Overview of the Data

Pediatric research is an important area of focus for the Regenstrief Institute. Regenstrief receives data from multiple sources, including health systems contributing to the Indiana Network for Patient Care (INPC) as well as state-level data sources such as Indiana birth and death records. Pediatric data within the INPC are particularly valuable because they include information from major academic health systems in Indiana, including Indiana University Health and Eskenazi Health, while also capturing data from community hospitals and clinics across the state.

The INPC can support a broad range of pediatric research studies, including investigations of healthcare utilization, vaccine effectiveness, chronic disease prevalence, health-related social needs, and social determinants of health. Pediatric data in the INPC are strongest for diagnostic, laboratory, and demographic information. In some cases, additional value comes from the ability to link maternal and child records, as well as the inclusion of statewide birth and death data.

Data Quality and Availability:

The INPC does not require participating health systems to contribute specific data elements. Instead, organizations contribute data voluntarily, and the type and completeness of data submitted can vary substantially between systems. Major academic health systems such as Indiana University Health and Eskenazi Health tend to contribute relatively comprehensive datasets, while smaller or more distant health systems may contribute fewer data elements. In this way, INPC data is broad but often shallower than individual hospital data warehouses. As such, it can serve as an important complement to these data, allowing for incorporation of patients from small community and rural care settings. Some specific data elements, as well as strengths and limitations of the INPC pediatrics data, are discussed below.

Vaccination data

Vaccine data are generally quite good within Indiana because of the statewide Children and Hoosiers Immunization Registry Program (CHIRP). Many health systems, including Epic-based systems, maintain bidirectional interfaces with CHIRP, allowing vaccination information to flow between clinical systems and the state registry. Notably, use of vaccine registry data in INPC often requires additional approvals. Also, vaccine data are not always perfectly synchronized. Delays or mismatches in patient identifiers may occur, particularly when vaccines are administered at outside sites. In some situations, vaccine information may only appear after a subsequent healthcare encounter triggers synchronization. Despite this, INPC vaccine information can still provide substantial detail regarding vaccine types, administration timing, and the clinical context in which vaccines were administered, such as well-child visits versus sick visits. The INPC has also contributed to CDC vaccine effectiveness studies, including the VISION network, where it serves as a major contributor of non-academic patient populations.

Vital signs and growth parameters

Vital signs and anthropometric measurements such as height, weight, and body mass index (BMI) are available within the INPC but require careful validation. Pediatric growth parameters can occasionally contain implausible values, such as negative BMI values or biologically impossible measurements. Investigators are likely to benefit from obtaining some values prior to the study date or range of interest, as all growth parameters may not be obtained at each visit. Investigators may also wish to think about how they would want to handle repeat measurements of growth parameters, similar to vital signs. Sometimes, only one measure may be obtained; in other cases, there may be multiple recordings. Observations with abnormal initial measurements may be more likely to trigger repeat measurements, meaning that a carefully considered approach is needed. Not all contributors to INPC data provide vital sign information, although efforts are ongoing to backfill this information where possible.

Social determinants of health and health-related social needs

Social determinants of health, measured at the community level and health-related social needs measured at the individual level are important in pediatric populations. Childhood environments strongly influence long-term physical health, mental health, educational attainment, and healthcare utilization across the lifespan. The INPC provides growing opportunities to study these factors within pediatric populations through geocoding capabilities and linkage to community-level measures such as the Social Vulnerability Index and the Child Opportunity Index through collaboration with the Polis Center. Investigators noted that these tools can support analyses of neighborhood disadvantage and structural contributors to pediatric outcomes across Indiana. Because the INPC captures healthcare utilization from both academic medical centers and community settings throughout the state, it offers an important opportunity to study social determinants of health beyond tertiary pediatric referral populations. In addition, maternal-child linkage capabilities and statewide birth data may allow researchers to examine how social and environmental factors influence outcomes beginning in the neonatal period and continuing through childhood. However, investigators should recognize that outpatient pediatric care capture remains incomplete in some regions and health systems, which may affect interpretation of longitudinal care patterns and healthcare access.

Strengths of the INPC data

One of the major strengths of the INPC for pediatric research is its breadth across healthcare settings and geographic regions. Because the INPC aggregates data from major academic health systems as well as community hospitals and clinics throughout Indiana, it allows investigators to study pediatric populations beyond tertiary referral centers. This can be particularly valuable for examining healthcare utilization, rural health disparities, and statewide patterns of disease and care delivery. The inclusion of data from systems such as Indiana University Health and Eskenazi Health provides strong representation of urban pediatric populations, while the broader statewide network captures care delivered in smaller and rural communities. Diagnostic coding data are generally comprehensive because they are closely tied to billing processes, and laboratory data tend to be relatively robust across institutions. Inpatient and emergency department utilization data are also considered strengths of the INPC and are often more complete than outpatient pediatric data.

Another important strength of the INPC is the availability of longitudinal and linked data resources relevant to pediatric populations. Maternal-child linkage capabilities, along with access to Indiana birth and death records, create opportunities to study neonatal outcomes, maternal-child dyads, and long-term pediatric trajectories. The availability of clinical notes within portions of the INPC also creates opportunities for natural language processing approaches using tools such as Regenstrief’s nDepth platform and large language models by working closely with Regenstrief Data Services (RDS) in a secure computing environment. In addition, ongoing improvements in geocoding and continuity of care documentation may further strengthen the ability to study social determinants of health, transitions of care, and longitudinal healthcare utilization. Investigators also noted that combining the broad reach of the INPC with deeper institutional datasets can create particularly powerful pediatric research cohorts.

Limitations of the data

One of the major limitations of pediatric INPC data is incomplete outpatient data capture. Much pediatric healthcare occurs in outpatient settings, including preventive visits, developmental assessments, and chronic disease management, yet outpatient data submission varies substantially across participating health systems. This can create challenges for studies requiring evidence of longitudinal follow-up or continuity of care, such as analyses of preventive healthcare utilization. In some cases, it may be difficult to distinguish children who truly did not receive care from those who obtained care outside the INPC network or from organizations that contribute limited outpatient data. Investigators therefore often need to carefully evaluate the completeness of outpatient encounters for the populations and institutions included in a study.

Additional limitations relate to the variability and incompleteness of certain data elements. Medication data may be difficult to interpret because not all health systems contribute medication information consistently, and many commonly used pediatric medications are available over the counter and therefore may not appear in medication lists. Detailed clinical variables such as respiratory support, procedure data, and some vital signs may also be inconsistently available across institutions. Neonatal and maternal-child linkage studies can require additional validation because newborn names frequently change after birth hospitalization, creating challenges for automated linkage. As with all EHR-based research, investigators should anticipate some degree of missingness, inconsistency, and variability across institutions and should work closely with Regenstrief analysts to develop appropriate phenotypes and validation strategies.