An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Alisa Khan, MD, MPH, Matthew W. Ramotar, BA, Christopher P. Landrigan, MD, MPH; Patient and Family Centered I-PASS Study Group.
Author Information and AffiliationsInterventions to improve handoff communications between health care providers are associated with improved patient safety. The association between safety and interventions to improve communication between health care providers and patients and families is poorly understood, particularly in pediatrics.
To determine if medical errors and adverse events (AEs) (primary outcomes), family experience, family and nurse engagement, and communication processes between health care providers and pediatric patients and families improve following implementation of a structured communication intervention (Patient- and Family-Centered I-PASS).
We conducted a prospective multicenter study of a patient and family-centered communication intervention, “Patient- and Family-Centered I-PASS,” implemented in a staggered fashion at 7 North American pediatric hospitals from December 2014-January 2017. The intervention included a health literacy-informed, structured communication framework for family-centered rounds (multidisciplinary bedside rounds that include families and integrate their perspectives into clinical decision-making); written rounds summaries for families; a staff training and learning program; and strategies to support teamwork and implementation. The primary outcome had 2 principal components: medical errors (ie, failures in care processes) and AEs (ie, harms to patients due to medical care). Rates of errors and AEs were measured using an established systematic surveillance methodology including family safety reporting. Family experience was measured through predischarge surveys. Communication processes (eg, family engagement, rounds duration, teaching on rounds) were assessed by means of direct observation. We used Poisson regression and generalized estimating equations clustered by site to compare pre- vs postimplementation data. To inform the development of a reliable safety surveillance methodology incorporating patient/family input, we also conducted a subpopulation analysis of the first 4 study sites in the preintervention cohort to compare rates of errors/AEs: (1) gathered systematically with vs without family reporting, (2) reported by families vs providers, and (3) reported by families vs hospital incident reports.
Implementation of a structured family-centered communication intervention was associated with improvements in patient safety, family experience, and multiple aspects of hospital communication processes, without negatively impacting teaching or rounds duration. Moreover, families provided unique information about hospital safety. Including families in hospital safety and communication has the potential to improve hospital safety and quality of care.
In 1999, the Institute of Medicine estimated that between 44 000 and 98 000 patients are killed each year in the United States due to medical error.1 Subsequent studies, including by our group, have found that medical errors remain widespread2,3 and that rates of harm due to care are not decreasing.4 In 2010, the US Office of the Inspector General estimated that 180 000 patients die annually due to medical care.5 In a controversial 2016 analysis, Makary and Daniel extrapolated data from epidemiologic studies to conclude that 251 000 deaths may occur each year in the United States due to medical errors,6 a figure that would make medical errors the third leading cause of death.7 Regardless of the precise numbers of deaths, there is no doubt that harms due to medical care are extremely common in pediatric and adult inpatient settings.8,9
Miscommunication is consistently identified as a primary root cause10 of medical errors (failures in the delivery of medical care11) and adverse events (AEs, harms due to medical care12). According to Joint Commission data, communication errors are a leading cause of “sentinel events”10—serious AEs and harms due to medical care—in hospitals. Communication errors include both handoff (eg, change-of-shift) failures between clinicians and miscommunications between clinicians and patients.13
Miscommunications between physicians, nurses, and families occur with high frequency, and are a major source of medical errors.13,14 Several developments in the health care system have exacerbated communication problems: (1) rising acuity of hospitalized patients; (2) need for increasingly complex teams of clinicians to care for sicker patients; and introduction of resident work hour reductions in response to studies demonstrating the hazards of excessive work hours, which have increased the frequency of handoffs.15-21 Although harm due to medical errors remains common, many interventions have been successfully demonstrated to reduce rates of errors and/or preventable AEs.19,20,22-27 Both written and verbal sign-out processes are prone to errors, and amenable to intervention.28 Adoption of standardized tools and practices can reduce the risk of miscommunications.29 Structured communications and checklists have been shown to improve patient safety and reduce medical errors,29-31 and verbal mnemonics have been strongly advocated for by TeamSTEPPS and others, including patient advocacy groups,32 both to structure verbal handoffs and to decrease omissions of key information.33
We previously found that implementation of a standardized resident-physician handoff communication program organized around the verbal mnemonic “I-PASS” (I: illness severity; P: patient summary; A: action list; S: situation awareness/contingency planning; S: synthesis by receiver) was associated with a marked reduction in harmful errors (ie, preventable AEs).34
Notably, while such efforts to redesign handoffs, structure communications, institute verbal mnemonics, and improve written information exchange have been promising, they have typically not included patients and families or members of the interprofessional team. Additionally, handoffs are only 1 example of communication in hospitals. Rounds are another key point at which communication occurs.
Rounds are the time every morning when hospital staff (usually doctors, sometimes nurses) review patients' interval events and come up with the day's plan. Plans are often made without direct patient or family input. Patients and families may be directly excluded from rounds (eg, when rounds occur in a conference room), or indirectly excluded from rounds (eg, when rounds occur in the patient room in the presence of the family, but in a manner that does not truly engage the patient or family).
Family-centered rounds (FCRs) is a model for rounds in which rounds occur in the presence of the family. In FCRs, physicians, nurses, and other hospital staff visit with family members and patients each morning to review updates and formulate plans.35 FCRs have been proposed to serve a critical role in interprofessional and patient/family communication. They are further associated with improvements in teamwork and engagement and satisfaction of families and providers.36-38 At their core an enhanced form of handoff, FCRs have been advocated for by the Institute of Medicine (IOM), the American Academy of Pediatrics (AAP), the Accreditation Council for Graduate Medical Education (ACGME), and the Society of Pediatric Nurses (SPN),39,40 and the communication techniques they espouse have been endorsed by patient and family advocacy groups.32 However, there is relatively little research on the relationship between FCRs and safety, or the role of particular FCR processes on safety or patient and family experience.41 Moreover, FCRs as commonly practiced are highly variable36 and often occur in a manner that does not fully incorporate families or interprofessional team members.
For instance, communication during FCRs often occurs in unidirectional fashion: staff may tell patients and families the plan but do not always truly listen or partner with them to coproduce it. Staff often use medical jargon and do not always effectively ask families for their input or questions. Staff may not effectively ensure that the family understands the plan. FCRs may occur with the family present, but not truly involved.
In sum, medical errors are a leading cause of patient death and harm,6,42 and communication failures are in turn a leading cause of sentinel events, the most serious types of harmful errors.10 Interventions to improve poor handoffs of patient care between health care providers have been shown to improve patient safety.34,43-45 We previously found that implementing a standardized resident-physician handoff communication program organized around the I-PASS mnemonic was associated with a marked reduction in medical errors and AEs.34 Improvements in handoffs and communications have been identified as a priority in nationwide efforts to improve patient safety.46-48 Similarly, improved communications, focusing specifically on the importance of redesigning health care communications in a manner that better integrates patient and family voices, and standardizes the structure of family-clinician communications, are important.26,27 However, relatively little research has addressed the effect on patient safety of interventions to improve communications between providers and families.
In a single-center pilot study (the Nighttime Communication Study), we found that communication quality, shared understanding, and family experience improved after implementing a bundle of nighttime structured family-centered communication.49 Similarly, 2 recent single-center studies have suggested that such provider-patient communication interventions in hospitals might lead to improved patient safety.50,51
To more definitively address this possibility, building on our prior work, we conducted a prospective, multicenter intervention study to determine whether implementation of a structured communication program between health care providers and patients and families that focused on improving FCRs and the entire 24-hour cycle of communication, not just nighttime communication—Patient- and Family-Centered I-PASS—was associated with:
We hypothesized that the implementation of Patient- and Family-Centered I-PASS would lead to:
Patient- and Family-Centered I-PASS engaged a variety of stakeholders, including nursing and physician leaders, family advisors, and health literacy (HL) experts. We partnered with stakeholders throughout the entirety of the study, from study design and data instrument development to intervention implementation and dissemination. The Patient- and Family-Centered I-PASS 15-member Family Advisory Council (FAC), 14-member Nursing Advisory Council (NAC), and 6-member HL Subcommittee were integrated into all aspects of the program (Figure 1).
Patient- and Family-Centered I-PASS Organizational Chart.
Engagement of stakeholders commenced with the inception of the Nighttime Communication Study, a single-center family-centered initiative that served to pilot Patient- and Family-Centered I-PASS, but which focused on solely nighttime communication, not daytime communication or FCRs. Stakeholder input by means of observation and feedback contributed toward informing refinements and prompting research questions that would ultimately lead to what is Patient- and Family-Centered I-PASS. We identified stakeholders for the Patient- and Family-Centered I-PASS program from the Nighttime Communication Study, as well as via introductions facilitated by study sites and partner institutions. Following identification, we hosted a kickoff meeting in Boston, Massachusetts, to provide invited stakeholders with information on the general aims of the study and expected commitment of time. Health literacy experts were subsequently identified through the Institute of Medicine and National Academy of Medicine's Roundtables on Health Literacy.
We have since constructed an infrastructure to facilitate the integration of stakeholders at the central level and ensure each group had a voice in all study decisions. Stakeholders ultimately contributed toward informing and guiding the establishment and refinement of all process and outcome measures and data collection instruments, as well as toward all analyses and dissemination efforts. In sum, stakeholders from all levels of care were intimately intertwined with the conception, design, and execution of the Patient- and Family-Centered I-PASS study and interwoven throughout all facets of the project.
We formed a FAC that engaged patient and family advocates from across all study sites. The FAC was chaired by Dale Micalizzi and Helen Haskell, national patient advocates who lead patient safety improvement and engagement organizations. The FAC included parents with backgrounds in patient engagement and patient safety, many of whom had a wealth of health care experience with their own children. The FAC met monthly and reported back to the larger group during quarterly study group calls. The parents also worked with family advisory councils at their own children's hospitals to assist in implementation efforts. Beyond their role in the FAC, families were involved in all subcommittees, including the Coordinating Council (CC), Scientific Oversight Committee (SOC), the Education Executive Committee (EEC), and the Team Communication Subcommittee. The CC oversaw and coordinated all aspects of the Patient- and Family-Centered I-PASS Study and met on a weekly basis. The Patient- and Family-Centered I-PASS SOC assembled weekly to biweekly and was responsible for study design, research tool and protocol development, and data management and analysis. The EEC, which met weekly, designed and oversaw all aspects of the intervention, training, and coordinated the operation of educational subcommittees (including the Rounds Subcommittee, Simulation and Educational Strategies Support Team, Campaign Subcommittee, Dissemination Core, and Written Communication Tool Subcommittee). The Team Communication Subcommittee, which provided insight on interprofessional hospital communication processes, also met every other week as the intervention was being developed and refined. We compensated family advisors for their involvement through payment of honoraria.
We formed an NAC that engaged nursing advocates and stakeholders from across all study sites. The NAC was chaired by Dr Jennifer Baird and Ms Jayne Rogers, both experts in interprofessional care and nurse engagement. The group met monthly over the course of the study and was similarly involved throughout all aspects of it, including development and implementation of the intervention, interpretation of findings, and formulation of future plans for dissemination. Like families, nurses were involved in our CC, SOC, EEC, and a variety of subcommittees, including the Simulation and Educational Strategies Support Team, the QI and Implementation, the Team Communication, the Rounds, the Campaign, and the Written Communication Tool Subcommittees. The CC, SOC, and EEC convened weekly to biweekly throughout the study, as described above (see Families), while the various Patient- and Family-Centered I-PASS Subcommittees met on a biweekly basis during the active intervention development and refinement phases of the study. We compensated nurse advisors for their involvement through payment of honoraria.
The HL Subcommittee was chaired by Drs Benard Dreyer and Shonna Yin, leading experts in HL-informed communication strategies. This 6-member working group, included an interpreter, a health policy and systems design innovation expert, and clinicians and researchers with expertise in cultural competency and HL. The group convened every other week at the study's onset and ad hoc as it progressed. Members worked closely with family stakeholders, including on a specific curriculum to define best practice for FCRs with low-English-proficiency patients. Health literacy experts provided invaluable insight on the design of our training program and materials, as well as on written materials used by providers and researchers throughout the course of the intervention, including parent- and researcher-specific assessment tools for evaluating its various components. Moreover, the HL subcommittee was consulted to provide expertise on focused questions that arose. For instance, HL experts helped train residents on how best to perform synthesis with families on FCRs and partnered with us in dissemination efforts, outlined below.
We additionally partnered with physician stakeholders and advocacy organizations. Physician stakeholders included frontline physicians and physician leaders (n = 70). Physician leaders included physician administrators in various senior hospital leadership roles (eg, chief medical officer, chief of graduate medical education, chief quality officer, residency program director) across North America, the bulk of whom were affiliated with a Patient- and Family-Centered I-PASS study site or attached to the I-PASS Study Group via the original Resident Handoff Study. We additionally engaged other physician stakeholders, such as leaders of national associations and advocacy organizations, including representatives of the Pediatric Research in Inpatient Settings (PRIS) network, Institute for Healthcare Improvement (IHI), and the National Institute for Children's Health Quality (NICHQ). We ultimately convened an advisory board composed of physician stakeholders as described above.
Frontline physicians at each of the study sites were engaged in the design and implementation of the study in a site-specific and central way. Information from frontline physicians on the current state of FCRs aided the evolution of the intervention and curriculum components. Formal quality improvement (QI) observations conducted by frontline physicians while the study was underway served to guide subsequent refinements. Physician leaders of all tiers were integrated into the central direction of the study and contributed toward informing and refining the study's design, implementation, and dissemination. Many physician leaders were furthermore additionally engaged at the site-level and with working groups that supported the advancement of the program. By way of an illustration, Dr Benard Dreyer, president of the AAP at the time, provided leadership on the Patient- and Family-Centered I-PASS Coordinating Council and served as chair of the Patient- and Family-Centered I-PASS HL Subcommittee. He thus ensured principles of HL were interwoven throughout our research strategy and implementation of the program across study sites.
To sum, stakeholder engagement occurred at all levels and throughout each phase of our study. Families, nurses, and HL experts aided in the development of our intervention, curriculum, written communication materials, and assessment tools. Additionally, each group of stakeholders provided insight on and revisions to our various curricular materials, such as the “Rounds Report” and Patient- and Family-Centered Rounds Brochure, and participated in QI observations. We intend to continue to actively include family and nurse stakeholders, as well as HL experts, in our future efforts, as outlined in the subsequent subsections of this chapter.
Our group has striven to and will continue to strive to actively include families, nurses, and HL experts in all scholarship from the Patient- and Family-Centered I-PASS Study, including national presentations (see Table A1 in Dataforms Appendix). In addition, we have submitted proposals to present at the 2018 Pediatric Academic Societies (PAS), Pediatric Hospital Medicine (PHM), and IHI annual meetings, all of which are currently under review and include families, nurses, and HL experts as speakers.
Comments from families and nurses support the relevance and significance of the research question, and reinforced the importance of engaging families and nurses as true partners in the care team (Table 1).
Stakeholder Engagement in Scholarship.
Partnering with families, nurses, and HL experts enabled us to develop a more patient-centric study. As described above, each stakeholder group was represented in our SOC and CC and thus had a voice in the design of the study, as well as an influence on its various processes and outcomes. For instance, family members and HL experts were asked to provide input on our consent forms, to help us better define terms essential to our intervention, such as “rounds” and “safety,” using language that families would understand. Additionally, we asked the FAC for guidance on how to administer surveys to family members, and when we observed suboptimal levels of nurse engagement (see below), we sought the advice of NAC members about how to alter rounds processes to promote the role of nurses on FCRs.
The involvement of HL experts and family and nurse stakeholders has greatly benefited the rigor and quality of our study. Notably, our primary outcome, patient safety, was directly measured using a 2-step methodology that included family safety reporting. The family safety reporting methodology relied heavily on input from family partners on how to best administer family safety interviews (FSIs) and phrase questions regarding sensitive matters in a manner that would gather the best possible data. For instance, when piloting the FSI dataform with family members, 1 parent suggested alternate wording to what we had initially proposed to help increase the yield of family safety reports. She suggested we add the following question to our dataform, which we consequently added, “Did anything else happen or almost happen that upset you or could have harmed your child during this hospital stay?”
In addition, parent and nurse stakeholders joined SOC meetings focused on interpreting results of our ongoing analyses. On several occasions, there were outcomes that deviated from our own expectations. However, by involving these stakeholders, clinical nurses and parents with years of experience caring for hospitalized children, we were able to view the unexpected occurrences from their own unique perspectives, helping us develop a richer understanding of why the particular phenomena might be occurring.
We additionally sought out input from family members and HL experts on the administration of the Newest Vital Sign (NVS) dataform, a tool to assess parent subject HL. During data collection, research assistants (RAs) expressed concerns that the NVS was awkward to administer to parents, requesting tips to improve its administration to parent participants. Consequently, we sought input from the HL Subcommittee about how best to administer the NVS to families (see below).
On admission to the study unit, family members were approached by unit staff and briefed on the Patient- and Family-Centered I-PASS Intervention, namely through a Patient- and Family-Centered I-PASS Study Information Sheet and Family Centered Rounds Brochure. We asked the FAC and HL Subcommittee for guidance on both documents. Family members and HL experts helped refine the language of the study information sheet, as well as the framing of the intervention and various aspects of the study (such as the parent's role on FCRs as described in the brochure) in a manner that was easy for parents to understand. The brochure specifically made it clear to parents that our intervention was of direct relevance to their child. Furthermore, through input from both FAC and NAC members, the brochure was able to reinforce to parents the important role they played on rounds and in communications with all members of the care team. We believe that the involvement of families and nurses in these efforts ultimately helped to drive family participation in our intervention.
Partnering with family and nurse stakeholders, as well as HL experts, encouraged us to consider their unique points of view and expertise throughout all phases of the study. Moreover, each stakeholder group was so intimately involved in the study that Patient- and Family-Centered I-PASS was truly coproduced by parents, nurses, physicians, educators, and researchers. Our relationship with stakeholders fostered a sense of transparency throughout the research process and helped ensure that we actively addressed issues that arose with the study and intervention in real-time. For instance, we struggled with engaging nurses on FCRs, encouraging families to perform synthesis on FCRs (eg, “teach back” the plan), and ensuring intervention buy-in at certain implementation sites. Our course of action was to always reach out to our family members, nurses, and HL experts, to inform them of concerns and request their expert guidance on how to address them.
The questions we asked and sought to answer were relevant to families and nurses, as these stakeholder groups are essential, but often overlooked members of the hospital care team. We found that having family and nurse advisors participate in all study subcommittees ensured their voices were heard throughout each stage of the project. The FAC and NAC each fostered feelings of unity on the project and promoted a forum in which members of each council could share ideas, questions, concerns, and feedback on any aspect of the study. By empowering family and nurse advisors to provide honest feedback, and by promoting exchange of experience across various study sites with diverse geography, patient-mix, and institutional culture, we were able to learn from diverse and unique perspectives. Their input directly influenced the research questions we asked and approaches we decided on. Our findings are higher impact and should be easier to implement because they are the result of a study that was grounded in the input of these essential members of the care team.
Working with stakeholder partners to widely disseminate our intervention and curriculum to national audiences is well underway. In 2017, we introduced Patient- and Family-Centered I-PASS to national audiences by means of plenary presentations and workshop sessions at the PAS and PHM annual conferences, as well as the IHI National Forum on Quality Improvement in Health Care. We presented at the Beryl Institute Patient Experience and Patient and Family-Centered Care annual conferences in March of 2018 and are invited to speak at the 2018 PAS conference in May, the IHI-National Patient Safety Foundation (NPSF) in May 2018, and the PHM yearly meeting in July 2018. We are in the process of publishing our Dissemination Toolkit to MedEdPORTAL (see Implementation of Study Results), a peer-reviewed curricular repository published by the Association of American Medical Colleges. We are also working to publish our curricula in high profile academic journals and disseminate our results to patients and families through partner organizations (eg, through Consumers Advancing Patient Safety and the Justin's Hope blog) and in the lay press. We are partnering with families, nurses, and HL experts in each of these efforts.
Moreover, we are proposing to conduct a dissemination and implementation study of the Patient- and Family-Centered I-PASS program in community and academic pediatric hospitals across North America. We have received letters of support and widespread interest in disseminating Patient- and Family-Centered I-PASS by 36 sites across the US. We have convened a multidisciplinary Advisory Board consisting of national experts in safety, quality, patient advocacy, HL, and patient experience, many of whom are new partners enthusiastic to disseminate Patient- and Family-Centered I-PASS. The board includes senior leadership of key physician, nursing, patient, and family advocacy organizations, including IHI, NPSF, American Academy of Pediatrics, Children's Hospital Association, Hospital Research Education Trust, PRIS, the Beryl Institute, the Patients' View Institute, Consumers Advancing Patient Safety, Mothers Against Medical Error, the Leapfrog Group, Connecticut Center for Patient Safety, American Academy of Nursing, and the Society of Pediatric Nurses. Our collaboration with these organizations will serve to facilitate our ongoing partnerships with and dissemination to stakeholders from all levels of care.
Family, nurse, and HL expert stakeholders will remain involved in all future efforts, including ancillary and sub analyses, manuscripts, scholarly presentations, and grants.
This study was a highly meaningful educational experience for all of us. Our collaboration with family members, nurses, and HL experts enriched the experience of every member of our study group, and allowed providers specifically to better understand the needs and concerns of the study's populations (notable examples included in Table 2). We learned how valuable it is to meaningfully engage families, nurses, and HL experts as true partners in research and care, and we have every intention of continuing such a partnership as we adapt and spread the Patient- and Family-Centered I-PASS intervention.
Examples of How Engagement with Stakeholders Changed the Research.
Patient- and Family-Centered I-PASS is a bundle of communication interventions to improve the quality of information exchange between physicians, nurses, and families, and to better integrate families into all aspects of daily decision making in hospitals. This project tested the hypothesis that rates of medical errors/AEs (primary outcomes), hospital experience, communication, and shared understanding would improve following implementation of Patient- and Family-Centered I-PASS, as compared with current practice.
We conducted an intervention study on pediatric inpatient units at 1 Canadian and 6 US teaching hospitals from December 17, 2014-January 3, 2017. Each site was assigned to 1 of 3 staggered waves. We implemented the intervention over a 9-month period of iterative refinement and assessed processes and outcomes of care for 3 months both before and after the intervention (Figure 2). The preintervention period occurred between December 17, 2014, and December 30, 2015; postintervention was between December 15, 2015, and January 3, 2017. We selected these time periods to ensure that time of year and resident training levels were consistent between pre- and postintervention. This allowed us to control for confounding due to level of resident-physician experience and seasonality. Resident-physicians and nurses included were those present on the study units, which, particularly for residents, typically varied between pre- and postintervention periods. The Boston Children's Hospital IRB approved the study. IRBs at participating institutions additionally approved the study through reliance agreements with Boston Children's Hospital or through approval of the individual site's IRB.
Patient- and Family-Centered I-PASS Study Timeline.
Our study team included nearly 100 collaborating health services researchers, hospitalists, medical educators, parents, nurses, and HL experts across North America (Figure 3). We included parent, nurse, and HL expert stakeholders in every aspect of the study, from intervention development and refinement to study instrument creation (see the Participation of Patients and Other Stakeholders section).
The Patient- and Family-Centered I-PASS Study Group Composed of Health Services Researchers, Hospitalists, Medical Educators, Parents, Nurses, HL Experts, and Stakeholders of All Levels of Care Across the United States and Canada.
All patients, nurses, medical students, and resident-physicians on the study units were eligible to participate as subjects in the study. Parents/guardians or caregivers of children under 18 years of age who spoke a study language (English, Arabic, Chinese, Russian, and Spanish) were likewise eligible to participate. All sites had rotating medical students and residents. We determined study languages according to the most commonly spoken languages across the 7 study sites. We obtained waivers of consent from Institutional Review Boards at participating institutions to review patient charts. We obtained verbal consent from parents and medical students and written informed consent from resident-physicians and nurses to conduct interviews and surveys. We offered participants small incentives (eg, snacks) to participate.
FCRs served as the primary venue for improving communication with families of patients through a series of structural changes to enhance the 2-way transfer of information about the patient between family members and providers. We developed the Patient- and Family-Centered I-PASS intervention iteratively, through a consensus-building process informed by our prior study,49 literature review, and partnership with collaborating parent stakeholders, nurse stakeholders, physician stakeholders and HL experts. The intervention included: (1) a structured high-reliability communication framework for patients and families on FCRs that was organized around the I-PASS mnemonic and bolstered by principles of family engagement, HL (such as use of plain language by providers, eg, “fever” instead of “febrile”),52 bidirectional communication, and interprofessional and nurse engagement; (2) a “Rounds Report” (ie, written daily summary of FCRs) provided for families in I-PASS format and completed in real-time on a whiteboard or on paper (see Figure A2 in Appendix); (3) a learning and training program for families and professional team members with interactive learner-specific workshops reinforced by simulation and role play exercises and computer-based video modules; and (4) strategies to support teamwork and implementation of the intervention, including midshift nurse-physician huddles to confer on patient care issues and family concerns, structured observations of rounds with feedback to improve team performance, and a sustainability campaign (Figure 4).
The Patient- and Family-Centered I-PASS Intervention.
For sites that already had FCRs, I-PASS was introduced as a high-reliability framework that was complementary to but built upon existing rounding models. Certain elements of I-PASS were clear improvements to existing FCRs because they were nonexistent (eg, Illness severity, synthesis by receiver) or inconsistently included on FCRs (eg, situational awareness) preintervention. Certain elements (patient summary and action items) aligned nicely with existing FCR practices, but were reframed to emphasize HL principles and problem-based assessments. Moreover, all sites already used I-PASS for resident-resident handoffs, which also helped facilitate buy-in.
Primary outcomes were rates of medical errors and AEs. Secondary outcomes included hospital experience, communication processes, and shared understanding, as described below. Additionally, we compared rates of errors/AEs: (1) gathered systematically with vs without family reporting; (2) reported by families vs providers; and (3) reported by families vs hospital incident reports.
We collected data at 7 North American inpatient pediatric medical (ie, non-intensive care) units, ranging in size from 189 to 540 beds. Study sites, which included free standing children's hospitals (n = 3), pediatric hospitals within a larger system (n = 3), and a pediatric unit within an adult system (n = 1), determined their medical units to select for inclusion in the program. Study units varied in average daily patient census, ranging from 10 to 30 patients, and clinical services represented, which included general pediatrics (n = 7), specialty care (n = 3), and complex care (ie, children with medical complexity including multiple health concerns/needs, multiservice/specialty involvement, multiple medications, medical technology-dependence, or frequent emergency department/hospital visits; n = 1). Five of the study sites practiced FCRs before the intervention, though practices and levels of experience differed.
We assessed rates of medical errors (failures in care processes11) and AEs (harms to patients due to medical care,12 eg, a reaction to penicillin in a patient with a known allergy) per 1000 patient-days using an established systematic safety surveillance methodology that included family safety reporting (Figure 5).34,43,53-56
Patient- and Family-Centered I-PASS Safety Surveillance Methodology.
Notably, overall errors are the sum of harmful errors (also known as preventable AEs) and nonharmful errors (Figure 6). Overall AEs are the sum of preventable AEs (also known as harmful errors) and nonpreventable AEs.
Participant Flow by Type.
Every weekday, across all sites, research clinicians collected potential medical errors and AEs by reviewing hospital incident reports, postshift provider (eg, resident-physician and nurse) surveys, and medical records of all study unit patients. To incorporate input from families into the error surveillance process, we additionally collected potential medical errors and AEs using FSIs of parents/guardians or caregivers of eligible hospitalized patients before discharge (and every 7 days for longer-stay patients) (see the “Family Safety Interviews” section), and ultimately merged family-reported safety data with safety data collected from other sources. Trained physician-reviewer pairs blinded to pre- vs postintervention status independently categorized all research clinician- and FSI-collected events as exclusions, nonharmful errors, or AEs. Physician-reviewer pairs further categorized AEs as preventable or nonpreventable. AEs that were clearly caused by a failure in a care process were deemed preventable, while all other cases were deemed nonpreventable. Discordant categorizations by physician-reviewers were reconciled through consensus.
We collected family-reported safety data by administering FSIs to families. We developed the FSI using a modified Delphi method with input from experts in survey methodology, patient safety, and HL, as well as family advisors. The FSI was piloted and iteratively refined through cognitive interviews with family members at Boston Children's Hospital.
To educate families on information to report, the FSI provided descriptions, definitions, and examples of safety events related to miscommunications, delays in care, and complications of care, among other incident classifications. Using closed- and open-ended questions, families were asked to indicate whether their child's illness worsened or almost worsened because: (1) of medical care (ie, AE); (2) something was not done that should have been (ie, preventable AE due to error of omission); or (3) something was done that should not have been (ie, preventable AE due to error of commission). Parents were also asked to indicate whether a mistake occurred that did not result in harm to their child (ie, nonharmful error) and if anything else happened or almost happened that was upsetting or could have harmed their child (see Engagement).
The FSI was administered to parents of eligible patients every 7 days while hospitalized and before their child's discharge (see Figure A8 in Dataforms Appendix). Interviews typically lasted between 3 and 5 minutes (when no safety concerns) and 10 and 15 minutes (when safety concerns). Research clinicians (nurses or physicians) then classified family concerns as safety, non-safety-related quality (eg, unpleasant interaction with provider), other (eg, difficult intravenous placement), or exclusions (eg, off-unit safety concerns).
We collected potential safety events from providers by modifying a reporting instrument used in previous studies34,43 (see Figure A3 in Dataforms Appendix). Survey respondents were asked to describe errors/AEs and procedures, medications, fluids, or other therapies that were: unnecessary or questionably beneficial, delayed, involved in an error/AE, or ordered erroneously but intercepted before reaching the patient (near-misses).57 Providers were additionally requested to report resident-physician or nurse sign-out omissions or inaccuracies that led to problems with patient care. Every weekday morning during both the pre- and postintervention periods, research clinicians at all sites administered the survey (on paper or verbally) to postshift overnight resident-physicians. Surveys were also posted on the study units to allow voluntary, anonymous error reporting from unit staff not enrolled in the study.
We measured family experience before and after implementation of the Patient- and Family-Centered I-PASS intervention using a survey administered to 2 randomly selected parents per week per site, before their child's discharge (see Figure A4 in Dataforms Appendix). The survey was developed, cognitively tested, and piloted at Boston Children's Hospital, and then translated into Arabic, Chinese, Russian, and Spanish. The <15-minute survey, which had a Flesch-Kincaid58 reading grade level of 5.9, was verbally administered by RAs at each site and asked families to rate their experience, using predominantly 5-point Likert scales: during and after FCRs, with written communications, with resident-physicians and nurses, and during the overall hospitalization.
To measure hospital communication processes, RAs conducted 1-hour weekly in-person observations of FCRs at each site, concurrently completing a real-time structured direct observation assessment tool for each patient and audio-recording rounds. RAs assessed and recorded a median of 3 (interquartile range, 2-5) patients during each observation session. RAs conducted real-time assessments using a structured observation assessment tool developed by our study team to rate various aspects of quality of communication during rounds processes, as well as engagement of parents and nurses, using Likert scales (see Figure A5 in Dataforms Appendix). They also kept track of rounds duration. In addition, research clinicians at each site who were blinded to pre- vs postintervention status conducted post hoc analyses of a subset of FCR audio recordings (2 patients per week per site) using a structured assessment tool that measured rounding team adherence to evidence-based communication processes (eg, plain language was used effectively).
We measured shared understanding between parents, residents, and nurses before and after implementation of the intervention. An RA at each site verbally administered the survey, which was cognitively tested, piloted, and translated as described above (see Figure A6 and A7 in Dataforms Appendix). The survey asked parents, as well as physicians and nurses, to report their understanding of key elements of the medical plan. Percentage concordance was assessed through a comparison of responses and mean concordance ratings were compared pre- vs postintervention.
We obtained patient characteristics from hospital administrative data and collected provider and parent demographic information through surveys.
Because we were interested in the role of family safety reporting, we also subanalyzed data from the first 4 study sites (using preintervention data only; selected because they participated in the study first chronologically) to further evaluate the utility of family safety reporting. In this subanalysis, we sought to compare rates of errors and AEs: (1) gathered systematically with vs without family reporting; (2) reported by families vs providers; and (3) reported by families vs hospital incident reports.
To compare error and AE rates (events per 1000 patient-days) pre- vs postintervention, we used Poisson regression estimated via generalized estimating equations (GEEs) to account for clustering of patients within site. For experience and rounds observation data, we compared percentage of top-box ratings (choosing the top-most response on a scale; eg, “excellent” or 5 out of 5) pre vs post using a GEE chi-square test for binary outcomes, clustered by site. For shared understanding, we used t tests to compare mean concordance ratings before and after implementation.
We analyzed all collected variables that could be potential confounders, including patient, parent, nurse, and physician characteristics, in the pre- vs postintervention cohorts to ensure none of these variables differed significantly and required adjustment in analyses. To account for missing data, we used multiple imputation appropriate for missing data in clustered studies (calculating P values for variables with more than 5% missing data using multiple imputations; 300 imputations used).59-61 Distributions of possible confounders were comparable in the pre and post periods. For simplicity, we present unadjusted results, accounting for clustering by site (see the Results section).
For our 2 primary outcomes (overall errors and overall AEs), we applied a Bonferroni correction and considered P < .025 significant. For our 3 secondary analyses (family experience, communication processes, and shared understanding), we considered P < .05 statistically significant. We used REDCap62 to collect and manage study data and SAS 9.4 (SAS Institute) to conduct analyses.
We made slight modifications to the original protocol, as outlined below (Table 3 and Study Protocol Appendix). Regarding our statistical analyses, we analyzed our data as per our analysis plan except for a few minor modifications. For errors and AEs, rather than using existing adult family safety reporting instruments, we adapted existing adult and pediatric family safety reporting instruments in order to develop a more robust inpatient pediatric family safety reporting methodology. We did so in conjunction with parent partners and HL experts and created a final methodology that was based on piloting and cognitive interviewing our instrument. We recently published this methodology and our findings in JAMA Pediatrics.
Changes to the Original Study Protocol for Patient and Family Centered I-PASS.
In addition, there was no way to cluster error and AE rates by resident as originally planned, because we were only able to attribute a fraction of all captured errors and AEs to a particular resident or other provider. Rather, we were examining system-wide rates of incidents. Therefore, it was not possible to cluster by resident. However, for any future analyses, where we examine only the subset of errors attributable to residents, we will cluster by resident.
To evaluate communication processes on FCRs, although our original plan was to do 25 FCR observations preintervention and 25 FCR observations postintervention, we ended up doing considerably more direct observations (a total of 653 across 7 sites). However, due to pragmatic limitations, we only analyzed audio recordings for a subset of these FCR observations (n = 164).
Moreover, we had originally planned to measure process and content changes in communication via direct observation and development of a validated Verbal Handoff Score (VHS) to assess FCRs (ie, a global numerical rating of FCRs). Although developing a VHS for FCRs is within the scope of our future work, for the time being, we felt it more salient to compare the individual elements we thought best represented the key components of our intervention (eg, statement of illness severity at start of rounds, synthesis by receiver, use of plain languages) (see Figure 9). Developing a global VHS score for FCRs is a future direction.
Sources of Medical Errors and AEs, With Incidents Validated Through 2-Step Methodology.
A total of 2148 parents (83.6% of those confirmed as eligible; generally, 1 parent per patient was enrolled), 435 nurses (97.3%), 203 medical students (99.5%), and 586 residents (95.6%) consented to participate in the study. No patients were lost to follow-up and missingness (which ranged from 2.0%-9.7%, except for patient race and annual household income) was similar in pre- and postintervention cohorts. In total, we collected 2034 FSIs (95.9% of patients approached) and 1224 family experience surveys (97.8% response rate), and we conducted 654 direct observations of FCRs. Participant flow is depicted below by participant type (Figure 7) and intervention cohort (Table 4).
Family Experience Before and After Implementation of Patient and Family Centered I-PASS.
Patient Admissions per Hospital by Intervention Cohort.
We reviewed a total of 3106 patient admissions (1574 preintervention, 1532 postintervention) for medical errors and AEs, representing 13,171 patient-days (6326 pre intervention, 6845 postintervention). Across both cohorts, mean patient age was 7.1 years; 50.5% were female; 50.4% were non-White; and 8.0% had ≥2 complex chronic conditions (CCC). Patient characteristics were similar in the pre- and postintervention cohorts (Table 5).
Patient, Parent, and Provider Characteristics Before and After Implementation of Patient and Family Centered I-PASS.
A total of 2148 parents (83.6% of those approached), 435 nurses (97.3%), 203 medical students (99.5%), and 586 residents (95.6%) consented to participate in the study. In total, we conducted 654 direct observations of FCRs processes and collected 1224 family experience surveys (97.8% response rate) and 2034 FSIs (95.9%). Across both cohorts, the mean ages were 36.2 years for parents, 34.8 years for nurses, and 28.4 years for physicians; 80.8% of parents, 90.5% of nurses, and 68.2% of residents were female; and 53.1% of parents, 90.5% of nurses, and 68.2% of residents were non-White. Most parents attended college and roughly half of families reported annual household incomes of <$50 000.
Kappa statistics of the physician reviewer categorization of all suspected events as exclusions, nonharmful errors, or AEs (preconsensus agreement = 68.9%; κ = 0.53; 95% CI, 0.48-0.58) and subsequent categorization of AEs as preventable or nonpreventable (preconsensus agreement = 83.5%; κ = 0.64; 95% CI, 0.54-0.74) indicated reliability. Examples of validated medical errors and AEs are provided in Table 6.
Examples of Validated Medical Errors and AEs.
From the preintervention period to the postintervention period, we found that implementation of Patient- and Family-Centered I-PASS was associated with a 45.6% reduction in the rate of overall AEs (per 1000 patient-days), from 34.0 (95% CI, 26.4-43.7) to 18.5 (13.3-25.6) (P = .002). The intervention was also associated with a 37.7% decrease in preventable AEs (also known as harmful errors) from 20.7 (95% CI, 15.3-28.1) to 12.9 (95% CI, 8.9-18.6) (P = .01). Nonpreventable AEs also decreased from 12.6 (95% CI, 8.9-17.9) to 5.2 (95% CI, 3.1-8.8) (P = .003).
The overall medical error rate (ie, the sum of harmful errors and nonharmful errors) did not change between the pre- and postintervention periods (41.2 [95% CI, 31.2-54.5] vs 35.8 [95% CI, 26.9-47.7], P = .21) (Table 7). This result was likely driven by the fact that nonharmful errors (20.0 [95% CI, 13.2-30.2] pre- vs 22.0 [95% CI, 15.1-32.1] post, P = .50) did not change pre- vs postintervention, as harmful errors (also known as preventable AEs, rates given above) significantly decreased postintervention. Notably, given the utility of family safety reporting, all of these numbers include family safety reports.
Rates of Medical Errors and AEs Before and After Implementation of Patient- and Family-Centered I-PASS (Includes Family Safety Reporting).
Top-box ratings for 6 of 25 components of family-reported experience improved significantly between the pre- and postintervention periods. Specifically, there were improvements in the percentage of pa rents reporting top-box (selecting the topmost Likert scale option, eg, “excellent”) scores for understanding of what was said on rounds (53.9% pre intervention vs 62.8% postintervention, P = .03), receipt of written updates (15.6% vs 33.7%, P < .001), and understanding of written updates (46.5% vs 57.9%, P = .04) (Figure 8). Several elements of family experience linked to communications with nurses also improved significantly following implementation of the intervention, such as shared understanding of the care plan (55.3% vs 65.4%, P = .02), and whether nurses addressed concerns (61.2% vs 70.2%, P = .02) and made family members feel a part of the health care team (63.2% vs 70.7%, P = .04). No elements of family experience worsened significantly between pre- and postintervention cohorts.
Rounds Process Data Before and After Implementation of Patient- and Family-Centered I-PASS.
Real-time structured direct observations of rounds indicated that FCRs (as opposed to rounding in a non-FCR-manner) occurred more frequently postintervention (72.2% of 305 patients observed on rounds vs 82.8% of 347 patients observed on rounds, P = .02), as did the provision of written summaries to families (17.8% vs 53.6%, P < .001). We additionally found that family engagement (55.6% vs 66.7%, P = .04) and nurse engagement (20.4% vs 35.5%, P = .03) on FCRs improved significantly following implementation of the intervention (Figure 9A). Frequency of teaching on rounds (73.4% vs 72.4%, P = .78) and average rounds duration (8.5 vs 10.2 minutes per patient, P = .13) did not significantly change between the pre- and postintervention periods.
Post hoc analyses of FCR audio recordings indicated that frequency of adherence to key communication behaviors improved postintervention, with increased levels of parents expressing concerns at the start of rounds (18.2% vs 37.7%, P = .03) and completing synthesis (4.7% vs 26.5%, P = .02) on rounds (Figure 9B). Family reporting of child's illness severity (eg, better or worse) on rounds trended upwards but did not increase significantly (28.8% vs 43.5%, P = .08). Effective use of plain language on rounds did not change significantly (28.8% vs 34.7%, P = .36).
Initial analyses indicate that mean concordance ratings for shared understanding between parent, nurse, and physician did not significantly change from the preintervention to postintervention period. Mean shared understanding (on a scale of 1=5, with 1=not at all agree, 5=completely agree) between parents and nurses about the overall medical plan for hospitalization went from 3.45 (SD 1.20) preintervention to 3.37 (1.29) postintervention (P = .60). Mean shared understanding between parents and physicians about the overall medical plan for hospitalization went from 3.28 (1.27) preintervention to 3.51 (1.24) postintervention (P = .14). Mean shared understanding between nurses and physicians about the overall medical plan for hospitalization went from 3.52 (1.21) preintervention to 3.70 (1.20) postintervention (P = .23).
In the subpopulation analysis of the first 4 study sites (using preintervention data only),57 a total of 746 parents (95.4% of those approached), 146 nurses (98.6%), and 207 resident-physicians (95.4%) consented to participate, and 717 family members completed 763 FSIs, while 77 residents completed 284 postshift provider event reporting surveys. The data from all residents rotating through these first 4 sites in the preintervention period were included. Of the family members that provided consent, median (interquartile range) age was 32.5 (26-40) years; 380 (53.0%) were non-White; roughly half had attended college (53.0%); and the majority were female (78.9%) and English-speaking (84.1%) (Supplementary Table 1).
This subanalysis of FSIs from preintervention data only from the first 4 study sites revealed a total of 255 family -reported incidents. These incidents were reported by 185 parents (25.8%). Among these 255 incidents, 132 (51.8%) were classified as safety concerns, 102 (40.0%) as non-safety-related quality concerns, and 21 (8.2%) as other concerns. Of those classified as safety concerns, 97 occurred on study units, while 35 occurred outside of study units and were consequently excluded from further analyses. Following 2-step review, 50 of the family-reported safety concerns were confirmed as medical errors (17 nonharmful medical errors) and/or (22 preventable AEs) AEs (11 nonpreventable AEs). Family reports included 13 otherwise unidentified medical errors and 8 otherwise unidentified AEs, including 7 preventable AEs. Medical errors and AEs uniquely reported by parents included delay in diagnosis and multiple needle sticks, inadequate suctioning, and side effects from medication, accordingly.
These family-reported incidents subsequently underwent the same 2-step validation as incidents detected by any source.
Upon 2-step validation of all errors and AEs collected by any source (family-reported or otherwise), we detected a total of 179 validated medical errors and 113 validated AEs overall (using preintervention data only from the first 4 study sites). Parents reported 21.8% (n = 39) of these 179 validated medical errors (10.6% uniquely and 11.2% in combination with another source) and 29.2% (n = 33) of these 113 validated AEs (7.1% uniquely and 22.1% in combination with another source) (Figure 10).
Impact of Family Safety Reporting.
In this subanalysis of preintervention only data from the first 4 study sites, we found that errors (per 1000 patient-days) were 15.5% higher with family reporting than without (45.9 vs 39.7, P < .001). AEs were 9.8% higher with family reporting than without (28.7 vs 26.1, P = .003) (Figure 11).
In this subanalysis of preintervention only data from the first 4 study sites, we found that family-reported rates of medical errors and AE were equivalent to provider-reported rates. Families reported 10.0 medical errors and 8.5 AEs per 1000 patient-days, while providers similarly reported 12.8 (P = .25) and 6.2 (P = .20) errors and AEs, respectively (Figure 11).
Among sites with available hospital incident report data (3 of the 4 subanalysis sites), hospital incident reports comprised 5% (n = 5), while family reports comprised 23% (n = 25) of errors detected through all sources (n = 111). Hospital incident reports comprised 10% (n = 7), while family reports comprised 28% (n = 20) of AEs detected through all sources (n = 71).
Family-reported error rates (per 1000 patient-days) across these sites were 5.0-fold (8.7 vs 1.7, P < .001) higher than hospital incident report rates, and family-reported AE rates were 2.9-fold (6.9 vs 2.4, P = .02) higher than hospital incident report rates (Figure 11). Examples of errors and AEs by reporting source are listed in Table 8.
Examples of Errors and Adverse Events by Reporting Source.
Bivariate predictors of family-reported errors included parent/caregiver proficiency in written and spoken English and higher parent education, as well as patient age and ≥1 CCC. Multivariate predictors of family-reported errors included older parent age and higher parent education, along with younger patient age and ≥1 CCC.
Bivariate predictors of family-reported AEs include higher parent/caregiver education and greater family presence during hospitalization, as well as ≥1 CCC. Multivariate predictors of family-reported AEs included higher parent education and ≥1 CCC.
In a 7-center study, we found that implementation of a structured patient and family-centered communication intervention emphasizing family engagement and HL—”Patient- and Family-Centered I-PASS”—was associated with a 38% reduction in harmful errors and 46% decrease in overall AEs. The intervention included: (1) a structured high-reliability communication framework to better engage patients and families on FCRs, organized around the I-PASS mnemonic and bolstered by principles of HL, bidirectional communication, and interprofessional and nurse engagement; (2) a written daily summary of FCRs for families in I-PASS format (the “Rounds Report”); (3) a learning and training program for families and professional team members with interactive learner-specific workshops; and (4) strategies to support teamwork and implementation of the intervention (Figure 4).
Family engagement on rounds improved significantly following implementation of the intervention, as did multiple aspects of family experience with communication. Moreover, improvements in safety, experience, and quality of communication on rounds occurred without a significant increase in rounds duration, or a decrease in the amount of teaching that occurred on rounds. We additionally found that by integrating family safety reporting into the 2-step (nurse review followed by 2-physician review, Figure 5) systematic safety surveillance typically used in research, medical errors and AEs that we would have previously missed were reported by families and included in our analyses. Furthermore, parents identified comparable rates of medical errors and AEs as health care providers and 5 times more errors and 3 times more AEs than the hospital incident reports used operationally in most hospitals. 57
This study addresses several limitations of our single-center pilot Nighttime Communication Study, in which we found improvements in parent and provider experience as well as shared understanding following implementation of a bundle of family-centered, inter professional nighttime communication interventions.49 We incorporated refinements to more effectively engage family members and nurses and better integrate the intervention into physician and nurse workflow. We focused on communication during the entire 24-hour cycle (particularly during morning rounds), not just nighttime communication. Moreover, we conducted a subpopulation analysis and found that family safety reporting increased overall error detection by 16% and AE detection by 10%, compared with what was typically considered the highest yield methodology in safety surveillance research, which typically excludes patients and families. Consequently, our study directly measured safety outcomes pre- and postintervention using a novel expanded active surveillance methodology that incorporated input from families.63
Prior studies have indicated that improved teamwork is associated with improved patient outcomes.64,65 However, there is relatively limited empirical evidence on the relationship between family-centric care and patient safety.50,51 A recent single-site adult intensive care unit (ICU) study showed that implementation of a multifaceted intervention emphasizing patient and caregiver engagement, communication, and information technology was associated with decreases in harmful errors and improvements in patient and provider experience.50 Our study expands on this work by indicating applicability beyond an adult critical-care setting and suggests the ability to generalize a family-centered communication program across multiple sites. Moreover, concepts of structured communication and adherence to HL principles that serve as the core of our intervention are likely to be applicable across settings. Our study provides evidence of generalizability to pediatric inpatient units in academic hospitals. Further study is warranted to confirm generalizability in other types of clinical units and nonpediatric hospitals.
In our subpopulation analysis of preintervention only data from the first 4 study sites, we found that family safety reporting was a valuable source of safety data. Although our study focused on pediatric patients, family and caregiver reporting may be broadly applicable for other patient populations where patients may be unable to fully advocate or speak for themselves (eg, ICU patients, geriatric patients). Additional research is required to confirm generalizability and feasibility in these and other settings.
Physician,66-68 nursing,69 and family70,71 advocacy organizations have called for improvements in the family-centeredness and quality of care. For instance, FCRs have been advocated by the IOM, the AAP, the ACGME, and the SPN.39,40 The communication techniques FCRs espouse have been endorsed by patient and family advocacy groups.32 However, there is relatively little research on FCRs and safety, or on FCR processes and safety and patient/family experience.41 Our intervention to improve family-centered communications and patient safety—”Patient- and Family-Centered I-PASS”—was directly responsive to these calls and fills these gaps in the literature. In this study, implementation of a bundle of communication interventions targeting FCRs and interprofessional hospital team communications across 7 sites was successful in that it was associated with a 38% reduction in harmful errors and 46% decrease in overall AEs. Additionally, family engagement on rounds improved significantly, as did multiple aspects of family experience with communication.
To facilitate future implementation of this successful program, we are finalizing a Patient- and Family-Centered I-PASS Mentored Implementation Toolkit that we will publish in MedEdPORTAL and hope to distribute, with additional support, through the Mentored Implementation methodology and PRIS network. The Toolkit will provide a summary of our findings and will describe the fundamental building blocks of a successful program. It will include details about our training, FCRs, and other communication processes, process measures, and details of how the program was implemented across sites using the Mentored Implementation methodology. Lessons learned in our study both from sites where adoption was optimal and suboptimal will be included, as will principles that other programs should consider to maximize the probability of successful adoption of the program. We believe that actively disseminating results will lay the groundwork for ensuring that our program will be successfully adopted widely, and that implementation efforts will inform improvement efforts. Additionally, our study group has an extensive track record of partnering with PRIS and Society of Hospital Medicine (SHM), with whom we hope to work to carry out future dissemination and implementation efforts.
We will continue to disseminate our findings directly to patients and family through multiple avenues, including the lay press, social media, blogs, patient advocacy networks, family advisory councils of our parent partners and their networks, and national organizations (including the Beryl Institute and the Patient- and Family-Centered Care Conference). We will also continue to actively engage our parent partners in these dissemination efforts.
Our group is currently completing an I-PASS handoff dissemination and implementation project in 32 community and academic hospitals based on results from our prior I-PASS Resident Handoff Study.34
Preliminary analyses of the first 16 sites demonstrate significant increases in adoption of I-PASS processes to a level (70%) that exceeded what we achieved in the prior multicenter I-PASS Resident Handoff study, as well as a significant reduction in mean rate of staff-reported harms (15% vs 4% of patients per month experiencing a major harm due to a handoff failure).72 These findings demonstrate our ability to disseminate and implement a rigorous communication improvement program with important patient safety implications in diverse settings, and collaborate successfully with PRIS and the SHM Mentored Implementation Program.
We hope to embark on a similar process in the present Patient- and Family-Centered I-PASS Study to likewise disseminate and implement our successful Patient- and Family-Centered I-PASS intervention across diverse pediatric centers. Based on the promising nature of the Patient- and Family-Centered I-PASS intervention, as well as our successful dissemination and implementation of our prior Resident Handoff Study using the SHM Mentored Implementation Program, broad dissemination of the Patient- and Family-Centered I-PASS intervention has the potential to directly improve patient safety and family experience across a wide range of clinical settings. Moreover, improving uptake of suboptimal elements of the intervention (eg, nurse engagement and use of plain language) will enable us to better improve the quality of information exchange between physicians, nurses, and families and consequently further improve patient safety, as well as hospital experience and communication. Additionally, as concepts of interprofessional communication, teamwork, and adherence to HL principles are thought to be universal, our intervention may be adaptable to a wide range of hospital settings, such as adult, geriatric, and critical care settings.
Our initial approach to data collection in this study was intentionally rigorous in order to provide a sound evidence base for the validity of our findings. However, we have had success in the past when transitioning from initial research to subsequent dissemination projects by using a systematic and rigorous, but far less labor-intensive methodology to capture safety events that is more widely reproducible and generalizable across hospitals. For instance, in the Mentored Implementation 32-hospital dissemination project of the original 10-center I-PASS Resident Handoffs Study, we used a far-less labor-intensive surveillance methodology that demonstrated a >40% reduction in AEs73 (a rate comparable to the >30% reduction in AEs we detected using more intensive safety surveillance methods in the original 10-center study).34 The results of Patient- and Family-Centered I-PASS can similarly be generalized and reproduced as we disseminate it to additional sites using less labor-intensive but still rigorous safety measurement methods.
Many individuals were involved in the initial process of thoughtfully developing and refining our initial Patient- and Family-Centered I-PASS intervention. However, at each given site, implementing the intervention required the buy-in and support of a much smaller group of ∼6 to 8 key individuals. This group generally included, for instance, the site lead, program director, parent champion, nurse champion, hospitalist director, and faculty champion. Thus, future dissemination of our intervention to a wide range of hospitals and settings should be feasible and practical, if each site is able to recruit such a group of ∼6 to 8 key individuals.
Our subpopulation analysis of preintervention only data from the first 4 study sites suggested the importance of including family safety reporting to accurately capture hospital safety. We found that errors and AE rates were significantly higher with family safety reporting than without. Families reported error incidents at rates similar to providers and ∼5 times (medical errors) and ∼3 times (AEs) higher than hospital incident reporting rates. More educated parents and parents of children with CCCs had higher rates of family-reported errors. Although family error reporting is promising, particularly for medically complex children, the best method for operationalizing family safety reporting in hospitals for QI and research safety surveillance remains to be seen.
Our study has several limitations. First, the study design limits our ability to definitively establish a causal link between the intervention and error rates or other outcomes. Rigorous studies of quality and safety improvement interventions are often not amenable to classic randomized trial designs.30,34 Because we believed our intervention was highly unlikely to harm patients, our lack of equipoise precluded patient-level randomization or a cross-over design. We opted against a step-wedge design for pragmatic reasons, as a step-wedge design would have required us to collect data continuously at all 7 sites, rather than in a staggered fashion for 3 months preintervention and 3 months postintervention at each site (see timeline in Figure 2). This magnitude of data collection was beyond our resource constraints. Additionally, there was logistical difficulty of randomizing the timing of introducing a complex intervention involving physicians, nurses, and families and ensuring that no simultaneous cointerventions were occurring that might confound results. While our pre-post intervention design is at risk of confounding due to secular trends, prior research2,4,74 strongly suggests that a change in AEs of the magnitude we observed is unlikely in such a short period due to secular trends alone. Additionally, given we observed simultaneous improvements in patient safety outcomes, family experience, and rounds communication processes as hypothesized, and because findings were derived from data gathered across 7 sites (lowering the likelihood that changes were due to unmeasured site-specific confounders), we believe it is likely the improvements seen were related to the implementation of the intervention.
Second, we cannot be sure that that our findings would be reproducible in other types of clinical units or nonpediatric hospitals. However, as above, concepts of structured communication and adherence to HL principles that served as the core of our intervention are likely to be applicable to other settings. Furthermore, although unit selection in our study might limit generalizability, systemic interventions such as ours must consider and account for the nuances of an intervention setting, given that proper unit selection and preparation is an essential part of the intervention.
Third, medical error/AE classification is a complex and imperfect process that could have resulted in mistakes in classifying AEs and medical errors.34,43 However, the reliability (ie, κ) of our ratings was similar to or better than prior studies.34,43 Additionally, our understanding of the preventability of AEs is constantly changing and interpretation of errors and AEs is not always crystal clear. Moreover, it is unlikely that primary reviewer bias could explain the significant differences in rates of AEs observed given that final error and AE classifications were ultimately made by secondary reviewers blinded to pre- vs postcohort status only after postintervention data collection was complete. Additionally, if our findings were primarily explained by primary reviewer bias, and our intervention was in reality ineffective, we would expect fewer “soft-call” nonharmful errors in the postintervention period, as biased primary reviewers may be less likely to include events that they were uncertain about in the postintervention period. Instead, we found no change in nonharmful errors in both periods. It is likely in fact that this lack of change in nonharmful errors pre- vs postintervention explained why overall medical error rates were unchanged (since the subset of errors that were harmful significantly decreased postintervention).
Finally, although our instruments for assessing safety have been validated in other patient populations and settings, those for measuring experience and communication were novel instruments that have not been validated outside of this study, but which generally should good reliability statistics, such as Cronbach's α.
Our finding that nonpreventable AEs decreased was unexpected. Physician-reviewers were instructed to classify AEs as preventable only when incidents were clearly caused by a failure in a care process. Therefore, it is possible that we may have been overly conservative in judging preventability of patient harm. However, it is generally accepted within the patient safety field that our understanding of the preventability of AEs is constantly changing. Hospital-acquired central line infections, for instance, were once generally considered to be nonpreventable because there is typically no observable error (eg, break in sterile technique) that results in infection. We now know, however, that interventions can prevent nearly 80% of central line infections.75,76 In a similar way, it is conceivable that by improving interprofessional communications or better engaging families and nurses as essential members of the care team, for instance, our intervention prevented or avoided some AEs from occurring through mechanisms not yet apparent to us. Additional research is required to further explore this possibility.
Other areas of future study including examining the impact of this intervention on children with medical complexity, further evaluating and engaging adolescents in FCRs, and return on investment analyses of our program. In addition, further work is warranted to engage low English proficiency families in FCRs and examine the resulting impacts on safety and experience. Families can also be more involved in QI observations to improve and provide feedback to teams on FCR performance. Families can also participate in future efforts to collect data as this will further enhance stakeholder engagement and may make respondents more likely to share impressions openly with other families.
Regarding our subpopulation analysis of family-reported errors, further research is needed to determine the best method for operationalizing family safety-reporting in hospitals, and to assess its feasibility, success, and safety implications.
In sum, we found that implementation of Patient- and Family-Centered I-PASS—a multifaceted, structured patient- and family-centered communication intervention designed to enhance patient and family engagement, interprofessional communication, and adherence to HL principles—was associated with a marked reduction in preventable and nonpreventable AEs as well as improvements in family experience and hospital communication processes. These improvements occurred without a significant increase in the duration of rounds or a reduction in the amount of teaching occurring during rounds.
We additionally developed and studied a novel family safety reporting methodology that suggests families are useful partners in hospital safety reporting. We performed a subpopulation analysis of preintervention only data from the first 4 study sites. We found in this subanalysis that families reported events not otherwise detected or documented, and that integrating family safety reporting allowed us to better detect medical errors and AEs than if we were to use either the voluntary hospital incident reporting systems currently employed by most hospitals or the systematic safety surveillance used in research. We therefore integrated family safety reporting into all subsequent error and AE analyses.
Our study supports the inclusion of family reporting in safety surveillance and provides important evidence that improved communication between health care providers and patients and families could be an important means to reducing failures in care processes and incidents of patient harm.
We thank the members of the Patient- and Family-Centered I-PASS FAC, NAC, and HL Subcommittee. We also thank all other members of our Study Group, as well as all patients, families, nurses, and residents who participated in the study.
“A wise family doctor once told me something that has stuck with me through the years. It went something like this: ‘Hospitals are not set up for patients. They are set up for doctors.’ As I struggled through years of care with my children, I saw firsthand how true this statement really was. We have to make it easier for families to be a true part of their children's care. When patients and families are true members of the medical team, care is more informed, more targeted, and more safe for everyone.”
-Sharon Cray, a mother of 2 children with cystic fibrosis and partner in our study
We extend our sincerest thanks to the patient and family members of the Patient- and Family-Centered I-PASS Study Group, who partnered with us every step along the way, from development and implementation to analysis and dissemination, and without whom, this study would not have been possible.
Research reported in this report was [partially] funded through a Patient-Centered Outcomes Research Institute® (PCORI®) Award (#CDR-1306-03556) Further information available at: https://www.pcori.org/research-results/2013/does-patient-and-family-centered-hospital-communications-program-reduce
Dataforms and Additional Tables/Figures (PDF, 666K)
Stakeholder Engagement in Scholarship (PDF, 171K)
Rounds Report (PDF, 117K)
Patient and Family-Centered I-PASS Study Event Reporting Form for Staff (PDF, 23K)
Patient and Family-Centered I-PASS Study Family Experience Survey (PDF, 174K)
Patient and Family-Centered I-PASS Study Provider Concordance Interview (PDF, 86K)
Patient and Family-Centered I-PASS Study Family Concordance Interview (PDF, 131K)
Patient and Family-Centered I-PASS Study Family Error-Reporting Interview Questions (PDF, 204K)
Khan A, Ramotar MW, Landrigan CP. (2019). Does a Patient- and Family-Centered Hospital Communications Program Reduce Medical Errors? Patient-Centered Outcomes Research Institute (PCORI). https://doi.org/10.25302/8.2019.CDR.130603556
The [views, statements, opinions] presented in this report are solely the responsibility of the author(s) and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute® (PCORI®), its Board of Governors or Methodology Committee.
This book is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License which permits noncommercial use and distribution provided the original author(s) and source are credited. (See https://creativecommons.org/licenses/by-nc-nd/4.0/
Your browsing activity is empty.
Activity recording is turned off.
See more...
