Contraindication Checker

The core reasoning engine must distinguish between absolute contraindications, where the risk of harm unequivocally outweighs any benefit and the intervention must be blocked, and relative contraindications, where caution is advised but the therapy may proceed under specific clinical circumstances. This requires a structured knowledge base where each drug-condition pair is tagged with a severity modifier. For example, isotretinoin is an absolute contraindication in pregnancy due to teratogenicity, while metformin is a relative contraindication in moderate renal impairment, requiring a dose adjustment rather than a hard stop.

The checker must ingest and normalize structured patient data from the EHR to parameterize its rules. Key data points include:

• Active Problem List: ICD-10-CM coded diagnoses mapped to contraindication rules.
• Allergy/Intolerance List: RxNorm-coded substances with reaction types, distinguishing true IgE-mediated allergies from intolerances.
• Pregnancy Status: A binary flag with estimated gestational age, critical for FDA pregnancy category logic.
• Laboratory Results: Quantitative values like estimated glomerular filtration rate (eGFR) for renal dosing checks and platelet counts for anticoagulant safety.

The checker operates synchronously within the Computerized Physician Order Entry (CPOE) workflow. When a clinician signs an order, the system invokes a stateless decision service via a FHIR CDS Hooks order-select or order-sign call. The service must return a response within sub-second latency to avoid disrupting clinical workflow. The response includes a structured CDS Hooks Card containing a decision (e.g., stop for absolute contraindications) and a human-readable summary of the triggering logic for clinician review.

The rule set must be sourced from authoritative, evidence-based compendia. Common sources include First Databank (FDB), Multum, and Lexicomp, which provide structured drug-disease interaction tables. Each rule must carry metadata tracking its provenance, including the source, last update timestamp, and the level of evidence (e.g., randomized controlled trial, case report). This allows the institution's Pharmacy and Therapeutics (P&T) Committee to audit and locally override rules, suppressing low-evidence alerts to reduce alert fatigue.

A high-sensitivity, low-specificity checker generates excessive interruptive alerts, leading clinicians to habitually override even critical warnings. Advanced checkers employ tiered severity signaling:

• Hard Stops: Block the order entirely for absolute contraindications.
• Soft Alerts: Display a non-interruptive warning for relative contraindications, allowing the clinician to proceed with a documented override reason.
• Silent Filtering: Suppress alerts for known inconsequential interactions (e.g., a documented allergy to a specific brand when the active ingredient is not present) using context-aware suppression logic.

Beyond drug-disease interactions, a comprehensive checker must evaluate multiple interaction vectors simultaneously:

• Drug-Drug: Pharmacokinetic interactions via CYP450 enzyme competition.
• Drug-Allergy: Cross-reactivity warnings for drug classes (e.g., penicillin and cephalosporins).
• Drug-Food: Critical interactions like tyramine-rich foods with MAOIs.
• Drug-Laboratory: Interference with test results, such as biotin skewing troponin assays.
• Duplicate Therapy: Preventing cumulative toxicity from two agents in the same therapeutic class.

A real-time safety notification triggered when a newly prescribed medication has a known adverse reaction potential with an existing active medication in a patient's profile. Unlike absolute contraindications, which represent binary stop functions, drug-drug interactions often involve relative risk that can be managed through dose adjustment or increased monitoring. Modern checkers use cytochrome P450 pathway analysis to predict metabolic competition between agents.

A specialized module that evaluates a prescribed medication against a patient's documented allergies, including immunologic cross-reactivity between structurally similar compounds. For example, a documented penicillin allergy triggers evaluation of cephalosporin cross-reactivity risk based on R1 side-chain similarity. Advanced systems incorporate graded challenge protocols and distinguish between true IgE-mediated hypersensitivity and non-allergic adverse effects to reduce unnecessary avoidance.

A clinical safety component that cross-references medications against FDA Pregnancy Categories and Lactation Risk Categories (L1-L5). This module integrates with the patient's gestational age and trimester-specific risk windows, as teratogenic vulnerability varies significantly across developmental stages. Modern implementations incorporate the Teratogen Information System (TERIS) and LactMed databases for evidence-based risk quantification rather than relying solely on categorical labels.

An automated process that verifies a prescribed medication against a health plan's approved drug list to ensure coverage and cost-effectiveness. While distinct from safety-oriented contraindication checking, formulary status often correlates with therapeutic appropriateness and evidence-based guidelines. Integration between formulary and contraindication modules enables therapeutic substitution suggestions that respect both safety constraints and payer requirements.

A clinical decision support function that validates a prescribed dose against established minimum and maximum safety limits based on patient-specific factors. Critical parameters include creatinine clearance for renally-excreted drugs, body surface area for chemotherapeutic agents, and gestational age for neonatal dosing. This module transforms a contraindication from a binary gate into a continuous safety envelope that accounts for therapeutic index and organ function.

A safety alert triggered when a new medication order is placed for a drug in the same therapeutic class as an existing active order. This prevents unintentional overdose from concurrent administration of agents with identical mechanisms of action, such as two NSAIDs or two SSRIs. Advanced implementations use ATC classification codes and VA class hierarchies to detect therapeutic duplication even when brand and generic names differ.