Chuan Zhou, PhD

Examining disparities in health-related quality of life (HRQoL) outcomes in children provides a unique patient-centered perspective on pediatric health services equity.^1,2 Prior studies have demonstrated the relationship between minority race, low socioeconomic status, and lower maternal education with poorer HRQoL outcomes in children.^3-6 Some studies have also shown a dose-response relationship between social disadvantage markers and poorer child health status.^7,8 Furthermore, the associations between social disadvantage and poor access to care,^9-11 and between poor access to care and lower HRQoL, are also well established.^12-14

Examining HRQoL before and after hospitalization can further our understanding of how disparities in HRQoL may change once children engage with the medical system for an acute illness.¹⁵ Children requiring hospitalization constitute a useful population for examination of this question as they represent a group of children with variable social disadvantage markers and access to outpatient care.¹⁶ Although interventions to address social determinants of health for patients with social disadvantages have been associated with within-group improvements in HRQoL, none have assessed changes in disparities as an outcome.¹⁷ Furthermore, many of these studies were conducted in the outpatient setting,^18,19 whereas hospitalization provides an additional point of care to address the social determinants of health for vulnerable families.²⁰ Even for short hospitalizations, the 24/7 nature of hospital care provides the opportunity for frequent interactions with clinicians, nurses, and support staff to clarify illness-related questions, discuss other health concerns and unmet needs, and connect with social services or community resources. These opportunities may be particularly important for families with a higher number of social disadvantage markers and even more beneficial to those with difficulty accessing needed care from their primary medical home.

In this study, we focused on children with common respiratory illnesses (asthma, bronchiolitis, and pneumonia), which constitute the majority of childhood hospitalizations.²¹ Additionally, we only focused on the child’s physical functioning component of HRQoL because this component is most likely to improve after hospitalization for children with an acute respiratory illness.²² A prior study examining HRQoL before and after hospitalization demonstrated that most children return to and/or surpass their baseline physical functioning by 1 month after hospital discharge.²³

Our primary objective was to examine associations between several markers of social disadvantage, access to care, and child physical functioning before and after hospitalization for acute respiratory illness. Second, we aimed to understand if access to care (defined as perceived difficulty/delays getting care) acts as an independent predictor of improvement in physical functioning from baseline to follow-up and/or if it modifies the relationship between social disadvantage and improvement in physical functioning (Appendix Figure).

Study Design and Population

This study was nested within a multicenter, prospective cohort study of children who were hospitalized for asthma, bronchiolitis, or pneumonia between July 2014 and June 2016 at one of five children’s hospitals in the Pediatric Research in Inpatient Settings Network.²⁴

We approached families for study participation within 72 hours of admission to the hospital using a standard protocol. Patients and their caregivers were eligible to participate in the study if the patient was 2 weeks to 16 years old and if the primary caregiver’s preferred language for medical communication was either English or Spanish. Patients with chronic medical conditions (except asthma), with moderate to severe developmental delay, with a history of prematurity <32 weeks, or who received care in the intensive care unit were excluded. Patients could only participate in the study once.

The study team set out to enroll an even number of patients across all three conditions. If a patient’s discharge diagnosis differed from their admission diagnosis (eg, from bronchiolitis to pneumonia), discharge diagnosis was used for condition group assignment. If the discharge diagnosis was not one of these three respiratory conditions, we excluded the patient from further analysis.

Data Collection

We collected data using two surveys. The first survey was administered within 72 hours of admission. This survey asked questions related to (1) caregiver-reported markers of social disadvantage, (2) caregiver perceptions of access to care, and (3) caregiver- and patient-reported assessments of physical functioning. The second survey was administered within 2 to 8 weeks after the patient’s discharge and included a second assessment of physical functioning.

Social Disadvantage

Patients were considered to have a marker of social disadvantage if their caregiver reported (1) being of non-White race and/or Hispanic ethnicity, (2) primarily speaking a language other than English at home and not speaking English very well (ie, limited English proficiency), (3) attaining at most a high school or equivalent degree, or (4) having a =/<$30,000 annual household income.

Access to Care

We used the following survey item from the 2009-2010 National Survey of Children with Special Health Care Needs²⁵ to measure caregiver perceptions of access to care: “In the last six months, did you have any difficulties or delays getting care for your child because there were waiting lists, backlogs, or other problems getting an appointment?” We narrowed the original assessment time frame from 12 months to 6 months to provide a more proximal assessment of access in relation to the hospitalization.

Child Physical Functioning

We assessed child physical functioning using the physical functioning domain of the Pediatric Quality of Life Inventory (PedsQL) 4.0 Generic Core Scales and PedsQL Infant Scales, which have been validated for use in the inpatient setting.²² Caregivers completed one of these scales based on their child’s age. Assenting patients 8 to 16 years old completed the self-report PedsQL 4.0 Generic Core Scales instrument. When completing the first PedsQL survey, caregivers and patients reflected on the previous month before their child (or they) became ill to obtain a baseline physical functioning assessment.²³ When completing the second PedsQL survey, caregivers and patients reflected on the past 7 days to obtain a follow-up assessment.

All study procedures were approved by the Western Institutional Review Board (IRB) or the participating hospitals’ IRB.

Statistical Analysis

Patients with no missing data for all four social disadvantage markers were categorized based on the number of markers they reported: none, one, two, or three or more markers. We combined patients with three and four social disadvantage markers into one group to maximize power for the analyses. We dichotomized the access to care variable and coded response options as “no difficulty/delays accessing care” if the caregiver chose “Never” and “any difficulty/delays accessing care” if they chose “Sometimes/Usually/Always.”

For each patient–caregiver dyad, PedsQL items were scored using a standard method in which higher scores reflected better functioning.²² A single set of PedsQL scores was used for each patient–caregiver dyad. We used self-reported patient scores if the patient completed the PedsQL instrument; otherwise, we used proxy-reported caregiver scores. Intraclass correlations between child self-report and parent proxy-report demonstrate moderate to good agreement above age 8 years.²⁶ We computed a change in the physical functioning score by subtracting the baseline score from the follow-up score. The minimal clinically important difference (MCID) for the PedsQL instrument is 4.5 points, which we used to identify clinically meaningful differences.¹³

Analysis of variance models were constructed to test for differences in mean baseline and follow-up PedsQL scores (dependent variable) between the following independent variables: (1) social disadvantage groups and (2) those who reported having any difficulty/delays accessing care compared with those who did not. Only patient–caregiver dyads with both baseline and follow-up assessments were included in these analyses. Mixed-effects linear regression models were constructed to identify clinically meaningful differences in PedsQL scores between groups (MCID =/> 4.5) with adjustment for patient age, gender, respiratory condition, days between surveys, and hospital site as fixed effects. Site-specific random effects were included to account for within-hospital clustering. A similarly adjusted mixed-effects linear regression model was constructed to examine whether having any difficulty/delays accessing care modified the association between social disadvantage and PedsQL change scores (eg, an improvement from baseline to follow-up).

Because 17% of respondents had missing data for at least one social disadvantage marker, sensitivity analyses were conducted using multiple imputation to account for missing social disadvantage markers using chained equations.²⁷ Sensitivity analyses were also conducted to adjust for severity of illness using vital sign data within the first 24 hours, which could only be validly captured on patients with asthma within our dataset. By restricting this latter analysis to patients with asthma, we were able to examine the relationships of interest in a population with chronic disease.

The study sample included 1,860 patients, of which 1,325 had both baseline and follow-up PedsQL data (71%). Descriptive statistics were similar between those who completed the baseline and follow-up surveys (Table 1).

Twenty-two percent of patients had >/=3 social disadvantages and 30% of caregivers reported having any difficulty/delays accessing care. The mean follow-up PedsQL score was higher than the baseline score (90.4 vs 82.5; Table 1).

Social Disadvantage Markers and PedsQL Scores

The number of social disadvantage markers was inversely related to mean baseline PedsQL scores, but there was no difference in mean follow-up PedsQL scores between social disadvantage groups (Table 2). In adjusted analyses, the mean baseline PedsQL score was −6.1 points (95% CI: −8.7, −3.5) lower for patients with >/= 3 social disadvantage markers compared with patients with no social disadvantage markers, which exceeded the scale’s MCID.

Difficulty/Delays Accessing Care and PedsQL Scores

Having any difficulty/delays accessing care was significantly associated with lower baseline and follow-up PedsQL scores (Table 2). In adjusted analyses, the difference in baseline scores was 5.2 points (95% CI: −7.2, −3.2), which exceedes the scale’s MCID.

Interaction Between Social Disadvantage Markers, Difficulty/Delays Accessing Care, and Change in PedsQL Scores from Baseline to Follow-Up

While having =/>2 social disadvantage markers and difficulty/delays accessing care were each positively associated with changes in PedsQL scores from baseline to follow-up (Table 3), only patients with =/> 3 social disadvantage markers exceeded the PedsQL MCID. In stratified analyses, patients with a combination of social disadvantage makers and difficulties/delays accessing care had lower baseline PedsQL scores and greater change in PedsQL scores from baseline to follow-up compared with those without difficulties/delays accessing care (Figure). However, having any difficulty/delays accessing care did not significantly modify the relationship between social disadvantage and change in PedsQL scores, as none of the interaction terms were significant (Table 3, Model 3).

Sensitivity Analysis

Baseline, follow-up, and change in PedsQL scores were similar to our main analysis after performing multiple imputation for missing social disadvantage markers (Supplemental Table 1). Findings were also similar for patients with a diagnosis of asthma only; however, changes in PedsQL scores were greater in magnitude (Appendix Table 2).

This study examined the relationship between social disadvantage and child physical functioning before and after hospitalization for acute respiratory illness. Study findings indicated that patients with higher numbers of social disadvantage markers reported lower PedsQL scores before hospitalization; however, differences in PedsQL scores were not apparent after hospitalization. Patients who experienced difficulty/delays accessing care also reported lower PedsQL scores at baseline. This difference was still significant but did not exceed the PedsQL MCID threshold after hospitalization. Difficulty/delays accessing care appeared to be an additional social disadvantage marker; however, it did not modify the relationship between social disadvantage and improvement in physical functioning.

The study findings at baseline are consistent with prior studies demonstrating a negative association between social disadvantage markers and HRQoL and a cumulative effect based on the number of social disadvantages.^3,4,7,8 This study adds to the existing literature by examining how this relationship changes after hospitalization. As evidenced by the lack of association between social disadvantage markers and follow-up PedsQL scores, our findings suggest that receipt of inpatient care improved perceptions of physical functioning to a greater extent for patients with more social disadvantage markers (especially patients with =/> 3 social disadvantage markers). There are several potential reasons for these findings.

Social disadvantage was associated with lower child physical functioning before hospitalization, but differences were not apparent after hospitalization for children with acute respiratory illness. Caregiver-perceived difficulty/delays accessing care was found to be an additional predictor of lower physical functioning at baseline but did not significantly alter the association between social disadvantage and improvement in physical functioning over time. Further studies are needed to understand how hospital care may differentially impact child physical functioning for patients with higher social disadvantage makers in order to sustain improvements in HRQoL disparities.

Acknowledgments

The authors thank the following individuals of the Pediatric Respiratory Illness Measurement System (PRIMES) study team for their contributions to this work: Karen M. Wilson, New York, New York; Ricardo A. Quinonez, Houston, Texas; Joyee G. Vachani, Houston, Texas; and Amy Tyler, Aurora, Colorado. We would also like to thank the Pediatric Research in Inpatient Settings Network for facilitating this work.

Examining disparities in health-related quality of life (HRQoL) outcomes in children provides a unique patient-centered perspective on pediatric health services equity.^1,2 Prior studies have demonstrated the relationship between minority race, low socioeconomic status, and lower maternal education with poorer HRQoL outcomes in children.^3-6 Some studies have also shown a dose-response relationship between social disadvantage markers and poorer child health status.^7,8 Furthermore, the associations between social disadvantage and poor access to care,^9-11 and between poor access to care and lower HRQoL, are also well established.^12-14

Examining HRQoL before and after hospitalization can further our understanding of how disparities in HRQoL may change once children engage with the medical system for an acute illness.¹⁵ Children requiring hospitalization constitute a useful population for examination of this question as they represent a group of children with variable social disadvantage markers and access to outpatient care.¹⁶ Although interventions to address social determinants of health for patients with social disadvantages have been associated with within-group improvements in HRQoL, none have assessed changes in disparities as an outcome.¹⁷ Furthermore, many of these studies were conducted in the outpatient setting,^18,19 whereas hospitalization provides an additional point of care to address the social determinants of health for vulnerable families.²⁰ Even for short hospitalizations, the 24/7 nature of hospital care provides the opportunity for frequent interactions with clinicians, nurses, and support staff to clarify illness-related questions, discuss other health concerns and unmet needs, and connect with social services or community resources. These opportunities may be particularly important for families with a higher number of social disadvantage markers and even more beneficial to those with difficulty accessing needed care from their primary medical home.

In this study, we focused on children with common respiratory illnesses (asthma, bronchiolitis, and pneumonia), which constitute the majority of childhood hospitalizations.²¹ Additionally, we only focused on the child’s physical functioning component of HRQoL because this component is most likely to improve after hospitalization for children with an acute respiratory illness.²² A prior study examining HRQoL before and after hospitalization demonstrated that most children return to and/or surpass their baseline physical functioning by 1 month after hospital discharge.²³

Our primary objective was to examine associations between several markers of social disadvantage, access to care, and child physical functioning before and after hospitalization for acute respiratory illness. Second, we aimed to understand if access to care (defined as perceived difficulty/delays getting care) acts as an independent predictor of improvement in physical functioning from baseline to follow-up and/or if it modifies the relationship between social disadvantage and improvement in physical functioning (Appendix Figure).

Study Design and Population

This study was nested within a multicenter, prospective cohort study of children who were hospitalized for asthma, bronchiolitis, or pneumonia between July 2014 and June 2016 at one of five children’s hospitals in the Pediatric Research in Inpatient Settings Network.²⁴

We approached families for study participation within 72 hours of admission to the hospital using a standard protocol. Patients and their caregivers were eligible to participate in the study if the patient was 2 weeks to 16 years old and if the primary caregiver’s preferred language for medical communication was either English or Spanish. Patients with chronic medical conditions (except asthma), with moderate to severe developmental delay, with a history of prematurity <32 weeks, or who received care in the intensive care unit were excluded. Patients could only participate in the study once.

The study team set out to enroll an even number of patients across all three conditions. If a patient’s discharge diagnosis differed from their admission diagnosis (eg, from bronchiolitis to pneumonia), discharge diagnosis was used for condition group assignment. If the discharge diagnosis was not one of these three respiratory conditions, we excluded the patient from further analysis.

Data Collection

We collected data using two surveys. The first survey was administered within 72 hours of admission. This survey asked questions related to (1) caregiver-reported markers of social disadvantage, (2) caregiver perceptions of access to care, and (3) caregiver- and patient-reported assessments of physical functioning. The second survey was administered within 2 to 8 weeks after the patient’s discharge and included a second assessment of physical functioning.

Social Disadvantage

Patients were considered to have a marker of social disadvantage if their caregiver reported (1) being of non-White race and/or Hispanic ethnicity, (2) primarily speaking a language other than English at home and not speaking English very well (ie, limited English proficiency), (3) attaining at most a high school or equivalent degree, or (4) having a =/<$30,000 annual household income.

Access to Care

We used the following survey item from the 2009-2010 National Survey of Children with Special Health Care Needs²⁵ to measure caregiver perceptions of access to care: “In the last six months, did you have any difficulties or delays getting care for your child because there were waiting lists, backlogs, or other problems getting an appointment?” We narrowed the original assessment time frame from 12 months to 6 months to provide a more proximal assessment of access in relation to the hospitalization.

Child Physical Functioning

We assessed child physical functioning using the physical functioning domain of the Pediatric Quality of Life Inventory (PedsQL) 4.0 Generic Core Scales and PedsQL Infant Scales, which have been validated for use in the inpatient setting.²² Caregivers completed one of these scales based on their child’s age. Assenting patients 8 to 16 years old completed the self-report PedsQL 4.0 Generic Core Scales instrument. When completing the first PedsQL survey, caregivers and patients reflected on the previous month before their child (or they) became ill to obtain a baseline physical functioning assessment.²³ When completing the second PedsQL survey, caregivers and patients reflected on the past 7 days to obtain a follow-up assessment.

All study procedures were approved by the Western Institutional Review Board (IRB) or the participating hospitals’ IRB.

Statistical Analysis

Patients with no missing data for all four social disadvantage markers were categorized based on the number of markers they reported: none, one, two, or three or more markers. We combined patients with three and four social disadvantage markers into one group to maximize power for the analyses. We dichotomized the access to care variable and coded response options as “no difficulty/delays accessing care” if the caregiver chose “Never” and “any difficulty/delays accessing care” if they chose “Sometimes/Usually/Always.”

For each patient–caregiver dyad, PedsQL items were scored using a standard method in which higher scores reflected better functioning.²² A single set of PedsQL scores was used for each patient–caregiver dyad. We used self-reported patient scores if the patient completed the PedsQL instrument; otherwise, we used proxy-reported caregiver scores. Intraclass correlations between child self-report and parent proxy-report demonstrate moderate to good agreement above age 8 years.²⁶ We computed a change in the physical functioning score by subtracting the baseline score from the follow-up score. The minimal clinically important difference (MCID) for the PedsQL instrument is 4.5 points, which we used to identify clinically meaningful differences.¹³

Analysis of variance models were constructed to test for differences in mean baseline and follow-up PedsQL scores (dependent variable) between the following independent variables: (1) social disadvantage groups and (2) those who reported having any difficulty/delays accessing care compared with those who did not. Only patient–caregiver dyads with both baseline and follow-up assessments were included in these analyses. Mixed-effects linear regression models were constructed to identify clinically meaningful differences in PedsQL scores between groups (MCID =/> 4.5) with adjustment for patient age, gender, respiratory condition, days between surveys, and hospital site as fixed effects. Site-specific random effects were included to account for within-hospital clustering. A similarly adjusted mixed-effects linear regression model was constructed to examine whether having any difficulty/delays accessing care modified the association between social disadvantage and PedsQL change scores (eg, an improvement from baseline to follow-up).

Because 17% of respondents had missing data for at least one social disadvantage marker, sensitivity analyses were conducted using multiple imputation to account for missing social disadvantage markers using chained equations.²⁷ Sensitivity analyses were also conducted to adjust for severity of illness using vital sign data within the first 24 hours, which could only be validly captured on patients with asthma within our dataset. By restricting this latter analysis to patients with asthma, we were able to examine the relationships of interest in a population with chronic disease.

The study sample included 1,860 patients, of which 1,325 had both baseline and follow-up PedsQL data (71%). Descriptive statistics were similar between those who completed the baseline and follow-up surveys (Table 1).

Twenty-two percent of patients had >/=3 social disadvantages and 30% of caregivers reported having any difficulty/delays accessing care. The mean follow-up PedsQL score was higher than the baseline score (90.4 vs 82.5; Table 1).

Social Disadvantage Markers and PedsQL Scores

The number of social disadvantage markers was inversely related to mean baseline PedsQL scores, but there was no difference in mean follow-up PedsQL scores between social disadvantage groups (Table 2). In adjusted analyses, the mean baseline PedsQL score was −6.1 points (95% CI: −8.7, −3.5) lower for patients with >/= 3 social disadvantage markers compared with patients with no social disadvantage markers, which exceeded the scale’s MCID.

Difficulty/Delays Accessing Care and PedsQL Scores

Having any difficulty/delays accessing care was significantly associated with lower baseline and follow-up PedsQL scores (Table 2). In adjusted analyses, the difference in baseline scores was 5.2 points (95% CI: −7.2, −3.2), which exceedes the scale’s MCID.

Interaction Between Social Disadvantage Markers, Difficulty/Delays Accessing Care, and Change in PedsQL Scores from Baseline to Follow-Up

While having =/>2 social disadvantage markers and difficulty/delays accessing care were each positively associated with changes in PedsQL scores from baseline to follow-up (Table 3), only patients with =/> 3 social disadvantage markers exceeded the PedsQL MCID. In stratified analyses, patients with a combination of social disadvantage makers and difficulties/delays accessing care had lower baseline PedsQL scores and greater change in PedsQL scores from baseline to follow-up compared with those without difficulties/delays accessing care (Figure). However, having any difficulty/delays accessing care did not significantly modify the relationship between social disadvantage and change in PedsQL scores, as none of the interaction terms were significant (Table 3, Model 3).

Sensitivity Analysis

Baseline, follow-up, and change in PedsQL scores were similar to our main analysis after performing multiple imputation for missing social disadvantage markers (Supplemental Table 1). Findings were also similar for patients with a diagnosis of asthma only; however, changes in PedsQL scores were greater in magnitude (Appendix Table 2).

This study examined the relationship between social disadvantage and child physical functioning before and after hospitalization for acute respiratory illness. Study findings indicated that patients with higher numbers of social disadvantage markers reported lower PedsQL scores before hospitalization; however, differences in PedsQL scores were not apparent after hospitalization. Patients who experienced difficulty/delays accessing care also reported lower PedsQL scores at baseline. This difference was still significant but did not exceed the PedsQL MCID threshold after hospitalization. Difficulty/delays accessing care appeared to be an additional social disadvantage marker; however, it did not modify the relationship between social disadvantage and improvement in physical functioning.

The study findings at baseline are consistent with prior studies demonstrating a negative association between social disadvantage markers and HRQoL and a cumulative effect based on the number of social disadvantages.^3,4,7,8 This study adds to the existing literature by examining how this relationship changes after hospitalization. As evidenced by the lack of association between social disadvantage markers and follow-up PedsQL scores, our findings suggest that receipt of inpatient care improved perceptions of physical functioning to a greater extent for patients with more social disadvantage markers (especially patients with =/> 3 social disadvantage markers). There are several potential reasons for these findings.

Social disadvantage was associated with lower child physical functioning before hospitalization, but differences were not apparent after hospitalization for children with acute respiratory illness. Caregiver-perceived difficulty/delays accessing care was found to be an additional predictor of lower physical functioning at baseline but did not significantly alter the association between social disadvantage and improvement in physical functioning over time. Further studies are needed to understand how hospital care may differentially impact child physical functioning for patients with higher social disadvantage makers in order to sustain improvements in HRQoL disparities.

Acknowledgments

The authors thank the following individuals of the Pediatric Respiratory Illness Measurement System (PRIMES) study team for their contributions to this work: Karen M. Wilson, New York, New York; Ricardo A. Quinonez, Houston, Texas; Joyee G. Vachani, Houston, Texas; and Amy Tyler, Aurora, Colorado. We would also like to thank the Pediatric Research in Inpatient Settings Network for facilitating this work.

Examining disparities in health-related quality of life (HRQoL) outcomes in children provides a unique patient-centered perspective on pediatric health services equity.^1,2 Prior studies have demonstrated the relationship between minority race, low socioeconomic status, and lower maternal education with poorer HRQoL outcomes in children.^3-6 Some studies have also shown a dose-response relationship between social disadvantage markers and poorer child health status.^7,8 Furthermore, the associations between social disadvantage and poor access to care,^9-11 and between poor access to care and lower HRQoL, are also well established.^12-14

Examining HRQoL before and after hospitalization can further our understanding of how disparities in HRQoL may change once children engage with the medical system for an acute illness.¹⁵ Children requiring hospitalization constitute a useful population for examination of this question as they represent a group of children with variable social disadvantage markers and access to outpatient care.¹⁶ Although interventions to address social determinants of health for patients with social disadvantages have been associated with within-group improvements in HRQoL, none have assessed changes in disparities as an outcome.¹⁷ Furthermore, many of these studies were conducted in the outpatient setting,^18,19 whereas hospitalization provides an additional point of care to address the social determinants of health for vulnerable families.²⁰ Even for short hospitalizations, the 24/7 nature of hospital care provides the opportunity for frequent interactions with clinicians, nurses, and support staff to clarify illness-related questions, discuss other health concerns and unmet needs, and connect with social services or community resources. These opportunities may be particularly important for families with a higher number of social disadvantage markers and even more beneficial to those with difficulty accessing needed care from their primary medical home.

In this study, we focused on children with common respiratory illnesses (asthma, bronchiolitis, and pneumonia), which constitute the majority of childhood hospitalizations.²¹ Additionally, we only focused on the child’s physical functioning component of HRQoL because this component is most likely to improve after hospitalization for children with an acute respiratory illness.²² A prior study examining HRQoL before and after hospitalization demonstrated that most children return to and/or surpass their baseline physical functioning by 1 month after hospital discharge.²³

Our primary objective was to examine associations between several markers of social disadvantage, access to care, and child physical functioning before and after hospitalization for acute respiratory illness. Second, we aimed to understand if access to care (defined as perceived difficulty/delays getting care) acts as an independent predictor of improvement in physical functioning from baseline to follow-up and/or if it modifies the relationship between social disadvantage and improvement in physical functioning (Appendix Figure).

Study Design and Population

This study was nested within a multicenter, prospective cohort study of children who were hospitalized for asthma, bronchiolitis, or pneumonia between July 2014 and June 2016 at one of five children’s hospitals in the Pediatric Research in Inpatient Settings Network.²⁴

We approached families for study participation within 72 hours of admission to the hospital using a standard protocol. Patients and their caregivers were eligible to participate in the study if the patient was 2 weeks to 16 years old and if the primary caregiver’s preferred language for medical communication was either English or Spanish. Patients with chronic medical conditions (except asthma), with moderate to severe developmental delay, with a history of prematurity <32 weeks, or who received care in the intensive care unit were excluded. Patients could only participate in the study once.

The study team set out to enroll an even number of patients across all three conditions. If a patient’s discharge diagnosis differed from their admission diagnosis (eg, from bronchiolitis to pneumonia), discharge diagnosis was used for condition group assignment. If the discharge diagnosis was not one of these three respiratory conditions, we excluded the patient from further analysis.

Data Collection

We collected data using two surveys. The first survey was administered within 72 hours of admission. This survey asked questions related to (1) caregiver-reported markers of social disadvantage, (2) caregiver perceptions of access to care, and (3) caregiver- and patient-reported assessments of physical functioning. The second survey was administered within 2 to 8 weeks after the patient’s discharge and included a second assessment of physical functioning.

Social Disadvantage

Patients were considered to have a marker of social disadvantage if their caregiver reported (1) being of non-White race and/or Hispanic ethnicity, (2) primarily speaking a language other than English at home and not speaking English very well (ie, limited English proficiency), (3) attaining at most a high school or equivalent degree, or (4) having a =/<$30,000 annual household income.

Access to Care

We used the following survey item from the 2009-2010 National Survey of Children with Special Health Care Needs²⁵ to measure caregiver perceptions of access to care: “In the last six months, did you have any difficulties or delays getting care for your child because there were waiting lists, backlogs, or other problems getting an appointment?” We narrowed the original assessment time frame from 12 months to 6 months to provide a more proximal assessment of access in relation to the hospitalization.

Child Physical Functioning

We assessed child physical functioning using the physical functioning domain of the Pediatric Quality of Life Inventory (PedsQL) 4.0 Generic Core Scales and PedsQL Infant Scales, which have been validated for use in the inpatient setting.²² Caregivers completed one of these scales based on their child’s age. Assenting patients 8 to 16 years old completed the self-report PedsQL 4.0 Generic Core Scales instrument. When completing the first PedsQL survey, caregivers and patients reflected on the previous month before their child (or they) became ill to obtain a baseline physical functioning assessment.²³ When completing the second PedsQL survey, caregivers and patients reflected on the past 7 days to obtain a follow-up assessment.

All study procedures were approved by the Western Institutional Review Board (IRB) or the participating hospitals’ IRB.

Statistical Analysis

Patients with no missing data for all four social disadvantage markers were categorized based on the number of markers they reported: none, one, two, or three or more markers. We combined patients with three and four social disadvantage markers into one group to maximize power for the analyses. We dichotomized the access to care variable and coded response options as “no difficulty/delays accessing care” if the caregiver chose “Never” and “any difficulty/delays accessing care” if they chose “Sometimes/Usually/Always.”

For each patient–caregiver dyad, PedsQL items were scored using a standard method in which higher scores reflected better functioning.²² A single set of PedsQL scores was used for each patient–caregiver dyad. We used self-reported patient scores if the patient completed the PedsQL instrument; otherwise, we used proxy-reported caregiver scores. Intraclass correlations between child self-report and parent proxy-report demonstrate moderate to good agreement above age 8 years.²⁶ We computed a change in the physical functioning score by subtracting the baseline score from the follow-up score. The minimal clinically important difference (MCID) for the PedsQL instrument is 4.5 points, which we used to identify clinically meaningful differences.¹³

Analysis of variance models were constructed to test for differences in mean baseline and follow-up PedsQL scores (dependent variable) between the following independent variables: (1) social disadvantage groups and (2) those who reported having any difficulty/delays accessing care compared with those who did not. Only patient–caregiver dyads with both baseline and follow-up assessments were included in these analyses. Mixed-effects linear regression models were constructed to identify clinically meaningful differences in PedsQL scores between groups (MCID =/> 4.5) with adjustment for patient age, gender, respiratory condition, days between surveys, and hospital site as fixed effects. Site-specific random effects were included to account for within-hospital clustering. A similarly adjusted mixed-effects linear regression model was constructed to examine whether having any difficulty/delays accessing care modified the association between social disadvantage and PedsQL change scores (eg, an improvement from baseline to follow-up).

Because 17% of respondents had missing data for at least one social disadvantage marker, sensitivity analyses were conducted using multiple imputation to account for missing social disadvantage markers using chained equations.²⁷ Sensitivity analyses were also conducted to adjust for severity of illness using vital sign data within the first 24 hours, which could only be validly captured on patients with asthma within our dataset. By restricting this latter analysis to patients with asthma, we were able to examine the relationships of interest in a population with chronic disease.

The study sample included 1,860 patients, of which 1,325 had both baseline and follow-up PedsQL data (71%). Descriptive statistics were similar between those who completed the baseline and follow-up surveys (Table 1).

Twenty-two percent of patients had >/=3 social disadvantages and 30% of caregivers reported having any difficulty/delays accessing care. The mean follow-up PedsQL score was higher than the baseline score (90.4 vs 82.5; Table 1).

Social Disadvantage Markers and PedsQL Scores

The number of social disadvantage markers was inversely related to mean baseline PedsQL scores, but there was no difference in mean follow-up PedsQL scores between social disadvantage groups (Table 2). In adjusted analyses, the mean baseline PedsQL score was −6.1 points (95% CI: −8.7, −3.5) lower for patients with >/= 3 social disadvantage markers compared with patients with no social disadvantage markers, which exceeded the scale’s MCID.

Difficulty/Delays Accessing Care and PedsQL Scores

Having any difficulty/delays accessing care was significantly associated with lower baseline and follow-up PedsQL scores (Table 2). In adjusted analyses, the difference in baseline scores was 5.2 points (95% CI: −7.2, −3.2), which exceedes the scale’s MCID.

Interaction Between Social Disadvantage Markers, Difficulty/Delays Accessing Care, and Change in PedsQL Scores from Baseline to Follow-Up

While having =/>2 social disadvantage markers and difficulty/delays accessing care were each positively associated with changes in PedsQL scores from baseline to follow-up (Table 3), only patients with =/> 3 social disadvantage markers exceeded the PedsQL MCID. In stratified analyses, patients with a combination of social disadvantage makers and difficulties/delays accessing care had lower baseline PedsQL scores and greater change in PedsQL scores from baseline to follow-up compared with those without difficulties/delays accessing care (Figure). However, having any difficulty/delays accessing care did not significantly modify the relationship between social disadvantage and change in PedsQL scores, as none of the interaction terms were significant (Table 3, Model 3).

Sensitivity Analysis

Baseline, follow-up, and change in PedsQL scores were similar to our main analysis after performing multiple imputation for missing social disadvantage markers (Supplemental Table 1). Findings were also similar for patients with a diagnosis of asthma only; however, changes in PedsQL scores were greater in magnitude (Appendix Table 2).

This study examined the relationship between social disadvantage and child physical functioning before and after hospitalization for acute respiratory illness. Study findings indicated that patients with higher numbers of social disadvantage markers reported lower PedsQL scores before hospitalization; however, differences in PedsQL scores were not apparent after hospitalization. Patients who experienced difficulty/delays accessing care also reported lower PedsQL scores at baseline. This difference was still significant but did not exceed the PedsQL MCID threshold after hospitalization. Difficulty/delays accessing care appeared to be an additional social disadvantage marker; however, it did not modify the relationship between social disadvantage and improvement in physical functioning.

The study findings at baseline are consistent with prior studies demonstrating a negative association between social disadvantage markers and HRQoL and a cumulative effect based on the number of social disadvantages.^3,4,7,8 This study adds to the existing literature by examining how this relationship changes after hospitalization. As evidenced by the lack of association between social disadvantage markers and follow-up PedsQL scores, our findings suggest that receipt of inpatient care improved perceptions of physical functioning to a greater extent for patients with more social disadvantage markers (especially patients with =/> 3 social disadvantage markers). There are several potential reasons for these findings.

Social disadvantage was associated with lower child physical functioning before hospitalization, but differences were not apparent after hospitalization for children with acute respiratory illness. Caregiver-perceived difficulty/delays accessing care was found to be an additional predictor of lower physical functioning at baseline but did not significantly alter the association between social disadvantage and improvement in physical functioning over time. Further studies are needed to understand how hospital care may differentially impact child physical functioning for patients with higher social disadvantage makers in order to sustain improvements in HRQoL disparities.

Acknowledgments

The authors thank the following individuals of the Pediatric Respiratory Illness Measurement System (PRIMES) study team for their contributions to this work: Karen M. Wilson, New York, New York; Ricardo A. Quinonez, Houston, Texas; Joyee G. Vachani, Houston, Texas; and Amy Tyler, Aurora, Colorado. We would also like to thank the Pediatric Research in Inpatient Settings Network for facilitating this work.

Acute kidney injury (AKI) is increasingly common in the hospitalized patient^1,2 with recent adult and pediatric multinational studies reporting AKI rates of 57% and 27%, respectively.^3,4 The development of AKI is associated with significant adverse outcomes including an increased risk of mortality.^5-7 For those that survive, the history of AKI may contribute to a lifetime of impaired health with chronic kidney disease.^8,9 This is particularly concerning for pediatric patients as AKI may impact morbidity for many decades, influence available therapies for these morbidities, and ultimately contribute to a shortened lifespan.¹⁰

AKI in the hospitalized patient is no longer accepted as an unfortunate and unavoidable consequence of illness or the indicated therapy. Currently, there is strong interest in this hospital-acquired condition with global initiatives aimed at increased prevention and early detection and treatment of AKI.^11,12 To this objective, risk stratification tools or prediction models could assist clinicians in decision making. Numerous studies have tested AKI prediction models either in particular high-risk populations or based on associated comorbidities, biomarkers, and critical illness scores. These studies are predominantly in adult populations, and few have been externally validated.¹³ While associations between certain medications and AKI are well known, an AKI prediction model that is applicable to pediatric or adult populations and is based on medication exposure is difficult. However, there is a growing recognition of the potential to develop such a model using the electronic health record (EHR).¹⁴

In 2013, Seattle Children’s Hospital (SCH) implemented a nephrotoxin and AKI detection system to assist in clinical decision making within the EHR. This system instituted the automatic ordering of serum creatinines to screen for AKI when the provider ordered three or more medications that were suspected to be nephrotoxic. Other clinical factors such as the diagnoses or preexisting conditions were not considered in the decision-tool algorithm. This original algorithm (Algorithm 1) was later modified and the list of suspected nephrotoxins was expanded (Table 1) in order to align with a national pediatric AKI collaborative (Algorithm 2). However, it was unclear whether the algorithm modification would improve AKI detection.

Context

SCH is a tertiary care academic teaching hospital affiliated with the University of Washington School of Medicine, Harborview Medical Center, and the Seattle Cancer Care Alliance. The hospital has 371 licensed beds and approximately 18 medical subspecialty services.

Study Population

This was a retrospective cohort study examining all patients ages 0-21 years admitted to SCH between December 1, 2013 and November 30, 2015. The detection system was modified to align with the national pediatric AKI collaborative, Nephrotoxic Injury Negated by Just-in-Time Action (NINJA) in November 2014. Both acute care and intensive care patients were included (data not separated by location). Patients who had end-stage kidney disease and were receiving dialysis and patients who were evaluated in the emergency department without being admitted or admitted as observation status were excluded from analysis. Patients were also excluded if they did not have a baseline serum creatinine as defined below.

Study Measures

AKI is defined at SCH using the Kidney Disease: Improving Global Outcomes Stage 1 criteria as a guideline. The diagnosis of AKI is based on an increase in the baseline serum creatinine by 0.3 mg/dL or an increase in the serum creatinine by >1.5 times the baseline assuming the incoming creatinine is 0.5 mg/dL or higher. For our definition, the increase in serum creatinine needs to have occurred within a one-week timeframe and urine output is not a diagnostic criterion.¹⁵ Baseline serum creatinine is defined as the lowest serum creatinine in the previous six months. Forty medications were classified as nephrotoxins based on previous analysis¹⁶ and adapted for our institutional formulary.

Statistical Analysis

To evaluate the efficacy of our systems in detecting nephrotoxin-related AKI, the sensitivity and the specificity using both our original algorithm (Algorithm 1) and the modified algorithm (Algorithm 2) were generated on our complete data set. To test sensitivity, the proportion of AKI patients who would trigger alert using Algorithm 1 and then with Algorithm 2 was identified. Similarly, to test specificity, the proportion of non-AKI patients who did not trigger an alert by the surveillance systems was identified. The differences in sensitivity and specificity between the two algorithms were evaluated using two-sample tests of proportion.

The statistical method of Combinatorial Inference has been utilized in studies of cancer biology¹⁷ and in genomics.¹⁸ A variation of this approach was used in this study to identify the specific medication combinations most associated with AKI. First, all of the nephrotoxic medications and medication combinations that were prescribed during our study period were identified from a data set (ie, a training set) containing 75% of all encounters selected at random without replacement. Using this training set, the prevalence of each medication combination and the rate of AKI associated with each combination were identified. The predicted overall AKI risk of an individual medication is the average of all the AKI rates associated with each combination containing that specific medication. Also incorporated into the determination of the predicted AKI risk was the prevalence of that medication combination.

To test our model’s predictive capability, the algorithm was applied to the remaining 25% of the total patient data (ie, the test set). The predicted AKI risk was compared with the actual AKI rate in the test data set. Our model’s predictive capability was represented in a receiver operator characteristic (ROC) analysis. The goal was to achieve an area under the ROC curve (AUC) approaching one as this would reflect 100% sensitivity and 100% specificity, whereas an AUC of 0.5 would represent a random guess (50% chance of being correct).

Lastly, our final step was to use our model’s ROC curve to determine an optimal threshold of AKI risk for which to trigger an alert. This predicted risk threshold was based on our goal to increase our surveillance system’s sensitivity balanced with maintaining an acceptable specificity.

An a priori threshold of P = .05 was used to determine statistical significance of all results. Analyses were conducted in Stata 12.1 (StataCorp LP, College Station, Texas) and R 3.3.2 (R Foundation for Statistical Computing, Vienna, Austria). A sample data set containing replication code for our model can be found in an online repository (https://dataverse.harvard.edu/dataverse/chuan). This study was approved by the Seattle Children’s Institutional Review Board.

Sensitivity and Specificity

Of the patient encounters, 14,779 were eligible during the study period. The sensitivity of the system’s ability to identify nephrotoxin-related AKI decreased from 46.9% using Algorithm 1 to 43.3% using Algorithm 2, a change of 3.6% (P = .22). The specificity increased from 73.6% to 89.3%, a change of 15.7% (P < .001; Table 2).

Improvement of Our Nephrotoxin-Related AKI Detection System Using a Novel AKI Prediction Strategy

A total of 838 medication combinations were identified in our training set and the predicted AKI risk for every medication combination was determined. By comparing the predicted risk of AKI to the actual AKI occurrence, an ROC curve with an AUC of 0.756 (Figure) was generated. An increase in system sensitivity was prioritized when determining the optimal AKI risk at which the model would trigger an alert. Setting an alert threshold at a predicted AKI risk of >8%, our model performed with a sensitivity of 74% while decreasing the specificity to 70%.

Identification of High-Risk Nephrotoxic Medications and Medication Combinations

Approximately 200 medication combinations were associated with >8% AKI risk, our new AKI prediction model’s alert threshold. Medication combinations consisting of up to 11 concomitantly prescribed medications were present in our data set. However, many of these combinations were infrequently prescribed. Further analysis, conducted in order to increase the clinical relevance of our findings, identified 10 medications or medication combinations that were both associated with a predicted AKI risk of >8% and that were prescribed on average greater than twice a month (Table 3).

The nephrotoxin-related AKI detection system at SCH automatically places orders for serum creatinines on patients who have met criteria for concomitant nephrotoxin exposure. This has given us a robust database from which to develop our clinical decision-making tool. Both our original and updated systems were based on the absolute number of concomitant nephrotoxic medications prescribed.¹⁶ This is a reasonable approach given the complexity of building a surveillance system¹⁹ and resource limitations. However, a system based on observed rather than theoretical or in vitro data, adaptable to the institution and designed for ongoing refinement, would be more valuable.

The interest in AKI prediction tools continues to be high. Bedford et al. employed numerous variables and diagnostic codes to predict the development of AKI in adults during hospitalization. They were able to produce a prediction model with a reasonable fit (AUC 0.72) to identify patients at higher risk for AKI but were less successful in their attempts to predict progression to severe AKI.²⁰ Hodgson et al. recently developed an adult AKI prediction score (AUC 0.65-0.72) also based on numerous clinical factors that was able to positively impact inpatient mortality.²¹ To our knowledge, our model is unique in that it focuses on nephrotoxins using a predicted AKI risk algorithm based on observed AKI rates of previously ordered medications/medication combinations (two to 11 medications). Having a decision tool targeting medications gives the clinician guidance that can be used to make a specific intervention rather than identifying a patient at risk due to a diagnosis code or other difficult to modify factors.

There are abundant case studies and reports using logistic regression models identifying specific medications associated with AKI. Our choice of methodology was based on our assessment that logistic regression models would be inadequate for the development of a real-time clinical decision-making tool for several reasons. Using logistic regression to explore every medication combination based on our medication list would be challenging as there are approximately 5.5 × 10¹⁰ potential medication combinations. Additionally, logistic regression ignores any potential interactions between the medications. This is an important point as medication interactions can be synergistic, neutral, or antagonist. Consequently, the outcome generated from a set of combined variables may be different from one generated from the sum of each variable taken independently. Logistic regression also does not account for the potential prescribing trends among providers as it assumes that all medications or medication combinations are equally available at the same time. However, in practice, depending on numerous factors, such as hospital culture (eg, the presence of clinical standard work pathways), local bacterial resistance patterns, or medication shortages; certain medication combinations may occur more frequently while others not at all. Finally, logistic regression cannot account for the possibility of a medication combination occurring; therefore, logistic regression may identify a combination strongly associated with AKI that is rarely prescribed.

We theorized that AKI detection would improve with the Algorithm 2 modifications, including the expanded nephrotoxin list, which accompanied alignment with the national pediatric AKI collaborative, NINJA. The finding that our surveillance sensitivity did not improve with this system update supported our subsequent objective to develop a novel nephrotoxin-related AKI decision tool or detection system using our EHR data to identify which specific medications and/or medication combinations were associated with a higher rate of AKI. However, it should be noted that two factors related to measurement bias introduce limitations to our sensitivity and specificity analyses. First, regarding the presence of the alert system, our system will order serum creatinines on patients when they have been exposed to nephrotoxins. Consequently, the proportion of patients with creatinines measured will increase in the nephrotoxin-exposed patients. Unexposed patients may have AKI that is not detected because creatinines may not be ordered. Therefore, there is the potential for a relative increase in AKI detection among nephrotoxin-exposed patients as compared with unexposed patients, which would then affect the measured sensitivity and specificity of the alert. Second, the automated alerts require a baseline creatinine in order to trigger therefore are unable to identify AKI among patients who do not have a baseline serum creatinine measurement.

Our new nephrotoxin-related AKI detection model performed best when an alert was triggered for those medications or medication combinations with a predicted AKI risk of >8%. Forty-six medication combinations consisting of exactly two medications were determined to have a predicted AKI risk of >8% therefore would trigger an alert in our new model system. These medication combinations would not have triggered an alert using either of the previous system algorithms as both algorithms are based on the presence of three or more concomitant nephrotoxic medications.

From the list of suspected nephrotoxins, we identified 11 unique medications in 10 different combinations with a predicted AKI risk of >8% that were prescribed frequently (at least twice a month on average; Table 3). Notably, six out of 10 medication combinations involved vancomycin. Piperacillin-tazobactam was also represented in several combinations. These findings support the concern that others have reported regarding these two medications particularly when prescribed together.^22,23

Acknowledgments

Dr. Karyn E. Yonekawa conceptualized and designed the study, directed the data analysis, interpreted the data, drafted, revised and gave final approval of the manuscript. Dr. Chuan Zhou contributed to the study design, acquired data, conducted the data analysis, critically reviewed, and gave final approval of the manuscript. Ms. Wren L. Haaland contributed to the study design, acquired data, conducted the data analysis, critically reviewed, and gave final approval of the manuscript. Dr. Davene R. Wright contributed to the study design, data analysis, critically reviewed, revised, and gave final approval of the manuscript.

The authors would like to thank Holly Clifton and Suzanne Spencer for their assistance with data acquisition and Drs. Derya Caglar, Corrie McDaniel, and Thida Ong for their writing support.

All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

Disclosures

The authors have no conflicts of interest to report.

Acute kidney injury (AKI) is increasingly common in the hospitalized patient^1,2 with recent adult and pediatric multinational studies reporting AKI rates of 57% and 27%, respectively.^3,4 The development of AKI is associated with significant adverse outcomes including an increased risk of mortality.^5-7 For those that survive, the history of AKI may contribute to a lifetime of impaired health with chronic kidney disease.^8,9 This is particularly concerning for pediatric patients as AKI may impact morbidity for many decades, influence available therapies for these morbidities, and ultimately contribute to a shortened lifespan.¹⁰

AKI in the hospitalized patient is no longer accepted as an unfortunate and unavoidable consequence of illness or the indicated therapy. Currently, there is strong interest in this hospital-acquired condition with global initiatives aimed at increased prevention and early detection and treatment of AKI.^11,12 To this objective, risk stratification tools or prediction models could assist clinicians in decision making. Numerous studies have tested AKI prediction models either in particular high-risk populations or based on associated comorbidities, biomarkers, and critical illness scores. These studies are predominantly in adult populations, and few have been externally validated.¹³ While associations between certain medications and AKI are well known, an AKI prediction model that is applicable to pediatric or adult populations and is based on medication exposure is difficult. However, there is a growing recognition of the potential to develop such a model using the electronic health record (EHR).¹⁴

In 2013, Seattle Children’s Hospital (SCH) implemented a nephrotoxin and AKI detection system to assist in clinical decision making within the EHR. This system instituted the automatic ordering of serum creatinines to screen for AKI when the provider ordered three or more medications that were suspected to be nephrotoxic. Other clinical factors such as the diagnoses or preexisting conditions were not considered in the decision-tool algorithm. This original algorithm (Algorithm 1) was later modified and the list of suspected nephrotoxins was expanded (Table 1) in order to align with a national pediatric AKI collaborative (Algorithm 2). However, it was unclear whether the algorithm modification would improve AKI detection.

Context

SCH is a tertiary care academic teaching hospital affiliated with the University of Washington School of Medicine, Harborview Medical Center, and the Seattle Cancer Care Alliance. The hospital has 371 licensed beds and approximately 18 medical subspecialty services.

Study Population

This was a retrospective cohort study examining all patients ages 0-21 years admitted to SCH between December 1, 2013 and November 30, 2015. The detection system was modified to align with the national pediatric AKI collaborative, Nephrotoxic Injury Negated by Just-in-Time Action (NINJA) in November 2014. Both acute care and intensive care patients were included (data not separated by location). Patients who had end-stage kidney disease and were receiving dialysis and patients who were evaluated in the emergency department without being admitted or admitted as observation status were excluded from analysis. Patients were also excluded if they did not have a baseline serum creatinine as defined below.

Study Measures

AKI is defined at SCH using the Kidney Disease: Improving Global Outcomes Stage 1 criteria as a guideline. The diagnosis of AKI is based on an increase in the baseline serum creatinine by 0.3 mg/dL or an increase in the serum creatinine by >1.5 times the baseline assuming the incoming creatinine is 0.5 mg/dL or higher. For our definition, the increase in serum creatinine needs to have occurred within a one-week timeframe and urine output is not a diagnostic criterion.¹⁵ Baseline serum creatinine is defined as the lowest serum creatinine in the previous six months. Forty medications were classified as nephrotoxins based on previous analysis¹⁶ and adapted for our institutional formulary.

Statistical Analysis

To evaluate the efficacy of our systems in detecting nephrotoxin-related AKI, the sensitivity and the specificity using both our original algorithm (Algorithm 1) and the modified algorithm (Algorithm 2) were generated on our complete data set. To test sensitivity, the proportion of AKI patients who would trigger alert using Algorithm 1 and then with Algorithm 2 was identified. Similarly, to test specificity, the proportion of non-AKI patients who did not trigger an alert by the surveillance systems was identified. The differences in sensitivity and specificity between the two algorithms were evaluated using two-sample tests of proportion.

The statistical method of Combinatorial Inference has been utilized in studies of cancer biology¹⁷ and in genomics.¹⁸ A variation of this approach was used in this study to identify the specific medication combinations most associated with AKI. First, all of the nephrotoxic medications and medication combinations that were prescribed during our study period were identified from a data set (ie, a training set) containing 75% of all encounters selected at random without replacement. Using this training set, the prevalence of each medication combination and the rate of AKI associated with each combination were identified. The predicted overall AKI risk of an individual medication is the average of all the AKI rates associated with each combination containing that specific medication. Also incorporated into the determination of the predicted AKI risk was the prevalence of that medication combination.

To test our model’s predictive capability, the algorithm was applied to the remaining 25% of the total patient data (ie, the test set). The predicted AKI risk was compared with the actual AKI rate in the test data set. Our model’s predictive capability was represented in a receiver operator characteristic (ROC) analysis. The goal was to achieve an area under the ROC curve (AUC) approaching one as this would reflect 100% sensitivity and 100% specificity, whereas an AUC of 0.5 would represent a random guess (50% chance of being correct).

Lastly, our final step was to use our model’s ROC curve to determine an optimal threshold of AKI risk for which to trigger an alert. This predicted risk threshold was based on our goal to increase our surveillance system’s sensitivity balanced with maintaining an acceptable specificity.

An a priori threshold of P = .05 was used to determine statistical significance of all results. Analyses were conducted in Stata 12.1 (StataCorp LP, College Station, Texas) and R 3.3.2 (R Foundation for Statistical Computing, Vienna, Austria). A sample data set containing replication code for our model can be found in an online repository (https://dataverse.harvard.edu/dataverse/chuan). This study was approved by the Seattle Children’s Institutional Review Board.

Sensitivity and Specificity

Of the patient encounters, 14,779 were eligible during the study period. The sensitivity of the system’s ability to identify nephrotoxin-related AKI decreased from 46.9% using Algorithm 1 to 43.3% using Algorithm 2, a change of 3.6% (P = .22). The specificity increased from 73.6% to 89.3%, a change of 15.7% (P < .001; Table 2).

Improvement of Our Nephrotoxin-Related AKI Detection System Using a Novel AKI Prediction Strategy

A total of 838 medication combinations were identified in our training set and the predicted AKI risk for every medication combination was determined. By comparing the predicted risk of AKI to the actual AKI occurrence, an ROC curve with an AUC of 0.756 (Figure) was generated. An increase in system sensitivity was prioritized when determining the optimal AKI risk at which the model would trigger an alert. Setting an alert threshold at a predicted AKI risk of >8%, our model performed with a sensitivity of 74% while decreasing the specificity to 70%.

Identification of High-Risk Nephrotoxic Medications and Medication Combinations

Approximately 200 medication combinations were associated with >8% AKI risk, our new AKI prediction model’s alert threshold. Medication combinations consisting of up to 11 concomitantly prescribed medications were present in our data set. However, many of these combinations were infrequently prescribed. Further analysis, conducted in order to increase the clinical relevance of our findings, identified 10 medications or medication combinations that were both associated with a predicted AKI risk of >8% and that were prescribed on average greater than twice a month (Table 3).

The nephrotoxin-related AKI detection system at SCH automatically places orders for serum creatinines on patients who have met criteria for concomitant nephrotoxin exposure. This has given us a robust database from which to develop our clinical decision-making tool. Both our original and updated systems were based on the absolute number of concomitant nephrotoxic medications prescribed.¹⁶ This is a reasonable approach given the complexity of building a surveillance system¹⁹ and resource limitations. However, a system based on observed rather than theoretical or in vitro data, adaptable to the institution and designed for ongoing refinement, would be more valuable.

The interest in AKI prediction tools continues to be high. Bedford et al. employed numerous variables and diagnostic codes to predict the development of AKI in adults during hospitalization. They were able to produce a prediction model with a reasonable fit (AUC 0.72) to identify patients at higher risk for AKI but were less successful in their attempts to predict progression to severe AKI.²⁰ Hodgson et al. recently developed an adult AKI prediction score (AUC 0.65-0.72) also based on numerous clinical factors that was able to positively impact inpatient mortality.²¹ To our knowledge, our model is unique in that it focuses on nephrotoxins using a predicted AKI risk algorithm based on observed AKI rates of previously ordered medications/medication combinations (two to 11 medications). Having a decision tool targeting medications gives the clinician guidance that can be used to make a specific intervention rather than identifying a patient at risk due to a diagnosis code or other difficult to modify factors.

There are abundant case studies and reports using logistic regression models identifying specific medications associated with AKI. Our choice of methodology was based on our assessment that logistic regression models would be inadequate for the development of a real-time clinical decision-making tool for several reasons. Using logistic regression to explore every medication combination based on our medication list would be challenging as there are approximately 5.5 × 10¹⁰ potential medication combinations. Additionally, logistic regression ignores any potential interactions between the medications. This is an important point as medication interactions can be synergistic, neutral, or antagonist. Consequently, the outcome generated from a set of combined variables may be different from one generated from the sum of each variable taken independently. Logistic regression also does not account for the potential prescribing trends among providers as it assumes that all medications or medication combinations are equally available at the same time. However, in practice, depending on numerous factors, such as hospital culture (eg, the presence of clinical standard work pathways), local bacterial resistance patterns, or medication shortages; certain medication combinations may occur more frequently while others not at all. Finally, logistic regression cannot account for the possibility of a medication combination occurring; therefore, logistic regression may identify a combination strongly associated with AKI that is rarely prescribed.

We theorized that AKI detection would improve with the Algorithm 2 modifications, including the expanded nephrotoxin list, which accompanied alignment with the national pediatric AKI collaborative, NINJA. The finding that our surveillance sensitivity did not improve with this system update supported our subsequent objective to develop a novel nephrotoxin-related AKI decision tool or detection system using our EHR data to identify which specific medications and/or medication combinations were associated with a higher rate of AKI. However, it should be noted that two factors related to measurement bias introduce limitations to our sensitivity and specificity analyses. First, regarding the presence of the alert system, our system will order serum creatinines on patients when they have been exposed to nephrotoxins. Consequently, the proportion of patients with creatinines measured will increase in the nephrotoxin-exposed patients. Unexposed patients may have AKI that is not detected because creatinines may not be ordered. Therefore, there is the potential for a relative increase in AKI detection among nephrotoxin-exposed patients as compared with unexposed patients, which would then affect the measured sensitivity and specificity of the alert. Second, the automated alerts require a baseline creatinine in order to trigger therefore are unable to identify AKI among patients who do not have a baseline serum creatinine measurement.

Our new nephrotoxin-related AKI detection model performed best when an alert was triggered for those medications or medication combinations with a predicted AKI risk of >8%. Forty-six medication combinations consisting of exactly two medications were determined to have a predicted AKI risk of >8% therefore would trigger an alert in our new model system. These medication combinations would not have triggered an alert using either of the previous system algorithms as both algorithms are based on the presence of three or more concomitant nephrotoxic medications.

From the list of suspected nephrotoxins, we identified 11 unique medications in 10 different combinations with a predicted AKI risk of >8% that were prescribed frequently (at least twice a month on average; Table 3). Notably, six out of 10 medication combinations involved vancomycin. Piperacillin-tazobactam was also represented in several combinations. These findings support the concern that others have reported regarding these two medications particularly when prescribed together.^22,23

Acknowledgments

Dr. Karyn E. Yonekawa conceptualized and designed the study, directed the data analysis, interpreted the data, drafted, revised and gave final approval of the manuscript. Dr. Chuan Zhou contributed to the study design, acquired data, conducted the data analysis, critically reviewed, and gave final approval of the manuscript. Ms. Wren L. Haaland contributed to the study design, acquired data, conducted the data analysis, critically reviewed, and gave final approval of the manuscript. Dr. Davene R. Wright contributed to the study design, data analysis, critically reviewed, revised, and gave final approval of the manuscript.

The authors would like to thank Holly Clifton and Suzanne Spencer for their assistance with data acquisition and Drs. Derya Caglar, Corrie McDaniel, and Thida Ong for their writing support.

All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

Disclosures

The authors have no conflicts of interest to report.

Acute kidney injury (AKI) is increasingly common in the hospitalized patient^1,2 with recent adult and pediatric multinational studies reporting AKI rates of 57% and 27%, respectively.^3,4 The development of AKI is associated with significant adverse outcomes including an increased risk of mortality.^5-7 For those that survive, the history of AKI may contribute to a lifetime of impaired health with chronic kidney disease.^8,9 This is particularly concerning for pediatric patients as AKI may impact morbidity for many decades, influence available therapies for these morbidities, and ultimately contribute to a shortened lifespan.¹⁰

AKI in the hospitalized patient is no longer accepted as an unfortunate and unavoidable consequence of illness or the indicated therapy. Currently, there is strong interest in this hospital-acquired condition with global initiatives aimed at increased prevention and early detection and treatment of AKI.^11,12 To this objective, risk stratification tools or prediction models could assist clinicians in decision making. Numerous studies have tested AKI prediction models either in particular high-risk populations or based on associated comorbidities, biomarkers, and critical illness scores. These studies are predominantly in adult populations, and few have been externally validated.¹³ While associations between certain medications and AKI are well known, an AKI prediction model that is applicable to pediatric or adult populations and is based on medication exposure is difficult. However, there is a growing recognition of the potential to develop such a model using the electronic health record (EHR).¹⁴

In 2013, Seattle Children’s Hospital (SCH) implemented a nephrotoxin and AKI detection system to assist in clinical decision making within the EHR. This system instituted the automatic ordering of serum creatinines to screen for AKI when the provider ordered three or more medications that were suspected to be nephrotoxic. Other clinical factors such as the diagnoses or preexisting conditions were not considered in the decision-tool algorithm. This original algorithm (Algorithm 1) was later modified and the list of suspected nephrotoxins was expanded (Table 1) in order to align with a national pediatric AKI collaborative (Algorithm 2). However, it was unclear whether the algorithm modification would improve AKI detection.

Context

SCH is a tertiary care academic teaching hospital affiliated with the University of Washington School of Medicine, Harborview Medical Center, and the Seattle Cancer Care Alliance. The hospital has 371 licensed beds and approximately 18 medical subspecialty services.

Study Population

This was a retrospective cohort study examining all patients ages 0-21 years admitted to SCH between December 1, 2013 and November 30, 2015. The detection system was modified to align with the national pediatric AKI collaborative, Nephrotoxic Injury Negated by Just-in-Time Action (NINJA) in November 2014. Both acute care and intensive care patients were included (data not separated by location). Patients who had end-stage kidney disease and were receiving dialysis and patients who were evaluated in the emergency department without being admitted or admitted as observation status were excluded from analysis. Patients were also excluded if they did not have a baseline serum creatinine as defined below.

Study Measures

AKI is defined at SCH using the Kidney Disease: Improving Global Outcomes Stage 1 criteria as a guideline. The diagnosis of AKI is based on an increase in the baseline serum creatinine by 0.3 mg/dL or an increase in the serum creatinine by >1.5 times the baseline assuming the incoming creatinine is 0.5 mg/dL or higher. For our definition, the increase in serum creatinine needs to have occurred within a one-week timeframe and urine output is not a diagnostic criterion.¹⁵ Baseline serum creatinine is defined as the lowest serum creatinine in the previous six months. Forty medications were classified as nephrotoxins based on previous analysis¹⁶ and adapted for our institutional formulary.

Statistical Analysis

To evaluate the efficacy of our systems in detecting nephrotoxin-related AKI, the sensitivity and the specificity using both our original algorithm (Algorithm 1) and the modified algorithm (Algorithm 2) were generated on our complete data set. To test sensitivity, the proportion of AKI patients who would trigger alert using Algorithm 1 and then with Algorithm 2 was identified. Similarly, to test specificity, the proportion of non-AKI patients who did not trigger an alert by the surveillance systems was identified. The differences in sensitivity and specificity between the two algorithms were evaluated using two-sample tests of proportion.

The statistical method of Combinatorial Inference has been utilized in studies of cancer biology¹⁷ and in genomics.¹⁸ A variation of this approach was used in this study to identify the specific medication combinations most associated with AKI. First, all of the nephrotoxic medications and medication combinations that were prescribed during our study period were identified from a data set (ie, a training set) containing 75% of all encounters selected at random without replacement. Using this training set, the prevalence of each medication combination and the rate of AKI associated with each combination were identified. The predicted overall AKI risk of an individual medication is the average of all the AKI rates associated with each combination containing that specific medication. Also incorporated into the determination of the predicted AKI risk was the prevalence of that medication combination.

To test our model’s predictive capability, the algorithm was applied to the remaining 25% of the total patient data (ie, the test set). The predicted AKI risk was compared with the actual AKI rate in the test data set. Our model’s predictive capability was represented in a receiver operator characteristic (ROC) analysis. The goal was to achieve an area under the ROC curve (AUC) approaching one as this would reflect 100% sensitivity and 100% specificity, whereas an AUC of 0.5 would represent a random guess (50% chance of being correct).

Lastly, our final step was to use our model’s ROC curve to determine an optimal threshold of AKI risk for which to trigger an alert. This predicted risk threshold was based on our goal to increase our surveillance system’s sensitivity balanced with maintaining an acceptable specificity.

An a priori threshold of P = .05 was used to determine statistical significance of all results. Analyses were conducted in Stata 12.1 (StataCorp LP, College Station, Texas) and R 3.3.2 (R Foundation for Statistical Computing, Vienna, Austria). A sample data set containing replication code for our model can be found in an online repository (https://dataverse.harvard.edu/dataverse/chuan). This study was approved by the Seattle Children’s Institutional Review Board.

Sensitivity and Specificity

Of the patient encounters, 14,779 were eligible during the study period. The sensitivity of the system’s ability to identify nephrotoxin-related AKI decreased from 46.9% using Algorithm 1 to 43.3% using Algorithm 2, a change of 3.6% (P = .22). The specificity increased from 73.6% to 89.3%, a change of 15.7% (P < .001; Table 2).

Improvement of Our Nephrotoxin-Related AKI Detection System Using a Novel AKI Prediction Strategy

A total of 838 medication combinations were identified in our training set and the predicted AKI risk for every medication combination was determined. By comparing the predicted risk of AKI to the actual AKI occurrence, an ROC curve with an AUC of 0.756 (Figure) was generated. An increase in system sensitivity was prioritized when determining the optimal AKI risk at which the model would trigger an alert. Setting an alert threshold at a predicted AKI risk of >8%, our model performed with a sensitivity of 74% while decreasing the specificity to 70%.

Identification of High-Risk Nephrotoxic Medications and Medication Combinations

Approximately 200 medication combinations were associated with >8% AKI risk, our new AKI prediction model’s alert threshold. Medication combinations consisting of up to 11 concomitantly prescribed medications were present in our data set. However, many of these combinations were infrequently prescribed. Further analysis, conducted in order to increase the clinical relevance of our findings, identified 10 medications or medication combinations that were both associated with a predicted AKI risk of >8% and that were prescribed on average greater than twice a month (Table 3).

The nephrotoxin-related AKI detection system at SCH automatically places orders for serum creatinines on patients who have met criteria for concomitant nephrotoxin exposure. This has given us a robust database from which to develop our clinical decision-making tool. Both our original and updated systems were based on the absolute number of concomitant nephrotoxic medications prescribed.¹⁶ This is a reasonable approach given the complexity of building a surveillance system¹⁹ and resource limitations. However, a system based on observed rather than theoretical or in vitro data, adaptable to the institution and designed for ongoing refinement, would be more valuable.

The interest in AKI prediction tools continues to be high. Bedford et al. employed numerous variables and diagnostic codes to predict the development of AKI in adults during hospitalization. They were able to produce a prediction model with a reasonable fit (AUC 0.72) to identify patients at higher risk for AKI but were less successful in their attempts to predict progression to severe AKI.²⁰ Hodgson et al. recently developed an adult AKI prediction score (AUC 0.65-0.72) also based on numerous clinical factors that was able to positively impact inpatient mortality.²¹ To our knowledge, our model is unique in that it focuses on nephrotoxins using a predicted AKI risk algorithm based on observed AKI rates of previously ordered medications/medication combinations (two to 11 medications). Having a decision tool targeting medications gives the clinician guidance that can be used to make a specific intervention rather than identifying a patient at risk due to a diagnosis code or other difficult to modify factors.

There are abundant case studies and reports using logistic regression models identifying specific medications associated with AKI. Our choice of methodology was based on our assessment that logistic regression models would be inadequate for the development of a real-time clinical decision-making tool for several reasons. Using logistic regression to explore every medication combination based on our medication list would be challenging as there are approximately 5.5 × 10¹⁰ potential medication combinations. Additionally, logistic regression ignores any potential interactions between the medications. This is an important point as medication interactions can be synergistic, neutral, or antagonist. Consequently, the outcome generated from a set of combined variables may be different from one generated from the sum of each variable taken independently. Logistic regression also does not account for the potential prescribing trends among providers as it assumes that all medications or medication combinations are equally available at the same time. However, in practice, depending on numerous factors, such as hospital culture (eg, the presence of clinical standard work pathways), local bacterial resistance patterns, or medication shortages; certain medication combinations may occur more frequently while others not at all. Finally, logistic regression cannot account for the possibility of a medication combination occurring; therefore, logistic regression may identify a combination strongly associated with AKI that is rarely prescribed.

We theorized that AKI detection would improve with the Algorithm 2 modifications, including the expanded nephrotoxin list, which accompanied alignment with the national pediatric AKI collaborative, NINJA. The finding that our surveillance sensitivity did not improve with this system update supported our subsequent objective to develop a novel nephrotoxin-related AKI decision tool or detection system using our EHR data to identify which specific medications and/or medication combinations were associated with a higher rate of AKI. However, it should be noted that two factors related to measurement bias introduce limitations to our sensitivity and specificity analyses. First, regarding the presence of the alert system, our system will order serum creatinines on patients when they have been exposed to nephrotoxins. Consequently, the proportion of patients with creatinines measured will increase in the nephrotoxin-exposed patients. Unexposed patients may have AKI that is not detected because creatinines may not be ordered. Therefore, there is the potential for a relative increase in AKI detection among nephrotoxin-exposed patients as compared with unexposed patients, which would then affect the measured sensitivity and specificity of the alert. Second, the automated alerts require a baseline creatinine in order to trigger therefore are unable to identify AKI among patients who do not have a baseline serum creatinine measurement.

Our new nephrotoxin-related AKI detection model performed best when an alert was triggered for those medications or medication combinations with a predicted AKI risk of >8%. Forty-six medication combinations consisting of exactly two medications were determined to have a predicted AKI risk of >8% therefore would trigger an alert in our new model system. These medication combinations would not have triggered an alert using either of the previous system algorithms as both algorithms are based on the presence of three or more concomitant nephrotoxic medications.

From the list of suspected nephrotoxins, we identified 11 unique medications in 10 different combinations with a predicted AKI risk of >8% that were prescribed frequently (at least twice a month on average; Table 3). Notably, six out of 10 medication combinations involved vancomycin. Piperacillin-tazobactam was also represented in several combinations. These findings support the concern that others have reported regarding these two medications particularly when prescribed together.^22,23

Acknowledgments

Dr. Karyn E. Yonekawa conceptualized and designed the study, directed the data analysis, interpreted the data, drafted, revised and gave final approval of the manuscript. Dr. Chuan Zhou contributed to the study design, acquired data, conducted the data analysis, critically reviewed, and gave final approval of the manuscript. Ms. Wren L. Haaland contributed to the study design, acquired data, conducted the data analysis, critically reviewed, and gave final approval of the manuscript. Dr. Davene R. Wright contributed to the study design, data analysis, critically reviewed, revised, and gave final approval of the manuscript.

The authors would like to thank Holly Clifton and Suzanne Spencer for their assistance with data acquisition and Drs. Derya Caglar, Corrie McDaniel, and Thida Ong for their writing support.

All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

Disclosures

The authors have no conflicts of interest to report.

Acute respiratory illnesses (ARIs), including acute exacerbations of asthma, croup, pneumonia, and bronchiolitis, are among the most common illnesses in childhood.¹ Although most ARIs can be managed in the outpatient setting, hospitalization is common with respiratory illnesses accounting for >425,000 hospitalizations annually.¹ Pneumonia, asthma, and bronchiolitis each rank among the top five reasons for pediatric hospitalization in the United States.¹ Successful efforts to prevent or mitigate the severity of ARIs could have a major impact on child health.

Exposure to secondhand smoke (SHS) is a preventable risk factor for ARI in children, particularly when there is regular exposure in the home.² Chronic exposure to SHS impacts systemic inflammation by suppressing serum interferon-gamma,³ which can lead to increased susceptibility to viral and bacterial infections,⁴ and increasing Th2 (atopic) cytokine expression, which is associated with asthma.⁵ SHS exposure in children has also been linked to diminished lung function.⁶ As a result, SHS exposure is associated with increased ARI susceptibility and severity in children.^7-10

Much research has focused on the clinical impact of SHS exposure on respiratory health in children, but little is known about the impact on patient-reported outcomes, such as health-related quality of life (HRQOL). Patient-reported outcomes help provide a comprehensive evaluation of the effectiveness of healthcare delivery systems. These outcomes are increasingly used by health service researchers to better understand patient and caregiver perspectives.¹¹ Given the known associations between SHS exposure and ARI morbidity, we postulated that regular SHS exposure would also impact HRQOL in children. In this study, we assessed the relationship between SHS exposure and HRQOL within a large, multicenter, prospective cohort of children presenting to the emergency department (ED) and/or hospital with ARI.

Study Population

This study was nested within the Pediatric Respiratory Illness Measurement System (PRIMES) study, a prospective cohort study of children with ARI in the ED and inpatient settings at five tertiary care children’s hospitals within the Pediatric Research in Inpatient Settings Network in Colorado, Pennsylvania, Tennessee, Texas, and Washington. Eligible children were two weeks to 16 years of age hospitalized after presenting to the ED with a primary diagnosis of asthma, croup, bronchiolitis, or pneumonia between July 1, 2014 and June 30, 2016. Because of an anticipated low frequency of croup hospitalizations, we also included children presenting to the ED and then discharged to home with this diagnosis. Children were assigned to a PRIMES diagnosis group based on their final discharge diagnosis. If there was a discrepancy between admission and discharge diagnoses, the discharge diagnosis was used. If a child had more than one discharge diagnosis for a PRIMES condition (eg, acute asthma and pneumonia), we chose the PRIMES condition with the lowest total enrollments overall. If the final discharge diagnosis was not a PRIMES condition, the case was excluded from further analysis. Patients with immunodeficiency, cystic fibrosis, a history of prematurity <32 weeks, chronic neuromuscular disease, cardiovascular disease, pulmonary diseases (other than asthma), and moderate to severe developmental delay were also excluded. Children admitted to intensive care were eligible only if they were transferred to an acute care ward <72 hours following admission. A survey was administered at the time of enrollment that collected information on SHS exposure, HRQOL, healthcare utilization, and demographics. All study procedures were reviewed and approved by the institutional review boards at each of the participating hospitals.

To ascertain SHS exposure, we asked caregivers, “How many persons living in the child’s home smoke?” Responses were dichotomized into non-SHS exposed (0 smokers) and SHS exposed (≥1 smokers). Children with missing data on home SHS exposure were excluded.

Health-Related Quality of Life Outcomes

We estimated HRQOL using the Pediatric Quality of Life (PedsQL^TM) 4.0 Generic Core and Infant Scales. The PedsQL instruments are validated, population HRQOL measures that evaluate the physical, mental, emotional, and social functioning of children two to 18 years old based on self- or caregiver-proxy report.^12-15 These instruments have also shown responsiveness as well as construct and predictive validity in hospitalized children.¹¹ For this study, we focused on the PedsQL physical functioning subscale, which assesses for problems with physical activities (eg, sports activity or exercise, low energy, and hurts or aches) on a five-point Likert scale (never to almost always a problem). Scores range from 0 to 100 with higher scores indicating a better HRQOL. The reported minimal clinically important difference (MCID), defined as the smallest difference in which individuals would perceive a benefit or would necessitate a change in management, for this scale is 4.5 points.^16,17

Children >8 years old were invited to complete the self-report version of the PedsQL. For children <8 years old, and for older children who were unable to complete them, surveys were completed by a parent or legal guardian. Respondents were asked to assess perceptions of their (or their child’s) HRQOL during periods of baseline health (the child’s usual state of health in the month preceding the current illness) and during the acute illness (the child’s state of health at the time of admission) as SHS exposure may influence perceptions of general health and/or contribute to worse outcomes during periods of acute illness.

Covariates collected at the time of enrollment included sociodemographics (child age, gender, race/ethnicity, and caregiver education), and healthcare utilization (caregiver-reported patient visits to a healthcare provider in the preceding six months). Insurance status and level of medical complexity (using the Pediatric Medical Complexity Algorithm)¹⁸ were obtained using the Pediatric Hospital Information System database, an administrative database containing clinical and resource utilization data from >45 children’s hospitals in the United States including all of the PRIMES study hospitals.¹³

Analysis

Descriptive statistics included frequency (%) and mean (standard deviation). Bivariate comparisons according to SHS exposure status were analyzed using chi-squared tests for categorical variables and analysis of variance for continuous variables. Multivariable linear mixed regression models were used to examine associations between home SHS exposure and HRQOL for baseline health and during admission, overall and stratified by diagnosis. Covariates in each model included age, sex, race/ethnicity, caregiver education, and healthcare visits in the preceding six months. We also included a hospital random effect to account for clustering of patients within hospitals and used robust standard errors for inference.

In a secondary analysis to explore potential dose-response effects of SHS exposure, we examined associations between an ordinal exposure variable (0 smokers, 1 smoker, ≥2 smokers) and HRQOL for baseline health and during admission for the acute illness. Because of sample size limitations, diagnosis-specific analyses examining dose-response effects were not conducted.

Study Population

Of the 2,334 children enrolled in the PRIMES study, 25 (1%) respondents did not report on home SHS exposure and were excluded, yielding a final study population of 2,309 children, of whom 728 (32%) had reported home SHS exposure. The study population included 664 children with asthma (mean age seven years [3.5]; 38% with home SHS exposure), 740 with bronchiolitis (mean age 0.7 years [0.5]; 32% with home SHS exposure), 342 with croup (mean age 1.7 [1.1]; 25% with home SHS exposure), and 563 with pneumonia (mean age 4.4 [3.8]; 27% with home SHS exposure; Table 1). Compared with non-SHS-exposed children, those with home SHS exposure tend to be slightly older (3.9 vs 3.4 years, P = .01), more likely to be non-Hispanic Black (29% vs 19%, P < .001), to have a chronic condition (52% vs 41%, P < .001), to come from a household where caregiver(s) did not graduate from college (45% vs 29%, P < .001), and to have public insurance (73% vs 49%, P < .001).

Home SHS Exposure and Health-related Quality of Life

The overall mean HRQOL score for baseline health was 83 (15), with a range across diagnoses of 82 to 87. Compared with non-SHS-exposed children, children with home SHS exposure had a lower mean HRQOL score for baseline health (adjusted mean difference –3.04 [95% CI -4.34, –1.74]). In analyses stratified by diagnosis, baseline health scores were lower for SHS-exposed children for all four conditions, but differences were statistically significant only for bronchiolitis (adjusted mean difference –2.94 [–5.0, –0.89]) and pneumonia (adjusted mean value –4.13 [–6.82, –1.44]; Table 2); none of these differences met the MCID threshold.

The overall mean HRQOL score at the time of admission was 56 (23), with a range across diagnoses of 49 to 61, with lower scores noted among SHS-exposed children compared with non-SHS-exposed children (adjusted mean difference –2.16 [–4.22, –0.10]). Similar to scores representing baseline health, admission scores were lower across all four conditions for SHS-exposed children. Only children with croup, however, had significantly lower admission scores that also met the MCID threshold (adjusted mean difference –5.71 [–10.67, –0.75]; Table 2).

To assess for potential dose-response effects of SHS exposure on HRQOL, we stratified SHS-exposed children into those with one smoker in the home (n = 513) and those with ≥2 smokers in the home (n = 215). Compared with non-SHS-exposed children, both HRQOL scores (baseline health and admission) were lower for SHS-exposed children. Consistent with a dose-response association, scores were lowest for children with ≥2 smokers in the home, both at baseline health (adjusted mean difference –3.92 [–6.03, –1.81]) and on admission (adjusted mean difference –3.67 [–6.98, –0.36]; Table 3).

Within a multicenter cohort of 2,309 children hospitalized with ARI, we noted significantly lower HRQOL scores among children exposed to SHS in the home as compared with nonexposed children. Differences were greatest for children living with ≥2 smokers in the home. In analyses stratified by diagnosis, differences in baseline health HRQOL scores were greatest for children with bronchiolitis and pneumonia. Differences in acute illness scores were greatest for children with croup.¹⁶

Our study provides evidence for acute and chronic impacts of SHS on HRQOL in children hospitalized with ARI. Although several studies have linked SHS exposure to reduced HRQOL in adults,^19,20 few similar studies have been conducted in children. Nonetheless, a wealth of studies have documented the negative impact of SHS exposure on clinical outcomes among children with ARI.^8,10,21-23 Our findings that home SHS exposure was associated with reduced HRQOL among our cohort of children with ARI are therefore consistent with related findings in adults and children. The observation that the effects of SHS exposure on HRQOL were greatest among children living with ≥2 smokers provides further evidence of a potential causal link between regular SHS exposure and HRQOL.

Although the magnitude and significance of associations between SHS exposure and HRQOL varied for each of the four diagnoses for baseline health and the acute illness, it is important to note that the point estimates for the adjusted mean differences were uniformly lower for the SHS-exposed children in each subgroup. Even so, only acute illness scores for croup exceeded the MCID threshold.¹⁶ Croup is the only included condition of the upper airway and is characterized by laryngotracheal inflammation leading to the typical cough and, in moderate to severe cases, stridor. Given that chronic SHS exposure induces a proinflammatory state,³ it is possible that SHS-exposed children with croup had more severe illness compared with nonexposed children with croup resulting in lower HRQOL scores on admission. Further, perceived differences in illness severity and HRQOL may be more readily apparent in children with croup (eg, stridor at rest vs intermittent or no stridor) as compared with children with lower respiratory tract diseases.

Of the four included diagnoses, the link between SHS exposure and asthma outcomes has been most studied. Prior work has demonstrated more frequent and severe acute exacerbations, as well as worse long-term lung function among SHS-exposed children as compared with nonexposed children.^22-24 It was, therefore, surprising that our study failed to demonstrate associations between SHS exposure and HRQOL among children with asthma. Reasons for this finding are unclear. One hypothesis is that caregivers of SHS-exposed children with asthma may be more aware of the impacts of SHS exposure on respiratory health (through prior education) and, thus, more likely to modify their smoking behaviors, or for their children to be on daily asthma controller therapy. Alternatively, caregivers of children with asthma may be more likely to underreport home SHS exposure. Thirty-eight percent of children with asthma, however, were classified as SHS-exposed. This percentage was greater than the other three conditions studied (25%-32%), suggesting that differential bias in underreporting was minimal. Given that children with asthma were older, on average, than children with the other three conditions, it may also be that these children spent more time in smoke-free environments (eg, school).

Nearly one-third of children in our study were exposed to SHS in the home. This is similar to the prevalence of exposure in other studies conducted among hospitalized children^8,10,21,25 but higher than the national prevalence of home SHS exposure among children in the United States.²⁶ Thus, hospitalized children represent a particularly vulnerable population and an important target for interventions aiming to reduce exposure to SHS. Although longitudinal interventions are likely necessary to affect long-term success, hospitalization for ARI may serve as a powerful teachable moment to begin cessation efforts. Hospitalization also offers time beyond a typical primary care outpatient encounter to focus on cessation counseling and may be the only opportunity to engage in counseling activities for some families with limited time or access. Further, prior studies have demonstrated both the feasibility and the effectiveness of smoking cessation interventions in hospitalized children.^27-30 Unfortunately, however, SHS exposure is often not documented at the time of hospitalization, and many opportunities to intervene are missed.^25,31 Thus, there is a need for improved strategies to reliably identify and intervene on SHS-exposed children in the hospital setting.

These findings should be considered in the context of several limitations. The observational nature of our study raises the potential for confounding, specifically with regard to socioeconomic status, as this is associated with both SHS exposure and lower HRQOL. Our modeling approach attempted to control for several factors associated with socioeconomic status, including caregiver education and insurance coverage, but there is potential for residual confounding. No single question is sufficient to fully assess SHS exposure as the intensity of home SHS exposure likely varies widely, and some children may be exposed to SHS outside of the home environment.³² The home, however, is often the most likely source of regular SHS exposure,^33,34 especially among young children (our cohort’s mean age was 3.6 years). Misclassification of SHS exposure is also possible due to underreporting of smoking.^35,36 As a result, some children regularly exposed to SHS may have been misclassified as nonexposed, and the observed associations between SHS exposure and HRQOL may be underestimated. Confirming our study’s findings using objective assessments of SHS exposure, such as cotinine, are warranted. Given the young age of our cohort, the PedsQL surveys were completed by the parent or legal guardian only in >90% of the enrolled subjects, and caregiver perceptions may not accurately reflect the child’s perceptions. Prior work, however, has demonstrated the validity of parent-proxy reporting of the PedsQL, including correlation with child self-report.³⁷ In our study, correlation between child and caregiver reporting (when available) was also very good (r = 0.72, 95% CI 0.64, 0.77). It is also possible that the timing of the HRQOL assessments (on admission) may have biased perceptions of baseline HRQOL, although we anticipate any bias would likely be nondifferential between SHS-exposed and nonexposed children and across diagnoses.

Nearly one-third of children in our study were exposed to SHS exposure in the home, and SHS exposure was associated with lower HRQOL for baseline health and during acute illness, providing further evidence of the dangers of SHS. Much work is needed in order to eliminate the impact of SHS on child health and families of children hospitalized for respiratory illness should be considered a priority population for smoking cessation efforts.

Acknowledgment

The authors wish to acknowledge the efforts of PRIS-PRIMES study team. The authors also wish to thank the children and families who consented to be a part of the PRIMES study.

Disclosures

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

Funding

This study was supported by NIH-NHLBI 1R01HL121067 to RMS.

Acute respiratory illnesses (ARIs), including acute exacerbations of asthma, croup, pneumonia, and bronchiolitis, are among the most common illnesses in childhood.¹ Although most ARIs can be managed in the outpatient setting, hospitalization is common with respiratory illnesses accounting for >425,000 hospitalizations annually.¹ Pneumonia, asthma, and bronchiolitis each rank among the top five reasons for pediatric hospitalization in the United States.¹ Successful efforts to prevent or mitigate the severity of ARIs could have a major impact on child health.

Exposure to secondhand smoke (SHS) is a preventable risk factor for ARI in children, particularly when there is regular exposure in the home.² Chronic exposure to SHS impacts systemic inflammation by suppressing serum interferon-gamma,³ which can lead to increased susceptibility to viral and bacterial infections,⁴ and increasing Th2 (atopic) cytokine expression, which is associated with asthma.⁵ SHS exposure in children has also been linked to diminished lung function.⁶ As a result, SHS exposure is associated with increased ARI susceptibility and severity in children.^7-10

Much research has focused on the clinical impact of SHS exposure on respiratory health in children, but little is known about the impact on patient-reported outcomes, such as health-related quality of life (HRQOL). Patient-reported outcomes help provide a comprehensive evaluation of the effectiveness of healthcare delivery systems. These outcomes are increasingly used by health service researchers to better understand patient and caregiver perspectives.¹¹ Given the known associations between SHS exposure and ARI morbidity, we postulated that regular SHS exposure would also impact HRQOL in children. In this study, we assessed the relationship between SHS exposure and HRQOL within a large, multicenter, prospective cohort of children presenting to the emergency department (ED) and/or hospital with ARI.

Study Population

This study was nested within the Pediatric Respiratory Illness Measurement System (PRIMES) study, a prospective cohort study of children with ARI in the ED and inpatient settings at five tertiary care children’s hospitals within the Pediatric Research in Inpatient Settings Network in Colorado, Pennsylvania, Tennessee, Texas, and Washington. Eligible children were two weeks to 16 years of age hospitalized after presenting to the ED with a primary diagnosis of asthma, croup, bronchiolitis, or pneumonia between July 1, 2014 and June 30, 2016. Because of an anticipated low frequency of croup hospitalizations, we also included children presenting to the ED and then discharged to home with this diagnosis. Children were assigned to a PRIMES diagnosis group based on their final discharge diagnosis. If there was a discrepancy between admission and discharge diagnoses, the discharge diagnosis was used. If a child had more than one discharge diagnosis for a PRIMES condition (eg, acute asthma and pneumonia), we chose the PRIMES condition with the lowest total enrollments overall. If the final discharge diagnosis was not a PRIMES condition, the case was excluded from further analysis. Patients with immunodeficiency, cystic fibrosis, a history of prematurity <32 weeks, chronic neuromuscular disease, cardiovascular disease, pulmonary diseases (other than asthma), and moderate to severe developmental delay were also excluded. Children admitted to intensive care were eligible only if they were transferred to an acute care ward <72 hours following admission. A survey was administered at the time of enrollment that collected information on SHS exposure, HRQOL, healthcare utilization, and demographics. All study procedures were reviewed and approved by the institutional review boards at each of the participating hospitals.

To ascertain SHS exposure, we asked caregivers, “How many persons living in the child’s home smoke?” Responses were dichotomized into non-SHS exposed (0 smokers) and SHS exposed (≥1 smokers). Children with missing data on home SHS exposure were excluded.

Health-Related Quality of Life Outcomes

We estimated HRQOL using the Pediatric Quality of Life (PedsQL^TM) 4.0 Generic Core and Infant Scales. The PedsQL instruments are validated, population HRQOL measures that evaluate the physical, mental, emotional, and social functioning of children two to 18 years old based on self- or caregiver-proxy report.^12-15 These instruments have also shown responsiveness as well as construct and predictive validity in hospitalized children.¹¹ For this study, we focused on the PedsQL physical functioning subscale, which assesses for problems with physical activities (eg, sports activity or exercise, low energy, and hurts or aches) on a five-point Likert scale (never to almost always a problem). Scores range from 0 to 100 with higher scores indicating a better HRQOL. The reported minimal clinically important difference (MCID), defined as the smallest difference in which individuals would perceive a benefit or would necessitate a change in management, for this scale is 4.5 points.^16,17

Children >8 years old were invited to complete the self-report version of the PedsQL. For children <8 years old, and for older children who were unable to complete them, surveys were completed by a parent or legal guardian. Respondents were asked to assess perceptions of their (or their child’s) HRQOL during periods of baseline health (the child’s usual state of health in the month preceding the current illness) and during the acute illness (the child’s state of health at the time of admission) as SHS exposure may influence perceptions of general health and/or contribute to worse outcomes during periods of acute illness.

Covariates collected at the time of enrollment included sociodemographics (child age, gender, race/ethnicity, and caregiver education), and healthcare utilization (caregiver-reported patient visits to a healthcare provider in the preceding six months). Insurance status and level of medical complexity (using the Pediatric Medical Complexity Algorithm)¹⁸ were obtained using the Pediatric Hospital Information System database, an administrative database containing clinical and resource utilization data from >45 children’s hospitals in the United States including all of the PRIMES study hospitals.¹³

Analysis

Descriptive statistics included frequency (%) and mean (standard deviation). Bivariate comparisons according to SHS exposure status were analyzed using chi-squared tests for categorical variables and analysis of variance for continuous variables. Multivariable linear mixed regression models were used to examine associations between home SHS exposure and HRQOL for baseline health and during admission, overall and stratified by diagnosis. Covariates in each model included age, sex, race/ethnicity, caregiver education, and healthcare visits in the preceding six months. We also included a hospital random effect to account for clustering of patients within hospitals and used robust standard errors for inference.

In a secondary analysis to explore potential dose-response effects of SHS exposure, we examined associations between an ordinal exposure variable (0 smokers, 1 smoker, ≥2 smokers) and HRQOL for baseline health and during admission for the acute illness. Because of sample size limitations, diagnosis-specific analyses examining dose-response effects were not conducted.

Study Population

Of the 2,334 children enrolled in the PRIMES study, 25 (1%) respondents did not report on home SHS exposure and were excluded, yielding a final study population of 2,309 children, of whom 728 (32%) had reported home SHS exposure. The study population included 664 children with asthma (mean age seven years [3.5]; 38% with home SHS exposure), 740 with bronchiolitis (mean age 0.7 years [0.5]; 32% with home SHS exposure), 342 with croup (mean age 1.7 [1.1]; 25% with home SHS exposure), and 563 with pneumonia (mean age 4.4 [3.8]; 27% with home SHS exposure; Table 1). Compared with non-SHS-exposed children, those with home SHS exposure tend to be slightly older (3.9 vs 3.4 years, P = .01), more likely to be non-Hispanic Black (29% vs 19%, P < .001), to have a chronic condition (52% vs 41%, P < .001), to come from a household where caregiver(s) did not graduate from college (45% vs 29%, P < .001), and to have public insurance (73% vs 49%, P < .001).

Home SHS Exposure and Health-related Quality of Life

The overall mean HRQOL score for baseline health was 83 (15), with a range across diagnoses of 82 to 87. Compared with non-SHS-exposed children, children with home SHS exposure had a lower mean HRQOL score for baseline health (adjusted mean difference –3.04 [95% CI -4.34, –1.74]). In analyses stratified by diagnosis, baseline health scores were lower for SHS-exposed children for all four conditions, but differences were statistically significant only for bronchiolitis (adjusted mean difference –2.94 [–5.0, –0.89]) and pneumonia (adjusted mean value –4.13 [–6.82, –1.44]; Table 2); none of these differences met the MCID threshold.

The overall mean HRQOL score at the time of admission was 56 (23), with a range across diagnoses of 49 to 61, with lower scores noted among SHS-exposed children compared with non-SHS-exposed children (adjusted mean difference –2.16 [–4.22, –0.10]). Similar to scores representing baseline health, admission scores were lower across all four conditions for SHS-exposed children. Only children with croup, however, had significantly lower admission scores that also met the MCID threshold (adjusted mean difference –5.71 [–10.67, –0.75]; Table 2).

To assess for potential dose-response effects of SHS exposure on HRQOL, we stratified SHS-exposed children into those with one smoker in the home (n = 513) and those with ≥2 smokers in the home (n = 215). Compared with non-SHS-exposed children, both HRQOL scores (baseline health and admission) were lower for SHS-exposed children. Consistent with a dose-response association, scores were lowest for children with ≥2 smokers in the home, both at baseline health (adjusted mean difference –3.92 [–6.03, –1.81]) and on admission (adjusted mean difference –3.67 [–6.98, –0.36]; Table 3).

Within a multicenter cohort of 2,309 children hospitalized with ARI, we noted significantly lower HRQOL scores among children exposed to SHS in the home as compared with nonexposed children. Differences were greatest for children living with ≥2 smokers in the home. In analyses stratified by diagnosis, differences in baseline health HRQOL scores were greatest for children with bronchiolitis and pneumonia. Differences in acute illness scores were greatest for children with croup.¹⁶

Our study provides evidence for acute and chronic impacts of SHS on HRQOL in children hospitalized with ARI. Although several studies have linked SHS exposure to reduced HRQOL in adults,^19,20 few similar studies have been conducted in children. Nonetheless, a wealth of studies have documented the negative impact of SHS exposure on clinical outcomes among children with ARI.^8,10,21-23 Our findings that home SHS exposure was associated with reduced HRQOL among our cohort of children with ARI are therefore consistent with related findings in adults and children. The observation that the effects of SHS exposure on HRQOL were greatest among children living with ≥2 smokers provides further evidence of a potential causal link between regular SHS exposure and HRQOL.

Although the magnitude and significance of associations between SHS exposure and HRQOL varied for each of the four diagnoses for baseline health and the acute illness, it is important to note that the point estimates for the adjusted mean differences were uniformly lower for the SHS-exposed children in each subgroup. Even so, only acute illness scores for croup exceeded the MCID threshold.¹⁶ Croup is the only included condition of the upper airway and is characterized by laryngotracheal inflammation leading to the typical cough and, in moderate to severe cases, stridor. Given that chronic SHS exposure induces a proinflammatory state,³ it is possible that SHS-exposed children with croup had more severe illness compared with nonexposed children with croup resulting in lower HRQOL scores on admission. Further, perceived differences in illness severity and HRQOL may be more readily apparent in children with croup (eg, stridor at rest vs intermittent or no stridor) as compared with children with lower respiratory tract diseases.

Of the four included diagnoses, the link between SHS exposure and asthma outcomes has been most studied. Prior work has demonstrated more frequent and severe acute exacerbations, as well as worse long-term lung function among SHS-exposed children as compared with nonexposed children.^22-24 It was, therefore, surprising that our study failed to demonstrate associations between SHS exposure and HRQOL among children with asthma. Reasons for this finding are unclear. One hypothesis is that caregivers of SHS-exposed children with asthma may be more aware of the impacts of SHS exposure on respiratory health (through prior education) and, thus, more likely to modify their smoking behaviors, or for their children to be on daily asthma controller therapy. Alternatively, caregivers of children with asthma may be more likely to underreport home SHS exposure. Thirty-eight percent of children with asthma, however, were classified as SHS-exposed. This percentage was greater than the other three conditions studied (25%-32%), suggesting that differential bias in underreporting was minimal. Given that children with asthma were older, on average, than children with the other three conditions, it may also be that these children spent more time in smoke-free environments (eg, school).

Nearly one-third of children in our study were exposed to SHS in the home. This is similar to the prevalence of exposure in other studies conducted among hospitalized children^8,10,21,25 but higher than the national prevalence of home SHS exposure among children in the United States.²⁶ Thus, hospitalized children represent a particularly vulnerable population and an important target for interventions aiming to reduce exposure to SHS. Although longitudinal interventions are likely necessary to affect long-term success, hospitalization for ARI may serve as a powerful teachable moment to begin cessation efforts. Hospitalization also offers time beyond a typical primary care outpatient encounter to focus on cessation counseling and may be the only opportunity to engage in counseling activities for some families with limited time or access. Further, prior studies have demonstrated both the feasibility and the effectiveness of smoking cessation interventions in hospitalized children.^27-30 Unfortunately, however, SHS exposure is often not documented at the time of hospitalization, and many opportunities to intervene are missed.^25,31 Thus, there is a need for improved strategies to reliably identify and intervene on SHS-exposed children in the hospital setting.

These findings should be considered in the context of several limitations. The observational nature of our study raises the potential for confounding, specifically with regard to socioeconomic status, as this is associated with both SHS exposure and lower HRQOL. Our modeling approach attempted to control for several factors associated with socioeconomic status, including caregiver education and insurance coverage, but there is potential for residual confounding. No single question is sufficient to fully assess SHS exposure as the intensity of home SHS exposure likely varies widely, and some children may be exposed to SHS outside of the home environment.³² The home, however, is often the most likely source of regular SHS exposure,^33,34 especially among young children (our cohort’s mean age was 3.6 years). Misclassification of SHS exposure is also possible due to underreporting of smoking.^35,36 As a result, some children regularly exposed to SHS may have been misclassified as nonexposed, and the observed associations between SHS exposure and HRQOL may be underestimated. Confirming our study’s findings using objective assessments of SHS exposure, such as cotinine, are warranted. Given the young age of our cohort, the PedsQL surveys were completed by the parent or legal guardian only in >90% of the enrolled subjects, and caregiver perceptions may not accurately reflect the child’s perceptions. Prior work, however, has demonstrated the validity of parent-proxy reporting of the PedsQL, including correlation with child self-report.³⁷ In our study, correlation between child and caregiver reporting (when available) was also very good (r = 0.72, 95% CI 0.64, 0.77). It is also possible that the timing of the HRQOL assessments (on admission) may have biased perceptions of baseline HRQOL, although we anticipate any bias would likely be nondifferential between SHS-exposed and nonexposed children and across diagnoses.

Nearly one-third of children in our study were exposed to SHS exposure in the home, and SHS exposure was associated with lower HRQOL for baseline health and during acute illness, providing further evidence of the dangers of SHS. Much work is needed in order to eliminate the impact of SHS on child health and families of children hospitalized for respiratory illness should be considered a priority population for smoking cessation efforts.

Acknowledgment

The authors wish to acknowledge the efforts of PRIS-PRIMES study team. The authors also wish to thank the children and families who consented to be a part of the PRIMES study.

Disclosures

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

Funding

This study was supported by NIH-NHLBI 1R01HL121067 to RMS.

Acute respiratory illnesses (ARIs), including acute exacerbations of asthma, croup, pneumonia, and bronchiolitis, are among the most common illnesses in childhood.¹ Although most ARIs can be managed in the outpatient setting, hospitalization is common with respiratory illnesses accounting for >425,000 hospitalizations annually.¹ Pneumonia, asthma, and bronchiolitis each rank among the top five reasons for pediatric hospitalization in the United States.¹ Successful efforts to prevent or mitigate the severity of ARIs could have a major impact on child health.

Exposure to secondhand smoke (SHS) is a preventable risk factor for ARI in children, particularly when there is regular exposure in the home.² Chronic exposure to SHS impacts systemic inflammation by suppressing serum interferon-gamma,³ which can lead to increased susceptibility to viral and bacterial infections,⁴ and increasing Th2 (atopic) cytokine expression, which is associated with asthma.⁵ SHS exposure in children has also been linked to diminished lung function.⁶ As a result, SHS exposure is associated with increased ARI susceptibility and severity in children.^7-10

Much research has focused on the clinical impact of SHS exposure on respiratory health in children, but little is known about the impact on patient-reported outcomes, such as health-related quality of life (HRQOL). Patient-reported outcomes help provide a comprehensive evaluation of the effectiveness of healthcare delivery systems. These outcomes are increasingly used by health service researchers to better understand patient and caregiver perspectives.¹¹ Given the known associations between SHS exposure and ARI morbidity, we postulated that regular SHS exposure would also impact HRQOL in children. In this study, we assessed the relationship between SHS exposure and HRQOL within a large, multicenter, prospective cohort of children presenting to the emergency department (ED) and/or hospital with ARI.

Study Population

This study was nested within the Pediatric Respiratory Illness Measurement System (PRIMES) study, a prospective cohort study of children with ARI in the ED and inpatient settings at five tertiary care children’s hospitals within the Pediatric Research in Inpatient Settings Network in Colorado, Pennsylvania, Tennessee, Texas, and Washington. Eligible children were two weeks to 16 years of age hospitalized after presenting to the ED with a primary diagnosis of asthma, croup, bronchiolitis, or pneumonia between July 1, 2014 and June 30, 2016. Because of an anticipated low frequency of croup hospitalizations, we also included children presenting to the ED and then discharged to home with this diagnosis. Children were assigned to a PRIMES diagnosis group based on their final discharge diagnosis. If there was a discrepancy between admission and discharge diagnoses, the discharge diagnosis was used. If a child had more than one discharge diagnosis for a PRIMES condition (eg, acute asthma and pneumonia), we chose the PRIMES condition with the lowest total enrollments overall. If the final discharge diagnosis was not a PRIMES condition, the case was excluded from further analysis. Patients with immunodeficiency, cystic fibrosis, a history of prematurity <32 weeks, chronic neuromuscular disease, cardiovascular disease, pulmonary diseases (other than asthma), and moderate to severe developmental delay were also excluded. Children admitted to intensive care were eligible only if they were transferred to an acute care ward <72 hours following admission. A survey was administered at the time of enrollment that collected information on SHS exposure, HRQOL, healthcare utilization, and demographics. All study procedures were reviewed and approved by the institutional review boards at each of the participating hospitals.

To ascertain SHS exposure, we asked caregivers, “How many persons living in the child’s home smoke?” Responses were dichotomized into non-SHS exposed (0 smokers) and SHS exposed (≥1 smokers). Children with missing data on home SHS exposure were excluded.

Health-Related Quality of Life Outcomes

We estimated HRQOL using the Pediatric Quality of Life (PedsQL^TM) 4.0 Generic Core and Infant Scales. The PedsQL instruments are validated, population HRQOL measures that evaluate the physical, mental, emotional, and social functioning of children two to 18 years old based on self- or caregiver-proxy report.^12-15 These instruments have also shown responsiveness as well as construct and predictive validity in hospitalized children.¹¹ For this study, we focused on the PedsQL physical functioning subscale, which assesses for problems with physical activities (eg, sports activity or exercise, low energy, and hurts or aches) on a five-point Likert scale (never to almost always a problem). Scores range from 0 to 100 with higher scores indicating a better HRQOL. The reported minimal clinically important difference (MCID), defined as the smallest difference in which individuals would perceive a benefit or would necessitate a change in management, for this scale is 4.5 points.^16,17

Children >8 years old were invited to complete the self-report version of the PedsQL. For children <8 years old, and for older children who were unable to complete them, surveys were completed by a parent or legal guardian. Respondents were asked to assess perceptions of their (or their child’s) HRQOL during periods of baseline health (the child’s usual state of health in the month preceding the current illness) and during the acute illness (the child’s state of health at the time of admission) as SHS exposure may influence perceptions of general health and/or contribute to worse outcomes during periods of acute illness.

Covariates collected at the time of enrollment included sociodemographics (child age, gender, race/ethnicity, and caregiver education), and healthcare utilization (caregiver-reported patient visits to a healthcare provider in the preceding six months). Insurance status and level of medical complexity (using the Pediatric Medical Complexity Algorithm)¹⁸ were obtained using the Pediatric Hospital Information System database, an administrative database containing clinical and resource utilization data from >45 children’s hospitals in the United States including all of the PRIMES study hospitals.¹³

Analysis

Descriptive statistics included frequency (%) and mean (standard deviation). Bivariate comparisons according to SHS exposure status were analyzed using chi-squared tests for categorical variables and analysis of variance for continuous variables. Multivariable linear mixed regression models were used to examine associations between home SHS exposure and HRQOL for baseline health and during admission, overall and stratified by diagnosis. Covariates in each model included age, sex, race/ethnicity, caregiver education, and healthcare visits in the preceding six months. We also included a hospital random effect to account for clustering of patients within hospitals and used robust standard errors for inference.

In a secondary analysis to explore potential dose-response effects of SHS exposure, we examined associations between an ordinal exposure variable (0 smokers, 1 smoker, ≥2 smokers) and HRQOL for baseline health and during admission for the acute illness. Because of sample size limitations, diagnosis-specific analyses examining dose-response effects were not conducted.

Study Population

Of the 2,334 children enrolled in the PRIMES study, 25 (1%) respondents did not report on home SHS exposure and were excluded, yielding a final study population of 2,309 children, of whom 728 (32%) had reported home SHS exposure. The study population included 664 children with asthma (mean age seven years [3.5]; 38% with home SHS exposure), 740 with bronchiolitis (mean age 0.7 years [0.5]; 32% with home SHS exposure), 342 with croup (mean age 1.7 [1.1]; 25% with home SHS exposure), and 563 with pneumonia (mean age 4.4 [3.8]; 27% with home SHS exposure; Table 1). Compared with non-SHS-exposed children, those with home SHS exposure tend to be slightly older (3.9 vs 3.4 years, P = .01), more likely to be non-Hispanic Black (29% vs 19%, P < .001), to have a chronic condition (52% vs 41%, P < .001), to come from a household where caregiver(s) did not graduate from college (45% vs 29%, P < .001), and to have public insurance (73% vs 49%, P < .001).

Home SHS Exposure and Health-related Quality of Life

The overall mean HRQOL score for baseline health was 83 (15), with a range across diagnoses of 82 to 87. Compared with non-SHS-exposed children, children with home SHS exposure had a lower mean HRQOL score for baseline health (adjusted mean difference –3.04 [95% CI -4.34, –1.74]). In analyses stratified by diagnosis, baseline health scores were lower for SHS-exposed children for all four conditions, but differences were statistically significant only for bronchiolitis (adjusted mean difference –2.94 [–5.0, –0.89]) and pneumonia (adjusted mean value –4.13 [–6.82, –1.44]; Table 2); none of these differences met the MCID threshold.

The overall mean HRQOL score at the time of admission was 56 (23), with a range across diagnoses of 49 to 61, with lower scores noted among SHS-exposed children compared with non-SHS-exposed children (adjusted mean difference –2.16 [–4.22, –0.10]). Similar to scores representing baseline health, admission scores were lower across all four conditions for SHS-exposed children. Only children with croup, however, had significantly lower admission scores that also met the MCID threshold (adjusted mean difference –5.71 [–10.67, –0.75]; Table 2).

To assess for potential dose-response effects of SHS exposure on HRQOL, we stratified SHS-exposed children into those with one smoker in the home (n = 513) and those with ≥2 smokers in the home (n = 215). Compared with non-SHS-exposed children, both HRQOL scores (baseline health and admission) were lower for SHS-exposed children. Consistent with a dose-response association, scores were lowest for children with ≥2 smokers in the home, both at baseline health (adjusted mean difference –3.92 [–6.03, –1.81]) and on admission (adjusted mean difference –3.67 [–6.98, –0.36]; Table 3).

Within a multicenter cohort of 2,309 children hospitalized with ARI, we noted significantly lower HRQOL scores among children exposed to SHS in the home as compared with nonexposed children. Differences were greatest for children living with ≥2 smokers in the home. In analyses stratified by diagnosis, differences in baseline health HRQOL scores were greatest for children with bronchiolitis and pneumonia. Differences in acute illness scores were greatest for children with croup.¹⁶

Our study provides evidence for acute and chronic impacts of SHS on HRQOL in children hospitalized with ARI. Although several studies have linked SHS exposure to reduced HRQOL in adults,^19,20 few similar studies have been conducted in children. Nonetheless, a wealth of studies have documented the negative impact of SHS exposure on clinical outcomes among children with ARI.^8,10,21-23 Our findings that home SHS exposure was associated with reduced HRQOL among our cohort of children with ARI are therefore consistent with related findings in adults and children. The observation that the effects of SHS exposure on HRQOL were greatest among children living with ≥2 smokers provides further evidence of a potential causal link between regular SHS exposure and HRQOL.

Although the magnitude and significance of associations between SHS exposure and HRQOL varied for each of the four diagnoses for baseline health and the acute illness, it is important to note that the point estimates for the adjusted mean differences were uniformly lower for the SHS-exposed children in each subgroup. Even so, only acute illness scores for croup exceeded the MCID threshold.¹⁶ Croup is the only included condition of the upper airway and is characterized by laryngotracheal inflammation leading to the typical cough and, in moderate to severe cases, stridor. Given that chronic SHS exposure induces a proinflammatory state,³ it is possible that SHS-exposed children with croup had more severe illness compared with nonexposed children with croup resulting in lower HRQOL scores on admission. Further, perceived differences in illness severity and HRQOL may be more readily apparent in children with croup (eg, stridor at rest vs intermittent or no stridor) as compared with children with lower respiratory tract diseases.

Of the four included diagnoses, the link between SHS exposure and asthma outcomes has been most studied. Prior work has demonstrated more frequent and severe acute exacerbations, as well as worse long-term lung function among SHS-exposed children as compared with nonexposed children.^22-24 It was, therefore, surprising that our study failed to demonstrate associations between SHS exposure and HRQOL among children with asthma. Reasons for this finding are unclear. One hypothesis is that caregivers of SHS-exposed children with asthma may be more aware of the impacts of SHS exposure on respiratory health (through prior education) and, thus, more likely to modify their smoking behaviors, or for their children to be on daily asthma controller therapy. Alternatively, caregivers of children with asthma may be more likely to underreport home SHS exposure. Thirty-eight percent of children with asthma, however, were classified as SHS-exposed. This percentage was greater than the other three conditions studied (25%-32%), suggesting that differential bias in underreporting was minimal. Given that children with asthma were older, on average, than children with the other three conditions, it may also be that these children spent more time in smoke-free environments (eg, school).

Nearly one-third of children in our study were exposed to SHS in the home. This is similar to the prevalence of exposure in other studies conducted among hospitalized children^8,10,21,25 but higher than the national prevalence of home SHS exposure among children in the United States.²⁶ Thus, hospitalized children represent a particularly vulnerable population and an important target for interventions aiming to reduce exposure to SHS. Although longitudinal interventions are likely necessary to affect long-term success, hospitalization for ARI may serve as a powerful teachable moment to begin cessation efforts. Hospitalization also offers time beyond a typical primary care outpatient encounter to focus on cessation counseling and may be the only opportunity to engage in counseling activities for some families with limited time or access. Further, prior studies have demonstrated both the feasibility and the effectiveness of smoking cessation interventions in hospitalized children.^27-30 Unfortunately, however, SHS exposure is often not documented at the time of hospitalization, and many opportunities to intervene are missed.^25,31 Thus, there is a need for improved strategies to reliably identify and intervene on SHS-exposed children in the hospital setting.

These findings should be considered in the context of several limitations. The observational nature of our study raises the potential for confounding, specifically with regard to socioeconomic status, as this is associated with both SHS exposure and lower HRQOL. Our modeling approach attempted to control for several factors associated with socioeconomic status, including caregiver education and insurance coverage, but there is potential for residual confounding. No single question is sufficient to fully assess SHS exposure as the intensity of home SHS exposure likely varies widely, and some children may be exposed to SHS outside of the home environment.³² The home, however, is often the most likely source of regular SHS exposure,^33,34 especially among young children (our cohort’s mean age was 3.6 years). Misclassification of SHS exposure is also possible due to underreporting of smoking.^35,36 As a result, some children regularly exposed to SHS may have been misclassified as nonexposed, and the observed associations between SHS exposure and HRQOL may be underestimated. Confirming our study’s findings using objective assessments of SHS exposure, such as cotinine, are warranted. Given the young age of our cohort, the PedsQL surveys were completed by the parent or legal guardian only in >90% of the enrolled subjects, and caregiver perceptions may not accurately reflect the child’s perceptions. Prior work, however, has demonstrated the validity of parent-proxy reporting of the PedsQL, including correlation with child self-report.³⁷ In our study, correlation between child and caregiver reporting (when available) was also very good (r = 0.72, 95% CI 0.64, 0.77). It is also possible that the timing of the HRQOL assessments (on admission) may have biased perceptions of baseline HRQOL, although we anticipate any bias would likely be nondifferential between SHS-exposed and nonexposed children and across diagnoses.

Nearly one-third of children in our study were exposed to SHS exposure in the home, and SHS exposure was associated with lower HRQOL for baseline health and during acute illness, providing further evidence of the dangers of SHS. Much work is needed in order to eliminate the impact of SHS on child health and families of children hospitalized for respiratory illness should be considered a priority population for smoking cessation efforts.

Acknowledgment

The authors wish to acknowledge the efforts of PRIS-PRIMES study team. The authors also wish to thank the children and families who consented to be a part of the PRIMES study.

Disclosures

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

Funding

This study was supported by NIH-NHLBI 1R01HL121067 to RMS.