Design of Experiments (DOE) analyzes a process, targeting the process variables that most affect product quality, allowing one to understand those variables. Then, by controlling or-even better-mistake-proofing those variables, it is possible to prevent product defects from occurring.

Today it goes without saying that manufacturers must provide low-cost products that are defect free and meet customer needs. The requirements for manufacturers of critical equipment are even more stringent; the equipment must operate reliably and leave no doubt in the consumer’s mind that it will provide the required function.

From MRIs to catheters to joint replacements, medical devices must meet this higher expectation for quality. At minimum, faulty medical equipment can lead to discomfort on the part of the patient. A patient misdiagnosed by defective diagnostic equipment, however, can experience more extreme negative outcomes-even death. So how should manufacturers of medical devices assure a high level of quality? They need to go back to the basics.

Elements of Quality Control

Quality control has become essential in modern management practices. It is aimed at maintaining or restoring an acceptable level of quality for products, services or information. Quality control can be defined as the maintenance or restoration of the operating status quo as measured by the acceptable level of defects and provision of customer needs. For medical devices, the acceptable level of defects is very low and the level of customer needs is very high, but the maintenance of the status quo takes the very same form as it does with less critical products.

All control focuses on control subjects, those product or process features that are measured and must be maintained within prescribed limits.

After a control subject is measured, the result is compared to the acceptable standard, whether it is specification limits or, in the case of statistical process control (SPC), control limits. If the measurement conforms to the standard, the control subject is then measured again at the interval specified in the control plan. It does not matter what the standards are; control activities are just as effective with stringent standards as they are in less critical products and processes. Effective control loops also incorporate an actuator, ideally automatic, that adjusts the process to quickly restore control if it has been lost.

Ideally, a manufacturer would like to control the process rather than the product, thereby preventing defects in the product from the beginning. This is particularly important in critical product manufacturing such as medical devices.

Control vs. Improvement

In order to meet the customer’s critical quality requirements, the producer may need to do more than control the process. What if, for example, the process is not capable of producing products that meet the customer’s desires? In that case, control will not achieve fulfillment of customer needs and an improvement activity will be required.

The difference between control and improvement is distinct. Control implies the restoration of the status quo. Improvement, on the other hand, implies the changing of the existing process, taking it to a new and better level of performance.

Arguably, a process must be in control before it can be improved. But, if it is recognized that the process is both out of control and in need of improvement, methodologies such as lean Six Sigma can be applied to restore process control.

What to Control?

Particularly in critical manufacturing processes such as medical devices, understanding the process is essential if one hopes to control it. Design of experiments (DOE) is a lean Six Sigma tool that often is applied in such a case.

DOE analyzes a process, targeting the process variables that most affect product quality, allowing one to understand those variables. Then, by controlling or-even better-mistake-proofing those variables, it is possible to prevent product defects from occurring.

DOE concerns itself with five main variables, known as dominate variables:

Set-up dominant. Some processes are highly stable, and their results can be reproduced over many cycles of a process. The design for control should provide the individuals with the means for precise setup and the means to validate it before a process begins. A common example is an operation.

Time-dominant. Here the process is known to change progressively with time: depletion of consumable supplies, equipment or components heating up, length of an operation, wear of equipment, for example. The design for control should provide means for periodic evaluation of the effect of any progressive change and for convenient readjustment.

Employee-dominant. In these processes, quality depends mainly on the skill of employees. The skilled trades and specialists are well-known examples. The design for control should emphasize aptitude testing of operators, training and certification, quality rating of operators, and mistake-proofing to reduce staff errors.

Component-dominant. Here the main variable is the quality of the input materials, assembly and components. An example is the assembly of complex equipment. For the short run, it may be necessary to resort to inspection of materials from a supplier. For the long run, the design for control should be directed at supplier relations, including joint planning with suppliers to upgrade their inputs.

Information-dominant. Here the processes are of a job-shop nature, meaning that there is frequent change in what product is produced. As a result, the job information changes frequently, as in the case of a service department. The design for control should concentrate on providing an information system that can deliver accurate, up-to-date information on how this job differs from its predecessors.

In the manufacture of critical products, it is essential to control the process rather than the product. The key to this is a clear understanding of which process variables are most important in their effect on product quality. DOE provides an understanding of dominant variables in a process, allowing one to improve and maybe even mistake-proof a process to ensure high levels of quality.