To evaluate dissolution from UV data, convert each absorbance into concentration with your calibration, multiply by the medium volume actually in the vessel at that time, add back the drug removed with earlier samples, and divide by the label claim. Then apply the USP <711> acceptance stages unit by unit, not on the mean. Most evaluation errors that hold up release come from the two steps in between: a missing volume correction and a stage decision taken on averaged data.
This walkthrough uses one constructed dataset from start to finish, so every number can be checked by hand. The companion article on UV dissolution testing covers method validation, f2 and sink conditions; here the focus is the arithmetic of a single release run.
Step 1: from absorbance to concentration
The example is a 25 mg immediate-release tablet in 900 mL medium, USP apparatus 2. The calibration from five standards gives:
- slope 0.0298 AU per µg/mL
- intercept 0.0010 AU
Concentration at each timepoint: c = (A − intercept) / slope. At 100 % release the vessel would hold 25 mg in 900 mL, which is 27.8 µg/mL or about 0.83 AU. That sits comfortably inside the linear range; a 200 mg tablet in the same set-up would need a dilution step, and the dilution factor then belongs in every line of the calculation.
Step 2: correct for the samples you took out
Every sample removes medium and the drug dissolved in it. If the medium is not replaced, the volume in the vessel shrinks with each sample; if it is replaced with fresh medium, the volume stays constant but the vessel is diluted. Either way, the simple formula c × 900 mL / dose is wrong after the first sample.
For a vessel without replacement, with sample volume v and n samples taken before the current one:
amount dissolved(n) = c(n) × (V − n × v) + v × [c(1) + c(2) + ... + c(n−1)]
The first term is what is in the vessel now; the second is what left with earlier samples. With replacement, the volume term stays V and the second term is the same sum.
A worked example with six timepoints
Mean absorbances for the run, 10 mL samples, no replacement:
| Time (min) | Absorbance (AU) | c (µg/mL) | Uncorrected (% LC) | Corrected (% LC) |
|---|---|---|---|---|
| 5 | 0.312 | 10.44 | 37.6 | 37.6 |
| 10 | 0.505 | 16.91 | 60.9 | 60.6 |
| 15 | 0.641 | 21.48 | 77.3 | 76.7 |
| 30 | 0.722 | 24.20 | 87.1 | 86.2 |
| 45 | 0.748 | 25.07 | 90.2 | 89.2 |
| 60 | 0.756 | 25.34 | 91.2 | 90.1 |
The uncorrected column uses 900 mL throughout and overstates release by 0.9 percentage points at 30 minutes. That sounds small, until the specification is Q = 80 % at 30 minutes and a single unit sits close to the stage limit, which is where the next step comes in.
Step 3: apply the acceptance stages per unit
USP <711> and Ph. Eur. 2.9.3 judge immediate-release products unit by unit in up to three stages:
| Stage | Units tested | Acceptance |
|---|---|---|
| S1 | 6 | each unit not less than Q + 5 % |
| S2 | 6 more (12) | mean of 12 at least Q, no unit less than Q − 15 % |
| S3 | 12 more (24) | mean of 24 at least Q, not more than 2 units less than Q − 15 %, no unit less than Q − 25 % |
The six individual vessels at 30 minutes, corrected: 86, 88, 84, 89, 85 and 85 % of label claim. Their mean is 86.2 %, well above Q = 80 %. But S1 asks for every unit at 85 % or more, and vessel 3 is at 84 %. The run does not pass S1 and goes to S2 with six more units.
Two things follow. First, a mean-based glance at this run would have released the batch. Second, the uncorrected calculation would have shown vessel 3 at 84.9 %, which rounds to 85 % against a limit stated in whole percent, and passed S1 on 0.9 percentage points that do not exist. Both errors are invisible in the summary and visible in an audit.
Checks before you sign the evaluation
- Calibration in range. Every sample absorbance between the lowest and highest standard; extrapolation above the top standard is a common source of low early timepoints after dilution changes.
- Blank and placebo. A placebo vessel absorbance subtracted or shown to be negligible; otherwise excipient absorbance inflates the early points.
- Variability. Coefficient of variation across units no more than 20 % at the early points and 10 % thereafter, if the profile is to be used for comparison.
- Plausibility. Released amount should not decrease between timepoints by more than the analytical precision; a drop usually means filter adsorption, degradation in the medium or a transcription error.
- Units and dilution. Dilution factor, sample volume and label claim taken from the method and batch record, not retyped.
Limits of the statement
- The volume correction assumes the sample is representative of the vessel at that moment. Poor sampling position or a filter that adsorbs drug breaks that assumption, and no calculation fixes it.
- UV gives the absorbance of everything in the cuvette at that wavelength. Selectivity has to be shown during validation; the release evaluation inherits it.
- The stage logic above is for immediate-release products. Extended-release and delayed-release products have their own acceptance tables.
Frequently asked questions
How do you calculate percent dissolved from absorbance?
Convert absorbance into concentration with the calibration line, multiply by the medium volume in the vessel at that time, add the drug removed with earlier samples, and divide by the label claim. Include any dilution factor. In the example above, 0.722 AU at 30 minutes corresponds to 86.2 % of label claim after volume correction.
Do I need a volume correction if I replace the medium?
Yes. Replacing the medium keeps the volume constant but removes drug with every sample and dilutes the vessel. The drug removed with earlier samples still has to be added back, otherwise later timepoints are understated.
Why did the run fail S1 although the mean was 86 %?
S1 is a test on individual units: each of the six must reach Q + 5 %. One vessel at 84 % with Q = 80 % fails S1 regardless of the mean. The run continues to S2 with six more units.
What is the difference between S2 and S3 in USP <711>?
S2 tests 12 units in total and requires a mean of at least Q with no unit below Q − 15 %. S3 tests 24 units and allows up to two units below Q − 15 % but none below Q − 25 %, again with a mean of at least Q.
Can I use the mean profile for f2 comparison?
Yes, f2 is calculated on mean profiles, but only if unit-to-unit variability stays within the limits of the guidance (coefficient of variation up to 20 % early, 10 % later). The individual values are still required for the release decision.