Inpatient Obesity Recognition and Diagnosis in Pediatric Patients Free

This retrospective cohort study used CERNER Health Facts, a database that captures and stores deidentified, longitudinal electronic health record patient data, and aggregates and organizes these data into consumable data sets to facilitate analysis and reporting. The data are generated from CERNER- and non-CERNER–participating contributing facilities, dating back to the year 2000. CERNER Health Facts is a Health Insurance Portability and Accountability Act-compliant database collected from participating clinical facilities and is a comprehensive source of deidentified, real-world data collected as a byproduct of patient care.

Included patients were those aged 2 to 18 years, admitted to inpatient or observation status under the pediatric service from January 2012 to July 2017. Excluded patients were those on subspecialty services, surgical services, and patients admitted to the ICU, as well as those lacking anthropomorphic data sufficient to calculate BMI or those whose anthropomorphic data were biologically implausible according to the Centers for Disease Control and Prevention’s biologically implausible value algorithm (Supplemental Table 3).⁷  Data collected from patient charts included demographics (age, sex, weight, height, race/ethnicity, and hospital number) and hospital descriptors (census region, size of hospital, academic affiliation, and rural versus urban settings). Race/ethnicity was self-reported in the electronic health record, and rural versus urban and nonteaching versus teaching were defined by CERNER Health Facts.

BMI was calculated on the basis of admission height and weight, or BMI was used if individual values were unavailable. The distribution of BMIs was summarized by normal, overweight, or obese, and obesity was further classified as class I, class II, and class III obesity. Age, sex, and race/ethnicity among BMI groups were summarized with χ² testing for group-level differences. Appropriate obesity diagnosis was defined as a patient with obesity having an International Classification of Diseases, Ninth (278.*) or 10th Revision (E66.0*, E66.1*, E66.2*, E66.8, E66.9, Z68.3*, Z68.4*, Z68.54), diagnosis code for obesity. The proportion of patients with obesity who had an appropriate obesity diagnosis was summarized overall, by demographic covariates and by hospital characteristics. Testing for associations with individual and hospital characteristics was performed using χ² for hypothesis testing.

We identified 131  700 total patients, which included the number of children aged 2 to 18 years hospitalized for any reason during the study period. We excluded 67  784 (51.4%) patients because of missing height or weight data and 6695 (5.1%) patients because of biologically implausible values. We included a total of 57  291 encounters (43.5%), representing 48  802 distinct patients (Supplemental Table 3).

Measured obesity was most common in ages 11 to 15 years (21.8%), followed by 6 to 10 years (19.9%), and 16 to 18 years (18.9%), and was least common in ages 2 to 5 years (17.2%) (P < .001). Approximately 20.1% of inpatient pediatric males were obese, and 18.8% of females were obese. The rate of children with obesity has fluctuated over time (P < .001), ranging from 18.7% (2015) to 21.1% (2012). The proportion of children with obesity varied in hospitals of different sizes with no clear trend (P < .001). There were statistically significant differences in obesity proportion by hospital size, census region, and teaching status, but not by urban versus rural hospital settings (Table 1).

Appropriate obesity diagnosis occurred most in ages 16 to 18 years, (24.6%), followed by 11 to 15 years (18.7%), 6 to 10 years (10.1%), and 2 to 5 years (2.3%) (P < .001). There was a significant difference (P < .001) among the appropriate obesity diagnosis by gender, with males at 11.3% and females at 15.4%. Appropriate obesity diagnosis varied over time (P < .001), ranging from 11.5% in 2014 to 16.7% in 2016. There was statistically significant variation (P < .001) of appropriate obesity diagnosis among hospitals of various sizes, but no clear trend relative to increasing size. Additionally, appropriate obesity diagnosis varied among census regions, with the poorest in the South (8.5%) and highest in the Midwest (19.5%) (P < .001). There was no statistical significance among appropriate obesity diagnosis in teaching versus nonteaching facilities or in rural versus urban settings (Table 2).

This retrospective study concluded that, despite childhood obesity prevalence in the United States tripling over the last 3 decades,⁸  according to our study, only 13.2% of the 19.5% of admitted pediatric patients who had obesity received an appropriate obesity diagnosis. Appropriate obesity diagnosis increased with higher class of obesity (class I 6.9%, class II 19.6%, class III 39.3%). This finding was also noted in a study performed at 2 large, tertiary care children’s hospitals, which evaluated appropriate obesity documentation in both inpatient and outpatient settings.⁹  Additionally, in our study, there was a significant difference among appropriate obesity documentation in genders, with females with obesity more likely to receive an appropriate obesity diagnosis. Possible theories in adults have been postulated for higher appropriate obesity diagnosis in female patients, including gender bias, obesity social stigma, more-frequent provider visits, or dissatisfaction with weight in females.

Appropriate obesity diagnosis proportion varies among census regions, with poorest documentation in the South and best documentation in the Midwest. Further research is needed as to why the South has such poor obesity diagnosis, but it is possible that the high prevalence of obesity in this area has caused it to be considered “normal.” Future areas of exploration include association between obesity diagnosis and clinical conditions during hospitalization, which has been examined on a hospital level, but not on a national level.

There are multiple limitations to our study. This study extracted data from CERNER Health Facts, which may not be fully representative of nationwide inpatient pediatric data. Additionally, more than half of the patients identified either were lacking anthropomorphic data or had biologically implausible values. The available characteristics were compared between included and excluded patients (Supplemental Table 3). Additionally, CERNER Health Facts has a single race/ethnicity identifier, which likely does not represent the true makeup of the cohort, particularly as it pertains to the surprisingly small number of patients identified as Hispanic. Patients admitted to the ICU, subspecialties, and surgical specialties were excluded, limiting the generalizability.

Pediatric hospitalizations are an area where appropriate recognition of obesity needs improvement, as evidenced by the disparities among genders and census regions. This process begins with obtaining both a height and a weight upon admission so that BMI, an area where significant improvement is needed, can be calculated. Research shows that pediatric patients and their parents want to be informed of obesity status while inpatient, which increases child and parent concern for weight status.¹⁰  Furthermore, inpatient dieticians provide an additional layer of support so families can make healthier choices at that point in time rather than months later while awaiting outpatient nutritional management. Lastly, recommended obesity screening laboratories (hemoglobin A1C, lipid panel, liver enzymes, and vitamin D levels) can be easily obtained and followed up while inpatient, with the potential to recognize comorbid conditions early. By addressing obesity and providing nutritional counseling, providers are equipping patients and their families with the tools they need to have a healthier future.