ON-Site Testing – Understanding and deciding

Statistical analysis can be a useful tool, but with microbial contaminations, data swings are very large. Even with a little statistical analysis of your testing program, you can understand a lot.



Regulatory compliance for prevention of infection by pathogens and overall contamination control is of course an important aspect of your business, and one of the reasons to run microbiological tests. However, it shouldn't be the only or even the main reason for testing.

Regular microbial testing, and recording the results, will allow you to build up a comprehensive picture of the microorganisms present in your environment and whether they are under control.  You can of course perform statistical analysis to obtain complex and comprehensive data on trends and performance. However, the advantage of regular microbial testing is that a contamination event or a loss of control results in a very large and noticeable swing in results, visible with the naked eye!

Maintaining regular monitoring or even an environmental monitoring program gives you a number of advantages :

  • If an auditor (customer, headquarters, inspector, …) sees data that supports practices, he may have suggestions regarding those practices, but will be sensitive to the fact there is data available and that it was put to good use
  • If you ever want to set up sophisticated analysis in the future, you will have historical data
  • If you want to conduct your own investigation into an event, any historical data you have available will most likely save you time and resources

Obtaining that valuable data is simple for DIY microbiologists like yourself!


You should define a question that your on-site testing and data collecting will attempt to answer, along with concrete actions to act on results.

Here are some examples :


A. Question : How « clean » is this tool after cleaning?

Potential actions: Intensify frequency / reinforce sanitation... or not

When to sample: after cleaning

B. Question : How much contamination builds-up during normal operations?

Potential actions: increase sanitation frequency, modify design or practices... or not

When to sample: prior to cleaning

C. Question : How efficient is the cleaning operation?

Potential actions: modify the cleaning protocol or change sanitant... or not

When to sample: prior and after cleaning


A. Question : what is the quality of the water produced by our treatment system?
Potential actions: Improve design, practices, intensify routine maintenance... or not
When to sample: after sanitation & flush of sampling point

B. Question : what is the quality of the water used by operations?
Potential actions: investigate water quality, contamination of Point-Of-Uses, sanitize piping... or not
When to sample: in normal use conditions, without sampling flush or sanitation

Also: adapt your program to the question(s) asked.

A) Random sampling: you suspect the body to control is homogeneous e.g. process water with a recirculated distribution loop
Sampling: after sanitation of sampling point & flush
Testing: Different moments in the day and in the week, from different points

B) Targeted sampling (worst case) e.g. process water
Sampling: in normal use conditions, without sampling flush or sanitation
Testing: intensified when the risk is highest, e.g. beginning / end of day, week, shut-down period, in hot summer period, just before routing maintenance
Photo of someone holding a red nomad tester and writing down the results


In an ideal statistical world, microorganisms are randomly distributed throughout your test matter!

Schema of a tank
Unfortunately in the real world, identical samples taken from a liquid won't contain the same number of microorganisms.

Schema of a tank
If you draw large numbers of identical samples from this fictional tank, your results will look like this.

Graphic showing an example of a number of results compared to colony counts
As you can see it doesn't follow a normal distribution. When the body that is being tested is not homogenous, individual results will naturally be more variable.


Collect larger samples:
With nomad Testers, the sample volume is fixed at 1 ml. testing more than that is not possible.

With nomad surface swabbing kits, larger surface areas can be tested.

Pooled samples:
Drawing several samples and pooling them together before testing will compensate to some extend for body heterogeneity

Repeat tests (replicate testing)

Run 3-5 tests on the same body and report results as the average of the 3-5 results. This practice compensates for natural distribution in results and for product heterogeneity.

You can also calculate the standard deviation which is valuable information for differentiating a normal result from a potentially problematic one

However, doing extra tests also significantly increases the cost of obtaining a result. We do still recommend this approach during the initial phase of testing, to determine baseline standard deviation and check protocol reproducibility

Finally, be sure to interpret count results knowing that two individual results that are half/double one of the other are potentially "the same" i.e. you can consider that a count of 50 is not necessarily a different result from a count of 100. "Helpfully", microbial growth typically translates in significant changes in counts!

Find out more about how you can implement nomad in your industry!


Once you have some data, you can set an alert level equal to the mean count + 3 or 5 times the standard deviation, and adjust as you get more data. Alternatively, the alert level can be set at 0.7 to 1 log above the log of your average value.

You can gradually refine your alert level once you get more data and observations under your belt on cause-and-effect relationships between events or situations and counts.

Sophisticated analytical tools can help detect slow tends or shifts and anticipate future issues.

Ideally you should be working based upon a year of historical data as microorganism growth is affected by temperature and humidity. These factors also affect the plant entrants (raw materials, air, staff) as well as the environmental contamination within the plant.Around 100 data points collected over at least a year allows solid statistical analysis.

Table showing the meaning of counts with nomad testers


We'd love to hear from you to answer any questions you might have!