FHIR SDOH Observation

Encodes a specific social risk finding—such as a food insecurity score or housing instability assessment—as a discrete, computable resource. Each observation captures the screening instrument used (e.g., PRAPARE, AHC-HRSN), the coded question, and the patient's answer using LOINC and SNOMED CT terminology bindings. This transforms narrative social history into structured data that can trigger clinical decision support, populate quality measures, and drive closed-loop referrals.

Leverages consensus-driven value sets developed by the Gravity Project to standardize SDOH data elements. Key bindings include:

• LOINC codes for screening questions (e.g., 88122-7 for food insecurity panel)
• SNOMED CT codes for findings (e.g., 733423003 for food insecurity)
• ICD-10-CM Z-codes for diagnosis mapping (e.g., Z59.4 for lack of adequate food) This alignment ensures semantic interoperability across the healthcare ecosystem and supports USCDI compliance.

Uses a parent-child component architecture to represent multi-question screening instruments within a single resource. The parent observation identifies the screening panel (e.g., 'PRAPARE Assessment'), while each component element captures an individual question-response pair. This structure preserves the contextual relationship between questions, supports scoring algorithms, and enables granular querying of specific social risk domains without losing the integrity of the original assessment.

The SDOHCC Observation Screening Response profile constrains the base FHIR Observation resource with mandatory elements specific to social risk screening:

• Requires Observation.code to identify the screening question using LOINC
• Mandates Observation.subject to reference the patient
• Enforces Observation.value[x] to capture the coded or numeric answer
• Supports Observation.derivedFrom to link findings back to the original questionnaire response These constraints ensure consistent implementation across vendors and use cases.

Employs standardized category codes and tagging mechanisms to classify observations by social risk domain. The Observation.category field uses the SDOH domain value set to tag findings as food-insecurity, housing-instability, transportation-insecurity, or other domains. This enables:

• Population-level aggregation for health equity analytics
• Risk stratification by specific social need categories
• Quality measure calculation for value-based care contracts
• Automated routing to domain-specific community resource platforms

Captures critical metadata about the screening event through extension elements and provenance resources. Key contextual data includes:

• Screening time and location (when and where the assessment occurred)
• Performer identity (who administered the screening)
• Patient consent status for data sharing with community organizations
• Conditional security labels for sensitive social data This provenance trail supports audit requirements, consent management, and longitudinal tracking of social risk over time.

An automated sequence of natural language processing components designed to extract, classify, and structure social risk factors from unstructured clinical narratives. The pipeline's output is the raw material for a FHIR SDOH Observation. Key stages include:

• Named Entity Recognition for SDOH: Identifying spans of text like 'homeless' or 'unemployed'.
• Negation Detection for SDOH: Distinguishing 'patient is homeless' from 'patient denies homelessness'.
• Experiencer Detection: Determining if the risk applies to the patient or a family member.
• Temporality Classification: Classifying the risk as current, historical, or future.

An automated workflow that tracks a patient's journey from a positive social risk screening through to a confirmed connection with a community-based service provider. A FHIR SDOH Observation with a positive finding acts as the trigger event. The workflow then creates a FHIR ServiceRequest and Task to manage the referral, and ultimately updates the Observation or creates a new one to document the intervention outcome, completing the closed loop.

The application of predictive models to segment a patient population by their level of social risk. A FHIR SDOH Observation provides the standardized, structured input features for these models. By aggregating multiple Observations (e.g., housing status, food security score, transportation access) alongside clinical data, a risk stratification algorithm can assign a composite Social Vulnerability Index or Area Deprivation Index score to each patient, enabling targeted resource allocation by population health teams.