CPIA Module 5, Section 1: The Five Rights of CDS and Types of Interventions
MODULE 5: CLINICAL DECISION SUPPORT (CDS) FUNDAMENTALS

Section 5.1: The Five Rights of CDS and Types of Interventions

Building the Architectural Blueprint for Effective Clinical Guidance.

SECTION 5.1

The Five Rights of CDS and Types of Interventions

From Clinical Philosophy to Technical Implementation: A Pharmacist’s Guide.

5.1.1 The “Why”: Establishing the Core Philosophy of CDS

Before a single line of code is written, before any alert is designed, we must begin with a foundational philosophy. In the world of health informatics, there is a concept so powerful and so essential that it governs every aspect of effective system design: The Five Rights of Clinical Decision Support. This framework is not merely a set of best practices; it is the constitution upon which all successful CDS is built. It serves as the ultimate antidote to the most pervasive and dangerous problem in clinical technology: alert fatigue.

Every clinician has experienced alert fatigue. It is the cognitive exhaustion and desensitization that occurs from being bombarded with a constant stream of low-value, irrelevant, or poorly timed digital interruptions. It is the digital “boy who cried wolf,” where a flood of meaningless alerts (e.g., a warning about a trivial interaction between multivitamins and calcium) drowns out the one critical, life-saving warning (e.g., a severe allergy alert). When users become habituated to clicking “override” without thinking, the entire safety system collapses. The goal of an informatics pharmacist is not to build more alerts. The goal is to design and implement meaningful interventions that clinicians welcome, trust, and act upon.

The Five Rights provide the architectural blueprint to achieve this. They force us to transition our thinking from a narrow, technical question (“Can I make an alert fire when Drug A and Drug B are ordered together?”) to a series of profound clinical and workflow-based questions. Mastering this framework is the first and most critical step in translating your clinical expertise into a system-level safeguard. It ensures that what you build is not just technologically functional, but clinically effective, contextually aware, and ultimately, a force for improving patient safety on a massive scale.

The RIGHT Information

The RIGHT People

The RIGHT Channels

The RIGHT Format

The RIGHT Point in Workflow

Retail Pharmacist Analogy: The Master-Level Consultation

Imagine a long-time patient, Mrs. Jones, comes to your pharmacy counter. She has a new prescription for amiodarone from her cardiologist. While you are processing it, she also places a box of over-the-counter cimetidine (Tagamet HB) on the counter for her recurring heartburn. This is a critical moment, a junction where a medication error is waiting to happen. How an expert pharmacist handles this is a perfect real-world demonstration of the Five Rights of CDS.

  • The RIGHT Information: Your brain doesn’t just see two drug names. It instantly synthesizes multiple data points into actionable knowledge. You know amiodarone has a narrow therapeutic index and a high risk of QTc prolongation. You know cimetidine is a potent inhibitor of the CYP3A4 enzyme. You know that this specific interaction can dramatically increase amiodarone levels, raising the risk of fatal arrhythmia and other toxicities. This is not generic data; it’s precise, high-stakes clinical knowledge.
  • The RIGHT People: The intervention is multi-targeted. The primary recipient is Mrs. Jones, who is about to make a dangerous mistake. The secondary recipient is yourself, as you are the professional legally and ethically required to intervene. A less-experienced technician might not recognize the danger, so the intervention must come from the pharmacist.
  • The RIGHT Channel: You don’t just hand her a printed leaflet and hope she reads it. That’s a passive, low-impact channel. You use the most effective channel available: a direct, face-to-face, interruptive conversation. You physically stop the transaction. This is an “active alert.”
  • The RIGHT Format: You don’t just say, “Don’t take these together.” That’s a “hard stop” without context. You format the information for maximum impact and understanding. You say, “Mrs. Jones, I’m glad you asked me about this. This heartburn medicine, cimetidine, can have a serious interaction with your new heart medication. It can cause your body to build up too much of the amiodarone, which could be dangerous. Instead, let me recommend an alternative like famotidine (Pepcid) or a PPI, which are much safer to take with your new prescription.” You’ve presented the problem, the risk, and an actionable solution in a clear, respectful format.
  • The RIGHT Point in Workflow: This entire intervention happens at the most critical moment: at the point of dispensing, before she has paid for and taken home the interacting medication. Intervening a week later during a refill check would be too late. The timing is perfect.

This elegant, expert consultation is what we strive to replicate with technology. Every CDS tool we build is an attempt to codify this exact process of applying the right knowledge to the right person in the right way at the right time. Your clinical judgment is the gold standard; our task in informatics is to embed that standard into the system itself.

5.1.2 Deep Dive: The First Right – The RIGHT Information

The entire edifice of Clinical Decision Support rests on a foundation of high-quality information. Without it, even the most brilliantly designed system will fail, or worse, cause harm. For an informatics pharmacist, understanding the nuances of clinical information—its sources, its hierarchy, and its attributes—is a non-negotiable core competency. We must move beyond thinking of information as just “data” and learn to see it as the raw material we shape into life-saving interventions.

From Data to Wisdom: The DIKW Pyramid in a Clinical Context

A foundational concept in information science is the Data, Information, Knowledge, Wisdom (DIKW) pyramid. It perfectly describes the process we must follow when designing CDS.

WISDOM

Applying knowledge to achieve optimal outcomes

KNOWLEDGE

Synthesizing information to create actionable rules

INFORMATION

Giving context to raw data

DATA

Discrete, objective facts

  • Level 1: Data. This is the raw material. It is discrete, unorganized, and without context. Examples: `Serum Creatinine = 2.4 mg/dL`, `Drug Ordered = Dabigatran`, `Age = 88 years`, `Weight = 52 kg`.
  • Level 2: Information. This is data that has been organized and given context. We can now make comparisons and see relationships. Example: “An 88-year-old, 52 kg patient with a serum creatinine of 2.4 mg/dL has an estimated CrCl of ~18 mL/min. The standard dose of Dabigatran is being ordered.”
  • Level 3: Knowledge. This is the application of rules and expertise to information. It answers the “so what?” question. Example: “Dabigatran is contraindicated in patients with a CrCl < 30 mL/min due to a significantly increased risk of life-threatening bleeding. The ordered dose is unsafe for this patient." This is where your clinical pharmacy knowledge resides.
  • Level 4: Wisdom. This is the highest level of abstraction. It involves applying knowledge in a nuanced, context-aware manner to produce the best possible outcome. Example: “For this 88-year-old patient requiring anticoagulation for atrial fibrillation, Apixaban would be a safer and more appropriate agent given her severe renal impairment. We should recommend a dose of 2.5 mg BID.” Wisdom is not just identifying the problem; it’s providing the optimal solution.

Our primary job in CDS design is to build systems that automate the climb up this pyramid—transforming raw EHR data into wise, actionable recommendations at the point of care.

Masterclass Table: Sources of Information for CDS

Effective CDS relies on pulling information from a wide variety of sources, each with its own strengths and challenges. As an informatics pharmacist, you must be fluent in the language of these data domains.

Information Source Description Example of Use in Pharmacy CDS Key Informatics Challenge
Patient-Specific Data (The EHR Core) Dynamic, real-time data unique to the individual patient. This is the most common fuel for CDS engines.
  • Demographics: Using age to flag pediatric or geriatric dosing. Using weight for mg/kg calculations.
  • Allergies: Firing an alert if a drug from the same class as a documented allergy is ordered.
  • Problem List: Alerting a provider who orders a beta-blocker for a patient with a documented history of severe asthma.
  • Lab Results: Triggering a renal dosing recommendation based on the latest serum creatinine.
  • Medication List: Detecting a drug-drug interaction between a new order and an existing medication.
 Interoperability and Standardization. Data must be discrete and coded (e.g., using LOINC for labs, SNOMED-CT for problems, RxNorm for drugs). Free-text entries are nearly useless for automated CDS. Ensuring data is accurate and up-to-date is a constant battle.
Clinical Knowledge Bases (The “Brains”) Third-party, curated databases containing comprehensive information on drugs, diseases, and clinical guidelines.
  • Drug Interactions: The core logic for nearly all drug-drug interaction alerts comes from vendors like First Databank (FDB) or Medi-Span.
  • Dosing Information: Providing reference information on standard, max, and organ-adjusted doses.
  • Clinical Guidelines: Embedding links or recommendations from sources like UpToDate, Lexicomp, or professional society guidelines directly into the workflow.
 Integration and Customization. The knowledge base must be seamlessly integrated with the EHR. The biggest challenge is “alert fatigue”—these databases often contain thousands of theoretical interactions, and the informatics pharmacist’s job is to customize and tune the system to only fire alerts for clinically significant issues.
Institutional Data & Policy (The “Local Rules”) Information specific to your hospital or health system.
  • Formulary: Guiding prescribers towards cost-effective, preferred medications and alerting them when a non-formulary drug is ordered.
  • Antibiogram: Recommending empiric antibiotics for sepsis based on the hospital’s local bacterial resistance patterns.
  • Clinical Protocols: Building order sets that reflect the institution’s approved protocols for conditions like VTE prophylaxis or glycemic control.
 Maintenance and Governance. Who is responsible for updating the formulary status in the EHR? How often is the antibiogram data refreshed? This requires a robust governance process involving the P&T Committee, IT, and pharmacy leadership to keep local information current.
Population-Level Data (The “Big Picture”) Aggregated data from patient populations, used to identify trends and risks.
  • Predictive Analytics: Using a patient’s combined risk factors (age, comorbidities, lab values) to generate a score predicting their risk of developing a condition like C. difficile, and alerting the team to implement preventative measures.
  • Registries: Creating dashboards that identify all diabetic patients in a clinic who are overdue for an A1c test or a statin prescription.
 Data Quality and Algorithmic Bias. These tools are highly dependent on massive amounts of clean, well-structured data. There is also a significant risk that algorithms trained on one population may not perform well on another, potentially exacerbating healthcare disparities. Validation and ongoing monitoring are critical.
Garbage In, Gospel Out: The Peril of Unvalidated Data

There is a dangerous phenomenon in health IT known as “Garbage In, Gospel Out.” This is an evolution of the classic “Garbage In, Garbage Out” problem. It occurs when flawed, incomplete, or inaccurate data is entered into an EHR (Garbage In), but because the computer then presents this information in a clean, authoritative, and official-looking format, clinicians begin to trust it as if it were truth (Gospel Out).

A Classic Pharmacy Example: The Outdated Allergy List. A patient had a mild rash to amoxicillin as a child. It was entered into the EHR as a severe, anaphylactic allergy to “Penicillin.” Ten years later, the patient is admitted with a life-threatening infection for which a beta-lactam is the drug of choice. The EHR fires a hard-stop, critical allergy alert based on the outdated, inaccurate data. The prescriber, fearing a lawsuit and trusting the “Gospel” on the screen, opts for a second-line agent that is less effective and more toxic. The CDS, built on a foundation of garbage data, has directly led to patient harm.

As an informatics pharmacist, you must be a profound skeptic. You must constantly question the provenance and quality of the data used to drive your CDS. Your role involves not just building the rules, but also championing the data governance and workflow processes (like regular medication and allergy reconciliation) that ensure the underlying data is as clean and reliable as possible.

5.1.3 Deep Dive: The Second Right – The RIGHT People

Once you have secured high-quality information, the next critical question is: who needs to see it? A “one-size-fits-all” approach to delivering CDS is a guaranteed path to irrelevance and user frustration. An alert that is critically important to a pharmacist during verification may be completely meaningless noise to a physician during order entry. A piece of guidance vital for a nurse at the bedside may be an unnecessary distraction for a respiratory therapist. Tailoring the CDS intervention to the specific role, responsibilities, and cognitive workflow of the recipient is paramount. This is the essence of role-based CDS.

Role-Based CDS: Thinking from the User’s Perspective

Effective CDS design requires a deep sense of empathy for the end-user. Before building any rule, you must ask yourself a series of questions from the perspective of the person who will receive the intervention:

  • What is this person’s primary goal at this exact moment? (e.g., A physician is trying to rapidly enter admission orders; a nurse is trying to administer medications safely and on time).
  • Is this person empowered to act on this information? (e.g., Alerting a nurse about a drug-drug interaction is less effective than alerting the prescriber who can change the order, or the pharmacist who can block it).
  • Does this information fit within their existing mental model and scope of practice? (e.g., A complex pharmacokinetic recommendation is perfect for a pharmacist but may be overly detailed for other roles).
  • What is the “signal-to-noise” ratio for this user? (e.g., Attending physicians in the ICU are already managing a high cognitive load; any CDS directed at them must be of the absolute highest value and urgency).
Masterclass Table: Tailoring CDS Interventions to the Healthcare Team
Role Primary Medication-Related Focus Example of Poorly-Targeted CDS Example of Well-Targeted CDS
Prescriber (Physician, NP, PA) Diagnosis, treatment selection, and ordering. They are focused on the “what” and “why” of therapy. An alert during order entry that says “Drug is stored in the north tower pharmacy.” This is logistical information that is irrelevant to the clinical decision being made. An alert during order entry: “Patient’s eGFR is 25 mL/min. The recommended dose of enoxaparin for this level of renal impairment is 30 mg daily, not 30 mg BID.” This is clinically relevant, actionable, and occurs at the precise moment of decision.
Pharmacist Safety, appropriateness, efficacy, and logistics. They conduct the final clinical review and are responsible for ensuring the order is perfect before it leaves the pharmacy. A basic drug-drug interaction alert for amlodipine and simvastatin. This is well-known and easily managed; a competent pharmacist doesn’t need an alert for it. This is noise. A highly specific, high-risk alert: “Patient has a documented G6PD deficiency. The order for rasburicase is contraindicated due to the risk of severe hemolytic anemia.” This is a rare but catastrophic error that a pharmacist is perfectly positioned to intercept.
Nurse Safe and timely administration, patient monitoring, and patient education. They are focused on the “how” and “when” of therapy at the bedside. An alert upon scanning a medication that details the full mechanism of action of the drug. This is academic information, not actionable at the point of administration. An alert upon scanning a vancomycin dose: “WARNING: This is a loading dose. Infuse over a minimum of 90 minutes to prevent Red Man Syndrome.” This provides a critical administration parameter necessary for immediate patient safety.
Respiratory Therapist (RT) Administration of inhaled medications, management of oxygen therapy and mechanical ventilation. An alert about a potential interaction between a patient’s oral medications. This is outside their primary scope of practice. An order set for asthma exacerbation that includes a default sequence of nebulized albuterol/ipratropium treatments and automatically pages the RT to the patient’s room. This directly supports and initiates their workflow.
Patient Adherence, understanding their condition, managing costs, and scheduling follow-ups. A complex clinical alert in their patient portal filled with medical jargon about pharmacokinetics. This is confusing and likely to cause anxiety. A clear, simple text message or portal notification: “It’s time to refill your Lisinopril. Click here to send a request to your pharmacy.” Or: “You are due for a blood test to monitor your warfarin. Please schedule an appointment with the lab this week.”
Team-Based CDS: Using Technology to Connect the Dots

The most sophisticated CDS moves beyond targeting a single individual and instead focuses on facilitating communication and coordinated action across the entire team. The EHR can act as a central hub, ensuring that an action taken by one person automatically and reliably triggers the necessary next step for another.

Example: The Pharmacist-Driven Dosing Protocol.

  1. A physician places an order for “Vancomycin per Pharmacy Protocol.” This is a trust-based handoff.
  2. CDS for the Pharmacist: The order generates a task in the pharmacist’s work queue. The pharmacist opens a specialized flowsheet (a form of CDS) that pulls in the patient’s weight, SCr, and other relevant data, and recommends an initial loading dose and maintenance regimen based on institutional guidelines.
  3. The pharmacist finalizes the dosing plan and signs the order.
  4. CDS for the Nurse: The signed order appears on the nurse’s Medication Administration Record (MAR). A rule attached to the order automatically schedules a future task for the nurse to draw a vancomycin trough level at the appropriate time (e.g., 30 minutes before the 4th dose).
  5. CDS for the Lab: The nurse’s action of drawing the lab sends the sample to be processed.
  6. CDS for the Pharmacist (Round 2): When the lab result is finalized, a new CDS rule fires, sending a high-priority task back to the pharmacist: “Vancomycin trough result available for Patient X. Please re-evaluate dosing.”

In this example, no single alert was a simple pop-up. Instead, the CDS was a choreographed series of tasks, order sets, and notifications that guided multiple people through a complex clinical process, ensuring no steps were missed. This is the future of truly effective decision support.