Trends in Health Care Use and Spending for Young Children With Neurologic Impairment

We performed a retrospective longitudinal cohort analysis of Medicaid-enrolled children with NI, using the IBM Medicaid MarketScan Database from years 2009–2017 (IBM Watson Health, Armonk, NY). Children were enrolled in Medicaid care plans from 12 deidentified states across the United States. The data set contains deidentified data on demographics, clinical diagnoses, and paid claims across the inpatient and outpatient care continuum.

Children were included in our sample if they received ≥1 NI diagnosis during their first 6 months of life, such as brain abnormalities, epilepsy, or intracranial hemorrhage. NI diagnoses were identified by using a previously established set of 604 International Classification of Diseases, Ninth Revision, codes that were originally identified by multiple pediatric neurologists, who reviewed and approved NI diagnoses using a modified Delphi technique.⁵  They used a NI definition of: “diagnosis is consistent with NI (static or progressive) and typically results in either functional and/or intellectual impairment.” We included subjects in the database from 2009 to 2012 if they had a birth hospitalization identified by using International Classification of Diseases, Ninth Revision, codes of V30-V39 and were continuously enrolled in Medicaid through age 5 (allowing for a 1-month lapse per year). This study was deemed not to be human subject research by our institutional review board.

Our primary outcomes were annual health care use (any versus none) and spending by service category across the care continuum for children with NI during their first 5 years of life. Service categories were provided by IBM Watson and grouped by health care setting (ie, inpatient, ED, and outpatient). Outpatient use was further subcategorized into 6 services: primary care; specialty care; occupational, physical, and speech therapy; mental health and behavioral counseling; home health; and outpatient medications.

Health care use was reported as the proportion of the study sample who received inpatient, ED, and outpatient subcategory services in each year of life. Because birth hospitalization was an inclusion criterion, use of inpatient services during the first year of life was defined as reuse of inpatient services after birth hospitalization, or prolonged birth hospitalization (>3 days).

Health care spending was reported as the per-member-per-year (PMPY) spending by service category.

For subjects, we report sex, reason for Medicaid enrollment, NI category, number of complex chronic conditions (CCCs), and use of NICU services. Seven NI categories were assessed: (1) static neurologic disease, (2) progressive neurologic disease, (3) anatomic abnormality, (4) epilepsy, (5) genetic or metabolic condition, (6) cerebrovascular disease, and (7) peripheral neurologic disease. NI categories were not mutually exclusive because some patients have diagnoses in multiple categories. CCCs are conditions that are expected to last ≥12 months, require specialty care, and are likely to require hospitalization because of medical diagnoses.¹⁸  Subjects were identified as using NICU services if their birth hospitalization included services billed under intensive care.

To explore trends in health care use and spending in a subset of children with NI who also had medical complexity, we performed a subanalysis of children with NI who had >2 CCCs. These children fit the generally accepted definition of children with medical complexity (CMC),¹⁹  and we refer to them as CMC with NI.

We examined the annual health care use and PMPY spending by service categories, which included inpatient, ED, and 6 outpatient subcategory services (see outcomes). To evaluate trends in use, we modeled annual use as a binary outcome for each service category, by using logistic regression with a fixed effect for year. Results are presented as an odds ratio (OR), which can be interpreted as the annual change in the odds of use for a one-year increase in age. To analyze trends in health care spending, we modeled annual spending as a continuous outcome using linear regression with a fixed effect for year. Given that annual spending was not normally distributed, we performed log transformation of the data for the trend analysis. This year-over-year analysis yielded a mean annual change (MAC), which can be interpreted as the MAC in spending for a one-year increase in age.

All analyses were completed by using SAS version 9.4 (SAS Institute, Cary, North Carolina). P values ≤0.05 were considered statistically significant.

There were 13 947 children with NI (5.0% prevalence) who were continuously enrolled in Medicaid from 0 to 5 years of age. In the cohort, 37.2% used NICU services after birth (Table 1).

The 3 most common NI categories were genetic or metabolic condition (24.2%), anatomic abnormality (22.1%), and epilepsy (21.9%) (see Table 1; individual diagnoses are shown in Supplemental Table 4). Overall, 72.6% of children with early-onset NI had ≥1 CCC, and 10.6% were technology dependent (eg, using a feeding tube or tracheostomy). In the sample, 15.9% met criteria for CMC with NI.

Whereas 66.8% of children with NI used inpatient services during their first year, only 5.8% did so during their fifth year. Across their first 5 years, annual inpatient service use showed a decreasing trend (OR: 0.35; 95% confidence interval [CI]: 0.34 to 0.36), with use decreasing by 80.7% between their first and second year (Table 2). Thereafter, the rate of use continued to decrease by 15% to 30% annually. For those who used inpatient services at any point during their first 5 years, the median length of stay was 3 days (interquartile range [IQR]: 2–9 days). The 4 most common diagnosis-related groups for inpatient use were asthma and bronchitis, dehydration, seizure, and gastroenteritis.

PMPY spending on inpatient services showed a decreasing trend (MAC: −$15 232; 95% CI: −$15 627 to −$14 837, Table 3). Annual spending on inpatient services was consistently the largest spending category, although the proportion of annual spending on inpatient services decreased over the time period (90.6% of annual spending in the first year, 25.5% in the fifth year).

We found a decreasing trend in both ED service use (OR: 0.78; 95% CI: 0.77 to 0.79) and spending (MAC: −$75; 95% CI: −$79 to −$72), although the decrease was more gradual than that observed for inpatient services. The rate of ED use decreased by 7% to 12% annually. For children with early-onset NI, the mean days of ED use was 2.0 during their first year, 1.2 days during their third year, and 0.9 days during their fifth year. The 4 most common principal diagnoses for ED use were fever, upper respiratory infection, otitis media, and vomiting. PMPY spending on ED services represented ∼4% of annual spending for children >1 year of age.

Across their first 5 years, the majority of children with NI used primary care services (>85% of children each year) but had a decreasing trend of use (OR: 0.69; 95% CI: 0.68 to 0.71). PMPY spending on primary care services represented ∼5% of annual spending for those >1 year of age.

Specialty services were used by >64% of children with NI each year. PMPY spending on specialty services was the second largest spending category annually and remained constant across the first 5 years (MAC: $23; 95% CI: −$15 to $61). Thus, as the PMPY spending on other services decreased, spending on specialty services represented an increasing proportion of annual spending: ∼20% by the fifth year (PMPY: $1463).

Annual use of other outpatient services shifted over the 5 years (Table 2). For example, the use of home health services decreased by 82.9% after the first year, from 22.3% in the first year to 3.8% in the second. Mental health and behavioral counseling was the one outpatient service that showed a trend of increased use (OR: 1.03; 95% CI: 1.01 to 1.04).

In total, per member health care spending for children with early-onset NI during their entire first 5 years of life was $122 544. The median per member health care spending for the entire first 5 years was $34 394 (IQR: $12 503–$133 521). As children with NI aged from 0 to 5 years, the distribution of spending transitioned from primarily inpatient during their first year (>90% of PMPY spending) to a more even distribution across service lines in subsequent years (Table 3).

CMC with NI, who had >2 CCCs and NI, had higher health care use and spending over their first 5 years than noncomplex children with NI did but demonstrated similar decreasing trends of use and spending.

During their first year of life, 95.5% of CMC with NI used inpatient services, compared with 16.7% in their fifth year. Across their first 5 years, annual inpatient service use showed a decreasing trend (OR: 0.41, 95% CI: 0.40 to 0.43), decreasing by 65.9% between the first and second year (Supplemental Table 5). Over 5 years, PMPY spending on inpatient services showed a decreasing trend (MAC: −$43 738; 95% CI: −$45 538 to −$41 938). PMPY spending on inpatient services was $219 555 during the first year (92.0% of annual spending) and $6644 during the fifth year (31.7%).

During each year, >52% of CMC with NI used ED services, with a decreasing trend (OR: 0.80; 95% CI: 0.78 to 0.82).

Over their first 5 years, the vast majority of CMC with NI consistently used primary care (>91% each year) and specialty services (>80%). Annual spending on specialty services was consistently the second largest category, representing ∼20% to 25% of annual spending for those >1 year of age.

Total PMPY spending on health care services was $238 535 in year 1, $27 666 in year 3, and $20 952 in year 5 (Supplemental Table 6). In total, per member health care spending for CMC with NI during their entire first 5 years of life was $345 702. The median per member health care spending for the entire first 5 years was $235 860 (IQR: $97 372–$450 520).

As we hypothesized, our study of Medicaid-enrolled children with early-onset NI found decreasing use of inpatient, ED, and outpatient services as they aged from 0 to 5 years. Spending on inpatient services remained the largest annual spending category in all years but decreased dramatically after the first year, when the distribution of spending shifted to a more even distribution across inpatient and outpatient care. To our knowledge, this study is the first to examine longitudinal health care use across inpatient and outpatient care settings for a cohort of young children with NI. Our subanalysis of CMC with NI revealed similar trends from birth to 5 years, although levels of use and spending were predictably higher for this group.

This study provides a population-based overview of health care use for children with early-onset NI over their first 5 years of life. Birth cohort analyses, such as ours, help to clarify the natural history of health care use and could inform development of innovative population-based programs. Targeted efforts to reduce inpatient use during the costly first year, which accounted for >90% of that year’s health care costs, may reduce overall spending. Strategies for clinicians to consider adopting to achieve this goal might include proactively educating families about how to recognize and manage acute medical problems including respiratory and dehydration-related conditions at home, optimizing family support from outpatient services such as home health, and improving timely access to outpatient specialty care through availability of same-day or urgent appointments. This study also suggests the need for further research to examine the specific reasons for inpatient and outpatient service use in this patient group. Such efforts could identify modifiable child, family, and health system-level factors to optimize use, maximize value, and decrease unnecessary costs.

Our study highlights the relatively small amount spent on primary care for children with NI (PMPY ∼$500), representing only 5% of annual spending. Other studies have found that two-thirds of children with NI do not have access to a primary care medical home and primary care systems struggle to coordinate services across the care spectrum for these patients with limited financial resources.²⁰  Given that more frequent well-child primary care visits have been associated with a lower likelihood of hospitalizations,²¹  in future studies, researchers should evaluate the optimal number of annual primary care visits for children with NI and examine enrollment in medical homes. It is telling that use of home health services decreased after these children’s first year by 82.9% to only <4% for those >1 year of age. Given home health’s potential promise to reduce expensive inpatient service use^22,23  and promote family functioning,¹⁶  future work is necessary to assess if this low level of use meets the needs of these children and their families.

The only service subcategory where use increased across the first 5 years was behavioral and mental health services, used most frequently by children receiving services for developmental delay, intellectual disabilities, attention-deficit/hyperactivity disorder, conduct disorders, and depressive disorders. Given that the majority of children with NI develop behavioral and mental health comorbidities by the age of 6,²  these children could benefit from proactive screening and optimization of behavioral and mental health conditions before school age, when behavioral problems interfere with academic and social success and lead to higher health and education resource use.^24–27 

In our study, we found that the majority of children with NI use multiple different health services across their early years, especially in their first year, which may highlight the potential need for proactive care coordination. Care coordination is defined as a team-based effort to organize and drive service integration to address medical, social, developmental, behavioral, educational, and financial needs.^28,29  Our finding that substantially fewer children with NI used inpatient services after their first year of life complements a growing body of evidence that many children with chronic conditions use fewer health care services after high-use periods.^3,4  As medical home interventions and value-based care models assess their impact on use and spending,^30–33  they should account for our findings of decreased use of health care services over time. Although this result may reflect medical stabilization of these children’s chronic diseases,^34,35  it might also be explained by more multifaceted health care processes.

Despite the decreasing trend in ED use over the first 5 years, the proportion of children with NI who used ED services each year remained substantial (>40% annually), even while use of inpatient services markedly decreased. This finding suggests that ED visits are prompted by different factors than hospitalizations. Previous studies have shown that families of children with chronic conditions such as NI are driven to seek emergency care by a complex mix of contextual factors: chronic disease decompensation, lack of access to ambulatory or specialty care, lack of caregiver self-efficacy, or absence of psychosocial supports.^36,37  These multiple factors result in variable patterns of ED use that may be modifiable.³⁸ 

Our study has several limitations. Because we used a deidentified claims-based database, we could not analyze family, community, or regional attributes that might influence use and spending. Analyzing use as a binary variable did not allow for examination of specific risk factors for intensity of use. Our findings do not reflect children’s use of services for which Medicaid claims were not submitted (eg, day care or transportation). Our definition of NI, although based on the most widely used NI coding system, also included disorders with heterogeneous levels of impairment. In addition, our sample is representative of children with NI enrolled in Medicaid and may not generalize to privately insured children.

Another limitation is that some level of selection bias may have been present, given that we assessed NI diagnoses during the first 6 months of life to define our study population and required participants to be continuously enrolled in Medicaid. Thus, we did not include those who were diagnosed with NI later in childhood, those who died before 5 years of age, or those with intermittent Medicaid coverage. Given these several possible sources of bias, the direction in either over or underestimating health care use is unknown. In addition, some early-onset NI conditions may be mild in severity or improve over time, whereas others could be persistent and may result in more consistent health care use.

Our findings may help to guide the design and evaluation plans of population-health based interventions for children with early-onset NI. Health system leaders could leverage our identified patterns of health care use to incentivize a redistribution of services for this population, and consider further optimization through early intervention, primary care, and/or home health. In future studies, researchers should continue exploration of subgroups of children with NI, particularly those with higher disease acuity and multimorbidity,³⁹  to examine how specific diagnoses, including mental and behavioral health, drive service use and spending.

We encourage providers to use our findings to counsel families of children with NI about how to plan for their child’s future health care needs. Parents appreciate receiving this information near the time of diagnosis, even if it is based on uncertain predictions, so they can prepare emotionally and logistically to navigate the health system.^40–44  For example, infants with NI are frequently admitted for respiratory conditions, so families of neonates with NI should be taught to recognize and manage respiratory decompensation. In addition, given our finding that approximately one-half of children with NI use ED services annually, clinicians could offer anticipatory guidance to help families implement plans for managing acute illness at home and proactively prepare for ED visits.

Children with NI enrolled in Medicaid showed trends of decreasing annual rates of use of inpatient, ED, and outpatient services (with the exception of mental and behavioral health) across their first 5 years of life. Annual use and spending on inpatient services markedly decreased after the first year but remained the largest spending category for these children each year. Health systems and policy makers should be encouraged to optimize care models for children with NI that anticipate longitudinal trends across service categories. Clinicians may leverage these findings when counseling families of children with NI to help them prepare in practical ways for shifting health care use patterns as their children age.

Constance D. Baldwin, PhD (Professor of Pediatrics, University of Rochester, Rochester, New York), offered valued mentorship and thoughtful revisions of the final manuscript.

CCC

complex chronic condition

CI

confidence interval

CMC

children with medical complexity

ED

emergency department

IQR

interquartile range

MAC

mean annual change

NI

neurologic impairment

OR

odds ratio

PMPY

per-member-per-year