ABPath Clinical Informatics cheat sheet

Clinical informatics applies information science, computer science, and clinical knowledge to the acquisition, storage, retrieval, and use of health information.

The data-information-knowledge-wisdom hierarchy progresses from raw data to organized information, synthesized knowledge, and applied judgment.

An EHR spans multiple organizations longitudinally; an EMR is managed within a single organization's encounters.

Interoperability levels are foundational transport, structural format, semantic meaning, and organizational governance.

HL7 version 2 uses message segments and pipe-delimited fields; FHIR uses resources and RESTful APIs; DICOM handles imaging objects.

LOINC codes laboratory observations; SNOMED CT codes clinical concepts; ICD-10-CM codes diagnoses; CPT codes procedures; UCUM codes units.

The sociotechnical model examines people, processes, technology, organization, and external environment as interacting components.

Usability is measured by effectiveness, efficiency, and satisfaction; high cognitive load and alert burden reduce safe use.

Health IT lifecycle phases include needs assessment, selection, implementation, optimization, and decommissioning.

The five rights of CDS are right information, right person, right format, right channel, and right time.

CDS interventions include alerts, order sets, templates, calculators, reference links, and workflow-embedded guidance.

Alert specificity and prevalence determine positive predictive value; low prevalence produces many false-positive alerts even at high sensitivity.

Positive predictive value equals true positives divided by true positives plus false positives; negative predictive value equals true negatives divided by true negatives plus false negatives.

Knowledge management requires provenance, version control, review dates, ownership, and retirement criteria for guidelines and rules.

CDS governance defines approval, testing, monitoring, and change control for rules that affect clinical decisions.

Interruptive alerts are reserved for high-severity, time-sensitive risks; noninterruptive displays are used for lower-urgency information.

Rule performance monitoring includes trigger volume, acceptance rate, override reason, patient outcomes, and unintended consequences.

Drug-drug interaction severity tiers commonly range from contraindicated to major, moderate, and minor, with display rules tied to tier.

The system development life cycle includes requirements, design, build, test, deploy, and maintain phases, with change control across phases.

Selection artifacts include requirements documents, RFI, RFP, scripted demonstrations, reference checks, and contract service-level agreements.

Validation for regulated systems includes installation qualification, operational qualification, and performance qualification.

User acceptance testing verifies that representative end users can complete required workflows against agreed requirements.

Big-bang cutover changes all users at once; phased cutover limits scope but adds interface and dual-process complexity.

Downtime procedures define paper workarounds, data capture, reconciliation, and communication during system unavailability.

Project controls use scope, schedule, budget, risk register, issue log, and RACI roles for decision rights.

Optimization uses usage logs, user feedback, workflow observation, and outcome metrics to prioritize changes through change control.

Interface testing validates message structure, code mapping, trigger events, error handling, and reconciliation between systems.

Data governance assigns ownership, stewardship, definitions, access rules, and lifecycle controls for data assets.

Data quality dimensions include completeness, accuracy, timeliness, consistency, validity, and uniqueness.

A data dictionary defines field names, data types, allowed values, provenance, and business rules.

Analytics levels are descriptive, diagnostic, predictive, and prescriptive.

Precision equals true positives divided by retrieved positives; recall equals true positives divided by all relevant items; F1 is the harmonic mean of precision and recall.

Cohort definitions require explicit inclusion, exclusion, time windows, and code sets; phenotype validation compares computed cohorts with chart review.

Dashboards use defined numerators, denominators, stratification, and display rules; statistical process control distinguishes signal from noise.

HIPAA Safe Harbor removes eighteen identifier categories; expert determination requires a qualified statistical or scientific analysis.

Data lifecycle controls cover acquisition, validation, storage, access, retention, archival, and destruction.

HIPAA regulates protected health information held by covered entities and business associates; permitted uses include treatment, payment, and health care operations.

The security triad is confidentiality, integrity, and availability; safeguards are administrative, physical, and technical.

Access control principles include least privilege, role-based permissions, unique user identity, multi-factor authentication, and audit logging.

A HIPAA risk analysis identifies threats, vulnerabilities, likelihood, impact, and controls for electronic protected health information.

Breach notification to individuals is due without unreasonable delay and no later than sixty days after discovery.

Breaches affecting five hundred or more individuals require notification to the Secretary of Health and Human Services and media where applicable.

Audit trails record access, modification, and disclosure events; HIPAA documentation retention is six years.

Incident response phases are preparation, detection, containment, eradication, recovery, and lessons learned.

CLIA governs clinical laboratory testing; CAP provides accreditation; FDA regulates medical devices and software functions meeting device definitions.

The Institute of Medicine six aims are safe, effective, patient-centered, timely, efficient, and equitable care.

The Donabedian model links structure and process to outcomes.

PDSA tests small changes through plan, do, study, and act cycles; DMAIC defines, measures, analyzes, improves, and controls.

Control charts separate common-cause variation from special-cause variation using center lines and control limits.

FMEA scores severity, occurrence, and detection; the risk priority number equals severity multiplied by occurrence multiplied by detection.

Root cause analysis examines system factors, human factors, and process conditions contributing to a sentinel event.

Just culture distinguishes human error, at-risk behavior, and reckless behavior, with responses matched to conduct.

Laboratory quality indicators include turnaround time, specimen rejection rate, corrected reports, critical value notification, and proficiency testing performance.

Run charts display data over time with a median line; nonrandom patterns such as shifts or trends indicate process change.

Kotter's change model includes urgency, coalition, vision, communication, barrier removal, short-term wins, gains consolidation, and anchoring.

ADKAR defines individual change as awareness, desire, knowledge, ability, and reinforcement.

Lewin's model frames change as unfreezing, moving, and refreezing.

A power-interest grid classifies stakeholders into manage closely, keep satisfied, keep informed, and monitor.

A RACI matrix assigns responsible, accountable, consulted, and informed roles for tasks and decisions.

Governance structures include a steering committee, executive sponsor, project management office, and clinical advisory groups.

Business cases compare costs, benefits, risks, and strategic alignment; return on investment equals net benefit divided by cost.

Change readiness depends on sponsorship, communication, training, workflow fit, incentives, and perceived usefulness.

Project escalation paths define thresholds and decision owners for scope, budget, risk, and clinical safety issues.

The laboratory information system supports order entry, accessioning, specimen tracking, result entry, verification, reporting, quality control, and interfaces.

Specimen labeling uses at least two patient identifiers and barcode scanning at collection and accessioning to prevent misidentification.

Autoverification rules release results meeting validated criteria; delta checks and critical value rules require review or notification.

LOINC maps laboratory tests and observations; UCUM specifies units; SNOMED CT supports pathology and clinical findings.

HL7 ORM messages support orders, ORU messages report results, and DICOM supports images and related metadata.

Westgard rules detect random and systematic error in quality control; Levey-Jennings charts display control results over time.

CLIA requires proficiency testing, quality control, personnel competency, and validation of laboratory methods.

Critical laboratory results require timely notification documented with recipient, time, result, and read-back.

Synoptic reporting uses structured pathology checklists with standardized elements and allowable values for diagnosis, staging, and analytics.