Pediatric Health and System Impacts of Mass Incarceration, 2009–2020: A Matched Cohort Study

Abstract

Objective

The US has the highest incarceration rate in the world; incarceration’s direct and indirect toll on the health and healthcare use of youth is rarely investigated. We sought to compare the health of youth with known personal or family justice involvement and a matched cohort of youth without known personal/family justice involvement.

Methods

A cross-sectional matched parallel cohort study was conducted. We queried electronic health records on youth (< 21 years) with a visit in a large Midwestern pediatric hospital-based institution from January 2009 to December 2020. Youth were located by searching for justice-related (e.g. prison, jail) keywords within all clinician notes. Health diagnostic profiles were measured using ICD 9/10 codes. Healthcare use included total admissions, inpatient days, emergent and urgent visits, and outpatient visits.

Results

Across all youth at one institution over an 11-year period, 2.2% (N = 38,263) were identified as having probable personal or family justice-involvement. Youth with personal or familial justice involvement had 1.5–16.2 times the prevalence of mental health and physical health diagnoses across all domain groupings compared to a matched sample and the total population sample. From 2009–2020, approximately two-thirds of behavioral healthcare and nearly a quarter of all hospital inpatient days were attributed to the 2.2% of youth with probable personal or familial justice system involvement.

Conclusion

The study illuminates the vast disparities between youth with indirect or direct contact with the criminal legal system and matched youth with no documented contact. Better investment in monitoring and prevention efforts are needed.

Introduction

The United States (US) has, for several decades, maintained the highest rate of incarceration in the world.¹ In 2019 alone, over 722,000 youth interacted with juvenile justice systems,² nearly 6.3 million adults were incarcerated in a prison or jail or on probation or parole,³ and 600,000 adult prison admissions and 10.6 million adult jail admissions occured.⁴ Over a third of US adults have criminal records,⁴ with dire consequences for community and family health⁵ – especially in rural areas.^6,7 Racial and economic disparities exist at every level in the US criminal legal system^8,9 with African Americans, Native Americans, and individuals with lower educational attainment more likely to experience prison incarceration.^10,11 One in five US adults have had a parent incarcerated¹² and one in seven have had an immediate family member spend one year or longer in prison.¹²

National agencies^13,14 acknowledge the need for greater collaboration among health, criminal legal, and child welfare systems for all youth to thrive. Despite high youth exposure to the criminal legal system (family or personal), gaps remain in understanding its prevalence and consequences. This is partially because few pediatric health systems screen for such exposures;¹⁵ California is an exception.¹⁶ Routine screening of adverse childhood experiences is controversial, partially due to health providers’ limited time and training, and limited institutional resources and responses upon disclosure.^17,18 In addition, it may be important to respect youth and families’ choice to not disclose criminal legal system involvement due to perceived intrusiveness, trauma, fear of child protection services involvement, stigma, and/or judgement.^19,20

Most of what we know about criminal legal system exposures and child and young adult health comes from household-based surveys (missing nontraditionally-housed youth).^21,22 Research is also limited by small sample sizes, inadequate comparison groups, and self or parent reported outcomes.^22,23 Recent reviews call for higher quality data, rigor and information on physical health and healthcare use.^22,23 Prior studies link parental incarceration to worse mental health,^22,24 physical health,²⁵ school outcomes,^26,27 and foster care participation.²⁸ Prior studies on the health of incarcerated youth have mostly focused on poorer mental health and increased substance use, neurodevelopmental disabilities, and high risk sexual behaviors.^23,29 Very few studies exist on the effects of sibling or peer incarceration, a key gap in the literature.

Greater understanding of the role of mass incarceration on health and health disparities in youth is needed to raise awareness and to effect change. Arguably, it is methodologically difficult to evaluate the effect of personal or familial justice involvement on child health because the risk factors preceding incarceration are also social determinants of poor health. However, studies show an independent effect of personal or familial incarceration on youth health and behavior, beyond other associated disadvantages,²² and strategic comparisons to youth who experience accumulative adversities.³⁰ Studies utilizing data that represent large, community-based populations to explore the health of youth with any type of contact with the justice system are needed. Using data from 2009–2020, we sought to identify youth with personal or family-justice involvement known to a large pediatric health institution and compare their health and healthcare service use characteristics to sociodemographically matched youth who do not have known personal or family justice-involvement (or were not identified as such). There is no standard way for a clinician to ask about personal or family justice involvement; thus, our comparison sample includes both individuals without justice involvement and those who have had justice involvement that is not known to the healthcare system, the most salient limitation of our study.

Methods

Study Design

A cross-sectional matched parallel cohort study was conducted using electronic health records (EHR) from a large Midwestern US pediatric health institution. Patient-level characteristics were drawn from the most recent EHR data available at the time of analysis. The University of Cincinnati (2021–0178) and Cincinnati Children’s Hospital Medical Center (2021–0256) Institutional Review Boards approved this study. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.

Setting

EHR data stored in Epic^® (Verona, WI) were queried for 1,747,336 youth under 21 years of age with at least one recorded healthcare encounter between January 1, 2009 and December 31, 2020. The healthcare system provides care for more than 1.5 million patient visits annually and includes a network of primary care centers, behavioral health clinics, urgent care clinics, two emergency departments and 622 total in-patient registered beds.

Youth with Probable Personal or Familial Justice Involvement

Like most pediatric hospital-based institutions in the US, this institution does not require routine screening for personal or familial incarceration. In turn, we electronically searched for instances in which families disclosed and providers reported justice involvement. To identify and extract information on youth with probable personal or familial justice system involvement, we queried justice and familial keywords over the EHR. Consistent with prior work,^31,32 we chose justice terms to capture all four types of detainments following arrest in the US. Justice-based query words included: (“prison”, “sentenced”, “incarcerated”, “probation”, “parole”, “jail”, and “criminal”) OR the following smartform text entered by providers or clinical staff (“Family Hx of Incarceration? -Yes” OR “Legal charges? -Yes”). We included “parent and guardian” keywords to capture the health records of youth exposed to parental incarceration. These familial keywords included: (“father”, “dad”, “mother”, “mom”, “parent”, “grandpa”, “grandma”, “caregiver”, and “legal guardian”). Every type of clinical note in the medical record was searched. We filtered notes that only had these keywords: “incarcerated hernia” and the following provider smartform text: (“Family Hx of Incarceration? No” OR “Criminal Justice Involvement? -No” OR “History of Legal Charges? -No” or “Legal Charges? -No”) to decrease the number of false-positives.

Matched Cohort

The comparison sample had no justice keywords in their chart and was 1:1 matched to the exposed sample via PL/SQL, matched on gender, date of birth within 3 months, race, ethnicity, insurance type, and zip code as reported by the patient or family via routine hospital data collection. Due to over representation of minorities, especially Black or African American, in juvenile and adult correctional facilities, race/ethnicity was used in our matching strategy. If patients were unable to be matched by zip code, then county was used.

Diagnostic Measures

Health diagnoses were measured via the International Classification of Diseases Version 9/Version 10 (ICD) codes and were extracted from the EHR as time-invariant total number of diagnoses from 2009–2020. Mental health disorders were grouped consistent with the Children’s Hospital Association Child and Adolescent Mental Health Disorders Classification System (CAMHD-CS).³³ The 30 groupings classify child mental health disorders across ICD-9/10 diagnostic codes and aligns with the Diagnostic and Statistical Manual of Mental Disorders (DSM-V) diagnostic groups.³³ Physical health and other chronic medical conditions were comprehensively grouped by the affected domain or system (e.g. abdominal pain, allergy, immunology, asthma), previously reported elsewhere.³⁴

Healthcare Use

Healthcare use included total hospital admissions, inpatient days, emergent and urgent care visits, behavioral health hospitalizations, behavioral health inpatient days, and primary care visits at this institution from 2009–2020. Total foster care clinic visits from 2012 (when it opened) to 2020 were also calculated.

Statistical Analysis

Data were aggregated from the EHR using Oracle PL/SQL. Summary statistics described the characteristics of youth identified by justice keywords, a matched comparison sample, and the total pediatric population. Our two primary cohorts were matched; therefore, all prevalence estimates reported are unadjusted. Prevalence ratios and prevalence differences were calculated to compare the likelihood of diagnostic characteristics and healthcare use between youth with a justice keyword in their chart and matched youth. We used Microsoft 365 Excel, version 2403 for the analysis.

We also extracted 1,000 random clinician notes for trained graduate students to manually review and annotate for type of justice involvement. N = 1,000 notes exceeds the required minimum sample of 791 to an estimate a 95% confidence interval with a +/− 1% margin of error using the baseline estimate of 2% from prior work.³¹

Results

Patient Characteristics

Out of the 1,747,336 unique patients, 38,263 (2.2%) had a justice keyword in their health records (Table 1). Nearly half of youth with a justice keyword identified as male, and 63.0% identified as white and non-Hispanic (95.3%). Age was calculated at the time of the data pull and therefore, most of the patients identified by the keyword search were ages 13 and older (80.0%). Most youth (82.8%) had Medicaid/SCHIP health insurance. and over a third of all patients with a Child in Welfare Custody z-code (Z62.21) also had a justice keyword in their medical chart (36.5% or 4,548 out of 12,462).

Table 1.

Demographic information of patients identified by justice keywords and all patients (ages 0–21) in electronic health record (EHR) database from January 2009–2020

Patient Characteristics	Patients with Justice Keywords in the EHR n = 38,263		All Patients in Database n = 1,747,336
	na	%	nb	%
Gender
Female	18695	48.9%	894064	51.2%
Male	19566	51.1%	850040	48.7%
Unknown	2	0.01%	1843	0.1%
No Value	0	0.00%	1389	0.1%
Age Range **
0–4 years	978	2.6%	141019	8.1%
5–9 years	3900	10.2%	162772	9.3%
10–12 years	2768	7.2%	90823	5.2%
13–18 years	8722	22.8%	246709	14.1%
19–21 years	21895	57.2%	1106013	63.3%
Race
Black/African American	10773	28.2%	223357	12.8%
White	24092	63.0%	1117766	64.0%
Unknown	520	1.4%	117617	6.7%
Multiple Race	1599	4.2%	26384	1.5%
Asian Race	175	0.46%	27353	1.6%
Other	921	2.4%	103565	5.9%
Native Hawaiian	30	0.08%	1643	0.1%
Refuse to Answer	52	0.14%	1423	0.1%
American Indian or Alaskan Native	35	0.09%	1692	0.1%
Middle Eastern	28	0.07%	2369	0.1%
No Value	38	0.10%	124167	7.1%
Ethnicity
Not Hispanic or Latino	36474	95.3%	1029071	58.9%
Hispanic or Latino	1286	3.4%	52184	3.0%
Unknown/Patient Refused	465	1.2%	119281	6.8%
No Value	38	0.10%	546800	31.3%
Health Insurance Coverage ***
Medicaid/SCHIP	31682	82.8%	526761	30.2%
Private/Commercial	17940	46.9%	830484	47.5%
Self-Pay	1024	2.7%	13041	0.8%
Other-Unknown	2350	6.1%	34319	2.0%
Medicare	543	1.4%	57566	3.3%
Foster Care or Child Welfare Involvement
Child in Welfare Custody - Z62.21	4548	11.9%	12462	0.7%

Notes: Analysis used electronic health record data of a large Midwestern pediatric hospital-based institution. (N = 1.7 million unique patients ages 0–21 years). Correctional query words algorithm used in keyword supported search: (“prison” or “sentenced” or “incarcerated” or “probation” or “parole” or “jail” or “criminal” ) AND (“father” or “dad” or “mother” or “mom” or “parent” or “grandpa” or “grandma” or “caregiver” or “legal guardian”) OR (“Family Hx of Incarceration Yes” OR “legal charges YES” ).

^*Number of patients with the given characteristic and a correctional keyword in their medical chart out of the total population.

^**Age range indicates current age of the youth in the system and not age at time of possible exposure to family history or personal history of correctional involvement.

^***Health insurance coverage: counts do not add up as there can be multiple types of insurance used for visit/per patient.

Mental Health Diagnoses

Among youth with a justice keyword (n = 38,263) in the health record (Table 2), the top three mental health diagnostic groupings were trauma and stressor-related disorders (36.7%), depressive disorders (36.7%), and attention deficit and hyperactivity disorders (36.0%). Youth with a justice keyword had 3.0 to 15.5 times the prevalence of mental health disorders in every diagnostic grouping. Youth with a justice keyword made up a large proportion of all youth ever diagnosed with mental health disorders, including 42.9% of all schizophrenia spectrum and other psychotic disorders, 42.1% of all bipolar and related disorders, 38.3% of suicide or self-injury disorders, 37.5% of all substance related and addictive disorders and 35.2% of all personality disorders. On average, youth with a justice keyword had 4.5 times the prevalence of a mental health diagnosis. Among youth with a justice keyword, there were approximately 269.2 more mental health diagnoses per 100 youth compared to matched youth.

Table 2).

Mental health disorder information (using Children’s Hospital Association ICD9/10 groupings) of patients identified by justice/family keywords and all patients (ages 0–21) in electronic health record (EHR) database from January 2009–2020 and compared to a matched sample of youth (with no justice keyword and matched on gender, race/ethnicity, age, insurance, and zip code).

Mental Health Disorder Groupings	Patients with Justice Keywords in the EHR n = 38,263		Matched Patients with NO Justice Keywords in the EHR* n = 38,263		Prevalence Ratio na / nb	Prevalence Difference (na - nb)* 100	All Patients in Database N = 1,747,336		% of All Patients with Justice Keyword**/of Total Population with Each Mental Health Disorder (na/nc)
	na	%	nb	%	na/ nb	(na - nb)* 100	nc	%	%
Accidental or Undetermined Poisoning	288	0.8%	36	0.1%	8.00	0.66	1054	0.1%	27.3%
Attention Deficit Hyperactivity Disorders	13773	36.0%	3755	9.8%	3.67	26.18	77900	4.5%	17.7%
Anxiety Disorders	11686	30.5%	3958	10.3%	2.95	20.20	89878	5.1%	13.0%
Autism Spectrum Disorder	2308	6.0%	773	2.0%	2.99	4.01	20705	1.2%	11.2%
Bipolar and Related Disorders	3280	8.6%	308	0.8%	10.65	7.77	7801	0.5%	42.1%
Communication Disorders	8658	22.6%	2533	6.6%	3.42	16.01	59470	3.4%	14.6%
Depressive Disorders	14027	36.7%	2475	6.5%	5.67	30.19	54526	3.1%	25.7%
Developmental Delay or Unspecified Neurodevelopmental Disorder	5892	15.4%	1870	4.9%	3.15	10.51	45234	2.6%	13.0%
Disruptive, Impulse Control and Conduct Disorders	12957	33.9%	2529	6.6%	5.12	27.25	50072	2.9%	25.9%
Dissociative Disorders	93	0.24%	6	0.0%	15.50	0.23	269	0.0%	34.6%
Fetal or Newborn Damage Related to Maternal Substance Abuse	1122	2.9%	228	0.6%	4.92	2.34	4732	0.3%	23.7%
Intellectual Disability	2700	7.1%	498	1.3%	5.42	5.75	11342	0.7%	23.8%
Mental Health Symptom	11133	29.1%	2362	6.2%	4.71	22.92	53527	3.1%	20.8%
Neurocognitive Disorders	1950	5.1%	462	1.2%	4.22	3.89	10738	0.6%	18.2%
Obsessive-Compulsive and Related Disorders	1387	3.6%	371	1.0%	3.74	2.66	9862	0.6%	14.1%
Personality Disorders	1821	4.8%	200	0.5%	9.11	4.24	5177	0.3%	35.2%
Schizophrenia Spectrum and Other Psychotic Disorders	2001	5.2%	204	0.5%	9.81	4.70	4668	0.3%	42.9%
Specific Learning Disorders	7243	18.9%	2305	6.0%	3.14	12.91	54702	3.1%	13.2%
Substance-Related and Addictive Disorders	5491	14.4%	610	1.6%	9.00	12.76	14628	0.8%	37.5%
Suicide or Self-Injury	10956	28.6%	1322	3.5%	8.29	25.18	28621	1.6%	38.3%
Trauma and Stressor-Related Disorders	14058	36.7%	3021	7.9%	4.65	28.85	57472	3.3%	24.5%
Total MH Disorder Codes	13824	3.47 codes per child	29826	0.78 codes per child	4.45	269.18	662378	0.38 codes per child	20.1%

Notes: Analysis of electronic health record data of a large Midwestern pediatric hospital-based institution. (N = 1.74 million unique patients ages 0–21 years). Justice query words used in keyword supported search: (“prison” or “sentenced” or “incarcerated” or “probation” or “parole” or “jail” or “criminal” ) AND (“father” or “dad” or “mother” or “mom” or “parent” or “grandpa” or “grandma” or “caregiver” or “legal guardian”) OR (“Family Hx of Incarceration Yes” OR “legal charges YES” ).

^*Patients matched on gender, race, ethnicity, age, insurance, and location (zip code match (34,474), then county (2,487), or no zip/count (1,302)). Number of patients with the given characteristic and a justice keyword in their medical chart out of the total population. All patient health characteristics are represented International Classification of Diseases Version 9 or Version 10 codes and the italicized words indicate the diagnostic keyword searched within the medical record unless the specific ICD/CPT code is listed. Diagnostic codes and characteristics are not mutually exclusive.

Physical Health Diagnoses

Among youth with a justice keyword in their health record (n = 38,263), the top three physical health diagnostic groups were neurodevelopmental diagnoses (69.7%), neurological diagnoses (59.7%), and pregnancy related diagnoses (59.7%) (Table 3). Youth with a justice keyword had 1.3 to 16.2 times the prevalence of specific physical health disorders in every diagnostic grouping. Youth with a justice keyword made up a large proportion of all youth diagnosed with physical health disorders including 44.9% of all shaken baby syndrome diagnoses, 37.4% of all sexual risk diagnoses, 26.7% of in utero exposure diagnoses, 15.0% of all sexually transmitted infections, and 14.2% of all abuse, neglect, or maltreatment diagnoses. Youth with a justice keyword had 1.93 times the prevalence of a physical health diagnosis; they also had 428.2 more physical health diagnoses per 100 youth than the matched youth.

Table 3).

Physical health disorder information of patients identified by justice keywords and all patients (ages 0–21) in electronic health record (EHR) database from January 2009–2020 and compared to a matched sample of youth (with no justice keyword and matched on gender, race/ethnicity, age, insurance, and zip code).

Physical Health Disorder Groupings	Patients with Justice Keywords in the EHR n = 38,263		Matched Patients with NO Justice Keywords in the EHR* n = 38,263		Prevalence Ratio na / nb	Prevalence Difference (na - nb)* 100	All Patients in Database N = 1,747,336		% of All Patients with Justice Keyword**/of Total Population with each Physical Health Disorder (na/nc)
	na	%	n	%	na/ nb	(na - nb)* 100	nc	%	%
Abdominal pain	11574	30.3%	6966	18.2%	1.66	12.04	150345	8.6%	7.7%
Abuse/Neglect/Maltreatment	15993	41.8%	5982	15.6%	2.67	26.16	112432	6.4%	14.2%
Allergy and Immunology	10989	28.7%	5515	14.4%	1.99	14.31	116720	6.7%	9.4%
Asthma	8459	22.1%	4407	11.5%	1.92	10.59	89125	5.1%	9.5%
Back pain	11392	29.8%	6623	17.3%	1.72	12.46	144697	8.3%	7.9%
Cardiology	11461	30.0%	5770	15.1%	1.99	14.87	139025	8.0%	8.2%
Chest pain	4970	13.0%	2548	6.7%	1.95	6.33	51020	2.9%	9.7%
Endocrinology	8438	22.1%	3501	9.2%	2.41	12.90	72055	4.1%	11.7%
Extremities, Head, Face, and Joint Pain	17577	45.9%	11684	30.5%	1.50	15.40	244893	14.0%	7.2%
Gastroenterology	19621	51.3%	12384	32.4%	1.58	18.91	292625	16.8%	6.7%
Genetics	3765	9.8%	1647	4.3%	2.29	5.54	46322	2.7%	8.1%
Gynecology	4168	10.9%	1905	5.0%	2.19	5.91	39154	2.2%	10.7%
Headache	9509	24.9%	4954	13.0%	1.92	11.90	96511	5.5%	9.9%
Hematology	3901	10.2%	1821	4.8%	2.14	5.44	61416	3.5%	6.4%
Hepatology	1395	3.7%	460	1.2%	3.03	2.44	13052	0.8%	10.7%
Infectious Disease	195	0.5%	47	0.1%	4.15	0.39	1400	0.1%	13.9%
Neonatology	282	0.7%	222	0.6%	1.27	0.16	6559	0.4%	4.3%
Nephrology	2214	5.8%	1041	2.7%	2.13	3.07	37072	2.1%	6.0%
Neurodevelopmental	26658	69.7%	13643	35.7%	1.95	34.01	312546	17.9%	8.5%
Neuorology	22854	59.7%	11854	31.0%	1.93	28.75	261119	14.9%	8.8%
Oncology	243	0.6%	78	0.2%	3.12	0.43	6138	0.4%	4.0%
Opthalmology	18660	48.8%	9752	25.5%	1.91	23.28	202608	11.6%	9.2%
Orthopeadics	7876	20.6%	5047	13.2%	1.56	7.39	123985	7.1%	6.4%
Other pain	14708	38.4%	8861	23.2%	1.66	15.28	194104	11.1%	7.6%
Otolaryngology	15474	40.4%	10774	28.2%	1.44	12.28	235010	13.5%	6.6%
Pregnancy	22853	59.7%	8989	23.5%	2.54	36.23	178251	10.2%	12.8%
Pulmonology	11741	30.7%	6273	16.4%	1.87	14.29	135331	7.7%	8.7%
Rheumatology	2033	5.3%	1200	3.1%	1.69	2.18	34254	2.0%	5.9%
Sexual Risk	4506	11.8%	612	1.6%	7.36	10.18	12047	0.7%	37.4%
Sports Medicine	941	2.5%	590	1.5%	1.59	0.92	13027	0.8%	7.2%
Sexually Transmitted Infections	4705	12.3%	2199	5.8%	2.14	6.55	31390	1.8%	15.0%
Surgery	1288	3.4%	396	1.0%	3.25	2.33	10278	0.6%	12.5%
Urology	5648	14.8%	3142	8.2%	1.80	6.55	71064	4.1%	8.0%
Hearing/Vision	6710	17.5%	4096	10.7%	1.64	6.83	90593	5.2%	7.4%
In Utero Exposure	1182	3.1%	231	0.6%	5.12	2.49	4450	0.3%	26.6%
Lead Exposure	2029	5.3%	813	2.1%	2.50	3.18	18237	1.0%	11.1%
Neurogenetic	11983	31.3%	6609	17.3%	1.81	14.04	165946	9.5%	7.2%
Shaken Baby Syndrome	97	0.3%	6	0.0%	16.17	0.24	216	0.0%	44.9%
Speech/Language	12415	32.5%	4010	10.5%	3.10	21.97	99046	5.7%	12.5%
Physical Health Disorder Totals	340507	8.90 codes per child	176652	4.62 codes per child	1.93	428.23	3914063	2.24 codes per child	8.7%

Notes: Analysis of electronic health record data of a large Midwestern pediatric hospital-based institution. (N = 1.74 million unique patients ages 0–21 years). Justice query words used in keyword supported search: (“prison” or “sentenced” or “incarcerated” or “probation” or “parole” or “jail” or “criminal” ) AND (“father” or “dad” or “mother” or “mom” or “parent” or “grandpa” or “grandma” or “caregiver” or “legal guardian”) OR (“Family Hx of Incarceration Yes” OR “legal charges YES” ).

^*Patients matched on gender, race, ethnicity, age, insurance, and location (zip code match (34,474), then county (2,487), or no zip/count (1,302)).

^**Number of patients with the given characteristic and a justice keyword in their medical chart out of the total population. All patient health characteristics are represented International Classification of Diseases Version 9 or Version 10 codes and the italicized words indicate the diagnostic keyword searched within the medical record unless the specific ICD/CPT code is listed. Diagnostic codes and characteristics are not mutually exclusive.

From 2009–2020, youth with a justice keyword had 4.0 times the prevalence of a hospital admission (non-behavioral health related), 2.1 times the prevalence of an emergent or urgent care visit, 21.7 times the prevalence of a behavioral health admission, 2.1 times the prevalence of a primary care visit, and 4.4 times the prevalence of a foster care visit compared to matched youth without a justice keyword (Table 4). Approximately 67.4% of the behavioral health hospital care, 23.7% of all hospitalization inpatient days (non-behavioral health-related), 11.5% of all emergency and urgent care visits, and 45.5% of all foster care visits from 2009–2020 were attributed to the 2.2% of youth who had justice keywords in their chart.

Table 4).

Healthcare use information of patients identified by justice/family keywords and all patients (ages 0–21) in electronic health record (EHR) database from January 2009–2020 and compared to a matched sample of youth

Healthcare Use Variable	Patients with Justice Keywords in the EHR n = 38,263		Matched Patients with NO Justice Keywords in the EHR* n = 38,263		Prevalence Ratio na/ nb	Prevalence Difference (na - nb)* 100	All Patients in Database N = 1,747,336		% of All Patients with Justice Keyword**/of Total Population (na/nc)
	na	na/n	nb	nb/n	na/ nb	(na - nb)* 100	nc	nc/N	%
Total hospital admissions 2009-2020 (non-behavioral health related)	41780.0	1.09 per child	10480.0	0.27 per child	3.98	81.80	248822.0	0.14 per child	16.8%
Total inpatient days 2009–2020 (non-behavioral health related)	384962.0	10.06 per child	65040.17	1.70 per child	5.92	836.11	1626182.2	0.93 per child	23.7%
Total emergency department 41 and urgent care visits, 2009–2020	216368.0	5.65 visits per child	100977	2.64 visits per child	2.14	301.57	1884722.0	1.08 visit per child	11.5%
Total behavioral health inpatient admissions, 2009–2020	26042.0	0.68 admissions per child	1231.0	0.03 admissions per child	21.16	64.84	41163.0	0.02 admissions per child	63.3%
Total behavioral health inpatient days, 2009–2020	239149.0	6.25 days per child	11369.40	0.30 days per child	21.03	595.30	354757.45	0.20 days per child	67.4%
Total primary care visits, 2009–2020	33254.0	0.87 primary care visits	15708.0	0.41 primary care visits	2.12	45.86	231668.0	0.13 primary care visits per child	14.4%
Total foster care visits, 2009–2020	11597.0	0.30 foster care visits per child	2630.0	0.07 foster care visits per child	4.41	23.44	25513.0	0.01 foster care visits per child	45.5%

Notes: Analysis of electronic health record data of a large Midwestern pediatric hospital-based institution. (N = 1.74 million unique patients ages 0-21 years). Justice query words used in keyword supported search: (“prison” or “sentenced” or “incarcerated” or “probation” or “parole” or “jail” or “criminal” ) AND (“father” or “dad” or “mother” or “mom” or “parent” or “grandpa” or “grandma” or “caregiver” or “legal guardian”) OR (“Family Hx of Incarceration Yes” OR “legal charges YES” ).

^*Patients matched on gender, race, ethnicity, age, insurance, and location (zip code match (34,474), then county (2,487), or no zip/count (1,302)).

^**Number of patients with the given characteristic and a justice keyword in their medical chart out of the total population

Chart Validation of Query: Types of Justice Involvement

About 66% (n = 664) of clinical notes with justice keywords indicated some type of personal or family involvement with the justice system, and 30.3% did not indicate any involvement, and 3.3% of notes were unclear (Figure 1). Of the 664 notes that indicated involvement with the justice system, 59.2% indicated parental justice-involvement (maternal or paternal), 25.6% indicated self/youth, and 8.0% indicated other family (e.g., sibling).

Figure 1.

Results of manually annotating 1000 random clinical notes with justice keywords for type of justice-system involvement

Of the 303 notes that did not indicate personal or family justice, most (56.1%) were because justice keywords were used as a description term (e.g., to describe academic probation, to describe food (“prison food”), to describe fears (“this will get me thrown in jail”), or to describe future goals (“to study criminal justice”). 21.8% of the notes were because of other provider smartform text, a semi-structured, or unstructured field that indicated no contact with the system. 16.5% of the notes were not indicative of justice system involvement because a justice keyword was in a medical condition not excluded from the analysis (e.g., “incarcerated” inguinal/umbilical hernia, “jailing”), 4.0% were because of a different tuberculosis risk assessment question that was not previously excluded.

Discussion

This study is the first to comprehensively examine the health records and healthcare use of youth with probable personal or familial justice involvement known to the healthcare system compared to a matched sample of youth at one large Midwestern US pediatric health institution using data from 2009–2020. Only 2.2% of all youth had a justice keyword in their health record (n = 38,263 patients out of 1,747,336 unique patients). This is likely a gross underestimation of the true exposure of personal or family incarceration in youth served by this institution. About 7% of all US youths have had a parent incarcerated,³⁵ and nearly half of all US adults have had a family member incarcerated.¹² Few pediatric health systems screen for personal or family incarceration; therefore, we were limited to electronically searching through clinical notes for instances of family and provider reporting. In turn, our measure of contact with the criminal legal system represents an unverified proxy measure of exposure to mass incarceration that is known to the healthcare system. Despite a gross underestimation, our results depicted a high prevalence of mental and physical health diagnoses and healthcare use in a small proportion of youth with justice keywords in their chart that indicate probable (66.4% confirmed) personal or family justice-system involvement. In turn, these findings represent the population of children for whom personal or family incarceration was known and documented in the electronic health record. Importantly, the sample identified by justice keywords were disproportionately Black/African American in line with national estimates but interestingly, the sample was nearly split by male/female gender despite more male youth being disproportionately incarcerated at higher rates compared to female youth. Broadening the measurement to include multiple types of exposures of the criminal legal system may allow for greater understanding of the impacts of mass incarceration.

Compared to a matched sample without justice-system documentation and to the total population, youth with any disclosed contact with the justice system had higher rates of physical and mental health disorder diagnoses across all groupings and higher healthcare use. From 2009–2020, nearly 63.3% of all behavioral health inpatient admissions, 23.7% of all hospitalization inpatient days and 45.5% of all foster care visits were attributed to the 2.2% of youth that had documented probable personal or family justice system involvement. These findings confirm and extend prior work³¹ at another pediatric institution. These population-level investigations are the only large-scale studies in the nation to attempt to aggregate the health records and clinician notes of youth with probable personal or family exposure to the justice system compared to a matched sample. Our findings reveal the need to address the ‘weathering’ and toxic stress impacts that stem from the criminal legal system on the health of youth, with implications for prevention and treatment of health conditions. For example, if a child is experiencing a mental health crisis and concurrently experiencing family-member incarceration – the treatment plan could include addressing the trauma of familial loss and mitigating the family’s social, economic, and legal strain related to this exposure. In line with recent recommendations from the American Academy of Pediatrics, youth who are justice-involved should also receive the same level and standards of medical, oral, mental health, and substance use care as non-detained youth and be confined in developmentally appropriate facilities with comprehensive youth programming.³⁶

Our study finds vast disparities between youth with documented probable contact to the justice system and matched youth. Youth with justice-involved family members and youth who have or had justice involvement represent important groups for close health and social service follow-up. Although parental incarceration and juvenile detainment are considered adverse childhood experiences, few health systems regularly screen for such exposures.^15,16 Few, if any, studies that address how to screen for personal or family justice-involvement in highly sensitive and protective ways. Importantly, upon arrest of a parent or family member, few families are followed up to address trauma screening and social service linkages that may assist in mitigating stress and physical and mental health problems. There are budding recommendations addressing how to care for youth with incarcerated parents³⁷ or youth in juvenile facilities,³⁸ yet few evidenced-based recommendations and scientific investigations exist.³⁹ Less than 0.5% of all research projects ever funded at the National Institutes of Health (NIH), National Sciences Foundation, and Department of Justice were about “incarceration” (0.11% total) or justice-involved youth (0.04% total) and only 23 projects were funded that related to “incarcerated parents” since 1985.³⁹ As the American Public Health Association⁴⁰ recommends, funding research related to the effectiveness of alternatives to incarceration and policy determinants of exposure to carceral system would likely benefit the public health and safety of the United States – especially for youth. If future research confirms our findings, it may provide support to address justice-health policy reforms (e.g. decarceration reforms, bail reforms to decrease parents who sit in jail longer due to poverty, enact 12 years as a minimum age of jurisdiction as advocated by leading national health organizations⁴¹). In addition to the urgent development of behavioral health interventions to decrease health disparities in youth,⁴² and other initiatives to ensure developmentally appropriate confinement facilities and better delivery of care for youth within the juvenile justice system.³⁶ Importantly, in order for health, criminal legal and child welfare systems to better collaborate for all youth to thrive as advocated by the National Academies of Science, Engineering, and Medicine,¹³ families affected by criminal legal system involvement should serve as leaders and partners in this greater work to adequately ensure that youth and families feel supported upon disclosure of criminal legal system involvement in the health care settings rather than experience continued punishment or further harm from child protective or criminal legal services involvement.⁴³

Limitations

Our study has limitations. Our counts of criminal legal involvement are markedly underestimated, as there is no routine screening for personal or familial incarceration at this institution. Data reflect families who disclosed and health providers who documented; our data miss families who refrain from disclosing, or whose information is not documented. In turn, these results only portray a proxy measure of unverified and probable exposure to family or personal justice involvement. Further, health providers are typically untrained in legal terminology and may be unreliable recorders on types of justice involvement. Considering the size, churn, and impact of mass incarceration on children and families in the United States,^22,44,45 it is unfortunate that we are limited to electronically searching clinician notes for justice-related keywords. Until appropriate routine incarceration-related screenings, health-justice cross-sector data linkages, or novel cohort identification methods³² are fully developed and implemented – it is still important to explore the data with methods that do exist (however limited) in efforts to uncover the magnitude and impact that mass incarceration has on the health of children and youth. Importantly, about a third of the 1000 clinician notes randomly sampled from the justice keywords group indicated a “false positive” and suggest that our results may overestimate total non-behavioral admissions (related to youth who have “incarcerated” hernias and other serious physical conditions that contain a justice keyword) and underestimate behavioral healthcare usage. We advise caution in interpretation of the results and direct solutions from this study. This study serves as a starting point to inform better ways to identify, and support children and families affected by criminal legal system involvement until better methods³² are available or routine screenings occur. Last, our matched comparator group of youth could have experienced justice involvement that was not disclosed or documented. Despite these limitations, this cross-sectional matched parallel cohort study provides the first diagnostic and healthcare use snapshot of youth with a probable history of personal or familial justice-involvement. Our sample size and time frame provide an important contribution and call to further public health action for better screening, support, and care coordination for children and families affected by incarceration. Future research is needed and must refine and expand these methods (e.g. prospective studies) to adjust for temporality and other socioeconomic factors, and to inform policy. If these findings are replicated in other samples and health institutions, strong justification exists for decarceration efforts and bail or other criminal legal system reforms, especially if we want all US children and families to thrive.

Conclusions

Pediatric health systems can do better in identifying and supporting youth and families affected by the criminal legal system. Alleviating the negative effects of mass incarceration is an important public health and justice issue that can no longer be ignored. Maintaining such high incarceration rates are bound to affect generations to come, and as such, it would behoove our communities and pediatric systems to better understand how to best intervene and inform comprehensive child health equity policy and programming for children and families affected by incarceration. Future research is needed to inform development and evaluation of preventative and behavioral health interventions for children directly or indirectly affected by the criminal legal system. If we want all children to thrive, we must ensure that youth and families directly or indirectly affected by criminal legal system involvement are also supported.

Abbreviations:

US

United States

ICD

International Classification of Diseases Version 9/Version 10

EHR

electronic health records

CAMHD-CS

Children’s Hospital Association Child and Adolescent Mental Health Disorders Classification System

DSM-V

Diagnostic and Statistical Manual of Mental Disorders