Background

Healthcare-Effectiveness-Data-and-Information-Set (HEDIS) performance measures rely solely on diagnosis codes to determine clinician attention to high body mass index (BMI). Our group previously published a validated method to identify guideline-recommended weight-management clinician behaviors from electronic phenotypes using extractable electronic-health-record (EHR) indicators (diagnosis codes, weight-loss medicines, and nutrition/weight-management referrals). Data indicate clinician attention to high BMI (coded by our method) increases the likelihood a child with overweight/obesity improves their BMI.

Methods

To determine proportion of clinician attention to high BMI missed by HEDIS-measure reliance on diagnosis codes alone, we applied our algorithm to retrospective EHR data. Inclusion criteria were: child age 2-18 years with ≥2 BMI-percentiles ≥85 (overweight), ≥1 well-child visit, and no major congenital/metabolic condition. Among visits (with child BMI% ≥85) coded as having evidence of attention to BMI, we ascertained the source of evidence (diagnosis code, referral, or medicine). Descriptive statistics described sample characteristics and the proportion of children and visits with attention to BMI in which the source of evidence did not include a diagnosis code.

Results

Of 53,146 2- to 18 year-old children, 42% had ≥2 BMI% ≥85. Of these, 82% (n=18,401) had ≥1 well-child visit and no exclusion condition. Thirty-eight percent were age 2-5 years, 43%, 6-11 years, and 19%, 12-18 years; 52% were overweight, 48% had obesity. Evidence of attention to high BMI was identified for 90% of children (at ≥1 visit) and at 32% of 125,435 visits. Referrals were the sole source of evidence of clinician attention to high BMI for 33% of children and 21% of visits. No child received an obesity medicine.

Conclusions

HEDIS performance measures that use diagnosis codes miss one-third of children and one-fifth of visits with nutrition/weight-management referrals that denote clinician attention to high BMI.