Microbiological methods are less precise than analytical ones, even with the advent of rapid microbiological methods. This is because our ability to detect specifically is limited. Microbiological samples are often less representative of what is actually present in a sample, reflecting the issue of microbial culturability. Finally, microbial samples are generally less accurate than chemical ones, due to the variations with the distribution of microorganisms in the environment.
In this article, the reasons for the limitations of method, specificity, and distribution are briefly considered. The aim is to present background information about the limitations of microbial data within pharmaceuticals and healthcare, for microbiologists and non-microbiologists alike.
THE REASONS FOR THE LIMITATIONS
There are three broad groupings for limitations pertaining to microbial data: the methods used; the ability anyway to detect what might be present; and the chances of accurately detecting what can actually be detected as a consequence of microorganisms in a niche generally not conforming to normal distribution.
Limitations With Microbial Methods
Limitations with microbial methods extend to both conventional and alternative methods. By conventional methods, this refers to the culture-based methods that have stood with microbiology for over a century. By alternative methods, this includes rapid methods, and these may or may not be reliant upon culture.
With conventional methods:
- Application of the methods is of a narrow scope since conventional methods are growth-dependent enumeration methods.
- The accuracy of detection is limited.
- This can be assessed using the dilution to extinction approach, whereby a concentrated microbial inoculum is stabilized in a small amount of the matrix of interest. Subsequent dilutions are made in the matrix itself until the recovery by the alternative method becomes, at first, fractional and then progresses to all negative results. This approach does not infer, however, whether microorganisms can be actually recovered from a sample.
- The accuracy of sampling methods is limited.
- For example:
- An active air sampler is only designed and calibrated to detect 50% of the largest particle size captured by the sampling device.
- Swabs and contact plates, when optimized, will only detect up to 60% or 70% of the recoverable organisms present.
- Settle plates can be prone to desiccation.
- For example:
- The colony forming unit is limited, in that number of cells in a sample do not match the number of colonies growing on a solid microbiological plate. This occurs due to organisms coming together to form a single colony (1). A similar effect can occur on a settle plate when a particle with more than one microorganism on it lands on the plate. Furthermore, plate counting becomes more difficult as numbers increase relatively to the size of the plate (such as >250 CFU on a 9cm Petri dish) as the consequence of confluent or over-crowded growth.
- Many sample sizes are small, relative to the population volume.
- Sampling times vary and are not often event specific.
- Recovery of microorganisms is limited to the capabilities of the culture medium, incubation time, and temperature.
- The brand and variations to the formulation of culture media affect microbial recovery and growth patterns, such as the degree of swarming of specific organisms (2).
- Methods that rely on humans counting colonies can be prone to error (3), and they rely heavily upon the counting analyst's subjective interpretation of colony count (4).
With rapid microbiological methods:
- Rapid methods provide faster ‘time to result’, but this not (necessarily) synonymous with accuracy.
- The same issues are sample sizes and time apply equally to rapid methods as conventional methods.
With alternative methods:
- Alternative methods can provide greater accuracy over microbial populations, addressing the colony forming unit challenges discussed above. An example is with flow cytometry, which can be used to analyze and sort cells, and dispense precise numbers of the cells in liquid or freeze-dried forms. A second example is with auto-fluorescent cells detected by spectrophotometric methods. However, the same sampling and distribution issues arise.
Limitations of Specificity
A second limitation is that not all microorganisms within an environment can be recovered, especially using a culture-based method. This is due to organisms being active but non-culturable, either because they cannot be cultured (either absolutely or within the context of the test method) or they are culturable under ideal conditions but not on the occasion of sampling (5).
Hence, across all types of methods, there are limitations in terms of microbial recovery, in that:
- Not all organisms are culturable, due to the cultural conditions applied.
- Some organisms require enrichment to promote recovery, which requires an understanding of what is intended to be recovered.
- Some organisms are simply not culturable.
- Other organisms may be culturable under ideal conditions, but they may not be recoverable from the general environment due to damage, stress, or issues relating to acclimatization (6).
The above affects the limit of detection applied to a method. This is because, to determine the detection limit of a qualitative method, the is tested on samples inoculated (or rinse solutions) with microorganisms above and below the anticipated detection limit. The laboratory strains used may not be represented of the wildtypes in a given sample for when the method is put into practice.
Limitations of Distribution
Even if a sample should be taken from which all microorganisms present could be enumerated and identified, the sample itself may not be representative. This is particularly so with a single sample and with samples of a small volume. This is a consequence of microbial distribution and hence there will always an unobserved element from any sample taken from the larger population when attempting to assess microbial numbers and species richness.
Furthermore, there is the distribution of microorganisms within a niche (7), where the pattern does not conform to normal distribution, thereby making the accuracy of an individual sample limited (even supposing all organisms captured by the sample can be recovered and counted).While larger and multiple sampling may yield more representative results, this is invariably impractical. The distribution of microorganisms in a dilute suspension or water generally, though not always, accords well with a random (Poisson) distribution, albeit subject to different temporal and spatial dynamics (8). Even here, biofilm formation complicates distributions, especially where there is microbial community alpha diversity and interspecies relationships since multispecies biofilms are generally more resistant to disinfection than single-species biofilms (9). In contrast, the true distribution of organisms within raw materials (active pharmaceuticals and excipients) is difficult o assess because the procedures used to assess their numbers disrupt microcolonies and clumps that have resulted from cell growth or death. Quite often the distribution of organisms is to a contagious (over-dispersed) pattern (a statistical term for the presence of greater variability in a data set than would be expected based on a given statistical model) (10).
Challenges also arise for product sampling, especially when a release test is required. This because more challenging under conditions of low-level contamination, since microbial dispersion through the samples follows Poisson distribution, the rate of detection of microbiologically defective samples decreases as the number of defective units per batch decreases (11).
Distribution also creates measurement uncertainty. This is the metrological concept required to demonstrate trueness (accuracy) and precision of analytical work undertaken in laboratories. Based on the identification of repeatability (r) and reproducibility (R) for all stages in an analytical procedure followed by their combination to provide an overall estimated approaches are challenging because viable microbial numbers often fluctuate and are not constant (12).
In presenting the three limitations, it is possible that methodology will evolve further but issues of growth and issues of distribution will be ever present. The impact of growth can be contextualized if those organisms that can grow from a sample are sufficiently representative to provide an indication of the microbial levels present. Sampling can be enhanced by being tied to events or specific times and with repeat sampling and trending. However, none of these measures is ideal.
Knowing these limitations explains why ‘detection’, especially in a risk assessment context does not add much to mitigation. Instead, we need to control contamination as we cannot easily detect. Furthermore, cognizance around these limitations can help with developing sampling plans, sampling frequencies, and trend systems, as well as the appropriate forms of statistical analysis and probability theory to deploy for the assessment of relative and absolute bioburden.
- Olsen, RH and Bakken, LR (1987) Viability of Soil Bacteria: Optimization of Plate-Counting Technique and Comparison between Total Counts and Plate Counts within Different Size Groups. Microb. Ecol. 3:59-74
- Gorsuch, J., LeSaint, D., Jones, Z. et al (2020) Culture medium brand choice impacts colony swarming behavior among industrial Bacillus isolates and the accuracy of aerobic plate counts, Journal of Microbiological Methods, 172, 105891, https://doi.org/10.1016/j.mimet.2020.105891.
- Basil Jarvis, B. (2016) Chapter 7 - Errors associated with colony count procedures . In Jarvis, B.(Ed.) Statistical Aspects of the Microbiological Examination of Foods (Third Edition), Academic Press, pp119-140
- Sutton, S. (2012) The limitations of CFU: compliance to CGMP requires good science J. GXP Compl. 16 (1): 74
- Sachidanandham, R., Yew- Hoong Gin, K. and Poh, C.(2004) Monitoring of active but non-culturable bacterial cells by flow cytometry, Biotechnology and Bioengineering, 89 (1): 24-31
- Brown MRW and P Gilbert (1995) Influence of the environment on the properties of vegetative microorganisms: an overview. In Microbiological Quality Assurance – A guide towards relevance and reproducibility of inocula CRC, Boca Raton
- Franklin, R. and Mills, A. (2007) Introduction, The Spatial Distribution of Microbes in the Environment, 10.1007/978-1-4020-6216-2, (1-30)
- Potgieter, S., Pinto, A., Sigudu, M. et al (2018) Long-term spatial and temporal microbial community dynamics in a large-scale drinking water distribution system with multiple disinfectant regimes, Water Research, 139: 406-419,
- Berry, D., Xi, C., and Lutgarde Raskin, L. (2006) Microbial ecology of drinking water distribution systems, Current Opinion in Biotechnology, 17 (3): 297-302,
- Jarvis, B.(2016) Chapter 4 - The distribution of microorganisms in foods in relation to sampling. In Jarvis, B.(Ed.) Statistical Aspects of the Microbiological Examination of Foods (Third Edition), Academic Press, pp47-70,
- Eissa, M. (2017) Quantitative Microbial Risk Assessment of Pharmaceutical Products, PDA Journal of Pharmaceutical Science and Technology, 71 (3) 245-251
- Jarvis, B. (2016) Chapter 10 - Measurement uncertainty in microbiological analysis. In Jarvis, B.(Ed.) Statistical Aspects of the Microbiological Examination of Foods (Third Edition),Academic Press, pp185-194