A Four Year Review of Legionella and Plate Count Analyses of Cooling Tower Water
G. Pinna B App Sc (Med Tech) B Sc MASM MAIRAH
Manager of Biotech Laboratories Pty Ltd, Brisbane, Queensland.

This paper is a review of the laboratory analysis of cooling tower water samples for Legionella bacteria by AS3896 [2] and heterotrophic bacteria by Plate count (PC). Samples included in the review were received for processing between 1/4/92 and 31/3/96. During the review period 28,898 samples were processed for Legionella analysis, of these 3,478 (12%) had detectable levels of Legionella equal to or above 10 colony forming units per millilitre (CFU/ml). A total of 49% of legionellae detected were equal to or above 100 CFU/ml. The highest level detected was 270,000 CFU/ml. The isolation rate peaked in the summer and autumn months. Of those samples with detectable legionellae levels, 1,988 had a PC analysis also performed. Of these, 32.7% had a PC of less than 10,000 CFU/ml. The results of this review suggest that there is no correlation between PC and the presence of legionellae. The evaluation of cooling tower water contaminated with legionellae using the Legionella/PC ratio in addition to the level of contamination may facilitate a more accurate rick assessment protocol. Further it is noted that maintaining a cooling tower to the Australian Standard AS3666 [1] and the Standards Australia handbook HB32 [10] alone is not sufficient to guarantee a continued absence of legionellae. The data presented showed that by performing monthly Legionella analyses the average levels of a positive detection was lower (90 CFU/ml) compared to quarterly analyses (972 CFU/ml). It is therefore recommended that monthly Legionella analysis and PC be incorporated into the preventative maintenance procedures of cooling tower water systems.



Sample collection and transport
Samples were collected into sterile containers, volumes received ranged from 50 to 250ml. All samples were collected and held prior to transport under the control and supervision of the client. Clients are advised to refrigerate samples prior to transport and to transport samples for LA at ambient temperature as described in AS3896 [2]. Some clients chose to transport samples under refrigerated conditions.

Sample processing
Samples were refrigerated upon receipt at the laboratory prior to processing. Samples received before 7 pm (Monday to Friday) were processed on the same day.

Methods of analysis
Legionella analysis - The method used for LA was the unmodified Australian Standard method AS3896 [2]. The biocide reduction step used was the centrifugation method (10ml centrifuged for 10 minutes at 6,000g) rather than the filtration method. Legionellae isolated were identified as L. pneumophila serogroup 1, L. pneumophila serogroup 2-14 or Legionella species (not pneumophila). This is a National Association of Testing Authorities (NATA) accredited method.
Plate Count analysis - The method used for PC was the spread plate method of the American Public Health Association (APHA Standard Method 9215 Water and Waste Water) [3]. Results were reported in the range <1,000 to >1,000,000 CFU/ml. This is a NATA accredited method.


Aim of Review
This review evaluates the laboratory findings of LA and PC of cooling tower water samples for:

• the examination of the legionellae isolation rate by the AS3896 method.
• the examination of the relationship between LA and PC.
• the evaluation of LA as a routine procedure in the preventative maintenance protocol of cooling towers.

Results of Legionella Anaylsis

Of the 28,898 samples processed for LA in the review period 3,478 (12%) had detectable levels of legionellae equal to or greater than 10 CFU/ml. The range of levels detected was from 10 to 270,000 CFU/ml. In general a high proportion of samples received are from towers which are tested quarterly. The average legionellae level of a positive detection was 972 CFU/ml. A subgroup of samples received from 25 cooling towers which were tested monthly during the review period had an average isolation rate of 10.1%, but had an average legionellae level of a positive detection of 90 CFU/ml. Using the groupings described by the NSW Health Department [8], Table 1 shows the breakdown of legionellae levels for the positive samples.

Table 1. Legionella levels detected.

 Table 1


The NSW Health Department guidelines classify Legionella levels between 100 and 1,000 CFU/ml as potentially hazardous and levels greater than or equal to 1,000 CFU/ml as a serious situation requiring immediate shutdown of the system for decontamination [8]. This grouping is similar to that proposed by Morris and Feeley [7] to the American Society for Heating, Refrigerating and Air-Conditioning Engineers, Inc. annual meeting in 1990. The proposed risk categories are as follows:

Legionellae (CFU/ml) Risk category

•<10 Very low
•10 - 99 Low
•100 - 999 Moderate
•>1,000 High

As shown in Table 1, approximately half of the positive water samples had legionellae detected at potentially hazardous or serious contamination levels. Table 2. shows the speciation of legionellae isolated.

Table 2. Legionella speciation.

Table 2

At present there are over 25 species and 48 serogroups of Legionella recognised, a substantial proportion are capable of causing disease. The United States public health authority, Centers for Disease Control, reports that when an isolate is recovered from a disease outbreak, 90% of cases of legionellosis are due to L. pneumophila [4]. Of these cases 82% are caused by L. pneumophila serogroup 1 [4]. Legionella micdadei, L. pneumophila serogroups 6 and 3, and L. longbeachae are the organisms isolated next most frequently from patients with legionellosis [4]. The Queensland Health department figures for the period 1988 to 1992 showed that 37% of legionellosis cases were due to L. pneumophila, 35% to L. longbeachae and 16% to L. bozemannii [9]. Based on both these findings it can be noted that the species of legionellae or the serogroup of L. pneumophila isolated from a cooling tower should not influence the interpretation of the laboratory findings. Most laboratories will report isolates as L. pneumophila serogroup 1, L. pneumophila serogroup 2 to 14 or Legionella species (not pneumophila). As shown, pathogenic legionellae have been reported from each of these groups and therefore the risk assessment should, at present, only be based on the concentration of the legionellae reported. To further influence this method of evaluation, our laboratory classifies positive legionellae isolates as either L. pneumophila or Legionella species (not pneumophila).

The monthly isolation rate ranged from 2.4% (October 1993) to 23.8% (November 1995). As seen in Graph 1, the isolation rate for legionellae peaked in the summer and autumn months.


 Table 3a


Results of Standard Plate Count Analysis
Of the 3,478 cooling tower water samples which had detectable levels of legionellae; 1,988 had a PC performed. Table 3 shows the distribution of the PC results of the water samples that had a positive LA and a PC performed.

Table 3. Standard Plate Count results of Legionella positive samples.

Table 5a

The Standards Australia publication HB32 [10] states that an acceptable level for PC of cooling tower water is below 100,000 CFU/ml. As shown in Table 3, 73.5% of cooling tower water samples with detectable levels of legionellae had a PC of less than 100,000 CFU/ml. During the review period it was not uncommon to have a Legionella count higher than the PC for a given sample. It should be noted that legionellae and other fastidious bacteria are not detectable by the PC method. An example of this was a sample with a Legionella count of 16,000 CFU/ml and a PC of 2,000 CFU/ml, this has also been reported by other workers [6]. These findings should not be interpreted as an inverse relationship between PC and LA. The increase in detection of legionellae at lower PC levels reflects the sensitivity of LA. The standard method used [2] utilises heat treatment, dilution procedures and selective agar plates to minimise the adverse affects of other bacteria present in the water sample. Despite this, the fact remains that these procedures are not perfect and contaminating bacteria are often grown on the Legionella culture plates. Where the PC is high, and other bacteria are not suppressed the sensitivity of the LA is reduced and a false negative result may be reported. It would therefore be of assistance to have a PC performed in unison with each LA.


Legionella/Plate Count Ratios
Miller and Kenepp [5] proposed that the risk status of a cooling tower be based on the ratio of Legionella to PC rather than solely on the CFU/ml of Legionella present. This was based on two observations; first, it was noted that legionellae were rarely observed in environmental specimens in excess of 1% of the total bacterial population; second, they observed that high levels of legionellae were occasionally found in towers with very few other detectable bacteria present [6]. Therefore the Legionella/PC ratio may reflect the tendency of the water system to select for legionellae. This may be due to variations in the resistance of biocides used to treat the water system. The distribution of the review data using the categories preposed by Miller and Kenepp. is shown in Table

Table 4. Legionella / Standard Plate Count ratios.

Table 6a

Using this criteria, 3% of samples that had detectable levels of legionellae and PC performed, would fall into the highest risk group. However, using the present risk categories based solely on CFU/ml levels, 12.3% of samples in this same group would be classified into the high risk category (ie >1,000 CFU/ml). On the basis that legionellae are found in a relatively high percentage of cooling towers in the absence of disease, it would appear the concentration of viable legionellae alone may be overestimating the risk associated with high levels. The criteria proposed by Miller and Kenepp [5] reflects the selection of legionellae in the cooling tower. A high Legionella/PC ratio (>50%) indicates a reduction of microbial competition which could lead to a rapid increase in Legionella overgrowth and perhaps this criteria should be considered as an additional factor when evaluating the risk based on laboratory findings. Using both Legionella/SPC ratio and the legionellae level risk assessment methods on this subgroup of samples, only 1.5% would be classified into both high risk categories. The concern with any laboratory based risk evaluation process is that it may underestimate the importance of low or moderate levels of legionellae and overestimate the risk associated with high levels. Also, all in vitro risk assessments ignore the physical characteristics of a cooling tower such as efficiency of drift eliminators, design, size and position in relation to air intake ducts or the public access areas.


During the review period an increase in the Legionella isolation rate for the summer and autumn months was observed (Graph 1). During the period from December to May the highest isolation rates were recorded. Further investigation of the variation in monthly isolation rates is required.

These investigations may further examine the effect of environmental factors such as temperature, precipitation and humidity on the colonisation of cooling tower waters with legionellae.

At present the three groups of legionellae reported routinely by testing laboratories (L. pneumophila serogroup 1, L. pneumophila serogroup 2 to 14 and Legionella species (not pneumophila) have all been implicated in disease processes [4,9]. Based on this, the risk assessment of a cooling tower contaminated with legionellae should, at present, be based solely on the level of Legionella present and not on the species or serogroup of the isolated microorganism.

The data presented shows that a low or an acceptable PC level of cooling tower water does not indicate that the system is free of legionellae (Table 3). That is, a PC test of a cooling tower should not be used as an indicator of the Legionella risk status of a tower. A PC should be used, however, as an indicator of the microbial load of a tower in relation to current acceptable standards. A PC should always be performed in unison with a LA, as a high SPC may reduce the sensitivity of the LA.

The routine analysis of cooling tower water for the presence of legionellae remains a controversial topic. On the basis that the bacteria is found commonly in cooling towers in the absence of an associated Legionnaires' Disease outbreak, most Australian public health departments do not recommend routine testing for this microorganism. However, building owners, managers and water treatment companies routinely have these tests performed. The ever-present threat of a disease outbreak and the associated legal and financial liabilities warrant routine risk assessments to be made. The data presented would suggest that reliance on maintaining cooling towers to the Australian Standard AS3666 and the Standards Australia publication HB32 does not prevent contamination of the system with legionellae. We would therefore recommend that the relevant authorities review the present guidelines with respect to including LA in the preventative maintenance protocol of cooling towers.

On this basis it should be noted that the interval between analysis must be selected carefully. If quarterly testing is performed, results provide only a random check for the presence of legionellae. It is our experience that inside a three month period legionellae levels may increase rapidly to high risk levels and fall to acceptable levels prior to the next analysis. This is reinforced by the results showing an equivalent isolation rate between towers tested monthly as compared to quarterly, but a dramatic decrease in the average Legionella level of a positive sample for the monthly analyses towers (90 CFU/ml as compared to 972 CFU/ml). To approach a system for monitoring the presence of legionellae, a monthly LA would be required. As the method requires a 10 day incubation, testing inside a four week period could be of limited use except to monitor a known positive cooling tower during the decontamination process. It should also be noted that a PC should be performed routinely with each LA to enable accurate interpretation of both negative and positive results.

Further it is proposed that research be initiated to evaluate the usefulness of Legionella/PC ratios as an additional factor for evaluating the risk associated with legionellae contaminated cooling towers. This would require a review of previous laboratory data for cooling towers implicated in Legionnaires' Disease outbreaks in Australia and overseas.


1. Australian Standard: AS3666-1989 "Air-handling and water systems of buildings- Microbial control" (1989).

2. Australian Standard: AS3896-1991 "Waters-Examination for legionellae" (1991).

3. APHA: Standard Methods for the Examination of Water and Wastewater 17th ed., American Public Health Association, p9-61, (1989).

4. Marston, B.J., Lipman, H and Breiman, R.F.: A decade of surveillance for Legionnaires' Disease: update on risk factors for morbidity and mortality due to infection with Legionella. Submitted for publication.

5. Miller, R.D. and Kenepp, K.A.: in Legionella Current State and Emerging Perspectives, Editors: Barbaree, J.M., Breiman, R.F. and Dufour, A.P.: American Society for Microbiology, pp 40-43, (1993).

6. Miller, R.D. and Kenepp, K.A.: Aerobiology, Abstr. Pan-Am. Aerobiol. Assoc. Annu. Meet., p20, (1991).

7. Morris, G.K., and Feeley, J.C.: Abstr. ASHRAE Annual Meeting, p76, (1990).

8. NSW Health Department: Code of Practice for the Control of Legionnaires' Disease, New South Wales Health Department, p35, (1991).

9. Pearce, M.: "Legionellosis in Queensland - A Review" Communicable Diseases Intelligence, Vol.16, No.14, p296. (1992).

10. Standards Australia: HB32-1992 "Control of microbial growth in air-handling and water systems in buildings" (1992).