Interviews

An Interview with the engineers and research and design team behind the Abbott Architect instrument line.

Interview by Sten Westgard, MS
With special thanks to Daniel Stredler, Gene Osikowicz, Esther Yang, and Tony Orzechowski.

Anyone following recent website postings over the last few months will have noticed that a number of QC applications have analyzed Abbott instruments (here, here, and here) specifically from the ARCHITECT product line. After a number of Sigma analyses produced favorable results, we began to wonder if there was something more to this story.

I had the opportunity to interview a number of Research and Design scientists at Abbott, to ask them questions about ARCHITECT performance. As it turns out, a lot of what they are doing in the research, design and manufacturing of their instruments can be applied directly to what we should be doing in the laboratory.

1. Set the goal: Do the Right Quality

At the laboratory level, it’s important to set goals for performance – it’s hard to know if you’ve made a touchdown or a safety if you don’t have any goal lines. As we have consistently recommended, laboratories must set the quality required by their tests and optimize their QC procedures to achieve that performance.

For manufacturers, establishing requirements and goals is quite easy; there is always a competitor's specifications to beat, or a "state-of-the-art" threshold to cross. But more-for-the-sake-of-more isn’t a recipe for a good instrument. Manufacturers must intelligently set the requirements for performance.

Abbott sets quality requirements for the ultimate performance of the instrument out-front. Before they begin working on the manufacturing, they’ve set the goal. From this performance goal, they engineer every sub-system and component of the instrument, so that the sum total of all the processes, sub-processes, and sub-sub-processes does not exceed that goal. This, we want to stress, is in stark contrast to most manufacturers, who slap their QC on the back of the instrument, like a bumper sticker, as the device is being shipped out the door. Abbott has made setting quality goals a first priority, not an afterthought.

Using these ultimate quality and performance specifications as a starting point, Abbott scientists created specifications for every sub-system, from sample delivery to reagent delivery, optics, temperature control, timing, carryover and more. One of posters that describes the subsystem performance can be found here.

Let’s take a detailed look at one subsystem, sample delivery: Abbott initially specified a goal for the c16000 at delivery of 2 ul with less than 3 CV% variation. Using Taguchi methods, engineering optimization of the sample wash cup and sample probe, and high speed robotics, Abbott was able to create a process that actually delivered 0.5% CV.

On the reagent side, Abbott identified Open On Board and Calibration Stability as key selection criteria for their CO2 reagents. Often suppliers do not make claims for this type of stability due to platform dependency, but Abbott worked with the suppliers to reiterate these requirements and ensure that the suppliers were capable of delivering the desired product.

2. Do the Right things to get Quality Right

As anyone stuck with a clunker in their lab knows, you can’t work miracles on a bad instrument. If you’ve got the wrong device, the wrong method, or the wrong reagent, you’re in a deep hole. No matter how good the technologists or how diligent the QC procedures, you can’t turn a toad into a prince. If you want good test results, you’ve got pick the right instrument in the first place.

This is true with instrument design as well. When Abbott began designing their instrument, they were very careful about picking the right partners. Having set quality requirements, they needed vendors and suppliers who could provide them products that would be able to achieve that quality.

Gene Osikowicz, R&D Director of Clinical Chemistry Systems, explained that Abbott selected Toshiba Medical Systems, a company with more than 25 years in the industry, to develop the Clinical Chemistry analytical modules. Partnering with Toshiba allowed Abbott to reap the benefits of years of continuous improvement, lessons learned, and time-tested hardware and software. As, Dr. Osikowicz put it, the ability to draw upon decades of engineering experience, "can't be over highlighted."

On the reagent side, Abbott employed an equally rigorous selection process. Dr. Esther Yang, R&D Director of Immunochemistry Systems, noted  that in "some cases we set the requirements beyond the supplier claims and in other cases we add the requirements that the suppliers do not provide." In addition to developing demanding requirements on their part, Abbott performed quality audits, risk assessments, and a review of supplier complaints, field actions, lot histories, and stability data of potential suppliers.

3. Assure you keep doing the Right Quality, don't assume it.

In the laboratory, you’re going to meet a lot of sales reps who will promise you best in class performance and a host of features, including the Brooklyn bridge, all at a cost that’s practically free. And you know better than that.

In fact, one of the reasons why CLIA and the accreditation agencies require method validation for non-waived methods is to put the pin to that balloon full of marketing hot air. The regulators require labs to perform specific experiments so that we can assure that the new instrument can achieve, if not the promised performance, at least the performance necessary for good medical care.

On the manufacturing side, there are even greater regulatory hurdles. There’s the infamous premarket notification (510k) of the FDA, a Byzantine application process, of which we will say no more. Obviously, since the ARCHITECT line is on the market, it cleared the 510k process. But of more interest are the quality assurance procedures undertaken by Abbott and Toshiba during instrument design.

Back in May 2005, Abbott began running 50,000 to150,000 assay results a month on their first feasibility prototypes. These runs would include comprehensive metabolic panels from established serum pools, in multiple replicates. Any value that went outside of 5 standard deviations was declared an outlier and had to be investigated.

Abbott tracked these outliers in ppm, the equivalent of defects per million. By statistical theory, you should expect 233 defects per million will exceed the 5 standard deviation limit. However, Abbott’s initial defect rate was approximately 2,000 dpm (or 4.4 Sigma). Using root cause analysis, the problem with each defects was established, and corrections and improvements were made. By the end of the feasibility stage, Abbott had lowered their defects over 88%, bringing the outlier rate back into accord with expected probabilities.

During their design control phase, spanning a year from January 2006 until January 2007, another million and a half assay results were completed and reviewed. In addition to these studies, external familiarization studies were being performed in real-world laboratories outside of Abbott. Many of those experiences became posters at the AACC conference.

The quality assurance processes have not ended just because the instrument is now on the market. Each and every finished analyzer is tested at the factory for assay performance.

On the reagent side, CLSI protocols have pretty clear requirements for the assessment of performance. But Abbott makes it harder on themselves, so that they can have a better idea of assay and system performance at the boundary limits. For instance, while CLSI guidelines find it perfectly acceptable to use one control lot to test reagent performance, Abbott will test multiple control lots, including lots near the end of their shelf life or near the end of their on-board life. This results in more realistic and robust estimates of performance.

Conclusion: Quality as Competitive Strategy, both in manufacturing and in the laboratory

Dr. Michael Porter wrote a classic article in the Harvard Business Review on the importance of strategy in business. He defined strategy as the creation of self-reinforcing synergy “among a company’s activities. The success of a strategy depends on doing many things well – not just a few – and integrating among them.”[1]

Abbott has made quality a competitive strategy. They have built a self-reinforcing system that not only promotes quality, but makes it easier to achieve. Setting quality requirements early in the instrument design process makes it easier to set sub-system goals for quality. Rigorous attention to sub-system performance in turn makes it easier to achieve the quality requirements set in the first place.

Dr. Porter notes that once a company achieves a competitive position, it is difficult for competitors to emulate it. “A competitor seeking to match an activity system gains little by imitating only some activities and not matching the whole.” Competing manufacturers will certainly attempt to copy the success of Abbott, but attempting to improve on one aspect of method performance won’t succeed unless they, too, invest in a similar quality system to support that.

Quality can be a competitive strategy in the laboratory, too. In an age of compliance, cost-cutting and “bottom line” thinking, any laboratory trying to compete only on cost and the lowest common denominator faces a losing proposition. The more you turn your testing into a commodity, the more cost pressures you will face. The solution is not to emulate the quality-slashing cost reductions of your competitor, but to differentiate your testing by providing more quality, more service, and better results.

Reference: Michael E. Porter, What is Strategy?, Harvard Business Review, November-December 1996.