BASELINE software tool for calculation of microbiological criteria and risk management metrics for selected foods and hazards WP6 Model Development Final.

BASELINE software tool for BASELINE software tool for calculation of calculation of

microbiological criteria and microbiological criteria and risk management metrics for risk management metrics for selected foods and hazards selected foods and hazards

WP6 Model DevelopmentWP6 Model Development

Final Conference BASELINE . Bologna 11-12 November 2013

Baseline Software tool: Data and Figures


58 users registered in the application

2% 2%

7%8% 5%

3%

15%

2%2%2%2%

50%

Brazil Canada Croatia France

Germany Greece Italy New Zealand

Norway The Netherlands Sweden Spain



Laboratories, R&D institutions, Universities and Official authorities

0 5 10 15 20 25

Labs

Public institution

R&D Technology Center

R&D company

Universities



Current position of Baseline software users

0

2

4

6

8

10

12

14



Main intended use: dissemination, training, teaching, research and training activities and official control.

It was presented at the International Conference on Predictive Modelling in Foods (ICPMF 8), Paris (France) being selected at the top five software tools

Webinar and training sessions were perfomed over 2013 (Oslo, Northern Spain, Bergamo)

EFSA workshop (September 26th, Parma)

Improvements and upgrades were carried out related to terminology, units and equations.

DERIVING AN MC FROM A PO THAT IS SET AS CONCENTRATION LIMIT OF THE PATHOGEN

L. monocytogenes in cold-smoked salmon

Raw material (fresh fish)

Manufacturing

Treatment Storage Distribution Consumption

PO PO PO FSO=100 cfu/g

PC

It is established a maximum concentration level of 100 cfu/g before consumptionFor simplification, PO is set after product elaboration / storage

It is assumed that a Competent Authority has established a PO for the concentration of a specific pathogen in a certain commodity.


Example I

Example I

L. monocytogenes in cold-smoked salmon: Input data

Initial concentration: just after packaging ~ 10-20 cfu/gStorage in the industry at 4ºC during 4 days (96h)Product formulation: 2ppm phenol + 3 mg/100g NaCl


8

Example II


9

Example II


10

Example II


Final concentration : 1.74 log cfu/g


Example I


Estimate the standard deviation of the product in such a way the following PO will be complied

P (log cfu/g >3 ) < 5 % of the samples conforming the lot

Using the NORMDIST function of MS Excel we obtain:

=NORMDIST(3; 1.74;σ; 1) σ ~ 0.76 log cfu/g

Therefore, the distribution for the concentration of Lm satisfying the PO would be log normal (1.74; 0.76)

One can use the ‘Solver” function by changing values for ‘standard deviation’, when starting with an unknown value for ‘standard deviation’ for a known probability (target cell value equal to 0.95)]


Example I


Establish a microbiological limit so that a practical and feasible sampling plan can be applied In this scenario, the value for m is chosen to be 2 log cfu/g (i.e., 100 cfu/g), considering that lower or higher values would be either not practical because of constraints regarding microbiological analysis

Calculate what the probability is for ‘n’ samples to be negative for a just compliant batch/lot

Decide on the probability with which a non compliant lot should be rejected (95%)

How many samples should be taken from the lot so that the probability of rejection is achieved?


14Final Conference BASELINE . Bologna 11-12 November 2013

n = 5; c= 0 Pacpp = 0.1023


n = 7; c= 0 Pacpp = 0.0411 OK


17

Input received by WP1-5


18



19



Example I

Code Analysis Standard/guideline Assessment Sampling plan calculations

n c m MSatisfactor

yUnsatisfactory

1L.

monocytogenes7 0 2 NA <m/g

>m/g in any of the subsamples

tested

Two class sampling plan based on concentration data


21

- Influence of processing time / temperature on the growth of Salmonella Enteritidis in egg yolk

- Establishment of sampling procedures in powdered eggs

Example II


22

Processing temperature: assume constant temperature of 20°C

Processing time: scenario analysis (8h; 15h)

Latimer et al. 2002

Example II

23

Estimation of the growth of S. Enteritidis at both storage times:

a) 8h:

b) 15h:

1.23 log cfu/ml = 17 cfu/ml

2.89 log cfu/ml = 776 cfu/ml

Example II


24

Number of samples collected

Sample size

Contaminated part of the lot

Lot weight

Concentration in contaminated

samples

Example II


25

SCENARIO ANALYSIS

a)Low contamination vs high contamination (17 – 776 cfu/g)b)Increasing proportion of the contaminated part of the lot (from 0.01 to 0.1)c)Lot size effect (1000, 10000, 100000 g)d)Combining number of samples and sample size (n and w)

Example II


26

n w (g) p N (g) c (cfu/g) Pacc Prej

20 50 0,01 10000 17 0,8179 0,182120 50 0,05 10000 17 0,4341 0,565920 50 0,1 10000 17 0,3075 0,692520 50 0,01 1000 17 0,8179 0,182120 50 0,01 100000 17 0,8915 0,108550 20 0,01 10000 17 0,6153 0,384720 50 0,01 10000 776 0,8179 0,182120 50 0,05 10000 776 0,3585 0,641520 50 0,1 10000 776 0,1216 0,878420 50 0,01 1000 776 0,8179 0,182120 50 0,01 100000 776 0,8179 0,182150 20 0,01 10000 776 0,605 0,395

Example II


27

a) Low contamination vs high contamination (17 – 776 cfu/g)

Sampling is more effective as p increases, since Pacc decreases. No significant impact of the lot size and the combination n/w

b) Increasing proportion of the contaminated part of the lot (from 0.01 to 0.1)

Sampling is more effective as p increases, especially from 0.01 to 0.05

c) Lot size effect (1000, 10000, 100000 g) No significant

d) Combining number of samples and sample size (n and w)

Increasing n and decreasing w is more effective to detect positives, regardless of the microbial contamination

Impact of high contamination when p > 0.05. At low values of p, sampling is mainly affected by the initial

prevalence

Example II


28

Dependence of probability of acceptance on variability in lot sizes (N, kg). The sampling plan is characterized by sample size (w) = 100 g; microbial concentration (c) = 1 CFU/g ; proportion of contamination (p) = 0.05. The red vertical line corresponds to number of samples = 30.

Example II


29

Dependence of probability of acceptance on variability in sample sizes (w, g). The sampling plan is characterized by lot size (N) = 3000 kg; microbial concentration (c) = 1 CFU/g ; proportion of contamination (p) = 0.05. The red vertical line corresponds to number of samples=30.

Example II


30

Example II

Dependence of probability of acceptance on variability in proportions of the contaminated lot (p). The sampling plan is characterized by lot size (N) = 3000 kg; sample size (w) = 100 g ; microbial concentration (c) = 1 CFU/g. The red vertical line corresponds to number of samples=30.


31Final Conference BASELINE . Bologna 11-12 November 2013

The software tool is currently free available

www.baselineapp.com

Further possibilities: on-demand training sessions, consulting, assistance to SMEs to develop MC and sampling plans according to the real production systems.Inclusion of new predictive models and models validation etc.

http://www.baselineapp.com/

THANK YOU FOR YOUR ATTENTION !!!

BASELINE software tool for calculation of microbiological criteria and risk management metrics for selected foods and hazards WP6 Model Development Final.

Documents

international conference

final concentration

concentration limit

concentration of lm

compliedp log cfug

example iexample il

software tools webinar

standard deviation