EDITOR’S INTRODUCTION: Last fall, Mark Smythe’s four-part DARK REPORT series about the thirteen “Perilous Parallels” common to commercial laboratory managers provoked widespread response among our clients and readers. We’ve invited him back to address management issues involving laboratory automation and robotics. As an industrial engineer with 35 years experience at some of America’s best-run companies, Mr. Smythe’s insights about the economics and usefulness of laboratory automation will surely stimulate animated discussion among laboratory executives currently considering laboratory automation and robotics.
CLINICAL LABORATORY AUTOMATION is a hot topic in the industry today. Open any lab publication and you will find prominent stories about how automation and robotics promise to transform clinical laboratories.
A careful reading of these stories reveals a very different conclusion. Laboratory automation is fraught with pitfalls and problems. It is a technology whose time has not yet come.
For unwary laboratory executives, the consequences of investing capital too soon, for the wrong reasons, or on the wrong technology, can bring about financial disaster if not bankruptcy.
I know of what I speak, because for almost four decades my career focused on improving manufacturing operations and introducing industrial automation and robots into actual production for well-known companies such as Control Data, Emerson Electric, NCR, Mallory, Philco Ford and others.
When I began working with clinical laboratories several years ago, I was struck by two things. First, clinical laboratories are exactly like factories. Raw materials (specimens) come in one door and finished product (test results) goes out the other door. Second, almost no lab executive sees his clinical laboratory as similar to a factory.
As a consequence, most laboratory executives fail to access invaluable sources of management wisdom and techniques that exist outside clinical laboratories. Because clinical laboratories and factories are alike, techniques used to slash costs, boost productivity and innovate in the factory can work with equal success in the laboratory.
Automation and robotics is one example where the experience of the manufacturing world can help clinical laboratory executives make informed decisions. The purpose of this article is to help you examine automation through the eyes of an Industrial/Manufacturing Engineer. In so doing, you may well save your laboratory millions of dollars in capital investments, wasted labor and dissatisfied physician clients.
Questions asked by industrial engineers affect two basic areas of the business. First, how will this automation enhance the efficiency and flexibility of our manufacturing process? Second, what return on investment and what increase to operating profit margins will accrue from automation?
A good engineer not only identifies what problems are expected to be solved, but anticipates the problems which will be created. He asks questions: How will the end product be improved… or diminished? Where will the new process position us regarding state of the art? Will the results produce measurable, tangible benefits? Finally, how extensive must the installation be to secure optimum results?
It is important to do this homework before making the commitment to automate your laboratory. Comprehensive planning avoids problems and unnecessary implementation expenses. Paying outside expertise to come to your laboratory and work your team through these issues before shopping for specific equipment will be money well spent.
Within the framework of the issues high-lighted above, industrial engineers evaluate automation’s potential to improve several specific production processes. The use of robots, for example, is typically recommended when the items being processed are too heavy to lift, too awkward to handle, too hot or too cold to touch, too delicate (40% of scrap is assignable to people losing their concentration) or in environments that are chemically, electrically or mechanically hazardous.
You may have a TQM (Total Quality Management) program of some sort where zero defects is a goal. Variability is a source of defects and errors. Robots are frequently used to reduce variability. In fact, new generations of robots have the sophistication to react to minute or gross variability in a process.
Often problems of variability are actually attributable to suppliers, not your laboratory staff. In such cases, robotics and automation would not necessarily solve those problems.
Cannot Solve Problems
Process problems of variability can also be caused by factors which automation cannot solve. For example, I had a serious variability problem at a production facility in Greencastle, Indiana. The plant manufactured solid tantalum capacitors for the electronics industry. Yields were below 50% and quality was erratic.
One obvious solution was to overwhelm the problem with additional production capacity and automated processing equipment, a very expensive option. Instead we attacked the problem using Value Analysis techniques. One process engineer recommended that we humidify the atmosphere. For minimal cost, we installed water lines with misting nozzles and yields immediately jumped to 95% with consistent quality.
This was a case where Value Analysis techniques identified a solution which required no equipment or automation. It was an extraordinarily cheap fix with huge profit impact.
Another area where automation contributes to lowering production costs is by reducing the time it takes to prepare for a production run. In the laboratory, set up and lead times for actually running tests are affected by order entry delays, missing data, illegible specimen labels, late or missed pick-ups, specimen integrity problems, poor scheduling of test runs, excessive instrument breakdowns, inadequate operator training and similar issues.
Regardless of how efficient the automated line runs in the laboratory, all these factors influence whether the specimens can actually be tested or not. Most of these factors cannot be solved by laboratory automation or robotics.
Next comes what engineers call “The Six-Tenths Rule.” A machine with twice the capacity has six-tenths of the unit cost per output. Productivity and cost are disproportionate. Even though the machine doubles productivity, the cost of production does not fall by half.
The reason that productivity and reduced costs do not change proportionally is because processing costs associated with the new equipment off-set savings by a significant factor. Increased costs include equipment service contracts, regular maintenance costs, retraining and site modifications. Walls may need to be moved and changes made to electrical wiring, floor tapes, hangers and tables. These costs are a direct consequence of bringing in the more productive equipment.
Volume Drives Automation
Industrial engineers look carefully at the volume going through a factory. High volume is essential if automation is to pay for itself. Most automation requires high volume to justify the capital expenditure and related implementation costs for automation equipment.
Manufacturing plants have two and three shifts running six and seven days a week. Continuous volume makes it feasible to amortize expensive automation technology.
However, laboratories do not operate work shifts like factories. Laboratories typically run only one shift at maximum capacity. Laboratories don’t work weekends. It is difficult for laboratories to create the continuous flow of high volume necessary to recoup the expense of the equipment.
Unit value of the product also drives automation. Factories making automobiles create production volume that generates sizeable dollar value, ranging in the millions of dollars per hour. With plants running 18-24 hours per day, seven days per week, potential savings from automation are huge.
Unit value again places laboratories at a disadvantage to factories when seeking to automate. The average unit cost of a laboratory test is typically $10-15 at a large commercial laboratory site. Compared to saving 10% on a $20,000 automobile, saving 10% on a $15 test makes it difficult for a laboratory to recoup the costs of automation.
Further, a laboratory site generating $50 million in annual net revenues is only doing 2-4 million billable tests per year. Unlike widget manufacturers who stamp out tens of millions of five-cent items, the relatively small annual throughput of product from a clinical laboratory reduces the opportunity to recover the cost of automation.
Another issue in automation is how it best serves the needs of your customers and clients. There should be a direct correlation between those needs and the proposed automation. Automation of both factories and laboratories can affect customers and clients in negative ways. Frequently these consequences are learned only after the automation is installed.
Laboratories do not operate work shifts like factories. Laboratories typically run only one shift at maximum capacity. Laboratories don’t work weekends.
Automated storage, long distance conveyors and similar automation enhancements add cost to the product but generally do not add value to the customer. Most laboratory executives would find it interesting to know that the greatest uses of robots throughout the world are in painting and welding (processes not used in a clinical laboratory).
Within the factory world, a key reason to automate involves solving two serious people issues. First is to reduce the ongoing labor required to produce goods and services, particularly where wages are high. Second is to eliminate or replace labor in situations where unions are militant and labor relations are uneasy. The automobile industry proves to be a great example on both points concerning labor issues.
In a manufacturing plant, a major cost is production labor. That is not necessarily true in a clinical laboratory. The ratio of medical technologists (“production workers”) to total laboratory staff is generally low compared to manufacturing plants.
A further difference with clinical laboratories compared to factories is that medical technologists tend to be more cooperative employees than factory workers. Issues of labor unrest and poor attitudes do not have the financial impact in laboratories that they do in factories.
Because of these facts, laboratories do not have the same potential gains from reducing high-priced labor hours and eliminating the management problems of dealing with unions or a labor pool that is uncooperative. Automation in the laboratory setting does not provide management with the same benefits in dealing with labor that it does in a manufacturing plant.
Up to this point, the examples I provided deal with the mechanical impact of automation to the workflow process. Laboratories and factories are alike in how they gather raw materials and process them into finished products. For that reason, automation interacts with workflows in clinical laboratories in the same way that it does in a factory.
However, once engineers evaluate the mechanical application of automation to the workflow process, they must also evaluate the financial impact of the proposed automation project. It is beyond the scope of this article to discuss how the financial analysis should be done, but I do want to highlight several key points.
First, whatever automation equipment is chosen for a laboratory, the manufacturer should specify an expected return on investment (ROI). Calculations to arrive at this number should be clearly understood. Both the laboratory buyer and the vendor should be prepared to work together to achieve that ROI.
As the project is implemented, there should be clear measures to monitor and evaluate efficiency, productivity and financial performance. All too often I find that laboratory administrators do not collect and report accurate data to guide their management decisions. Yet it is precisely this information which their hospital CEO and CFO use to evaluate capital requests and authorize major expenditures.
Like most DARK REPORT readers, I eagerly scan the clinical laboratory press for financial documentation as to how laboratory automation has reduced costs, improved productivity and delivered a market return on investment to those few laboratories which have pioneered the installation of such technology.
Such documentation has not been forthcoming. Consequently, it would be a reasonable conclusion that neither the automation vendor nor the laboratory customer is totally satisfied with the performance of their laboratory automation installations to date.
Quest Automates Labs
Only two years ago, Quest Diagnostics Inc. (formerly Corning Clinical Laboratories) announced that they planned to introduce automation into their St. Louis, Denver and Detroit laboratories. After automating St. Louis and Denver, Quest has yet to automate Detroit. That can be interpreted to mean that, based on the financial return of the first two automation projects, Quest determined that eco- nomic performance of the current generation of automation did not justify installation in the Detroit laboratory.
Several knowledgeable observers told me that Mayo Medical Laboratories flirted with an automation vendor and apparently went so far as to sign a contract and begin design work. But at some point they got cold feet and stopped the project. That could be another sign that the economics of laboratory automation are still marginal, at best.
However, should the Mayo story be true, then the management team at Mayo should be complimented. They had the courage to pull the plug on something that looked uncertain and wait until there was more documentation as to the cost-effectiveness of laboratory automation.
SmithKline Beecham Clinical Laboratories’ automation project at their Norristown, Pennsylvania facility was launched several years ago. Insiders say it has proven to be prohibitively expensive. SmithKline has yet to publish data on either the productivity performance or financial return generated by the automation.
The lack of published documentation and the anecdotal stories mentioned here indicate that laboratory automation is still in its infancy. Were I to wear my industrial engineer’s hat and give advice to a laboratory administrator looking at automation, I would bring out two points.
First, even with projected ROI payout over five years, rapid changes to both the technology of automated laboratory equipment and to the tests themselves may render today’s generation of laboratory automation systems obsolete within five years. Include those scenarios in your planning process.
Second, I would use an old engineer’s trick. I like to calculate my savings per day and see whether such an automation investment really puts me ahead or not.
To do this, take the annual net projected savings that the automation project is supposed to deliver and divide that by 240, which is the number of working days in the year. The resulting number is the savings per day to be expected from automation.
I compare this to two numbers. The first comparison is against the daily cost of a full-time medical technologist. The second is to calculate my billable tests per day, divide it into projected savings per day and see how much money per test I would be saving.
Comparing your savings per day from automation against both the med tech cost per day and savings per lab test will probably surprise you. Assuming that your laboratory already has overcapacity, daily savings may be minimal when viewed against the huge capital cost of the automated equipment. The cost to access that overcapacity may simply be a few extra medical technologists on staff, not $2-4 million worth of automated equipment.
Further, as an engineer, situations such as this would cause me to consider acquiring newer instruments which have multiple test capability as well as random access, then reconfigure the workflow through the laboratory to maximize existing production assets and staffing. This approach applies the engineering concepts of efficiency and performance, not productivity.
Third, before considering automation, I would apply industrial techniques known as Value Analysis and Deliberate Methods Change to the laboratory’s workflow and design. Most laboratory administrators are unfamiliar with how these proven techniques work. That is unfortunate, because these are powerful tools that can help them slash costs by 15% to 40% while improving quality and maintaining employment stability of the staff.
Although the subject of laboratory automation now gets wide exposure in the clinical laboratory industry, there is an abundance of misinformation and misunderstanding about how it works, what it does and how to use it effectively.
The goal of this article is to provide you with a new perspective on the topic of industrial automation and robotics. With a better understanding about the engineering and financial principles underlying laboratory automation, you can make better decisions.
My experience through almost four decades of work in factories and laboratories teaches me that careful decision making is the best way to save money and create a high-performance laboratory organization.
I believe in the benefits of automation. But I have seen too frequently that an ill-considered automation project spells disaster. General Motors’ decision to spend $40 billion to automate its manufacturing plants during the 1980s proved to be one of the most expensive mistakes made by an executive team in corporate history.
Until clinical laboratories with installed automation publish unequivocal data as to the financial effectiveness of laboratory automation, I would judge the current range of products as unproven in commercial use.
Changes To Technology
I do believe that technological developments will help make automation cost- effective. At the same time, I wonder how changes in the technology of medical testing may eliminate the economic justification for huge, centralized laboratories that suck in specimens from vast regions. Should “lab on a microchip” and similar technologies succeed, then large centralized laboratories may well disappear in favor of localized clusters of small laboratory sites.
With laboratory automation costing upwards of $2-4 million dollars, making a bet on today’s technology would make me uncomfortable as a laboratory administrator. That is especially true when so many accepted industrial techniques for process improvement, cost reduction and profit enhancement exist, but are unknown or unused by most laboratory administrators.
Having introduced you to the methods used by industrial engineers to look at automation, I would be extremely interested to hear from those readers who are developing automation plans for their laboratories.