Stability Studies — ICH Q1E

Shelf-Life Estimation Calculator

Fit a linear regression to stability time-point data and extrapolate to the point where the trend (and its one-sided confidence bound) crosses your specification limit.

The ICH Q1E Approach

When a quality attribute shows a trend over time (e.g., assay slowly declining), ICH Q1E recommends fitting a linear regression to the stability data and using the intersection of the 95% one-sided confidence limit with the acceptance criterion to estimate shelf life — not simply the mean trend line. This builds statistical conservatism into the estimate, since the mean line alone could underestimate risk of early failure.

Linear regression: y = a + b·t Shelf life (mean line) = time where mean trend crosses specification limit Shelf life (95% CI) = time where the one-sided confidence bound crosses the limit → The more conservative (earlier) of these is normally used as the proposed shelf life

This tool fits an ordinary least-squares regression to your data and estimates both the mean-line crossing point and an approximate one-sided 95% confidence bound crossing point.

Interactive Calculator

Stability Data (time in months, result — one pair per line) 0, 100.2 3, 99.6 6, 99.1 9, 98.5 12, 98.0 18, 96.9 24, 95.8

Format: time, result — comma-separated, one point per line

Specification Limit

Limit Direction Lower limit (result must stay above) Upper limit (result must stay below)

Slope

—

Intercept

—

R²

—

Shelf Life (Mean Line)

—

Shelf Life (95% Bound, Approx.)

—

—

Notes on Interpretation

Situation	Consideration
No significant trend (slope ≈ 0, low R²)	Poolability/batch-to-batch analysis and a mean-based shelf life approach may be more appropriate than extrapolation
95% bound shelf life much shorter than mean line	High data variability — consider additional time points or tighter process control
Multiple batches available	ICH Q1E requires assessing poolability across batches before combining data for a single regression

This calculator uses a simplified approximation of the ICH Q1E one-sided confidence bound (using a fixed z-value rather than the exact t-distribution and full regression confidence interval formula) and is intended for educational/preliminary estimation only. Formal shelf-life claims must follow a validated statistical methodology, proper batch poolability testing, and regulatory review.