Clinical laboratories are, not surprisingly, facing myriad challenges in today’s health care climate, not the least of which involve staffing. Decreased numbers of new MT and MLS graduates combined with an increase in staff retirement is not a novel problem for clinical laboratories. However, the onset of the pandemic and its enduring impact has only served to couple these realities with further staff attrition (due to a variety of pandemic-related issues) and exacerbated pressure via louder calls for faster “STAT” turnaround times and increased on-boarding of new testing to drive revenue and accommodate all things COVID. To alleviate some of this pressure, laboratory directors and leaders across the country are seeking new ways to meet the needs of their providers and patients.

In Part 1 of this article, we discussed the concepts of targeting specific microbiology tasks for automation support, bringing AI capabilities to bear in expediting microbiology workflow, total laboratory automation (TLA) systems, and current options for incorporating automated and AI-equipped instruments into the laboratory. Herein, we will discuss modular microbiology systems, further impacts of automation on specific workflow, and final considerations when reviewing and selecting microbiology automation for your unique laboratory.

Automation & AI in Modular Systems

As discussed in Part 1, currently, there are a substantial number of customization allowances built into the design of laboratory automation. For some labs, TLA systems—designed to encompass the entire microbiology workflow from specimen processing through culture reporting—may prove to be the best option. Alternatively, modular systems may be a better option for small to mid-size laboratories, as they tend to have a smaller footprint overall, and their modular and mobile configurations can be incorporated easily into existing spaces. Modular systems also tend to be less expensive than, for example, TLA systems, depending on usage. That said, despite their relatively small size, these instruments offer excellent benefits to microbiology workflow.

Some of these instruments, such as automated culture plate reading systems, may lack the front-end processing automation seen in TLA systems, yet substantial benefit can be gained through the use of artificial intelligence (AI) and machine learning capabilities within a system’s plate imaging software, for example. When cultures are loaded onto such a system, it evaluates the plates and places them into one of three categories:

1. Significant Growth
2. Review
3. No Growth (or No Significant Growth; based on the amount of CFUs present and the colony morphology)

At the same time, parameters can be set within a hospital’s LIS that allow for the autoverification and reporting of cultures designated to the No Growth or No Significant Growth category, effectively removing them from the workflow all together. This automated step enables bench techs to prioritize cultures with greater clinical significance falling under the Significant Growth and Review categories. As a specific example, when this type of automated instrument is used with cultures featuring known high-negative rates, such as urine cultures (60% negative rate)1 or MRSA/VRE screens (90% negative rate),2 this pre-sorting feature becomes even more valuable in maximizing workflow efficiencies.

Automate Removal of No Growth Cultures

Hennepin County Medical Center (HCMC) is a mid-sized, safety net hospital in Minneapolis, Minnesota. Accordingly, we opted for a modular automation system to bring technology advancement into the microbiology laboratory. While budget and system footprint certainly played significant roles in the decision, we were ultimately intrigued by the system’s AI capabilities and the possibilities we envisioned by integrating it with our other lab systems.

One of the primary goals of incorporating an automated culture plate-reading system was to gauge whether we could effectively remove No Growth cultures from our workflow, and thereby better utilize the significant skillsets of our techs. Knowing that urine cultures have a higher incidence rate of negative results compared to other sample types, we conducted a study to see how many No Growth cultures could be removed from our workflow.

The study looked at 6,200 clinical specimens that were processed through the system and reviewed the samples removed by the system via manual clinical interpretation (see the FIGURE). From the original specimen group, the system removed 1,860 cultures as true No Growth; cultures we considered as having no significant growth. Cultures that were categorized this way had less than 104 CFUs/mL and lacked clinically significant morphology (ie, no beta hemolysis/no gram-negative rods), which represented roughly 30% of our urine cases. Currently, HCMC is in the process of conducting further studies to qualify and quantify the time savings and efficiencies gained through this process. We have noticed a clear increase in the amount of time and speed with which our techs have to work on significant cultures compared to when negative cultures were read and sorted manually.

Click here to view a larger version of this FIGURE.

Among the operational results of performing this study included changes to our incubation process for urine cultures. Cultures are now separated into intervals that allow plates to be incubated for 18 hours and then read and set time frames throughout the day shift. This allowed cultures to be worked up when ready, eliminating the need for reincubation.

Key Considerations

Among the important, overarching aspects to consider when reviewing automated microbiology systems are the logistics involved in a full implementation of the system within the lab. Beginning the project with cooperation and support from your information services or LIS team is crucial to a timely and proper system implementation. Likewise, be sure to agree on the level of support you can expect from your system vendor, as they will be key in assisting with full system integration, as well as maintenance moving forward.

Automation and instrument vendors typically work with a wide variety of hospital systems and should be able to provide useful insight on best practices and troubleshooting. Finally, establish a system validation plan as early as possible (as soon as a platform is selected) to enable a smooth transition with minimal stress and disturbance to the department.

Conclusion

Seeking support through automation in microbiology is vital to addressing several current challenges. It can help alleviate pressure caused by staffing shortages, minimize turnaround times, help reduce error, and enable better utilization of skilled personnel. Furthermore, automation can increase workflow efficiencies by streamlining front-end processing, decreasing the number of staff needed to accession and prepare cultures, and redirecting them to clinically important cultures and complex testing.

Less time spent on tasks that are below the skill level of certain staff members, such as specimen processing and screening negative cultures, will help maximize the efficiency of the microbiology department. When properly reviewed, compared, and implemented, automation and related AI and software not only can improve the practice of microbiology departments, but also provide better care for patients.

References

1. Millan-Lou MI, Garcia-Lechuz JM, Ruiz-Andres MA, et al. Comparing Two Automated Techniques for the Primary Screening-Out of Urine Culture. Front Med (Lausanne). 2018;5:353.
2. Mejuto P, Luengo M, Diaz-Gigante J. Automated Flow Cytometry: An Alternative to Urine Culture in a Routine Clinical Microbiology Laboratory? Int J Microbiol. 2017:8532736.

Jacqueline Getty, MLS(ASCP), is a medical technologist in the microbiology department at Hennepin County Medical Center in Minneapolis, Minnesota. She received her MLS degree from the University of Minnesota. In addition to her clinical work in microbiology Jacqueline is involved in Hennepin County’s rapid testing protocols for COVID-19 at the hospital level.