Measure & KPI Pack: Definitions, Calculation Notes, Owners, and Use Guidance

Measure & KPI Pack — Practical Definitions and Use Guidance

Purpose: Provide a single, consistent reference so teams measure the same thing the same way. Each metric entry below follows a clear template you can copy into local measurement plans, dashboards, audits, or improvement charters.

How to use this pack

• Adopt the metric template as the canonical definition for your team or unit; change only with documented governance and version history.
• Keep definitions short but precise: numerator, denominator, time window, exclusions, data source, owner, cadence.
• Use the provided pseudocode/SQL as starting points for data engineering and dashboarding.
• Combine measures: pair lagging safety outcomes with leading process indicators (guidance below).

Standard metric template

Every metric entry here follows this structure so it is easy to compare and reuse:

• Metric name
• Definition — short, plain-language description
• Numerator — exact inclusion rules
• Denominator — population / exposure measure
• Time window — how events are dated and which period they count toward
• Exclusions / Notes — important edge cases and coding guidance
• Suggested data source(s)
• Owner — recommended role for stewardship
• Cadence — reporting / review frequency
• SQL / pseudocode — example extraction or calculation
• Visualization suggestions — charts, filters, and comparisons
• Recommended use — how teams should act on the metric
• Cautions & anti-gaming notes

Core metric examples

1. 30-day Readmission Rate (All-Cause, Inpatient)

Definition: Percent of patients discharged from an inpatient admission who are readmitted for any cause within 30 days.

Numerator: Count of index discharges where the patient had an unplanned inpatient readmission within 30 days of discharge date.

Denominator: Count of index inpatient discharges (exclude transfers to another acute facility when specified by policy).

Time window: 30 days from discharge date; attribute readmission to the original discharge month for monthly reporting.

Exclusions / Notes: Planned readmissions (e.g., scheduled chemo, planned procedures) excluded per list of CPT/DRG codes; deaths within 30 days are still counted as non-readmitted.

Suggested data sources: EHR admission/discharge data, case management system.

Owner: Quality & case management.

Cadence: Monthly, with quarterly trending and quarterly case-review meetings for high-volume conditions.

SQL/pseudocode:

-- identify index discharges
SELECT patient_id, discharge_date, admission_id
FROM admissions
WHERE discharge_type = 'inpatient'
  AND NOT transfer_to_acute
-- identify readmissions within 30 days
SELECT a.admission_id AS index_adm, count(r.admission_id) AS readmit_count
FROM index_adm a
LEFT JOIN admissions r ON r.patient_id = a.patient_id
  AND r.admit_date > a.discharge_date
  AND r.admit_date <= a.discharge_date + INTERVAL '30 days'
  AND r.admission_type = 'unplanned'
GROUP BY a.admission_id;
  

Visualization: Monthly rate line with control limits and denominator volume beneath; cohort breakdown by service/condition.

Recommended use: Identify high-risk cohorts for transitional care interventions and monitor impact of discharge bundles.

Cautions: Case-mix differences and outpatient follow-up access affect rates. Avoid penalizing legitimate planned returns.

2. Hospital-Acquired Infection (HAI) Rate — CLABSI (per 1,000 central line days)

Definition: Count of central-line associated bloodstream infections per 1,000 central line days for the specified unit.

Numerator: Confirmed CLABSI events meeting NHSN/CDC criteria during the measurement period.

Denominator: Total central line days (sum of patients with an eligible central line each day).

Time window: Typically monthly.

Exclusions / Notes: Follow local infection prevention rules aligned with NHSN; include only eligible device types.

Suggested data sources: Infection prevention surveillance, device-tracking logs, nursing documentation.

Owner: Infection Prevention.

Cadence: Monthly reporting; review at infection prevention committee.

SQL/pseudocode:

SELECT SUM(cl_days) AS total_line_days,
       COUNT(clabsi_event_id) AS clabsi_count,
       (COUNT(clabsi_event_id) * 1000.0) / SUM(cl_days) AS rate_per_1000
FROM device_days dd
LEFT JOIN clabsi_events ce ON ce.unit = dd.unit AND ce.date BETWEEN dd.date AND dd.date
WHERE dd.unit = 'ICU-A' AND dd.date BETWEEN '2026-01-01' AND '2026-01-31';
  

Visualization: Bar chart of rate per 1,000 line-days by unit with denominator overlay; run chart for trend detection.

Cautions: Small denominators create volatile rates—use rolling averages or aggregate by quarter for small units.

3. Medication Errors per 1,000 Doses

Definition: Number of reported medication errors (all severities or defined severity threshold) per 1,000 medication doses dispensed/administered.

Numerator: Confirmed medication error reports during the period (specify severity levels to include).

Denominator: Total medication doses administered or dispensed (depending on what is measurable).

Time window: Monthly recommended.

Exclusions / Notes: Clarify whether near-misses are included; if voluntary reporting undercounts events, triangulate with audits.

Suggested data sources: Incident reporting system, pharmacy dispensing logs, BCMA (bar-code med administration).

Owner: Pharmacy & Safety.

Cadence: Monthly with monthly safety huddles to review incidents and improvement actions.

4. ED Boarding Hours

Definition: Sum of hours patients remain in the ED after decision-to-admit until transfer to an inpatient bed.

Numerator: For each admitted ED patient, boarding time = inpatient bed assignment or physical departure time - decision-to-admit time. Sum boarding hours for period.

Denominator: Number of admitted ED patients (or report as total boarding hours and median boarding time).

Suggested data sources: ED tracking system, admission orders, bed management system.

Owner: ED operations / patient flow.

Cadence: Daily operational dashboard + weekly tactical review.

5. Time-to-Antibiotics for Suspected Sepsis

Definition: Median or percent within target (e.g., 60 minutes) from sepsis recognition (time zero) to first appropriate antibiotic dose.

Numerator: Number of sepsis events where antibiotics were administered within target window.

Denominator: Eligible sepsis events (per clinical definition being used).

Notes: Define time zero consistently (e.g., first sepsis bundle trigger or time of triage if documented). Use clinical validation for denominators.

Owner: Sepsis program / quality.

6. Patient Satisfaction Index (Composite)

Definition: Composite score from patient-experience survey (e.g., percent top-box on overall rating and recommend-question).

Numerator: Count of respondents giving top-box answers across selected items.

Denominator: Total completed eligible surveys in the period.

Owner: Patient experience.

Cadence: Monthly or rolling 90-day to smooth sampling variability.

Guidance on balancing leading and lagging indicators

Pair outcome metrics (readmissions, HAIs) with process and structural indicators (timely follow-up calls, central-line insertion checklists, staff training completion). Leading measures help teams act earlier and avoid reactive cycles.

Governance, versioning, and tailoring

• Designate an owner for each metric who approves local adaptations, maintains a short rationale, and records the version/date.
• When tailoring, copy the canonical definition into a local measurement plan and document which fields changed and why.
• Maintain a change log so downstream dashboards, analytics, and contracts remain traceable.

Implementation checklist

1. Agree on canonical definitions and owners.
2. Implement SQL/pseudocode in a test view and validate against known cases (spot-check at least 20 events).
3. Choose visualizations and set denominators and smoothing rules.
4. Publish dashboard with definition hover-text and link to this reference.
5. Review metrics quarterly for relevance, threshold adjustments, and gaming risks.

Common pitfalls & anti-gaming

• Unclear denominators produce misleading comparisons — always show denominator volume.
• Avoid rewarding coding tricks — focus on clinical process improvements rather than raw numbers alone.
• Small sample sizes: use rolling averages or aggregate by longer periods for reliability.

Next steps & reuse

Copy metric templates into your local domain, assign owners, and consider implementing an interactive KPI-definition form (see capability notes) so units can create validated local copies that remain traceable to the canonical library.