01 / Start with a decision
A dashboard should lead somewhere.
Choose a question before a chart: What is waiting? Where is coverage missing? Which device needs attention? Give each measure an owner and a decision it supports. A larger collection of numbers is not necessarily a better operating report.
Keep a small set of process, outcome and balancing measures. For example, a faster intake process should be considered alongside correction work and staff burden. This follows the general measurement approach described by the Institute for Healthcare Improvement; the EEG examples below are Axiom's proposed operational definitions, not endorsed national measures.
02 / A working metric dictionary
Six useful measures. Defined before compared.
Open a measure for its formula, eligible population, example and interpretation. All numbers are illustrative. Adopt definitions through your facility's governance process; these are not a claim that every measure is already available as a product report.
01Intake review timeWhere do incoming orders wait? · Median hours
Calculation
Median of (validation timestamp − receipt timestamp)
- Include
- Orders validated during the reporting window with both timestamps present and an end at or after the start.
- Worked example
- Five eligible orders take 1, 2, 3, 4 and 20 hours. The median is 3 hours; the mean is 6 hours. Show the count alongside either result.
- Evidence needed
- Receipt and human-validation events. Name the receipt event precisely: PDF upload, fax receipt and the date written on an order are different events.
- Interpret carefully
- This completed-order cohort omits work still waiting. Pair it with an open-queue count and oldest waiting age. Report missing or invalid timestamp counts separately.
- Next action
- Separate intake channels and priority groups before deciding whether the delay is staffing, information completeness or a handoff.
02First-review completenessHow much intake arrives ready for review? · Percentage of reviewed orders
Calculation
Orders with no required-information gaps at first review ÷ all orders first reviewed × 100
- Include
- Orders first reviewed in the period. Use the same version of the required-field definition for every order being compared.
- Worked example
- Of 40 first reviews, 30 have no required-information gaps: 30 ÷ 40 × 100 = 75%. This describes completeness, not clinical appropriateness.
- Evidence needed
- A recorded first-review result and the required-field rule in effect at that time. A current corrected record cannot reliably reconstruct its original state.
- Interpret carefully
- Do not count later corrections as first-pass success. If original review evidence is unavailable, show unavailable rather than infer a rate.
- Next action
- Group gaps by field and source to identify an order-form or intake-instruction improvement.
03Reader turnaroundHow long does interpretation follow-up take? · Median hours
Calculation
Median of (read completion timestamp − sent-to-reader timestamp)
- Include
- Reads completed during the period with valid start and end events. State whether the endpoint is signed interpretation or another recorded completion event.
- Worked example
- Completed-read intervals of 2, 4, 6, 8 and 30 hours have a 6-hour median. The longest interval remains visible rather than disappearing into the average.
- Evidence needed
- Reader-routing and completion events. Keep this distinct from recording-end-to-report time, which starts at a different handoff.
- Interpret carefully
- Completed-only results omit the current backlog. Include pending reads and oldest pending age; distinguish routine and urgent work. No universal clinical target is proposed here.
- Next action
- Review the handoff and coverage pattern with the accountable clinical leader before setting local response expectations.
04Equipment readinessHow many devices can actually be used now? · Snapshot count and percentage
Calculation
Ready devices ÷ active, non-retired devices in the defined pool × 100
- Include
- A named device pool at one stated timestamp. Readiness must follow the facility's identity, cleaning, inspection and custody rules.
- Worked example
- Three ready devices in an active pool of five gives 60% readiness at that moment. Two devices in use may explain the remainder without indicating a failure.
- Evidence needed
- Device state, required identity and readiness evidence. Report in-use, cleaning, maintenance and setup-needed counts alongside ready devices.
- Interpret carefully
- A snapshot is not a utilization rate. Utilization requires observed time intervals and a defined denominator of available device-hours.
- Next action
- Check tomorrow's planned demand against likely returns and maintenance constraints; do not treat all unavailable devices as overdue.
05Qualification-aware coverageWhich required staffing slots have eligible cover? · Covered slots / required slots
Calculation
Required slots filled by eligible staff ÷ all required slots × 100
- Include
- The same unit, time window and version of the coverage requirements. Check qualifications and availability at the scheduled time.
- Worked example
- Eleven eligible filled slots out of twelve required slots gives 91.7%. One slot is still uncovered; the percentage should not hide its location or timing.
- Evidence needed
- Published requirements, scheduled assignments, effective qualifications and availability. Resolve overlapping assignments before treating someone as covering two places.
- Interpret carefully
- Slots of different duration are not equivalent. If using hours, label the measure covered required hours and use hours in both numerator and denominator. No requirements means not applicable, not 100%.
- Next action
- Open the specific uncovered slot. A coverage rate supports planning; it does not certify safe staffing or authorize an unqualified assignment.
06Overdue operational workWhich open tasks have passed their due time? · Snapshot count
Calculation
Count open tasks where snapshot time is later than the recorded due time
- Include
- Open tasks of one named type at a stated timestamp. Keep disconnects, equipment returns and administrative follow-up distinguishable.
- Worked example
- At 10:00, open tasks due at 09:00 and 09:30 are overdue; one due at 10:00 is due now under this definition. Overdue count: 2.
- Evidence needed
- Task identity, status, due timestamp and facility timezone. Avoid counting the same task once for each alert or reminder.
- Interpret carefully
- A count mixes neither elapsed hours nor completed events. List tasks without a due time separately so they do not disappear as apparently on-time work.
- Next action
- Show the owner, location and next action beside the oldest items. A useful exception report helps someone resolve work.
03 / Make the report trustworthy
Show the denominator. Keep the gaps visible.
- Use one cohort. Do not divide completions this month by orders received this month and call it a completion rate. They can be different populations.
- Name the clock. Publish the facility timezone, start and end events, and whether the measure uses elapsed or business hours. Do not mix them.
- Show missing data. Zero eligible observations means “not available” or “not applicable,” not zero delay or perfect performance. Display the number excluded and why.
- Compare like with like. Separate study type, urgency and setting where the work differs. A routine outpatient study and continuous monitoring are not interchangeable units of effort.
- Keep definitions versioned. Record the rule, source, owner, refresh time and changes. A rate moving after a rule change may not represent operational improvement.
AHRQ's measurement resources emphasize defining numerators and denominators when selecting measures. Use that discipline for local operational reporting without presenting these examples as AHRQ quality indicators. Read AHRQ's measurement resource.
04 / Put the evidence to work
Daily action. Weekly learning. Monthly decisions.
At the daily huddle, review open work and assign the next action. Each week, examine a consistent trend with sample sizes and exceptions. Each month, review whether the chosen change helped and whether it created more rework or burden elsewhere. These are suggested review rhythms, not clinical response deadlines.
The EHR owns the clinical record. The acquisition system owns the waveform. Axiom owns the operation. Operational measures organize the work around care; they do not measure diagnostic accuracy or replace clinical judgment.
Axiom OrderFlow
Follow the work
Explore intake, operational handoffs and exception visibility.
Explore the workflow →Axiom Shift
Plan the coverage
Explore scheduling around qualifications and department requirements.
Explore scheduling →Sources and scope
This guide applies general improvement-measurement principles to EEG operations. IHI and AHRQ do not endorse Axiom or the six example definitions. No staffing minimums, clinical deadlines, benchmark targets or guaranteed savings are prescribed. Confirm source availability and local definitions before adopting a measure.