative medicine interventions were temporally associated with favourable trends in neurological outcomes, parental psychological states, and provider well-being. The proposed concepts of clinical narrative signals, a window of narrative intervention, and bidirectional healing offer a preliminary framework for integrating narrative practices into pediatric neurocritical care. These findings are hypothesis-generating and warrant prospective investigation in larger controlled studies. Introduction Anti-N-methyl-D-aspartate receptor (NMDAR) encephalitis is the most common autoimmune encephalitis in children, with a complex and multifactorial etiology, accounting for approximately 80% of pediatric cases.1–3 The disease typically follows a fulminant course: previously healthy children develop rapidly progressive neuropsychiatric symptoms, often requiring pediatric intensive care unit admission.4,5 Despite timely first-line immunotherapy, a substantial proportion of patients remain refractory and require second-line agents such as rituximab.6 Even among survivors, recovery is often protracted, with many children left with persistent neurological deficits, cognitive impairment, and functional disability.7 Beyond the direct medical impact, anti-NMDAR encephalitis can impose a severe psychological and socioeconomic burden on families. Parents face weeks to months of profound uncertainty, frequently developing clinically significant anxiety, depression, and post-traumatic stress symptoms, along with high levels of caregiver burden.8,9 At the same time, healthcare providers—particularly nurses and physicians working in PICU and pediatric neurology units—are at high risk of compassion fatigue, moral distress, and burnout.10–13 Therefore, effective interventions must address the neurological recovery of the child, the psychological well-being of parents, and the professional resilience of the care team. Narrative medicine offers a potential framework for addressing these multidimensional needs.14 As defined by Charon,15 narrative medicine is “medicine practiced with narrative competence to recognize, absorb, interpret, and be moved by the stories of illness”. It emphasizes three core movements: attention, representation, and affiliation.16 Attention refers to empathic listening—not only to symptoms but also to the patient’s broader illness story and social context. Representation is the act of rendering that story into reflective writing such as parallel charts. Affiliation is the relational outcome: the formation of a therapeutic alliance. These movements distinguish narrative medicine from general empathic communication by its deliberate, reflective, and relational methodology.17 Narrative approaches have been applied in difficult communication contexts, including organ donation discussions, end-of-life care, and resident burnout prevention,18–20 and in pediatrics to support families of children with complex chronic conditions.21–23 However, its application to pediatric autoimmune encephalitis remains largely unexplored. No study has systematically evaluated the triadic impact of narrative interventions on children, parents, and healthcare providers simultaneously, nor have the mechanisms—such as a critical “window of narrative intervention”or “a bidirectional healing effect”—been proposed or empirically tested. Three specific gaps motivated this study. First, quantitative evidence is lacking: few studies have provided data on narrative medicine’s effects in pediatric neurocritical care. Second, previous studies typically adopt a dyadic perspective, overlooking potential synergistic effects across the complete caregiving triad. Third, the underlying mechanisms remain poorly understood. To address these gaps, we conducted a mixed-methods evaluation of narrative medicine interventions in five children with severe anti-NMDAR encephalitis and their families. The study had three objectives: (1) to quantitatively assess changes in clinical outcomes (modified Rankin Scale, mRS), parental psychological states (anxiety, depression, caregiver burden), and healthcare provider well-being (empathy, burnout, meaningful work); (2) to qualitatively explore lived experiences through parallel charts, reflective writings, and narrative interviews; and (3) to integrate findings into a preliminary model, generating novel concepts—window of narrative intervention (weeks 2–4 post-admission), clinical narrative signals (subtle non-verbal responses indicating early consciousness recovery), and bidirectional healing (reciprocal emotional restoration experienced by providers)—for future investigation. We hypothesized that narrative interventions would be associated with favourable trends in neurological recovery, reduced parental distress, and enhanced provider well-being. By adopting a triadic, mixed-methods approach, this study aims to generate hypotheses and provide preliminary evidence for integrating narrative practices into standard care for pediatric neuroinflammatory disorders. Material and Methods Study Design, Setting, and Participants This retrospective mixed-methods study with an explanatory sequential design evaluated children with severe anti‑NMDAR encephalitis who received narrative medicine interventions at the Department of Pediatric Neurology, Sun Yat-sen Memorial Hospital, Sun Yat-sen University, Guangzhou, China, between January 2025 and June 2025. All clinical data and narrative interventions were derived from routine clinical practice during this period. The study was approved by the Institutional Review Board of Sun Yat‑sen Memorial Hospital, Sun Yat-sen University (Approval Number: SYSKY-2026-452-01), which waived the requirement for prospective informed consent due to the retrospective nature of the analysis. All patient and provider data were anonymized prior to analysis. The study was conducted in accordance with the Declaration of Helsinki. Ethical approval was sought and obtained retrospectively for the analysis of routinely collected clinical data, as per institutional policy for minimal-risk retrospective studies. Children Five children with severe anti-NMDAR encephalitis who received narrative medicine interventions as part of routine clinical care were identified from the hospital’s pediatric neurology database. Inclusion criteria were: (1) diagnosis confirmed by cerebrospinal fluid antibody testing; (2) age 1–18 years; (3) admission modified Rankin Scale (mRS) score ≥4; (4) narrative medicine interventions documented in medical or nursing parallel charts. Exclusion criteria were: (1) other causes of encephalitis; (2) pre-existing neurodevelopmental disorders. All eligible children during the study period were included. Parents The primary caregiver (mother in all five cases) of each child participated in the narrative interventions and had documented psychological assessments (retrospective or follow‑up). Healthcare Providers Two physicians (the lead author and a second pediatric neurologist) and six nurses (all registered nurses with ≥3 years of experience in pediatric neurology or intensive care) who provided direct care to these children and participated in narrative medicine training and reflective writing sessions were included. The two physicians engaged in self‑directed training in narrative medicine through literature review, collaborative case discussions, and iterative refinement of their narrative approaches during the study period. The six nurses received a structured 2‑hour training session covering the core principles of narrative medicine (attention, representation, and affiliation), the use of parallel charts for reflective documentation, and narrative communication skills for engaging with families in distress. All team members also participated in weekly narrative sharing sessions, which served as ongoing experiential learning and peer supervision. Narrative Medicine Interventions As part of routine clinical care, the following narrative medicine interventions were delivered to all five children and their families throughout hospitalization, beginning within 48 hours of admission and continuing until discharge. Although not part of a prospective protocol, these interventions were documented in parallel charts and nursing notes. The interventions were designed to operationalize the three core movements of narrative medicine—attention, representation, and affiliation—distinguishing them from general supportive care through their deliberate structure, reflective documentation, and team-based processing. Physician-Led Practices The attending physician routinely performed: (a) daily bedside narrative attention (5–10 minutes/day), which involved speaking to the child about non-medical topics and listening to parental concerns; (b) parallel chart writing, with one narrative parallel chart per child focusing on the illness story and family emotional journey; and (c) narrative family conferences (1–2 per child) to facilitate sharing of experiences, fears, and hopes. A parallel chart is a reflective narrative written by the clinician alongside the standard medical record. Unlike the conventional chart, which documents objective clinical data (eg, vital signs, laboratory results, treatment administered), the parallel chart captures the patient’s illness story, the family’s emotional journey, and the clinician’s personal reflections. It is not a substitute for the medical record but a complementary narrative that provides insights into the lived experience of illness and care. Nurse-Led Practices Nurses received a 2-hour training session on narrative medicine principles before the study period, covering the core concepts of attention, representation, and affiliation, the use of parallel charts for reflective documentation, and narrative communication skills for engaging with families in distress. During routine care, they: (a) maintained daily nursing parallel charts documenting behavioral responses and parental emotional states; (b) engaged in narrative communication with parents during daily care; and (c) participated in weekly, physician-facilitated narrative sharing sessions (60 minutes each). Family Narrative Engagement Parents were encouraged to speak, read, or sing to their child daily and to document small progress. Key narrative turning points (eg, first eye closure, first smile, first swallow) were recorded in parallel charts. Data Collection (Retrospective) Data were extracted from electronic medical records, nursing parallel charts, physician parallel charts, and follow-up clinic visits at three predefined time points: admission (within 48 hours), discharge (within one week before discharge), and follow-up (6 months after discharge). The retrospective nature of this study means that the intervention was not delivered according to a prospective protocol; rather, the data presented reflect routine clinical practice during the study period. Baseline clinical data (admission mRS) were recorded prospectively as part of routine clinical documentation. However, parental psychological scales were not part of routine care; these were collected retrospectively at follow-up using a structured anchor-point method (see Section 2.5). Clinical Outcomes Modified Rankin Scale (mRS) scores at admission, discharge, and 6-month follow-up were extracted from physician notes. Hospital length of stay, ICU stay, and second-line immunotherapy use were obtained from medical records. Parental Psychological Outcomes The three scales (SAS, SDS, ZBI) had not been administered prospectively. During the 6-month follow-up visit, mothers therefore completed each scale twice: first retrospectively, recalling their state at the time of the child’s admission (using a validated anchor-point procedure), and then reporting their current state. The Chinese versions of the scales showed good internal consistency in this sample (Cronbach’s α: SAS 0.92, SDS 0.90, ZBI 0.94). Healthcare Provider Outcomes Physicians and nurses retrospectively completed the IRI, NBS, and WAMI scales, recalling their status before the first narrative session (pre-intervention) and reporting their current status at discharge (post-intervention). Providers were not reassessed at 6 months. All scales were paper-based. The WAMI was administered only to physicians. Qualitative Data Existing qualitative documents were collected: (a) five physician parallel charts (one per child); (b) reflective writings from six nurses (2–3 reflections each); and (c) semi-structured interview transcripts from the five mothers, conducted at the 6-month follow-up (15–25 minutes each). These interviews had originally been conducted as part of routine clinical follow-up to assess family satisfaction and were stored in the department’s quality improvement archive. Data Analysis Quantitative Analysis Data were analyzed using SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Descriptive statistics (mean, standard deviation, frequency) were calculated for all variables. Normality was assessed using the Shapiro–Wilk test. For normally distributed paired data (parental SAS, SDS, ZBI; nurse IRI, NBS, WAMI), paired t-tests were used to compare pre- and post-intervention scores. For non-normally distributed paired data (child mRS), the Wilcoxon signed-rank test was applied. Effect sizes were calculated as Cohen’s d for t-tests (d ≥ 0.2 small, ≥0.5 medium, ≥0.8 large) and as r = Z/√n for Wilcoxon tests. Statistical significance was set at a two-tailed p < 0.05. Qualitative Analysis All qualitative data (parallel charts, reflective writings, interview transcripts) were analyzed using thematic analysis following Braun and Clarke’s six-phase framework.24 Two researchers (the lead physician and a research assistant) independently coded the data using NVivo 12 (QSR International, Melbourne, Australia). Codes were grouped into potential themes, which were reviewed and refined iteratively. Disagreements were resolved through discussion. Themes were identified separately for physicians, nurses, and parents, then integrated into overarching meta-themes. Representative quotes were selected to illustrate each theme. Integration (Mixed-Methods) Quantitative and qualitative findings were integrated using a joint display matrix to compare results across the three stakeholder groups and to identify convergence, complementarity, or dissonance. This triangulation approach strengthened the validity of the findings. Handling of Retrospective Bias To minimize recall bias in the retrospective collection of parental admission scores, a structured anchor-point procedure was used (eg, “Think back to the first week after your child was admitted”). For healthcare providers pre-intervention scores, participants were asked to recall their status “before any narrative medicine training or practice”. Consistency between retrospective admission scores and contemporaneous clinical notes (eg, notes describing “mother crying”, “father anxious”) was also assessed. The 6-month recall period for admission status was considered acceptable for retrospective assessment of psychological distress. These limitations are addressed in the Discussion. We acknowledge that retrospective recall may be influenced by subsequent events and that the absence of prospective baseline measures limits causal interpretation. The findings should therefore be viewed as exploratory and hypothesis-generating rather than confirmatory. Results Clinical Outcomes All five children with severe anti-NMDAR encephalitis showed favourable neurological recovery from admission to the 6-month follow-up. The median modified Rankin Scale (mRS) score decreased from 5 (range 4–5) at admission to 1 (range 0–2) at follow-up (Wilcoxon signed-rank test, Z = –2.06, p = 0.039). At follow-up, three children (60%) had returned to school or kindergarten, while the remaining two (40%) continued home-based rehabilitation but achieved functional gains. Detailed clinical characteristics and outcomes are presented in Table 1.Table 1Clinical Characteristics and Treatment OutcomesVariables0102030405Mean ± SDAge (years)/Sex5/F9/M9/F13/M7/M8.6 ± 3.0Antibody typeAnti-NMDARAnti-NMDARAnti-NMDARAnti-NMDARAnti-NMDAR—Admission mRS545554.8 ± 0.4Discharge mRS112231.8 ± 0.8Follow-up mRS001120.8 ± 0.8Hospital stay(days)1509013775148120.0 ± 43.6ICU stay (days)21018352018.8 ± 11.6Second-line immunotherapyRituximabRituximabRituximabRituximabRituximab5/5 (100%)Narrative turning point (day)20 (first smile)21 (first eye contact)1 (eye closure on command)35 (first vocalization)30 (first swallow)21.4 ± 11.8Follow-up duration (months)666666.0 ± 0Return to schoolYesYesNoYesNo3/5 (60%)Note: Wilcoxon signed‑rank test comparing admission mRS vs follow‑up mRS: Z = –2.06, p = 0.039.Abbreviations: mRS, modified Rankin Scale; F, female, M, male; NMDAR, N‑methyl‑D‑aspartate receptor; ICU, intensive care unit; SD, standard deviation. Table 1Clinical Characteristics and Treatment OutcomesVariables0102030405Mean ± SDAge (years)/Sex5/F9/M9/F13/M7/M8.6 ± 3.0Antibody typeAnti-NMDARAnti-NMDARAnti-NMDARAnti-NMDARAnti-NMDAR—Admission mRS545554.8 ± 0.4Discharge mRS112231.8 ± 0.8Follow-up mRS001120.8 ± 0.8Hospital stay(days)1509013775148120.0 ± 43.6ICU stay (days)21018352018.8 ± 11.6Second-line immunotherapyRituximabRituximabRituximabRituximabRituximab5/5 (100%)Narrative turning point (day)20 (first smile)21 (first eye contact)1 (eye closure on command)35 (first vocalization)30 (first swallow)21.4 ± 11.8Follow-up duration (months)666666.0 ± 0Return to schoolYesYesNoYesNo3/5 (60%)Note: Wilcoxon signed‑rank test comparing admission mRS vs follow‑up mRS: Z = –2.06, p = 0.039.Abbreviations: mRS, modified Rankin Scale; F, female, M, male; NMDAR, N‑methyl‑D‑aspartate receptor; ICU, intensive care unit; SD, standard deviation. Clinical Characteristics and Treatment Outcomes Note: Wilcoxon signed‑rank test comparing admission mRS vs follow‑up mRS: Z = –2.06, p = 0.039. Abbreviations: mRS, modified Rankin Scale; F, female, M, male; NMDAR, N‑methyl‑D‑aspartate receptor; ICU, intensive care unit; SD, standard deviation. Key Nodes of Narrative Medicine Interventions The narrative interventions were temporally associated with identifiable turning points in each child’s hospital course. The mean time to the first clear narrative turning point was 21.4 days (range 1–35 days). These turning points included second-line treatment decision support (Patient 1), transfer crisis communication (Patient 2), consciousness signal capture (Patient 3), conflict resolution (Patient 4), and long-term rehabilitation planning (Patient 5). The intervention types encompassed clinical decision support, crisis communication, clinical observation with emotional support, conflict management, and long-term rehabilitation support. Final outcomes were favourable recovery for Patients 1 and 2, partial recovery for Patients 3 and 5, and functional improvement for Patient 4. These data are summarized in Table 2.Table 2Key Nodes of Narrative Medicine InterventionsPatient NumberTurning Point (day)Core Intervention StrategyIntervention TypeFinal Outcome01Day 20 (first smile)Second-line treatment decision supportClinical decision supportComplete recovery02Day 21 (Week 3) (first eye contact)Transfer crisis communication, recovery case sharingCrisis communicationComplete recovery03Day 1 (eye closure on command)Consciousness signal capture, daily bedside companionshipClinical observation + emotional supportPartial recovery04Day 35 (first vocalization)Conflict resolution, paper airplane narrativeConflict managementSignificant improvement05Day 30 (first swallow)Long-term rehabilitation planning, micro-progress documentationLong-term rehabilitation supportPartial recoveryNotes: Intervention types were categorized into five domains: clinical decision support, crisis communication, clinical observation, conflict management, and long-term rehabilitation support. Complete recovery was defined as follow-up mRS = 0–1 and return to school/kindergarten. Partial recovery was defined as follow-up mRS = 1. Significant improvement was defined as follow-up mRS = 2 with functional gains. Table 2Key Nodes of Narrative Medicine InterventionsPatient NumberTurning Point (day)Core Intervention StrategyIntervention TypeFinal Outcome01Day 20 (first smile)Second-line treatment decision supportClinical decision supportComplete recovery02Day 21 (Week 3) (first eye contact)Transfer crisis communication, recovery case sharingCrisis communicationComplete recovery03Day 1 (eye closure on command)Consciousness signal capture, daily bedside companionshipClinical observation + emotional supportPartial recovery04Day 35 (first vocalization)Conflict resolution, paper airplane narrativeConflict managementSignificant improvement05Day 30 (first swallow)Long-term rehabilitation planning, micro-progress documentationLong-term rehabilitation supportPartial recoveryNotes: Intervention types were categorized into five domains: clinical decision support, crisis communication, clinical observation, conflict management, and long-term rehabilitation support. Complete recovery was defined as follow-up mRS = 0–1 and return to school/kindergarten. Partial recovery was defined as follow-up mRS = 1. Significant improvement was defined as follow-up mRS = 2 with functional gains. Key Nodes of Narrative Medicine Interventions Notes: Intervention types were categorized into five domains: clinical decision support, crisis communication, clinical observation, conflict management, and long-term rehabilitation support. Complete recovery was defined as follow-up mRS = 0–1 and return to school/kindergarten. Partial recovery was defined as follow-up mRS = 1. Significant improvement was defined as follow-up mRS = 2 with functional gains. To provide a clearer treatment timeline in relation to narrative medicine interventions, the timing of key therapeutic components—including first-line immunotherapy, second-line rituximab, ICU care, seizure control, and rehabilitation—was also summarized for each patient (as shown in Table 1). All five children had received first-line immunotherapy (corticosteroids and/or intravenous immunoglobulin) prior to or upon admission to our center. Second-line rituximab was initiated at different time points across the five patients: Day 21 (Patient 1), Day 18 (Patient 2), Day 1 (Patient 3), Day 7 (Patient 4), and Day 30 (Patient 5). ICU stays ranged from 0 to 35 days, and seizure control was achieved in all patients prior to or during hospitalization. Rehabilitation was initiated upon admission in all five cases and continued throughout hospitalization. The narrative turning points emerged at Day 20 (Patient 1), Day 21 (Patient 2), Day 1 (Patient 3), Day 35 (Patient 4), and Day 30 (Patient 5). Notably, in all five cases, the narrative signals preceded measurable improvement on the mRS, suggesting that these signals may capture an early indicator of consciousness recovery independent of the timing of immunotherapy response. Parental Psychological States Mothers of the five children completed the SAS, SDS, and ZBI scales retrospectively for admission and at the 6‑month follow‑up. All three measures showed improvements (Table 3). Mean anxiety (SAS) scores decreased from 42.8 ± 1.1 at admission to 23.4 ± 2.5 at follow‑up (paired t‑test, t = 21.2, p < 0.001; Cohen’s d = 9.7). Similarly, depression (SDS) scores fell from 54.8 ± 2.2 to 27.4 ± 5.1 (t = 14.6, p < 0.001; d = 6.5), and caregiver burden (ZBI) scores declined from 59.6 ± 3.4 to 29.6 ± 10.4 (t = 7.7, p < 0.001; d = 3.4). All five mothers reported a return to work and positive psychological states at follow‑up, with qualitative descriptions shifting from “hopelessness” at admission to “hope restored” or “grateful”.Table 3Evolution of Parental Psychological States with Quantitative AssessmentsVariables0102030405Mean ± SDStatistical ResultAdmission scores (retrospective)SAS score414344434342.8 ± 1.1—SDS score515556575554.8 ± 2.2—ZBI score555763616259.6 ± 3.4—Follow-up scores (actual)SAS score222227212523.4 ± 2.5t = 21.2, p <0.001SDS score222234283127.4 ± 5.1t = 14.6, p < 0.001ZBI score162141294129.6 ± 10.4t = 7.7, p < 0.001Notes: SAS = Self-Rating Anxiety Scale (Zung, 1971); SDS = Self-Rating Depression Scale (Zung, 1965); ZBI = Zarit Burden Interview (Zarit et al, 1980). Admission scores were collected retrospectively and may be subject to recall bias. Follow‑up scores were measured at the time of interview. All paired t‑tests were statistically significant with large effect sizes (Cohen’s d > 0.8). Table 3Evolution of Parental Psychological States with Quantitative AssessmentsVariables0102030405Mean ± SDStatistical ResultAdmission scores (retrospective)SAS score414344434342.8 ± 1.1—SDS score515556575554.8 ± 2.2—ZBI score555763616259.6 ± 3.4—Follow-up scores (actual)SAS score222227212523.4 ± 2.5t = 21.2, p <0.001SDS score222234283127.4 ± 5.1t = 14.6, p < 0.001ZBI score162141294129.6 ± 10.4t = 7.7, p < 0.001Notes: SAS = Self-Rating Anxiety Scale (Zung, 1971); SDS = Self-Rating Depression Scale (Zung, 1965); ZBI = Zarit Burden Interview (Zarit et al, 1980). Admission scores were collected retrospectively and may be subject to recall bias. Follow‑up scores were measured at the time of interview. All paired t‑tests were statistically significant with large effect sizes (Cohen’s d > 0.8). Evolution of Parental Psychological States with Quantitative Assessments Notes: SAS = Self-Rating Anxiety Scale (Zung, 1971); SDS = Self-Rating Depression Scale (Zung, 1965); ZBI = Zarit Burden Interview (Zarit et al, 1980). Admission scores were collected retrospectively and may be subject to recall bias. Follow‑up scores were measured at the time of interview. All paired t‑tests were statistically significant with large effect sizes (Cohen’s d > 0.8). Healthcare Provider Assessments Both physicians and nurses showed improvements in empathy and burnout, and physicians also reported increased meaningful work (Table 4). Among the two physicians, mean IRI (empathy) increased from 52.0 to 68.5 (+16.5), mean NBS (burnout) decreased from 87.0 to 63.0 (–24.0), and mean WAMI (meaningful work) rose from 22.5 to 38.0 (+15.5). Among the six nurses, mean IRI increased from 63.7 ± 5.7 to 78.0 ± 8.0 (paired t‑test, t = 5.21, p = 0.003; Cohen’s d = 1.9); mean NBS decreased from 73.0 ± 8.7 to 56.3 ± 7.0 (t = 5.98, p = 0.002; d = 2.1); and mean WAMI increased from 33.2 ± 3.5 to 40.8 ± 5.2 (t = 4.28, p = 0.008; d = 1.5). These quantitative changes were consistent with the qualitative themes of enhanced empathic witnessing, reduced compassion fatigue, and renewed professional meaning.Table 4Quantitative Assessments of Healthcare ProvidersVariablesPhysician-1Physician-2Nurse-1Nurse-2Nurse-3Nurse-4Nurse-5Nurse-6Work experience (years)1033510864IRI (Empathy)Pre-intervention6143575469686668Post-intervention7265726988857379NBS (Burnout)Pre-intervention9381695984817471Post-intervention6759585071596056WAMI (Meaningful work)Pre-intervention2025293531383135Post-intervention3442364438503641Notes: IRI = Interpersonal Reactivity Index (range 22-110); NBS = Nurse Burnout Scale (range 27-135); WAMI = Work as Meaning Inventory (range 10-50). Higher IRI and WAMI scores indicate better outcomes; higher NBS scores indicate worse outcomes. Table 4Quantitative Assessments of Healthcare ProvidersVariablesPhysician-1Physician-2Nurse-1Nurse-2Nurse-3Nurse-4Nurse-5Nurse-6Work experience (years)1033510864IRI (Empathy)Pre-intervention6143575469686668Post-intervention7265726988857379NBS (Burnout)Pre-intervention9381695984817471Post-intervention6759585071596056WAMI (Meaningful work)Pre-intervention2025293531383135Post-intervention3442364438503641Notes: IRI = Interpersonal Reactivity Index (range 22-110); NBS = Nurse Burnout Scale (range 27-135); WAMI = Work as Meaning Inventory (range 10-50). Higher IRI and WAMI scores indicate better outcomes; higher NBS scores indicate worse outcomes. Quantitative Assessments of Healthcare Providers Notes: IRI = Interpersonal Reactivity Index (range 22-110); NBS = Nurse Burnout Scale (range 27-135); WAMI = Work as Meaning Inventory (range 10-50). Higher IRI and WAMI scores indicate better outcomes; higher NBS scores indicate worse outcomes. Qualitative Findings Thematic analysis of physician parallel charts (n = 5), nurse reflective writings (6 nurses, 2–3 reflections each), and family narrative interviews (5 mothers) identified three overarching meta‑themes that complemented and explained the quantitative results. Codes were initially generated independently by two researchers, then grouped into potential themes and refined iteratively until consensus was reached. The three final themes—clinical narrative signals, window of narrative intervention, and bidirectional healing—emerged consistently across all three stakeholder groups. Theme 1: Capturing Clinical Narrative Signals Clinicians and nurses consistently reported that subtle, non‑standardized responses—such as a first eye closure, a fleeting smile, or an attempted swallow—served as early indicators of consciousness and engagement. One physician wrote: “When I asked her to close her eyes if she could hear me, she did. That moment changed everything”. A nurse reflected: “The first time he swallowed a drop of water, I knew he was still there”. Another nurse observed: “It was week three, and nothing had changed. Then one morning, I saw her eyes follow her mother’s face. That was the first sign that she was coming back”. A physician caring for a child with severe dystonia added: “These signals—a blink, a turn of the head, a quieting of the body when you speak—they are the patient’s voice when the patient cannot speak”. These signals often preceded measurable improvements on the mRS by several days to weeks. Theme 2: The Window of Narrative Intervention Across all five cases, families appeared most receptive to narrative support between the second and fourth week after admission—a period characterized by maximal uncertainty and emotional distress. A mother recalled: At week three, I was losing hope. But the doctor sat with us, showed us photos of other children who had recovered, and said ‘we are on the right path.’ That was the turning point. At week three, I was losing hope. But the doctor sat with us, showed us photos of other children who had recovered, and said ‘we are on the right path.’ That was the turning point. Another mother described her own shift: “The first two weeks were a blur, but by the third week, the waiting became unbearable. That was when something shifted”. A physician reflected: I began to see a pattern. By the third and fourth weeks, families were desperate for something to hold onto—and that was when narrative support seemed to make the most difference. I began to see a pattern. By the third and fourth weeks, families were desperate for something to hold onto—and that was when narrative support seemed to make the most difference. Nurses also noted that this window required intensified empathic presence. Theme 3: Bidirectional Healing Healthcare providers described emotional restoration and renewed professional meaning as they witnessed children’s recovery and family gratitude. A nurse wrote: “I came home exhausted for weeks, but when I heard that little girl say ‘thank you’ on a voice message, I cried—and I knew why I do this job”. A physician reflected: “She no longer needed the Bible at her bedside. She said, “With you here, it’s been put away”. That healed me as much as I healed her”. Another nurse described a moment of renewal: Seeing him walk into school after a year of rehabilitation—that healed something in me too. I had doubted whether we were making any difference. That moment answered the question. Seeing him walk into school after a year of rehabilitation—that healed something in me too. I had doubted whether we were making any difference. That moment answered the question. A physician added: “When a family says ‘we trust you,’ it reminds me that we are in this together. That is not just good for the patient—it is good for the doctor”. These qualitative themes provide explanatory depth to the quantitative improvements in parental psychological states and provider well‑being, and they are consistent with the proposed concepts of “clinical narrative signals”, “window of narrative intervention”, and “bidirectional healing”. Detailed illustrative quotes are provided in Supplementary Table S1. Integration of Quantitative and Qualitative Findings The joint display (Supplementary Table S2) summarizes the convergence between quantitative and qualitative results across the three stakeholder groups. In all instances, qualitative themes aligned with the direction and magnitude of quantitative changes, providing convergent support for the findings. For example, the large effect sizes for reductions in parental anxiety (d = 9.7) and depression (d = 6.5) were mirrored in mothers’ narratives of transitioning from hopelessness to hope. Similarly, the improvements in nurse empathy (IRI) and burnout (NBS) were echoed in reflective writings describing renewed professional meaning. Discussion This mixed‑methods study of five children with severe anti‑NMDAR encephalitis suggests that narrative medicine interventions may be associated with favourable trends across children, parents, and healthcare providers. Below we discuss the key findings in a focused, point‑by‑point manner. Core Findings in Context All five children had severe disability at admission (mRS ≥ 4). At six‑month follow‑up, median mRS improved to 1, with three children returning to school. Parental anxiety, depression, and caregiver burden decreased with effect sizes of notable magnitude (Cohen’s d 3.4–9.7). Nurses’ empathy increased (IRI d=1.9) and burnout decreased (NBS d=2.1); physicians showed parallel improvements and increased meaningful work (WAMI +15.5). In this small sample, these effect sizes were substantial and suggest meaningful patterns that warrant further investigation in larger prospective studies.25 Clinical Narrative Signals: A Novel Early Indicator In all five cases, subtle, non‑standardized responses—eye closure, first smile, attempted swallow—preceded measurable mRS improvement by days to weeks. We term these clinical narrative signals. They may represent the patient’s “voice” when verbal communication is impossible. In locked‑in or minimally conscious states, such signals may be the only available evidence of consciousness.26–28 Documenting them in parallel charts provides a high‑resolution, longitudinal record that complements episodic, scale‑based assessments.29 We propose that parallel charts may function as story‑based patient‑reported outcomes for non‑communicative patients—a concept that appears not to have been previously operationalized in neurocritical care. A Window of Narrative Intervention (weeks 2–4) In our observations, families showed maximal receptivity to narrative support between the second and fourth week after admission, a period marked by peak diagnostic uncertainty and emotional distress. This temporal pattern suggests a window of narrative intervention.30 The concept parallels well‑established therapeutic windows in neurorehabilitation and psychopharmacology. Mechanistically, this window may represent a transition from acute shock to active meaning‑making, in which narrative reconstruction may be most effective.31,32 Clinically, it implies that psychosocial resources could be intensified during weeks 2–4, rather than uniformly distributed—a potential precision approach to narrative support. The timing of narrative signals relative to immunotherapy further supports this conceptualization. Across the five patients, narrative signals—eye closure, first smile, first eye contact, first swallow, and first vocalization—emerged at Day 1, Day 20, Day 21, Day 30, and Day 35, respectively, with a mean of 21.4 days. Notably, in all cases, these signals appeared prior to measurable mRS improvement, regardless of the timing of second-line rituximab initiation (which ranged from Day 1 to Day 30). This temporal pattern suggests that narrative signals may not be a direct consequence of immunotherapy response, but rather may reflect a distinct dimension of recovery—conscious engagement and relational responsiveness—that is not captured by conventional neurological scales such as mRS. This observation strengthens the rationale for intensifying narrative support during weeks 2–4, when families are most receptive and when early signs of consciousness recovery may be detectable through attentive narrative listening. Bidirectional Healing: Providers Benefit as Well Nurses and physicians not only delivered care but also experienced improved empathy, reduced burnout, and increased meaningful work. This bidirectional healing effect—providers being emotionally restored through the process of helping families—has been hypothesized but rarely quantified. Our data provide initial empirical support. This finding aligns with post‑traumatic growth theory applied to healthcare professionals, suggesting that narrative sharing sessions may transform emotional labor into professional reward.33,34 From a health systems perspective, narrative medicine could serve as a low‑cost intervention to mitigate workforce burnout.35,36 Integration with Emerging Mechanistic Theories We observed a temporal association between parental narrative reconstruction and subsequent clinical improvement (approximately one‑week lag). Whether this reflects a direct influence of parental stress reduction on the child’s neuroimmune environment, as suggested by psychoneuroimmunological research, remains speculative and would require biomarker‑based investigations. Nevertheless, the active elicitation of clinical narrative signals (eg, “close your eyes if you can hear me”) created a cycle of recognition and hope that distinguishes this approach from unidirectional narrative interventions such as photo‑narrative.37 These observations suggest that bidirectional narrative communication may offer advantages over unidirectional communication approaches. Ketogenic diet therapy, while sometimes considered as an adjunctive option for controlling refractory seizures secondary to encephalitis, was not used in any of the patients in this cohort.38 Strengths, Limitations, and Implications This study has several strengths. To our knowledge, it is the first to provide triadic, mixed‑methods evidence for narrative medicine in pediatric anti‑NMDAR encephalitis. The integration of validated quantitative instruments with qualitative data allowed us to identify three novel concepts—clinical narrative signals, a window of narrative intervention, and bidirectional healing—that may be clinically relevant. The effect sizes for parental psychological outcomes (Cohen’s d up to 9.7) and provider well‑being (d up to 2.1) were substantial in this small sample and warrant replication in larger studies. These findings suggest potential avenues for low‑cost, psychosocial interventions in pediatric neurocritical care. Limitations include the retrospective design (recall bias partly mitigated by structured anchor points), small sample (n=5), lack of a control group, single‑center setting, and potential allegiance bias (the first author was a participant). Causality cannot be established from this study. The modified Rankin Scale, while widely used in neurological research, may not fully capture the nuanced functional and quality‑of‑life changes most relevant to children with autoimmune encephalitis and their families. Nevertheless, the convergence of quantitative and qualitative data and the temporal association between narrative turning points and clinical improvement support the plausibility of an intervention effect. Clinically, we suggest routine documentation of clinical narrative signals, intensified narrative support during weeks 2–4 after admission, weekly nurse sharing sessions, and narrative family conferences for families experiencing distrust or anger. Future research should include prospective controlled trials with baseline assessments, longer follow‑up (12–24 months), multi‑center designs, and mechanistic studies incorporating stress biomarkers and neuroimaging. The proposed concepts merit validation in larger cohorts using time‑series designs. Conclusion This mixed‑methods study provides preliminary triadic evidence suggesting that narrative medicine interventions may be associated with favourable outcomes in children with severe anti‑NMDAR encephalitis, their parents, and their healthcare providers. Children showed notable neurological recovery, parents experienced reductions in anxiety, depression, and caregiver burden, and nurses and physicians demonstrated enhanced empathy, reduced burnout, and increased work meaningfulness. We introduce three novel concepts—clinical narrative signals, a window of narrative intervention (weeks 2–4 after admission), and bidirectional healing—that may offer a practical framework for integrating narrative practices into pediatric neurocritical care. These preliminary findings suggest that narrative medicine may be a low‑cost, high‑value adjunct to standard care, but prospective, controlled studies are needed to confirm causality and explore underlying mechanisms.