Environmental monitoring and assessment pdf

To assess comparability of these methods, quantify withinlaboratory variability for each method, and place that variability into context of variability among laboratories using the same method, 22 southern California laboratories participated in a series of intercalibration exercises.

Each laboratory processed three to five replicates from thirteen samples, with total coliforms, fecal coliforms or enterococci measured depending on the sample. Results were generally comparable among methods, though membrane filtration appeared to underestimate the other two methods for fecal coliforms, possibly due to clumping. Variability was greatest for the multiple tube fermentation method. For all three methods, within laboratory variability was greater than among laboratories variability.

Keywords: microbiology, intercalibration, variability, bacterial indicators, beach water quality 1. Introduction Coastal beaches are the subject of extensive water quality monitoring to detect fecal contamination from human activities, such as wastewater discharge, industrial input, and surface runoff. Included in many of these monitoring programs is measurement of indicator bacteria, such as total coliforms, fecal coliforms, and enterococci. While indicator bacteria are not necessarily pathogenic, they are found abundantly in human wastes, where pathogenic organisms may exist.

Bacterial indicators are used in preference to direct tests for pathogens because bacteria indicator measurements are less expensive and correlate with the incidence of illness in swimmers Cabelli, ; Haile et al.

IDEXX , have recently been gaining popularity. The three methods each are based upon measuring different products of bacterial growth. MF enumerates bacterial colonies on a specific growth substrate. MTF measures metabolic activity as determined by fermentation and the production of gas. Chromogenic methods measure the ability of organisms to metabolize a specific labeled substrate, thereby releasing a chromogen. These differences in analytical endpoint provide the potential for differing results among methods.

Environmental Monitoring and Assessment —, Printed in the Netherlands. No study has quantified response among all three methods, nor placed differences among methods into context of variability among laboratories that use the same methods.

Furthermore, no study has compared measurement precision among the three methods. IDEXX kits have the advantages of being less expensive and faster than the historically used methods, but these advantages are of little value if the results produced by IDEXX are not comparable to that from the historic methods.

Here we present a series of intercalibration studies that were conducted among 22 southern California laboratories. We use these studies to assess comparability of results among the three bacterial indicator measurement methods, quantify within laboratory variability for each method and place that variability into context of variability among laboratories using the same method.

Methods Thirteen experiments were conducted on five separate occasions Table 1. The first set of experiments involved quantification of total coliforms in transport media at three bacterial indicator concentrations.

The second set of experiments involved quantification of fecal coliforms [or Escherichia coli E. The third set of experiments involved quantification of enterococci in transport media at three bacterial indicator concentrations. The fourth set of experiments involved quantification of total coliforms and fecal coliforms or E.

The fifth set of experiments involved quantification of fecal coliforms or E. Ten of the thirteen experiments were performed by seeding 24 hour-old stock cultures of E. Transport media was prepared prior to the day of the experiment in two-liter volumes and sterilized. Carboys were sterilized separately.

Bacteria were added to the transport media and mixed for twenty minutes on a magnetic mixer prior to dispensing the first sample. The amount of stock culture necessary to achieve the target concentrations was based on MF analyses from the preceding day.

In experiments 5 and 10, stock cultures of E. Median bacterial indicator count and standard deviation in each experiment. Numbers with the same letter code are not significantly different. Before inoculating the seawater sample, the primary treated wastewater was filtered through Whatman Grade filter paper to remove fine particulates. On the morning of each experiment, the contents of the carboys were aliquotted into subsamples, which were packed in ice and distributed to the participating laboratories.

All laboratories began their analyses at the same time, approximately five hours after the stock sample was prepared. The originating laboratory analyzed the first and last sample dispensed from each carboy by MF and MTF to validate homogeneity of samples aliquotted from the carboy. Each laboratory used its own standard operating procedures.

Three to five replicates were performed for each indicator at each concentration. Several laboratories used more than one analytical method, which resulted in more than 22 sets of results for some experiments. To test the hypothesis that the median within-laboratory values were the same among methods, we performed an ANOVA on ranks, separately for each experiment Conover and Iman, A Bonferoni adjustment to significance level was used to account for pair-wise testing of methods.

Similar analyses were performed to test for difference in within-laboratory variance among methods. After log transforming the data, least squares regression was used to model the within laboratory variances as a function of within laboratory means.

Confidence intervals were then back-transformed to original scale. Separate regressions were performed for each indicator. Results Comparability of bacterial densities measured among methods was indicator specific Figure 1; Table 1. For enterococci and total coliforms, we saw no significant differences among methods in any of the individual experiments, though we did observe that MTF results were higher in five of the seven experiments.

We also observed that IDEXX yielded the lowest median count in all three of the experiments conducted with enterococci. Log-transformed bacterial density vs. Laboratory Lab number.

Laboratory numbering was random and arranged by method. Experiments 1—6 are for Fecal Coliforms. Experiments 7—10 are for Total Coliforms. Experiments 11—13 are for Enterococci. The variability associated with MF was generally a little lower than that with IDEXX, but not significantly different in any of the individual experiments. The coefficient of variation appeared to be independent of which bacterial indicator was measured.

The confidence interval for MTF was about twice as large as for the other two methods Table 2. The confidence intervals are all asymmetric around the threshold, reflecting the lognormal distribution of the data.

Also, the confidence intervals presented in Table 2 are all specific to the threshold concentration, as the data displayed a significant variance: mean ratio.

Table 2. One price for increased consistency, though, can be loss in flexibility. For instance, California, which has the most beach monitoring in the United States Schiff et al. Examination of data from Los Angeles County reveals that a majority of the single sample standard exceedences for the five-year period between and Schiff et al. These findings suggest there is a great deal of uncertainty associated with warnings based on a single sample standard.

The magnitude of our variability estimates, as well as our finding of higher variability for MTF, is consistent with that of previous studies Fleisher, MTF is a most probable number technique, which is a statistical estimate based on the percentage of test tubes eliciting a positive response to the presence of bacteria.

As the estimate is based on a binomial distribution, its variance is primarily a function of the number of tubes used. As detailed earlier, environmental monitoring is performed inside the isolator during testing. This monitoring, which has an action level of 1 CFU Colony Forming Unit , is designed to detect any potential contamination inside the isolator environment.

Then, by using a largely numerically- driven set of tools, repeatability and reproducibility can be ensured. Examples of individual out-of-limits results and data-sets relating to an operation are examined below using examples from an aseptic filling process. Following this, an example of an overall assessment of different processes over time is explored. Numerical approaches are useful in applying a level of consistency between one decision and another.

Individual Assessments The section below details some methods that can be used to quantify the risk of contamination in pharmaceutical cleanrooms. The models out- lined are based on the work performed by Whyte and Eaton a and b. There is no available guide as to what percentage constitutes which level of risk. The 0. This is based on the Parenteral Drug Association Survey of Aseptic Filling Practices , where it is common in the pharmaceutical industry to allow 0.

This would constitute a high risk. Logically, the range between 0. Where the operator is only present in the Grade B room and has no impact on the Grade A operation, this is automatically considered to be low risk if there are no other special factors.

Low risk does not imply lack of action or assess- ment. However, it aims conceptualisation of the result in terms of probable risk to the batch. The bottles and vials are partially stoppered and utensils are normally used. The likelihood of contamination is considered to be low. On-load 1. The bottles and vials are not stoppered, although utensils are normally used. The likelihood of contamination is higher than for off-load.

Stopper bowl 1. A direct intervention into the bowl could result in micro-organisms being deposited onto stoppers. The risk of this is considered higher than the risk with on-load or off-load activities, although such an intervention is rare. Freeze-dryer loading 1. However, vials and bottles are partially stoppered and are contained with cassettes.

Point-of-fill: air sample 2. However, as a direct intervention into the Grade A zone, it is a higher risk than those parts of the filling machine previously examined. Filtration transfer 2. If this process becomes contaminated, this could affect the product. The time taken to perform the connection is normally very short under 30 seconds , which reduces the risk.

Machine connection 2. If the transfer line is contaminated, this could cause contamination to the product. Point-of-fill: 2. Counts associated with such activities require detailed examination. Hand 1. However, it is procedure to sanitise hands prior to undertaking the operation. Plus seconds 1. Microbial count x Location x Method of intervention x Duration of operation 1 x 2. Based on historical data over the past six-months, the highest record example of a Grade A intervention finger plate is a count of 2 cfu: using forceps to retrieve a fallen vial and lasting for more than seconds.

This would have given a score of 7. The user should develop a scheme that fits his or her facility Whyte and Eaton, b. The logic demonstrated in Figure 9 may be used to determine risk factor A. General filling room 0. In this event, the risk factor increases to 1. Machine general 1.

In this event, the risk factor increases to 2. Machine product 2. See Figure 10 for an example of the reasoning that would support Risk Factor B.

General filling room or 1. Risk of transfer is low. Some risk of transfer exists, but protective measures should prevent this. Direct contact with product: highest risk. See Figure 11 for an example of this assessment. Barrier exists by way of filling or filtration room area machine doors and UDAF. Grade B Machine general 1.

Machine product 1. A count of 1 cfu on one of these product contact site locations would give a score of 9. In most filling zones and clean zones, sample results from product contact sites would be expected to record zero counts for samples out of every Whereas, a count of 3 from a non-product contact site would result in a medium risk category.

However, the formulae associated with these are difficult to calculate in practice because often all information is not available and assessment of variables, such as impaction speed, are not readily calculable. Therefore, a qualitative assessment, such as the one included in the example of the numerical approach, may be more suitable.

See Figure 12 for this example. The location can be given a risk rating in relation to its proximity to the critical zone, ease of dispersion or transfer, and effectiveness of control methods.

The table shown in Figure 13 is proposed as a tool for risk assess- ment and to aid investigations. It supplements the risk assessment tools that have been previously examined. See Figure 14 for an example. Figure 14 Product-related Risk Assessment Factors Product Rank Reason Freeze-dried 1 Freeze-drying will destroy most micro-organisms Liquid product, heat treated 2 Undergoes pasteurisation - effective against most non spore-forming micro-organisms Intra-muscular product 3 Small volume; intra-muscular route Intravenous product, 4 Intravenous route; no further processing no further treatment An Overall Assessment The approach taken for an overall assessment involves the historical examination of a number of operations and assigning a value above which the operation is considered to be atypical.

This may be, for example, a batch fill. The data from the session is examined and points are awarded for each result above a pre-set warning or action level. The total score is then summed and the results obtained are compared to a set level at which atypical sessions are indicated. The pre-set level would be assessed from historical data over a reasonable time period such as one year. An example of such a scheme follows: For Grade A The results from a filling operation are examined for individual viable counts and for the mean particle counts taken during the fill.

Using the criteria presented in Figures 15 and 16 produces the Grade A score. For Grade B The results from a filling operation are examined for the individual viable counts and the mean particle counts taken during the fill.

Each result that equals or exceeds a warning level is scored according to the criteria in Figure 17 and Figure This produces the Grade B score. Figure 19 displays a simple representation of this assessment. For the data set, the criticality score was calculated at 25, which corresponded with the 95th percentile for a set of data from the filling of an example drug.

The graph in Figure 19 indicates that some fills exceeded the cut-off criticality value during a particular time period see fill numbers 12 through