Weight Status and Risk of Inpatient Admission for Children With Lower Respiratory Tract Disease

We performed a retrospective cohort study using electronic health record data from 6 children’s health systems across the United States that participate in the PEDSnet clinical research network.²¹  The institutional review board at an academic children’s hospital deemed this study exempt from human subjects review.

The cohort included children 2 to <18 years old evaluated for LRTD, defined as a discharge diagnosis of bronchiolitis, viral pneumonia, asthma, or CAP, in the PED between January 1, 2009, and November 30, 2019. In a separate manuscript (unpublished observations), the authors evaluated the associations of LRTD outcomes with weight category in hospitalized children. Discharge diagnosis codes of interest were identified from International Classification of Diseases, Ninth Revision (ICD-9) codes, which were translated to Systematized Nomenclature of Medicine Clinical Terms codes (Supplemental Tables 2 and 3).²²  We excluded children younger than 2 years because of lack of consistent definitions for weight categories for this age group and excluded children older than 17 years because this population is often not admitted to children’s hospitals.²³  We excluded children missing height and weight measurements within 3 months of the PED visit. Finally, we used the Pediatric Medical Complexity Algorithm Version 3.0 to exclude those with medical complexity because of anticipated effects of underlying diagnoses on weight status.²⁴ 

We obtained clinical, demographic, and anthropometric data from the PEDSnet database. Primary exposure was weight status. We calculated each patient’s BMI from height and weight recorded in the PED; if either value was missing, we accepted the closest available recorded height or weight within 3 months of that PED visit to minimize selection bias. We used age, sex, and BMI to categorize weight status per Centers for Disease Control and Prevention guidelines: <fifth percentile, underweight; fifth percentile to <85th percentile, normal weight; 85th percentile to <95th percentile, overweight; 100% to <120% of 95th percentile, class 1 obesity; 120% to <140% of 95th percentile, class 2 obesity; and ≥140% of 95th percentile, class 3 obesity.^23,25,26  Outliers were excluded by Centers for Disease Control and Prevention guidelines.²⁷  Age was categorized as 2–5 years, 6–11 years, and 12–18 years. Payer was categorized as private/military, public, or none. Primary outcome was hospital admission; secondary outcomes were 7- and 30-day PED revisits.

Descriptive statistics determined the prevalence of underweight, normal weight, overweight, and obesity by demographic and clinical characteristics. χ² tests or analysis of variance procedures were performed as appropriate. We modeled hospital admission, 7-day PED revisits, and 30-day PED revisits using multivariable mixed-effects logistic regression, which included site and patient as random effects and age, sex, race and ethnicity, and payer as covariates. A 2-sided P value of .05 was considered statistically significant. Given incomplete payer data at one site due to electronic health record migration, we performed a sensitivity analysis with all payer data from that site set to missing. We used SAS version 9.4 (SAS Institute, Inc, Cary, NC) for all analyses.

Of 258 709 patients with at least 1 PED visit for LRTD during 2009 and 2019, our final cohort included 107 446 children with 218 180 PED encounters, as shown in Fig 1. Demographic and clinical characteristics are shown in Table 1. At the time of the PED visit, 4.5% had underweight, 56.4% had normal weight, 16.1% had overweight, 14.6% had class 1 obesity, 5.5% had class 2 obesity, and 3.0% had class 3 obesity. PED visits for children with underweight were more likely to include an infectious LRTD discharge diagnosis (bronchiolitis, CAP, or viral pneumonia) compared with those for children of other weight categories. Nearly half of PED encounters for children with underweight resulted in hospital admission, higher than any other weight category.

Our primary outcome was hospital admission, which occurred in 29.9% of PED evaluations overall. After multivariable mixed-effects logistic regression, underweight had a positive association with hospital admission (odds ratio [OR] 1.76; 95% confidence interval [CI] 1.69–1.84) compared with normal weight. In contrast, overweight, class 1 obesity, and class 2 obesity all had slight negative associations with hospital admission, as shown in Fig 2. These findings persisted in a sensitivity analysis performed to account for missing insurance data at a single site.

Our secondary outcomes were 7- and 30-day PED revisits (Fig 2). We found small positive associations between 7-day revisits and class 1 obesity (OR 1.07; 95% CI 1.01–1.14) and class 2 obesity (OR 1.12; 95% CI 1.02–1.23). Findings were similar for 30-day revisits, except there were significant associations with overweight (OR 1.04; 95% CI 1.00–1.08) as well as class 1 obesity (OR 1.10; 95% CI 1.05–1.15) and class 2 obesity (OR 1.18; 95% CI 1.10–1.26).

In this large cohort representing multiple children’s hospitals across the United States, underweight was associated with increased hospital admission among children evaluated in the PED for LRTD. In contrast to our hypothesis, overweight, class 1 obesity, and class 2 obesity had slight negative associations with hospital admission. We found only small positive associations with 7- and 30-day PED revisits for patients with class 1 and class 2 obesity.

The outcome of hospital admission could represent increased severity of illness.^5,16,28,29  Studies from both developed and developing countries have revealed associations between malnutrition and severity of respiratory tract infections, which could be related to immune system effects.^9,30  Notably, vitamin D deficiency has been associated with severe respiratory tract infections.^31,32  In our study, we assessed underweight rather than specific nutrition or micronutrient deficiencies; however, this association may be an important area of further research. Underweight specifically has been associated with severe illness in Japanese children hospitalized for bronchiolitis and influenzalike illness.^33,34  In one multisite study in the United States, both underweight and obesity were associated with measures of asthma severity; hospitalization was not considered as an outcome in that study. Understanding the mechanisms behind our findings could help improve clinical outcomes for this population.

The need for hospital admission could vary with other factors in addition to illness severity. It could reflect health care use, specifically in terms of use of the PED. Patients with more limited health care access may use the PED more frequently, resulting in a larger denominator of children seeking care and a lower percentage who require hospital admission. Obesity is more prevalent among households with lower income and lower education, which could have implications for access to health care; existing studies on differential PED use by children with obesity reveal conflicting results.^35–37  We did not specifically assess access to care in this study. Hospital admission could also reflect decision-making by PED physicians that is unrelated to illness severity, including social concerns or physician biases, which were not specifically assessed here.

Our large multicenter cohort displays many similarities to descriptions of the general US population. We found similar disparities in weight category by age and race and ethnicity as have been demonstrated previously.¹  The prevalence of overweight and class 1 obesity is slightly lower than in the general pediatric population; this likely reflects the younger age of our cohort, which has been described previously in children being treated for LRTD.¹  The geographic diversity of our sites may enhance the generalizability of our results.

Several limitations apply to this study. Although multiple hospitals contributed data to the study, all are academic children’s hospitals, which does not reflect all PEDs providing care to children. Missing anthropometric data excluded approximately half of all potential encounters in the PEDSnet database. Although we anticipate that these missing data would most likely bias our results toward the null hypothesis because of likely exclusion of more children with relatively normal height and weight, it should still be considered in interpretation of our results. BMI and BMI categorization do not fully capture body composition or adiposity. Because of the focus of this study on patient BMI category, we excluded children <2 years old, which represents a large population of children with LRTD, especially bronchiolitis. We did not account for asthma severity in this analysis, which could affect these results. Socioeconomic status and health care use are represented only with payer in this data set; other social barriers to health were not captured.

In this multicenter cohort study, hospital admission for LRTD is positively associated with underweight and negatively associated with overweight and obesity in children seeking care in the PED. Further study should be focused on mechanisms behind this finding as a next step to improving clinical outcomes.