1Overview of Decision Support

Medication Decision Support Systems (CDSS) are computerized systems that help healthcare providers make safer medication decisions by automatically screening for potential safety issues before medications are prescribed or dispensed.

Critical Use Cases:

• Preventing adverse drug events and medication errors
• Alerting providers to drug-drug interactions
• Screening for drug allergies and sensitivities
• Detecting contraindications based on patient conditions
• Identifying duplicate therapy and therapeutic overlap
• Dose range checking and renal/hepatic adjustments
• Age-specific and pregnancy/lactation warnings

Clinical Impact

Studies show that CDSS can prevent up to 86% of potential adverse drug eventswhen properly implemented and used. However, alert fatigue from excessive or poorly designed alerts can reduce effectiveness.

2Types of Medication Alerts

Alert Type	Severity	Example
Drug-Drug Interaction	High	Warfarin + Aspirin = Bleeding risk
Drug Allergy	Critical	Patient allergic to penicillin
Contraindication	High	Beta-blocker in severe asthma
Duplicate Therapy	Medium	Two ACE inhibitors prescribed
Dose Range	Medium	Dose exceeds maximum daily limit
Age/Pregnancy	Variable	Category X drug in pregnancy
Renal/Hepatic	High	Drug requires dose adjustment for CrCl <30

⚠️ Alert Severity Levels

• Critical: Life-threatening - must be addressed before proceeding
• High: Significant risk - strong recommendation to reconsider
• Medium: Moderate risk - clinical judgment required
• Low: Minor concern - informational only

3Database Design

Core Tables for Decision Support

-- Patient Allergies
CREATE TABLE patient_allergies (
  allergy_id VARCHAR(50) PRIMARY KEY,
  patient_id VARCHAR(50) NOT NULL,
  allergen_type VARCHAR(50), -- DRUG, FOOD, ENVIRONMENTAL
  allergen_nm VARCHAR(255) NOT NULL,
  allergen_cd VARCHAR(50), -- RxNorm or custom code
  reaction_desc TEXT,
  severity_cd VARCHAR(20), -- MILD, MODERATE, SEVERE, LIFE_THREATENING
  onset_dt DATE,
  verified_ind CHAR(1) DEFAULT 'N',
  created_dttm TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

-- Drug Interaction Rules
CREATE TABLE drug_interaction_rules (
  interaction_id VARCHAR(50) PRIMARY KEY,
  drug_a_rxcui VARCHAR(20) NOT NULL,
  drug_a_nm VARCHAR(255),
  drug_b_rxcui VARCHAR(20) NOT NULL,
  drug_b_nm VARCHAR(255),
  severity_cd VARCHAR(20), -- CRITICAL, HIGH, MODERATE, LOW
  interaction_desc TEXT,
  clinical_effect TEXT,
  management_rec TEXT, -- Recommended management
  evidence_level VARCHAR(20), -- PROVEN, PROBABLE, POSSIBLE
  source_nm VARCHAR(100), -- First DataBank, Micromedex, etc.
  created_dttm TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

-- Contraindication Rules
CREATE TABLE contraindication_rules (
  contraindication_id VARCHAR(50) PRIMARY KEY,
  drug_rxcui VARCHAR(20) NOT NULL,
  drug_nm VARCHAR(255),
  condition_cd VARCHAR(50), -- ICD-10 or SNOMED
  condition_nm VARCHAR(255),
  severity_cd VARCHAR(20),
  contraindication_desc TEXT,
  alternative_therapy TEXT,
  created_dttm TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

-- Alert Log (for tracking which alerts were shown)
CREATE TABLE alert_log (
  alert_log_id VARCHAR(50) PRIMARY KEY,
  patient_id VARCHAR(50),
  prescriber_id VARCHAR(50),
  alert_type VARCHAR(50), -- DDI, ALLERGY, CONTRAINDICATION, etc.
  alert_severity VARCHAR(20),
  alert_msg TEXT,
  drug_rxcui VARCHAR(20),
  override_ind CHAR(1) DEFAULT 'N',
  override_reason TEXT,
  alert_dttm TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

-- Indexes
CREATE INDEX idx_allergies_patient ON patient_allergies(patient_id);
CREATE INDEX idx_allergies_allergen ON patient_allergies(allergen_cd);
CREATE INDEX idx_ddi_drug_a ON drug_interaction_rules(drug_a_rxcui);
CREATE INDEX idx_ddi_drug_b ON drug_interaction_rules(drug_b_rxcui);
CREATE INDEX idx_contra_drug ON contraindication_rules(drug_rxcui);
CREATE INDEX idx_contra_condition ON contraindication_rules(condition_cd);

4Drug-Drug Interactions (DDI)

Drug-drug interactions occur when one medication affects the activity of another, potentially leading to adverse effects or reduced efficacy.

Common DDI Mechanisms:

Pharmacokinetic

Changes in absorption, distribution, metabolism, or excretion

Example: Ketoconazole inhibits CYP3A4, increasing simvastatin levels

Pharmacodynamic

Additive or opposing effects at receptor sites

Example: Combining two CNS depressants increases sedation

SQL: Check for Drug Interactions

-- Check if new prescription interacts with patient's current meds
WITH patient_current_meds AS (
  SELECT DISTINCT rxcui
  FROM prescriptions
  WHERE patient_id = '12345'
    AND active_ind = 'Y'
)
SELECT 
  ddi.severity_cd,
  ddi.drug_a_nm,
  ddi.drug_b_nm,
  ddi.interaction_desc,
  ddi.clinical_effect,
  ddi.management_rec
FROM drug_interaction_rules ddi
WHERE (
  -- New drug vs existing meds
  (ddi.drug_a_rxcui = '197361' -- New prescription RxCUI
   AND ddi.drug_b_rxcui IN (SELECT rxcui FROM patient_current_meds))
  OR
  (ddi.drug_b_rxcui = '197361'
   AND ddi.drug_a_rxcui IN (SELECT rxcui FROM patient_current_meds))
)
AND ddi.severity_cd IN ('CRITICAL', 'HIGH')
ORDER BY 
  CASE ddi.severity_cd
    WHEN 'CRITICAL' THEN 1
    WHEN 'HIGH' THEN 2
    ELSE 3
  END;

5Allergy Screening

Allergy checking is one of the most critical safety checks. Systems must screen for:

Direct Match

Exact match between prescribed drug and documented allergy

Class Allergy

Cross-reactivity within drug classes (e.g., all penicillins)

Ingredient Allergy

Allergies to specific inactive ingredients (dyes, preservatives)

SQL: Allergy Screening

-- Screen for drug allergies before prescribing
SELECT 
  pa.allergen_nm,
  pa.reaction_desc,
  pa.severity_cd,
  'ALLERGY ALERT: Patient has documented allergy' as alert_msg
FROM patient_allergies pa
WHERE pa.patient_id = '12345'
  AND pa.allergen_type = 'DRUG'
  AND (
    -- Direct match on drug name
    pa.allergen_nm ILIKE '%amoxicillin%'
    OR
    -- Match on drug class (would need class mapping table)
    pa.allergen_cd IN (
      SELECT drug_class_cd 
      FROM drug_classes 
      WHERE rxcui = '723' -- Amoxicillin RxCUI
    )
  )
ORDER BY 
  CASE pa.severity_cd
    WHEN 'LIFE_THREATENING' THEN 1
    WHEN 'SEVERE' THEN 2
    WHEN 'MODERATE' THEN 3
    ELSE 4
  END;

6Contraindications

Contraindications are conditions or factors that make a particular treatment inadvisable. They can be absolute (never use) or relative (use with caution).

Common Contraindication Types:

• Disease-based: Drug contraindicated in specific medical conditions
• Age-based: Pediatric or geriatric contraindications
• Pregnancy/Lactation: Teratogenic or harmful in breastfeeding
• Lab-based: Contraindicated with abnormal lab values (e.g., low K+)
• Organ dysfunction: Renal or hepatic impairment

Example Contraindications:

Metformin in severe renal impairmentAbsolute contraindication (lactic acidosis risk)

Beta-blockers in severe asthmaRelative contraindication (bronchospasm risk)

Isotretinoin in pregnancyAbsolute contraindication (teratogenic)

7Duplicate Therapy Detection

Duplicate therapy alerts warn when a patient is prescribed multiple drugs from the same therapeutic class, which may be unintentional or potentially harmful.

Detection Scenarios:

• Same Ingredient: Two formulations of the same drug (e.g., Lipitor + atorvastatin generic)
• Same Class: Two drugs from same therapeutic class (e.g., two statins)
• Same Mechanism: Drugs with identical or overlapping mechanisms
• Additive Effects: Drugs that produce same physiologic effect

SQL: Detect Duplicate Therapy

-- Find duplicate therapy in patient's medication list
WITH patient_meds_with_class AS (
  SELECT 
    p.rx_id,
    p.drug_nm,
    dc.therapeutic_class,
    dc.mechanism_of_action
  FROM prescriptions p
  JOIN drug_classes dc ON p.rxcui = dc.rxcui
  WHERE p.patient_id = '12345'
    AND p.active_ind = 'Y'
)
SELECT 
  m1.drug_nm as drug_1,
  m2.drug_nm as drug_2,
  m1.therapeutic_class,
  'DUPLICATE THERAPY: Multiple drugs from same class' as alert_msg
FROM patient_meds_with_class m1
JOIN patient_meds_with_class m2 
  ON m1.therapeutic_class = m2.therapeutic_class
  AND m1.rx_id < m2.rx_id -- Avoid duplicate pairs
WHERE m1.therapeutic_class IS NOT NULL
ORDER BY m1.therapeutic_class;

8Implementation Guide

Best Practices for CDSS Implementation:

1. Reduce Alert Fatigue

Show only clinically significant alerts. Studies show override rates >90% for low-priority alerts.

2. Provide Context

Include patient-specific data in alerts (current labs, vital signs, concurrent medications).

3. Suggest Alternatives

When alerting to a problem, provide recommended alternative medications.

4. Track Overrides

Log all alert overrides with reasons to identify alert quality issues.

5. Use Tiered Severity

Critical alerts should be interruptive; low-priority alerts can be passive.

📚 Clinical Knowledge Sources

• First DataBank (FDB): Comprehensive drug interaction database
• Micromedex: Evidence-based clinical decision support
• Lexicomp: Drug information and interaction checking
• RxNorm: NLM's normalized drug vocabulary (free)
• DailyMed: FDA medication information (free)

mdatool.

Medication Decision Support Systems

Abbreviations Used in This Guide

Table of Contents