KPI Pack: Measures, Data Mappings & Run‑Chart Templates for Readmissions Reduction

Purpose

This KPI pack gives frontline teams a concise, practical, and consistent way to define, measure, visualize, and interpret readmissions so improvement work can be evaluated reliably. It supports a 90‑day improvement journey by standardizing definitions, data sources, calculation examples, run‑chart practices, and suggested leading indicators.

1. Standard readmission definition and exclusions

Primary measure (unplanned readmissions within 30 or 90 days): The proportion of index discharges from the target population that result in an unplanned inpatient readmission to any facility within X days (30 or 90) of the index discharge date.

• Index discharge: A completed inpatient discharge (not observation) from the target service/ward during the reporting period.
• Unplanned readmission: Any subsequent inpatient admission within X days whose admission type or documented intent is emergency/unplanned, or that meets unplanned admission diagnosis rules. Exclude planned admissions per a pre‑specified planned‑admission list (e.g., elective procedures, scheduled chemotherapy, planned dialysis access revision).
• Observation stays and ED visits: Decide up front whether to include ED revisits and observation stays. If included, measure and report them separately as composite events.
• Transfers: Transfers between acute facilities within 24 hours should be treated as part of the index stay, not a readmission.
• Exclusions: hospice patients at index discharge, patients who die before readmission window ends, and admissions for unrelated scheduled care defined by your planned‑admission list.

2. Data source mapping

Map your measure consistently across systems. At minimum capture:

• Patient identifier (MRN) – unique across settings where possible
• Index admission start and discharge datetime
• Admission type / admission source (elective vs emergency)
• Encounter disposition, transfer flags
• Readmission admission datetime, facility code
• Planned admission flag or procedure codes indicating planned care
• Death date (to apply exclusion if patient died within the window)

Typical EHR fields: encounter_id / visit_id, encounter_class (inpatient/observation), admit_datetime, discharge_datetime, disposition_code, scheduled_flag, primary_diagnosis_code. Claims fields: claim_id, admission_date, discharge_date, admission_type, diagnosis/procedure codes, claim_payment_indicator.

3. KPI calculation examples

Use clear numerator/denominator language and calculate consistently.

1. Primary KPI (30‑day unplanned readmission rate):
   • Numerator: number of index discharges with an unplanned inpatient readmission within 30 days.
   • Denominator: number of eligible index discharges during the measurement period (after exclusions).
   • Calculation: (Numerator / Denominator) × 100 = readmission rate (%)
2. Alternative: readmission count per 1,000 discharges useful for small volumes: (Numerator / Denominator) × 1000.

Example: If 12 of 400 eligible discharges are readmitted within 30 days, rate = (12/400)*100 = 3.0%.

4. Denominator & risk‑adjustment notes

• Keep the denominator consistent with your improvement scope (unit, service line, condition cohort, or whole hospital).
• For fair comparisons over time, stratify by key risk factors rather than attempting full statistical risk adjustment in early improvement cycles. Useful strata: age group, primary diagnosis or condition cohort (e.g., CHF, COPD), presence of complex comorbidity, payer type for social‑risk scanning.
• Document any exclusions and changes to the definition before comparing periods.

5. Run chart template & interpretation tips

Use run charts (time series of points) to show performance across the 90‑day journey. Weekly or biweekly points are common; choose cadence that balances signal with sample size.

1. Plot the KPI value on the Y axis and time periods (weeks) on the X axis.
2. Include a center line (median of baseline period) and annotate interventions (PDSA cycles) on the chart.
3. Use simple rules to detect non‑random change (special cause):
   • Shift: six or more consecutive points all above or below the median.
   • Run: too few or too many runs than expected for the number of points (use run chart tables or simple run rules).
   • Trend: five or more consecutive points all increasing or decreasing.
4. Interpretation tips: look for consistent shifts after an intervention. Short temporary drops that don’t persist likely reflect chance; investigate sudden jumps for data or process changes.

6. Suggested leading indicators (definitions & suggested collection)

• Medication reconciliation completed before discharge: percent of index discharges with completed med reconciliation documented.
• Follow‑up appointment scheduled on discharge: percent with appointment within 7 days documented prior to discharge.
• Follow‑up completed: percent of patients contacted or seen within 7–14 days (dependent on cohort).
• Social needs escalated: percent with documented social needs screen and active referral (transportation, food, meds access) when needed.
• Home‑service or transitional care referral completed: percent where home nursing or transitional program accepted the referral within 48 hours.

Collect leading indicators weekly if possible. They give faster feedback than readmission endpoints.

7. Suggested targets and baseline cadence

Suggested approach:

1. Run a 4–8 week baseline at chosen cadence (weekly or biweekly) to establish the median.
2. Set an achievable 90‑day target: for example a 10–30% relative reduction from baseline readmission rate, depending on baseline variation and clinical context. Smaller absolute improvements may be significant for low baseline rates.
3. Set process targets for leading indicators (e.g., med reconciliation ≥ 95%, follow‑up scheduled ≥ 80%) to support the outcome target.

8. Reporting & governance

• Include both outcome KPI and selected leading indicators in weekly improvement huddles.
• Annotate run charts with intervention dates and ownership for each change tested.
• Agree on escalation rules when readmission rate rises above pre‑agreed control limits or when leading indicators fall below thresholds.

9. Quick start checklist for a 90‑day team

1. Agree scope (service line, cohort) and readmission window (30 vs 90 days).
2. Extract 4–8 weeks of baseline data using the standard definition and create a run chart.
3. Select 2–4 leading indicators to track weekly and set process targets.
4. Plan 1–2 rapid pilots (PDSA cycles) focused on highest‑impact failure modes (meds, follow‑up, social needs).
5. Hold weekly huddles, update run charts, and document learnings.

10. Common pitfalls

• Changing the measure definition mid‑project without documenting it.
• Using small weekly denominators that create noisy signals—consider biweekly aggregation for small volumes.
• Relying only on outcome data without tracking leading indicators.

Appendix: Suggested fields for data export (CSV)

Include columns: patient_mrn, index_admit_date, index_discharge_date, index_service, planned_flag, readmit_flag_30d, readmit_date, readmit_facility, readmit_type, death_date, medrec_completed_flag, followup_scheduled_flag, social_needs_flag.

Use this pack as the standard measurement spine for your 90‑day readmission improvement journey. Keep definitions, exclusions, and data mappings documented and versioned so comparisons over time are meaningful.