Method Validation Parameters
On this page
Direct answer
Each validation parameter is a defined experiment with an acceptance expectation. Accuracy is recovery of added analyte — spiking placebo at 80, 100 and 120 per cent of the test concentration, with assay recoveries conventionally 98-102 per cent. Precision splits into repeatability (six replicate preparations at 100 per cent, reported as per cent RSD), intermediate precision (different day, analyst, instrument within one lab) and reproducibility (between laboratories), each attacking a different source of variability. Specificity shows the analyte is measured free of interference — placebo, degradants, process impurities — by peak purity and separation studies. Linearity spans five-plus levels across the claimed range (commonly 80-120 per cent for assay, quantitation limit to 150 per cent of specification for impurities), judged by regression statistics, not a high correlation coefficient alone. Detection and quantitation limits follow DL = 3.3σ/S and QL = 10σ/S; robustness deliberately varies method factors — mobile phase pH by about 0.2 units, organic composition, flow, column temperature, detection wavelength — to prove small changes do not break the method.
What you must remember
- Accuracy protocol: placebo spiking at 80/100/120 per cent; conventional acceptance 98-102 per cent for assay (about 80-120 per cent for trace impurities).
- Precision hierarchy: repeatability (six replicates, one analyst, one day), intermediate precision (day-analyst-instrument varied), reproducibility (inter-laboratory) — per cent RSD is the universal metric.
- Linearity evidence: five levels minimum bracketing the range; report slope, intercept, residuals and r — a significant intercept or curved residuals disqualify a high r.
- Specificity battery: blank and placebo injection, stressed samples, peak purity by PDA or mass, resolution from the nearest impurity.
- DL/QL practice: 3.3σ/S and 10σ/S; at QL, precision of about ten replicate injections is commonly demonstrated with RSD near or below 10 per cent.
- Robustness factors: pH ±0.2, organic phase composition ±2 per cent absolute, flow ±0.1 mL/min or 10 per cent, column temperature ±5 °C, wavelength ±2 nm, plus column lots.
- Range: the interval between upper and lower concentrations validated as accurate, precise and linear — claimed no wider than the data.
- System suitability: the daily sentinel (replicate RSD commonly ≤2 per cent, tailing and resolution limits) that inherits robustness conclusions into routine sequences.
Where students slip
Accuracy and precision are interchangeable in weak answers; the examiner's diagram is the archery target — truth of the mean versus scatter of shots — and a method can be precise and inaccurate, recovering 94 per cent every time. Second, repeatability versus intermediate precision is the most confused pair: same-day replicates versus day-analyst-instrument variation. Third, linearity by r alone: 0.999 over a narrow range proves little, and the standard answer demands residual inspection and a justified intercept. Fourth, robustness is a designed experiment with named factors and pre-defined criteria — name the factors or lose the mark. Finally, QL is a claimed quantity: precision at the QL must be demonstrated, since a limit that cannot be measured precisely cannot report trace impurities.
Frequently asked questions
How is accuracy experimentally demonstrated?
Recovery studies — spiking placebo with known analyte at 80, 100 and 120 per cent and comparing found versus added; assay recoveries conventionally 98-102 per cent.
What distinguishes repeatability from intermediate precision?
Repeatability is replicate preparations under identical conditions (commonly six samples, one analyst, one day); intermediate precision varies day, analyst and instrument within the same laboratory.
Why is a high correlation coefficient alone insufficient for linearity?
r hides curvature and non-zero intercepts; validation demands residual plots, slope and intercept assessment across at least five levels bracketing the claimed range.
Which factors are varied in a robustness study?
Mobile phase pH (±0.2), organic composition, flow rate (±10 per cent), column temperature (±5 °C), wavelength (±2 nm) and column lots — with pre-set windows.
What must be demonstrated at the quantitation limit?
Beyond QL = 10σ/S arithmetic, the method must measure reliably at that level — signal-to-noise near 10 and replicate precision commonly within about 10 per cent RSD.