Colleen Mangeot, MS

Adverse Childhood Experiences, or ACEs, include exposure to abuse, neglect, or household dysfunction (eg, having a parent who is mentally ill) as a child.¹ Exposure to ACEs affects health into adulthood, with a dose-response relationship between ACEs and a range of comorbidities.¹ Adults with 6 or more ACEs have a 20-year shorter life expectancy than do those with no ACEs.¹ Still, ACEs are static; once experienced, that experience cannot be undone. However, resilience, or positive adaptation in the context of adversity, can be protective, buffering the negative effects of ACEs.^2,3 Protective factors that promote resilience include social capital, such as positive relationships with caregivers and peers.³

With their clear link to health outcomes across the life-course, there is a movement for pediatricians to screen children for ACEs⁴ and to develop strategies that promote resilience in children, parents, and families. However, screening a child for adversity has challenges because younger children may not have experienced an adverse exposure, or they may be unable to voice their experiences. Studies have demonstrated that parental adversity, or ACEs, may be a marker for childhood adversity.^5,6 Biological models also support this potential intergenerational effect of ACEs. Chronic exposure to stress, including ACEs, results in elevated cortisol via a dysregulated hypothalamic-pituitary-adrenal axis, which results in chronic inflammation.⁷ This “toxic stress” is prolonged, severe in intensity, and can lead to epigenetic changes that may be passed on to the next generation.^8,9

Hospitalization of an ill child, and the transition to home after that hospitalization, is a stressful event for children and families.¹⁰ This stress may be relevant to parents that have a history of a high rate of ACEs or a current low degree of resilience. Our previous work demonstrated that, in the inpatient setting, parents with high ACEs (≥4) or low resilience have increased coping difficulty 14 days after their child’s hospital discharge.¹¹ Our objective here was to evaluate whether a parent’s ACEs and/or resilience would also be associated with that child’s likelihood of reutilization. We hypothesized that more parental ACEs and/or lower parental resilience would be associated with revisits the emergency room, urgent care, or hospital readmissions.

Participants and Study Design

We conducted a prospective cohort study of parents of hospitalized children recruited from the “Hospital-to-Home Outcomes” Studies (H2O I and H2O II).^12,13 H2O I and II were prospective, single-center, randomized controlled trials designed to determine the effectiveness of either a nurse-led transitional home visit (H2O I) or telephone call (H2O II) on 30-day unplanned healthcare reutilization. The trials and this study were approved by the Cincinnati Children’s Institutional Review Board. All parents provided written informed consent.

Details of H2O I and II recruitment and design have been described previously.^12,13 Briefly, children were eligible for inclusion in either study if they were admitted to our institution’s general Hospital Medicine or the Hospital Medicine Complex Care Services; for H2O I, children hospitalized on the Neurology and Neurosurgery services were also eligible.^12,13 Patients were excluded if they were discharged to a residential facility, if they lived outside the home healthcare nurse service area, if they were eligible for skilled home healthcare services (eg, intravenous antibiotics), or if the participating caregiver was non-English speaking.^12,13 In H2O I, families were randomized either to receive a single nurse home visit within 96 hours of discharge or standard of care. In H2O II, families enrolled were randomized to receive a telephone call by a nurse within 96 hours of discharge or standard of care. As we have previously published, randomization in both trials successfully balanced the intervention and control arms with respect to key demographic characteristics.^12,13 For the analyses presented here, we focused on a subset of caregivers 18 years and older whose children were enrolled in either H2O I or II between August 2015 and October 2016. In both H2O trials, face-to-face and paper-based questionnaires were completed by parents during the index hospitalization.

Outcome and Predictors

Our primary outcome was unanticipated healthcare reutilization defined as return to the emergency room, urgent care, or unplanned readmission within 30 days of hospital discharge, consistent with the H2O trials. This was measured using the primary institution’s administrative data supplemented by a utilization database shared across regional hospitals.¹⁴ Readmissions were identified as “unplanned” using a previously validated algorithm,¹⁵ and treated as a dichotomous yes/no variable.

Our primary predictors were parental ACEs and resilience (see Appendix Tables). The ACE questionnaire addresses abuse, neglect, and household dysfunction in the first 18 years of life.¹ It is composed of 10 questions, each with a yes/no response.¹ We defined parents as low (ACE 0), moderate (ACE 1-3), or high (ACE ≥4) risk a priori because previous literature has described poor outcomes in adults with 4 or more ACEs.¹⁶

Given the sensitive nature of the questions, respondents independently completed the ACE questionnaire on paper instead of via the face-to-face survey. Respondents returned the completed questionnaire to the research assistant in a sealed envelope. All families received educational information on relevant hospital and community-based resources (eg, social work).

Parental resilience was measured using the Brief Resilience Scale (BRS). The BRS is 6 items, each on a 5-point Likert scale. Responses were averaged, providing a total score of 1-5; higher scores are representative of higher resilience.¹⁷ We treated the BRS score as a continuous variable. BRS has been used in clinical settings; it has demonstrated positive correlation with social support and negative correlation with fatigue.¹⁷ Parents answered BRS questions during the index pediatric hospitalization in a face-to-face interview.

Parent and Child Characteristics

Parent and child sociodemographic variables were also obtained during the face-to-face interview. Parental variables included age, gender, educational attainment, household income, employment status, and financial and social strain.¹¹ Educational attainment was analyzed in 2 categories—high school or less vs more than high school—because most discharge instructions are written at a high school reading level.¹⁸ Parents reported their annual household income in the following categories: <$15,000; $15,000-$29,999; $30,000-$44,999; $45,000-$59,999; $60,000-$89,999; $90,000-$119,999; ≥$120,000. Employment was dichotomized as not employed/student vs any employment. Financial and social strain were assessed using a series of 9 previously described questions.¹⁹ These questions assessed, via self-report, a family’s ability to make ends meet, ability to pay rent/mortgage or utilities, need to move in with others because of financial reasons, and ability to borrow money if needed, as well as home ownership and parental marital status.^15,19 Strain questions were all dichotomous (yes/no, single/not single). A composite variable was then constructed that categorized those reporting no strain items, 1 to 2 items, 3 to 4 items, and 5 or more items.²⁰

Child variables included race, ethnicity, age, primary care access,²¹ payer, and H2O treatment arm. Race categories were white/Caucasian, black/African American, American Indian or Alaskan Native, Asian or Pacific Islander, and other; ethnicity categories were Hispanic/Latino, non-Hispanic/Latino, and unknown. Given relatively low numbers of children reported to be Hispanic/Latino, we combined race and ethnicity into a single variable, categorized as non-Hispanic/white, non-Hispanic/black, and multiracial/Hispanic/other. Primary care access was assessed using the access subscale to the Parent’s Perception of Primary Care questionnaire. This includes assessment of a family’s ability to travel to their doctor, to see their doctor for routine or sick care, and to get help or advice on evenings or weekends. Scores were categorized as always adequate, almost always adequate, or sometimes/never adequate.²¹ Payer was dichotomized to private or public/self-pay.

Statistical Analyses

We examined the distribution of outcomes, predictors, and covariates. We compared sociodemographic characteristics of those respondents and nonrespondents to the ACE screen using the chi-square test for categorical variables or the t test for continuous variables. We used logistic regression to assess for associations between the independent variables of interest and reutilization, adjusting for potential confounders. To build our adjusted, multivariable model, we decided a priori to include child race/ethnicity, primary care access, financial and social strain, and trial treatment arm. We treated the H2O I control group as the referent group. Other covariates considered for inclusion were caregiver education, household income, employment, and payer. These were included in multivariable models if bivariate associations were significant at the P < .1 level. We assessed an ACE-by-resilience interaction term because we hypothesized that those with more ACEs and lower resilience may have more reutilization outcomes than parents with fewer ACEs and higher resilience. We also evaluated interaction terms between trial arm assignment and predictors to assess effects that may be introduced by the randomization. Predictors in the final logistic regression model were significant at the P < .05 level. Logistic regression assumption of little or no multicollinearity among the independent variables was verified in the final models. All analyses were performed with Stata v16 (Stata Corp, College Station, Texas).

There were a total of 1,787 parent-child dyads enrolled in the H2O I and II during the study period; 1,320 parents (74%) completed the ACE questionnaire and were included the analysis. Included parents were primarily female and employed, as well as educated beyond high school (Table 1). Overall, 64% reported one or more ACEs (range 0 to 9); 45% reported 1to 3, and 19% reported 4 or more ACEs. The most commonly reported ACEs were divorce (n = 573, 43%), exposure to alcoholism (n = 306, 23%), and exposure to mental illness (n = 281, 21%; Figure 1). Parents had a mean BRS score of 3.97 (range 1.17-5.00), with the distribution shown in Figure 2.

Of the 1,320 included patients, the average length of stay was 2.5 days, and 82% of hospitalizations were caused by acute medical issues (eg, bronchiolitis). A total of 211 children experienced a reutilization event within 30 days of discharge. In bivariate analysis, children with parents with 4 or more ACEs had a 2.02-times (95% CI 1.35-3.02) higher odds of experiencing a reutilization event than did those with parents reporting no ACEs. Parents with higher resilience scores had children with a lower odds of reutilization (odds ratio [OR] 0.77 95% CI 0.63-0.95).

In addition to our a priori variables, parental education, employment, and insurance met our significance threshold for inclusion in the multivariable model. The ACE-by-resilience interaction term was not significant and not included in the model. Similarly, there was no significant interaction between ACE and resilience and H2O treatment arm; the interaction terms were not included in the final adjusted model, but treatment arm assignment was kept as a covariate. A total of 1,292 children, out of the 1,320 respondents, remained in the final multivariable model; the excluded 28 had incomplete covariate data but were not otherwise different. In this final adjusted model, children with parents reporting 4 or more ACEs had a 1.69-times (95% CI 1.11-2.60) greater odds of reutilization than did those with parents reporting no ACEs (Table 2). Resilience failed to reach statistical significance in the adjusted model (OR 0.86, 95% CI 0.70-1.07).

We found that high-risk parents (4 or more ACEs) had children with an increased odds of healthcare reutilization, suggesting intergenerational effects of ACEs. We did not find a similar effect relating to parental resilience. We also did not find an interaction between parental ACEs and resilience, suggesting that a parent’s reported degree of resilience does not modify the effect of ACEs on reutilization risk.

Parental adversity may be a risk factor for a child’s unanticipated reutilization. We previously demonstrated that parents with 4 or more ACEs have more coping difficulty than a parent with no ACEs after a child’s hospitalization.¹¹ It is possible that parents with high adversity may have poorer coping mechanisms when dealing with a stressful situation, such as a child’s hospitalization. This may have resulted in inequitable outcomes (eg, increased reutilization) for their children. Other studies have confirmed such an intergenerational effect of adversity, linking a parent’s ACEs with poor developmental, behavioral, and health outcomes in their children.^6,22,23 O’Malley et al showed an association of parental ACEs to current adversities,²⁴ such as insurance or housing concerns, that affect the entirety of the household, including children. In short, it appears that parental ACEs may be a compelling predictor of current childhood adversity.

Resilience buffers the negative effects of ACEs; however, we did not find significant associations between resilience and reutilization or an interaction between ACEs and resilience. The factors that may contribute to reutilization are complex. In our previous work, parental resilience was associated with coping difficulty after discharge; but again, did not interact with parental ACEs.¹¹ Here, we suggest that while resilience may buffer the negative effects of ACEs, that buffering may not affect the likelihood of reutilization. It is also possible that the BRS tool is of less relevance on how one handles the stress of a child’s hospitalization. While the BRS is one measure of resilience, there are many other relevant constructs to resilience, such as connection to social supports, that also may also contribute to risk of reutilization.²⁵

Reducing the stress of a hospitalization itself and promoting a safe transition from hospital to home is critical to improving child health outcomes. Our data here, and in our previous work, demonstrate that a history of adversity and one’s current coping ability may drive a parent’s response to a child’s hospitalization and affect their capacity to care for that child after hospital discharge.¹¹ Additional in-hospital supports like child life, behavioral health, or pastoral care could reduce the stress of the hospitalization while also building positive coping mechanisms.^26-29 A meta-analysis demonstrated that such coping interventions can help alleviate the stress of a hospitalization.³⁰ Hill et al demonstrated successful stress reduction in parents of hospitalized children using a “Coping Kit for Parents.”³¹ Further studies are warranted to understand which interventions are most effective for children and families and whether they could be more effectively deployed if the inpatient team knew more about parental ACEs.

Screening for parental ACEs could help to identify patients at highest risk for a poor transition to home. Therefore, screening for parental adversity in clinical settings, including inpatient settings, may be relevant and valuable.³² Additionally, by recognizing the high prevalence of ACEs in an inpatient setting, hospitals and healthcare organizations could be motivated to develop and enact trauma-informed approaches. A trauma-informed care approach recognizes the intersection of trauma with health and social problems. With this recognition, care teams can more sensitively address the trauma as they provide relevant services.³³ Trauma-informed care is a secondary public health prevention approach that would help team members identify the prevalence and effects of trauma via screening, recognize the signs of a maladaptive response to stress, and respond by integrating awareness of trauma into practice management.^28,34 Both the National Academy of Medicine and the Agency for Healthcare Research and Quality have called for such a trauma-informed approach in primary care.³⁵ In response, many healthcare organizations have developed trauma-informed practices to better address the needs of the populations they serve. For example, provider training on this approach has led to improved rapport in patient-provider relationships.³⁶

Although ACE awareness is a component of trauma-informed care, there are still limitations of the original ACE questionnaire developed by Felitti et al. The existing tool is not inclusive of all adversities a parent or child may face. Moreover, its focus is on past exposures and experiences and not current health-related social needs (eg, food insecurity) which have known linkages with a range of health outcomes and health disparities.³⁷ Additionally, the original ACE questionnaire was created as a population level tool and not as a screening tool. If used as a screening tool, providers may view the questions as too sensitive to ask, and parents may have difficulty responding to and understanding the relevance to their child’s care. Therefore, we suggest that more evidence is required to understand how to best adapt ACE questions into a screening processes that may be implemented in a medical setting.

More evidence is also needed to determine when and where such screening may be most useful. A primary care provider would be best equipped to screen caregivers for ACEs given their established relationship with parents and patients. Given the potential relevance of such information for inpatient care provision, information could then flow from primary care to the inpatient team. However, because not all patients have established primary care providers and only 4% of pediatricians screen for ACEs,³⁸ it is important for inpatient medical teams to understand their role in identifying and addressing ACEs during hospital stays. Development of a screening tool, with input from all stakeholders—including parents—that is valid and feasible for use in a pediatric inpatient setting would be an important step forward. This tool should be paired with training in how to discuss these topics in a trauma-informed, nonjudgmental, empathic manner. We see this as a way in which providers can more effectively elicit an accurate response while simultaneously educating parents on the relevance of such sensitive topics during an acute hospital stay. We also recommend that screening should always be paired with response capabilities that connect those who screen positive with resources that could help them to navigate the stress experienced during and after a child’s hospitalization. Furthermore, communication with primary care providers about parents that screen positive should be integrated into the transition process.

This work has several limitations. First, our study was a part of randomized controlled trials conducted in one academic setting, which thereby limits generalizability. For example, we limited our cohort to those who were English-speaking patients only. This may bias our results because respondents with limited English proficiency may have different risk profiles than their English-speaking peers. In addition, the administration of the both the ACE and resilience questionnaires occurred during an acutely stressful period, which may influence how a parent responds to these questions. Also, both of the surveys are self-reported by parents, which may be susceptible to memory and response biases. Relatedly, we had a high number of nonrespondents, particularly to the ACE questionnaire. Our results are therefore only relevant to those who chose to respond and cannot be applied to nonrespondents. Further work assessing why one does or does not respond to such sensitive questions is an important area for future inquiry. Lastly, our cohort had limited medical complexity; future studies may consider links between parental ACEs (and resilience) and morbidity experienced by children with medical complexity.

Parents history of adversity is linked to their children’s unanticipated healthcare reutilization after a hospital discharge. Screening for parental stressors during a hospitalization may be an important first step to connecting parents and children to evidence-based interventions capable of mitigating the stress of hospitalization and promoting better, more seamless transitions from hospital to home.

Group Members: The following H2O members are nonauthor contributors: JoAnne Bachus, BSN, RN; Monica Borell, BSN, RN; Lenisa V Chang, MA, PhD; Patricia Crawford, RN; Sarah Ferris, BA; Jennifer Gold, BSN, RN; Judy A Heilman, BSN, RN; Jane C Khoury, PhD; Pierce Kuhnell, MS; Karen Lawley, BSN, RN; Margo Moore, MS, BSN, RN; Lynne O’Donnell, BSN, RN; Sarah Riddle, MD; Susan N Sherman, DPA; Angela M Statile, MD, MEd; Karen P Sullivan, BSN, RN; Heather Tubbs-Cooley, PhD, RN; Susan Wade-Murphy, MSN, RN; and Christine M White, MD, MAT.

The authors also thank David Keller, MD, for his guidance on the study.

The authors have no financial relationships or conflicts of interest relevant to this article to disclose.

Supported by funds from the Academic Pediatric Young Investigator Award (Dr A Shah) and the Patient-Centered Outcomes Research Institute Award (IHS-1306-0081, to Dr K Auger, Dr S Shah, Dr H Sucharew, Dr J Simmons), the National Institutes of Health (1K23AI112916, to Dr AF Beck), and the Agency for Healthcare Research and Quality (1K12HS026393-01, to Dr A Shah, K08-HS024735- 01A1, to Dr K Auger). Dr J Haney received Summer Undergraduate Research Fellowship funding through the Summer Undergraduate Research Fellowship at Cincinnati Children’s Hospital Medical Center.

All statements in this report, including findings and conclusions, are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute, its Board of Governors, or the Methodology Committee.

Adverse Childhood Experiences, or ACEs, include exposure to abuse, neglect, or household dysfunction (eg, having a parent who is mentally ill) as a child.¹ Exposure to ACEs affects health into adulthood, with a dose-response relationship between ACEs and a range of comorbidities.¹ Adults with 6 or more ACEs have a 20-year shorter life expectancy than do those with no ACEs.¹ Still, ACEs are static; once experienced, that experience cannot be undone. However, resilience, or positive adaptation in the context of adversity, can be protective, buffering the negative effects of ACEs.^2,3 Protective factors that promote resilience include social capital, such as positive relationships with caregivers and peers.³

With their clear link to health outcomes across the life-course, there is a movement for pediatricians to screen children for ACEs⁴ and to develop strategies that promote resilience in children, parents, and families. However, screening a child for adversity has challenges because younger children may not have experienced an adverse exposure, or they may be unable to voice their experiences. Studies have demonstrated that parental adversity, or ACEs, may be a marker for childhood adversity.^5,6 Biological models also support this potential intergenerational effect of ACEs. Chronic exposure to stress, including ACEs, results in elevated cortisol via a dysregulated hypothalamic-pituitary-adrenal axis, which results in chronic inflammation.⁷ This “toxic stress” is prolonged, severe in intensity, and can lead to epigenetic changes that may be passed on to the next generation.^8,9

Hospitalization of an ill child, and the transition to home after that hospitalization, is a stressful event for children and families.¹⁰ This stress may be relevant to parents that have a history of a high rate of ACEs or a current low degree of resilience. Our previous work demonstrated that, in the inpatient setting, parents with high ACEs (≥4) or low resilience have increased coping difficulty 14 days after their child’s hospital discharge.¹¹ Our objective here was to evaluate whether a parent’s ACEs and/or resilience would also be associated with that child’s likelihood of reutilization. We hypothesized that more parental ACEs and/or lower parental resilience would be associated with revisits the emergency room, urgent care, or hospital readmissions.

Participants and Study Design

We conducted a prospective cohort study of parents of hospitalized children recruited from the “Hospital-to-Home Outcomes” Studies (H2O I and H2O II).^12,13 H2O I and II were prospective, single-center, randomized controlled trials designed to determine the effectiveness of either a nurse-led transitional home visit (H2O I) or telephone call (H2O II) on 30-day unplanned healthcare reutilization. The trials and this study were approved by the Cincinnati Children’s Institutional Review Board. All parents provided written informed consent.

Details of H2O I and II recruitment and design have been described previously.^12,13 Briefly, children were eligible for inclusion in either study if they were admitted to our institution’s general Hospital Medicine or the Hospital Medicine Complex Care Services; for H2O I, children hospitalized on the Neurology and Neurosurgery services were also eligible.^12,13 Patients were excluded if they were discharged to a residential facility, if they lived outside the home healthcare nurse service area, if they were eligible for skilled home healthcare services (eg, intravenous antibiotics), or if the participating caregiver was non-English speaking.^12,13 In H2O I, families were randomized either to receive a single nurse home visit within 96 hours of discharge or standard of care. In H2O II, families enrolled were randomized to receive a telephone call by a nurse within 96 hours of discharge or standard of care. As we have previously published, randomization in both trials successfully balanced the intervention and control arms with respect to key demographic characteristics.^12,13 For the analyses presented here, we focused on a subset of caregivers 18 years and older whose children were enrolled in either H2O I or II between August 2015 and October 2016. In both H2O trials, face-to-face and paper-based questionnaires were completed by parents during the index hospitalization.

Outcome and Predictors

Our primary outcome was unanticipated healthcare reutilization defined as return to the emergency room, urgent care, or unplanned readmission within 30 days of hospital discharge, consistent with the H2O trials. This was measured using the primary institution’s administrative data supplemented by a utilization database shared across regional hospitals.¹⁴ Readmissions were identified as “unplanned” using a previously validated algorithm,¹⁵ and treated as a dichotomous yes/no variable.

Our primary predictors were parental ACEs and resilience (see Appendix Tables). The ACE questionnaire addresses abuse, neglect, and household dysfunction in the first 18 years of life.¹ It is composed of 10 questions, each with a yes/no response.¹ We defined parents as low (ACE 0), moderate (ACE 1-3), or high (ACE ≥4) risk a priori because previous literature has described poor outcomes in adults with 4 or more ACEs.¹⁶

Given the sensitive nature of the questions, respondents independently completed the ACE questionnaire on paper instead of via the face-to-face survey. Respondents returned the completed questionnaire to the research assistant in a sealed envelope. All families received educational information on relevant hospital and community-based resources (eg, social work).

Parental resilience was measured using the Brief Resilience Scale (BRS). The BRS is 6 items, each on a 5-point Likert scale. Responses were averaged, providing a total score of 1-5; higher scores are representative of higher resilience.¹⁷ We treated the BRS score as a continuous variable. BRS has been used in clinical settings; it has demonstrated positive correlation with social support and negative correlation with fatigue.¹⁷ Parents answered BRS questions during the index pediatric hospitalization in a face-to-face interview.

Parent and Child Characteristics

Parent and child sociodemographic variables were also obtained during the face-to-face interview. Parental variables included age, gender, educational attainment, household income, employment status, and financial and social strain.¹¹ Educational attainment was analyzed in 2 categories—high school or less vs more than high school—because most discharge instructions are written at a high school reading level.¹⁸ Parents reported their annual household income in the following categories: <$15,000; $15,000-$29,999; $30,000-$44,999; $45,000-$59,999; $60,000-$89,999; $90,000-$119,999; ≥$120,000. Employment was dichotomized as not employed/student vs any employment. Financial and social strain were assessed using a series of 9 previously described questions.¹⁹ These questions assessed, via self-report, a family’s ability to make ends meet, ability to pay rent/mortgage or utilities, need to move in with others because of financial reasons, and ability to borrow money if needed, as well as home ownership and parental marital status.^15,19 Strain questions were all dichotomous (yes/no, single/not single). A composite variable was then constructed that categorized those reporting no strain items, 1 to 2 items, 3 to 4 items, and 5 or more items.²⁰

Child variables included race, ethnicity, age, primary care access,²¹ payer, and H2O treatment arm. Race categories were white/Caucasian, black/African American, American Indian or Alaskan Native, Asian or Pacific Islander, and other; ethnicity categories were Hispanic/Latino, non-Hispanic/Latino, and unknown. Given relatively low numbers of children reported to be Hispanic/Latino, we combined race and ethnicity into a single variable, categorized as non-Hispanic/white, non-Hispanic/black, and multiracial/Hispanic/other. Primary care access was assessed using the access subscale to the Parent’s Perception of Primary Care questionnaire. This includes assessment of a family’s ability to travel to their doctor, to see their doctor for routine or sick care, and to get help or advice on evenings or weekends. Scores were categorized as always adequate, almost always adequate, or sometimes/never adequate.²¹ Payer was dichotomized to private or public/self-pay.

Statistical Analyses

We examined the distribution of outcomes, predictors, and covariates. We compared sociodemographic characteristics of those respondents and nonrespondents to the ACE screen using the chi-square test for categorical variables or the t test for continuous variables. We used logistic regression to assess for associations between the independent variables of interest and reutilization, adjusting for potential confounders. To build our adjusted, multivariable model, we decided a priori to include child race/ethnicity, primary care access, financial and social strain, and trial treatment arm. We treated the H2O I control group as the referent group. Other covariates considered for inclusion were caregiver education, household income, employment, and payer. These were included in multivariable models if bivariate associations were significant at the P < .1 level. We assessed an ACE-by-resilience interaction term because we hypothesized that those with more ACEs and lower resilience may have more reutilization outcomes than parents with fewer ACEs and higher resilience. We also evaluated interaction terms between trial arm assignment and predictors to assess effects that may be introduced by the randomization. Predictors in the final logistic regression model were significant at the P < .05 level. Logistic regression assumption of little or no multicollinearity among the independent variables was verified in the final models. All analyses were performed with Stata v16 (Stata Corp, College Station, Texas).

There were a total of 1,787 parent-child dyads enrolled in the H2O I and II during the study period; 1,320 parents (74%) completed the ACE questionnaire and were included the analysis. Included parents were primarily female and employed, as well as educated beyond high school (Table 1). Overall, 64% reported one or more ACEs (range 0 to 9); 45% reported 1to 3, and 19% reported 4 or more ACEs. The most commonly reported ACEs were divorce (n = 573, 43%), exposure to alcoholism (n = 306, 23%), and exposure to mental illness (n = 281, 21%; Figure 1). Parents had a mean BRS score of 3.97 (range 1.17-5.00), with the distribution shown in Figure 2.

Of the 1,320 included patients, the average length of stay was 2.5 days, and 82% of hospitalizations were caused by acute medical issues (eg, bronchiolitis). A total of 211 children experienced a reutilization event within 30 days of discharge. In bivariate analysis, children with parents with 4 or more ACEs had a 2.02-times (95% CI 1.35-3.02) higher odds of experiencing a reutilization event than did those with parents reporting no ACEs. Parents with higher resilience scores had children with a lower odds of reutilization (odds ratio [OR] 0.77 95% CI 0.63-0.95).

In addition to our a priori variables, parental education, employment, and insurance met our significance threshold for inclusion in the multivariable model. The ACE-by-resilience interaction term was not significant and not included in the model. Similarly, there was no significant interaction between ACE and resilience and H2O treatment arm; the interaction terms were not included in the final adjusted model, but treatment arm assignment was kept as a covariate. A total of 1,292 children, out of the 1,320 respondents, remained in the final multivariable model; the excluded 28 had incomplete covariate data but were not otherwise different. In this final adjusted model, children with parents reporting 4 or more ACEs had a 1.69-times (95% CI 1.11-2.60) greater odds of reutilization than did those with parents reporting no ACEs (Table 2). Resilience failed to reach statistical significance in the adjusted model (OR 0.86, 95% CI 0.70-1.07).

We found that high-risk parents (4 or more ACEs) had children with an increased odds of healthcare reutilization, suggesting intergenerational effects of ACEs. We did not find a similar effect relating to parental resilience. We also did not find an interaction between parental ACEs and resilience, suggesting that a parent’s reported degree of resilience does not modify the effect of ACEs on reutilization risk.

Parental adversity may be a risk factor for a child’s unanticipated reutilization. We previously demonstrated that parents with 4 or more ACEs have more coping difficulty than a parent with no ACEs after a child’s hospitalization.¹¹ It is possible that parents with high adversity may have poorer coping mechanisms when dealing with a stressful situation, such as a child’s hospitalization. This may have resulted in inequitable outcomes (eg, increased reutilization) for their children. Other studies have confirmed such an intergenerational effect of adversity, linking a parent’s ACEs with poor developmental, behavioral, and health outcomes in their children.^6,22,23 O’Malley et al showed an association of parental ACEs to current adversities,²⁴ such as insurance or housing concerns, that affect the entirety of the household, including children. In short, it appears that parental ACEs may be a compelling predictor of current childhood adversity.

Resilience buffers the negative effects of ACEs; however, we did not find significant associations between resilience and reutilization or an interaction between ACEs and resilience. The factors that may contribute to reutilization are complex. In our previous work, parental resilience was associated with coping difficulty after discharge; but again, did not interact with parental ACEs.¹¹ Here, we suggest that while resilience may buffer the negative effects of ACEs, that buffering may not affect the likelihood of reutilization. It is also possible that the BRS tool is of less relevance on how one handles the stress of a child’s hospitalization. While the BRS is one measure of resilience, there are many other relevant constructs to resilience, such as connection to social supports, that also may also contribute to risk of reutilization.²⁵

Reducing the stress of a hospitalization itself and promoting a safe transition from hospital to home is critical to improving child health outcomes. Our data here, and in our previous work, demonstrate that a history of adversity and one’s current coping ability may drive a parent’s response to a child’s hospitalization and affect their capacity to care for that child after hospital discharge.¹¹ Additional in-hospital supports like child life, behavioral health, or pastoral care could reduce the stress of the hospitalization while also building positive coping mechanisms.^26-29 A meta-analysis demonstrated that such coping interventions can help alleviate the stress of a hospitalization.³⁰ Hill et al demonstrated successful stress reduction in parents of hospitalized children using a “Coping Kit for Parents.”³¹ Further studies are warranted to understand which interventions are most effective for children and families and whether they could be more effectively deployed if the inpatient team knew more about parental ACEs.

Screening for parental ACEs could help to identify patients at highest risk for a poor transition to home. Therefore, screening for parental adversity in clinical settings, including inpatient settings, may be relevant and valuable.³² Additionally, by recognizing the high prevalence of ACEs in an inpatient setting, hospitals and healthcare organizations could be motivated to develop and enact trauma-informed approaches. A trauma-informed care approach recognizes the intersection of trauma with health and social problems. With this recognition, care teams can more sensitively address the trauma as they provide relevant services.³³ Trauma-informed care is a secondary public health prevention approach that would help team members identify the prevalence and effects of trauma via screening, recognize the signs of a maladaptive response to stress, and respond by integrating awareness of trauma into practice management.^28,34 Both the National Academy of Medicine and the Agency for Healthcare Research and Quality have called for such a trauma-informed approach in primary care.³⁵ In response, many healthcare organizations have developed trauma-informed practices to better address the needs of the populations they serve. For example, provider training on this approach has led to improved rapport in patient-provider relationships.³⁶

Although ACE awareness is a component of trauma-informed care, there are still limitations of the original ACE questionnaire developed by Felitti et al. The existing tool is not inclusive of all adversities a parent or child may face. Moreover, its focus is on past exposures and experiences and not current health-related social needs (eg, food insecurity) which have known linkages with a range of health outcomes and health disparities.³⁷ Additionally, the original ACE questionnaire was created as a population level tool and not as a screening tool. If used as a screening tool, providers may view the questions as too sensitive to ask, and parents may have difficulty responding to and understanding the relevance to their child’s care. Therefore, we suggest that more evidence is required to understand how to best adapt ACE questions into a screening processes that may be implemented in a medical setting.

More evidence is also needed to determine when and where such screening may be most useful. A primary care provider would be best equipped to screen caregivers for ACEs given their established relationship with parents and patients. Given the potential relevance of such information for inpatient care provision, information could then flow from primary care to the inpatient team. However, because not all patients have established primary care providers and only 4% of pediatricians screen for ACEs,³⁸ it is important for inpatient medical teams to understand their role in identifying and addressing ACEs during hospital stays. Development of a screening tool, with input from all stakeholders—including parents—that is valid and feasible for use in a pediatric inpatient setting would be an important step forward. This tool should be paired with training in how to discuss these topics in a trauma-informed, nonjudgmental, empathic manner. We see this as a way in which providers can more effectively elicit an accurate response while simultaneously educating parents on the relevance of such sensitive topics during an acute hospital stay. We also recommend that screening should always be paired with response capabilities that connect those who screen positive with resources that could help them to navigate the stress experienced during and after a child’s hospitalization. Furthermore, communication with primary care providers about parents that screen positive should be integrated into the transition process.

This work has several limitations. First, our study was a part of randomized controlled trials conducted in one academic setting, which thereby limits generalizability. For example, we limited our cohort to those who were English-speaking patients only. This may bias our results because respondents with limited English proficiency may have different risk profiles than their English-speaking peers. In addition, the administration of the both the ACE and resilience questionnaires occurred during an acutely stressful period, which may influence how a parent responds to these questions. Also, both of the surveys are self-reported by parents, which may be susceptible to memory and response biases. Relatedly, we had a high number of nonrespondents, particularly to the ACE questionnaire. Our results are therefore only relevant to those who chose to respond and cannot be applied to nonrespondents. Further work assessing why one does or does not respond to such sensitive questions is an important area for future inquiry. Lastly, our cohort had limited medical complexity; future studies may consider links between parental ACEs (and resilience) and morbidity experienced by children with medical complexity.

Parents history of adversity is linked to their children’s unanticipated healthcare reutilization after a hospital discharge. Screening for parental stressors during a hospitalization may be an important first step to connecting parents and children to evidence-based interventions capable of mitigating the stress of hospitalization and promoting better, more seamless transitions from hospital to home.

Group Members: The following H2O members are nonauthor contributors: JoAnne Bachus, BSN, RN; Monica Borell, BSN, RN; Lenisa V Chang, MA, PhD; Patricia Crawford, RN; Sarah Ferris, BA; Jennifer Gold, BSN, RN; Judy A Heilman, BSN, RN; Jane C Khoury, PhD; Pierce Kuhnell, MS; Karen Lawley, BSN, RN; Margo Moore, MS, BSN, RN; Lynne O’Donnell, BSN, RN; Sarah Riddle, MD; Susan N Sherman, DPA; Angela M Statile, MD, MEd; Karen P Sullivan, BSN, RN; Heather Tubbs-Cooley, PhD, RN; Susan Wade-Murphy, MSN, RN; and Christine M White, MD, MAT.

The authors also thank David Keller, MD, for his guidance on the study.

The authors have no financial relationships or conflicts of interest relevant to this article to disclose.

Supported by funds from the Academic Pediatric Young Investigator Award (Dr A Shah) and the Patient-Centered Outcomes Research Institute Award (IHS-1306-0081, to Dr K Auger, Dr S Shah, Dr H Sucharew, Dr J Simmons), the National Institutes of Health (1K23AI112916, to Dr AF Beck), and the Agency for Healthcare Research and Quality (1K12HS026393-01, to Dr A Shah, K08-HS024735- 01A1, to Dr K Auger). Dr J Haney received Summer Undergraduate Research Fellowship funding through the Summer Undergraduate Research Fellowship at Cincinnati Children’s Hospital Medical Center.

All statements in this report, including findings and conclusions, are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute, its Board of Governors, or the Methodology Committee.

Adverse Childhood Experiences, or ACEs, include exposure to abuse, neglect, or household dysfunction (eg, having a parent who is mentally ill) as a child.¹ Exposure to ACEs affects health into adulthood, with a dose-response relationship between ACEs and a range of comorbidities.¹ Adults with 6 or more ACEs have a 20-year shorter life expectancy than do those with no ACEs.¹ Still, ACEs are static; once experienced, that experience cannot be undone. However, resilience, or positive adaptation in the context of adversity, can be protective, buffering the negative effects of ACEs.^2,3 Protective factors that promote resilience include social capital, such as positive relationships with caregivers and peers.³

With their clear link to health outcomes across the life-course, there is a movement for pediatricians to screen children for ACEs⁴ and to develop strategies that promote resilience in children, parents, and families. However, screening a child for adversity has challenges because younger children may not have experienced an adverse exposure, or they may be unable to voice their experiences. Studies have demonstrated that parental adversity, or ACEs, may be a marker for childhood adversity.^5,6 Biological models also support this potential intergenerational effect of ACEs. Chronic exposure to stress, including ACEs, results in elevated cortisol via a dysregulated hypothalamic-pituitary-adrenal axis, which results in chronic inflammation.⁷ This “toxic stress” is prolonged, severe in intensity, and can lead to epigenetic changes that may be passed on to the next generation.^8,9

Hospitalization of an ill child, and the transition to home after that hospitalization, is a stressful event for children and families.¹⁰ This stress may be relevant to parents that have a history of a high rate of ACEs or a current low degree of resilience. Our previous work demonstrated that, in the inpatient setting, parents with high ACEs (≥4) or low resilience have increased coping difficulty 14 days after their child’s hospital discharge.¹¹ Our objective here was to evaluate whether a parent’s ACEs and/or resilience would also be associated with that child’s likelihood of reutilization. We hypothesized that more parental ACEs and/or lower parental resilience would be associated with revisits the emergency room, urgent care, or hospital readmissions.

Participants and Study Design

We conducted a prospective cohort study of parents of hospitalized children recruited from the “Hospital-to-Home Outcomes” Studies (H2O I and H2O II).^12,13 H2O I and II were prospective, single-center, randomized controlled trials designed to determine the effectiveness of either a nurse-led transitional home visit (H2O I) or telephone call (H2O II) on 30-day unplanned healthcare reutilization. The trials and this study were approved by the Cincinnati Children’s Institutional Review Board. All parents provided written informed consent.

Details of H2O I and II recruitment and design have been described previously.^12,13 Briefly, children were eligible for inclusion in either study if they were admitted to our institution’s general Hospital Medicine or the Hospital Medicine Complex Care Services; for H2O I, children hospitalized on the Neurology and Neurosurgery services were also eligible.^12,13 Patients were excluded if they were discharged to a residential facility, if they lived outside the home healthcare nurse service area, if they were eligible for skilled home healthcare services (eg, intravenous antibiotics), or if the participating caregiver was non-English speaking.^12,13 In H2O I, families were randomized either to receive a single nurse home visit within 96 hours of discharge or standard of care. In H2O II, families enrolled were randomized to receive a telephone call by a nurse within 96 hours of discharge or standard of care. As we have previously published, randomization in both trials successfully balanced the intervention and control arms with respect to key demographic characteristics.^12,13 For the analyses presented here, we focused on a subset of caregivers 18 years and older whose children were enrolled in either H2O I or II between August 2015 and October 2016. In both H2O trials, face-to-face and paper-based questionnaires were completed by parents during the index hospitalization.

Outcome and Predictors

Our primary outcome was unanticipated healthcare reutilization defined as return to the emergency room, urgent care, or unplanned readmission within 30 days of hospital discharge, consistent with the H2O trials. This was measured using the primary institution’s administrative data supplemented by a utilization database shared across regional hospitals.¹⁴ Readmissions were identified as “unplanned” using a previously validated algorithm,¹⁵ and treated as a dichotomous yes/no variable.

Our primary predictors were parental ACEs and resilience (see Appendix Tables). The ACE questionnaire addresses abuse, neglect, and household dysfunction in the first 18 years of life.¹ It is composed of 10 questions, each with a yes/no response.¹ We defined parents as low (ACE 0), moderate (ACE 1-3), or high (ACE ≥4) risk a priori because previous literature has described poor outcomes in adults with 4 or more ACEs.¹⁶

Given the sensitive nature of the questions, respondents independently completed the ACE questionnaire on paper instead of via the face-to-face survey. Respondents returned the completed questionnaire to the research assistant in a sealed envelope. All families received educational information on relevant hospital and community-based resources (eg, social work).

Parental resilience was measured using the Brief Resilience Scale (BRS). The BRS is 6 items, each on a 5-point Likert scale. Responses were averaged, providing a total score of 1-5; higher scores are representative of higher resilience.¹⁷ We treated the BRS score as a continuous variable. BRS has been used in clinical settings; it has demonstrated positive correlation with social support and negative correlation with fatigue.¹⁷ Parents answered BRS questions during the index pediatric hospitalization in a face-to-face interview.

Parent and Child Characteristics

Parent and child sociodemographic variables were also obtained during the face-to-face interview. Parental variables included age, gender, educational attainment, household income, employment status, and financial and social strain.¹¹ Educational attainment was analyzed in 2 categories—high school or less vs more than high school—because most discharge instructions are written at a high school reading level.¹⁸ Parents reported their annual household income in the following categories: <$15,000; $15,000-$29,999; $30,000-$44,999; $45,000-$59,999; $60,000-$89,999; $90,000-$119,999; ≥$120,000. Employment was dichotomized as not employed/student vs any employment. Financial and social strain were assessed using a series of 9 previously described questions.¹⁹ These questions assessed, via self-report, a family’s ability to make ends meet, ability to pay rent/mortgage or utilities, need to move in with others because of financial reasons, and ability to borrow money if needed, as well as home ownership and parental marital status.^15,19 Strain questions were all dichotomous (yes/no, single/not single). A composite variable was then constructed that categorized those reporting no strain items, 1 to 2 items, 3 to 4 items, and 5 or more items.²⁰

Child variables included race, ethnicity, age, primary care access,²¹ payer, and H2O treatment arm. Race categories were white/Caucasian, black/African American, American Indian or Alaskan Native, Asian or Pacific Islander, and other; ethnicity categories were Hispanic/Latino, non-Hispanic/Latino, and unknown. Given relatively low numbers of children reported to be Hispanic/Latino, we combined race and ethnicity into a single variable, categorized as non-Hispanic/white, non-Hispanic/black, and multiracial/Hispanic/other. Primary care access was assessed using the access subscale to the Parent’s Perception of Primary Care questionnaire. This includes assessment of a family’s ability to travel to their doctor, to see their doctor for routine or sick care, and to get help or advice on evenings or weekends. Scores were categorized as always adequate, almost always adequate, or sometimes/never adequate.²¹ Payer was dichotomized to private or public/self-pay.

Statistical Analyses

We examined the distribution of outcomes, predictors, and covariates. We compared sociodemographic characteristics of those respondents and nonrespondents to the ACE screen using the chi-square test for categorical variables or the t test for continuous variables. We used logistic regression to assess for associations between the independent variables of interest and reutilization, adjusting for potential confounders. To build our adjusted, multivariable model, we decided a priori to include child race/ethnicity, primary care access, financial and social strain, and trial treatment arm. We treated the H2O I control group as the referent group. Other covariates considered for inclusion were caregiver education, household income, employment, and payer. These were included in multivariable models if bivariate associations were significant at the P < .1 level. We assessed an ACE-by-resilience interaction term because we hypothesized that those with more ACEs and lower resilience may have more reutilization outcomes than parents with fewer ACEs and higher resilience. We also evaluated interaction terms between trial arm assignment and predictors to assess effects that may be introduced by the randomization. Predictors in the final logistic regression model were significant at the P < .05 level. Logistic regression assumption of little or no multicollinearity among the independent variables was verified in the final models. All analyses were performed with Stata v16 (Stata Corp, College Station, Texas).

There were a total of 1,787 parent-child dyads enrolled in the H2O I and II during the study period; 1,320 parents (74%) completed the ACE questionnaire and were included the analysis. Included parents were primarily female and employed, as well as educated beyond high school (Table 1). Overall, 64% reported one or more ACEs (range 0 to 9); 45% reported 1to 3, and 19% reported 4 or more ACEs. The most commonly reported ACEs were divorce (n = 573, 43%), exposure to alcoholism (n = 306, 23%), and exposure to mental illness (n = 281, 21%; Figure 1). Parents had a mean BRS score of 3.97 (range 1.17-5.00), with the distribution shown in Figure 2.

Of the 1,320 included patients, the average length of stay was 2.5 days, and 82% of hospitalizations were caused by acute medical issues (eg, bronchiolitis). A total of 211 children experienced a reutilization event within 30 days of discharge. In bivariate analysis, children with parents with 4 or more ACEs had a 2.02-times (95% CI 1.35-3.02) higher odds of experiencing a reutilization event than did those with parents reporting no ACEs. Parents with higher resilience scores had children with a lower odds of reutilization (odds ratio [OR] 0.77 95% CI 0.63-0.95).

In addition to our a priori variables, parental education, employment, and insurance met our significance threshold for inclusion in the multivariable model. The ACE-by-resilience interaction term was not significant and not included in the model. Similarly, there was no significant interaction between ACE and resilience and H2O treatment arm; the interaction terms were not included in the final adjusted model, but treatment arm assignment was kept as a covariate. A total of 1,292 children, out of the 1,320 respondents, remained in the final multivariable model; the excluded 28 had incomplete covariate data but were not otherwise different. In this final adjusted model, children with parents reporting 4 or more ACEs had a 1.69-times (95% CI 1.11-2.60) greater odds of reutilization than did those with parents reporting no ACEs (Table 2). Resilience failed to reach statistical significance in the adjusted model (OR 0.86, 95% CI 0.70-1.07).

We found that high-risk parents (4 or more ACEs) had children with an increased odds of healthcare reutilization, suggesting intergenerational effects of ACEs. We did not find a similar effect relating to parental resilience. We also did not find an interaction between parental ACEs and resilience, suggesting that a parent’s reported degree of resilience does not modify the effect of ACEs on reutilization risk.

Parental adversity may be a risk factor for a child’s unanticipated reutilization. We previously demonstrated that parents with 4 or more ACEs have more coping difficulty than a parent with no ACEs after a child’s hospitalization.¹¹ It is possible that parents with high adversity may have poorer coping mechanisms when dealing with a stressful situation, such as a child’s hospitalization. This may have resulted in inequitable outcomes (eg, increased reutilization) for their children. Other studies have confirmed such an intergenerational effect of adversity, linking a parent’s ACEs with poor developmental, behavioral, and health outcomes in their children.^6,22,23 O’Malley et al showed an association of parental ACEs to current adversities,²⁴ such as insurance or housing concerns, that affect the entirety of the household, including children. In short, it appears that parental ACEs may be a compelling predictor of current childhood adversity.

Resilience buffers the negative effects of ACEs; however, we did not find significant associations between resilience and reutilization or an interaction between ACEs and resilience. The factors that may contribute to reutilization are complex. In our previous work, parental resilience was associated with coping difficulty after discharge; but again, did not interact with parental ACEs.¹¹ Here, we suggest that while resilience may buffer the negative effects of ACEs, that buffering may not affect the likelihood of reutilization. It is also possible that the BRS tool is of less relevance on how one handles the stress of a child’s hospitalization. While the BRS is one measure of resilience, there are many other relevant constructs to resilience, such as connection to social supports, that also may also contribute to risk of reutilization.²⁵

Reducing the stress of a hospitalization itself and promoting a safe transition from hospital to home is critical to improving child health outcomes. Our data here, and in our previous work, demonstrate that a history of adversity and one’s current coping ability may drive a parent’s response to a child’s hospitalization and affect their capacity to care for that child after hospital discharge.¹¹ Additional in-hospital supports like child life, behavioral health, or pastoral care could reduce the stress of the hospitalization while also building positive coping mechanisms.^26-29 A meta-analysis demonstrated that such coping interventions can help alleviate the stress of a hospitalization.³⁰ Hill et al demonstrated successful stress reduction in parents of hospitalized children using a “Coping Kit for Parents.”³¹ Further studies are warranted to understand which interventions are most effective for children and families and whether they could be more effectively deployed if the inpatient team knew more about parental ACEs.

Screening for parental ACEs could help to identify patients at highest risk for a poor transition to home. Therefore, screening for parental adversity in clinical settings, including inpatient settings, may be relevant and valuable.³² Additionally, by recognizing the high prevalence of ACEs in an inpatient setting, hospitals and healthcare organizations could be motivated to develop and enact trauma-informed approaches. A trauma-informed care approach recognizes the intersection of trauma with health and social problems. With this recognition, care teams can more sensitively address the trauma as they provide relevant services.³³ Trauma-informed care is a secondary public health prevention approach that would help team members identify the prevalence and effects of trauma via screening, recognize the signs of a maladaptive response to stress, and respond by integrating awareness of trauma into practice management.^28,34 Both the National Academy of Medicine and the Agency for Healthcare Research and Quality have called for such a trauma-informed approach in primary care.³⁵ In response, many healthcare organizations have developed trauma-informed practices to better address the needs of the populations they serve. For example, provider training on this approach has led to improved rapport in patient-provider relationships.³⁶

Although ACE awareness is a component of trauma-informed care, there are still limitations of the original ACE questionnaire developed by Felitti et al. The existing tool is not inclusive of all adversities a parent or child may face. Moreover, its focus is on past exposures and experiences and not current health-related social needs (eg, food insecurity) which have known linkages with a range of health outcomes and health disparities.³⁷ Additionally, the original ACE questionnaire was created as a population level tool and not as a screening tool. If used as a screening tool, providers may view the questions as too sensitive to ask, and parents may have difficulty responding to and understanding the relevance to their child’s care. Therefore, we suggest that more evidence is required to understand how to best adapt ACE questions into a screening processes that may be implemented in a medical setting.

More evidence is also needed to determine when and where such screening may be most useful. A primary care provider would be best equipped to screen caregivers for ACEs given their established relationship with parents and patients. Given the potential relevance of such information for inpatient care provision, information could then flow from primary care to the inpatient team. However, because not all patients have established primary care providers and only 4% of pediatricians screen for ACEs,³⁸ it is important for inpatient medical teams to understand their role in identifying and addressing ACEs during hospital stays. Development of a screening tool, with input from all stakeholders—including parents—that is valid and feasible for use in a pediatric inpatient setting would be an important step forward. This tool should be paired with training in how to discuss these topics in a trauma-informed, nonjudgmental, empathic manner. We see this as a way in which providers can more effectively elicit an accurate response while simultaneously educating parents on the relevance of such sensitive topics during an acute hospital stay. We also recommend that screening should always be paired with response capabilities that connect those who screen positive with resources that could help them to navigate the stress experienced during and after a child’s hospitalization. Furthermore, communication with primary care providers about parents that screen positive should be integrated into the transition process.

This work has several limitations. First, our study was a part of randomized controlled trials conducted in one academic setting, which thereby limits generalizability. For example, we limited our cohort to those who were English-speaking patients only. This may bias our results because respondents with limited English proficiency may have different risk profiles than their English-speaking peers. In addition, the administration of the both the ACE and resilience questionnaires occurred during an acutely stressful period, which may influence how a parent responds to these questions. Also, both of the surveys are self-reported by parents, which may be susceptible to memory and response biases. Relatedly, we had a high number of nonrespondents, particularly to the ACE questionnaire. Our results are therefore only relevant to those who chose to respond and cannot be applied to nonrespondents. Further work assessing why one does or does not respond to such sensitive questions is an important area for future inquiry. Lastly, our cohort had limited medical complexity; future studies may consider links between parental ACEs (and resilience) and morbidity experienced by children with medical complexity.

Parents history of adversity is linked to their children’s unanticipated healthcare reutilization after a hospital discharge. Screening for parental stressors during a hospitalization may be an important first step to connecting parents and children to evidence-based interventions capable of mitigating the stress of hospitalization and promoting better, more seamless transitions from hospital to home.

Group Members: The following H2O members are nonauthor contributors: JoAnne Bachus, BSN, RN; Monica Borell, BSN, RN; Lenisa V Chang, MA, PhD; Patricia Crawford, RN; Sarah Ferris, BA; Jennifer Gold, BSN, RN; Judy A Heilman, BSN, RN; Jane C Khoury, PhD; Pierce Kuhnell, MS; Karen Lawley, BSN, RN; Margo Moore, MS, BSN, RN; Lynne O’Donnell, BSN, RN; Sarah Riddle, MD; Susan N Sherman, DPA; Angela M Statile, MD, MEd; Karen P Sullivan, BSN, RN; Heather Tubbs-Cooley, PhD, RN; Susan Wade-Murphy, MSN, RN; and Christine M White, MD, MAT.

The authors also thank David Keller, MD, for his guidance on the study.

The authors have no financial relationships or conflicts of interest relevant to this article to disclose.

Supported by funds from the Academic Pediatric Young Investigator Award (Dr A Shah) and the Patient-Centered Outcomes Research Institute Award (IHS-1306-0081, to Dr K Auger, Dr S Shah, Dr H Sucharew, Dr J Simmons), the National Institutes of Health (1K23AI112916, to Dr AF Beck), and the Agency for Healthcare Research and Quality (1K12HS026393-01, to Dr A Shah, K08-HS024735- 01A1, to Dr K Auger). Dr J Haney received Summer Undergraduate Research Fellowship funding through the Summer Undergraduate Research Fellowship at Cincinnati Children’s Hospital Medical Center.

All statements in this report, including findings and conclusions, are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute, its Board of Governors, or the Methodology Committee.

Alarms from bedside continuous physiologic monitors (CPMs) occur frequently in children’s hospitals and can lead to harm. Recent studies conducted in children’s hospitals have identified alarm rates of up to 152 alarms per patient per day outside of the intensive care unit,^1-3 with as few as 1% of alarms being considered clinically important.⁴ Excessive alarms have been linked to alarm fatigue, when providers become desensitized to and may miss alarms indicating impending patient deterioration. Alarm fatigue has been identified by national patient safety organizations as a patient safety concern given the risk of patient harm.^5-7 Despite these concerns, CPMs are routinely used: up to 48% of pediatric patients in nonintensive care units at children’s hospitals are monitored.²

Although the low number of alarms that receive responses has been well-described,^8,9 the reasons why clinicians do or do not respond to alarms are unclear. A study conducted in an adult perioperative unit noted prolonged nurse response times for patients with high alarm rates.¹⁰ A second study conducted in the pediatric inpatient setting demonstrated a dose-response effect and noted progressively prolonged nurse response times with increased rates of nonactionable alarms.^4,11 Findings from another study suggested that underlying factors are highly complex and may be a result of excessive alarms, clinician characteristics, and working conditions (eg, workload and unit noise level).¹² Evidence also suggests that humans have difficulty distinguishing the importance of alarms in situations where multiple alarm tones are used, a common scenario in hospitals.^13,14 Understanding the factors that contribute to clinicians responding or not responding to CPM alarms will be crucial for addressing this serious patient safety issue.

An enhanced understanding of why nurses respond to alarms in daily practice will inform intervention development and improvement work. In the long term, this information could help improve systems for monitoring pediatric inpatients that are less prone to issues with alarm fatigue. The objective of this qualitative study, which employed structured observation, was to describe how bedside nurses think about and act upon bedside monitor alarms in a general pediatric inpatient unit.

Study Design and Setting

This prospective observational study took place on a 48-bed hospital medicine unit at a large, freestanding children’s hospital with >650 beds and >19,000 annual admissions. General Electric (Little Chalfont, United Kingdom) physiologic monitors (models Dash 3000, 4000, and 5000) were used at the time of the study, and nurses could be notified of monitor alarms in four ways: First, an in-room auditory alarm sounds. Second, a light positioned above the door outside of each patient room blinks for alarms that are at a “warning” or “critical level” (eg ventricular tachycardia or low oxygen saturation). Third, audible alarms occur at the unit’s central monitoring station. Lastly, another staff member can notify the patient’s nurse via in-person conversion or secure smart phone communication. On the study unit, CPMs are initiated and discontinued through a physician order.

This study was reviewed and approved by the hospital’s institutional review board.

Study Population

We used a purposive recruitment strategy to enroll bedside nurses working on general hospital medicine units, stratified to ensure varying levels of experience and primary shifts (eg, day vs night). We planned to conduct approximately two observations with each participating nurse and to continue collecting data until we could no longer identify new insights in terms of responses to alarms (ie, thematic saturation¹⁵). Observations were targeted to cover times of day that coincided with increased rates of distraction. These times included just prior to and after the morning and evening change of shifts (7:00 am and 7:00 pm), during morning rounds (8:00 am-12:00 pm), and heavy admission times (12:00 pm-10:00 pm). After written informed consent, a nurse was eligible for observation during his/her shift if he/she was caring for at least one monitored patient. Enrolled nurses were made aware of the general study topic but were blinded to the study team’s hypotheses.

Data Sources

Prior to data collection, the research team, which consisted of physicians, bedside nurses, research coordinators, and a human factors expert, created a system for categorizing alarm responses. Categories for observed responses were based on the location and corresponding action taken. Initial categories were developed a priori from existing literature and expanded through input from the multidisciplinary study team, then vetted with bedside staff, and finally pilot tested through >4 hours of observations, thus producing the final categories. These categories were entered into a work-sampling program (WorkStudy by Quetech Ltd., Waterloo, Ontario, Canada) to facilitate quick data recording during observations.

The hospital uses a central alarm collection software (BedMasterEx by Anandic Medical Systems, Feuerthalen, Switzerland), which permitted the collection of date, time, trigger (eg, high heart rate), and level (eg, crisis, warning) of the generated CPM alarms. Alarms collected are based on thresholds preset at the bedside monitor. The central collection software does not differentiate between accurate (eg, correctly representing the physiologic state of the patient) and inaccurate alarms.

Observation Procedure

At the time of observation, nurse demographic information (eg, primary shift worked and years working as a nurse) was obtained. A brief preobservation questionnaire was administered to collect patient information (eg, age and diagnosis) and the nurses’ perspectives on the necessity of monitors for each monitored patient in his/her care.

The observer shadowed the nurse for a two-hour block of his/her shift. During this time, nurses were instructed to “think aloud” as they responded to alarms (eg, “I notice the oxygen saturation monitor alarming off, but the probe has fallen off”). A trained observer (AML or KMT) recorded responses verbalized by the nurse and his/her reaction by selecting the appropriate category using the work-sampling software. Data were also collected on the vital sign associated with the alarm (eg, heart rate). Moreover, the observer kept written notes to provide context for electronically recorded data. Alarms that were not verbalized by the nurse were not counted. Similarly, alarms that were noted outside of the room by the nurse were not classified by vital sign unless the nurse confirmed with the bedside monitor. Observers did not adjudicate the accuracy of the alarms. The session was stopped if monitors were discontinued during the observation period. Alarm data generated by the bedside monitor were pulled for each patient room after observations were completed.

Analysis

Descriptive statistics were used to assess the percentage of each nurse response category and each alarm type (eg, heart rate and respiratory rate). The observed alarm rate was calculated by taking the total number of observed alarms (ie, alarms noted by the nurse) divided by the total number of patient-hours observed. The monitor-generated alarm rate was calculated by taking the total number of alarms from the bedside-alarm generated data divided by the number of patient-hours observed.

Electronically recorded observations using the work-sampling program were cross-referenced with hand-written field notes to assess for any discrepancies or identify relevant events not captured by the program. Three study team members (AML, KMT, and ACS) reviewed each observation independently and compared field notes to ensure accurate categorization. Discrepancies were referred to the larger study group in cases of uncertainty.

Nine nurses had monitored patients during the available observations and participated in 19 observation sessions, which included 35 monitored patients for a total of 61.3 patient-hours of observation. Nurses were observed for a median of two times each (range 1-4). The median number of monitored patients during a single observation session was two (range 1-3). Observed nurses were female with a median of eight years of experience (range 0.5-26 years). Patients represented a broad range of age categories and were hospitalized with a variety of diagnoses (Table). Nurses, when queried at the start of the observation, felt that monitors were necessary for 29 (82.9%) of the observed patients given either patient condition or unit policy.

A total of 207 observed nurse responses to alarms occurred during the study period for a rate of 3.4 responses per patient per hour. Of the total number of responses, 45 (21.7%) were noted outside of a patient room, and in 15 (33.3%) the nurse chose to go to the room. The other 162 were recorded when the nurse was present in the room when the alarm activated. Of the 177 in-person nurse responses, 50 were related to a pulse oximetry alarm, 66 were related to a heart rate alarm, and 61 were related to a respiratory rate alarm. The most common observed in-person response to an alarm involved the nurse judging that no intervention was necessary (n = 152, 73.1%). Only 14 (7% of total responses) observed in-person responses involved a clinical intervention, such as suctioning or titrating supplemental oxygen. Findings are summarized in the Figure and describe nurse-verbalized reasons to further assess (or not) and then whether the nurse chose to take action (or not) after an alarm.

We characterized responses to physiologic monitor alarms by a group of nurses with a range of experience levels. We found that most nurse responses to alarms in continuously monitored general pediatric patients involved no intervention, and further assessment was often not conducted for alarms that occurred outside of the room if the nurse noted otherwise reassuring clinical context. Observed responses occurred for 36% of alarms during the study period when compared with bedside monitor-alarm generated data. Overall, only 14 clinical interventions were noted among the observed responses. Nurses noted that they felt the monitors were necessary for 82.9% of monitored patients because of the clinical context or because of unit policy.

Our study findings highlight some potential contradictions in the current widespread use of CPMs in general pediatric units and how clinicians respond to them in practice.² First, while nurses reported that monitors were necessary for most of their patients, participating nurses deemed few alarms clinically actionable and often chose not to further assess when they noted alarms outside of the room. This is in line with findings from prior studies suggesting that clinicians overvalue the contribution of monitoring systems to patient safety.^16,17 Second, while this finding occurred in a minority of the observations, the presence of family members at the patient’s bedside was cited by nurses as a rationale for whether they responded to alarms. While family members are capable of identifying safety issues,¹⁸ formal systems to engage them in patient safety and physiologic monitoring are lacking. Finally, clinical interventions or responses to the alerts of deteriorating patients, which best represented the original intent of CPMs, were rare and accounted for just 7% of the responses. Further work elucidating why physicians and nurses choose to use CPMs may be helpful to identify interventions to reduce inappropriate monitor use and highlight gaps in frontline staff knowledge about the benefits and risks of CPM use.

Our findings provide a novel understanding of previously observed phenomena, such as long response times or nonresponses in settings with high alarm rates.^4,10 Similar to that in a prior study conducted in the pediatric setting,¹¹ alarms with an observed response constituted a minority of the total alarms that occurred in our study. This finding has previously been attributed to mental fatigue, caregiver apathy, and desensitization.⁸ However, even though a minority of observed responses in our study included an intervention, the nurse had a rationale for why the alarm did or did not need a response. This behavior and the verbalized rationale indicate that in his/her opinion, not responding to the alarm was clinically appropriate. Study participants also reflected on the difficulties of responding to alarms given the monitor system setup, in which they may not always be capable of hearing alarms for their patients. Without data from nurses regarding the alarms that had no observed response, we can only speculate; however, based on our findings, each of these factors could contribute to nonresponse. Finally, while high numbers of false alarms have been posited as an underlying cause of alarm fatigue, we noted that a majority of nonresponse was reported to be related to other clinical factors. This relationship suggests that from the nurse’s perspective, a more applicable framework for understanding alarms would be based on clinical actionability⁴ over physiologic accuracy.

In total, our findings suggest that a multifaceted approach will be necessary to improve alarm response rates. These interventions should include adjusting parameters such that alarms are highly likely to indicate a need for intervention coupled with educational interventions addressing clinician knowledge of the alarm system and bias about the actionability of alarms may improve response rates. Changes in the monitoring system setup such that nurses can easily be notified when alarms occur may also be indicated, in addition to formally engaging patients and families around response to alarms. Although secondary notification systems (eg, alarms transmitted to individual clinician’s devices) are one solution, the utilization of these systems needs to be balanced with the risks of contributing to existing alarm fatigue and the need to appropriately tailor monitoring thresholds and strategies to patients.

Our study has several limitations. First, nurses may have responded in a way they perceive to be socially desirable, and studies using in-person observers are also prone to a Hawthorne-like effect,^19-21 where the nurse may have tried to respond more frequently to alarms than usual during observations. However, given that the majority of bedside alarms did not receive a response and a substantial number of responses involved no action, these effects were likely weak. Second, we were unable to assess which alarms were accurately reflecting the patient’s physiologic status and which were not; we were also unable to link observed alarm response to monitor-recorded alarms. Third, despite the use of silent observers and an actual, rather than a simulated, clinical setting, by virtue of the data collection method we likely captured a more deliberate thought process (so-called System 2 thinking)²² rather than the subconscious processes that may predominate when nurses respond to alarms in the course of clinical care (System 1 thinking).²² Despite this limitation, our study findings, which reflect a nurse’s in-the-moment thinking, remain relevant to guiding the improvement of monitoring systems, and the development of nurse-facing interventions and education. Finally, we studied a small, purposive sample of nurses at a single hospital. Our study sample impacts the generalizability of our results and precluded a detailed analysis of the effect of nurse- and patient-level variables.

We found that nurses often deemed that no response was necessary for CPM alarms. Nurses cited contextual factors, including the duration of alarms and the presence of other providers or parents in their decision-making. Few (7%) of the alarm responses in our study included a clinical intervention. The number of observed alarm responses constituted roughly a third of the alarms recorded by bedside CPMs during the study. This result supports concerns about the nurse’s capacity to hear and process all CPM alarms given system limitations and a heavy clinical workload. Subsequent steps should include staff education, reducing overall alarm rates with appropriate monitor use and actionable alarm thresholds, and ensuring that patient alarms are easily recognizable for frontline staff.

Disclosures

The authors have no conflicts of interest to disclose.

Funding

This work was supported by the Place Outcomes Research Award from the Cincinnati Children’s Research Foundation. Dr. Brady is supported by the Agency for Healthcare Research and Quality under Award Number K08HS23827. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality.

Alarms from bedside continuous physiologic monitors (CPMs) occur frequently in children’s hospitals and can lead to harm. Recent studies conducted in children’s hospitals have identified alarm rates of up to 152 alarms per patient per day outside of the intensive care unit,^1-3 with as few as 1% of alarms being considered clinically important.⁴ Excessive alarms have been linked to alarm fatigue, when providers become desensitized to and may miss alarms indicating impending patient deterioration. Alarm fatigue has been identified by national patient safety organizations as a patient safety concern given the risk of patient harm.^5-7 Despite these concerns, CPMs are routinely used: up to 48% of pediatric patients in nonintensive care units at children’s hospitals are monitored.²

Although the low number of alarms that receive responses has been well-described,^8,9 the reasons why clinicians do or do not respond to alarms are unclear. A study conducted in an adult perioperative unit noted prolonged nurse response times for patients with high alarm rates.¹⁰ A second study conducted in the pediatric inpatient setting demonstrated a dose-response effect and noted progressively prolonged nurse response times with increased rates of nonactionable alarms.^4,11 Findings from another study suggested that underlying factors are highly complex and may be a result of excessive alarms, clinician characteristics, and working conditions (eg, workload and unit noise level).¹² Evidence also suggests that humans have difficulty distinguishing the importance of alarms in situations where multiple alarm tones are used, a common scenario in hospitals.^13,14 Understanding the factors that contribute to clinicians responding or not responding to CPM alarms will be crucial for addressing this serious patient safety issue.

An enhanced understanding of why nurses respond to alarms in daily practice will inform intervention development and improvement work. In the long term, this information could help improve systems for monitoring pediatric inpatients that are less prone to issues with alarm fatigue. The objective of this qualitative study, which employed structured observation, was to describe how bedside nurses think about and act upon bedside monitor alarms in a general pediatric inpatient unit.

Study Design and Setting

This prospective observational study took place on a 48-bed hospital medicine unit at a large, freestanding children’s hospital with >650 beds and >19,000 annual admissions. General Electric (Little Chalfont, United Kingdom) physiologic monitors (models Dash 3000, 4000, and 5000) were used at the time of the study, and nurses could be notified of monitor alarms in four ways: First, an in-room auditory alarm sounds. Second, a light positioned above the door outside of each patient room blinks for alarms that are at a “warning” or “critical level” (eg ventricular tachycardia or low oxygen saturation). Third, audible alarms occur at the unit’s central monitoring station. Lastly, another staff member can notify the patient’s nurse via in-person conversion or secure smart phone communication. On the study unit, CPMs are initiated and discontinued through a physician order.

This study was reviewed and approved by the hospital’s institutional review board.

Study Population

We used a purposive recruitment strategy to enroll bedside nurses working on general hospital medicine units, stratified to ensure varying levels of experience and primary shifts (eg, day vs night). We planned to conduct approximately two observations with each participating nurse and to continue collecting data until we could no longer identify new insights in terms of responses to alarms (ie, thematic saturation¹⁵). Observations were targeted to cover times of day that coincided with increased rates of distraction. These times included just prior to and after the morning and evening change of shifts (7:00 am and 7:00 pm), during morning rounds (8:00 am-12:00 pm), and heavy admission times (12:00 pm-10:00 pm). After written informed consent, a nurse was eligible for observation during his/her shift if he/she was caring for at least one monitored patient. Enrolled nurses were made aware of the general study topic but were blinded to the study team’s hypotheses.

Data Sources

Prior to data collection, the research team, which consisted of physicians, bedside nurses, research coordinators, and a human factors expert, created a system for categorizing alarm responses. Categories for observed responses were based on the location and corresponding action taken. Initial categories were developed a priori from existing literature and expanded through input from the multidisciplinary study team, then vetted with bedside staff, and finally pilot tested through >4 hours of observations, thus producing the final categories. These categories were entered into a work-sampling program (WorkStudy by Quetech Ltd., Waterloo, Ontario, Canada) to facilitate quick data recording during observations.

The hospital uses a central alarm collection software (BedMasterEx by Anandic Medical Systems, Feuerthalen, Switzerland), which permitted the collection of date, time, trigger (eg, high heart rate), and level (eg, crisis, warning) of the generated CPM alarms. Alarms collected are based on thresholds preset at the bedside monitor. The central collection software does not differentiate between accurate (eg, correctly representing the physiologic state of the patient) and inaccurate alarms.

Observation Procedure

At the time of observation, nurse demographic information (eg, primary shift worked and years working as a nurse) was obtained. A brief preobservation questionnaire was administered to collect patient information (eg, age and diagnosis) and the nurses’ perspectives on the necessity of monitors for each monitored patient in his/her care.

The observer shadowed the nurse for a two-hour block of his/her shift. During this time, nurses were instructed to “think aloud” as they responded to alarms (eg, “I notice the oxygen saturation monitor alarming off, but the probe has fallen off”). A trained observer (AML or KMT) recorded responses verbalized by the nurse and his/her reaction by selecting the appropriate category using the work-sampling software. Data were also collected on the vital sign associated with the alarm (eg, heart rate). Moreover, the observer kept written notes to provide context for electronically recorded data. Alarms that were not verbalized by the nurse were not counted. Similarly, alarms that were noted outside of the room by the nurse were not classified by vital sign unless the nurse confirmed with the bedside monitor. Observers did not adjudicate the accuracy of the alarms. The session was stopped if monitors were discontinued during the observation period. Alarm data generated by the bedside monitor were pulled for each patient room after observations were completed.

Analysis

Descriptive statistics were used to assess the percentage of each nurse response category and each alarm type (eg, heart rate and respiratory rate). The observed alarm rate was calculated by taking the total number of observed alarms (ie, alarms noted by the nurse) divided by the total number of patient-hours observed. The monitor-generated alarm rate was calculated by taking the total number of alarms from the bedside-alarm generated data divided by the number of patient-hours observed.

Electronically recorded observations using the work-sampling program were cross-referenced with hand-written field notes to assess for any discrepancies or identify relevant events not captured by the program. Three study team members (AML, KMT, and ACS) reviewed each observation independently and compared field notes to ensure accurate categorization. Discrepancies were referred to the larger study group in cases of uncertainty.

Nine nurses had monitored patients during the available observations and participated in 19 observation sessions, which included 35 monitored patients for a total of 61.3 patient-hours of observation. Nurses were observed for a median of two times each (range 1-4). The median number of monitored patients during a single observation session was two (range 1-3). Observed nurses were female with a median of eight years of experience (range 0.5-26 years). Patients represented a broad range of age categories and were hospitalized with a variety of diagnoses (Table). Nurses, when queried at the start of the observation, felt that monitors were necessary for 29 (82.9%) of the observed patients given either patient condition or unit policy.

A total of 207 observed nurse responses to alarms occurred during the study period for a rate of 3.4 responses per patient per hour. Of the total number of responses, 45 (21.7%) were noted outside of a patient room, and in 15 (33.3%) the nurse chose to go to the room. The other 162 were recorded when the nurse was present in the room when the alarm activated. Of the 177 in-person nurse responses, 50 were related to a pulse oximetry alarm, 66 were related to a heart rate alarm, and 61 were related to a respiratory rate alarm. The most common observed in-person response to an alarm involved the nurse judging that no intervention was necessary (n = 152, 73.1%). Only 14 (7% of total responses) observed in-person responses involved a clinical intervention, such as suctioning or titrating supplemental oxygen. Findings are summarized in the Figure and describe nurse-verbalized reasons to further assess (or not) and then whether the nurse chose to take action (or not) after an alarm.

We characterized responses to physiologic monitor alarms by a group of nurses with a range of experience levels. We found that most nurse responses to alarms in continuously monitored general pediatric patients involved no intervention, and further assessment was often not conducted for alarms that occurred outside of the room if the nurse noted otherwise reassuring clinical context. Observed responses occurred for 36% of alarms during the study period when compared with bedside monitor-alarm generated data. Overall, only 14 clinical interventions were noted among the observed responses. Nurses noted that they felt the monitors were necessary for 82.9% of monitored patients because of the clinical context or because of unit policy.

Our study findings highlight some potential contradictions in the current widespread use of CPMs in general pediatric units and how clinicians respond to them in practice.² First, while nurses reported that monitors were necessary for most of their patients, participating nurses deemed few alarms clinically actionable and often chose not to further assess when they noted alarms outside of the room. This is in line with findings from prior studies suggesting that clinicians overvalue the contribution of monitoring systems to patient safety.^16,17 Second, while this finding occurred in a minority of the observations, the presence of family members at the patient’s bedside was cited by nurses as a rationale for whether they responded to alarms. While family members are capable of identifying safety issues,¹⁸ formal systems to engage them in patient safety and physiologic monitoring are lacking. Finally, clinical interventions or responses to the alerts of deteriorating patients, which best represented the original intent of CPMs, were rare and accounted for just 7% of the responses. Further work elucidating why physicians and nurses choose to use CPMs may be helpful to identify interventions to reduce inappropriate monitor use and highlight gaps in frontline staff knowledge about the benefits and risks of CPM use.

Our findings provide a novel understanding of previously observed phenomena, such as long response times or nonresponses in settings with high alarm rates.^4,10 Similar to that in a prior study conducted in the pediatric setting,¹¹ alarms with an observed response constituted a minority of the total alarms that occurred in our study. This finding has previously been attributed to mental fatigue, caregiver apathy, and desensitization.⁸ However, even though a minority of observed responses in our study included an intervention, the nurse had a rationale for why the alarm did or did not need a response. This behavior and the verbalized rationale indicate that in his/her opinion, not responding to the alarm was clinically appropriate. Study participants also reflected on the difficulties of responding to alarms given the monitor system setup, in which they may not always be capable of hearing alarms for their patients. Without data from nurses regarding the alarms that had no observed response, we can only speculate; however, based on our findings, each of these factors could contribute to nonresponse. Finally, while high numbers of false alarms have been posited as an underlying cause of alarm fatigue, we noted that a majority of nonresponse was reported to be related to other clinical factors. This relationship suggests that from the nurse’s perspective, a more applicable framework for understanding alarms would be based on clinical actionability⁴ over physiologic accuracy.

In total, our findings suggest that a multifaceted approach will be necessary to improve alarm response rates. These interventions should include adjusting parameters such that alarms are highly likely to indicate a need for intervention coupled with educational interventions addressing clinician knowledge of the alarm system and bias about the actionability of alarms may improve response rates. Changes in the monitoring system setup such that nurses can easily be notified when alarms occur may also be indicated, in addition to formally engaging patients and families around response to alarms. Although secondary notification systems (eg, alarms transmitted to individual clinician’s devices) are one solution, the utilization of these systems needs to be balanced with the risks of contributing to existing alarm fatigue and the need to appropriately tailor monitoring thresholds and strategies to patients.

Our study has several limitations. First, nurses may have responded in a way they perceive to be socially desirable, and studies using in-person observers are also prone to a Hawthorne-like effect,^19-21 where the nurse may have tried to respond more frequently to alarms than usual during observations. However, given that the majority of bedside alarms did not receive a response and a substantial number of responses involved no action, these effects were likely weak. Second, we were unable to assess which alarms were accurately reflecting the patient’s physiologic status and which were not; we were also unable to link observed alarm response to monitor-recorded alarms. Third, despite the use of silent observers and an actual, rather than a simulated, clinical setting, by virtue of the data collection method we likely captured a more deliberate thought process (so-called System 2 thinking)²² rather than the subconscious processes that may predominate when nurses respond to alarms in the course of clinical care (System 1 thinking).²² Despite this limitation, our study findings, which reflect a nurse’s in-the-moment thinking, remain relevant to guiding the improvement of monitoring systems, and the development of nurse-facing interventions and education. Finally, we studied a small, purposive sample of nurses at a single hospital. Our study sample impacts the generalizability of our results and precluded a detailed analysis of the effect of nurse- and patient-level variables.

We found that nurses often deemed that no response was necessary for CPM alarms. Nurses cited contextual factors, including the duration of alarms and the presence of other providers or parents in their decision-making. Few (7%) of the alarm responses in our study included a clinical intervention. The number of observed alarm responses constituted roughly a third of the alarms recorded by bedside CPMs during the study. This result supports concerns about the nurse’s capacity to hear and process all CPM alarms given system limitations and a heavy clinical workload. Subsequent steps should include staff education, reducing overall alarm rates with appropriate monitor use and actionable alarm thresholds, and ensuring that patient alarms are easily recognizable for frontline staff.

Disclosures

The authors have no conflicts of interest to disclose.

Funding

This work was supported by the Place Outcomes Research Award from the Cincinnati Children’s Research Foundation. Dr. Brady is supported by the Agency for Healthcare Research and Quality under Award Number K08HS23827. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality.

Alarms from bedside continuous physiologic monitors (CPMs) occur frequently in children’s hospitals and can lead to harm. Recent studies conducted in children’s hospitals have identified alarm rates of up to 152 alarms per patient per day outside of the intensive care unit,^1-3 with as few as 1% of alarms being considered clinically important.⁴ Excessive alarms have been linked to alarm fatigue, when providers become desensitized to and may miss alarms indicating impending patient deterioration. Alarm fatigue has been identified by national patient safety organizations as a patient safety concern given the risk of patient harm.^5-7 Despite these concerns, CPMs are routinely used: up to 48% of pediatric patients in nonintensive care units at children’s hospitals are monitored.²

Although the low number of alarms that receive responses has been well-described,^8,9 the reasons why clinicians do or do not respond to alarms are unclear. A study conducted in an adult perioperative unit noted prolonged nurse response times for patients with high alarm rates.¹⁰ A second study conducted in the pediatric inpatient setting demonstrated a dose-response effect and noted progressively prolonged nurse response times with increased rates of nonactionable alarms.^4,11 Findings from another study suggested that underlying factors are highly complex and may be a result of excessive alarms, clinician characteristics, and working conditions (eg, workload and unit noise level).¹² Evidence also suggests that humans have difficulty distinguishing the importance of alarms in situations where multiple alarm tones are used, a common scenario in hospitals.^13,14 Understanding the factors that contribute to clinicians responding or not responding to CPM alarms will be crucial for addressing this serious patient safety issue.

An enhanced understanding of why nurses respond to alarms in daily practice will inform intervention development and improvement work. In the long term, this information could help improve systems for monitoring pediatric inpatients that are less prone to issues with alarm fatigue. The objective of this qualitative study, which employed structured observation, was to describe how bedside nurses think about and act upon bedside monitor alarms in a general pediatric inpatient unit.

Study Design and Setting

This prospective observational study took place on a 48-bed hospital medicine unit at a large, freestanding children’s hospital with >650 beds and >19,000 annual admissions. General Electric (Little Chalfont, United Kingdom) physiologic monitors (models Dash 3000, 4000, and 5000) were used at the time of the study, and nurses could be notified of monitor alarms in four ways: First, an in-room auditory alarm sounds. Second, a light positioned above the door outside of each patient room blinks for alarms that are at a “warning” or “critical level” (eg ventricular tachycardia or low oxygen saturation). Third, audible alarms occur at the unit’s central monitoring station. Lastly, another staff member can notify the patient’s nurse via in-person conversion or secure smart phone communication. On the study unit, CPMs are initiated and discontinued through a physician order.

This study was reviewed and approved by the hospital’s institutional review board.

Study Population

We used a purposive recruitment strategy to enroll bedside nurses working on general hospital medicine units, stratified to ensure varying levels of experience and primary shifts (eg, day vs night). We planned to conduct approximately two observations with each participating nurse and to continue collecting data until we could no longer identify new insights in terms of responses to alarms (ie, thematic saturation¹⁵). Observations were targeted to cover times of day that coincided with increased rates of distraction. These times included just prior to and after the morning and evening change of shifts (7:00 am and 7:00 pm), during morning rounds (8:00 am-12:00 pm), and heavy admission times (12:00 pm-10:00 pm). After written informed consent, a nurse was eligible for observation during his/her shift if he/she was caring for at least one monitored patient. Enrolled nurses were made aware of the general study topic but were blinded to the study team’s hypotheses.

Data Sources

Prior to data collection, the research team, which consisted of physicians, bedside nurses, research coordinators, and a human factors expert, created a system for categorizing alarm responses. Categories for observed responses were based on the location and corresponding action taken. Initial categories were developed a priori from existing literature and expanded through input from the multidisciplinary study team, then vetted with bedside staff, and finally pilot tested through >4 hours of observations, thus producing the final categories. These categories were entered into a work-sampling program (WorkStudy by Quetech Ltd., Waterloo, Ontario, Canada) to facilitate quick data recording during observations.

The hospital uses a central alarm collection software (BedMasterEx by Anandic Medical Systems, Feuerthalen, Switzerland), which permitted the collection of date, time, trigger (eg, high heart rate), and level (eg, crisis, warning) of the generated CPM alarms. Alarms collected are based on thresholds preset at the bedside monitor. The central collection software does not differentiate between accurate (eg, correctly representing the physiologic state of the patient) and inaccurate alarms.

Observation Procedure

At the time of observation, nurse demographic information (eg, primary shift worked and years working as a nurse) was obtained. A brief preobservation questionnaire was administered to collect patient information (eg, age and diagnosis) and the nurses’ perspectives on the necessity of monitors for each monitored patient in his/her care.

The observer shadowed the nurse for a two-hour block of his/her shift. During this time, nurses were instructed to “think aloud” as they responded to alarms (eg, “I notice the oxygen saturation monitor alarming off, but the probe has fallen off”). A trained observer (AML or KMT) recorded responses verbalized by the nurse and his/her reaction by selecting the appropriate category using the work-sampling software. Data were also collected on the vital sign associated with the alarm (eg, heart rate). Moreover, the observer kept written notes to provide context for electronically recorded data. Alarms that were not verbalized by the nurse were not counted. Similarly, alarms that were noted outside of the room by the nurse were not classified by vital sign unless the nurse confirmed with the bedside monitor. Observers did not adjudicate the accuracy of the alarms. The session was stopped if monitors were discontinued during the observation period. Alarm data generated by the bedside monitor were pulled for each patient room after observations were completed.

Analysis

Descriptive statistics were used to assess the percentage of each nurse response category and each alarm type (eg, heart rate and respiratory rate). The observed alarm rate was calculated by taking the total number of observed alarms (ie, alarms noted by the nurse) divided by the total number of patient-hours observed. The monitor-generated alarm rate was calculated by taking the total number of alarms from the bedside-alarm generated data divided by the number of patient-hours observed.

Electronically recorded observations using the work-sampling program were cross-referenced with hand-written field notes to assess for any discrepancies or identify relevant events not captured by the program. Three study team members (AML, KMT, and ACS) reviewed each observation independently and compared field notes to ensure accurate categorization. Discrepancies were referred to the larger study group in cases of uncertainty.

Nine nurses had monitored patients during the available observations and participated in 19 observation sessions, which included 35 monitored patients for a total of 61.3 patient-hours of observation. Nurses were observed for a median of two times each (range 1-4). The median number of monitored patients during a single observation session was two (range 1-3). Observed nurses were female with a median of eight years of experience (range 0.5-26 years). Patients represented a broad range of age categories and were hospitalized with a variety of diagnoses (Table). Nurses, when queried at the start of the observation, felt that monitors were necessary for 29 (82.9%) of the observed patients given either patient condition or unit policy.

A total of 207 observed nurse responses to alarms occurred during the study period for a rate of 3.4 responses per patient per hour. Of the total number of responses, 45 (21.7%) were noted outside of a patient room, and in 15 (33.3%) the nurse chose to go to the room. The other 162 were recorded when the nurse was present in the room when the alarm activated. Of the 177 in-person nurse responses, 50 were related to a pulse oximetry alarm, 66 were related to a heart rate alarm, and 61 were related to a respiratory rate alarm. The most common observed in-person response to an alarm involved the nurse judging that no intervention was necessary (n = 152, 73.1%). Only 14 (7% of total responses) observed in-person responses involved a clinical intervention, such as suctioning or titrating supplemental oxygen. Findings are summarized in the Figure and describe nurse-verbalized reasons to further assess (or not) and then whether the nurse chose to take action (or not) after an alarm.

We characterized responses to physiologic monitor alarms by a group of nurses with a range of experience levels. We found that most nurse responses to alarms in continuously monitored general pediatric patients involved no intervention, and further assessment was often not conducted for alarms that occurred outside of the room if the nurse noted otherwise reassuring clinical context. Observed responses occurred for 36% of alarms during the study period when compared with bedside monitor-alarm generated data. Overall, only 14 clinical interventions were noted among the observed responses. Nurses noted that they felt the monitors were necessary for 82.9% of monitored patients because of the clinical context or because of unit policy.

Our study findings highlight some potential contradictions in the current widespread use of CPMs in general pediatric units and how clinicians respond to them in practice.² First, while nurses reported that monitors were necessary for most of their patients, participating nurses deemed few alarms clinically actionable and often chose not to further assess when they noted alarms outside of the room. This is in line with findings from prior studies suggesting that clinicians overvalue the contribution of monitoring systems to patient safety.^16,17 Second, while this finding occurred in a minority of the observations, the presence of family members at the patient’s bedside was cited by nurses as a rationale for whether they responded to alarms. While family members are capable of identifying safety issues,¹⁸ formal systems to engage them in patient safety and physiologic monitoring are lacking. Finally, clinical interventions or responses to the alerts of deteriorating patients, which best represented the original intent of CPMs, were rare and accounted for just 7% of the responses. Further work elucidating why physicians and nurses choose to use CPMs may be helpful to identify interventions to reduce inappropriate monitor use and highlight gaps in frontline staff knowledge about the benefits and risks of CPM use.

Our findings provide a novel understanding of previously observed phenomena, such as long response times or nonresponses in settings with high alarm rates.^4,10 Similar to that in a prior study conducted in the pediatric setting,¹¹ alarms with an observed response constituted a minority of the total alarms that occurred in our study. This finding has previously been attributed to mental fatigue, caregiver apathy, and desensitization.⁸ However, even though a minority of observed responses in our study included an intervention, the nurse had a rationale for why the alarm did or did not need a response. This behavior and the verbalized rationale indicate that in his/her opinion, not responding to the alarm was clinically appropriate. Study participants also reflected on the difficulties of responding to alarms given the monitor system setup, in which they may not always be capable of hearing alarms for their patients. Without data from nurses regarding the alarms that had no observed response, we can only speculate; however, based on our findings, each of these factors could contribute to nonresponse. Finally, while high numbers of false alarms have been posited as an underlying cause of alarm fatigue, we noted that a majority of nonresponse was reported to be related to other clinical factors. This relationship suggests that from the nurse’s perspective, a more applicable framework for understanding alarms would be based on clinical actionability⁴ over physiologic accuracy.

In total, our findings suggest that a multifaceted approach will be necessary to improve alarm response rates. These interventions should include adjusting parameters such that alarms are highly likely to indicate a need for intervention coupled with educational interventions addressing clinician knowledge of the alarm system and bias about the actionability of alarms may improve response rates. Changes in the monitoring system setup such that nurses can easily be notified when alarms occur may also be indicated, in addition to formally engaging patients and families around response to alarms. Although secondary notification systems (eg, alarms transmitted to individual clinician’s devices) are one solution, the utilization of these systems needs to be balanced with the risks of contributing to existing alarm fatigue and the need to appropriately tailor monitoring thresholds and strategies to patients.

Our study has several limitations. First, nurses may have responded in a way they perceive to be socially desirable, and studies using in-person observers are also prone to a Hawthorne-like effect,^19-21 where the nurse may have tried to respond more frequently to alarms than usual during observations. However, given that the majority of bedside alarms did not receive a response and a substantial number of responses involved no action, these effects were likely weak. Second, we were unable to assess which alarms were accurately reflecting the patient’s physiologic status and which were not; we were also unable to link observed alarm response to monitor-recorded alarms. Third, despite the use of silent observers and an actual, rather than a simulated, clinical setting, by virtue of the data collection method we likely captured a more deliberate thought process (so-called System 2 thinking)²² rather than the subconscious processes that may predominate when nurses respond to alarms in the course of clinical care (System 1 thinking).²² Despite this limitation, our study findings, which reflect a nurse’s in-the-moment thinking, remain relevant to guiding the improvement of monitoring systems, and the development of nurse-facing interventions and education. Finally, we studied a small, purposive sample of nurses at a single hospital. Our study sample impacts the generalizability of our results and precluded a detailed analysis of the effect of nurse- and patient-level variables.

We found that nurses often deemed that no response was necessary for CPM alarms. Nurses cited contextual factors, including the duration of alarms and the presence of other providers or parents in their decision-making. Few (7%) of the alarm responses in our study included a clinical intervention. The number of observed alarm responses constituted roughly a third of the alarms recorded by bedside CPMs during the study. This result supports concerns about the nurse’s capacity to hear and process all CPM alarms given system limitations and a heavy clinical workload. Subsequent steps should include staff education, reducing overall alarm rates with appropriate monitor use and actionable alarm thresholds, and ensuring that patient alarms are easily recognizable for frontline staff.

Disclosures

The authors have no conflicts of interest to disclose.

Funding

This work was supported by the Place Outcomes Research Award from the Cincinnati Children’s Research Foundation. Dr. Brady is supported by the Agency for Healthcare Research and Quality under Award Number K08HS23827. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality.

Alarm fatigue is a patient safety hazard in hospitals¹ that occurs when exposure to high rates of alarms leads clinicians to ignore or delay their responses to the alarms.^2,3 To date, most studies of physiologic monitor alarms in hospitalized children have used data from single institutions and often only a few units within each institution.⁴ These limited studies have found that alarms in pediatric units are rarely actionable.² They have also shown that physiologic monitor alarms occur frequently in children’s hospitals and that alarm rates can vary widely within a single institution,⁵ but the extent of variation between children’s hospitals is unknown. In this study, we aimed to describe and compare physiologic monitor alarm characteristics and the proportion of patients monitored in the inpatient units of 5 children’s hospitals.

We performed a cross-sectional study using a point-prevalence design of physiologic monitor alarms and monitoring during a 24-hour period at 5 large, freestanding tertiary-care children’s hospitals. At the time of the study, each hospital had an alarm management committee in place and was working to address alarm fatigue. Each hospital’s institutional review board reviewed and approved the study.

We collected 24 consecutive hours of data from the inpatient units of each hospital between March 24, 2015, and May 1, 2015. Each hospital selected the data collection date within that window based on the availability of staff to perform data collection.⁶ We excluded emergency departments, procedural areas, and inpatient psychiatry and rehabilitation units. By using existing central alarm-collection software that interfaced with bedside physiologic monitors, we collected data on audible alarms generated for apnea, arrhythmia, low and high oxygen saturation, heart rate, respiratory rate, blood pressure, and exhaled carbon dioxide. Bedside alarm systems and alarm collection software differed between centers; therefore, alarm types that were not consistently collected at every institution (eg, alarms for electrode and device malfunction, ventilators, intracranial and central venous pressure monitors, and temperatures probes) were excluded. To estimate alarm rates and to account for fluctuations in hospital census throughout the day,⁷ we collected census (to calculate the number of alarms per patient day) and the number of monitored patients (to calculate the number of alarms per monitored-patient day, including only monitored patients in the denominator) on each unit at 3 time points, 8 hours apart. Patients were considered continuously monitored if they had presence of a waveform and data for pulse oximetry, respiratory rate, and/or heart rate at the time of data collection. We then determined the rate of alarms by unit type—medical-surgical unit (MSU), neonatal intensive care unit (NICU), or pediatric intensive care unit (PICU)—and the alarm types. Based on prior literature demonstrating up to 95% of alarms contributed by a minority of patients on a single unit,⁸ we also calculated the percentage of alarms contributed by beds in the highest quartile of alarms. We also assessed the percentage of patients monitored by unit type. The Supplementary Appendix shows the alarm parameter thresholds in use at the time of the study.

Physiologic monitor alarm rates and the proportion of patients monitored varied widely between unit types and among the tertiary-care children’s hospitals in our study. We found that among MSUs, the hospital with the lowest proportion of beds monitored had the highest alarm rate, with over triple the rate seen at the hospital with the lowest alarm rate. Regardless of unit type, a small subgroup of patients at each hospital contributed a disproportionate share of alarms. These findings are concerning because of the patient morbidity and mortality associated with alarm fatigue¹ and the studies suggesting that higher alarm rates may lead to delays in response to potentially critical alarms.²

We previously described alarm rates at a single children’s hospital and found that alarm rates were high both in and outside of the ICU areas.⁵ This study supports those findings and goes further to show that alarm rates on some MSUs approached rates seen in the ICU areas at other centers.⁴ However, our results should be considered in the context of several limitations. First, the 5 study hospitals utilized different bedside monitors, equipment, and software to collect alarm data. It is possible that this impacted how alarms were counted, though there were no technical specifications to suggest that results should have been biased in a specific way. Second, our data did not reflect alarm validity (ie, whether an alarm accurately reflected the physiologic state of the patient) or factors outside of the number of patients monitored—such as practices around ICU admission and transfer as well as monitor practices such as lead changes, the type of leads employed, and the degree to which alarm parameter thresholds could be customized, which may have also affected alarm rates. Finally, we excluded alarm types that were not consistently collected at all hospitals. We were also unable to capture alarms from other alarm-generating devices, including ventilators and infusion pumps, which have also been identified as sources of alarm-related safety issues in hospitals.^9-11 This suggests that the alarm rates reported here underestimate the total number of audible alarms experienced by staff and by hospitalized patients and families.

While our data collection was limited in scope, the striking differences in alarm rates between hospitals and between similar units in the same hospitals suggest that unit- and hospital-level factors—including default alarm parameter threshold settings, types of monitors used, and monitoring practices such as the degree to which alarm parameters are customized to the patient’s physiologic state—likely contribute to the variability. It is also important to note that while there were clear outlier hospitals, no single hospital had the lowest alarm rate across all unit types. And while we found that a small number of patients contributed disproportionately to alarms, monitoring fewer patients overall was not consistently associated with lower alarm rates. While it is difficult to draw conclusions based on a limited study, these findings suggest that solutions to meaningfully lower alarm rates may be multifaceted. Standardization of care in multiple areas of medicine has shown the potential to decrease unnecessary utilization of testing and therapies while maintaining good patient outcomes.^12-15 Our findings suggest that the concept of positive deviance,¹⁶ by which some organizations produce better outcomes than others despite similar limitations, may help identify successful alarm reduction strategies for further testing. Larger quantitative studies of alarm rates and ethnographic or qualitative studies of monitoring practices may reveal practices and policies that are associated with lower alarm rates with similar or improved monitoring outcomes.

We found wide variability in physiologic monitor alarm rates and the proportion of patients monitored across 5 children’s hospitals. Because alarm fatigue remains a pressing patient safety concern, further study of the features of high-performing (low-alarm) hospital systems may help identify barriers and facilitators of safe, effective monitoring and develop targeted interventions to reduce alarms.

The authors thank Melinda Egan, Matt MacMurchy, and Shannon Stemler for their assistance with data collection.

Disclosure

Dr. Bonafide is supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award Number K23HL116427. Dr. Brady is supported by the Agency for Healthcare Research and Quality under Award Number K08HS23827. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Agency for Healthcare Research and Quality. There was no external funding obtained for this study. The authors have no conflicts of interest to disclose.

Alarm fatigue is a patient safety hazard in hospitals¹ that occurs when exposure to high rates of alarms leads clinicians to ignore or delay their responses to the alarms.^2,3 To date, most studies of physiologic monitor alarms in hospitalized children have used data from single institutions and often only a few units within each institution.⁴ These limited studies have found that alarms in pediatric units are rarely actionable.² They have also shown that physiologic monitor alarms occur frequently in children’s hospitals and that alarm rates can vary widely within a single institution,⁵ but the extent of variation between children’s hospitals is unknown. In this study, we aimed to describe and compare physiologic monitor alarm characteristics and the proportion of patients monitored in the inpatient units of 5 children’s hospitals.

We performed a cross-sectional study using a point-prevalence design of physiologic monitor alarms and monitoring during a 24-hour period at 5 large, freestanding tertiary-care children’s hospitals. At the time of the study, each hospital had an alarm management committee in place and was working to address alarm fatigue. Each hospital’s institutional review board reviewed and approved the study.

We collected 24 consecutive hours of data from the inpatient units of each hospital between March 24, 2015, and May 1, 2015. Each hospital selected the data collection date within that window based on the availability of staff to perform data collection.⁶ We excluded emergency departments, procedural areas, and inpatient psychiatry and rehabilitation units. By using existing central alarm-collection software that interfaced with bedside physiologic monitors, we collected data on audible alarms generated for apnea, arrhythmia, low and high oxygen saturation, heart rate, respiratory rate, blood pressure, and exhaled carbon dioxide. Bedside alarm systems and alarm collection software differed between centers; therefore, alarm types that were not consistently collected at every institution (eg, alarms for electrode and device malfunction, ventilators, intracranial and central venous pressure monitors, and temperatures probes) were excluded. To estimate alarm rates and to account for fluctuations in hospital census throughout the day,⁷ we collected census (to calculate the number of alarms per patient day) and the number of monitored patients (to calculate the number of alarms per monitored-patient day, including only monitored patients in the denominator) on each unit at 3 time points, 8 hours apart. Patients were considered continuously monitored if they had presence of a waveform and data for pulse oximetry, respiratory rate, and/or heart rate at the time of data collection. We then determined the rate of alarms by unit type—medical-surgical unit (MSU), neonatal intensive care unit (NICU), or pediatric intensive care unit (PICU)—and the alarm types. Based on prior literature demonstrating up to 95% of alarms contributed by a minority of patients on a single unit,⁸ we also calculated the percentage of alarms contributed by beds in the highest quartile of alarms. We also assessed the percentage of patients monitored by unit type. The Supplementary Appendix shows the alarm parameter thresholds in use at the time of the study.

Physiologic monitor alarm rates and the proportion of patients monitored varied widely between unit types and among the tertiary-care children’s hospitals in our study. We found that among MSUs, the hospital with the lowest proportion of beds monitored had the highest alarm rate, with over triple the rate seen at the hospital with the lowest alarm rate. Regardless of unit type, a small subgroup of patients at each hospital contributed a disproportionate share of alarms. These findings are concerning because of the patient morbidity and mortality associated with alarm fatigue¹ and the studies suggesting that higher alarm rates may lead to delays in response to potentially critical alarms.²

We previously described alarm rates at a single children’s hospital and found that alarm rates were high both in and outside of the ICU areas.⁵ This study supports those findings and goes further to show that alarm rates on some MSUs approached rates seen in the ICU areas at other centers.⁴ However, our results should be considered in the context of several limitations. First, the 5 study hospitals utilized different bedside monitors, equipment, and software to collect alarm data. It is possible that this impacted how alarms were counted, though there were no technical specifications to suggest that results should have been biased in a specific way. Second, our data did not reflect alarm validity (ie, whether an alarm accurately reflected the physiologic state of the patient) or factors outside of the number of patients monitored—such as practices around ICU admission and transfer as well as monitor practices such as lead changes, the type of leads employed, and the degree to which alarm parameter thresholds could be customized, which may have also affected alarm rates. Finally, we excluded alarm types that were not consistently collected at all hospitals. We were also unable to capture alarms from other alarm-generating devices, including ventilators and infusion pumps, which have also been identified as sources of alarm-related safety issues in hospitals.^9-11 This suggests that the alarm rates reported here underestimate the total number of audible alarms experienced by staff and by hospitalized patients and families.

While our data collection was limited in scope, the striking differences in alarm rates between hospitals and between similar units in the same hospitals suggest that unit- and hospital-level factors—including default alarm parameter threshold settings, types of monitors used, and monitoring practices such as the degree to which alarm parameters are customized to the patient’s physiologic state—likely contribute to the variability. It is also important to note that while there were clear outlier hospitals, no single hospital had the lowest alarm rate across all unit types. And while we found that a small number of patients contributed disproportionately to alarms, monitoring fewer patients overall was not consistently associated with lower alarm rates. While it is difficult to draw conclusions based on a limited study, these findings suggest that solutions to meaningfully lower alarm rates may be multifaceted. Standardization of care in multiple areas of medicine has shown the potential to decrease unnecessary utilization of testing and therapies while maintaining good patient outcomes.^12-15 Our findings suggest that the concept of positive deviance,¹⁶ by which some organizations produce better outcomes than others despite similar limitations, may help identify successful alarm reduction strategies for further testing. Larger quantitative studies of alarm rates and ethnographic or qualitative studies of monitoring practices may reveal practices and policies that are associated with lower alarm rates with similar or improved monitoring outcomes.

We found wide variability in physiologic monitor alarm rates and the proportion of patients monitored across 5 children’s hospitals. Because alarm fatigue remains a pressing patient safety concern, further study of the features of high-performing (low-alarm) hospital systems may help identify barriers and facilitators of safe, effective monitoring and develop targeted interventions to reduce alarms.

The authors thank Melinda Egan, Matt MacMurchy, and Shannon Stemler for their assistance with data collection.

Disclosure

Dr. Bonafide is supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award Number K23HL116427. Dr. Brady is supported by the Agency for Healthcare Research and Quality under Award Number K08HS23827. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Agency for Healthcare Research and Quality. There was no external funding obtained for this study. The authors have no conflicts of interest to disclose.

Alarm fatigue is a patient safety hazard in hospitals¹ that occurs when exposure to high rates of alarms leads clinicians to ignore or delay their responses to the alarms.^2,3 To date, most studies of physiologic monitor alarms in hospitalized children have used data from single institutions and often only a few units within each institution.⁴ These limited studies have found that alarms in pediatric units are rarely actionable.² They have also shown that physiologic monitor alarms occur frequently in children’s hospitals and that alarm rates can vary widely within a single institution,⁵ but the extent of variation between children’s hospitals is unknown. In this study, we aimed to describe and compare physiologic monitor alarm characteristics and the proportion of patients monitored in the inpatient units of 5 children’s hospitals.

We performed a cross-sectional study using a point-prevalence design of physiologic monitor alarms and monitoring during a 24-hour period at 5 large, freestanding tertiary-care children’s hospitals. At the time of the study, each hospital had an alarm management committee in place and was working to address alarm fatigue. Each hospital’s institutional review board reviewed and approved the study.

We collected 24 consecutive hours of data from the inpatient units of each hospital between March 24, 2015, and May 1, 2015. Each hospital selected the data collection date within that window based on the availability of staff to perform data collection.⁶ We excluded emergency departments, procedural areas, and inpatient psychiatry and rehabilitation units. By using existing central alarm-collection software that interfaced with bedside physiologic monitors, we collected data on audible alarms generated for apnea, arrhythmia, low and high oxygen saturation, heart rate, respiratory rate, blood pressure, and exhaled carbon dioxide. Bedside alarm systems and alarm collection software differed between centers; therefore, alarm types that were not consistently collected at every institution (eg, alarms for electrode and device malfunction, ventilators, intracranial and central venous pressure monitors, and temperatures probes) were excluded. To estimate alarm rates and to account for fluctuations in hospital census throughout the day,⁷ we collected census (to calculate the number of alarms per patient day) and the number of monitored patients (to calculate the number of alarms per monitored-patient day, including only monitored patients in the denominator) on each unit at 3 time points, 8 hours apart. Patients were considered continuously monitored if they had presence of a waveform and data for pulse oximetry, respiratory rate, and/or heart rate at the time of data collection. We then determined the rate of alarms by unit type—medical-surgical unit (MSU), neonatal intensive care unit (NICU), or pediatric intensive care unit (PICU)—and the alarm types. Based on prior literature demonstrating up to 95% of alarms contributed by a minority of patients on a single unit,⁸ we also calculated the percentage of alarms contributed by beds in the highest quartile of alarms. We also assessed the percentage of patients monitored by unit type. The Supplementary Appendix shows the alarm parameter thresholds in use at the time of the study.

Physiologic monitor alarm rates and the proportion of patients monitored varied widely between unit types and among the tertiary-care children’s hospitals in our study. We found that among MSUs, the hospital with the lowest proportion of beds monitored had the highest alarm rate, with over triple the rate seen at the hospital with the lowest alarm rate. Regardless of unit type, a small subgroup of patients at each hospital contributed a disproportionate share of alarms. These findings are concerning because of the patient morbidity and mortality associated with alarm fatigue¹ and the studies suggesting that higher alarm rates may lead to delays in response to potentially critical alarms.²

We previously described alarm rates at a single children’s hospital and found that alarm rates were high both in and outside of the ICU areas.⁵ This study supports those findings and goes further to show that alarm rates on some MSUs approached rates seen in the ICU areas at other centers.⁴ However, our results should be considered in the context of several limitations. First, the 5 study hospitals utilized different bedside monitors, equipment, and software to collect alarm data. It is possible that this impacted how alarms were counted, though there were no technical specifications to suggest that results should have been biased in a specific way. Second, our data did not reflect alarm validity (ie, whether an alarm accurately reflected the physiologic state of the patient) or factors outside of the number of patients monitored—such as practices around ICU admission and transfer as well as monitor practices such as lead changes, the type of leads employed, and the degree to which alarm parameter thresholds could be customized, which may have also affected alarm rates. Finally, we excluded alarm types that were not consistently collected at all hospitals. We were also unable to capture alarms from other alarm-generating devices, including ventilators and infusion pumps, which have also been identified as sources of alarm-related safety issues in hospitals.^9-11 This suggests that the alarm rates reported here underestimate the total number of audible alarms experienced by staff and by hospitalized patients and families.

While our data collection was limited in scope, the striking differences in alarm rates between hospitals and between similar units in the same hospitals suggest that unit- and hospital-level factors—including default alarm parameter threshold settings, types of monitors used, and monitoring practices such as the degree to which alarm parameters are customized to the patient’s physiologic state—likely contribute to the variability. It is also important to note that while there were clear outlier hospitals, no single hospital had the lowest alarm rate across all unit types. And while we found that a small number of patients contributed disproportionately to alarms, monitoring fewer patients overall was not consistently associated with lower alarm rates. While it is difficult to draw conclusions based on a limited study, these findings suggest that solutions to meaningfully lower alarm rates may be multifaceted. Standardization of care in multiple areas of medicine has shown the potential to decrease unnecessary utilization of testing and therapies while maintaining good patient outcomes.^12-15 Our findings suggest that the concept of positive deviance,¹⁶ by which some organizations produce better outcomes than others despite similar limitations, may help identify successful alarm reduction strategies for further testing. Larger quantitative studies of alarm rates and ethnographic or qualitative studies of monitoring practices may reveal practices and policies that are associated with lower alarm rates with similar or improved monitoring outcomes.

We found wide variability in physiologic monitor alarm rates and the proportion of patients monitored across 5 children’s hospitals. Because alarm fatigue remains a pressing patient safety concern, further study of the features of high-performing (low-alarm) hospital systems may help identify barriers and facilitators of safe, effective monitoring and develop targeted interventions to reduce alarms.

The authors thank Melinda Egan, Matt MacMurchy, and Shannon Stemler for their assistance with data collection.

Disclosure

Dr. Bonafide is supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award Number K23HL116427. Dr. Brady is supported by the Agency for Healthcare Research and Quality under Award Number K08HS23827. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Agency for Healthcare Research and Quality. There was no external funding obtained for this study. The authors have no conflicts of interest to disclose.