Patterns of Children With Complex Chronic Conditions: A Latent Class Analysis

Area of Science:

• Pediatric Health Services Research
• Health Informatics
• Biostatistics

Background:

• Complex chronic conditions (CCCs) affect a significant pediatric population.
• Understanding heterogeneity in CCCs is crucial for predicting healthcare needs.
• Previous classifications may not fully capture the nuances of care utilization.

Purpose of the Study:

• To identify distinct empirical classes of children with CCCs.
• To evaluate if these identified classes predict future healthcare use and spending.
• To inform targeted healthcare resource allocation for pediatric populations with CCCs.

Main Methods:

• Latent class analysis of Medicaid claims data (2017-2019) for children aged 1-18 with CCCs.
• Inclusion of demographic factors, clinical characteristics, and 2017 healthcare utilization.
• Negative binomial and logistic regression models to assess 2018-2019 healthcare spending and use.

Main Results:

• A 3-class solution emerged from 185,672 children with CCCs.
• Class 1 (9.1%): High neuro-disability, high technology dependence, high multimorbidity.
• Class 2 (14.8%): High neuro-disability, low technology dependence.
• Class 3 (76.0%): Low neuro-disability, low technology dependence.
• Classes 1 and 2 exhibited significantly higher healthcare spending compared to Class 3 (RR 6.9 and 2.5, respectively).
• Inpatient and outpatient specialist services drove costs in Class 1; outpatient drugs, specialists, and mental health in Class 2.

Conclusions:

• Children with CCCs can be meaningfully categorized into distinct classes.
• These classes are identifiable using readily available data.
• The identified classes demonstrate differential patterns of future healthcare use and associated costs.