Excursion-to-Disposition Decision Accuracy (EDDA)

GMI-PH-024Pharmaceutical OperationsResearch-Forward

Definition

Percentage of temperature excursion dispositions later confirmed correct: released product testing within specification on retained samples, and destroyed product that would in fact have degraded under the actual excursion profile.

The Measurement Gap

A cold chain monitoring system detects that 847 units of a temperature-sensitive biologic were exposed to 9.2°C for 23 minutes: above the 2–8°C specification. The stability data available was generated under ICH-compliant accelerated conditions that do not match the actual excursion profile. The QA team extrapolates, and the product is released with documentation.

Three months later, a similar excursion (10.1°C for 18 minutes) affects 1,200 units of the same product. A different reviewer destroys them: $340K in product loss. Both decisions were defensible. Neither can be validated because nobody tracks whether excursion dispositions turn out to be right.

The result: inconsistent decisions, unquantified safety risk from incorrect releases, and unquantified financial waste from unnecessary destructions.

Decision enabled:

When EDDA falls below threshold, the organization develops product-specific stability data at common excursion profiles to replace extrapolation with evidence. Patterns of incorrect-release trigger immediate CAPA. Patterns of incorrect-destruction trigger revised disposition criteria to reduce waste.

Formula

EDDA = (Σ Correct_Dispositions / Σ Total_Excursion_Dispositions) × 100

A disposition is correct if: released product subsequently confirmed within specification through retained sample testing, or destroyed product would have degraded based on stability modeling at actual excursion conditions.

Worked Example:

Over 24 months, 31 excursion events required disposition decisions:

18 units released: 16 subsequently confirmed within spec through retained sample testing at expiry. 2 showed quality degradation at later time points. These were incorrect releases (safety risk).

13 units destroyed: stability modeling at excursion conditions suggests 9 would have been acceptable (unnecessary destruction worth $1.2M total). 4 would likely have degraded (destruction was appropriate).

Correct dispositions: 16 (correctly released) + 4 (correctly destroyed) = 20.
Incorrect: 2 (incorrectly released) + 9 (incorrectly destroyed) = 11.

EDDA = 20/31 × 100 = 64.5%

Below the 85% threshold, with risk on both sides: 2 cases of product that should not have been released and 9 cases of unnecessary product destruction.

Sign Convention:

Higher is better. Always stratify by error type: incorrect-release is a safety concern; incorrect-destruction is a financial concern.

Proposed formula, pending validation pilots.

At a Glance

Unit
Percentage
Trend Direction
Higher
Reporting Frequency
Per excursion event; quarterly decision-accuracy review
Category
Disposition Decision Quality
← Patient Matching Homogenization Score (PMHS)Sub-Threshold Event Response Consistency (STERC) →

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