Social Determinants of Health ICD-10 Code Use in Inpatient Pediatrics

We conducted a retrospective cohort study of the Kid’s Inpatient Database (KID) using 2016 and 2019 data.²²  The KID was developed as part of the Healthcare Cost and Utilization Project sponsored by the Agency for Healthcare Research and Quality and is released approximately every 3 years. The KID is the largest publicly available, all-payer pediatric inpatient care database in the United States, and includes deidentified administrative data from approximately 4000 hospitals in 48 states plus the District of Columbia. Hospitalizations are sampled at a rate of 10% of uncomplicated in-hospital births and 80% of all other pediatric discharges (aged <21 years) to yield data on approximately 3 million pediatric discharges each year.²³  This sample is then weighted to create a nationally representative estimate of the roughly 7 million hospital discharges per year for patients younger than 21 years of age.²³ 

The primary ICD-10 codes related to SDOH fall within the Z-code categories of Z55 to Z65. These codes include social risk factors (ie, food insecurity), social needs (ie, unsheltered homelessness), and ACEs (ie, substance use among family members).^1–5  Because this group of Z-codes was designed primarily to address adult needs, our team used an iterative process to identify a more comprehensive list of SDOH ICD-10 codes relevant to pediatric care.

First, we included all 99 Z-codes in categories Z55 to Z65 that were active during the study period. These Z-codes are grouped into 10 categories used in our analysis: Problems related to education and literacy (Z55.×), employment and unemployment (Z56.×), occupational exposure to risk factors (Z57.×), physical environment (Z58.×), housing and economic circumstances (Z59.×), social environment (Z60.×), upbringing (Z62.×), other problems related to primary support group including family circumstances (Z63.×), certain psychosocial circumstances (Z64.×), and other psychosocial circumstances (Z65.×).

Next, we reviewed a list of ICD-10 codes related to SDOH published by the AAP²⁴  and included all codes on this list unless they fit 1 of the following 3 exclusions: related to a medical diagnosis rather than SDOH (excluding “Alcohol abuse counseling and surveillance of alcoholic” and “Drug abuse counseling and surveillance of drug abuser”); not specific to SDOH (excluding “Sleep deprivation,” “Inadequate sleep hygiene,” and “Dietary counseling and surveillance”); or overly general (excluding “Stress, not elsewhere classified” and “Other specified personal risk factors, not elsewhere classified”). These exclusions were made to more directly reflect the World Health Organization definition of SDOH.¹  The resulting 13 AAP-identified SDOH codes were then mapped onto the existing Z-code categories (Supplemental Table 6). SDOH code use was defined as having at least 1 relevant ICD-10 code documented during the hospital encounter.

Demographic characteristics included age, race/ethnicity, sex, income quartile of ZIP code, and payer. Clinical characteristics included clinical service lines and the presence of a complex chronic condition (CCC). Clinical service lines are maintained by the Children’s Hospital Association (Lenexa, KS) and include 12 higher level groupings of 3M Healthcare’s All Patient Refined Diagnosis Related Groups of ICD-10 codes.²⁵  The pediatric CCC classification system was used to define the presence of a CCC.²⁶  Hospital characteristics included location and teaching status, ownership type, bed size, census region, and freestanding children’s hospital status.

We compared SDOH code use between 2016 and 2019 overall and by each Z-code category using χ² test and present the changes as odds ratios. We then compared overall SDOH code use between 2016 and 2019 by demographic, clinical, and hospital characteristics to identify populations with higher SDOH code documentation or significant change over time.

We evaluated the distribution of SDOH code use across hospitals to determine whether use was evenly distributed across hospitals or driven by a small number of “high documenting” hospitals. This analysis included data from hospitals with >10 pediatric discharges per year. Hospitals were subdivided by frequency of SDOH code utilization: 0%, 1% to 5%, 6% to 10%, and >10% of discharges with at least 1 SDOH code, and this distribution was compared between 2016 and 2019. We modeled the odds of a hospital having >5% of discharges with at least 1 SDOH code using multivariable logistic regression including year, hospital location and teaching status, hospital bed size, and census region.

All analyses were conducted using SAS v.9.4 (SAS Institute, Cary, NC), and P < .001 was considered statistically significant because of the large sample size. This study was determined to be nonhuman subjects research by the Children’s National Hospital Institutional Review Board because of the deidentified nature of the data.

The study population included >12 million discharges: 6.2 million discharges in 2016 and 5.9 million discharges in 2019. Overall SDOH code documentation remained low but increased slightly from 1.4% in 2016 to 1.9% in 2019 (P < .001). Table 1 compares SDOH code use for each of the 10 Z-code categories between 2016 and 2019 demonstrating a statistically significant increase in use for 7 of the 10 categories. The most frequently used Z-code category was “Problems related to upbringing” (Z62); this category also had the largest increase from 2016 to 2019, 0.8% to 1.1% (P < .001).

Adolescents had the highest rates of SDOH code documentation with 5.2% in 2016 and 7.1% in 2019, which also accounts for the highest relative increase in use from 2016 to 2019 (P < .001; Table 2). The highest rates of SDOH code use by race/ethnicity were among Native Americans (2.4% in 2016 and 3.8% in 2019; P < .001). Components representing economic standing including income quartile of ZIP code and payer showed even distribution of SDOH code use across all quartiles and payers, with statistically significant increases overall from 2016 to 2019 (P < .001).

Mental health-related admissions used SDOH codes significantly more often than all other service lines and had the largest absolute increase in use between years, 23.0% in 2016 and 26.1% in 2019 (P < .001; Table 3). All other service lines had ≤2% SDOH code use across years. There was no difference in SDOH code use for patients with and without CCCs.

When comparing SDOH code use in hospitals by location and teaching status, ownership type, bed size, and census region, all groups increased use of SDOH codes from 2016 to 2019 (P < .001; Table 4). The largest increase occurred in nonfederal government hospitals 1.8% to 3.0% with an odds ratio of 1.72 (95% CI [1.69–1.76]), whereas the remainder of the hospital cohorts had ≤2% SDOH code use in 2016 and 2019. Comparison of freestanding and nonfreestanding children’s hospitals also showed an increase in both hospital types from 2016 to 2019, with the larger increase over time in nonfreestanding children’s hospitals (1.2% to 1.8%, P < .001) but overall use was higher in freestanding children’s hospitals (2.6% to 3.0%, P < .001).

In an analysis of SDOH code use distribution across hospitals, SDOH codes were most often used in 1% to 5% of all pediatric discharges (Supplemental Table 7). There was a decrease in the overall number of hospitals who never used an SDOH code, from 38.5% in 2016 to 30.9% in 2019. In multivariate logistic regression analysis (Table 5), hospitals that had >5% of discharges with at least 1 SDOH code were more likely to be large (adjusted odds ratio [aOR], 1.59; 95% confidence interval [CI], 1.3–1.95), urban teaching hospitals (aOR, 1.87; 95% CI, 1.5–2.31), and located in the northeast (aOR, 2.2; 95% CI, 1.68–2.89).

This study evaluates the use of SDOH codes nationally for hospitalized pediatric patients after the coding policy change in 2018. We found that, although SDOH code use increased from 2016 to 2019 across most social needs domains, the overall documentation rate remained low (<2%) despite expansion in coding ability for all clinicians on the care team. SDOH codes were used most frequently to document “Problems related to upbringing,” a broad category inclusive of both social needs and ACEs (Supplemental Table 6). Overall code use was most commonly applied to adolescents, Native Americans, and patients with mental health diagnoses. Frequency of SDOH code documentation did not appear to differ between patients with complex chronic conditions and those without. Although overall SDOH code use remained low, the number of all hospitals using any SDOH codes increased by nearly 8% from 61.5% to 69.1% between 2016 and 2019.

Screening for SDOH and providing resources to address needs is increasingly practiced in the outpatient pediatric setting.^27–30  An SDOH process framework was initiated in 2013 for adolescent patients, calling for universal screening and resource prioritization,³¹  and again in 2016 for pediatric primary care.²⁸  Meanwhile, an emergency department randomized trial indicated that almost all families had at least 1 social need, and families were more likely to disclose needs via private electronic disclosure.³²  With the majority of inpatient admissions coming through the emergency department, this is potentially reflective of inpatient social needs and an argument for inpatient SDOH screening, documentation, and resource allocation.

Although not all institutions that screen for SDOH use Z-code documentation, our findings may support previous literature demonstrating that, though there is a presence of unmet social needs in the inpatient pediatric population, only a minority of pediatric inpatient care teams perform SDOH documentation.^{14,15,18,21,33–37}  It is possible that the stressful conditions of hospitalization itself can unmask social needs, making identification of these conditions an important part of inpatient care.³⁶  Screening for SDOH can be controversial if referral or access to resources is not readily available.^28,38  However, screening itself can provide valuable insight into population health needs³⁹  and a more holistic appreciation for the challenges individual patients/families are facing in terms of addressing their overall health. Furthermore, pediatric patient caregivers find the use of screening both acceptable and appropriate during inpatient hospitalization.⁴⁰ 

Despite this growing body of evidence around the importance of understanding SDOH across different health care settings, our study reinforced the ongoing limited use of SDOH codes even after the coding policy change in 2018. Several barriers limit the use of SDOH codes for inpatients. First, there is a strong possibility that overall awareness of SDOH ICD-10 codes themselves remains low, as indicated by a recent CMS interdisciplinary listening session on SDOH coding.⁴¹  Moreover, there may be a lack of awareness of the expansion of the term “clinician” to include documentation of SDOH from interdisciplinary care team members including nurses, social workers, and others, as clarified by the AHA Coding Clinic in 2018.¹⁷  Even in the event of knowledge of both the SDOH code’s existence and its expanded use, a coding team may not have the training or bandwidth to scan a full medical record in search of this documentation. Novel approaches to SDOH documentation such as using volunteers or multidisciplinary care team members to enter SDOH ICD-10 codes,⁴²  implementing structured screening with automatic SDOH ICD-10 coding,^43–45  or development of natural language processing data extraction tools⁴⁶  could be used to reduce the burden on providers.

Another barrier is that SDOH code claims are not typically part of the payment process, and institutions lack a financial incentive to code for SDOH but there is work toward changing this pattern. The 2021 Current Procedural Terminology Evaluation and Management outpatient- and office-visit coding guidelines now allow physicians to incorporate the presence of SDOH factors into their billing for level of service, specifically raising a visit to a moderate level of complexity if the physician can connect a specific SDOH factor to a clinical diagnosis and thus resulting in greater reimbursement.⁴⁷  The inpatient system is structured differently, but in the future could be modified to more easily recognize the role of SDOH in acute illness. Currently, some Medicaid Accountable Care Organizations (ACOs) are piloting equity-focused value-based care models that highlight SDOH documentation and resource allocation as a principle associated with Medicaid reimbursement, and some state ACOs have already implemented incentives toward equity-focused care, targeting SDOH.^20,48–50 

The ultimate goal of this work is to identify patients with social needs affecting their health and connect them with resources. We argue that to reach this goal, each step is important (screening, documentation, and the provision of resources) and that SDOH ICD-10 codes are a useful tool to accomplish the step of documentation. If the number of health systems screening for and documenting SDOH continues to rise, it is critical that referral systems and community resources increase to address identified needs. A recent systematic review of Medicaid ACOs found that, despite growing interest in interventions to address SDOH, such efforts have been scattered and published without key information on implementation, preventing the identification of best practices for widespread dissemination.⁵¹  To address this gap, CMS launched the Accountable Health Communities Model in 2016, a program with 28 participating sites testing whether systematically identifying and addressing SDOH through screening, referral, and community navigation services will affect health care costs and reduce health care use among the Medicare and Medicaid populations.^{19,20,52–54}  Best practices learned from this initiative may inform future inpatient pediatric programs to address SDOH.

In our analysis, 1 of the most striking findings was the increased frequency of documentation of SDOH codes among patients with mental health diagnoses. This is likely multifactorial, including the known relationship between unmet social needs and mental health diagnoses,^6,55  the increased emphasis on environment and psychosocial factors during psychiatric evaluation and diagnosis,^56,57  and the use of interdisciplinary teams including social work and case management in psychiatric care.^58,59  Future research should explore mental health providers as potential “high performers” for screening and documenting SDOH, elucidate which SDOH codes are most commonly applied to this group, and see whether this translates into improved resource referral. In addition, SDOH were coded more frequently for Native Americans. Although there is a deep history of systemic racism and health disparities for the Native American population,^60–62  this group is also developing innovative community-based participatory research to address SDOH and improve health disparities.⁶³  Future research should seek to learn from and address SDOH needs within the Native American community.^62,63 

Our study had several limitations. First, our dataset relied on health care team members to input SDOH codes, which may not reflect the number of patients who underwent screening and provision of resources for social needs. Related diagnoses may be documented in clinical notes, yet not be identified by billing codes,¹⁵  and there is variability in hospital coding practices that could not be broken down in the dataset. For example, nonphysicians are allowed to enter these codes, but the extent to which they are doing so is unknown. Second, we did not identify a baseline rate of pediatric social needs by hospital to establish a known social needs prevalence, although large scale estimates for pediatric ACEs and pediatric social needs exist.^64,65  Furthermore, previous studies have demonstrated a significant underuse of SDOH codes when compared with known rates of social needs.^14,15,18,21 

Future research is warranted to further investigate the patterns of SDOH code use revealed in this study, such as differences based on age, race, ethnicity, gender, clinical characteristics, and among different SDOH domains. With the next KID data release set for next year, trending overall Z-code use should be considered. Finally, future efforts could expand on the success of current high-performing sites using structured SDOH screening, which when entered into the electronic medical record, results in automatic SDOH ICD-10 coding.^43–45 

Despite the 2018 policy change allowing for a multidisciplinary approach to SDOH documentation, ICD-10 codes remain an underused tool to track SDOH within the inpatient pediatric setting. SDOH code use is more prevalent among patients presenting with mental health concerns, suggesting that this may be an area to look to for strategies to improve SDOH code documentation. Future research should explore whether SDOH code documentation is associated with increased response to unmet social needs and, if so, how to improve the use of SDOH codes in the inpatient setting.

AAP

American Academy of Pediatrics

ACE

adverse childhood experiences

AHA

American Hospital Association

aOR

adjusted odds ratio

CCC

complex chronic condition

CI

confidence interval

CMS

Center for Medicare and Medicaid Services

ICD-10

International Classification of Diseases, 10th Revision

KID

Kid’s Inpatient Database

SDOH

Social Determinants of Health