Implementing a Family-Centered Rounds Intervention Using Novel Mentor-Trios

Abstract

Background and Objectives:

Patient and Family Centered I-PASS (PFC I-PASS) emphasizes family/nurse engagement, health literacy, and structured communication on family-centered rounds, organized around the I-PASS framework. We assessed adherence, safety, and experience after implementing PFC I-PASS using a novel “Mentor-Trio” implementation approach with multidisciplinary parent-nurse-physician teams coaching sites.

Methods:

Hybrid Type II effectiveness-implementation study from 4/29/2019–3/13/2022 with ≥3 months of baseline and 12 months of post-implementation data collection/site across 21 US community and tertiary pediatric teaching hospitals. We conducted rounds observations and surveyed nurses, physicians, and Arabic/Chinese/English/Spanish-speaking patients/parents.

Results:

We conducted 4557 rounds observations and received 2285 patient/family, 1240 resident, 819 nurse, and 378 attending surveys. Adherence to all I-PASS components, bedside rounding, written rounds summaries, family/nurse engagement, and plain language improved post-intervention (13.0–60.8% absolute increase by item), all p<.05. Except for written summary, improvements sustained 12 months post-implementation. Resident-reported harms/1000-resident-days were unchanged overall but improved in larger hospitals (116.9 to 86.3 to 72.3 pre vs early vs late-implementation, p=.006), hospitals with greater nurse engagement on rounds (110.6 to 73.3 to 65.3, p<.001), and greater adherence to I-PASS structure (95.3 to 73.6 to 72.3, p<.05). Twelve of 12 measures of staff safety climate improved (e.g., excellent/very good safety grade improved from 80.4% to 86.3% to 88.0%), all p<.05. Patient/family experience and teaching were unchanged.

Conclusions:

Hospitals successfully used Mentor-Trios to implement PFC I-PASS. Family/nurse engagement, safety climate, and resident-reported harms improved in larger hospitals and hospitals with better nurse engagement and intervention adherence. Patient/family experience and teaching were not affected.

Article Summary:

This article describes changes in safety and adherence after implementing a family-centered rounds intervention across 21 hospitals using a novel multidisciplinary parent-nurse-physician “Mentor-Trio” coaching strategy.

Introduction

Communication failures frequently cause harmful errors.¹ High-reliability structured provider communication improves patient safety.² However, communication with patients/families is understudied. Patient/family-centered rounds (PFCR) improve provider-patient/family communication^3–6 and are a best practice per the Institute for Patient and Family-Centered Care (IPFCC)³ and the AAP.^7,8 Although common in pediatrics, PFCR have not fully achieved their potential to improve communication because they are often implemented inconsistently, more physician- than family-centered, replete with medical jargon, and inconsistently engage patients/families and nurses.^4,5,9–12 Support for PFCR also varies among residents, attendings, and nurses.¹³

We previously coproduced with parents/nurses/physicians a PFCR intervention, “Patient and Family Centered I-PASS” (PFC I-PASS).^14,15 PFC I-PASS emphasized family/nurse engagement, health literacy, and high-reliability structured bidirectional written/verbal communication organized around I-PASS (Illness severity-Patient summary-Action items-Situational awareness-Synthesis by receiver) framework. After implementing PFC I-PASS in 7 centers, harmful medical errors dropped 38% and hospital experience improved; rounds duration and teaching remained unchanged.¹⁴

To inform widespread dissemination, we sought to test PFC I-PASS implementation and real-world effectiveness in a broader group of US community and tertiary pediatric teaching hospitals. Recognizing that hospital-based QI interventions are often physician-centric or nurse-centric,¹⁶ we implemented PFC I-PASS using a novel inter-disciplinary implementation strategy involving coaching by parent, nurse, and physician mentors (Mentor-Trios). We hypothesized that intervention adherence/sustainment, safety climate, harmful errors, and hospital experience would improve post-implementation.

Methods

Design

We conducted a prospective hybrid-type-II-effectiveness-implementation¹⁷ study on 21 US teaching hospitals’ pediatric inpatient units from 4/29/2019–3/13/22. Hybrid-type II-effectiveness-implementation studies equally focus on intervention effectiveness and implementation.¹⁷ We collected staggered data across sites (≥3 months pre-intervention, then 12-months intensive mentored-implementation data collection/site). Boston Children’s Hospital’s IRB provided approval.

Sites

Thirty sites applied to participate, providing information about hospital, residency program, study team, PFCRs and I-PASS handoffs experience. A multidisciplinary parent-nurse-physician team selected 21 sites (eFigure 1) based on geographic diversity, hospital type (tertiary/community), residency size, and PFCR experience.

Sites included 14 nested children’s hospitals, 4 freestanding children’s hospitals, and 3 general hospitals. Twelve were tertiary care teaching hospitals, nine community teaching hospitals. Residency size ranged from 16–150. Rounding team structure varied (e.g., interns/third-year-residents vs interns/second-year-residents). Twenty hospitals implemented on general pediatric (non-intensive care) units, one on pediatric neurology. Sites variably practiced PFCRs (<5 years [n=5], 5–10 years [n=9], >10 years [n=7]); none previously used I-PASS for PFCR.

Intervention

Our intervention, implementation of PFC I-PASS,^14,18 is a structured verbal/written rounds communication framework organized around I-PASS structure, supported by communication, teamwork, and health literacy best practices to ensure a shared mental model (Figure 1).¹⁴ We adapted PFC I-PASS for larger-scale implementation by streamlining resident training and incorporating novel educational techniques (e.g., flipped-classroom, virtual learning).^15,19

Figure 1. PFC I-PASS Intervention.

Key intervention supporting activities include education and training activities for medical students, nurses, residents, and attending physicians using a flipped classroom and interactive simulations, as well as implementation and sustainment activities that ensure practice change, including observations and feedback.

The PFC I-PASS rounding process included several required elements, including pre-rounds planning, accounting for interpretation needs, infection-control precautions, and ensuring all key team members were present. During rounds, providers introduced team members, invited patient/family to share first, nurse to share second, and then resident/advanced care provider to share next. Health literacy best principles (e.g., avoiding medical jargon, using simple vocabulary), active listening, and nonverbal cues were incorporated into rounds. Rounding communication was structured through the I-PASS format, with Illness severity provided by the patient/family, Patient summary provided by the nurse and resident/advanced care provider, Action items and Situational Awareness provided by the resident/advanced care provider, and Synthesis provided by the patient/family with input from nursing. Providers then completed a written summary of rounds (Rounds Report).

The mentored implementation model, pioneered by the Society of Hospital Medicine, involves dedicated mentorship of each site by a mentor with expertise in the intervention. In our study, we modified this mentored implementation to include a novel “Mentor-Trio” approach where each site is paired with a Mentor-Trio consisting of a parent, nurse, and physician mentor. Mentor-Trios meet with Mentee-trios (parent, nurse, and physician champions) at each of the 21 sites monthly during the duration of the project to review stakeholder engagement, assess intervention adherence through review of monthly run charts, brainstorm potential plan-do-study-act (PDSA) cycles, and discuss intervention challenges and successes.

Our implementation strategy (Mentored Implementation), pioneered by the Society of Hospital Medicine²⁰ in multiple domains,^21–23 involves content experts (typically physicians) intensively coaching hospitals to implement large-scale QI projects. Informed by coproduction²⁴ with families and nurses, we refined mentored implementation using a novel “Mentor-Trio” approach, where triads of parent-nurse-physician mentors drawn from 7 parents, 6 nurses, and 16 physicians coached sites in implementing PFC I-PASS. Most Mentor-Trios previously participated in our 7-center PFC I-PASS Study.¹⁴ Experienced senior physician-mentors further guided Mentor-Trios. Mentors participated in monthly all-mentor calls for mentor training/support on PFC I-PASS and received pre-templated slide decks with standardized, tailored prompts for monthly site calls. We paired central Mentor-Trios with local “Mentee-Trios” (parent/nurse/physician project champions) leading site-level implementation. Mentor-Trios met with sites monthly via teleconference to review stakeholder engagement, intervention adherence (run charts), plan-do-study-act cycles, and intervention challenges/strategies.

When COVID-19 began, sites varied in implementation phase: pre-implementation (n=1), early-implementation (months 1–6)(n=13), late-implementation (months 7–12)(n=7). We made several ensuing rounds adaptions (e.g., incorporating telemedicine, eFigure 2) and paused data collection for 3–6 months across sites (3/2020–9/2020).

Data Collection/Measurement

To measure intervention adherence, site providers and, rarely, site parents observed rounds in real-time for ≥3 patients/week/site using a previously developed observation tool.^2,5,14 During the pandemic, some observations occurred virtually. Observers underwent video- and simulation-based training and rated key rounding aspects, including format (i.e., rounding at bedside)(closed-ended); I-PASS component adherence (dichotomous); communication quality (4-point-Likert-extent-scale); nurse presence (4-point-categorical-scale), family/nurse engagement (5-point-behaviorally-anchored-scale); and teaching (dichotomous).

To measure safety climate and resident-reported harms, we surveyed resident-physicians at end-of-rotation (~every 2–4 weeks) and nurses/attendings at staff meetings. Staff provided completion-based-consent, completing surveys on paper/REDCap. For safety climate, we assessed select AHRQ Survey on Patient Safety Culture²⁵ 5-point-Likert-scale items, including working as a team (Strongly Disagree-Strongly Agree), staff freely speaking up (Never-Always), and safety grade (Failing-Excellent). Participants also reported minor or major harms (adverse events) due to communication failures using items modified from prior studies.^14,21 Minor harms had limited clinical consequence (i.e., requiring more frequent monitoring or transient discomfort, without prolonging hospitalization, significant organ dysfunction, or worsening clinical condition). Major harms had significant clinical consequence (i.e., deteriorating clinical status, organ dysfunction, prolonged hospitalization, disability beyond discharge, death). We refined staff surveys using cognitive interviewing/piloting.

For patient/family experience, we surveyed 2–4 randomly selected Arabic/Chinese/English/Spanish-speaking inpatients (≥13-years-old) or parents/caregivers (“families”) weekly/site about safety climate, experience, and demographics. Families provided verbal consent with interpreters when necessary; patients ≥13-years-old provided assent. Patients/families self-completed paper (all 4 languages) or REDCap (English) surveys. Surveys included select 5-point Likert-scale Children’s Hospital Safety Climate Questionnaire²⁶ items about speaking up, questioning decisions, and fear asking questions (Strongly Disagree-Strongly Agree). They also included hospital experience items modified from prior research relating to how well doctors/nurses addressed concerns and made them feel part of the team (Not at all-Extremely).¹⁴ They also completed Universal Health Literacy Precautions Toolkit items for language proficiency,²⁷ the 3-item screener for health literacy,²⁸ and self-reported demographics (including race/ethnicity). Surveys had a 6^th grade Flesch-Kincaid reading level and underwent piloting/cognitive interviewing and professional translation into Arabic/simplified Chinese/Spanish.

Outcomes

Our primary implementation outcome was adherence to all 5 I-PASS components on rounds (dichotomous). Our primary clinical-effectiveness outcome was overall resident-reported harm (major+minor) rates.²¹ Secondary outcomes included top-2-box-rated safety climate and patient/family experience scores. (“top-2-box”= top-two Likert-scale options, e.g., Strongly Agree/Agree). We defined implementation, clinical-effectiveness, service, and experience outcomes using implementation research standards.²⁹ Process outcomes included family engagement (top-2-box), nurse engagement (top-2-box), plain language (top-2-box), and teaching (dichotomous).

Analyses

We compared respondent sociodemographic characteristics by intervention time-period (pre-/early-/late-intervention) using chi-square tests or Fisher’s exact tests. We estimated adherence rates to key intervention elements for each implementation month (normalized to intervention go-live, not calendar-month) using mixed-effects logistic regression with random intercepts controlling for site-level clustering. These included individual I-PASS components (e.g., illness severity) and composite outcomes (adherence to all I-PASS components). Interrupted-time-series³⁰ tested for intervention effect on linear trend over time of estimated monthly adherence rates.

We used mixed-effects Poisson regression to estimate resident-reported harm rates, testing for changes over time via tests for linear trend over time-periods.

For safety climate and experience, we dichotomized the sample into participants with and without top-2-box scores (4–5 vs 1–3 of 5). We examined bivariate associations between having top-2-box scores and intervention time-period using mixed-effects logistic regression and tested for trends over time. We adjusted top-2-box staff safety climate scores for characteristics varying across time-periods.

We explored whether the intervention’s effect on key outcomes in the late-intervention period differed by hospital size (number of pediatric staffed beds), family engagement, nurse engagement, or adherence to all I-PASS components. We dichotomized hospital size, family engagement, nurse engagement, and adherence by median hospital values for each characteristic. To test for effect-modification of hospital characteristics, we evaluated differences in trends in outcomes over time by hospital characteristic by including an interaction-term between the characteristic and time-period in regression models. Statistical significance was achieved with a two-sided p-value <.05. We used REDCap³¹ for data collection/management and SAS v9.4 (SAS Institute) for analyses.

Results

Participant Characteristics

We conducted 4457 rounds observations, and received surveys from 2285 patients/families, 1240 resident-physicians, 819 nurses, and 378 attending-physicians. Survey response rates were 72.8%, 50.6%, 33.1%, and 41.7%, respectively (53.0% overall). Most patient/family participants were parents (89.2%), female (82.6%), 18–34 years-old (48.1%), had some college (64.4%), and an income <$50,000/year (45.2%). The most common race/ethnicity was White non-Hispanic (49.1%), followed by Hispanic (23.0%), and Black non-Hispanic (14.0%); 13.8% had limited English proficiency and 24.9% had limited health literacy. Patient/family characteristics did not differ significantly between implementation-periods. Staff were predominantly resident-physicians (50.9%), female (75.4%), 18–34 years-old (69.2%), White non-Hispanic (65.0%), and in practice 1–5 years (40.2%). Staff role, gender, age, and years in practice significantly varied by implementation-period (Table 1).

Table 1a.

Patient/Family Characteristics (n=2285)

Characteristic	Overall N (2285) (%)	Pre-Implementation N (619) (%)	Early-Implementation N (866) (%)	Late-Implementation N (800) (%)
Relationship to patient (n=2256)
Patient	155 (6.9)	46 (7.5)	61 (7.1)	48 (6.1)
Parent	2012 (89.2)	541 (88.3)	765 (89.6)	706 (89.5)
Grandparent	58 (2.6)	19 (3.1)	16 (1.9)	23 (2.9)
Othera	31 (1.4)	7 (1.1)	12 (1.4)	12 (1.5)
Age, years (n=2248)
<18	100 (4.4)	33 (5.4)	38 (4.5)	29 (3.7)
18–34	1081 (48.1)	299 (49.0)	422 (49.6)	360 (45.7)
35–54	1000 (44.5)	261 (42.8)	368 (43.2)	371 (47.1)
≥55	67 (3.0)	17 (2.8)	23 (2.7)	27 (3.4)
Gender (n=2230)
Female	1843 (82.6)	501 (82.3)	693 (82.2)	649 (83.4)
Male	387 (17.4)	108 (17.7)	150 (17.8)	129 (16.6)
Race and Ethnicity (n=2238)
Asian, non-Hispanic	132 (5.9)	35 (5.8)	48 (5.7)
Black, non-Hispanic	314 (14.0)	81 (13.4)	110 (13.0)	123 (15.7)
Hispanic or Latino	515 (23.0)	142 (23.5)	195 (23.0)	178 (22.7)
White, non-Hispanic	1099 (49.1)	300 (49.7)	412 (48.5)	387 (49.3)
Other, non-Hispanicb	178 (8.0)	46 (7.6)	84 (9.9)	48 (6.1)
Language Proficiency (n=2253)
Limited English proficiency (LEP)c	312 (13.8)	83 (13.7)	121 (14.1)	108 (13.7)
English proficient	1941 (86.2)	524 (86.3)	736 (85.9)	681 (86.3)
Language preferred by patients/families with LEP (n=312)
Arabic	2 (0.6)	1 (12)	1 (0.8)	0 (0)
Chinese	4 (1.3)	1 (12)	1 (0.8)	2 (1.9)
English	159 (51.0)	48 (57.8)	61 (50.4)	50 (46.3)
Portuguese	1 (0.3)	1 (12)	0 (0)	0 (0)
Spanish	121 (38.8)	29 (34.9)	45 (37.2)	47 (43.5)
Otherd	14 (4.5)	3 (3.6)	8 (6.6)	3 (2.8)
Health Literacy (n=2247)
Limited health literacye	560 (24.9)	149 (24.6)	231 (27.1)	180 (22.8)
Adequate health literacy	1687 (75.1)	456 (75.4)	623 (73.0)	608 (77.2)
Education (n=2232)
No college	794 (35.6)	220 (36.5)	308 (36.5)	266 (33.8)
≥Some college	1438 (64.4)	383 (63.5)	535 (63.5)	520 (66.2)
Annual household income (n=2285)
<$14,999	285 (12.5)	71 (11.5)	112 (12.9)	102 (12.8)
$15,000 - $49,999	747 (32.7)	225 (36.4)	276 (31.9)	246 (30.8)
$50,000 - $99,999	540 (23.6)	138 (22.3)	200 (23.1)	202 (25.3)
$100,00	479 (21.0)	119 (19.2)	196 (22.6)	164 (20.5)
Missing	234 (10.2)	66 (10.7)	82 (9.5)	86 (10.8)

Patient/parent characteristics derived from self-reported survey data. Counts may not add up to column totals due to missing data (<3% missing except for annual household income). Patient/parent characteristics derived from self reported survey data.

^aOther relationship category includes 31 respondents who provided the following write-in categories: l/g (n=1); foster parent (n=4); primo (n=1); sister/primary caregiver (n=1); cousin (n=2); aunt, legal guardian (n=1); aunt (n=2); ant [sic] (n=1); guardian (n=3); uncle (n=1); foster mom (n=2); foster parent (n=1); brother (n=1); sister (n=4); hermana (n=1); girlfriend (n=1); sister of patient (n=1), guardian/caregiver (n=1); 2 who did not further specify the “other” relationship.

^bThe “Other, non-Hispanic” category for race and ethnicity includes 69 respondents who selected more than one race: White and Other (n=6); Native Hawaiian/Other Pacific Islander and Other (n=1); Native Hawaiian/Other Pacific Islander and White (n=5); Native Hawaiian/Other Pacific Islander and White and Other (n=1); Black and Other (n=2); Black and White (n=17); Black and White and Other (n=1); Black and Native Hawaiian/Other Pacific Islander (n=1); Asian and Other (n=2); Asian and White (n=4); Asian and Native Hawaiian/Other Pacific Islander (n=6); Asian and Native Hawaiian/Other Pacific Islander and Other (n=1); Asian and Native Hawaiian/Other Pacific Islander and White (n=3); Asian and Black (n=1); American Indian or Alaskan Native and White (n=13); American Indian or Alaskan Native and Native Hawaiian/Other Pacific Islander (n=1); American Indian or Alaskan Native and Black (n=1); American Indian or Alaskan Native and Black and White (n=2); American Indian or Alaskan Native and Asian and Black (n=1).

44 respondents also provided the following write-in categories: Middle Eastern (n=4); from middle east (n=1); Mixed (Indian/Black Jamaican) (n=1); Arabian (n=1); Arabic (n=2); Samoan (n=2); French Indian/German (n=1); Mix Hawaiian (?)[sic] (n=1); India (n=1); Ukrainian (n=1), Human (n=2); Multiracial (n=2); Native Hawaiian (n=1); Filipino (n=2); Aboriginal American (n=1), Pakistani (n=1); Indian (n=2); Mixed (n=1); Micronesian (n=1); Latino (n=1); Latina (n=1); Hispanic (n=2); Armenian (n=1); Native (n=1), Puerto Rican (n=1); Jewish (n=1); Phillipino [sic] (n=1); Everything - Mix Plate (n=1); Moroccan (n=1); Somali (n=2); North African/Middle Eastern (n=1); Samoans (n=1); Indian, Black, and White (n=1). The remainder of respondents selecting “Other” did not further specify their race.

^cLimited English proficiency was defined as responding anything other than “very well” to the question “how well do you speak English?” Response items range from very well, well, not well, to not at all. This item was taken from the Guide to Implementing the Health Literacy Universal Precautions Toolkit.²⁷

^dOther languages preferred by patients/families with LEP includes 14 respondents who provided the following write-in categories: American sign language (n=1); Bengali (n=1); Haitian creole (n=1); Hindi (n=1); Polish (n=1); Russian (n=2); Somali (n=4); Urdu (n=1); 2 who did not further specify the “other” language.

^eLimited health literacy was defined as selecting a non-top-box answer for 1 or more of the following 3 questions: (1) “How often do you have someone (like a family member, friend, hospital/clinic worker, or caregiver) help you read hospital materials?” (non-top-box responses: all, most, some, or a little of the time); (2) “How often do you have problems learning about your medical condition because of difficulty understanding written information?” (non-top-box responses: all, most, some, or a little of the time); and (3) “How confident are you filling out forms by yourself?” (non-top-box responses: quite a bit, somewhat, a little bit, or not at all). This item was taken from the Chews 3-item screener.²⁸

Intervention Adherence

At time of intervention, proportion of observations demonstrating bedside rounding with families improved (from 76.5% to 89.5%), as did adherence to all 5 I-PASS components on rounds (25.5% to 60.8%). Particularly strong improvements were seen in Illness severity (65.8% to 95.4%) and Synthesis by receiver (53.6% to 84.1%). Immediate improvements in written rounds summary (29.8% to 72.4%), family engagement (70.9% to 87.6%), nurse engagement (33.4% to 63.0%), and plain language (50.2% to 60.6%) also occurred, all p<.05. Except for written summary, improvements sustained during early- and late-implementation (Figure 2). Observed trainee teaching did not change by implementation period.

Figure 2. Adherence to Key PFC I-PASS Intervention Elements by Implementation Month.

Harms

Resident-reported overall, minor, and major harms were unchanged by implementation-period (Figure 3, etable1). However, there were differential decreases in overall harms in larger-sized hospitals and hospitals with greater nurse engagement and I-PASS adherence, compared to their counterparts. In larger hospitals, overall harms per 1000-resident days fell from 116.9 to 86.3 to 72.3 (p=.006), a 38.2% reduction. In smaller hospitals, harms were unchanged: 74.9 to 88.8 to 86.4 (p=.85). There was a significant differential change in harms by size (p=.03). In hospitals with greater nurse engagement, overall harms fell from 110.6 to 73.3 to 65.3 (p<.001), a 41.0% reduction. In hospitals with lesser nurse engagement, harms were unchanged: 80.0 to 102.8 to 93.9 (p=.39). There was a significant differential change in harms by nurse engagement (p<.001). In hospitals with higher adherence to I-PASS structure, overall harms fell from 95.3 to 73.6 to 72.3 (p<.05), a 24.1% reduction. In hospitals with lower adherence to I-PASS structure, harms were unchanged: 91.8 to 100.6 to 85.9 (p=.62). There was no significant differential change by I-PASS adherence (p=.26).

Figure 3. Resident-Reported Harms by Implementation Time Period, Overall and by Hospital Characteristics.

Safety Climate

Adjusted top-2-box staff safety climate scores significantly improved pre- vs early- vs late-implementation for all 12 items (p<.05), including working together as a team (90.4% to 93.1% to 94.4%), staff freely speak up (83.1% to 84.5% to 89.0%), and overall safety grade (80.5% to 86.3% to 88.0% excellent/very good)(Figure 4). Staff safety grade improved in smaller hospitals (74.6% to 86.2% to 87.8%, p<.001, but not in larger hospitals 86.2% to 86.0% to 88.1%, p=.44 [p=.02 for difference by size])(eFigure 3).

Figure 4. Staff Safety Climate (% Top-2-box).

Top-2-box patient/family safety climate scores for overall safety grade remained unchanged, though ratings for staff are afraid to ask questions (never/rarely [reverse-coded] went from 68.0% to 72.0% to 73.2% (p=.06)(eFigure 4).

Experience

Top-2-box patient/family experience did not significantly change by implementation-period (eTable 2). However, top-2-box patient/family experience for nurses making you feel part of the team improved for larger hospitals (90.4% to 93.2% to 95.1%, p=.02) vs 96.8% to 95.7% to 94.8% for smaller hospitals, p=.19, (p=.01 for difference by size).

Discussion

In this 21-center hybrid-effectiveness-implementation study, a novel parent-nurse-physician Mentor-Trio implementation strategy was associated with sustained high adherence to core PFC I-PASS components. Tests of “real-world” effectiveness of PFC I-PASS showed that resident-reported patient harms improved for larger-sized hospitals and those with higher PFC I-PASS adherence and greater nurse engagement. Staff safety climate improved, without negatively affecting patient/family experience or trainee teaching. Our findings indicate that the coproduced Mentor-Trio strategy was a promising strategy to implement PFC I-PASS on a larger scale and in more diverse contexts. Our study also provides further evidence about feasibility and effectiveness of structured PFCR in diverse settings and authentic partnership between patients/families, nurses, and physicians to successfully implement complex hospital-based QI interventions.

Our novel implementation approach for PFCR, , in which parents, nurses, and physicians ledtogether was feasible and well-received, even during a global pandemic, when many hospitals limited PFCR and bedside caregivers.^32–34 Our study demonstrates how family partnership can be accomplished virtually (not just in-person), allowing family co-investigators to more easily participate in hospital-based projects. Co-production between patients/families, clinicians, and hospitals is important²⁴ but underutilized, particularly in QI efforts. Our Mentor-Trio approach, which embraces coproduction principles called for by the IPFCC and AAP,^3,7,8 can be applied to other projects. By authentically involving physicians, nurses, and patients/families together rather than utilizing siloed QI efforts, hospitals may more successfully implement interventions. Nursing and parent participation in Mentor-Trios also helped address nursing-specific and family-specific implementation challenges (e.g., nursing bedside presence, family perceptions of trainee redirection on rounds).

Robust studies examining PFCR implementation are lacking: most are single-site QI projects.¹¹ Our findings and prior research show that adherence to complex I-PASS-structured communication interventions during resident handoffs^2,35 and PFCR¹⁴ is possible. Compared to the prior 7-center PFC I-PASS Study, adherence to PFC I-PASS components, like synthesis by receiver (teach-back), a health literacy best-practice that improves outcomes, was higher in the current study.³⁶ This may be due to several modifications to the intervention, including use of the Mentor-Trio model, and better training about teach-back. The growing evidence-base around PFC I-PASS may also have contributed, as providers may be more willing to implement interventions with a stronger evidence-base. Harms improved more in hospitals with better nurse engagement and patient/family experience with nursing improved in larger hospitals; this result is consistent with prior research showing structured communication interventions with modest nursing investments significantly improve nursing-related measures.^14,37,38 Indeed, nurse engagement in QI efforts is crucial in promoting hospital safety and quality.^39–41

We again found no decrease in resident teaching, a balancing measure and common PFCR criticism.^6,13 Although this study did not measure rounds duration and results of other PFCR studies vary,¹¹ our 7-center study found no increase in rounding duration with PFC I-PASS.¹⁴ Ensuring all team members, including nurses and families, are present and engaged in rounds can promote shared understanding and save time¹² by obviating downstream clarifications and streamlining communication throughout the day.

Our findings differ somewhat from our 7-center study, in which harmful errors decreased overall post-implementation.¹⁴ The resident-reported, communication-related harms effectiveness measure in this study is less sensitive than the intensive systematic safety surveillance measure in the prior study, but was chosen as a more practical, real-world alternative. The lack of overall harm improvement in this study may be due to decreased measure sensitivity, or differential effectiveness of the intervention across different centers, or situations where nurse engagement, intervention adherence, or other unmeasured factors vary. We may have been underpowered to detect more subtle improvements in harms. Additionally, COVID-19 may have adversely affected safety,^42–45 further confounding findings. That staff safety climate improved post-intervention across centers is promising, as safety climate is associated with directly measured safety outcomes.^46–48 Given provider burnout^49–51 following COVID-19, improving safety climate is important in its own right.

We observed no improvements in patient/family experience post-implementation. However, hospital experience was high at baseline. Additionally, our intervention focused on communication structure, not other aspects of interpersonal communication like trust-building, which may influence experience. While patient/family experience did not improve, it did not worsen post-intervention, despite the pandemic; it is possible (though unknown) that the intervention might have mitigated a pandemic-driven deterioration in patient/family experience.^32,52,53

Operationalizing effective PFCR interventions requires support from front-line staff and leaders, key elements of our implementation process. Barriers to nursing PFCR involvement exist, like competing time-demands (e.g., 8am medications).^11,12 Frontline nursing/ leadership support and nurse-centric PFCR workflow changes may be needed (e.g., rounding by nursing assignment, notifying nurses 1–2 patients in advance).

Our study had limitations. Our design—appropriate for a hybrid effectiveness-implementation study—limits conclusions about causality. However, our interrupted-time-series design strengthens rigor of analyses compared with simple pre-post designs and our study remains one of the most robust PFCR studies. Although this study provides evidence supporting broad PFC I-PASS implementation, we focused on teaching hospitals’ pediatric services. Additional research is required to determine whether PFC I-PASS generalizes to other specialties/hospital types. Although we surveyed Arabic, Chinese, and Spanish-speakers, patients speaking less common languages may have different experiences. Similarly, in our study (and much of medicine), physicians’/nurses’ racial/ethnic composition was different than patients’. Hospital experience and other outcomes might be improved with a more diverse physician/nursing workforce with greater patient-provider racial/ethnic concordance.^54,55 This is an important area for future study by our study group and the field. Rounds observations were primarily conducted by physician observers, so may have been subject to bias. Measurement of resident-reported harms was incomplete and potentially subject to reporting and social desirability bias but pragmatic based on study resources and the desire to more closely mimic usual operational conditions in data collection; similar resident-reported harm measures have been used successfully previously.²¹ COVID-19 may have confounded findings due to provider burnout, patient experience, and staffing; if so, our intervention might have been more beneficial than the current study suggests. While we made strides in family and nurse engagement, further improvement in nursing engagement on rounds is required, perhaps by understanding attitudes, beliefs, and competing demands.¹² Finally, we may have been underpowered to detect differences in harms, patient/family safety climate, and experience, and a ceiling effect may have existed with the latter two. Future work should study nurse engagement and PFCR disparities and best practices for patients speaking other languages (an ongoing focus of our study group), adults, and subspecialties.

Conclusions

A novel Mentor-Trio modification of mentored implementation was an effective strategy to implement PFC I-PASS in pediatric inpatient units. This larger-scale implementation was associated with significant improvements in provider and patient/family communication processes and safety climate. Additionally, in larger hospitals and hospitals with better nursing engagement and adherence to PFC I-PASS structure, PFC I-PASS was associated with reductions in resident-reported harms. The current intervention was successful in diverse teaching hospital settings despite challenges of a global pandemic. Our findings suggest that hospitals should better engage patients/families and nurses at the bedside during rounds. Doing so can improve patient safety, safety climate, and patient/family and staff engagement, increasing self-efficacy, coproduction,²⁴ and family-centeredness. Lastly, our findings provide evidence that coproduction of QI efforts by patients/families, nurses, and physicians working as a team can drive sustained transformational changes in hospitals.

Table 1b.

Resident Characteristics (n=1240)

Characteristic	Overall n (1240) (%)	Pre-Implementation N (363) (%)	Early-Implementation N (457) (%)	Late-Implementation N (420) (%)	p-value
Age, years (n=1221)					.22
18–34	1155 (94.6)	336 (94.9)	425 (94.2)	394 (94.7)
35–44	58 (4.8)	13 (3.7)	24 (5.3)	21 (5.1)
≥ 45	8 (0.7)	5 (1.4)	2 (0.4)	1 (0.2)
Gender (n=1225)					.09
Female	826 (67.4)	240 (67.2)	307 (67.6)	279 (67.4)
Male	352 (28.7)	109 (30.5)	127 (28.0)	116 (28.0)
Othera	2 (0.2)	2 (0.6)	0 (0)	0 (0)
Declined	45 (3.7)	6 (1.7)	20 (4.4)	19 (4.6)
Race and Ethnicity (n=1187)					.80
Asian, non-Hispanic	221 (18.6)	62 (18.2)	76 (17.2)	83 (20.4)
Black, non-Hispanic	72 (6.1)	21 (6.2)	23 (5.2)	28 (6.9)
Hispanic or Latino	106 (8.9)	32 (9.4)	40 (9.1)	34 (8.4)
White, non-Hispanic	675 (56.9)	192 (56.5)	261 (59.2)	222 (54.7)
Other, non-Hispanicb	113 (9.5)	33 (9.7)	41 (9.3)	39 (9.6)
Years of Experience (n=1227)					.16
<1 year	637 (51.9)	173 (48.5)	248 (54.9)	216 (51.7)
1–5 years	586 (47.8)	184 (51.5)	201 (44.5)	201 (48.1)
6–15 years	4 (0.3)	0 (0)	3 (0.7)	1 (0.2)

Provider characteristics derived from self-reported survey data.

^aThe “Other” category for gender includes 2 respondents who did not further specify their gender.

^bThe “Other, non-Hispanic” category for race and ethnicity includes 48 respondents who selected more than one race: White and Other (n=3); Native Hawaiian/Pacific Islander and White (n=1); Black and White (n=3); Asian and Other (n=1); Asian and White (n=20); Asian, White and Other (n=1); Asian and Native Hawaiian/Pacific Islander (n=2); Asian, Black, and White (n=2); American Indian/Alaskan Native and Other (n=1); American Indian/Alaskan Native and White (n=9); American Indian/Alaskan Native and Black (n=2); American Indian/Alaskan Native and Asian (n=1); American Indian/Alaskan Native, Asian, Black, Native Hawaiian/Pacific Islander, and White (n=2). Additionally, 65 respondents provided the following write-in categories for race and ethnicity: Arab (n=3); Indian (n=1); ME (n=1); Middle Eastern (n=5); Mixed (n=1); Pakistani (n=1); South Asian (n=2); N/A (n=1); Unknown (n=1); If I answer this you will know who I am (n=1); and I don’t think it’s appropriate to ask about race (n=1). The remainder of respondents selecting “Other” did not further specify their race and ethnicity.

Table 1c.

Attending Characteristics (n=378)

Characteristic	Overall n (378) (%)	Pre-Implementation N (194) (%)	Early-Implementation N (74) (%)	Late-Implementation N (110) (%)	p-value
Age, years (n=375)					.01
18–34	89 (23.7)	50 (26.0)	13 (17.8)	26 (23.6)
35–44	184 (49.1)	100 (52.1)	28 (38.4)	56 (50.9)
45–74	102 (27.2)	42 (21.9)	32 (43.8)	28 (25.5)
Gender (n=374)					.24
Female	241 (64.4)	131 (68.6)	40 (54.8)	70 (63.6)
Male	122 (32.6)	56 (29.3)	29 (39.7)	37 (33.6)
Declined	11 (2.9)	4 (2.1)	4 (5.5)	3 (2.7)
Race and Ethnicity (n=370)					.86
Asian, non-Hispanic	74 (20.0)	34 (18.1)	18 (24.7)	22 (20.2)
Black, non-Hispanic	17 (4.6)	11 (5.9)	3 (4.1)	3 (2.8)
Hispanic or Latino	21 (5.7)	12 (6.4)	4 (5.5)	5 (4.6)
White, non-Hispanic	228 (61.6)	117 (62.2)	41 (56.2)	70 (64.2)
Other, non-Hispanica	30 (8.1)	14 (7.5)	7 (9.6)	9 (8.3)
Years of Experience (n=375)					.19
<1 year	7 (1.9)	5 (2.6)	0 (0)	2 (1.8)
1–5 years	86 (22.9)	48 (25.0)	10 (13.7)	28 (25.5)
6–15 years	189 (50.4)	95 (49.5)	38 (52.1)	56 (50.9)
≥ 16 years	93 (24.8)	44 (22.9)	25 (34.3)	24 (21.8)

Provider characteristics derived from self-reported survey data.

^aThe “Other, non-Hispanic” category for race and ethnicity includes 14 respondents who selected more than one race: White and Other (n=1); Black and White (n=1); Asian and White (n=7); and American Indian/Alaskan Native and White (n=5). Additionally, 4 respondents provided the following write-in categories: Brown (n=1); Indian (n=1); and Middle Eastern (n=2). The remainder of respondents selecting “Other” did not further specify their race and ethnicity.

Table 1d.

Nurse Characteristics (n=819)

Characteristic	Overall n (819) (%)	Pre-Implementation N (410 ) (%)	Early-Implementation N (174 ) (%)	Late-Implementation N (235 ) (%)	p-value
Age, years (n=800)					<.0001
18–34	413 (51.6)	193 (48.0)	119 (68.8)	101 (44.9)
35–44	181 (22.6)	96 (23.9)	24 (13.9)	61 (27.1)
45–74	206 (25.8)	113 (28.1)	30 (17.3)	63 (28.0)
Gender (n=803)					.26
Female	746 (92.9)	375 (93.5)	158 (91.3)	213 (93.0)
Male	38 (4.73)	16 (4.0)	13 (7.5)	9 (3.9)
Declined	19 (2.4)	10 (2.5)	2 (1.2)	7 (3.1)
Race and Ethnicity (n=794)					<.001
Asian, non-Hispanic	46 (5.8)	35 (8.8)	1 (0.6)	10 (4.4)
Black, non-Hispanic	28 (3.5)	13 (3.3)	3 (1.8)	12 (5.3)
Hispanic or Latino	36 (4.5)	21 (5.3)	3 (1.8)	12 (5.3)
White, non-Hispanic	623 (78.5)	296 (74.6)	151 (88.3)	176 (77.9)
Other, non-Hispanica	61 (7.6)	32 (8.0)	13 (7.6)	16 (7.1)
Years of Experience (n=807)					<.0001
<1 year	70 (8.7)	28 (6.9)	32 (18.5)	10 (4.4)
1–5 years	293 (36.3)	143 (35.3)	81 (46.8)	69 (30.1)
6–15 years	242 (30.0)	127 (31.4)	32 (18.5)	83 (36.2)
≥ 16 years	202 (25.0)	107 (26.4)	28 (16.2)	67 (29.3)

Provider characteristics derived from self-reported survey data.

^aThe “Other, non-Hispanic” category for race and ethnicity includes 40 respondents who selected more than one race: White and Other (n=1); Native Hawaiian/Pacific Islander and White (n=2); Asian and White (n=9); Asian and Native Hawaiian/Pacific Islander (n=3); American Indian/Alaskan Native and Other (n=1); American Indian/Alaskan Native and White (n=23); Asian, Native Hawaiian/Pacific Islander, and White (n=1). Additionally, 4 respondents provided the following write-in categories: East Indian (n=1); Korean (n=1); Not Relevant (n=1); Unspecified (n=1). The remainder of respondents selecting “Other” did not further specify their race and ethnicity.

What’s Known on This Subject:

Family-centered rounds improve patient safety and hospital experience. However, family-centered rounds buy-in, adherence, and implementation strategies vary. Mentored-implementation has been used to successfully coach sites in implementing hospital-based interventions but is typically physician-centric and does not include patients/families and nurses.

What This Study Adds:

We successfully implemented a family-centered rounds intervention across 21 hospitals using novel “Mentor-Trios” involving parent-nurse-physician triads coaching sites. Safety climate and rounds adherence improved post-implementation; resident-reported harms decreased in larger hospitals and those with greater intervention adherence and nurse engagement.

Abbreviations:

AAP

American Academy of Pediatrics

IPFCC

Institute for Patient and Family-Centered Care

I-PASS

Illness severity-Patient summary-Action items-Situational awareness-Synthesis by receiver

LEP

Limited English proficiency

PFC I-PASS

Patient and Family Centered I-PASS

PFCR

Patient/family-centered rounds

QI

quality improvement