Oregon’s approach to leveraging system-level data to guide a social determinants of health-informed approach to children’s healthcare

Abstract

BackgroundChildren’s health and healthcare use are impacted by both medical conditions and social factors, such as their home and community environment. As healthcare systems manage a pediatric population, information about these factors is crucial to providing quality care coordination.MethodsThe authors developed a novel methodology combining medical complexity (using the Pediatric Medical Complexity Algorithm) and social complexity (using available family social factors known to impact a child’s health and healthcare use) to create a new health complexity model at both the population-level and individual-level. System-level data from Oregon’s Medicaid Management Information Systems and Integrated Client Services database was analysed, examining claims data and service utilization, to calculate the health complexity of children enrolled in Medicaid/Child Health Insurance Program (CHIP) across Oregon.ResultsOf the 390 582 children ages 0 to 17 enrolled in Medicaid/CHIP in Oregon from July 2015 to June 2016, 83.4% (n=325 900) had some level of medical and/or social complexity and 22.1% (n=85 839) had health complexity (both medical and social complexity). Statistically significant (p<0.05) differences in health complexity were observed among attributed patients by Oregon’s 16 Coordinated Care Organizations, as well as by a child’s age, county of residence and race/ethnicity.ConclusionsGiven the high proportion of children with health complexity, these findings demonstrate that a large number of Medicaid/CHIP-insured children could benefit from targeted care coordination and differential resource allocation. Reports have been shared with state, county and health system leaders to drive work across the state. This paper describes the collaborative process necessary for other states considering similar work.

Toward Measurement for a Whole Child Health Policy: Validity and National and State Prevalence of the Integrated Child Risk Index

“…The ICRI complements emerging state-level approaches to assess complex risks of children using system-level data, like that set forth in Oregon. 43 Interestingly, our Oregon ICRI findings using the NSCH were consistent with findings from the Oregon systems-level data algorithm. The ICRI identified 15.8% of Oregon's publicly insured children with risks on all three of its medical, social and relational health domains whereas the Oregon algorithm identified 22.1% of such children meeting both its medical and social risk criteria.…”

“…41,42 The NSCH is an annual survey led by the US Health Resources and Services Administration's (HRSA) Maternal and Child Health Bureau (MCHB) in collaboration with the US Bureau of the Census. 41 Unlike some state-level efforts to assess children's medical and social risks using system-level data, 43 the NSCH has the advantage of providing state data at the child level across numerous topics and enables stratification across a wide range of child characteristics. Here we used the combined 2016-2018 NSCH data (n=102,341) to create and validate the Integrated Child Risk Index (ICRI), and then used the 2019 NSCH data (n=29,344) to assess the reliability of study findings.…”

Identifying Priorities to Improve the System of Care for Children With Complex Health Needs in North Carolina: Process and Outcomes of Systematic Stakeholder Engagement

“…1). 2 Unfortunately, the services CCHN and their families require have traditionally operated in siloes, placing burdens on families to navigate and secure them. For families this responsibility often is stressful and burdensome.…”

An algorithm using administrative data to measure adverse childhood experiences (ADM‐ACE)

“…For example, Washington and Oregon each have developed integrated client databases, including data from state health services (to identify parental mental illness and substance use disorder), departments of corrections (to identify parental involvement in the criminal punishment system), and departments of health (to identify parental death through death certificates) along with child welfare records (to identify maltreatment and foster care), and Medicaid records (to correlate these ACEs with medical complexity). 22 , 23 These integrated client databases contain rich information on children's circumstances and health and have provided important insights such as the prevalence of ACEs by age groups within Oregon 23 and showing that child maltreatment and instability in foster care placements were most predictive of higher health care costs for children in Washington, compared to ACEs from parent risk factors. 22 However, the development of these state‐level linked administrative databases required years of interagency coordination to share and link data; thus, measures developed with these databases are not replicable in other states which have no linked data or by health services researchers with access to claims data but no state interagency datasets.…”

“…At present, the most common methods for identifying ACEs include the use of population‐based cross‐sectional surveys with parent‐reported questions on child maltreatment 19 and household exposures 20 or administrative data from child welfare records of official reports of maltreatment 21 . Some states have undergone years‐long efforts to develop linked administrative databases across state agencies to enable identification of child and family ACEs, 22,23 and other researchers have used diagnosis codes from administrative health encounter data to identify child maltreatment but no other ACEs, 24 which we seek to build on here. While surveys are important tools for understanding population‐level health, including the prevalence of ACE exposure, they are subject to social desirability and recall bias, are expensive to administer, and it is infeasible to administer surveys over the entire populations of interest 25 .…”