Obesity identified by discharge ICD-9 codes underestimates the true prevalence of obesity in hospitalized children

Abstract

Objectives: To define inpatient care of obese children with or without an obesity diagnosis.

Study design: A total of 29,352 inpatient discharges (18,459 unique inpatients) from a tertiary children's hospital were analyzed. Body mass index (BMI) was calculated from measured height and weight. "Obesity" was defined as BMI >or=95th percentile by using Centers for Disease Control and Prevention 2000 growth charts. "Diagnosed obesity" was defined by primary, secondary or tertiary International Classification of Diseases, Ninth Revision codes for "obesity" or "overweight." Analyses controlled for multiple inpatient records per individual.

Results: A total of 5989 discharges from the hospital (20.4%) were associated with obesity, but only 512 discharges (1.7%) indicated obesity as a diagnosis. An obesity diagnosis identified only 5.5% of inpatient days for obese inpatients. Obese patients with an obesity diagnosis (Ob/Dx) had fewer hospital discharges per person and shorter lengths of stay than obese patients without an obesity diagnosis (Ob/No Dx). Patients with Ob/Dx had higher odds of mental health, endocrine, and musculoskeletal disorders than non-obese inpatients, but Ob/No Dx patients generally did not.

Conclusions: Inpatient obesity diagnoses underestimate inpatient utilization and misidentify patterns of care for obese children. Extreme caution is warranted when using obesity diagnoses to study healthcare utilization by obese children.

Figure 1

Compilation of final dataset.

Figure 2

Inpatient utilization by obesity category. A, Adjusted mean length of stay ± 95% CI from mixed models, adjusted for sex, age, and fiscal year and accounting for multiple inpatient stays per unique individual. a: significantly different from Non-Ob (P ≤ .01), b: significantly different from Ob/No Dx (P < .0001). B, Number of inpatient stays per unique individual. Adjusted means ± 95% CI from general linear modeling, adjusted for sex, age, and fiscal year. a: significantly different from Non-Ob (P ≤ .05), b: significantly different from Ob/No Dx (P < .0001).