Designing the Intake Workflow—From First Contact to a Clean Initial Case

Multiple front doors, one process. Cases arrive via EDC triggers, phone, email, ePRO/eCOA alerts, home-health providers, pharmacists, or imaging readers. Standardize the front-door script: confirm minimum criteria; confirm identity and consent to be contacted; triage seriousness; collect immediate life-saving details; then step through the core data set. Whether the conversation is five minutes or fifty, the order is the same, and the outcome is a clean, valid initial case.

Required intake data set (initial). Patient initials/identifier and demographics; reporter details and relationship to patient; suspect product/device (name, lot/serial, dose, route, regimen; for devices, model/firmware/software); concomitants and relevant history; event term(s) in reporter’s words; onset date/time and course; seriousness criterion; site/sponsor relatedness (best available); and actions taken (dose changed, drug stopped, device removed/adjusted), plus immediate outcome. Capture the clock-starting timestamp and the communication channel used.

Follow-up data set (targeted, not fishing). Use a structured list tailored to the event type. For hepatic signals, lab trends (ALT/AST/bili, INR), imaging, viral serologies; for cardiac, ECGs/QTc method and rate; for device malfunctions, usage context, user role, training status, log files, photos, returned unit tracking, and whether the malfunction could cause serious injury if repeated. Follow-up requests should be purposeful, not generic; excessive fishing wastes goodwill and delays clear cases.

Narratives as structured evidence, not prose. Use a standard format: baseline condition and relevant history; exposure timeline; event onset and clinical course; tests/procedures and results; alternative etiologies considered; dechallenge/rechallenge; device specifics (configuration, environment, human-factors); outcome; and a one-sentence causality rationale. The narrative should reference attachments (e.g., “see discharge summary dated …”) and align with coded fields. If the narrative says “syncope,” the coded term should not be “fall.”

Medical coding and version locks. Code the event using the current dictionary version (e.g., MedDRA), lock the version at case intake, and record the dictionary/date so expectedness mapping and aggregate reports are reproducible. When the dictionary changes mid-study, set change-control rules: whether (and when) to re-code historic cases for aggregate outputs without rewriting the history of individual transmissions.

Duplicate detection and linking. The same hospitalization can appear as a phone call from the PI, an EDC SAE form, and an email from a family member. Employ deterministic keys (site ID + subject ID + onset date) and fuzzy logic (same hospital, near-identical terms) to merge or link related records. Never delete duplicates; cross-reference them. Duplicates that slip through generate inspection questions and sometimes double expedited submissions.

Privacy, identity, and decentralized workflows. Remote reporting introduces identity-verification risk. Require two-factor checks for tele-consent and for safety calls that request protected details; store only necessary data. Home-health and courier partners must use immutable logs and synchronized clocks; time drift breaks the chain-of-custody narrative for product returns, device retrievals, and doorstep temperature exceptions. Participant messaging must be factual, non-promotional, and respectful; scripts are part of the TMF evidence.

Device specifics at intake. A malfunction without injury may still be reportable if it could cause serious harm if repeated. Intake scripts must parse hardware vs. software vs. use error; capture IFU version, battery/power status, environment (EMI, fluids), and any recent firmware updates. Record whether a returned unit is expected, assign a tracking ID, and create a placeholder for the engineering analysis so the recurrence-risk decision arrives before an expedited deadline.

Case ownership and hand-offs. Assign a named case owner at creation with a visible SLA clock. Handoffs (intake → medical review → submissions) must be timestamped and justified (“awaiting imaging log from core lab,” “engineering analysis ETA 24 h”). When SLA risk appears (e.g., a cross-border holiday will land on day 3), escalate early and add temporary containment (provisional narrative with “best available” relatedness; update in follow-up submission).

Timelines, Buffers, and Escalation—Making Deadlines Boring

Principle: clocks start on awareness with a valid case. The moment the sponsor (or designee) has the four minimum criteria, the regulatory clock starts. Calendars do not pause for weekends or local holidays; build internal SLAs to be stricter than external deadlines and to anticipate cross-border time zones. Record awareness time precisely and keep it immutable; it anchors the submission story.

Routing logic drives the pace. Time-sensitive routes include expedited transmission of serious, related, and unexpected events for drugs/biologics; expedited reportability for serious, unanticipated device effects and certain malfunctions; and prompt notifications to relevant ethics bodies per local rules. Non-expedited SAEs still require timely documentation and follow-up. Your workflow engine should map intake triage to the appropriate route automatically, with human override logged and justified.

Design internal SLAs that feel generous—but are not. Set internal targets that comfortably beat external timelines. Practical examples many sponsors adopt: triage and case validity confirmation within hours of receipt; initial case entry within one business day; medical review within a short, fixed window; expectedness mapping and country routing prepared same day for likely expedited cases; and a submission buffer that avoids last-minute bottlenecks in case of translation or portal downtime. Make the SLA dates visible on the dashboard so operations know the plan, not just the rule.

Buffers and “day zero” ambiguity. Some regulators define “day 0” as the day the sponsor (or designee) becomes aware; others apply similar logic. Avoid philosophical debates by designing buffers that eliminate the difference between “calendar” and “working days” in practice. When awareness occurs after working hours, treat it as the same calendar day for calculations—this conservative approach simplifies storytelling and avoids threshold arguments.

Weekends, holidays, and cross-border coverage. Publish a coverage plan that names who answers the phone after hours, who can approve unblinding for safety, who can sign the case, and who can file in national portals that are open on weekends. Maintain a rolling holiday calendar across relevant countries and route submissions to backup staff when a local office is closed.

Submission mechanics and proof. Whether via E2B(R3) gateway, national web portals, or email with secure attachments (where permitted), test the route before a live case. Capture portal receipts, acknowledgments, checksum logs, and copies of narratives sent. Keep a one-click evidence chain: dashboard date → narrative → attachments → transmission proof. During inspection, the first question after “why did you classify this way?” is often “show me proof you sent it in time.”

Interim vs. final narratives and version control. If the clock forces an expedited transmission with partial information, file an interim narrative clearly labeled as such, then send follow-up with new data. Version narratives with a short “what changed and why” header: “Added creatinine trend and ultrasound; expectedness unchanged; causality refined from possible to probable based on dechallenge.” Never overwrite history; append and explain.

Early escalation rules. Define objective triggers for management attention: multiple open expedited clocks, cross-border portal outage, engineering analysis overdue on a device malfunction with serious potential, or a spike in “unknown seriousness” cases at intake. Escalation is useful only if it leads to resource movement—backup medical reviewer, alternate engineer, or a prioritized site contact—so the clock is protected.

Country routing and translation. Prepare translation vendors and glossaries in advance so “case-critical” phrases are consistent. For countries with specific templates or portal fields, maintain checklists and dry-run screenshots in the TMF. The ability to pre-fill static sections (sponsor details, contact, product dictionary) saves minutes that matter at the edge of a deadline.

Ethics body notifications. Even when not on the expedited clock, many programs require rapid notification to IRBs/IECs for SAEs deemed related and unexpected, or for any event that alters the risk-benefit profile. Keep ethics notifications on the same timeline board as regulatory submissions; evidence of consistency bolsters credibility.

Governance, Metrics, Common Pitfalls, and a Ready-to-Use Checklist

Ownership and meaning of approval. Keep the board small and named: Safety Operations Lead (accountable), Safety Physician (causality/medical accuracy), Regulatory Submissions Lead, Device Engineer (when applicable), Data Management (reconciliation), and Quality (ALCOA++/traceability). Each approval records its meaning—“medical accuracy verified,” “country routing confirmed,” “engineering assessment reviewed,” “ALCOA++ check passed.” Signatures that say what was approved are easier to defend than signatures that only show who.

Dashboards that change behavior. Track: average hours from awareness to validity confirmation; percentage of cases with complete minimum criteria on first contact; intake-to-entry time; intake-to-medical review time; expedited clock burn-down; proportion of interim narratives; translation cycle time; duplicate rate; reconciliation gap rate between safety and EDC; and proof-of-submission click-through rate. Each number must click to the artifacts: call notes, narratives, attachments, and acknowledgments.

KRIs and QTLs that force decisions. Monitor early-warning indicators: spike in “unassessable” causality; high mismatch rate between narrative and coded fields; >0% “unknown seriousness” at the time of submission; repeated engineering delays in device malfunctions; or recurring portal rejection reasons. Convert the most consequential to Quality Tolerance Limits. Examples: “≥5% of expedited cases missing proof of submission” or “≥10% of cases submitted with narrative/field inconsistency”—crossing the limit triggers a documented review and corrective action.

Safety–EDC reconciliation made routine. Schedule periodic listings that reconcile subject ID, onset date, term, seriousness, relatedness, and outcome between systems. Resolve mismatches with audit-trailed change notes. If the study collects site causality in EDC and sponsor causality in safety, label both clearly, display them together, and document which governs expedited routing (the most conservative plausible assessment).

Training and calibration library. Build a set of anonymized cases with final adjudications and timelines intact. Train new staff with paired examples differentiated by one fact (e.g., onset 2 hours vs. 2 weeks; present vs. absent dechallenge). After any IB/RSI update or device IFU change, run a micro-refresher because expectedness and malfunction logic can flip for the same clinical picture.

Common pitfalls—and durable fixes.

• Clocks started late because “the report was incomplete.” Fix with the minimum criteria rule at intake and a visible awareness timestamp.
• Narratives that don’t match fields. Fix by generating narratives from structured fields and forcing a “consistency check” before lock.
• Duplicate submissions. Fix with deterministic keys and fuzzy matching; link rather than delete duplicates; teach sites how to update the same case.
• Portal rejections at hour eleven. Fix with dry-runs, prefilled static fields, and a submission buffer; capture rejection reasons for trend analysis.
• Device malfunction cases without recurrence risk analysis. Fix by creating an engineering placeholder at intake and a 24-hour analysis SLA.
• Weekend/holiday gaps. Fix with an explicit after-hours roster, cross-border calendar, and a submissions backup.
• Unblinding creep. Fix with a safety-only unblinded unit and minimum-necessary disclosure; log who saw what and why.

30–60–90-day implementation plan. Days 1–30: finalize intake scripts and the minimum criteria checklist; wire the awareness timestamp; publish internal SLAs and escalation rules; build narrative templates; test submission routes; create the duplicate-detection logic. Days 31–60: pilot the workflow in two countries; run weekend drills; measure intake-to-entry and intake-to-medical review time; tune engineering and translation SLAs; start monthly reconciliation with EDC. Days 61–90: scale to all sites; lock KRIs/QTLs; institute weekly risk huddles; expand the calibration library; and complete a five-minute retrieval drill from dashboard case ID → narrative → submission proof.

Ready-to-use SAE intake & timelines checklist (paste into your safety plan or SOP).

• Minimum case criteria enforced at intake (patient, reporter, suspect product/device, event) with immutable awareness timestamp.
• Structured triage of seriousness, relatedness, and provisional expectedness; device malfunction taxonomy captured where relevant.
• Initial and targeted follow-up data sets defined; narrative template aligned to coded fields; dictionary version/date recorded.
• Duplicate detection and cross-reference rules implemented; never delete, always link.
• Privacy and identity verification defined for remote reporting; unblinding firewall and access logs in place.
• Internal SLAs stricter than external deadlines; weekend/holiday roster and cross-border calendar active.
• Submission mechanics dry-run completed; proof (receipts, acknowledgments, checksums) stored with one-click retrieval.
• Interim and final narrative version control with “what changed and why” headers.
• Reconciliation listings with EDC scheduled and defects closed with audit trails.
• KRIs/QTLs monitored; red thresholds trigger documented corrective actions and resource moves.

Bottom line. SAE intake works when it is engineered as a small, disciplined system: confirm minimum criteria, capture the right facts in the right order, lock ALCOA++ evidence, and route to the correct timeline with buffers and proof. Build that system once—scripts, templates, dashboards, SLAs, reconciliation—and you will make deadlines boring, decisions defensible, and inspections straightforward across drugs, biologics, devices, and decentralized workflows.