Using State Administrative Data to Identify Social Complexity Risk Factors for Children

Kimberly C. Arthur; Barbara A. Lucenko; Irina V. Sharkova; Jingping Xing; Rita Mangione-Smith

doi:10.1370/afm.2134

Article Figures & Data

Tables

Additional Files

View popup

Table 1

Social Complexity Risk Factors Having Demonstrated or Hypothesized Association With Suboptimal Health Care Use Among Children

Risk Factor	Definition of Risk Factor in Administrative Data
Parent domestic violence^13,14	Any arrest for which a charge recorded is in a domestic violence crime category
Parent death^a	State health department death certificates for biologic parents
Parent mental illness^15–18	Mental health diagnosis, service encounters, procedures, or prescribed psychotropic medications recorded in medical claims or publicly funded mental health records for either biologic parent^b
Parent criminal justice involvement^a	Any arrest or conviction recorded by state patrol or court filings, including adjudication in state court database, for any crime category for either biologic parent
Child abuse/neglect¹⁷	Any family involvement (child or either biologic parent) in child welfare or child protective services system in Washington State
Homelessness^a	Indicates at least 1 period of homelessness, including shelter stays, recorded by a financial eligibility worker during eligibility (re)determination for public assistance for child or either biologic parent
Poverty^12,19–21	TANF benefit recorded for child^c
Parent has limited English proficiency or speaks a language other than English at home^12,22	Primary language other than English and indicated need for interpreter for biologic parent and/or child
Child mental illness^a	Mental health diagnosis, service encounters, procedures, or prescribed psychotropic medications recorded in medical claims or publicly funded mental health records for child aged ≥5 y^b
Child substance abuse^a	Substance-related diagnosis, service or encounters recorded in medical claims or publicly funded mental health records; any arrest for which a charge recorded is in a substance-related crime category (eg, driving under the influence, possession of controlled substance) for child aged ≥12 y^b
Child juvenile or criminal justice involvement^a	Any arrest or conviction recorded by state patrol or court filings, including adjudication in state court database, for any crime category for child aged ≥12 y

TANF = Temporary Assistance for Needy Families.
Note: Based on Washington State administrative data. Four risk factors could not be identified using the available administrative data: (1) low parent educational attainment, (2) single parent in household, (3) adolescent exposure to intimate partner violence, and (4) discontinuous insurance coverage.
↵a Risk factor hypothesized to have an association with suboptimal health care use.
↵b Full list of codes used to identify parent/child mental illness and child substance abuse is given in the Supplemental Appendix, available at http://www.annfammed.org/content/16/1/62/suppl/DC1/.
↵c The income of a TANF family of 3 in Washington State with any housing expenses was approximately 38% of the 2008 Department of Health and Human Services Poverty Guidelines.

View popup

Table 2

Cohort Characteristics (N = 505,367)

Characteristic	All Children		Children With ≥1 Emergency Department Visits

	<5 Years (n=180,198)	5–17 Years (n=325,169)	<5 Years (n=62,512)	5–17 Years (n=69,015)
Sex, female, %	48.8	49.1	47.2	49.0
Age, mean (SD), y	1.8 (1.4)	10.3 (3.6)	1.5 (1.4)	10.5 (3.8)
Race or ethnicity, %
Hispanic	29.5	27.3	34.6	32.0
Non-Hispanic
White	36.1	48.3	32.8	44.8
African American	5.7	8.4	6.1	8.7
Asian, Native Hawaiian, or other Pacific Islander	3.6	4.4	2.9	2.7
American Indian or Alaska Native	3.3	4.8	3.9	6.0
Two or more races	1.6	2.4	1.8	2.8
Unknown race/ethnicity	20.1	4.5	17.8	3.1
Medicaid eligibility, mean (SD), mo	9.9 (3.2)	10.3 (3.2)	11.5 (1.6)	11.5 (1.7)
Medical complexity, %
Healthy (no chronic disease)	84.2	78.8	81.2	70.6
Noncomplex chronic disease	11.0	15.9	12.8	20.9
Complex chronic disease	4.8	5.3	6.0	8.5
Social complexity risk factors, %^a
Parent domestic violence	4.5	4.4	5.4	5.4
Parent death	0.4	2.1	0.3	2.4
Parent mental illness	31.1	32.7	36.8	40.3
Parent criminal justice involvement	44.0	40.4	50.2	45.6
Child abuse/neglect	27.7	34.5	32.1	41.6
Homelessness	17.0	17.0	20.4	20.7
Poverty	26.8	23.4	34.3	30.8
Limited English proficiency	25.7	18.5	28.9	20.4
Child mental illness	n/a	18.4	n/a	24.1
Child substance abuse	n/a	1.9	n/a	3.7
Child juvenile or criminal justice involvement	n/a	3.0	n/a	5.3
Social complexity risk factors
0 factors, %	21.2	19.8	13.9	12.3
1 factor, %	30.4	26.8	28.2	23.3
2 factors, %	20.2	20.7	22.2	21.3
3 factors, %	13.5	14.7	16.3	17.1
4 factors, %	9.0	9.8	11.4	13.0
≥5 factors, %	5.8	8.2	7.9	13.0
Mean (SD), No.	1.8 (1.5)	2.0 (1.6)	2.1 (1.5)	2.4 (1.7)
Outcomes
Emergency department visits, mean (SD), No.	0.7 (1.3)	0.3 (0.8)	1.9 (1.4)	1.6 (1.2)

n/a=not applicable.
↵a Assessable factors from Table 1. A total of 8 factors were assessed in children aged <5 years and 11 factors were assessed in children aged 5–17 years, Supplemental Appendix, http://www.annfammed.org/content/16/1/62/suppl/DC1/.

View popup

Table 3

Adjusted Analysis Predicting the Number of Emergency Department Visits for Children Aged Younger Than 5 Years

Predictor	Model 1: Incidence Rate Ratio (95% Wald CI)	Model 2: Incidence Rate Ratio (95% Wald CI)
Social complexity risk factor
Parent domestic violence	0.96 (0.92–1.00)	–
Parent death	0.85 (0.74–0.98)	–
Parent mental illness	1.21 (1.19–1.24)^a	–
Parent criminal justice involvement	1.11 (1.09–1.13)^a	–
Child abuse/neglect	1.08 (1.06–1.10)^a	–
Homelessness	1.05 (1.03–1.08)^a	–
Poverty	1.25 (1.23–1.28)^a	–
Limited English proficiency	1.17 (1.14–1.19)^a	–
Number of social complexity risk factors
1 factor	–	1.24 (1.21–1.28)^a
2 factors	–	1.48 (1.43–1.52)^a
3 factors	–	1.62 (1.57–1.67)^a
4 factors	–	1.76 (1.70–1.83)^a
≥5 factors	–	1.92 (1.85–2.00)^a

Note: Adjusted for child sex, race/ethnicity, age, level of medical complexity, and length of Medicaid coverage (months).
↵a Significant at P <.0001 level.

View popup

Table 4

Adjusted Analysis Predicting the Number of Emergency Department Visits for Children Aged 5 to 17 Years

Predictor	Model 1: Incidence Rate Ratio (95% Wald CI)	Model 2: Incidence Rate Ratio (95% Wald CI)
Social complexity risk factor
Parent domestic violence	1.01 (0.97–1.05)	–
Parent death	1.03 (0.97–1.10)	–
Parent mental illness	1.17 (1.15–1.20)^a	–
Parent criminal justice involvement	1.02 (1.00–1.04)^a	–
Child abuse/neglect	1.13 (1.11–1.15)^a	–
Homelessness	1.09 (1.06–1.11)^a	–
Poverty	1.25 (1.23–1.28)^a	–
Limited English proficiency	1.04 (1.01–1.07)^a	–
Child mental illness	1.10 (1.08–1.13)^a	–
Child substance abuse	1.37 (1.29–1.45)^a	–
Child juvenile or criminal justice involvement	1.40 (1.33–1.46)^a	–
Number of social complexity risk factors
1 factor	–	1.21 (1.18–1.25)^a
2 factors	–	1.37 (1.32–1.41)^a
3 factors	–	1.52 (1.47–1.57)^a
4 factors	–	1.71 (1.65–1.77)^a
≥5 factors	–	2.06 (1.99–2.14)^a

Note: Adjusted for child sex, race/ethnicity, age, level of medical complexity, and length of Medicaid coverage (months).
↵a Significant at P <.0001 level.

Additional Files

Tables

Supplemental Appendix

Supplemental appendix

Files in this Data Supplement:

Supplemental data: Appendix - PDF file
The Article in Brief

Using State Administrative Data to Identify Social Complexity Risk Factors for Children

Kimberly C. Arthur , and colleagues

Background Identifying children with adverse childhood experiences is challenging but critically important, because early intervention has the potential to improve health across the lifespan. This study tests the feasibility of using an integrated state agency administrative database to identify childrens' social complexity risk factors and examine their relationship to emergency department usage.

What This Study Found State administrative data can be used to identify social risk factors for children. Researchers linked administrative data for more than 500,000 children receiving Washington State Medicaid insurance coverage with parent data to identify social complexity risk factors (individual, family, or community characteristics that can affect health outcomes), such as poverty and parent mental illness. They found that social complexity risk factors frequently co-occurred, with approximately one-half the study population having two or more risk factors. Of 11 identifiable risk factors, nine were associated with a higher rate of emergency department utilization. The magnitude of the association with the rate of emergency department utilization was small for individual risk factors, but the rate increased as the number of risk factors increased independent of medical complexity.

Implications

Providing primary care physicians with a social complexity flag or score, the authors suggest, could facilitate targeted screening of families who are likely to have social risk, with the goal of identifying families who could benefit from care coordination or other supportive services. This, in turn, would help make best use of limited time and resources in primary care.