Medical Device Link.

MANUFACTURING

The vision

Image: iStockphoto

This article returns to the original vision of Six Sigma. It illustrates how Six Sigma has a symbiotic relationship with validation and risk management. In doing so it provides an insight into bridging the gap between Six Sigma metrics and expressing high degrees of assurance. It also suggests some of the main first steps in the deployment of Lean Six Sigma in medical device manufacturing organisations.

The vision of Six Sigma is the creation of near perfect products and processes. It is helpful that near perfection is quantified: it is operating processes that are able to produce less than 3.4 defective parts per million (ppm).

By having well understood, highly repeatable and reproducible processes, defects are eliminated, complexity is reduced, financial return is improved, customer satisfaction is maintained and therefore business survival is assured. Furthermore, product development is accelerated, its output is improved and the dynamic, language and tools for a positive cultural shift within an organisation are realised.

Delivering sustained improvement

Good quality products are essential to patient safety, customer retention and commercial success. Companies use a variety of approaches to achieve this, but generally, in the absence of getting it right first time, they use a combination of screening, inspection and rework. In critical applications the inability to get it right first time results in long delays in getting products to market, the possibility of inadequate validations, poorly managed risk, repaired/rebuilt processes and the associated costs and ongoing corrective action activity. In a minority, but measurable number, of cases it can result in field events and recalls. All these activities introduce cost, complexity and further risk. These costs contribute to the overall cost of poor quality (COPQ) in an organisation. Elimination of these costs is the most visible benefit of Six Sigma.

Although companies such as Motorola pioneered Six Sigma, it was the deployment successes with a fiscal focus at AlliedSignal (Honeywell) (www.honeywell.com) and GE (www.ge.com) that popularised it, made it attractive to Wall Street and ensured its survival. In an organisation dedicated to Six Sigma, the “hidden factory,” that is, the activities described above that exist to service the consequences of not getting it right first time is tackled through Six Sigma projects. Individuals and teams are mobilised to special action through training. These individuals are resourced and given authority, and with this comes accountability to deliver tangible, measured and sustained improvement. At AlliedSignal and GE, once people started viewing processes from the perspective of Six Sigma, the calculated COPQ was enormous, and in some cases estimated at 30% of the value of the products. Soon Six Sigma practitioners were recording million dollar process improvements through Six Sigma projects. One Black Belt certification project at an AlliedSignal automotive turbo-charging components facility annualised £300 (e378) ongoing savings through defect reduction. Another individual project recorded £1.2 million (e1.5 million) through rework elimination. There were many similar projects making the deployments a success and contributing savings, as a percent of revenue of 3.3%. The financial focus of the projects was an important factor in contributing to their success.

Achieving the vision

Simply put, Six Sigma is achieved by getting processes on target and reducing variation. By “centring” processes and controlling sources of variation, the possibility of defects is reduced to levels where there is no longer a need to have over inspection and rework. Then the infrastructure, that is, the systems, equipment and dedicated resources needed to sustain the rework and extra inspection are no longer required. People’s effort can be redirected to other activities such as product and process development, training and forward planning rather than dealing with the results of poor quality. It is interesting to calculate how much time is spent daily on dealing with preventable issues and worth keeping a detailed diary to measure it.

By removing the defect management infrastructure, known as the hidden factory, complexity is reduced. The processes can be reorganised into simpler, leaner configurations leading to improved flow and opportunities to further minimise waste; this is the definition of Lean Six Sigma. The confidence needed to realise this ambition results from Six Sigma levels of capability and control.

For existing products, the effective way to do this is by employing the Six Sigma methodology of define, measure, analyse, improve and control (DMAIC). The DMAIC approach uses scientific method to understand the cause and effect relationships in a process and through this to improve it. Its target is to centre the process so that the mean of its output is less than six times the average variation or standard deviation. The goal is to attain capability that yields less than 3.4 ppm outside specification in the long term.

Validation and Six Sigma

DMAIC is an effective approach to corrective and preventative action. A basic problem solving element is critical to effective corrective action. With Six Sigma it becomes reasonably practicable to lower defect occurrences to 3.4 ppm. Minimising the occurence of an adverse event to As Low As Reasonably Practicable level is risk management. ISO 14971 Medical Devices, Application of Risk Management to Medical Devices, describes process validation as risk management verification. It establishes documented evidence (Installation, Operational and Process Qualification, software, design verification and test method validation) that provides a high degree of assurance that a process will consistently produce a product that meets predetermined specification and quality attributes. The risk management actions are the establishment, monitoring and control of process inputs and outputs. Six Sigma and validation both have the same end: doing the right thing, right first time. Planning validation activities is guided by potential risk to the end user using hazard analysis such as Failure Mode and Effect and Criticality Analysis (FMEcA). The same rationale that applies to using validation to verify risk management can be used to remove wasteful inspections and improve process flow. Financially focused risk management can follow the same approach as patient risk.

Management of the validation infrastructure (procedures and personnel) requires an ability to qualify that processes are suitable for their intended use without having to perform 100% inspection, but instead to control the inputs and monitor the output. Every medical device manufacturing company that conducts validations has the ability to measure and document the management of process risk, which is a component of Six Sigma. By having capable processes, companies can eliminate wasteful over inspections. It is no surprise then that leading medical device companies such as J&J, (www.jnj.com) Medtronic (www.medtronic.com) and GE Healthcare have adopted Six Sigma and why progressive institutions such as the Limerick Institute of Technology in Ireland (www.lit.ie) have established life science courses based on it. Attainment of a Six Sigma belt is fast becoming the industry standard for individual progression.

Measuring the results

Table I: Calculated using a 1.5 sigma shift. Process sigma level verses ppm defective and yield. The reported process Sigma is highly dependant on the number of samples used to calculate them. (click image to enlarge)

There is sometimes an uncomfortable discussion concerning the metrics used to establish sigma levels of processes and other statistical tools used to provide confidence that processes are suitable for their intended use. The empirical goal of Six Sigma is to target the process and reduce the variation so that the mean is six standard deviations or six sigma away from the specification. Statistically this equates to a likelihood of approximately two defective parts per billion. Controversially, Six Sigma suggests that the mean varies over the long term and an empirically derived correction factor of 1.5 sigma is used in calculating the likelihood of defect so that it is stated that Six Sigma performance equates to less than 3.4 ppm. The empirical foundation for this assertion does not endear it to the statistical purist. That debate is for another forum, but the 1.5 sigma shift does have practical uses.

Process capability indices such as Ppk and Cpk are used to express the spread and location of the data. These indices are highly dependant on sample size. For example, using 50 samples (n = 50), the 95% confidence interval for a Cpk of 1 (chosen for illustration purposes) is between 0.78 and 1.22, that is, a range of 44% of the recorded index. As the sample size is increased, the precision of the index improves so that at n = 500 the precision is a more manageable 14% (0.93 to 1.07). For every day purposes, testing 50 samples to destruction is expensive, impractical, time consuming and difficult. Five hundred is often just not feasible. Yet, frequently capability indices are discussed as fact without regard to their reliability and they are used as the basis to calculate the sigma level of a process. The fact of the matter is that only a tolerance interval approach can confidently determine process reliability.

Figure 1: Illustration of 3.4 ppm tolerance interval. By sampling the process users can express a range that they are 95% confident in for approximately 99.99966% of the data. As the sample size n increases the width of the interval decreases. This was constructed using a mean of 0 and standard deviation of 1. The tolerance interval takes the form x-bar +/- ks. X-bar = mean, k = factor, s = sample standard deviation. The k factors were generated using TiNorm. (click image to enlarge)

Critical products have mandated performance levels. For example, the rated burst pressure of a percutaneous transluminal coronary angioplasty balloon is based on the inflation pressure that 99.9% of the balloons will meet with 95% confidence. Tolerance intervals that combine confidence and proportion are elegant statistical statements about expected performance that vary according to the sample size used to calculate them. They are intuitive and accessible. They can be calculated easily for a range using tables such as those contained in ISO 16269: 2005, Statistical Interpretation of Data, Part 6: Determination of Statistical Tolerance Intervals. The effort needed to calculate precise intervals manually is considerable. The equations to produce precise tolerance intervals are complex,¹ but this is addressed by easy to use validated software such as TiNorm (www.tinorm.com). Tolerance intervals can be calculated for proportions equivalent to the likely percentage of defective products expected at the different sigma levels (Table I). The calculation determines the number of standard deviations away from the mean required to express a certain proportion of data. The main difference is that tolerance intervals do this as a function of sample size. For example, using software the range can be extended to encompass 99.99966% or a likelihood of defect of less than 3.4 ppm opportunities. It is practical to do this for samples as low as eight, based on the minimum the standard requires for testing for departure from normality,² thereby eliminating empirical correction factors and minimising the material cost of the study. For a two sided interval, that is, one that addresses upper and lower specifications, this is graphically represented in Figure 1. A prerequisite is that the process is capable and in control, and that is achieved by black and green belts in their projects using the DMAIC approach. Therefore, using tools such as TiNorm, qualification studies of capable processes can be documented to six sigma levels of performance using economic sample sizes and with a high degree of statistical assurance.

Armed with the right tools and training, black and green belt, confident and accountable individuals can address the COPQ, simplify processes and realise the goal of regulations, that is, safe and effective products.

Deployment

Some organisations struggle with validations. For many the reality is that they are time consuming, sometimes frustrating and expensive in terms of product and resources. The result is that if they can be avoided, then they are, often by implementing 100% inspection. In addition, process improvement is tempered because it can, for some, require what amounts to an “unlocking” of the existing validated state. There is often a reticence to change settings that have been “validated” and a preference to make do with the present imperfection. Therefore, barriers to validation are barriers to the flow of process improvement and operational excellence. Often this is contributed to by perceptions and practices around process validation, in particular its documentation and approval processes. Thus, the first step is to be open to change and a lean validation infrastructure. This will eliminate complexity and improve throughput speed. A mechanism for doing this is through a transactional Six Sigma project.

Variations within and between suppliers can contribute to inhouse process volatility. It is important that suppliers are encouraged to resource their Six Sigma problem solving and process improvement capabilities, and that they have the personnel available and trained to engage with medical device manufacturers in the pursuit of operational excellence. The wide availability of open enrolment courses such as those at Limerick Institute of Technology makes Six Sigma accessible to small- and medium sized companies.

The black belt tool kit and training certificate gives individuals and managers the confidence to know that they have the ability to effectively characterise their processes and plan improvement. It is useful if Six Sigma trainers have an insight into the medical device manufacturing requirements and better if they have a dedicated curriculum such as MDLeanSigma.

Medical Device Lean Six Sigma

Six Sigma in medical device manufacturing requires and achieves a blend of accountability, reliability and velocity excellence. Through Six Sigma, medical device manufacturers can address risk, reduce complexity, improve process reliability, increase throughput velocity, remove waste and costs and thereby achieve customer and stakeholder satisfaction. Medical device manufacturing, by creating a lean validation system and harnessing it to address fiscal risk and user risk, and by using appropriate statistical tools, can more readily and effectively deploy Six Sigma.

Eoin Barry is a Consultant with ARV Excellence Ltd in Limerick, Ireland, tel. +353 6 144 8508, e-mail: eoin@arvexcellence.com, www.arvexcellence.com