A microbiology information system is focused on the particular information processing needs of the clinical microbiology laboratory, and should be distinguished from a general laboratory information system, which has been configured to attempt to provide functionality to support the operation of a microbiology section. However, in some cases the designers of a general clinical laboratory information system (LIS) have undertaken a separate design effort to create a comprehensive microbiology system. Our understanding of microbiology laboratory information systems has continued to evolve since the previous edition of this book [13].

20.1 In general, microbiology laboratory information systems fit one of three categories

1. Complete microbiology LISs provide the entire range of functionality needed in a clinical microbiology laboratory from accessioning, receiving, and processing to reporting of specimens for direct examinations and cultures. It includes clinical workup and reporting as well as management analysis such as quality control and quality assurance monitoring.
   
2. Instrument management and middleware modules focus on connectivity between analytical instruments and a general laboratory information system. Typically, these provide additional capabilities in the form of decision rules, report formulation, and many other features that a general laboratory information system may lack.
   
3. Specialized modules for a broad spectrum of particular tasks. Examples include infection control management and more advanced “business intelligence” management reports. These are in addition to reporting, which is standard in a comprehensive microbiology system, such as antibiogram and trend analysis reports.

In any consideration of microbiology information systems, it is important to recognize how the data generated and used in microbiology differ from other laboratory disciplines such as chemistry and hematology. We list here a few contributing factors. There are doubtless others:

1. A very high level of technical skill is required to perform bacteriology, mycology, mycobacteriology, parasitology, and virology. A huge knowledge base is required to transform raw observations in the microscope and on the culture plate into putative organism identifications.
   
2. Microbes evolve, while chemistries do not. Once rules are established to accept and validate glucose or creatinine results, those remain very stable over years, but knowledge around microorganisms changes. For example, the FDA (Food and Drug Administration) updates recommended antimicrobial susceptibility testing breakpoints for the commercial assays it approves and reviews. In addition, the CLSI (Clinical and Laboratory Standards Institute) annually publishes new criteria for what levels of antibiotic response constitute sensitive, intermediate, and resistant organisms. An issue arises when CLSI differs from FDA breakpoints. Information system decision tables must be updated for each of these, each time new criteria are published.
   
3. Timing: most chemistries produce results in a couple of hours. A microbiology culture may “hang around” for days, or even weeks. While a culture is being monitored, there is often more than one stage of reporting (preliminary, final). Therefore, the information system must have multiple “slots” into which to insert results, and the significance of those “slots” (preliminary, final) must be clearly distinguished.
   
4. Textual versus numeric: chemistry and hematology values are rather standard from one laboratory to the next, and follow well defined rules. Microbiology organisms, on the other hand, are represented with idiosyncratic abbreviations, and these textual codes will often differ between laboratories even within the same health system. Microbiology results take a variable number of characters. Often, a block of text commentary may be included.
   
5. Clusters of data are linked together in structured ways from specimen type, to culture focus (e.g., aerobic vs. viral), to colony count, to microorganisms, and to sets of susceptibilities.
   
6. There is a hierarchical arrangement of results rather than a simple linear listing of values (as you would have in a chemistry panel).
   
   
   
   1. The overall specimen is at the top level.
      
   2. For each specimen/source, there may be a series of specific types of cultures (aerobic, anaerobic, fungal, acid-fast bacteria (AFB), viral).
      
   3. For each culture there are a series of observations, which can include direct examinations or smear results (i.e. Gram stain, Ziehl Nielsen, or other stain pertinent to the culture type), plate observations (colony count and colony morphology), and isolated organisms.
      
   4. Identification tests for those organisms.
      
   5. Susceptibility testing of an organism to various antibiotics.
   
   
7. Need to aggregate data across microorganisms, and across clusters over time.
   
   
   
   1. Tracking the same organism across multiple specimens from the same patient helps assess the clinical significance of those isolates.
      
   2. If multiple bacteria are of concern, a collation of the susceptibility pattern for all of them is needed to select an antibiotic that will be active against all.
      
   3. Tracking the same organism across multiple patients is essential for infection control systems.
   
   
8. Many specialized rules must be applied. Certain combinations of organism identification and susceptibility dictate a changed interpretation. For example, if an organism is resistant to certain antibiotics, then certain antibiotics that appear effective may be contraindicated.
   
9. Limits on frequency of repeating certain assays: some laboratory assays are quite expensive, and are needed clinically only at a certain frequency. For example, it would be a waste of resources to perform an HIV viral load daily. A set of maximum ordering frequencies should be built into the order entry subsystem.
   
10. What constitutes an “abnormal” result as we feed information to the electronic health record (EHR)? This has been an issue, also, as we send results to the public health surveillance system. Typically, microbiology systems have not marked results as “normal” or “abnormal”. As long as only humans were reviewing and acting on the results, this worked fine. But in the age of automated case triage, we need to build in algorithms that will classify the typical textual results into “review not needed” and “need to review” buckets (if one prefers to avoid the “abnormal” terminology). Further, any textual result that the algorithm does not recognize should be routed to the “need to review” category. Furthermore, the microbiology section is often responsible for performing and reporting immunology and serology assays. Since the characteristics of these test results are more akin to chemistry and hematology, it may be advisable to use the “general laboratory” section of the LIS. However, this requires that microbiology staff be comfortable in rather different segments of the LIS.
    
11. Other laboratory sections have a finite list of possible outcomes versus the almost infinite scope of Bergey’s Manual for microbiology results. This implies that the automated rules will handle the common cases, and the cases not recognized would be flagged. Not only is there a wide scope of names, but also the taxonomists frequently change the name of common (and rare) bacteria and other microorganisms. While this may make good sense for molecular genetic purists, it can wreak havoc in a system that is not configured to deal with three or four differently named organisms as equivalent entities. Of course, this also adds complications to reporting of cultures, as we try to explain to our clinician colleagues that this is not a new, unfamiliar organism, but rather one that they know well, but under a different name.

20.2 What are the key features of software to support management of microbiology?

Systems must produce and handle the clustered data described above. Even if a comprehensive system has been designed, it will only evolve to usability by implementing a prototype in a real laboratory, and then incrementally improve the ability of that prototype to effectively handle microbiology data. The concept of developing theoretical specifications, writing a final system to those specifications, then implementing is ludicrous.