Confidence and reliability. Those are two words you’ll hear a lot when in discussions to test medical device packaging. They relate to the number of samples needed to test properly. Here’s the catch: There is no perfect sample size answer for every situation. But guidelines do exist to determine sample size. We’ll break this process down to better inform your medical device packaging needs and how you will work with your testing provider.
Guidelines for medical package testing sample size focus on two areas: Risk, and correlating risk to confidence and liability. We’ll review both before discussing common sample sizes.
Medical device package testing samples: Calculating risk
Calculating risk is a priority for medical package testing. Risk is defined by ISO as “the combination of occurrence of harm and the severity of that harm that can occur due to failure.” How is risk calculated? This is where the Risk Priority Number (RPN) from your FMEA (Failure Modes and Effects Analysis) comes into play. The RPN calculation is based on three things:
- Severity: Rates how severe the potential effect of failure will be
- Occurrence: Rates how likely failure will occur
- Detection value: Rates how likely a problem will be detected before reaching the end user
Rating scales for those three risks typically range from 1 – 5, or 1 – 10. The higher number is an indicator for highest risk for either range.
They will be multiplied once all three risk rating numbers have been assigned: severity x occurrence x detection value. This will be the RPN value. Mitigation will be recommended if the RPN falls in a specific range.
Above and beyond standard RPN are optional RPN methods such as “revised RPNs” and “percent reduction in RPNs.”
Calculating risk continues with additional steps, though these are the basics to understand exactly how it impacts your valid medical device samples. Next up is correlating risk.
Medical device package testing samples: Correlating risk
After calculating risk, your testing provider will correlate the RPN to confidence and reliability:
- Confidence: Relates to an unknown constant
- Reliability: Estimates how many packaged units will meet pass/fail criteria with success
A Non-Parametric Binomial Reliability Chart is often used to determine sample size based on the confidence and reliability measurements.
These are calculated in percentages. The percentage correlates to the number. For example, reliability at 90% indicates 90 – 100 units will meet pass/fail criteria. Confidence at 95% indicates the manufacturer (or testing provider) is 95% certain they will have less than or equal to 10 failures.
Once those numbers have been identified, then move on to next determining steps.
Sample sizes needed: Combining data
Other factors can aid in determining the ideal number of samples to test (for instance, implantable devices are high risk and can impact the number). But there is a common number most medical device package testing professionals turn to ensure an adequate number; 29 and 59. LSO uses 30 and 60 as samples sizes for 95%/90% and 95%/95%.
By calculating risk, assessing correlating risk, identifying additional factors, and considering the common number of 30 or 60, your medical package testing provider can determine the ideal sample size to test your device packaging.