Assessing (dermal) exposure loading in the evaluation process of biocidal products
TNO Quality of Life
Joop J. van Hemmen
TNO Senior Research Fellow in Occupational Toxicology TNO Chemistry
Th is i s a sh ip a nd the pa int co nta ins
a b ioc ide
Contents
• Use patterns and exposure
• Modelling exposure loading
• Current predictive dermal exposure models
• Uncertainties in the risk assessment process
• Data needs
• Current developments
Use patterns and exposure
• Exposure is strongly related to a task but there are exceptions
• There are many different tasks
• Duration of the task (and/or amount handled)
• Frequency of the task occurring
Emission, transfer, deposition
Adapted from Schneider et al. (1999)
Source Surface
Contaminan t Layer
Air Floor,
worktables, machinery,
tools
Outer clothing
Inner clothing
Phases of activity / time budget
• Manufacturing
• Formulating
• Mixing/loading
• Application/handling
• Post-application (re-entry)
• Removal
Modelling exposure loading
• Single study of a task (or series of tasks)
• Single survey (representative and robust)
• Several surveys
Exposure distributions
“width” and “typical value”
Individual data for boom sprayers
0 10 20 30 40 50 60 70 80
1 10 100 1000 10000 100000
Exposure µg/kg a.s. applied
S tud y num be r
Indicative distribution exposure model (I)
R ate P ro file
L ow M e d ium H ig h V er y h ig h
N arrow C ab b ed orch ard sp ra yin g
TP T (so lven t)
N et d ep lo ym en t An tifo ula nt b ru sh in g
M ed ium TP T (w ater)
AF m ix & lo ad
AF sp ra y S h eep
dip pin g
W id e P H I sp ra yin g
D ip p in g
R e m e dial
b io cid e sp ra ying U n ca bb ed
o rc ha rd sp ra yin g
Structure of BEAT
• Integration of expert judgement and objective measurements
• Based on what an expert would do anyhow, whenever he/she thinks
• BEAT consists of 3 elements:
• Bayesian framework
• Rule base (expert judgement)
• Database (with measurements)
Rule base
• Algorithm describing the similarity / analogy between exposure scenarios
• Algorithm contains the most important determinants of exposure
• Every determinant contains an “uncertainty factor”
• When the uncertainty factors increase, the similarity
between scenarios decreases
Rule base
Process
Work practice Working
environment Controls
Agent
Scenario
Exposure
distribution
Rule base
Determinant 1 Determinant 2 Determinant 3 Determinant n
Determinant 1 Determinant 2 Determinant 3 Determinant n
Rul e ba se
Scenario A Scenario B
Analogy score
Database
• Quantitative measurements for different
scenarios with background information on
exposure determinants (over 1,000 data
points)
Structure of BEAT
Scenario X
Database
Bayesian framework
Rule base
Posterior exposure distribution
Measurement data
Analogy
Models / databases
• Study / survey results, i.e. set of data points
• Several similar studies / surveys, i.e. database
• Humanexposition bei Holzschutzmitteln
• Mechanistic / theoretical model
• SprayExpo: Inhalative und dermale Expositionsdaten für das Versprühen von flüssigen Biozid-Produkten
• Combined mechanistic and database model
Technical Notes for Guidance (TNsG)
• Examples: http://ECB.JRC.IT/biocides
User Guidance
• Revisited set of models
Guidance for choice of surrogate value (percentile)
• PT 8 and PT 14 (rodenticides and wood
preservatives)
Geometric standard deviation
2 4 6 8 10
10 1.49 2.22 2.81 3.31 3.77
20 1.33 1.76 2.08 2.33 2.56
50 1.20 1.43 1.59 1.71 1.81
Sample size
100 1.13 1.29 1.39 1.46 1.52
Table 1: Scaling factors to obtain a 90% confidence interval for a 75 th
percentile with a variety of sample sizes and GSDs
variance
Variability
mean
mean
Uncertainty
variance
Determines magnitude & frequency of effects
Determines confidence limits for effects
variance
Variability
mean
mean
Uncertainty
variance
Key variables for control (risk mitigation)
Key variables for research (prioritisation)
Sensitivity analysis
Sensitivity analysis
Variability and uncertainty
Knowledge about both variability and uncertainty is essential for reliable
predictions
Variability and uncertainty (2)
• Variation
• What fraction of the population is above a certain value (AOEL/MAC/TLV)?
• A point on a cumulative distribution curve
• Uncertainty
• What is the confidence in this estimate?
• Expressed as confidence limits or error bars
• How do you assess these?
• Monte Carlo (probabilistic) analysis
1.0 10.0 100.0 1000.0 Dos e [µg/kg/day]
0 10 20 30 40 50 60 70 80 90 100
Percentile
1.0 10.0 100.0 1000.0
Dose [µg/kg/day]
0 10 20 30 40 50 60 70 80 90 100
Percentile
0 10 20 30 40 50 60 70 80 90 100
Percentile
Variability inner loop
Uncertainty outer loop
Median
90% confidence
limits
Determinants of processes
(Deliverable 36 RISKOFDERM)
Categories
• substance/ product characteristics
• tasks
• process, techniques/
equipment
• exposure control
• worker characteristics
• area and situation
Examples ( for deposition)
• physical state
• amount of substance
• type of aerosol generation
• ventilation
• position relative to source
• wind direction and speed
Types of situations studied: examples 1
Handling contaminated objects
Manual dispersion of products
Dispersion with a
hand-held tool
Types of situations studied: examples 2
Spray dispersion of products
Immersion of objects
Mechanical treatment
of solid objects
Extensive body contact?
Application rate (L/min)
Results – rate (µL/min) Percentile
Results - mass or volume
Duration (min)
No
75.0
50 0.5
Percentile distribution
646.
58.8
Median
1750.
160.
Percentile distribution
Yes
Median
25.0
50 0.068
Percentile distribution
297.
744.
Median
109.
274.
Percentile distribution Median
Hands Body
Example model
wiping
Example model spray painting
Direction spraying
Proximity to source
Results – rate (µL/min) Percentile
Results - mass / volume Duration (min)
Level
75.0
60 0.12
Percentile distribution
4.54 12.
Median
15.7 51.4
Percentile distribution Median
Away from the worker
Air flow
Up to 1 meter No Indoors
Indoors/outdoors
Not highly volatile
Segregation Volatility
Liquid