2.1 Electronic Health Records (EHR) & Health Informatics Foundations

Electronic Health Records (EHR) & Health Informatics Foundations

Health informatics sits at the intersection of information science, computer science, healthcare delivery, and data analytics. For the Certified Health Data Analyst (CHDA), mastering how clinical data is captured, structured, stored, and retrieved within Electronic Health Record (EHR) systems is essential. A robust analytics solution depends entirely on understanding the underlying architectural constraints, documentation workflows, and data models of source systems.

---

The Health Informatics Hierarchy: DIKW Framework

The Data-Information-Knowledge-Wisdom (DIKW) hierarchy provides the conceptual foundation for healthcare data analytics, illustrating how raw, uninterpreted elements progress into informed clinical and organizational action.

                    ┌──────────────┐
                    │    WISDOM    │  Applied Clinical Judgment & Ethics
                    ├──────────────┤
                    │  KNOWLEDGE   │  Synthesized Rules & Clinical Guidelines
                    ├──────────────┤
                    │ INFORMATION  │  Structured Data with Clinical Context
                    ├──────────────┤
                    │     DATA     │  Raw Discrete Values & Symbols
                    └──────────────┘

1. Data (Raw Facts & Symbols)

Data represents discrete, unorganized, and uninterpreted facts, numbers, symbols, or observations without contextual meaning.

• Clinical Example: A numeric value of 142, a string 82, or a code E11.9 stored in a database table without timestamps, units of measure, or patient linkage.
• Analytic Consideration: Raw data cannot inform clinical decision-making until it is paired with appropriate metadata (e.g., physiological parameter, measurement unit, reference range).

2. Information (Data in Context)

Information is data that has been processed, structured, organized, or contextualized to answer fundamental questions: Who? What? Where? When?

• Clinical Example: Associating 142 with systolic blood pressure in mmHg, measured at 08:30 AM on patient ID #98432 in the outpatient cardiology clinic, indicating a reading above normal adult thresholds.
• Analytic Consideration: Information enables descriptive analytics, such as identifying a cohort of patients with elevated blood pressure readings across a health system.

3. Knowledge (Synthesized Understanding & Patterns)

Knowledge emerges when information is synthesized, integrated with clinical domain expertise, compared against established medical guidelines, and evaluated for longitudinal patterns.

• Clinical Example: Recognizing that patient ID #98432 has maintained resting blood pressure readings exceeding 140/90 mmHg across three consecutive visits despite first-line thiazide diuretic therapy, fulfilling the clinical diagnostic criteria for Stage 2 Hypertension with inadequate control per AHA/ACC guidelines.
• Analytic Consideration: Knowledge is encoded into Clinical Decision Support (CDS) rule engines, clinical pathways, and predictive models.

4. Wisdom (Actionable Judgment & Ethical Application)

Wisdom represents the highest tier of the hierarchy: the pragmatic, ethical, and personalized application of knowledge to make optimal clinical decisions considering patient preferences, comorbidities, social determinants of health, and potential risks.

• Clinical Example: A physician, recognizing the patient's Stage 2 hypertension, chronic kidney disease (CKD Stage 3a), and limited financial means, prescribes an affordable generic ACE inhibitor (lisinopril) while scheduling renal function monitoring (serum creatinine and potassium) at 4 weeks, rather than initiating a costly brand-name combination agent.
• Analytic Consideration: Wisdom drives prescriptive analytics, shared decision-making tools, and holistic health equity interventions.

---

EHR System Architectures & Database Engines

Healthcare data analysts must navigate the technical architectures and database engines that power modern EHR systems, as these dictate extraction performance, data freshness, and analytical query design.

System Deployment Models

1. On-Premise / Client-Server Architecture:
   • Core database and application servers are housed in the healthcare organization's physical data center.
   • Characteristics: High initial Capital Expenditure (CapEx); internal IT team manages hardware maintenance, OS patching, backups, and disaster recovery; provides high local network speed and direct control over database access.
2. Cloud-Hosted / Software-as-a-Service (SaaS):
   • Infrastructure and applications are hosted in dedicated or multi-tenant public/private cloud environments (e.g., AWS, Microsoft Azure, Google Cloud).
   • Characteristics: Operational Expenditure (OpEx) subscription model; vendor manages infrastructure scaling, high availability, uptime SLAs, and routine software upgrades; facilitates rapid deployment and multi-facility data centralization.

Transactional (OLTP) vs. Analytical (OLAP) Database Engines

EHR systems are optimized for Online Transaction Processing (OLTP)—rapid, concurrent reads and writes of individual patient records with sub-second response times during clinical encounters. However, executing large-scale population analytics directly against OLTP databases degrades clinical performance. Consequently, healthcare organizations utilize dedicated Online Analytical Processing (OLAP) or dimensional data warehouses populated via Extract, Transform, Load (ETL) pipelines.

CHDA Exam Tip: Remember that Epic's transactional database (Chronicles) uses hierarchical MUMPS/IRIS globals and cannot be queried with standard SQL. Analysts query Clarity (relational schema) or Caboodle (dimensional star schema), which are populated via ETL processes.

---

Structured vs. Unstructured Healthcare Data

Clinical documentation generates a hybrid of structured discrete elements and unstructured narrative text. Effective data analysts must recognize the strengths, limitations, and extraction methods for each data type.

                               HEALTHCARE DATA SPECTRUM
 ┌──────────────────────────────┬─────────────────────────────┬─────────────────────────────┐
 │       STRUCTURED DATA        │    SEMI-STRUCTURED DATA     │      UNSTRUCTURED DATA      │
 ├──────────────────────────────┼─────────────────────────────┼─────────────────────────────┤
 │ • Discrete numeric values    │ • HL7 v2 messages           │ • Free-text progress notes  │
 │ • Standardized code sets     │ • C-CDA XML documents       │ • Narrative H&P notes       │
 │ • Dropdown / radio options   │ • FHIR JSON resources       │ • Pathology reports         │
 │ • Vital signs & lab panels   │ • Tagged diagnostic reports │ • Diagnostic imaging text   │
 └──────────────────────────────┴─────────────────────────────┴─────────────────────────────┘

Structured (Discrete) Data

Structured data resides in fixed, predefined fields with explicit data types (integer, float, date, boolean) or standardized clinical code sets.

• Standardized Terminologies: ICD-10-CM/PCS (diagnoses/procedures), CPT/HCPCS (outpatient procedures/services), SNOMED CT (clinical findings/anatomy), LOINC (laboratory/observational measurements), RxNorm (clinical medications).
• Advantages: Directly accessible via standard SQL queries (SELECT, WHERE, GROUP BY); easily aggregated; enables deterministic automated quality metric computation (e.g., CMS eCQMs); supports automated threshold alerting.
• Disadvantages: Rigid data entry templates can slow clinical workflow, lead to "click fatigue," and fail to capture complex, non-standard clinical scenarios.

Unstructured (Narrative) Data

Unstructured data comprises free-text clinical narratives, physician progress notes, consultation notes, operative reports, discharge summaries, and radiology narrative impressions. Estimates indicate that 70% to 80% of clinically valuable information resides in unstructured text.

• Advantages: Allows clinicians to document detailed diagnostic reasoning, subtle physical exam nuances, chronological illness progression, and subjective patient concerns without arbitrary template constraints.
• Disadvantages: Cannot be queried directly using standard relational SQL operators; contains clinical jargon, non-standard abbreviations, typos, and extensive copy-paste note bloat; requires Natural Language Processing (NLP) or Large Language Models (LLMs) to extract discrete facts.

Semi-Structured Data

Semi-structured data contains organizational markers or tags that separate semantic elements but do not conform to rigid relational database schemas. Common examples include HL7 v2 pipe-delimited messages, C-CDA (Consolidated Clinical Document Architecture) XML files, and HL7 FHIR (Fast Healthcare Interoperability Resources) JSON objects.

---

Clinical Workflows, CPOE, and CDS Alert Fatigue

Clinical Documentation Workflows

Clinical workflows follow a standardized sequence: patient registration/scheduling → triage and vital signs capture → physician history and physical examination (H&P) → computerized ordering → nursing administration → diagnostic interpretation → discharge summary and clinical coding.

Computerized Provider Order Entry (CPOE)

CPOE is the direct electronic entry of medical orders—including medications, laboratory tests, diagnostic imaging, physical therapy, and blood products—by licensed healthcare practitioners into the EHR.

• Safety Benefits: Eliminates handwriting illegibility errors; standardizes dosing units; accelerates order transit to pharmacy, laboratory, and radiology departments; automatically checks for formulary compliance.
• Analytic Metrics: CPOE adoption rates, turnaround time from order entry to medication administration, and order set utilization percentages.

Clinical Decision Support (CDS) Systems

CDS provides clinicians, staff, patients, or other individuals with knowledge and person-specific information, intelligently filtered or presented at appropriate times, to enhance health and healthcare delivery.

• Interruptive Alerts (Hard & Soft Stops): Modal pop-up dialogs that halt the clinician's workflow, requiring active acknowledgement, justification, or cancellation (e.g., severe drug-drug interaction alerts, anaphylactic drug-allergy alerts).
• Non-Interruptive Reminders (Passive CDS): Side-panel banners, infobuttons, recommended order sets, and subtle visual flags that inform without freezing the user interface.

CDS Alert Fatigue: Etiology and Remediation

Alert fatigue is the state of cognitive overload and desensitization experienced by clinicians when exposed to a high volume of frequent, clinically insignificant, or false-positive electronic alerts. When alert fatigue sets in, clinicians rapidly override, ignore, or click past critical safety warnings without reading them.

  High Alert Volume (Low Specificity) ──> Cognitive Overload ──> Desensitization ──> 80-90%+ Override Rate ──> Patient Safety Risk

The "Five Rights" of Clinical Decision Support (CDS)

To minimize alert fatigue and maximize clinical efficacy, CDS interventions must adhere to the CDS Five Rights Framework:

1. The Right Information: Evidence-based, clinically valid, actionable guidance.
2. To the Right Person: The specific clinician, nurse, pharmacist, or patient empowered to act on the recommendation.
3. In the Right Format: Appropriate alert style (e.g., passive notification vs. interruptive hard stop).
4. Through the Right Channel: Integrated directly into the clinical EHR interface, mobile clinical device, or patient portal.
5. At the Right Point in the Workflow: Delivered at the precise moment of clinical decision-making (e.g., during order entry or medication reconciliation, not after discharge).

---

Secondary Uses of Healthcare Data

Primary data use refers to direct patient care delivery. Secondary data use (or data reuse) involves analyzing aggregated, de-identified, or identifiable clinical data for non-direct care purposes.

                               ┌─────────────────────────────────────────┐
                               │    PRIMARY DATA USE: Direct Patient     │
                               │      Care, Bedside Clinical Workflow    │
                               └────────────────────┬────────────────────┘
                                                    │
                       ┌────────────────────────────┴────────────────────────────┐
                       ▼                                                         ▼
      ┌─────────────────────────────────┐                       ┌─────────────────────────────────┐
      │     CLINICAL & RESEARCH USES    │                       │    OPERATIONAL & FINANCIAL      │
      ├─────────────────────────────────┤                       ├─────────────────────────────────┤
      │ • Observational Cohort Studies  │                       │ • Capacity & Bed Management     │
      │ • Clinical Trial Feasibility    │                       │ • Staffing & Scheduling Models  │
      │ • Quality Measurement (eCQMs)   │                       │ • Case Mix Index (CMI) Analysis │
      │ • Population Health Outreach    │                       │ • Claims Billing & CDI Auditing │
      └─────────────────────────────────┘                       └─────────────────────────────────┘

1. Clinical Research & Epidemiology: Retrospective observational cohort studies, randomized clinical trial recruitment feasibility, post-market drug surveillance, and real-world evidence (RWE) generation.
2. Healthcare Operations & Capacity Management: Operating room throughput analysis, Emergency Department (ED) length-of-stay tracking, nurse staffing ratios, and inpatient bed occupancy forecasting.
3. Quality Measurement & Regulatory Reporting: Computing CMS Hospital Inpatient Quality Reporting (IQR) metrics, electronic Clinical Quality Measures (eCQMs), Merit-based Incentive Payment System (MIPS) scores, and The Joint Commission ORYX benchmarks.
4. Population Health Management: Identifying chronic disease cohorts (e.g., uncontrolled diabetics), calculating risk scores, tracking preventive cancer screening completion, and closing clinical care gaps.
5. Financial & Revenue Cycle Analytics: Monitoring Case Mix Index (CMI), identifying clinical documentation improvement (CDI) opportunities, analyzing claim denial patterns, and optimizing fee-for-service and value-based contract performance.