What is the difference between passive and active CDS?

Passive CDS presents information without interrupting the provider's workflow. Examples include order sets, documentation templates, and reference links. Active CDS interrupts the workflow with alerts that require acknowledgment. Most CDS failures stem from making everything active. Default to passive presentation and reserve active alerts for situations where the clinical risk justifies the interruption.

How do I handle CDS in a multi-EHR environment?

If your product integrates with multiple EHR systems, abstract your CDS logic into a rules engine that is EHR-agnostic. Use FHIR CDS Hooks (an HL7 standard) as the integration layer. Each EHR calls your rules engine at defined workflow touchpoints, and your engine returns cards with recommendations. This avoids rebuilding logic per EHR.

When does CDS require FDA clearance?

CDS that meets all four criteria in the 21st Century Cures Act exemption is not considered a medical device: (1) it is not intended to acquire or analyze medical images or signals, (2) it is intended for a healthcare professional, (3) it is intended for the purpose of enabling that professional to independently review the basis for the recommendation, and (4) it is intended for the purpose of not replacing the clinical judgment of a healthcare professional. If your CDS meets all four criteria, it is likely exempt. If it does not meet all four, consult your regulatory team.

How do I validate CDS rules before going live?

Run a retrospective validation against 6-12 months of historical patient data. Count how many times each rule would have fired, what the provider actually did, and whether the alert would have changed the outcome. This gives you a baseline override rate estimate and helps you tune triggering conditions before providers ever see the alerts.

What is the ideal number of CDS alerts per encounter?

Research consistently shows that provider attention drops sharply after 3-5 alerts per encounter. Above 10, providers adopt "alert dismissal autopilot" and stop reading entirely. Aim for fewer than 5 unique alerts per encounter. If your system generates more, re-evaluate rule specificity and severity classification. ---

Clinical Decision Support Template

What This Template Is For

Clinical decision support (CDS) features surface relevant clinical knowledge at the point of care to help providers make better decisions. Done well, CDS reduces diagnostic errors, catches drug interactions, and flags gaps in preventive care. Done poorly, it floods providers with irrelevant alerts until they click through everything without reading.

This template helps health tech PMs spec CDS features that balance clinical value with usability. It covers rule definition, evidence sourcing, alert severity tiers, provider workflow integration, and the measurement plan you need to prove the system is helping rather than hurting. If you are building within a regulated health product, pair this with the HIPAA Compliance Checklist Template to ensure your data handling meets requirements. For context on how CDS fits into the broader care experience, see the Patient Journey Template.

How to Use This Template

Copy the template into your documentation tool.
Define the clinical use case and target provider population.
Identify the clinical rules, their evidence base, and the triggering conditions.
Design the alert presentation: severity tier, content, and recommended actions.
Map the provider workflow to determine where and when alerts appear.
Define suppression and override logic to prevent alert fatigue.
Build the measurement plan: clinical outcomes, alert metrics, and provider feedback loops.
Review with your clinical advisory team before development. Use the stakeholder management guide to plan your review cadence.

The Template

CDS Feature Context

Field	Details
Clinical Domain	[e.g., Medication Safety, Preventive Care, Diagnosis Support]
Target Condition(s)	[e.g., Type 2 Diabetes, Hypertension, Sepsis screening]
Target Provider Roles	[e.g., Primary Care Physician, Emergency Medicine, Pharmacist]
Care Setting	[e.g., Ambulatory, Inpatient, Pharmacy, Telehealth]
EHR Integration Points	[e.g., CPOE, Chart review, Problem list, Medication reconciliation]
Regulatory Classification	[e.g., FDA Class II Medical Device, Exempt, Clinical workflow tool]
Product Owner	[Name]
Clinical Advisor	[Name, credentials]
Target Release	[Date or sprint]

Clinical Rules Definition

For each CDS rule, document the triggering condition, evidence basis, and expected clinical action.

Rule 1: [Rule Name]

Rule ID. [Unique identifier, e.g., CDS-MED-001]

Clinical Intent. [One sentence: what clinical outcome does this rule support?]

Triggering Condition

IF [patient data condition]
AND [additional patient data condition]
AND [contextual condition, e.g., provider is ordering medication]
THEN [trigger alert]

Example Trigger

IF patient.medications CONTAINS warfarin
AND new_order.medication IN [NSAIDs list]
AND patient.age >= 65
THEN trigger drug-interaction alert (severity: HIGH)

Evidence Source

Source	Reference	Strength
[e.g., FDA Drug Safety Communication]	[URL or citation]	[Strong / Moderate / Weak]
[e.g., UpToDate clinical pathway]	[URL or citation]	[Strong / Moderate / Weak]
[e.g., Local formulary committee]	[Internal reference]	[Consensus-based]

Data Requirements

☐ [Data element needed, e.g., Current medication list from EHR]
☐ [Data element needed, e.g., Patient age and weight]
☐ [Data element needed, e.g., Lab results within last 30 days]
☐ [Data element needed, e.g., Problem list / diagnosis codes]

Recommended Action for Provider

[Primary action, e.g., "Consider alternative analgesic. Acetaminophen is first-line for this patient."]
[Secondary action, e.g., "If NSAID is clinically necessary, add PPI for GI protection."]
[Documentation action, e.g., "Document clinical rationale if overriding this alert."]

Rule 2: [Rule Name]

(Repeat the same structure for each clinical rule)

Alert Design Specification

Severity Tiers

Define how alerts are presented based on clinical urgency. Fewer tiers are better. Three tiers is the recommended maximum.

Tier	Severity	Presentation	Provider Action Required	Example
1	Critical	Hard stop. Modal dialog. Cannot proceed without acknowledging.	Must select override reason or modify order	Contraindicated drug-drug interaction
2	Warning	Interruptive alert with recommended action. Can dismiss with one click.	Review recommended action, dismiss or accept	Dose outside normal range for age/weight
3	Informational	Non-interruptive. Inline display in chart or order context.	No action required. Read at provider discretion.	Preventive care gap reminder

Alert Content Structure

Each alert should include these elements, regardless of severity:

☐ Headline. [One sentence: what is the clinical concern?]
☐ Patient context. [Relevant patient data that triggered the alert]
☐ Evidence summary. [One sentence: why does this matter clinically?]
☐ Recommended action. [Specific, actionable step the provider can take]
☐ Override options. [For Tier 1-2: structured reasons for overriding the alert]
☐ Learn more link. [Link to full clinical reference for providers who want detail]

Alert Fatigue Prevention

☐ Define maximum alerts per encounter: [e.g., 5 unique alerts per patient visit]
☐ Define suppression rules: [e.g., Tier 3 alert suppressed if same alert fired within 30 days for this patient]
☐ Define provider-level preferences: [e.g., Provider can mute Tier 3 alerts for categories they have reviewed]
☐ Define override tracking: [e.g., If provider overrides same Tier 2 alert > 90% of the time, flag for rule review]

Provider Workflow Integration

Map where alerts appear in the provider's existing workflow. CDS that interrupts at the wrong moment gets ignored.

Workflow Touchpoints

Workflow Step	CDS Trigger Point	Alert Tier(s)	Rationale
[e.g., Opening patient chart]	Chart review	Tier 3 (informational)	Provider scanning context; non-interruptive reminders fit here
[e.g., Entering medication order]	CPOE order entry	Tier 1-2	Provider is making an active decision; interruptive alerts are appropriate
[e.g., Signing encounter note]	Encounter close	Tier 3	Final checklist items; preventive care gaps
[e.g., Reviewing lab results]	Results inbox	Tier 2	Abnormal results that need follow-up action

Workflow Diagram

Describe or link to a diagram showing where each CDS rule fires within the provider's workflow. Include the EHR screen or module name.

[Provider opens chart] → Tier 3 reminders display in sidebar
         ↓
[Provider enters order] → Tier 1-2 alerts check against rules engine
         ↓
[Provider reviews alert] → Accept recommendation OR Override with reason
         ↓
[Provider signs note] → Tier 3 care gap checklist

Data Architecture

Data Sources

Data Source	Data Elements	Refresh Frequency	Integration Method
[e.g., EHR medication module]	Current medications, allergies	Real-time	[FHIR API / HL7v2 / Direct DB]
[e.g., Lab system]	Recent lab results, reference ranges	[Batch nightly / Real-time]	[Interface type]
[e.g., Problem list]	Active diagnoses (ICD-10 codes)	Real-time	[Interface type]
[e.g., Clinical knowledge base]	Drug interaction database, clinical guidelines	[Monthly update]	[Vendor subscription]

Data Quality Requirements

☐ Define minimum data completeness thresholds: [e.g., medication list must be reconciled within 90 days]
☐ Define handling for missing data: [e.g., if allergy status is "unknown," fire Tier 3 reminder to update allergies]
☐ Define data latency tolerance: [e.g., lab results must be available within 2 hours of resulted status]

Measurement Plan

Clinical Outcome Metrics

Metric	Baseline	Target	Measurement Method
[e.g., Rate of contraindicated drug orders reaching pharmacy]	[Current %]	[Target %, e.g., < 0.1%]	[Pharmacy dispense audit]
[e.g., Preventive care screening compliance rate]	[Current %]	[Target %, e.g., > 85%]	[EHR quality reporting]
[e.g., Adverse drug event rate]	[Current rate per 1000 encounters]	[Target rate]	[Incident reporting system]

Alert Performance Metrics

Metric	Target	Red Flag
Alert override rate	< 70% for Tier 2, < 10% for Tier 1	> 90% override rate signals alert is not useful
Time to dismiss	< 5 seconds for Tier 2-3	< 1 second suggests provider is not reading the alert
Unique alerts per encounter	< 5	> 10 signals alert fatigue risk
Alert-to-action rate	> 30% for Tier 2	< 10% suggests poor clinical relevance

Use product metrics tracking to monitor these over time. Consider running an A/B test on alert presentation formats to optimize the action rate.

Provider Feedback Loop

☐ Schedule quarterly review with clinical advisory committee
☐ Collect structured provider feedback via in-app rating (thumbs up/down on each alert)
☐ Publish monthly CDS performance dashboard to clinical leadership
☐ Define rule retirement criteria: [e.g., if override rate > 95% for 3 consecutive months, retire or redesign the rule]

Filled Example: Warfarin Drug Interaction CDS

CDS Feature Context (Example)

Field	Details
Clinical Domain	Medication Safety
Target Condition(s)	Patients on warfarin anticoagulation therapy
Target Provider Roles	Prescribers (MDs, DOs, NPs, PAs) in ambulatory and inpatient settings
Care Setting	Ambulatory + Inpatient + Pharmacy
EHR Integration Points	CPOE medication ordering module
Regulatory Classification	Clinical workflow tool (not FDA-regulated medical device)

Rule: NSAID-Warfarin Interaction (Example)

Rule ID. CDS-MED-001

Clinical Intent. Prevent gastrointestinal bleeding in patients on warfarin who are prescribed an NSAID.

Triggering Condition

IF patient.active_medications CONTAINS warfarin
AND new_order.medication IN [ibuprofen, naproxen, diclofenac, meloxicam, ketorolac, indomethacin]
THEN trigger drug-interaction alert (severity: HIGH)

Evidence Source

Source	Reference	Strength
FDA Drug Safety Communication	FDA.gov/drugs/drug-safety	Strong
CHEST Antithrombotic Guidelines 2024	DOI: 10.1016/j.chest.2024.xx	Strong
Local P&T Committee recommendation	Internal formulary policy FP-2025-14	Consensus

Recommended Actions

Primary: "Consider acetaminophen (up to 2g/day) as an alternative analgesic."
Secondary: "If NSAID is clinically necessary, add omeprazole 20mg daily for GI protection and monitor INR within 3-5 days."
Override options: (a) Patient has documented NSAID tolerance, (b) Short course < 3 days with GI prophylaxis, (c) No suitable alternative per clinical judgment.

Tips for Effective CDS Design

Start with one high-impact rule, not 50. The most common CDS failure mode is launching with too many rules at once. Pick the single rule with the strongest evidence base and highest clinical impact. Prove the workflow works before scaling. Use the RICE Calculator to prioritize your rule backlog.

Design for the 30-second encounter. Providers spend an average of 15-30 seconds per alert. Your alert headline and recommended action must be scannable in that time. If it takes more than 30 seconds to understand the alert and decide what to do, the override rate will approach 100%.

Measure the override rate from day one. The override rate is the single best leading indicator of CDS value. A high override rate does not always mean the rule is wrong. It may mean the alert fires in the wrong context or the recommended action is impractical. Investigate before removing.

Involve pharmacists and nurses, not just physicians. Different clinical roles interact with CDS differently. A pharmacist reviewing a dispensing alert has different context than the prescriber who wrote the order. Map workflows for each role involved.

Plan for knowledge maintenance. Clinical guidelines change. Drug databases update. Your CDS rules will need regular review. Budget ongoing clinical informaticist time for rule maintenance, not just initial build. Review the clinical workflow template for mapping ongoing maintenance processes.

Key Takeaways

Design CDS for the 30-second encounter. If the alert is not scannable, it will be overridden
Start with one high-evidence rule and prove the workflow before adding more
The override rate is your most important leading indicator. Monitor it from day one
Alert fatigue kills clinical value. Fewer, more relevant alerts outperform an exhaustive rule set
Budget for ongoing knowledge maintenance. Clinical guidelines change and your rules must keep pace

About This Template

Created by: Tim Adair

Last Updated: 3/5/2026

Version: 1.0.0

License: Free for personal and commercial use

Clinical Decision Support Template