This is a ready-to-use calculation sheet for one run of a validated relative-potency bioassay or ELISA. It reproduces the parallelism equivalence test and the relative-potency arithmetic step by step, in the order the method actually requires them (parallelism first, potency only if parallelism passes), so a reviewer can check the logic and the numbers without redoing the statistics from scratch. Replace every <<FILL: ...>> placeholder. The model, weighting, parallelism metric, and equivalence bounds used here must already be fixed in the validated method; this worksheet applies them, it does not set them. This content is educational reference, not statistical consulting advice; have a qualified statistician confirm the method and any assumptions for your actual data.
Header
| Field | Entry |
|---|---|
| Method / protocol reference | <<FILL: validated method or protocol number, version>> |
| Run / plate ID | <<FILL>> |
| Product / sample ID and lot | <<FILL>> |
| Reference standard ID and lot | <<FILL>> |
| Curve model locked in the method | <<FILL: 4PL or 5PL>> |
| Weighting scheme locked in the method | <<FILL: e.g. 1/Y^2>> |
| Parallelism metric and equivalence bounds (from validation) | <<FILL: e.g. slope ratio, bounds 0.80 to 1.25>> |
| Completed by, date | <<FILL>> |
Part A: System suitability (complete before any potency or parallelism calculation)
| Parameter | Validated limit | Observed value | Pass / Fail |
|---|---|---|---|
| Reference EC50 within validated window | <<FILL>> | <<FILL>> | <<FILL>> |
| Signal window (D minus A) | <<FILL: minimum>> | <<FILL>> | <<FILL>> |
| Reference-vs-reference parallelism (where applicable) | <<FILL>> | <<FILL>> | <<FILL>> |
| Edge / positional controls within limits | <<FILL>> | <<FILL>> | <<FILL>> |
If any row fails, stop here. The run is invalid; do not proceed to Part B. Record the failure and route it per the governing SOP.
Part B: Fitted curve parameters
Report the parameters exactly as the locked model produces them. Do not drop or adjust a parameter to improve the apparent fit.
| Parameter | Reference standard | Test sample |
|---|---|---|
| A, lower asymptote | <<FILL>> | <<FILL>> |
| D, upper asymptote | <<FILL>> | <<FILL>> |
| B, slope (Hill slope) | <<FILL>> | <<FILL>> |
| C, EC50 | <<FILL>> | <<FILL>> |
| Fit diagnostic (residual pattern, random or structured) | <<FILL>> | <<FILL>> |
Part C: Parallelism, the equivalence test
Parallelism is evaluated first. A relative-potency result computed before parallelism passes is not a result; it is a number with no defined meaning.
| Field | Entry |
|---|---|
| Parallelism metric used | <<FILL: e.g. slope ratio (sample B / reference B); or a combined similarity score>> |
| Observed metric value | <<FILL>> |
| Confidence interval on the metric (typically 90%) | <<FILL: lower>> to <<FILL: upper>> |
| Pre-defined equivalence bounds (from the validated method, not chosen now) | <<FILL: lower>> to <<FILL: upper>> |
| Is the entire confidence interval inside the bounds? | <<FILL: Yes / No>> |
| Parallelism disposition | <<FILL: Parallel, proceed to Part D / Non-parallel, stop, do not report potency>> |
If the disposition is non-parallel: stop. Do not compute or report a relative-potency value. Route the run to an investigation per the OOS/OOT process. Re-running the sample until a parallel result appears, without an investigation, is testing into compliance.
Part D: Relative potency calculation (only if Part C disposition is Parallel)
| Field | Entry |
|---|---|
| C (EC50), reference standard | <<FILL>> |
| C (EC50), test sample | <<FILL>> |
| Dilution / nominal-concentration correction factor applied | <<FILL: e.g. sample prepared at same nominal concentration as reference = 1.0; state the factor if different>> |
| Relative potency = C(reference) / C(sample), corrected | <<FILL>> |
| Expressed as percent of label | <<FILL>> |
| Confidence interval on relative potency | <<FILL: lower>> to <<FILL: upper>> |
Part E: Disposition
| Field | Entry |
|---|---|
| Final disposition | <<FILL: Relative potency reported / Run invalid, non-parallel / Run invalid, system suitability failure>> |
| Reportable value contribution | <<FILL: this run's RP feeds the geometric mean of N independent assays per the validated reportable-value format; state N and which runs>> |
| Analyst, signature, date | <<FILL>> |
| QA / reviewer, signature, date | <<FILL>> |
Acceptance criteria
- Part A is complete and passing before Part B is read for anything other than the suitability check itself.
- The parallelism equivalence bounds in Part C are copied from the validated method, never chosen or adjusted after seeing the data on this run.
- A non-parallel disposition in Part C stops the worksheet; Part D is left blank and no relative-potency number is calculated, let alone reported.
- The relative-potency calculation in Part D uses the same locked model and dilution logic applied to both curves; a 4PL result is never compared against a 5PL result, or vice versa.
- The reportable value is stated as the format the method defines (commonly a geometric mean across N independent runs), not as this single run’s number presented as final.
Common inspection findings this worksheet prevents
- A relative-potency number calculated and reported from a run that never passed the parallelism check.
- Equivalence bounds that shift from run to run, or that trace to no documented development data.
- A parallelism metric computed one way in validation and a different way, or on a different parameter, in routine testing.
- The reportable value quietly reported as a single run’s result when the validated method defines it as a multi-run geometric mean.
References
USP General Chapter <1034>, Analysis of Biological Assays (parallelism, equivalence testing, 4PL/5PL models). USP General Chapter <1033>, Validation of Biological Assays (reportable value, acceptance criteria format). ICH Q2(R2), Validation of Analytical Procedures, and ICH Q14, Analytical Procedure Development.
Confirm the current version of each reference before issue.
Filled specimen
A completed worksheet for a reporter-gene bioassay run, using the worked example from the governing article. Illustrative only.
Header: Method PROT-BA-014 v2.0, run BA-2026-0913-02, sample lot DS-2026-118, reference lot RS-2026-04, 4PL model, 1/Y^2 weighting, parallelism metric slope ratio with equivalence bounds 0.80 to 1.25 (from validation).
Part A:
| Parameter | Validated limit | Observed | Pass / Fail |
|---|---|---|---|
| Reference EC50 | 3.6 to 4.6 ng/mL | 4.10 ng/mL | Pass |
| Signal window (D minus A) | Not less than 30,000 RLU | 47,150 RLU | Pass |
| Reference-vs-reference parallelism | Inside bounds | Inside bounds | Pass |
| Edge controls | Within limits | Within limits | Pass |
Part B:
| Parameter | Reference standard | Test sample |
|---|---|---|
| A | 1,050 | 1,080 |
| D | 48,200 | 47,900 |
| B (slope) | 1.18 | 1.21 |
| C (EC50) | 4.10 ng/mL | 4.32 ng/mL |
| Fit diagnostic | Random residuals | Random residuals |
Part C:
| Field | Entry |
|---|---|
| Parallelism metric | Slope ratio (sample B / reference B) |
| Observed value | 1.21 / 1.18 = 1.025 |
| 90% confidence interval | 0.93 to 1.13 |
| Equivalence bounds | 0.80 to 1.25 |
| Entire interval inside bounds | Yes |
| Disposition | Parallel, proceed to Part D |
Part D:
| Field | Entry |
|---|---|
| C, reference | 4.10 ng/mL |
| C, sample | 4.32 ng/mL |
| Correction factor | 1.0, sample prepared at the same nominal concentration as reference |
| Relative potency | 4.10 / 4.32 = 0.949 |
| Percent of label | 94.9% |
| Confidence interval | 89.1% to 100.8% |
Part E: Final disposition: relative potency reported, 94.9%. This run is 1 of 3 independent assays feeding the reportable value (geometric mean across runs BA-2026-0906-01, BA-2026-0910-01, BA-2026-0913-02) per PROT-BA-014 section 8.
Note what makes this defensible: parallelism was evaluated and passed before the relative-potency ratio was calculated at all, the equivalence bounds came from the validated method rather than a judgment call on this run, and the single-run result is explicitly labeled as one input to a multi-run reportable value rather than presented as the final answer on its own.
How to adapt this worksheet
- Set the method reference, curve model, weighting, and parallelism metric in the header to match your validated method exactly; this worksheet does not define them.
- If your validated parallelism metric is a combined similarity score rather than a slope ratio, replace Part C’s fields with the specific statistic your method uses, keeping the same before-potency sequencing.
- Where the reportable value format differs (a different N, a different combination rule), state it exactly as validated in Part E.
- Have a second reviewer confirm the arithmetic and the parallelism disposition before this worksheet supports a release decision.