Schedule Equity Distribution Index (SEDI)

GMI-HC-017HealthcarePilot-Ready

Definition

SEDI measures the statistical uniformity of undesirable shift assignments (nights, weekends, holidays) across a scheduling cohort. A perfectly equitable schedule produces a SEDI of 1.0; increasing disparity drives the index toward zero. The metric detects whether algorithmic scheduling (optimizing for coverage, cost, or preference satisfaction) systematically concentrates burden on specific individuals or demographic groups, creating inequity invisible to aggregate staffing reports.

The Measurement Gap

Measurement Gap and Decision

Decision Question: Does the scheduling algorithm distribute undesirable shifts equitably, or does it systematically burden specific staff members in ways that aggregate metrics cannot see?

What Goes Unmeasured Today:

Scheduling systems report coverage: shifts are filled, ratios are met, overtime is managed. What they do not report is distribution fairness. A unit can show perfect schedule coverage while one nurse works 40% nights and another works 5%, and the system treats both as equivalent filled shifts. When AI-assisted scheduling enters the picture, the problem intensifies. Algorithms optimize for stated objectives (minimize overtime, honor seniority preferences, reduce agency spend) and the optimization can systematically route undesirable shifts to staff with less seniority, fewer formal preference submissions, or lower perceived scheduling flexibility. The disparity compounds over months. Staff who receive disproportionate undesirable shifts experience higher burnout, increased turnover intent, and degraded clinical performance during those shifts. But because no metric captures distribution equity, the problem manifests only as exit interviews citing scheduling dissatisfaction: long after the damage is done.

Invisible Failure Mode:

Algorithmic scheduling bias hides behind aggregate compliance. The unit meets every staffing target while specific individuals absorb inequitable burden. Without measuring distribution uniformity, organizations cannot distinguish between equitable scheduling and systematically unfair scheduling that happens to produce compliant aggregate numbers.

Formula

Primary Formula:

SEDI = 1 − CV(Undesirable_Shift_Pct)

Where CV = coefficient of variation across all staff members in the scheduling cohort.

Variable Definitions:

Variable Definition Source
Undesirable_Shift_Pct_i Percentage of staff member i's total shifts classified as undesirable (nights, weekends, holidays) Scheduling system
CV Standard deviation of Undesirable_Shift_Pct across all staff divided by the mean Undesirable_Shift_Pct Calculated
Scheduling cohort All staff eligible for the same shift pool within a unit or department Scheduling system

Data Requirements:

Shift assignment records with shift type classification (day, evening, night, weekend, holiday) from scheduling system. Staff roster with eligibility flags for each shift type. Minimum 90 days of scheduling data for stable measurement.

Edge Cases and Limitations:

Staff with voluntary night-shift preferences will appear to carry disproportionate burden even when the assignment is desired. Flag voluntary preferences and report SEDI with and without voluntary assignments. Part-time staff with restricted availability will naturally receive different distributions; calculate SEDI within FTE-equivalent cohorts. Very small units (fewer than 8 staff) produce unstable CV calculations.

Proposed formula, pending validation pilots.

At a Glance

Category
AI Governance, Equity, and Workforce Impact
← Regulatory Event Classification Accuracy (RECA-HC)Risk Model Discrimination Currency (RMDC) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment