Why this matters
A dashboard without context is a Rorschach test—each person sees something different. Clinical Data Management (CDM) teams need dashboards that drive decisions, not debates. The missing ingredient is usually narrative:
- Explain the movement: A Key Performance Indicator (KPI) went red—which input changed and why?
- Point to action: Who will do what, and by when?
- Protect trust: A visible change log prevents “metric drift,” where definitions slowly shift and stakeholders lose confidence.
Plain‑language rule: Define all abbreviations on first use (e.g., CRO = Contract Research Organization; SAE = Serious Adverse Event). Use the same wording across studies.
Action notes that work
Every KPI that is yellow/red needs one short, action‑oriented note. A good note answers: What changed? Why? Who acts? By when? It also links to evidence.
Recommended format
Owner → Action → Due date → Status → Evidence link
What to include
- Owner: Specific role or person (e.g., “CRO Lead,” “Site PM,” “DM for Country X”).
- Action: One clear verb (“coach,” “re‑deliver,” “reconcile,” “close queries”).
- Due date: A date, not “ASAP.”
- Status: Planned, In progress, Blocked, or Done.
- Evidence link: Ticket, email thread, meeting notes, or vendor portal reference.
Examples (good)
- CRO Lead → Coach Site B on Visit 3 expectedness; complete by Friday → In progress → Link to training deck.
- Lab Vendor PM → Re‑deliver Week‑22 data cut including missing sodium results → Due Wed → Scheduled → Ticket #123.
Non‑examples (weak)
- “Investigating late data.” (No owner, date, or action.)
- “Site B is slow.” (Complaint, not a plan.)
- “Team to review.” (Who is “team”? What will they do?)
When to write a note
- KPI changes status (green→yellow or yellow→red).
- KPI direction contradicts last note’s expectation (e.g., trend worsened despite action).
- A dependency exists (vendor delivery, protocol amendment, site retraining).
Tone & structure tips
- Keep it to one sentence plus a link.
- Use numbers in plain words (e.g., “12 records,” “three sites”), not formulas.
- Avoid blame; focus on unblocking.
Change logs & definitions
Dashboards age well when their definitions and thresholds are versioned in public.
What to maintain
Metric glossary
Short, stable descriptions for terms used in CDM dashboards:
- Currency: How up‑to‑date data entries and query closures are.
- Cleanliness: Proportion of data without open discrepancies.
- Expectedness: Whether a visit/form/lab was due based on schedule.
- Aging: How long queries or tasks remain open.
- Reconciliation: Agreement between sources (e.g., CRF vs. central lab).
Threshold history
- Record when and why a threshold changed (e.g., late‑data threshold moved from “10%” to “12%” due to surge onboarding).
- Capture who approved and what the predicted impact is (e.g., fewer false alarms during startup).
Version manifest (per dashboard release)
- Data cut timestamp (UTC and study timezone).
- Dictionary versions (e.g., MedDRA, WHO Drug).
- Derivation versions (which business rules/scripts generated KPIs).
- Known limitations for that cut (e.g., “Site C lab feeds paused; no chemistry panels 2025‑03‑01 to 2025‑03‑10”).
Simple templates
Metric glossary entry
Name: Query Aging %>14 days
Definition (plain words): Percent of all open queries that have been open longer than 14 days.
Owner: CDM Analytics
First effective date: 2025‑01‑15
Related view(s): Query Aging Trend, Site Drilldown
Threshold change entry
Date: 2025‑04‑02
Item: Query Aging %>14 days (yellow->red threshold)
Change: Red threshold increased from 10% to 12% for startup phase
Reason: 12 new sites added; temporary backlog expected; reduce false red alarms
Approved by: Study DM Lead + CRO PM
Effective until: 2025‑05‑31 (review at IPR meeting)
Version manifest snippet
Dashboard version: v1.6 (released 2025‑04‑08)
Data cut: 2025‑04‑07 23:00 UTC
Dictionaries: MedDRA 27.0; WHO Drug B3 Q1‑2025
Derivations: ruleset v3.2
Limitations: No Week‑14 labs for Sites D/E due to courier strike
Where to put it: Same workspace as the dashboard (tab or collapsible panel named “Glossary & Change Log”). Don’t bury it in a separate repository.
Worked examples (CV trial)
Context: A cardiovascular (CV) outcomes trial is onboarding new sites. KPI in focus: Query Aging over 14 days = the percent of open data queries that have been open longer than 14 days. Roles mentioned:
- CRO Lead (Contract Research Organization Lead)
- Lab Vendor PM (Project Manager at the central laboratory)
- Study DM Lead (Lead Clinical Data Manager)
1) Explaining a temporary spike
What happened: Query Aging %>14 days rose from green to red this week.
Why: 12 new sites went live with first data entry; staff were still learning visit expectedness and query workflow.
Action plan:
- Owner: Study DM Lead
- Action: Add 1.5 full‑time equivalents (FTE) of temporary query triage for 8 weeks; run two coaching sessions for new sites on expectedness and common edit checks.
- Due date: Start Monday; complete coaching by Day 10.
- Status: In progress
- Evidence link: “New‑Site Coaching Plan” slides + resource ticket.
Dashboard note (as it should appear)
Study DM Lead → Add 1.5 FTE for query triage and coach 12 new sites on expectedness; start Mon, finish Day 10 → In progress → Coaching deck + staffing ticket.
Expectation set in the note: “Burndown will normalize within two weeks; red should drop to yellow by next cut and to green within one month.”
2) Vendor data cut missing results
What happened: Week‑22 lab cut arrived without sodium results for Sites F–H.
Why: Vendor transformation step skipped a column after a schema change.
Action plan:
- Owner: Lab Vendor PM
- Action: Re‑deliver corrected cut; add validation to prevent column drops.
- Due date: Wednesday 17:00 local
- Status: Scheduled
- Evidence link: Vendor Ticket #123 + corrected file location
Dashboard note
Lab Vendor PM → Re‑deliver Week‑22 cut including missing sodium; add transform check → Due Wed 17:00 → Scheduled → Ticket #123.
Change log entry (same week)
Date: 2025‑04‑10
Item: “Lab Cleanliness” KPI
Change: Exclude Week‑22 sodium from trendline; annotate chart
Reason: Vendor omission; backfill scheduled
Approval: Study DM Lead
Impact: Prevents false red signal this week
3) Adjusting a threshold responsibly
What happened: Startup surge creates predictable backlogs; red threshold for Query Aging is too sensitive.
Decision: Temporarily adjust red from 10% to 12% for 8 weeks.
Governance:
- Recorded in change log with date, reason, approver, review date.
- Visible on the dashboard (“Threshold history” tooltip).
- Revert plan documented: if trend stays below 8% for two consecutive cuts, revert early.
Why this matters: Users can see the rule change, understand the rationale, and know when it will be revisited—trust preserved.
Pitfalls & misconceptions
- Anti‑pattern: No change log. Definitions drift, memories differ, and trust erodes.
- Anti‑pattern: Notes without owners or dates. They become commentary, not commitments.
- Anti‑pattern: Over‑stuffed dashboards. More charts do not mean more clarity; highlight the few that matter.
- Misconception: Status meetings should review everything. CDM status should emphasize currency/cleanliness and actions, not safety narratives (SAEs), enrollment forecasts, or staffing rosters.
- Misconception: Thresholds are “set and forget.” Startup, mid‑study, and lock phases may need different, documented thresholds—with end dates.
Practice checklist
- Pair each yellow/red KPI with one concise action note.
- Put a Metric Glossary and Change Log on the dashboard, not in a hidden file.
- Version every release with data cut time, dictionary versions, and derivation versions.
- Snapshot dashboards weekly (PDF or image) for auditability and rollback.
- Use consistent terms (currency, cleanliness, expectedness, aging, reconciliation) across studies.
Key points to remember
- Dashboards drive action when every yellow/red KPI has a one‑sentence note with an owner and due date. Use: Owner → Action → Due date → Status → Evidence link.
- Write a note whenever a KPI changes status, a trend stalls/contradicts the last expectation, or a dependency (e.g., vendor cut, protocol change, site coaching) affects progress.
- Keep a visible change log to prevent metric drift—capture at minimum: date, item, change, reason, approver, and review/end date.
- Version each dashboard release with the data cut timestamp, dictionary versions (e.g., MedDRA/WHO Drug), derivation version, and known limitations for that cut.
- Use consistent, plain‑language definitions and vocabulary across studies—currency, cleanliness, expectedness, aging, reconciliation—to avoid re‑interpretation.
- In CDM status, prioritize currency/cleanliness and next actions; route non‑CDM content (e.g., safety events, enrollment, staffing) to their respective forums.