Obesity and Health-Related Quality of Life in Children Hospitalized for Acute Respiratory Illness

This study was nested within a previously published prospective cohort study examining the quality of care and outcomes for children hospitalized with ARI at 5 tertiary care children’s hospitals.²²  The original study population enrolled eligible children 2 weeks to 16 years of age hospitalized from the ED and discharged with a primary diagnosis of asthma, croup, bronchiolitis, or pneumonia between July 1, 2014, and June 30, 2016. Because of the predominantly young age of children presenting with bronchiolitis and croup, along with the inability to calculate BMI for children aged <2 years, we limited this study to children aged 2–16 years discharged with acute asthma exacerbation or pneumonia. Diagnoses were determined by abstractor review of the medical record, both at the time of admission and again at discharge. If a child’s diagnosis changed during their hospitalization (eg, from asthma to pneumonia), the primary discharge diagnosis noted on the discharge summary was used for the condition group assignment. If a child’s diagnosis changed between enrollment and discharge and the final diagnosis was not a previously defined ARI condition, the case was excluded. Patients with ICU admission, 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 excluded. Underweight children were excluded from the multivariable regression models because of the heterogeneity of contributing health conditions. Enrolled children were limited to children of families who were English and/or Spanish speaking. All study procedures were approved by the institutional review boards at each of the participating hospitals and the Western Institutional Review Board.

Weight and height data at the time of admission were abstracted from the medical record. BMI categories were defined by using BMI-for-age z scores and included underweight (z score < −1.64), normal weight (z score −1.64 to 1.04), overweight (z score 1.04 to 1.64), and obesity (z score >1.64).^23,24  The z scores were determined by using Centers for Disease Control and Prevention growth charts and calculated by using the z-anthro package in Stata (Stata Corp, College Station, TX).²⁵  In an attempt to capture extreme data entry errors, Stata did not provide BMI-for-age z scores for children with z scores <−5 or >5. Children with obesity were further described as having class 1, class 2, or class 3 obesity, with class 1 defined as BMI ≥95th percentile to <120% of 95th percentile for age and sex, class 2 as BMI ≥ 120% to < 140% of 95th percentile for age and sex, and class 3 as BMI ≥ 140% of 95th percentile for age and sex.²⁶  These subclasses of obesity were determined by using SAS version 9.4 (SAS Institute, Inc, Cary, NC).

We measured HRQOL using the Pediatric Quality of Life 4.0 Generic Core Scales (PedsQL), pediatric HRQOL measures with demonstrated validity across outpatient and inpatient populations.^27–29  We measured Physical and Psychosocial Functioning subscales, which range from 0 to 100, with higher scores indicating better HRQOL. The reported minimal clinically important difference for this scale is 4.5 points.²⁸  Children aged ≥8 years were invited to complete the self-report version of the PedsQL. For children aged <8 years, and for older children who were unable to complete the survey, HRQOL surveys were completed by a parent or legal guardian. In-hospital surveys were completed as a self-administered survey on an iPad or as a phone interview if the respondent preferred. The follow-up survey was completed as a self-administered online survey sent as an E-mail link or as a telephone interview. For each patient, the same respondent who had completed the initial survey completed the follow-up survey. Respondents were asked to assess perceptions of their (or their child’s) HRQOL at the time of admission and at 2–6 weeks after discharge from the hospital.

Sociodemographic covariates were collected by caregiver report at the time of enrollment, including child age, sex, race and ethnicity, English proficiency,³⁰  caregiver education,³¹  secondhand smoke exposure, and delayed access to care, defined as any parent-reported difficulty or delay in accessing care over the previous 6 months. Caregiver-reported race and ethnicity categories included white, Black, Hispanic, Asian American or Pacific Islander, American Indian or Alaskan native, or other. Low English proficiency was defined as caregiver survey responses of “Not well” or “Not at all” regarding how well they speak English. Caregiver education was defined as less than high school (HS), HS, or greater than HS. Insurance status, medical complexity, and hospital outcomes (30-day reuse and LOS) were obtained by using the Pediatric Hospital Information System database, an administrative database containing clinical and resource use data from >45 US children’s hospitals, including all of the study hospitals.³²  Medical complexity was defined by the Pediatric Medical Complexity Algorithm (PMCA), which identifies children with complex chronic conditions on the basis of hospital discharge or administrative claims data. The PMCA identifies children as nonchronic (eg, no chronic disease), noncomplex chronic (eg, asthma, attention-deficit/hyperactivity disorder), and complex chronic (eg, cystic fibrosis).³³  Thirty-day reuse was defined as an ED visit or hospital readmission with the same discharge diagnosis as the index admission within 30 days of discharge.

Bivariate comparisons according to BMI category (normal weight, overweight, and obesity) were analyzed by using χ² tests for categorical variables. Unadjusted differences in HRQOL by weight category were analyzed by using a 2-tailed t test. Multivariable linear mixed regression models were used to examine associations between BMI category and HRQOL at the time of admission and 2–6 weeks after discharge, stratified by diagnosis. Multivariable models evaluating HRQOL 2–6 weeks after discharge included the covariate of admission HRQOL score to better assess the change from baseline HRQOL. Models also included time, in days, between admission and follow-up surveys as a covariate because patients with longer time-periods between the 2 surveys were expected to have a greater HRQOL recovery. Multivariable logistic regression models were used to examine the relationship between BMI category and 30-day reuse. Given that LOS was positive and skewed, we applied a generalized linear model with logarithm link and γ family distribution to evaluate the relationship between BMI category and hospital LOS. Covariates in each model included age, sex, race and ethnicity, caregiver education, insurance status, and child level of medical complexity. We also included a hospital random effect to account for clustering of patients within hospitals and used robust standard errors for inference. For children missing HRQOL follow-up assessments, we performed sensitivity analysis using linear regression models to impute the missing HRQOL scores at 2–6 weeks after discharge with variables used in the main regression analysis. Results from individual replicated data sets were then pooled into overall estimates. χ² tests were used to evaluate for differences between those lost to follow-up and those not lost to follow-up, as well as between those with and without height data. For regression results, we considered a coefficient significant if its 95% confidence interval (CI) did not cover the null. Analyses were conducted in Stata version 10.

Of the 2187 children initially enrolled in the study, 716 (32.7%) had a calculable BMI z score. Of these, 429 (60.1%) had normal weight, 82 (11.4%) had overweight, and 138 (19.3%) had obesity. 67 (9.4%) had underweight and were excluded. The remaining study cohort of 649 children included 445 children hospitalized with asthma exacerbation and 204 with pneumonia (Fig 1). The characteristics of this cohort are revealed in Table 1. Of children with obesity, 21 (15.2%) had class 2 obesity and 17 (12.3%) had class 3 obesity. Compared with children with normal weight, children with overweight or obesity were more likely to be older and have public insurance, parents without a college education, secondhand smoke exposure, and chronic disease comorbidities (P < .05).

At the time of enrollment, 638 (98.3%) eligible children completed the physical function HRQOL survey and 647 (99.7%) completed the psychosocial HRQOL survey. Among those aged ≥8 years, 147 (67.1%) surveys were completed by the child, including 101 (65.6%) children with asthma and 46 (70.8%) children with pneumonia. For the follow-up survey, 183 (28.7%) children had missing physical functioning HRQOL measures and 191 (29.5%) had missing psychosocial functioning HRQOL measures. Those lost to follow-up were more likely to be Black (P = .001), report lower income (P = .011), and report higher English proficiency (P = .028). Comparing those lost to follow-up to those with complete HRQOL data, there were no differences in age, sex, delayed access to care, insurance type, secondhand smoke exposure, or medical complexity. Of the original eligible cohort, 21.6% of children with asthma and 28.4% of children with pneumonia had missing height data and could not be included in the cohort. There were no significant differences in demographics between these children and those with available height data.

Table 2 displays the unadjusted and adjusted associations between weight status and HRQOL scores at presentation and 2–6 weeks after discharge. Across each of the presenting diagnoses, there were no significant differences between BMI groups in physical functioning HRQOL scores at presentation or 2–6 weeks after discharge. Among children admitted with pneumonia, having overweight was associated with lower psychosocial HRQOL scores at presentation (β-coefficient −7.47, 95% CI:, −13.0 to −1.94; P = .008), although this association was not seen among children with obesity. There was no difference in psychosocial function HRQOL scores by BMI category for children presenting with asthma.

Nineteen (2.9%) children returned to care within 30 days of discharge, including 10 (2.3%) children with asthma and 9 (4.4%) with pneumonia (Table 3). Among children admitted with either asthma exacerbation or pneumonia, there was no increased likelihood of reuse among children with overweight or obesity (Table 3). The mean hospital LOS was 1.6 days (SD, 0.9) for children with asthma exacerbation and 2.4 days (SD, 1.7) for those with pneumonia. The mean LOS by diagnosis and weight category are revealed in Table 3. There were no significant differences in LOS by weight category (Table 3).

Our study of children hospitalized with acute asthma exacerbation or pneumonia addresses a gap in the literature regarding the impact of obesity on HRQOL among children hospitalized for these ARIs. Despite growing evidence of obesity’s adverse effects on HRQOL and the respiratory system,^6–8  we did not find an association between obesity and worse HRQOL, longer LOS, or increased 30-day reuse among children hospitalized for asthma or pneumonia.

Previous meta-analyses have revealed a strong association between obesity and a diagnosis of asthma, with increasing incidence of each condition over the past 2 decades.^34–36  The temporal and causal relationship of this association remains unclear, although several longitudinal studies support the hypothesis that increased BMI leads to increased risk of developing asthma.^34,36  Despite the wealth of data suggesting decreased HRQOL among children with asthma or with obesity, studies evaluating the combined impact of both obesity and asthma are limited, particularly during an acute exacerbation. Although results are inconsistent, previous studies suggest that, at baseline, there may be an additive negative effect of obesity on the HRQOL of children with asthma.^13,37,38  However, we were unable to identify any studies evaluating the association of obesity and HRQOL among children with acute exacerbation of asthma. Our study revealed no significant difference in HRQOL at the time of admission or follow-up by BMI category for children hospitalized with acute asthma exacerbation.

For patients hospitalized with pneumonia, we were unable to identify any previous studies evaluating the association between obesity and HRQOL among children or adults. In our study of hospitalized children with pneumonia, having overweight was associated with lower psychosocial HRQOL scores at presentation, although this association was not observed among children with obesity. It is unclear why this association was observed for children with overweight and not obesity, although it is possible that there was an unmeasured confounder that was more common in children with overweight than those with obesity (eg, comorbidities such as anxiety or depression) that might explain worse psychosocial HRQOL scores in this population. For children hospitalized with pneumonia, we found no association between obesity and physical functioning HRQOL.

Given the known associations of both obesity and asthma with lower HRQOL and the association of obesity with worsened baseline severity of asthma, it is interesting that our study found no relationship between obesity and worsened HRQOL for children hospitalized with asthma or pneumonia.^6,12,13,16  Regardless of their weight status, children require hospitalization when the severity of their respiratory illness reaches a certain threshold of impact on their physical functioning. At this threshold, patients may experience similar impacts on the HRQOL, regardless of their weight status. At the time of hospitalization, the impact on HRQOL of acute illness (eg, pneumonia) or an exacerbation of chronic disease (eg, asthma), may diminish any differences in baseline quality of life between weight classes.

Among children with a diagnosis of asthma managed in the outpatient setting, obesity has been associated with increased frequency of exacerbations and decreased responsiveness to inhaled corticosteroids.^30,31  However, the impact of obesity on asthma-related hospitalization and hospital outcomes has been inconsistent, with contradictory data regarding hospital LOS and the rate of hospitalization, ICU admission, and hospital readmission.^14–19  In our study, we found no differences in the hospital outcomes of LOS or reuse by BMI category for children hospitalized with asthma.

Studies evaluating the association of obesity and hospital outcomes among children hospitalized with pneumonia are more limited. Bramley et al found that, among children hospitalized for pneumonia, the odds of mechanical ventilation and ICU admission were higher for children with obesity; however, when stratified by asthma status, this relationship remained significant only among those children with asthma.²⁰  Okubo et al found that, among children hospitalized for pneumonia and bronchitis, obesity was associated with increased mechanical ventilation, bacteremia, cost, and LOS.²¹  Of note, this study defined obesity using diagnosis codes alone, and, on the basis of an unusually low prevalence of obesity (1.1%), it is possible that the study population disproportionately represented children with severe obesity. In our study, we found no differences in LOS or reuse by BMI category for children hospitalized with pneumonia.

This study has several important limitations. Because of the observational nature, there is the potential for residual confounding, although our model adjusted for available covariates related to socioeconomic status, including income, parental education, and insurance status. Second, because of the limited number of children with class 2 and class 3 obesity in this study population, we were unable to evaluate the association between the severity of obesity and outcomes. Given the findings of Okubo et al described above, evaluation of outcomes among children with severe obesity should be an area of future investigation.²¹  Third, reuse rates in this cohort were low, limiting our ability to draw meaningful conclusions from these data. Fourth, limitations of our study population, including the inclusion of only children admitted to tertiary care children’s hospitals and the exclusion of children admitted to the ICU, children with certain chronic illnesses, and children of families who do not speak English or Spanish, limit the generalizability of our findings. Fifth, we relied on caregiver report for the majority of our HRQOL assessments, which may not have accurately reflected the physical or psychosocial functioning of their children. Sixth, ∼29% of patients were lost to follow-up, which may have biased the HRQOL results, given that these children were more likely to be Black and from lower income families. Finally, ∼24% of eligible patients had missing height data and could not be included in our cohort; however, among the available demographics, there were no significant differences between those with and without available height data. Despite these limitations, this study has the strengths of a large, multicenter, prospective cohort.

In this study of children hospitalized for acute exacerbation of asthma or pneumonia in one of 5 tertiary care children’s hospitals, obesity was not associated with worse HRQOL, increased LOS, or increased reuse. Future studies should evaluate these relationships for children with greater variability in severity of illness, more severe obesity, and, to the extent possible, use child and adolescent self-report to assess HRQOL to better determine the need for targeted inpatient ARI management strategies in this vulnerable population.