The acronym DMAIC stands for: 1) Define, 2) Measure, 3) Analyze, 4) Improve, and 5) Control. I am suggesting the use of these 5 steps to target implementation of process risk controls to prevent complaints.

Step 1 (Define) – Identify the most likely complaints during the design process.
The goal of step one is to define the most likely quality issues before they happen. Most medical device manufacturers are currently using a design failure modes and effects analysis (dFMEA) as a risk analysis tool during the design process. The dFMEA is a bottom-up approach for risk analysis that starts with how each component of the device might fail. This is great when you are designing a product, but it makes it hard to review your risk analysis when you are performing a complaint investigation. Complaints are the result of a device or component malfunction. Therefore, a top-down risk analysis tool would be more helpful. If you start with hazards and harms, instead of components, then you should be able to identify the most likely complaints during your risk analysis. This is one of the reasons why I have integrated risk analysis tools into my design requirements matrix.

Step 2 (Measure) – Identify quantitative data related to each potential complaint.
The goal of this step is to establish a receiving inspection, in-process metric or final inspection (i.e., quality objectives) that can be used to monitor the effectiveness of your process and controls to ensure products meet the design specifications. Wherever you anticipate potential complaints, you should also be able to identify some related quality metric that can be used to predict when there may be a quality issue before it occurs. In Part 2 of the case study blog I published today, the peel testing of pouches at incoming inspection is a measurement method used to predict quality issues with packaging materials.

Step 3 (Analyze) – Analyze quality data and establish alert limits.
Every process has a validated process specification–typically an upper and lower limit. However, you should establish alert limits as well. Many companies choose +/- 3 sigma as a specification limit and +/- 2 sigma as an alert limit. You can deviate from this standard approach–especially if you want to set a more stringent alert limit. If you have a process limit which is not based upon statistics, you should analyze the data to determine process capability. Typically process capability is measured as Cpk and a typical acceptance criteria is > 1.33.

Step 4 (Improve) – Improve process controls whenever process capability is < 1.33.
If your process capability is less than the target of 1.33, you can use your alert limit to identify and contain possible defects. However, you should be using the Cpk as a factor to prioritize your process improvements. You can implement tighter raw material specifications, improve the design of equipment or focus on operator training to reduce variability.

Step 5 (Control) – Remeasure the process after implementation of improvements.
Each improvement you implement in step 4 is a potential preventive action. However, you still need to verify effectiveness of preventive actions. The best way to do this is to compare the Cpk of the process prior to the preventive action with the Cpk after the preventive action is implemented. If the Cpk increases substantially, then the preventive action is effective. Ideally, you will decide what an acceptable Cpk value is prior to implementation–possibly as part of a process revalidation protocol.

If you are interested in learning more about the risk analysis tools I have developed, please read last week’s blog on this topic: http://medicaldeviceacademy.com/risk-management-traceability-for-ce-marking/.

If you are interested in learning more about risk-based CAPAs, please visit my webinar page on the topic: http://medicaldeviceacademy.com/create-a-risk-based-capa-process/.

source: https://www.linkedin.com/groups/2070960/2070960-6070175701326184450