Examine family safety-reporting after implementing a parent–nurse–physician–leader coproduced, health literacy-informed, family safety-reporting intervention for hospitalized families of children with medical complexity.
We implemented an English and Spanish mobile family-safety-reporting tool, staff and family education, and process for sharing comments with unit leaders on a dedicated inpatient complex care service at a pediatric hospital. Families shared safety concerns via predischarge surveys (baseline and intervention) and mobile tool (intervention). Three physicians with patient safety expertise classified events. We compared safety-reporting baseline (via survey) versus intervention (via survey and/or mobile tool) with generalized estimating equations and sub-analyzed data by COVID-19-era and educational attainment. We also compared mobile tool-detected event rates with hospital voluntary incident reporting.
232 baseline and 208 intervention parents participated (78.2% consented); 29.5% of baseline families versus 38.2% of intervention families reported safety concerns (P = .09). Adjusted odds ratio (95% CI) of families reporting safety concerns intervention versus baseline was 1.6 (1.0–2.6) overall, 2.6 (1.3–5.4) for those with < college education, and 3.1 (1.3–7.3) in the COVID-19–era subgroup. Safety concerns reported via mobile tool (34.6% of enrolled parents) included 42 medical errors, 43 nonsafety-related quality issues, 11 hazards, and 4 other. 15% of mobile tool concerns were also detected with voluntary incident reporting.
Family safety-reporting was unchanged overall after implementing a mobile reporting tool, though reporting increased among families with lower educational attainment and during the COVID-19 pandemic. The tool identified many events not otherwise captured by staff-only voluntary incident reporting. Hospitals should proactively engage families in reporting to improve safety, quality, and equity.
Hospitals rely on voluntary incident reporting, which underdetects events, particularly in patients from disadvantaged backgrounds, and excludes patients/families. Families of children with medical complexity report high rates of safety events. Interventions to engage families in hospital-safety-reporting have had limited success.
Safety-reporting among families with lower educational attainment and during the COVID-19 pandemic improved after implementing a mobile family safety-reporting tool, though rates were unchanged overall. Hospitals should proactively engage families in reporting to improve safety and reduce disparities.
An estimated 250 000 US patients die annually from medical errors, making them a leading cause of death.1 Patients and families are vigilant partners in care2 who identify otherwise unrecognized safety concerns,3,4 including undocumented errors and adverse events.5–8 However, disparities in reporting exist: Families with lower educational attainment report safety concerns one-fifth to one-third as often as their counterparts.2,3
Families of children with medical complexity (CMC, children with lifelong, limiting, complex chronic conditions affecting multiple organ systems) are adept at identifying medical errors.2,3 This is likely because CMC are vulnerable to errors9–11 and their families are attuned to care nuances.
Despite ubiquity of medical errors, compelling data about patient and family safety-reporting,3,5,7,8 and patient and family ("family" hereafter) willingness to engage in safety efforts,12–14 most hospitals do not proactively include families in safety-reporting.15 Instead, hospitals rely largely on staff-only voluntary incident reporting to evaluate safety. Voluntary incident reporting captures 1% to 10% of events detected through systematic safety surveillance,16 and is prone to systemic bias, undercapturing events in patients who are Black17 or who speak languages other than English.18 Failing to comprehensively capture safety information, including from families, is a barrier to improving hospital safety, quality, and equity.
Efforts to engage families in hospital safety are increasing.19–21 However, interventions to engage families in hospital safety-reporting have low uptake.22,23 Additionally, families are frequently unaware how to report safety concerns.24 For instance, ratings of whether families are told how to report hospital safety concerns are the single-lowest rated experience domain.25
Building on previous research2,3 and disparities in family-safety-reporting, we coproduced26 with families, nurses, physicians, and hospital leaders an equity-focused, safety-reporting intervention for families of hospitalized CMC to share safety concerns and suggestions. Given our emphasis on coproduction, health literacy, and equity, we hypothesized that family safety-reporting would improve after intervention implementation, particularly among groups who might not otherwise report, and we would identify events not captured in voluntary incident reporting.
Methods
We conducted this pre–post intervention trial from April 2018 to February 2022 at a quaternary academic center with a dedicated, multiunit, inpatient complex care service (for patients with multiorgan involvement, multispecialty needs, technology-dependence, and/or neurologic impairment) with around 500 admissions per year.
The hospital has a voluntary incident reporting system that patients and families cannot directly report into, although staff can select whether the family was involved in identifying the concern.
We recruited Spanish- and English-speaking parents/caregivers of patients of all ages on weekdays. We excluded international patients (because much of their care depends on embassy approval outside the medical team’s control), as well as patients in state custody or living in residential facilities.
Parents provided verbal consent facilitated by an information sheet and received gift cards upon completion of (or verbalized intent to later complete) surveys. We consented Spanish-speakers via in-person or video interpreters and reconsented families of readmitted patients. When families were not present at bedside, we returned to the bedside several times and checked with nurses to determine when families might be present. The hospital’s institutional review board approved the study.
Coronavirus disease 2019 (COVID-19) interrupted our study 1.5 months postintervention, leading to a 3-month data collection pause followed by 6 months of additional COVID-19–era baseline data collection before we reimplemented the intervention with COVID-19–conscious protocols.
Intervention
The Family Activation and Communication About Errors and Safety (FACES) (Figs 1–2) intervention involved:
The multicomponent FACES intervention. FACES involved: (1) a Spanish and English mobile reporting tool (an electronic family safety comment card) for families, (2) family and staff education, and (3) a process for sharing comments and learnings in real-time with unit and hospital leaders. FACES was developed over a 2-year period and informed by qualitative methods, communication science, organizational behavior, and health literacy principles. Qualitative interviews of parents, nurses, physicians, and hospital leaders revealed unclear/nontransparent reporting processes, variable staff/leader buy-in, misaligned staff–family priorities, and a complex family decision-making calculus about whether to report.27 Health literacy principles included using plain, nonmedical language with a fifth- to sixth-grade reading level and audiovisual redundancy (bolstering written materials with verbal explanations).
The multicomponent FACES intervention. FACES involved: (1) a Spanish and English mobile reporting tool (an electronic family safety comment card) for families, (2) family and staff education, and (3) a process for sharing comments and learnings in real-time with unit and hospital leaders. FACES was developed over a 2-year period and informed by qualitative methods, communication science, organizational behavior, and health literacy principles. Qualitative interviews of parents, nurses, physicians, and hospital leaders revealed unclear/nontransparent reporting processes, variable staff/leader buy-in, misaligned staff–family priorities, and a complex family decision-making calculus about whether to report.27 Health literacy principles included using plain, nonmedical language with a fifth- to sixth-grade reading level and audiovisual redundancy (bolstering written materials with verbal explanations).
The FACES mobile reporting tool (English and Spanish). Communication science principles included using exemplars (representative pictures of patients from the unit) on materials,50 gain framing (eg, “help us make care safer for your child and all children”),51 and issue framing (eg, “families are experts”).52 The 4 Frames model of organizational behavior (structural, political, symbolic, and human resources) also informed FACES. Structurally, we created a formal mechanism for safety-reporting and had a process to hand off concerns to existing hospital structures, like patient relations and unit leadership. Politically, we garnered buy-in through continual feedback from leaders in safety, quality, and leadership, as well as front-line staff. Symbolically, we created a shared vision of our intervention for families and staff alike on the basis of framing that resonated with all stakeholders (“Families are experts. Let’s work together to make care safer for all children.”) and developed a brochure and signage with photos of actual providers and families from the units. From a human resources perspective, we partnered with physician and nursing leadership. First page of the FACES mobile family safety reporting tool, available as a secure REDCap link accessible by e-mail, text, or quick response (QR) code.
The FACES mobile reporting tool (English and Spanish). Communication science principles included using exemplars (representative pictures of patients from the unit) on materials,50 gain framing (eg, “help us make care safer for your child and all children”),51 and issue framing (eg, “families are experts”).52 The 4 Frames model of organizational behavior (structural, political, symbolic, and human resources) also informed FACES. Structurally, we created a formal mechanism for safety-reporting and had a process to hand off concerns to existing hospital structures, like patient relations and unit leadership. Politically, we garnered buy-in through continual feedback from leaders in safety, quality, and leadership, as well as front-line staff. Symbolically, we created a shared vision of our intervention for families and staff alike on the basis of framing that resonated with all stakeholders (“Families are experts. Let’s work together to make care safer for all children.”) and developed a brochure and signage with photos of actual providers and families from the units. From a human resources perspective, we partnered with physician and nursing leadership. First page of the FACES mobile family safety reporting tool, available as a secure REDCap link accessible by e-mail, text, or quick response (QR) code.
1. an English and Spanish mobile family-safety-reporting tool (“mobile tool” hereafter) available through personalized e-mail, personalized text, or anonymous quick response (QR) code;
2. staff and family education; and
3. a process for sharing comments with unit and hospital leaders.
For pragmatic reasons and feasibility of staff training, we implemented FACES on the 5 most common units that admitted patients from the complex care service.
A team of parents with patient-safety expertise; parents, nurses, and physicians with experience caring for CMC; and physicians with quality, safety, health literacy, and leadership expertise coproduced FACES. FACES was also informed by qualitative interviews,27 communication science, and organizational behavior. The intervention was equity-focused through its emphasis on meaningful partnership with families from different backgrounds, inclusion of English and Spanish speakers, and attention to health literacy. To develop materials, we used universal health literacy precautions, checked Flesch-Kincaid readability statistics, and piloted materials with end-users. We also professionally translated study materials into Spanish and a bilingual team member reviewed them.
Data Collection
During baseline and intervention periods, we surveyed families predischarge about safety concerns. This survey, adapted from previous research2,3 and described elsewhere,24 provided examples of safety issues and asked whether any occurred. Next, families completed questions about safety experience, including the Child Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) measure about whether the hospital told them how to report safety concerns.15 They then completed the Children’s Hospital Safety Climate Questionnaire,28 the Parent–Patient Activation Measure29 (0–100 score; 100 = highest activation), demographics (including self-reported race and ethnicity), and finally, the Newest Vital Sign30 to measure health literacy (categorized as adequate health literacy versus possibility of limited health literacy versus high likelihood of limited health literacy).
In the intervention period, families were also prompted to submit safety comments via mobile tool, which included examples of events with definitions and prompted families to share every 2 business days x 3, then weekly until discharge (Fig 2). Families were instructed to report concerns both via mobile tool and survey, and could report multiple concerns each time. Hospital interpreter services translated Spanish mobile tool and survey responses within 24 business hours.
Event Review
A parent–nurse–physician team with patient safety expertise, including a parent–patient safety advocate with decades of experience, reviewed safety concerns from the mobile tool in real-time and biweekly. They reviewed patient charts for further information, brainstormed quality improvement projects, and notified clinical staff if patient care was at risk. All safety concerns (from survey and mobile tool) also underwent post-hoc review by 3 trained physicians with patient safety expertise. They separated and consolidated comments into unique concerns, then reviewed and classified concerns on the basis of previous research2,3,24 into nonpreventable adverse events, harmful errors, nonharmful errors, hazards, nonsafety-related quality issues, or neither (kappa = 0.40 [0.16–0.63]). They also rated events by category (eg, diagnosis), National Coordinating Council for Medical Error Reporting and Prevention (NCC MERP)3 classification, and communication-relatedness.
Analyses
Our primary outcome was family-reported safety concerns, defined as reporting safety concern(s) via predischarge survey (baseline and intervention) or mobile tool (intervention). Safety concerns were counted only once if reported both via survey and mobile tool. Secondary outcomes were safety experience and climate scores. Given that our previous research showed disparities in reporting by parental education,2 our preplanned analyses included evaluating reporting by educational attainment.
We descriptively summarized family characteristics and study outcomes using means (SD) for continuous variables and proportions for categorical variables. We categorized income and education into roughly even groups on the basis of our previous research.2 χ2 tests compared categorical characteristics according to baseline versus intervention group. The Shapiro-Wilk test tested normality for continuous variables. We used t tests for normally distributed continuous variables and Wilcoxon Rank-Sum tests for nonparametric continuous variables.
We used generalized estimating equation models to examine associations between safety concerns and intervention period, adjusting for within-patient clustering and patient and family characteristics varying by intervention period. We conducted subgroup analyses by educational attainment and cases occurring during the COVID-19 pandemic (July 2020 onwards), hypothesizing that the pandemic may have changed reporting. We also examined top-box (top-most, eg, 5 of 5, Likert scale) safety climate scores baseline versus intervention and proportion of parents reporting “yes definitely” or “yes somewhat” versus “no” to the Child HCAHPS “tell you how to report” question.31 Two-sided P values <.05 were considered statistically significant.
Finally, we examined staff-reported voluntary incident reports to determine percentage of events reported via mobile tool that were also captured in voluntary incident reports.
We used Research Electronic Data Capture (REDCap) for study management and SAS v.9.0 for analyses.
Results
Sample
We enrolled 440 participants (baseline = 232, intervention = 208), of which 364 (82.7%) completed surveys (baseline = 207, intervention = 157) (Supplemental Fig 3).
Family participants were mostly parents (94.5%), female (83.9%), and 46 years old (SD = 11) on average. Overall, 5.0% were non-Hispanic Black, 17.8% Hispanic, and 71.4% non-Hispanic white; 14.8% had limited English proficiency, 19.5% had limited health literacy, 52.6% had not completed college, and 43.9% had an annual income <$50 000. Mean Parent–Patient Activation Measure (P-PAM) score was 77.0 (SD = 18.4).
Patients’ mean characteristics were an age of 13 years old (SD = 7), 9-day length-of-stay (SD = 13), 23 lifetime admissions (SD = 39), and 3 complex chronic conditions (SD = 1). Overall, 29.9% required ICU during admission and 54.8% had public insurance.
Patient and parent age differed significantly between baseline and intervention periods; other characteristics did not (Table 1, Table 2). Parents of males and patients with longer lengths of stay reported more frequently than their counterparts (Supplemental Table 5 and Supplemental Table 6).
Family Participant Characteristics by Intervention Period
| A. Family Respondent Characteristic | Baseline(N = 207) n (%) | Intervention(N = 157) n (%) | Total(N = 364) n (%) | Pd |
|---|---|---|---|---|
| Relationship to patient (n = 362) | ||||
| Parent | 190 (92.7) | 152 (96.8) | 342 (94.5) | .3d |
| Grandparent | 4 (2.0) | 1 (0.6) | 5 (1.4) | |
| Guardian | 7 (3.4) | 4 (2.6) | 11 (3.0) | |
| Foster parent | 3 (1.5) | 0 (0.0) | 3 (0.8) | |
| Othera | 1 (0.5) | 0 (0.0) | 1 (0.3) | |
| Age in y (n = 352) | ||||
| Mean (SD) | 48 (11) | 45 (10) | 46 (11) | .02 |
| Gender (n = 360) | ||||
| Female | 168 (82.0) | 134 (86.5) | 302 (83.9) | .3 |
| Male | 37 (18.1) | 21 (13.6) | 58 (16.1) | |
| Race and ethnicity (n = 360) | ||||
| American Indian or Alaskan Native, non-Hispanic | 1 (0.5) | 0 (0.0) | 1 (0.3) | .5 |
| Asian American/Pacific Islander, non-Hispanic | 5 (2.4) | 4 (2.6) | 9 (2.5) | |
| Black/African American, non-Hispanic | 9 (4.4) | 9 (5.8) | 18 (5.0) | |
| Hispanic | 30 (14.6) | 34 (21.9) | 64 (17.8) | |
| Multi-racial, non-Hispanicb | 4 (2.0) | 1 (0.7) | 5 (1.4) | |
| Other, non-Hispanicc | 4 (2.0) | 2 (1.3) | 6 (1.7) | |
| White, non-Hispanic | 152 (74.2) | 105 (67.7) | 257 (71.4) | |
| English proficiency (n = 359) | ||||
| Proficient | 174 (84.9) | 132 (85.7) | 306 (85.2) | .8 |
| Limited English proficient | 31 (15.1) | 22 (14.3) | 53 (14.8) | |
| Education attainment (n = 361) | ||||
| Less than college | 105 (51.0) | 85 (54.8) | 190 (52.6) | .5 |
| Completed college or higher | 101 (49.0) | 70 (45.2) | 171 (47.4) | |
| Annual household income (n = 351) | ||||
| <$50 000 | 86 (43.2) | 68 (44.7) | 154 (43.9) | .9 |
| $50 000–$100 000 | 52 (26.1) | 36 (23.7) | 88 (25.1) | |
| ≥$100 000 | 61 (30.7) | 48 (31.6) | 109 (31.1) | |
| Health literacy (newest vital sign) (n = 308) | ||||
| High likelihood (50% or more) of limited literacy | 11 (6.2) | 6 (4.6) | 17 (5.5) | .8 |
| Possibility of limited literacy | 26 (14.6) | 17 (13.1) | 43 (14.0) | |
| Adequate literacy | 141 (79.2) | 107 (82.3) | 248 (80.5) | |
| Parent–patient activation measure score (n = 327) | ||||
| Mean (SD) | 77.3 (17.1) | 76.7 (20.2) | 77.0 (18.4) | .8 |
| A. Family Respondent Characteristic | Baseline(N = 207) n (%) | Intervention(N = 157) n (%) | Total(N = 364) n (%) | Pd |
|---|---|---|---|---|
| Relationship to patient (n = 362) | ||||
| Parent | 190 (92.7) | 152 (96.8) | 342 (94.5) | .3d |
| Grandparent | 4 (2.0) | 1 (0.6) | 5 (1.4) | |
| Guardian | 7 (3.4) | 4 (2.6) | 11 (3.0) | |
| Foster parent | 3 (1.5) | 0 (0.0) | 3 (0.8) | |
| Othera | 1 (0.5) | 0 (0.0) | 1 (0.3) | |
| Age in y (n = 352) | ||||
| Mean (SD) | 48 (11) | 45 (10) | 46 (11) | .02 |
| Gender (n = 360) | ||||
| Female | 168 (82.0) | 134 (86.5) | 302 (83.9) | .3 |
| Male | 37 (18.1) | 21 (13.6) | 58 (16.1) | |
| Race and ethnicity (n = 360) | ||||
| American Indian or Alaskan Native, non-Hispanic | 1 (0.5) | 0 (0.0) | 1 (0.3) | .5 |
| Asian American/Pacific Islander, non-Hispanic | 5 (2.4) | 4 (2.6) | 9 (2.5) | |
| Black/African American, non-Hispanic | 9 (4.4) | 9 (5.8) | 18 (5.0) | |
| Hispanic | 30 (14.6) | 34 (21.9) | 64 (17.8) | |
| Multi-racial, non-Hispanicb | 4 (2.0) | 1 (0.7) | 5 (1.4) | |
| Other, non-Hispanicc | 4 (2.0) | 2 (1.3) | 6 (1.7) | |
| White, non-Hispanic | 152 (74.2) | 105 (67.7) | 257 (71.4) | |
| English proficiency (n = 359) | ||||
| Proficient | 174 (84.9) | 132 (85.7) | 306 (85.2) | .8 |
| Limited English proficient | 31 (15.1) | 22 (14.3) | 53 (14.8) | |
| Education attainment (n = 361) | ||||
| Less than college | 105 (51.0) | 85 (54.8) | 190 (52.6) | .5 |
| Completed college or higher | 101 (49.0) | 70 (45.2) | 171 (47.4) | |
| Annual household income (n = 351) | ||||
| <$50 000 | 86 (43.2) | 68 (44.7) | 154 (43.9) | .9 |
| $50 000–$100 000 | 52 (26.1) | 36 (23.7) | 88 (25.1) | |
| ≥$100 000 | 61 (30.7) | 48 (31.6) | 109 (31.1) | |
| Health literacy (newest vital sign) (n = 308) | ||||
| High likelihood (50% or more) of limited literacy | 11 (6.2) | 6 (4.6) | 17 (5.5) | .8 |
| Possibility of limited literacy | 26 (14.6) | 17 (13.1) | 43 (14.0) | |
| Adequate literacy | 141 (79.2) | 107 (82.3) | 248 (80.5) | |
| Parent–patient activation measure score (n = 327) | ||||
| Mean (SD) | 77.3 (17.1) | 76.7 (20.2) | 77.0 (18.4) | .8 |
P values < .05 are bolded.
The other relationship to patient category includes adoptive parent (n = 1).
The multiracial category includes Black/African American and Caribbean (n = 1), American Indian or Alaskan Native and white (n = 2), American Indian or Alaskan Native and Asian Indian (n = 1), white and Middle Eastern (n = 1).
The other non-Hispanic category for race and ethnicity includes Cape Verdean (n = 1), Somalian (n = 1), Portuguese and American (n = 1), Jewish (n = 1), unknown (n = 2).
Summary statistics include only nonmissing data. Two-sided P values are from χ2 tests for categorical variables and t tests (when normally distributed) or Wilcoxon Rank-Sum tests for continuous variables.
Patient Characteristics by Intervention Period
| B. Patient Characteristic | Baseline(N = 207) n (%) | Intervention(N = 157) n (%) | Total(N = 364) n (%) | Pc |
|---|---|---|---|---|
| Age at admission in y (n = 364) | ||||
| Mean (SD) | 14 (7) | 12 (7) | 13 (7) | .01 |
| Gender (n = 364) | ||||
| Female | 91 (44.0) | 67 (42.7) | 158 (43.4) | .8 |
| Male | 116 (56.0) | 90 (57.3) | 206 (56.6) | |
| Race and ethnicity (n = 354) | ||||
| Asian American/PI, non-Hispanic | 2 (1.0) | 7 (4.6) | 9 (2.5) | .07 |
| Black, non-Hispanic | 20 (9.9) | 13 (8.6) | 33 (9.3) | |
| Hispanic | 29 (14.2) | 34 (22.5) | 63 (17.8) | |
| Multiracial, non-Hispanica | 1 (0.5) | 0 (0.0) | 1 (0.3) | |
| Other, non-Hispanic | 5 (2.5) | 2 (1.3) | 7 (2.0) | |
| White, non-Hispanic | 146 (71.9) | 95 (62.9) | 241 (68.1) | |
| Length of stay in d (n = 364) | ||||
| Mean (SD) | 10 (14) | 9 (12) | 9 (13) | .7 |
| Insurance (n = 363) | ||||
| Commercial | 101 (49.0) | 62 (39.5) | 163 (44.9) | .1 |
| Public | 105 (51.0) | 94 (59.9) | 199 (54.8) | |
| Other | 0 (0.0) | 1 (0.6) | 1 (0.3) | |
| Required ICU during admission (n = 364) | ||||
| No | 148 (71.5) | 107 (68.2) | 255 (70.1) | .5 |
| Yes | 59 (28.5) | 50 (31.9) | 109 (29.9) | |
| 12-mo overnight hospitalization in visits (n = 329) | ||||
| Mean (SD) | 3 (3) | 3 (3) | 3 (3) | 0.96 |
| Lifetime overnight hospitalization in visits (n = 299) | ||||
| Mean (SD) | 22 (24) | 25 (53) | 23 (39) | .3 |
| Complex chronic condition number (n = 364) | ||||
| 0 or 1 | 18 (8.7) | 15 (9.6) | 33 (9.1) | .5 |
| 2 or 3 | 82 (39.6) | 71 (45.2) | 153 (42.0) | |
| ≥4 | 107 (51.7) | 71 (45.2) | 178 (48.9) | |
| Complex chronic condition typesb (n = 364) | ||||
| Cardiovascular | 16 (7.7) | 13 (8.3) | 29 (8.0) | |
| Gastrointestinal | 157 (75.9) | 111 (70.7) | 268 (73.6) | |
| Hematologic or immunologic | 17 (8.2) | 18 (11.5) | 35 (9.6) | |
| Malignancy | 1 (0.5) | 2 (1.3) | 3 (0.8) | |
| Metabolic | 58 (28.0) | 43 (27.4) | 101 (27.8) | |
| Neurologic and neuromuscular | 160 (77.3) | 122 (77.7) | 282 (77.5) | |
| Other congenital or genetic defects | 57 (27.5) | 35 (22.3) | 92 (25.3) | |
| Renal and urologic | 30 (14.5) | 22 (14.0) | 52 (14.3) | |
| Respiratory | 37 (17.9) | 21 (13.4) | 58 (15.9) | |
| Premature and neonatal | 5 (2.4) | 9 (5.7) | 14 (3.9) | |
| Technology dependent | 177 (85.5) | 124 (79.0) | 301 (82.7) | |
| Transplant | 0 (0.0) | 0 (0.0) | 0 (0.0) | |
| B. Patient Characteristic | Baseline(N = 207) n (%) | Intervention(N = 157) n (%) | Total(N = 364) n (%) | Pc |
|---|---|---|---|---|
| Age at admission in y (n = 364) | ||||
| Mean (SD) | 14 (7) | 12 (7) | 13 (7) | .01 |
| Gender (n = 364) | ||||
| Female | 91 (44.0) | 67 (42.7) | 158 (43.4) | .8 |
| Male | 116 (56.0) | 90 (57.3) | 206 (56.6) | |
| Race and ethnicity (n = 354) | ||||
| Asian American/PI, non-Hispanic | 2 (1.0) | 7 (4.6) | 9 (2.5) | .07 |
| Black, non-Hispanic | 20 (9.9) | 13 (8.6) | 33 (9.3) | |
| Hispanic | 29 (14.2) | 34 (22.5) | 63 (17.8) | |
| Multiracial, non-Hispanica | 1 (0.5) | 0 (0.0) | 1 (0.3) | |
| Other, non-Hispanic | 5 (2.5) | 2 (1.3) | 7 (2.0) | |
| White, non-Hispanic | 146 (71.9) | 95 (62.9) | 241 (68.1) | |
| Length of stay in d (n = 364) | ||||
| Mean (SD) | 10 (14) | 9 (12) | 9 (13) | .7 |
| Insurance (n = 363) | ||||
| Commercial | 101 (49.0) | 62 (39.5) | 163 (44.9) | .1 |
| Public | 105 (51.0) | 94 (59.9) | 199 (54.8) | |
| Other | 0 (0.0) | 1 (0.6) | 1 (0.3) | |
| Required ICU during admission (n = 364) | ||||
| No | 148 (71.5) | 107 (68.2) | 255 (70.1) | .5 |
| Yes | 59 (28.5) | 50 (31.9) | 109 (29.9) | |
| 12-mo overnight hospitalization in visits (n = 329) | ||||
| Mean (SD) | 3 (3) | 3 (3) | 3 (3) | 0.96 |
| Lifetime overnight hospitalization in visits (n = 299) | ||||
| Mean (SD) | 22 (24) | 25 (53) | 23 (39) | .3 |
| Complex chronic condition number (n = 364) | ||||
| 0 or 1 | 18 (8.7) | 15 (9.6) | 33 (9.1) | .5 |
| 2 or 3 | 82 (39.6) | 71 (45.2) | 153 (42.0) | |
| ≥4 | 107 (51.7) | 71 (45.2) | 178 (48.9) | |
| Complex chronic condition typesb (n = 364) | ||||
| Cardiovascular | 16 (7.7) | 13 (8.3) | 29 (8.0) | |
| Gastrointestinal | 157 (75.9) | 111 (70.7) | 268 (73.6) | |
| Hematologic or immunologic | 17 (8.2) | 18 (11.5) | 35 (9.6) | |
| Malignancy | 1 (0.5) | 2 (1.3) | 3 (0.8) | |
| Metabolic | 58 (28.0) | 43 (27.4) | 101 (27.8) | |
| Neurologic and neuromuscular | 160 (77.3) | 122 (77.7) | 282 (77.5) | |
| Other congenital or genetic defects | 57 (27.5) | 35 (22.3) | 92 (25.3) | |
| Renal and urologic | 30 (14.5) | 22 (14.0) | 52 (14.3) | |
| Respiratory | 37 (17.9) | 21 (13.4) | 58 (15.9) | |
| Premature and neonatal | 5 (2.4) | 9 (5.7) | 14 (3.9) | |
| Technology dependent | 177 (85.5) | 124 (79.0) | 301 (82.7) | |
| Transplant | 0 (0.0) | 0 (0.0) | 0 (0.0) | |
P values <.05 are bolded. PI, pacific Islander.
The multiracial, non-Hispanic category for race and ethnicity includes white and other (n = 1).
Numbers in each category are not mutually exclusive. P value therefore not provided because of multiple comparisons.
Summary statistics include only nonmissing data. Two-sided P values are from χ2 tests for categorical variables and t tests (when normally distributed) or Wilcoxon Rank-Sum tests for continuous variables.
Family-Safety-Reporting
The percentage of families reporting safety concerns went from 29.5% (61 of 207) at baseline to 38.2% (60 of 157) during intervention (P = .09). Adjusted odds ratio (aOR [95% confidence interval]) of families reporting safety concerns during intervention versus baseline was 1.6 (1.0–2.6), P = .06. (Table 3)
Family Safety-Reporting Pre Versus Postintervention (Overall, by Educational Attainment, and COVID-19 Subgroup)
| Yes, Total (%) | OR (95% CI) | P | OR (95% CI) | P | |||
|---|---|---|---|---|---|---|---|
| Baselinea | Interventionb | Interaction P | (Unadjusted) [Ref: Baseline] | (Adjusted for age) [Ref: Baseline] | |||
| Overall (N = 364) | 61 of 207 (29) | 60 of 157 (38) | 1.5 (0.9 − 2.4) | .09 | 1.6 (1.0 − 2.6) | .06 | |
| Educational attainment (N = 361) | .10 | ||||||
| Less than college (N = 190) | 20 of 105 (19) | 30 of 85 (35) | 2.4 (1.2 − 4.8) | .02 | 2.6 (1.3 − 5.4) | .01 | |
| Completed college or higher (N = 171) | 40 of 101 (40) | 29 of 70 (41) | 1.1 (0.5 − 2.1) | .86 | 1.1 (0.5 − 2.1) | .8 | |
| COVID-19 time period (N = 364) | .20 | ||||||
| Before COVID-19 (N = 171) | 52 of 155 (34) | 6 of 16 (38) | 1.2 (0.4 − 3.8) | .75 | 1.2 (0.4 − 3.9) | .7 | |
| During COVID (N = 193) | 9 of 52 (17) | 54 of 141 (38) | 2.9 (1.2 − 1.8) | .02 | 3.1 (1.3 − 7.3) | .01 | |
| Yes, Total (%) | OR (95% CI) | P | OR (95% CI) | P | |||
|---|---|---|---|---|---|---|---|
| Baselinea | Interventionb | Interaction P | (Unadjusted) [Ref: Baseline] | (Adjusted for age) [Ref: Baseline] | |||
| Overall (N = 364) | 61 of 207 (29) | 60 of 157 (38) | 1.5 (0.9 − 2.4) | .09 | 1.6 (1.0 − 2.6) | .06 | |
| Educational attainment (N = 361) | .10 | ||||||
| Less than college (N = 190) | 20 of 105 (19) | 30 of 85 (35) | 2.4 (1.2 − 4.8) | .02 | 2.6 (1.3 − 5.4) | .01 | |
| Completed college or higher (N = 171) | 40 of 101 (40) | 29 of 70 (41) | 1.1 (0.5 − 2.1) | .86 | 1.1 (0.5 − 2.1) | .8 | |
| COVID-19 time period (N = 364) | .20 | ||||||
| Before COVID-19 (N = 171) | 52 of 155 (34) | 6 of 16 (38) | 1.2 (0.4 − 3.8) | .75 | 1.2 (0.4 − 3.9) | .7 | |
| During COVID (N = 193) | 9 of 52 (17) | 54 of 141 (38) | 2.9 (1.2 − 1.8) | .02 | 3.1 (1.3 − 7.3) | .01 | |
P values <.05 are bolded. CI, confidence interval; Ref, reference.
Defined as reporting through predischarge survey.
Defined as reporting, either through comment card or through predischarge survey.
Among those with less than college education (n = 190), 19.0% (20 of 105) reported at baseline versus 35.3% (30 of 85) during intervention (aOR 2.6 [1.3–5.4], P = .01). Among those with greater than or equal to college education (n = 171), 39.6% (40 of 101) reported at baseline versus 41.4% (29 of 70) during intervention (aOR 1.1 [0.5–2.1], P = .83).
In the COVID-19–era subgroup (n = 193), 17.3% (9 of 52) parents reported at baseline versus 38.3% (54 of 141) during intervention (aOR of 3.1 [1.3–7.3], P = .01).
Mobile Tool
Overall, 93.4% (n = 157) of families recommended the mobile tool be routinely available. Seventy-two families (34.6%) reported 121 unique safety comments via mobile tool (83% concerns [n = 100], 17% compliments [n = 21]) (Supplemental Fig 4). They typically submitted comments in the afternoon or evening a few days into admission but before discharge. Four of 19 (21%) Spanish-speaking families and 18 of 69 (26%) families with lower educational attainment shared via mobile tool. Among the 100 safety concerns reported via mobile tool, physician-reviewers coded 42% medical errors (15 harmful; 27 nonharmful), 43% nonsafety-related quality issues, 11% hazards, and 4% other. The research team acted on concerns in 18% of cases. Patient-level actions taken included notifying clinical staff that family felt their medical concerns were dismissed. Unit- and systems-level actions included sharing recurring concerns (eg, formula errors, bed rails being left down, medication reconciliation errors) with unit and hospital leadership to inform systems-level improvements.
Events reported via mobile tool (Table 4) were most commonly in the other cares/therapy (24%), environment/equipment (22%), and medication (18%) categories. Although 46% of events were nonclassifiable by NCC MERP (ie, not errors), those events reported by mobile tool that were deemed errors had NCC MERP ratings from A to F. The most common NCC MERP rating was level C (error reached patient but did not cause harm, 20%) followed by A (circumstances or events with capacity to cause error, 10%).
Examples of Family Safety Comments by Category and Potential QI Projects
| Category | Family Safety Comment | Event Type | Potential QI Project |
|---|---|---|---|
| Other cares/therapy | “The charge nurse was very rude. I asked to speak with her in regards to discharge food plan (changing foods). She was very mean and I did make a report to clinical coordinator and will be calling patient relations.” | Nonsafety-related quality | Customer service training for staff |
| “Nurses allow a lot of things to fall on the floor, like caps. To my untrained eyes, this seems like a slipping hazard… maybe picking them up would mean expos[ure] to germs.” | Hazard | Staff education around discarding supplies | |
| “Twice today I have called the nurse button to help with bed transfers and they never got the message. It wasn't urgent, but I cannot move my son in traction on his own, so I wish the message was relayed better.” | Hazard | Call bell policy, managing family expectations around response times | |
| “Staff was great about sitting with my daughter while I showered or grabbed food since she cannot be left alone.” | Compliment | Planning family breaks (eg, around nursing cares) | |
| “His surgeon from the moment we met him listened to our situation, prioritized [our] son, and assured us of his ongoing support.” | Compliment | Communication training | |
| Diagnosis | “His face was so swollen, and he started intermittent gasping, which was new. I had trouble finding the call bell and getting his nurse, but eventually got help. His nurse gave possible reasons but did not respond with urgency. Over the next hours, his BP dropped very low (30s over 40s), HR and respirations, up… I…knew these were not good signs. When I expressed concern repeatedly about his BP, the nurse kept say[ing] it was OK and ‘this happens.’ A pain provider came in, saw his BP, and immediately offered to get help. She dismissed her also. Soon, a team was outside his door, problem-solving, but did not include me until later at bedside. I did not appreciate the nurse dismissing my concerns at all. Turns out he was septic.” | Harmful error | Staff education around listening to families when concerned, family education around escalating concerns |
| “Came for [spinal surgery], became septic, wound infection and heart infection (he had no heart issues before this stay). I don’t know if this could have been prevented or not. It was E. coli that infected his wound, so could there have been better dressings IDK.” | Harmful error | Staff education around postop diaper changes, standard perioperative bowel regimens | |
| “Laboratories: Wrong laboratories were ordered (duplicate from the day before).” | Nonharmful error | Laboratory order training | |
| Environment/ equipment | “Understandably, the cleaning staff has their hands full these days. However, boogers on the TV keyboard is just gross. I know the hospital doesn’t like to leave out the Oxivir wipes, but is there some other wipe or cleanser that could be stocked in patient rooms [so] that families can clean something on their own…?” | Nonsafety-related quality | Patient access to cleaning supplies |
| “Purel[l] dispensers in the rooms are empty. If they are not able to be refilled, it would be helpful to have a container of hand sanitizer at the sink in each room. Or for scheduled surgeries, ask families to bring one into the hospital.” | Hazard | Ensuring all rooms stocked with hand sanitizer, standardizing refill policies | |
| “We had a great experience with the ortho tech helping us get equipment for the halo traction.” | Compliment | Equipment checklists | |
| Fall/fall risk | “My child was a fall risk 100% and was never labeled as one. His bed was not kept in the low position unless I lowered it.” | Nonharmful error | Admission checklist with fall risk, functional status |
| Feeds | “Alimentación iniciada con fórmula errada no la que le corresponda al Niño. Se inició Complete Pediatric, pero El Niño consume Complete Pediatric Reduce Calories.” [They started feeding my son with the wrong formula rather than using the one my son needed. They used Complete Pediatric instead of Complete Pediatric Reduced Calories, which my son normally takes.] | Nonharmful error | “Formula reconciliation” akin to medication reconciliation |
| “Le dieron la leche equivocada y esto provocó sáchelo de kitosis, por lo tanto también tubo un desorden de convulsiones.” [They gave the wrong milk, and this caused him to go out of ketosis, so he had a seizure.] | Harmful error | Ketogenic diet “time-out” before medicines or feeds in patients on ketogenic diet | |
| Hospital-acquired infection | “Admitted on a Tuesday chest x-ray is taken…symptoms persistent rest of week with no diagnosis…another x-ray on Sunday was used for comparison with a diagnosis of aspiration pneumonia started a course of antibiotics on which he appears to be turning the corner…Maybe a[n] earlier x-ray would of accelerated his healing? Hard to say.” | Harmful error | Aspiration pneumonia clinical education |
| Medicine/IV fluids | “No le dieron unos de los medicamentos para convulsiones (VPA) y El Niño tuvo un desorden de convulsiones por falta del medicamento.” [They did not give some of the medicine for seizures (VPA), and the child had a seizure because they did not give the medicine.] | Harmful error | Improving medication reconciliation processes for patients who speak languages other than English |
| “The nurses were so careful with my son's medications. They check[ed] and double-checked every time to make sure they were giving the right med to the right child in the right dosage.” | Compliment | Confirm medicines with family before giving | |
| “I feel my son may had gotten way too many medications antibiotics.” | Other | Communication training, e.g., family-centered rounds | |
| Procedure or surgery | “X-ray was taken in a position comfortable for the patient, keeping in mind of her pain and comfort.” | Compliment | Asking families of CMC re: preferred positioning |
| Category | Family Safety Comment | Event Type | Potential QI Project |
|---|---|---|---|
| Other cares/therapy | “The charge nurse was very rude. I asked to speak with her in regards to discharge food plan (changing foods). She was very mean and I did make a report to clinical coordinator and will be calling patient relations.” | Nonsafety-related quality | Customer service training for staff |
| “Nurses allow a lot of things to fall on the floor, like caps. To my untrained eyes, this seems like a slipping hazard… maybe picking them up would mean expos[ure] to germs.” | Hazard | Staff education around discarding supplies | |
| “Twice today I have called the nurse button to help with bed transfers and they never got the message. It wasn't urgent, but I cannot move my son in traction on his own, so I wish the message was relayed better.” | Hazard | Call bell policy, managing family expectations around response times | |
| “Staff was great about sitting with my daughter while I showered or grabbed food since she cannot be left alone.” | Compliment | Planning family breaks (eg, around nursing cares) | |
| “His surgeon from the moment we met him listened to our situation, prioritized [our] son, and assured us of his ongoing support.” | Compliment | Communication training | |
| Diagnosis | “His face was so swollen, and he started intermittent gasping, which was new. I had trouble finding the call bell and getting his nurse, but eventually got help. His nurse gave possible reasons but did not respond with urgency. Over the next hours, his BP dropped very low (30s over 40s), HR and respirations, up… I…knew these were not good signs. When I expressed concern repeatedly about his BP, the nurse kept say[ing] it was OK and ‘this happens.’ A pain provider came in, saw his BP, and immediately offered to get help. She dismissed her also. Soon, a team was outside his door, problem-solving, but did not include me until later at bedside. I did not appreciate the nurse dismissing my concerns at all. Turns out he was septic.” | Harmful error | Staff education around listening to families when concerned, family education around escalating concerns |
| “Came for [spinal surgery], became septic, wound infection and heart infection (he had no heart issues before this stay). I don’t know if this could have been prevented or not. It was E. coli that infected his wound, so could there have been better dressings IDK.” | Harmful error | Staff education around postop diaper changes, standard perioperative bowel regimens | |
| “Laboratories: Wrong laboratories were ordered (duplicate from the day before).” | Nonharmful error | Laboratory order training | |
| Environment/ equipment | “Understandably, the cleaning staff has their hands full these days. However, boogers on the TV keyboard is just gross. I know the hospital doesn’t like to leave out the Oxivir wipes, but is there some other wipe or cleanser that could be stocked in patient rooms [so] that families can clean something on their own…?” | Nonsafety-related quality | Patient access to cleaning supplies |
| “Purel[l] dispensers in the rooms are empty. If they are not able to be refilled, it would be helpful to have a container of hand sanitizer at the sink in each room. Or for scheduled surgeries, ask families to bring one into the hospital.” | Hazard | Ensuring all rooms stocked with hand sanitizer, standardizing refill policies | |
| “We had a great experience with the ortho tech helping us get equipment for the halo traction.” | Compliment | Equipment checklists | |
| Fall/fall risk | “My child was a fall risk 100% and was never labeled as one. His bed was not kept in the low position unless I lowered it.” | Nonharmful error | Admission checklist with fall risk, functional status |
| Feeds | “Alimentación iniciada con fórmula errada no la que le corresponda al Niño. Se inició Complete Pediatric, pero El Niño consume Complete Pediatric Reduce Calories.” [They started feeding my son with the wrong formula rather than using the one my son needed. They used Complete Pediatric instead of Complete Pediatric Reduced Calories, which my son normally takes.] | Nonharmful error | “Formula reconciliation” akin to medication reconciliation |
| “Le dieron la leche equivocada y esto provocó sáchelo de kitosis, por lo tanto también tubo un desorden de convulsiones.” [They gave the wrong milk, and this caused him to go out of ketosis, so he had a seizure.] | Harmful error | Ketogenic diet “time-out” before medicines or feeds in patients on ketogenic diet | |
| Hospital-acquired infection | “Admitted on a Tuesday chest x-ray is taken…symptoms persistent rest of week with no diagnosis…another x-ray on Sunday was used for comparison with a diagnosis of aspiration pneumonia started a course of antibiotics on which he appears to be turning the corner…Maybe a[n] earlier x-ray would of accelerated his healing? Hard to say.” | Harmful error | Aspiration pneumonia clinical education |
| Medicine/IV fluids | “No le dieron unos de los medicamentos para convulsiones (VPA) y El Niño tuvo un desorden de convulsiones por falta del medicamento.” [They did not give some of the medicine for seizures (VPA), and the child had a seizure because they did not give the medicine.] | Harmful error | Improving medication reconciliation processes for patients who speak languages other than English |
| “The nurses were so careful with my son's medications. They check[ed] and double-checked every time to make sure they were giving the right med to the right child in the right dosage.” | Compliment | Confirm medicines with family before giving | |
| “I feel my son may had gotten way too many medications antibiotics.” | Other | Communication training, e.g., family-centered rounds | |
| Procedure or surgery | “X-ray was taken in a position comfortable for the patient, keeping in mind of her pain and comfort.” | Compliment | Asking families of CMC re: preferred positioning |
BP, blood pressure; HR, heart rate; IDK, I don’t know; IV, intravenous; postop, postoperative; QI, quality improvement; re, regarding; VPA, valproic acid.
Only 15% of comments reported via mobile tool were also detected in voluntary incident reports. These included missed home antiepileptics, incorrect fluids, postoperative pain, and wound infections.
Safety Climate and Experience
Family-reported safety climate and safety experience were unchanged baseline versus intervention. For the Child HCAHPS safety experience item asking whether hospital staff told how to report concerns about mistakes, 36% (66 of 182) answered “yes definitely” or “yes somewhat” at baseline versus 46% (69 of 151) during intervention (aOR 1.4 [0.9–2.4], P = .15).
Discussion
In this pre–post trial of an equity-focused family safety-reporting intervention coproduced by parents, nurses, physicians, and hospital leaders, one-third of families of hospitalized CMC shared safety comments via mobile tool. Nearly half of comments were medical errors or hazards, most not captured by staff-only voluntary incident reporting. Family-safety-reporting did not change overall after intervention implementation. However, reporting approximately tripled in both the COVID-19–era intervention subgroup and in the subgroup of intervention families with lower educational attainment (who typically report safety concerns markedly less than their counterparts2,3 ). Thus, our intervention may help hospitals identify otherwise undocumented safety concerns, particularly during times of upheaval, and from families hospitals might not otherwise hear from.
Common safety concerns raised by families of CMC pertained to feeds, antiepileptics, and bedrails. Spanish-speaking families described safety lapses with feeds, antiepileptics, and the ketogenic diet. Quality improvement projects resulting from our study include a formula reconciliation project (considering formula akin to medication), a family-centered medication safety check (families reviewing medication orders for errors 24 hours postadmission), and individualized inpatient care guides co-created by outpatient providers and families (documenting preferences like positioning and communication).
Contextualizing Findings
Other studies also find that families are adept at reporting safety concerns during hospitalization,2,3,5,7,32 particularly families of CMC.2,3,24 However, previous studies show that families with lower education are 3 to 5 times less likely to report safety concerns.2,3 In the present study, comparable reporting by education and language suggests our efforts to encourage speaking up, normalize reporting, and use clear, easy-to-understand language may have encouraged reporting by families who might not otherwise share.
Unless designed with a focus on equity, interventions can worsen inequities by disproportionately improving care for already advantaged groups.33,34 To ensure our intervention would not increase reporting only among privileged families who might already be comfortable reporting, we designed our intervention based on interviews with Spanish-speaking families and those with lower educational attainment, families who might experience systemic, explicit, and implicit bias in how they are treated, communicated with, and perceived. Additionally, we emphasized plain language, used activating language (“families are experts”), included individuals from diverse backgrounds as exemplars51 in visual materials, and applied other communication science principles (eg, gain framing52 ) as we coproduced the intervention.
Our observed rate of family reporting via mobile tool was 30%, comparable to error rates detected through systematic surveillance2,35–37 and much higher than rates detected through voluntary incident reporting.16 Though most events reported by families were errors, hazards, or nonsafety-related quality concerns—all important information for hospitals—very few were detected in voluntary incident reporting. Skeptics of family safety-reporting argue it will identify mostly low-level events and “noise.” However, our study suggests that families are excellent reporters of safety and quality issues, including many that hospitals might not otherwise know about. These events might not be captured in voluntary incident-reporting systems for many reasons: staff fear of reporting, diffusion of responsibility, belief that certain events are too commonplace or low-level to report, or lack of knowledge of events.38,39 Additionally, families report concerns that matter to them. By proactively soliciting family-safety reporting, hospitals can show they care by asking, listening, and in turn addressing family safety and quality concerns.
Legal Concernsi
Some worry that family safety-reporting may have legal ramifications for hospitals and clinicians. However, giving families a clear mechanism to share existing concerns and suggestions they are often already raising to staff verbally24 is unlikely to increase litigation. Rather, giving families a voice and validating concerns40 might instead decrease litigation given that transparency and disclosure can help protect against litigation.41–43
Study Implications
FACES allows hospitals to track, learn from, and address concerns of importance to families that might not otherwise be prioritized, thereby fostering systems-level improvement and change. Additionally, it could help families, particularly those hospitals might not otherwise hear from, be more active participants in care, in keeping with principles of coproduction26 and family-centeredness espoused by the Institute for Patient and Family Centered Care and the American Academy of Pediatrics.44–46
Coproducing care with patients and families is not merely a “feel good” idea; it can improve safety and quality, as evidenced by other coproduced interventions.35 Moreover, transparency around safety can help change institutional culture and promote staff psychological safety, which in turn affects patient safety.47–49
Hospitals should proactively engage all families, particularly those of CMC and other high-risk populations like oncology, and end-stage renal disease, where parents provide a great deal of specialized care at home, in reporting safety concerns. Families are a captive audience who may not be subject to staff barriers to reporting (e.g., diffusion of responsibility, belief certain errors are too commonplace to report). Hospitals should make processes more transparent and multilingual, with health literacy and equity in mind to ensure that disparities do not widen. Safety can and should be framed in a positive, transparent, nonpunitive manner that emphasizes partnership. Interventions like FACES are promising for hospitals to implement, as are efforts to integrate family safety-reporting into experience surveys and voluntary incident reporting.
Our intervention admittedly has service recovery implications outside our study’s scope that must be accounted for when moving from research to operations. These might involve integrating family-safety-reporting into existing unit-based service recovery efforts or hiring additional staff. Such investments are likely to be a valuable use of resources given the potential positive impacts of family-safety reporting interventions on patient experience, safety, quality, and equity.
Limitations
Our study’s pre–post design and setting (quaternary hospital’s dedicated complex care service) may limit causal inference and generalizability. However, our intervention’s true impact may have been even greater than observed, given we were proactively surveying families about safety even in our baseline condition, thereby gathering more reports from families in our baseline period than routinely occurs. This would bias our intervention toward the null. Additionally, COVID-19 interrupted our study and may have affected survey response rates and changed reporting behaviors. For instance, our predischarge survey response rates were lower postintervention, perhaps because of the pandemic or staffing changes. However, we obtained additional COVID-19–era baseline data and conducted a subgroup analysis which suggested that reporting increased after our intervention, even when constrained to COVID-19–era data. This may have been because safety worsened and/or parents’ bedside role became even more critical during the pandemic.
We were surprised that the Child HCAHPS31 hospital telling families how to report concerns item did not improve post-intervention. We may have been underpowered to detect changes. Alternatively, we intentionally framed our intervention as an opportunity to improve safety for all, which contrasts with the Child HCAHPS question's framing around mistakes. Perhaps the Child HCAHPS question could be revised with more positive or neutral framing, or hospitals could acknowledge that mistakes happen and embrace such terminology with patients and families. Additionally, because our intervention was introduced by research assistants, not clinical staff, as a study, parents may not have considered it to represent the hospital. We also included only English- and Spanish-speakers. Expanding the mobile tool to other languages is an important focus of ongoing research.
Conclusions
Family safety-reporting did not change overall after implementing an equity-focused, coproduced, mobile family-safety reporting tool for hospitalized CMC. However, the intervention was associated with increased reporting among families with lower educational attainment and during the COVID-19 pandemic. It identified high rates of medical errors and captured events largely absent from staff-only voluntary incident reporting. Hospitals should proactively engage families in safety-reporting to improve safety, quality, and equity.
Acknowledgments
We thank the patients, families, nurses, and physicians who participated in this study.
Dr Khan conceptualized and designed the study, coordinated and supervised data collection, designed study instruments, drafted the initial manuscript, obtained funding, and participated in acquisition, analysis, or interpretation of data, and critical revision of the manuscript for important intellectual content; Drs Baird and Mauskar helped conceptualize and design the study, supported implementation and data collection efforts, and participated in acquisition, analysis, or interpretation of data, and critical revision of the manuscript for important intellectual content; Ms Haskell, Ms Mallick, and Ms Rogers helped conceptualize and design the study, helped support implementation and data collection efforts, and participated in acquisition, analysis, or interpretation of data, and critical revision of the manuscript for important intellectual content; Ms Aldarondo, Ms Copp, Ms Hennessy, Ms Matherson, Ms McGeachey, and Ms Pinkham helped support implementation and data collection efforts, participated in acquisition, analysis, or interpretation of data, and critically revised the manuscript for important intellectual content; Drs Berry, Humphrey, Quiñones-Pérez, Stoeck, Toomey, and Wilder helped support implementation and data collection efforts, participated in acquisition, analysis, or interpretation of data, and critically revised the manuscript for important intellectual content; Ms Habibi, Ms Ngo, and Ms Elder provided administrative, technical, or material support, and participated in acquisition, analysis, or interpretation of data, and critical revision of the manuscript for important intellectual content; Ms Liu and Ms Melvin performed statistical analyses, and participated in acquisition, analysis, or interpretation of data, and critical revision of the manuscript for important intellectual content; Dr Gray oversaw the statistical analyses and participated in acquisition, analysis, or interpretation of data, and critical revision of the manuscript for important intellectual content; Drs Luff, Singer, Viswanath, and Schuster helped conceptualize and design the study, and participated in acquisition, analysis, or interpretation of data, and critical revision of the manuscript for important intellectual content; Dr Landrigan conceptualized and designed the study, supervised the study, and participated in acquisition, analysis, or interpretation of data, and critical revision of the manuscript for important intellectual content; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
This study is registered at ClinicalTrials.gov, NCT02877017, https://clinicaltrials.gov/ct2/show/NCT02877017.
FUNDING: This project was supported by grant K08HS025781 from the Agency for Healthcare Research and Quality, a Charles H. Hood Foundation grant, and a grant from the Boston Children’s Hospital Academy for Teaching and Educational Innovation and Scholarship. The funders had no role in the design or conduct of this study.
CONFLICT OF INTEREST DISCLOSURES: Dr Landrigan holds equity in and has consulted with the I-PASS Patient Safety Institute. Dr. Baird has also consulted with the I-PASS Patient Safety Institute. The I-PASS Patient Safety Institute is a company that seeks to train institutions in best handoff practices and aid in their implementation. The I-PASS Patient Safety Institute had no role in the design or conduct of this study. Analyses were conducted by a statistical team who do not have any involvement with the I-PASS Patient Safety Institute. In addition, Dr Landrigan has received monetary awards, honoraria, and travel reimbursement from multiple academic and professional organizations for teaching and consulting on sleep deprivation, physician performance, handoffs, and safety, and has served as an expert witness in cases regarding patient safety and sleep deprivation. All other authors have indicated they have no conflicts of interest relevant to this article to disclose.
- aOR
adjusted odds ratio
- CMC
children with medical complexity
- COVID-19
coronavirus disease 2019
- FACES
Family Activation and Communication about Errors and Safety
- HCAHPS
Hospital Consumer Assessment of Healthcare Providers and Systems
- NCC MERP
National Coordinating Council for Medical Error Reporting and Prevention
Comments