Do you really need to choose a single worst-case?
The worst-case approach frequently comes up in cleaning process validation: for acceptance criteria, for parameter evaluation, for defining hold times (DHT, CHT). The founding principle is simple: "if it works for the hardest, it works for all."
Why look for a worst-case product?
The goal is to prove that the cleaning process meets the expected cleanliness standards, even against the most stubborn product. If you succeed with the most difficult one, you succeed with all the others. The selection requires examining composition, toxicity, solubility, persistence on equipment, and cleaning difficulty.
This approach avoids validating each product individually, saves time, and strengthens credibility during inspections.
Two ways to define a worst-case
Method 1: the risk analysis approach
This method combines several criteria from GMP Annex 15: solubility, cleanability, toxicity, activity, etc. An overall scoring allows products to be compared.
Important note: for formulated products, the quantities of each substance must be taken into account. An excipient that is difficult to clean in large quantities will carry more weight than a highly active API present only in trace amounts.
Example — Product compositions:
| Component | Product A | Product B | Product C | Product D |
|---|---|---|---|---|
| Drug Substance | 0.07 mg/g | 0.03 mg/g | 0.01 mg/g | 5 mg/g |
| Sucrose | 60 mg/g | 60 mg/g | 95 mg/g | — |
| Mannitol | — | — | — | 50 mg/g |
| Tween | 0.2 mg/g | 0.2 mg/g | 0.2 mg/g | — |
| NaH₂PO₄ | 0.5 mg/g | 0.5 mg/g | 0.5 mg/g | — |
| NaCl | — | — | — | 8 mg/g |
FMEA scoring results:
| Product | Solubility | Cleanability | Activity | Toxicity | Score |
|---|---|---|---|---|---|
| Product A | 1 | 1 | 5 | 1 | 5 |
| Product B | 5 | 1 | 1 | 5 | 25 |
| Product C | 5 | 5 | 1 | 5 | 125 |
| Product D | 10 | 5 | 5 | 1 | 250 |
Product D emerges as the worst-case for validations.
Method 2: the experimental laboratory approach (coupon study)
Products are tested directly at small scale under controlled cleaning conditions. Residue quantities are measured before and after cleaning (generally by gravimetry). Parameters are adjusted to allow comparability — for example, replacing mechanical action with standardised soaking.
Example — Results:
| Measurement | Product A | Product B | Product C | Product D |
|---|---|---|---|---|
| Quantity deposited | 10.25 mg | 9.56 mg | 9.85 mg | 10.12 mg |
| Remaining quantity | 2.56 mg | 1.54 mg | 0.98 mg | 2.89 mg |
| % loss | 75.02% | 83.89% | 90.05% | 71.44% |
| Visual appearance | Clean | Clean | Clean | Traces observed |
Product D emerges as the worst-case with the lowest loss rate.
Coupon studies differ fundamentally from cleaning efficacy tests. The objective is to evaluate the intrinsic cleaning difficulty of the product, not to demonstrate the effectiveness of the cleaning process.
What when multiple products are "worst-case"?
Real-world scenarios frequently involve multiple products with similar scores. Example:
- Product 1: Score 5
- Product 2: Score 25
- Product 3: Score 250
- Product 4: Score 250
Products 3 and 4 both qualify as worst-case. Choosing one requires adding discriminating criteria:
- Maximum API concentration
- Highest toxicity level
- Greatest activity
- Largest batch size
- Highest production frequency
However, these additional criteria may lack a real connection to cleaning effectiveness, potentially obscuring realistic considerations. This is the worst-case trap: trying to force a single unique choice at all costs.
Accepting complexity also means gaining robustness
Seeking a single worst-case at any price risks being counterproductive. While presenting a single worst-case simplifies discussions with an inspector, accepting multiple alternatives proves more realistic. Multiple choices bring operational and scientific flexibility during validation testing, strengthen the credibility of the cleaning strategy and better prepare for audits. The balance between scientific rigour and industrial pragmatism must be found.