Successful implementation will rely on standards development, education, and careful planning

Budgets are tightening and expenses are being carefully watched. Is laboratory automation a realistic way to increase productivity under those constraints?

Take a look around your laboratory. Now imagine it without any automated equipment. What would your productivity be like in a facility where all the work was done manually, without the benefits of any automation?

• The recording spectrophotometer is an example of automation. The non-automated process requires manually selecting the wavelength, reading dark currents, reading the intensity with and without the sample present, and repeating as needed.
• The strip chart on your chromatograph is an automated recording of detector output, as is the data system that captures and processes analytical data. How productive would your lab be if peak parameters were measured by hand from charts?
• No hyphenated techniques, liquid-handling systems, high-throughput screening, microplate-based assay techniques, or automated sample preparation would be available without automation.
• Your lab would be back to double-pan balances for weighing. There is a long list of automated equipment that we take for granted.
• What samples need to be worked on? … Flip through the sample log book.

The introduction of automated instrumentation, equipment, and software has had a major impact on a lab’s ability to carry out work, whether in an analytical testing lab, a materials lab, or a lab focused on primary research. Automation in the form of Web applications has sped up the process of placing orders, searching, finding products and contact information, and so on. We really wouldn’t want to go back to ordering products by phone with endless phone menus and holding.

Automated equipment has provided a significant economic benefit to lab operations, but many of us have just scratched the surface. The real benefits, both economic and functional, will come when we change our thinking about how to plan for, choose, and apply the technologies. Interest in the topic is increasing. A 2008 survey¹ shows that automation is part of the lab’s future planning: 88 percent of those surveyed said that they would be more reliant on lab automation in the future and 12 percent planned on staying at the same level of automation.

The reasons are demonstrated by two questions, one from the survey noted above, and the second from the 2006 survey.²

From the 2006 survey: What reasons do people give for their interest in lab automation?

The 2008 survey looked at the factors driving labs to consider automation technologies.

In addition, the 2008 survey found that 74 percent of those responding wanted to use lab automation to streamline operations, and 71 percent wanted to add new capabilities. If you change “Improve Data Quality” to “Improve Product Quality” (in many labs, the data and information produced is part of the product), you’ll find yourself with a set of management pressures and requirements that are common to any production facility. And in many cases, automation is a practical means of meeting those demands. But does it work in laboratories? One good example shows it can work.

Clinical laboratories have a lot in common with testing and analytical labs in other industries. Samples are submitted, tests are scheduled, the analysis is performed and the results are reported. There are three major differences. The first is that their samples are drawn from us in the form of fluids, tissues, etc., that are evaluated according to standardized methods. Secondly, their charge rates are set by contract or the federal government and they have to operate within that limit. The third point we get to later. The response of clinical labs to increasing sample loads with a fixed cost structure—not dissimilar from the points noted above—was to automate their protocols. The results are impressive.³ The following bullets are quoted from the article’s abstract:

• Between 1965 and 2000, the Consumer Price Index increased by a factor of 5.5 in the United States.
• During the same 36 years, at our institution’s chemistry department [Mt. Sinai Medical Center], the productivity (indicated as the number of reported test results/employee/year) increased from 10,600 to 104,558 (9.3-fold).
• When expressed in constant 1965 dollars, the total cost per test decreased from $0.79 to $0.15.

Another report,⁴ from The Ohio State University, gave an overall productivity labor increase of 26 percent with an increase of 72 percent in the number of specimens processed per Full Time Equivalent Person. These improvements are due in large measure to an industry shift to the concept of “Total Laboratory Automation.”⁵ An article⁶ by George Streitberg discusses the impact that the concept has had at the Monash Medical Centre in Melbourne, Australia. The article covers the technology, changes in personnel duties, training, etc. The bottom line is this: Properly done, automation works in the laboratory environment.

“Properly done” is the key element. In the clinical laboratory industry the needed work was done to establish a framework for communication that allowed systems, including instrument-data-to-laboratory-information systems, to exchange information. That standard “glue” holds things together and makes them work; that is the third key difference between the clinical lab environment and the environments we commonly encounter in analytical, testing, pharmaceutical, biotech, materials characterization, and other labs.

If you’d like more evidence, look at the ease of microplate-based assays. Samples are processed and read using a variety of devices from a mix of vendors that work on a standardized sample format.

And for the rest of us…

For those not working in a clinical lab setting, how does this apply? The work on clinical laboratory and hospital information standards goes back to 1987 with the initiation of the HL7 program (www.HL7.org), which continues its work today. In addition, three ASTM standards were designed for clinical laboratories:

• ASTM E1238 – Standard Specification for Transferring Clinical Observations Between Independent Computer Systems
• ASTM E1381 – Specification for Low-Level Protocol to Transfer Messages Between Clinical Laboratory Instruments and Computer Systems
• ASTM E1394 – Standard Specification for Transferring Information Between Clinical Instruments and Computer Systems

These ASTM standards provided the initial basis for standardizing instrument-to-LIS communication.” (Note: Clinical people use LIS where industrial people use LIMS.) The successful implementation of lab automation in clinical labs provides a model for what could be done in your lab (the ASTM 1394 standard carries the note that it “does not necessarily apply to general analytical instruments in an industrial analytical or R&D setting.”) These ASTM standards have been replaced with work by the Clinical Standards and Laboratory Institute (CSLI - www.clsi.org) in the form of the “Laboratory Automation: Communications with Automated Clinical Laboratory Systems, Instruments, Devices, and Information Systems” electronic document. This approved standard is in its second edition. From this author’s point of view, the structure of the standards is an appropriate model, although the specific language and terminology will differ.

Three issues have to be addressed:

• Standards development
• Education
• Planning for automation

Standards Development

First we’ll look at the long-term consideration of standards development. The need for standards is addressed along with several other points in the web article “Integration in Laboratory Automation & Informatics.”⁷ If we are going to bring the maturity to lab automation necessary to allow products to exchange information without the need for custom coding, we have to move toward standards, and we must do so rapidly. Since the operational structure of clinical labs is similar to those of other testing and analytical labs, it may be possible to jump-start a standards development program by building on the HL7 and ASTM / CSLI work; this should be explored, and those interested in pursuing this should contact the author. Consider the following quote: “These individuals also believed that the total market for automation systems and equipment would be significantly greater with standards than without standards, especially if customers were not forced to purchase everything from one vendor, and that there might be competitive pricing and new technology fostered via the standards.”⁸ Sound like something you might believe? That was part of the rationale for developing the data interchange standards at the lab level for the HL7 program, which includes the same instruments as LIMS, and LIMS to management databases [note: the ASTM standards noted above are frequently referenced as a data interchange methodology]. Developing a similar standards program would dramatically improve the economics of lab automation. This isn’t wishful thinking—it has been done successfully in one significant area of work.

The alternative is to continue developing programs that may forge a connection between an instrument or data system and a LIMS or Electronic Laboratory Notebook (ELN) that works, possibly at considerable expense, but ties you not only to a specific product but to a specific version of that product as well, since upgrades may obviate that work and require that it be re-implemented. While the idea of instrument-to-LIMS/ELN communication is nice, doing it outside the facilities provided by the vendor (vendor-supported instrument interfaces) is both risky, in terms of project failure, and expensive.

Education

Education is a major differentiator between successful lab automation programs and those that fail. Education includes user-level material (working with lab automation systems, what does a LIMS/ELN do, etc.) as well as management-level material that looks at how to plan for automation, implementation considerations (including project management), product life cycle management and its impact, regulatory issues, and so on.

The cost of projects can be severely impacted by failing to fully understand the ramifications of decisions such as linking instruments to LIMS/ELNs outside the capabilities provided by the vendor, by not properly defining the scope of a project or missing key requirements, and by not fully evaluating the range of technology options. For example, in the LIMS/ELN environment, implementations can be the traditional system on site or perhaps the increasingly popular software-as-aservice model in which the application is run on external equipment. Both implementations are viable options, but corporate considerations may make one more appropriate than the other.

If we use the clinical environment as a potential model for what an automated lab can be, then we need to look at how it changed the nature of lab work, and how that change could be reflected in your lab. Much of laboratory work consists of carrying out tests and experiments. In a fully automated lab—including a research lab—much of that effort will be done by systems. We can see elements of this today in fully automated microplate-based assays. The tasks will change from manually conducting the work to planning work to be done, making sure systems are performing properly, doing data analysis, and doing what laboratory people do best: thinking, developing, and being innovative. This was the promise held out all along by lab-automation advocates.

An investment in education will pay for itself many times over in more-effective lab personnel, betterplanned programs, improvements in lab operations, and better science.

Planning for Automation

There are several ways of approaching lab automation. One is the gradual introduction of automated equipment into the lab, replacing manual tasks, such as liquid handling, with more-efficient automated components. This equipment can speed up aspects of lab work, but people still make the process work.

Another is to make a commitment to automation and reevaluate the procedures you are using: Are they suitable for automation? Are there steps in the procedure that, due to materials handling or the nature of the processing, prevent the use of automation, and if so, are there alternative methods that may lend themselves to automated systems? Within your industry, is it possible for several companies to evaluate and standardize procedures for automation (this is a role that ASTM fills for some applications)? This latter point is one of the elements that has led to automation successes in the clinical lab and can guide your own developments.

Laboratory economics is not just about saving money. It is about the efficient and effective use of resources, especially those working in the lab.

References

1. Hamilton, S. D. 2009. 2008 ALA Survey on Laboratory Automation. Journal of the Association for Laboratory Automation (JALA) 14: 308-319.
2. Hamilton, S. D. 2007. 2006 ALA Survey on Laboratory Automation. Journal of the Association for Laboratory Automation (JALA) 12: 239-246.
3. Sarkozi, L., Simson, E., and L. Ramanathan. 2003. “The Effects of Total Laboratory Automation on the Management of a Clinical Chemistry Laboratory. Retrospective Analysis of 36 Years.” Clinica Chimica Acta 329: 89-94.
4. Bissel, M. 2007. Total Laboratory Automation. http:// www.researchchannel.org
5. Young, D. 2000. “Laboratory Automation: Smart Strategies and Practical Applications.” Clinical Chemistry 46(5):740-745.
6. Streitberg, G. et al. 2009. “Automation and Expert Systems in a Core Clinical Chemistry Laboratory.” Journal of the Association for Laboratory Automation (JALA) 14: 94-105.
7. Liscouski, J. G. 2009. Integration in Laboratory Automation & Informatics. http://theintegratedlab.com/ integration-in-laboratory-automation-informatics/
8. Hawker, C. D. and M. R. Schlank. 2000. “Development of Standards for Laboratory Automation.” Clinical Chemistry 46(5): 746-750.