In the medical device industry, reliability is a critical factor because the failure of a device can have serious consequences for patient safety and health outcomes. Manufacturers often conduct extensive reliability testing during the development and validation phases to ensure that their devices meet the required reliability standards.

One common standard requires manufacturers to show 95/95 or 95/99 confidence and reliability. Let’s unpack what this important guideline means and how to use Minitab Statistical Software to determine whether you’ve met it.

### CONFIDENCE LEVEL AND RELIABILITY LEVEL RELATIONSHIP

You may have seen the numbers thrown around: 90/80, 95/95, 95/99, 99/99 etc. In the context of confidence and reliability, the first number is the confidence level, and the second number is the reliability level.

Every time you take a random sample of units from a larger population, you take a risk that the sample you obtain is atypical and conclusions you make about a population parameter based on this sample are not reflective of the entire population. A 95% confidence level means that you have a 5% risk of incorrectly concluding that you have demonstrated your reliability goal based on the specific units in your sample.

The reliability level is the target level of reliability that your system or product is expected to achieve. So, to achieve 95/95 reliability, you must demonstrate that at least 95% of the units in your population are conforming. Similarly, to achieve 95/99 reliability, you must demonstrate that at least 99% of the units in your population are conforming.

Putting it all together, if a process has achieved 95/95 confidence and reliability, you know that you can be 95% confident that 95% of the units in your population are good. With 90/80 confidence and reliability you can be 90% confident that 80% of the units in your population are good. Got it? Good.

### WHICH RELIABILITY STANDARD SHOULD I USE?

While a 95% confidence interval is by far the most common, the choice of reliability level – 95/90, 95/95, 95/99 – is often determined by the level of risk you can accept. These standards are application specific but using a tool such as the Failure Modes and Effects Analysis (FMEA) available in Minitab Workspace, you can identify the causes of failure and evaluate the risks associated with each cause.

Using the Risk Priority Number obtained from an FMEA, you can then categorize each type of failure as low, medium, or high risk and use that as a guide to determine the reliability level you want to achieve for withstanding that specific mode of failure. For example, you could set the reliability standards as shown in the table below:

### CONFIDENCE AND RELIABILITY IN MINITAB STATISTICAL SOFTWARE

Once you know the level of reliability you need, it is time to determine the best tool for the job. Let’s assume you have determined that your risk level is relatively high, so you want to demonstrate 95/99 confidence and reliability.

It turns out that there are many ways to do this, and the best approach depends on a few practical and statistical considerations such as:

The flow chart below will guide you to common tools in Minitab Statistical Software for demonstrating 95/99 or another confidence and reliability goal:

Download the Minitab Handbook for 95/95 Confidence and Reliability to look at the details behind each tool and how to use it to show that you have demonstrated the reliability you are looking for.