Document LgOKb1r38NXRzEkznynyb16Bq
Issues Motivating the Collection of Occupational Exposure Data
S. M. Rappaport, Ph.D. University of California, Berkeley
srappaport@berekeley.edu
1
Topics
Why measure airborne chemicals?
Hazard control v. health surveillance (epidemiology) Air measurements v. exposures
Evolution of air and exposure measurements Exposure variability and its sources How many measurements? If you build it will they come? Biomarkers of exposure
2
Why Measure Airborne Chemicals?
Hazard control: Setting and enforcing standards (OELs) Health surveillance (epidemiology): Investigate exposure-(dose)-response relationships
3
Health Surveillance v. Hazard Control
Health surveillance (epidemiology)
Primarily focus upon longterm effects (yearsdecades)
Investigate exposureresponse relationships Best with extensive longterm quantitative exposure data
Hazard control
Focus upon both shortterm and longterm hazards
Shortterm: (secondshours, e.g., H2S, CO, NO2, NH3) Require air (not exposure) measurements in some cases
Longterm: (yearsdecades, e.g., benzene, PAHs, asbestos, heavy metals) Require extensive longterm exposure data
4
Evolution of Air and Exposure Measurements
Type
Air /Exposure Time frame Weight (g)
Area
Breathing zone
Personal
Directreading (handheld or personal)
Air Exposure Exposure Air or exposure
1920 present 1940 present 1960 present 1980 present
>1000 100 1000 10 1000 1 1000
Assay
Lab Lab or direct Lab or direct Direct
5
Area Sampling: 1920sPresent (Air measurements not exposures)
A
BB
A: Collection of benzene by adsorption on charcoal about 1926 (Greenburg, 1926).
B: Greenburg-Smith impinger used for
dust sampling about 1930 (Drinker and
Hatch, 1936).
C: Collection of benzene by aspiration and
absorption in acid about 1928 (Smyth and
Smyth, 1928).
D: Collection of asbestos with a high-
volume sampler and a cascade impactor
about 1953 (Photograph courtesy of R.
Herrick).
CD
Rappaport and Kupper (2008) 6 Quantitative Exposure Assessment
Breathingzone Sampling: 1940s Present (Moving from air measurements to exposures)
Two examples of breathingzone sampling. Left: sampling with a midget impinger of explosive vapors (probably nitroglycerine) at an ordinance plant USA (1943). US PHS . (Photograph courtesy of R. Herrick). Right: Benzene sampling with an explosimeter and silica gel tubes during manufacture of mechanical seals UK (1950). (Photograph courtesy of R.J. Sherwood).
Rappaport and Kupper (2008)
7
Quantitative Exposure Assessment
Personal Sampling: 1960sPresent (Exposure measurements)
Personal samplers to measure styrene, styrene oxide and MEKP in the reinforced-plastics industry- USA (1986). Left: active sampling with sorbent tubes and micro-impinger; Right: passive sampling with activated carbon (styrene and styrene oxide only).
Rappaport and Kupper (2008) Quantitative Exposure Assessment
8
Direct Measurements: 1980s Present
(Air measurements or exposures)
Hand-held devices (air measurements)
Direct-reading personal monitors (exposures)
9
Exposure Variability
Styrene levels in a boat factory (162 personal measurements - 1986-87)
1000-Fold Range
Air Sample
Data from: Rappaport, et al. Cancer Res, 56: 5410-5416 (1996)
Assay error (0.2-fold)
10
Within- and Between-Worker Variability
Within-person variability (about 10-fold)
Between-person variability (more than 100 fold)
Data from: Rappaport et al., Ann. Occup. Hyg. 43:457469, 1999
11
Exposure Variability
Welding-fume exposures among construction workers
(198 measurements from 62 workers in 4 trades 1996-97)
Data from: Rappaport et al. Ann. Occup. Hyg. 43:457-469, 1999
1000-Fold range
12
Exposure Variability
Welding-fume exposures among construction workers
(198 measurements from 62 workers in 4 jobs)
Group variability (4-fold)
Data from: Rappaport et al. Ann. Occup. Hyg. 43:457-469, 1999
13
Exposure Variability
Welding-fume exposures among construction workers
(198 measurements from 62 workers in 4 jobs)
Within-person variability (10-fold)
<2-fold
Between-person variability
5-fold
5-fold
27-fold
Data from: Rappaport et al. Ann. Occup. Hyg. 43:457-469, 1999
14
Determinants of Exposure to Welding Fumes
(From fixed effect in mixed models)
56
40 10
Fume Conc. (mg/m3)
Fume Conc. (mg/m3)
30 20 10
0 0,0 0,1 1,0 1,1
8 6 4 2 0
0,0 0,1 1,0 1,1
78
10 5
Fume Conc. (mg/m3)
Fume Conc. (mg/m3)
84
6 4 2 0
0,0 0,1 1,0
1,1
3 2 1 0
0,0 0,1 1,0 1,1
Fig. 7.6 Predicted mean exposures to welding fumes for Groups 5 - 8 (5=BM; 6=IW; 7=PF; 8=WF), based upon the model shown in Table 7.5. Activity (IO, TW): (0,0)=outdoor brazing/cutting; (0,1)=outdoor welding; (1,0)=indoor brazing/cutting; (1,1)=indoor welding. Controls consisted of local-exhaust or mechanical ventilation (VE = 1) and reduction of hot work to less than 50% (CI = 1). (Note that magnitudes of the y-axes differ across groups).
Rappaport and Kupper (2008) Quantitative Exposure Assessment
Variability Within and Between Workers
Cumultive Percent
100
80 Between
60 Within
40
20
0 1 10 100
b R^0.95 and w R^0.95
Between-worker = 4 fold: 95% of the workers in a given job group have mean exposure levels spanning a 4-fold range.
1000
Within-worker = 15 fold: 95% of a typical worker's exposure levels vary 15-fold from day to day.
Rappaport and Kupper (2008) , Quantitative Exposure Assessment, based upon work by Krommout et al. (1993)
16
Variability Within and Between Workers: Not a New Idea!
Within-person and between-person sources of variability in exposure levels were recognized as early as 1952 when Oldham and Roach applied ANOVA models to breathing zone samples of dust in British coal mines (Oldham and Roach, 1952). They made the following observation:
"It was found that significant variation was occurring in the dust concentrations from one collier's experience to another's, and from one day to another in the same collier's experience." (Note that a `collier' is a coal miner).
Yet, this finding was largely ignored at the time, and the issue of within-person and between-person variability was not revisited again until some 35 years later when personal exposure measurements became available in occupational studies (Kromhout et al., 1987; Rappaport et al., 1988b; Spear et al., 1987).
Rappaport and Kupper (2008) Quantitative Exposure Assessment
17
Why so variable?
Multiplicative effects of several variables
Jobs (fixed) Time (fixed) Locations (fixed or random) Sources of contamination (fixed or random) Activities and equipment (fixed or random) Worker/source mobility (mostly random) Environmental conditions (mostly random)
18
Implications of Exposure Variability
Health surveillance (focus upon chronic health effects)
Many personal measurements needed to characterize longterm exposures
Longitudinal studies, evaluate variation within and between persons and across groups
Cannot assume all workers in a group are equally exposed
Advanced statistical models (mixedeffects models)
Hazard control
Longterm hazards: same issues as above for health surveillance (repeated personal measurements, mixed modeling, etc.)
Shortterm hazards: focus shifts to air levels/warnings of immediate dangers (high air concentrations not exposures)
Only most acutely toxic substances Area sampling sufficient (e.g., confined spaces or at point of release)
19
Sample Sizes for Air Measurements (1920s 1950s)
Focus upon health effects (exposureresponse) Few professionals (mostly governmental) Cumbersome equipment No OELs Few studies but relatively large sample sizes (hundreds of
measurements)
Variability recognized Desired accurate estimates of average levels for each location or factory
Classic study of Oldham and Roach (1952)
779 Breathing zone measurements (3min) randomly collected repeatedly from Welsh coal miners
20
Repeated Random Measurements
A portion of an appendix originally published by Oldham and Roach (Oldham and Roach, 1952). Each entry represents the dust level for a random 3-min sample obtained from a coal worker. Note that several such measurements were obtained from each subject on a given day.
Rappaport and Kupper (2008) 21 Quantitative Exposure Assessment
First Application of Lognormal Distribution to Occupational Data
Histogram of logged deviations of 779 breathingzone measurements of dust in British coal mines [from Oldham (1953)].
Provides basis for advanced statistical modeling of data
Rappaport and Kupper (2008) Quantitative Exposure Assessme2n2t
Exposure Data in Modern Epidemiological Studies
Only 13% of studies used quantitative measurements
From: B.K. Armstrong et al. Principles of Exposure Measurement in Epidemiology, Oxford Med. Pubs., 1992
23
Sample Sizes for Workplace Measurements after OSH Act of 1970
Many professionals (mostly employerbased) Advanced personal samplers and directreading monitors Focus upon hazard control rather than health surveillance
Only 16 new OSHA PELs since 1971
Almost all air monitoring for acute hazards (`safety'), e.g., confined spaces, LEL, O2 deficiency, substances IHTL
Few measurements for chronic health effects
Median = 4 meas. from 696 published studies reviewed by Symanski et al. (19671996)
What about industrial surveys?
24
Sample Sizes for Industrial Surveys
85% 4 Median = 1
Numbers of measurements obtained in 4864 annual surveys of occupational groups of workers in the nickel producing industry 1970 1990. [From (TorneroVelez et al., 1997)].
25
Better Equipment & More Professionals but Fewer Measurements Why?
Current trends
From government inspectors to employerbased inspectors (vested interests)
Increasing reliance on measurementfree methods (`exposure models', `controlbanding', etc.)
From exposureresponse (long time frame) to compliance with OELs (short time frame)
OELs have existed since the 1950s
Prior to 1970 OELs were guides After OSH Act they became legal limits
26
Compliance Testing
Rarely performed by OSHA inspectors
Fewer than 10,000 health inspections per year in 2.5 million US workplaces (P{health inspection} < 0.004/year)
Vast majority performed by employers who must provide workplaces "...free from recognized hazards."
Company representatives (e.g., IH) can measure personal levels of persons in all groups with potential for excessive exposures
Onetoone comparison of observed air levels with PEL
Compliance: All measurements < PEL
No additional measurements needed
27
Probability of Compliance
Let h represent the probability that a person in
Group h would be exposed on one day above the
OEL (exceedance of Group h)
h = P{Xhij > OEL}
PDF
0.14
0.12 i=1
Person (i) hi
1 0.001
A
2 0.003
0.10
3 0.009
2 0.08
4 0.090
0.06
3 Group h
5
0.203
4 0.04 OEL
0.02 5 h=0.047
0.00
0
20 Xhij
40
60
Then the probability of compliance for Group h is
P{Ch} = (1- h )Nh
Rappaport and Kupper (2008) Quantitative Exposure Assessment
28
Noncompliance and Sample Size
1-P{Ch}
1.00
Nh=20 10
0.80 5
3
0.60 2
0.40 1
0.20
0.00 0
0.2 0.4 h 0.6
0.8
1
Rappaport and Kupper (2008) Quantitative Exposure Assessment
29
Noncompliance and Sample Size
1.00
Nh=20 10
0.80 5
3
0.60 2
0.40
1
Hazardous exposures
where h > 0.2
and P{Compliance} > 50%
0.20
1-P{Ch}
0.00 0
0.2 0.4 h 0.6 0.8
1
Since P{Ch} depends greatly on Nh, employers have incentive to maximize P{Ch} by making very few measurements (Compliance testing can only be applied with small sample
sizes)
Rappaport and Kupper (2008) Quantitative Exposure Assessment
30
The Advantage of Air Monitoring
OSHA standards mandate air monitoring for conditions IHTL (safety standards)
Employers have incentives to avoid situations IHTL (confined spaces, trigger alarms, etc.)
Acute hazards easily assessed (reduces employer liability) Hundreds of directreading air monitors commercially available
Employers prefer to measure air levels of chronic toxicants because they are not clearly tied to exposures
Use of area rather than personal measurements Evaluating air levels for tasks rather than workers Can ignore betweenworker differences in exposure
31
The Problem with Exposure Monitoring
Under the OSH Act the burden of proof is upon the government to prove noncompliance with PELs
But there are essentially no government inspections (the OSH Act is based upon voluntary compliance!)
The use of compliance testing implicitly discourages exposure monitoring
But chronic health effects are slow to develop and difficult to relate to exposures without extensive exposure monitoring
Many employers don't want extensive exposure data in their files
The situation is unlikely to change without a paradigm shift, such as REACH (Registration, Evaluation, Authorization and Restriction of Chemical Hazards) which places the burden of proof on the manufacturer to prove the safety of its products
32
If you build it will they come? (You've got a DREAM ...)
Nanosensors of the future:
Ultra-small personal monitors Multiple analytes Data logging Measure all workers' exposures to everything - all the time!
I. Simon et al. / Sensors and Actuators B 73 (2001) 126
Air monitoring
1920
1970
2008
Trends for measurements
1920 2008
Exposure monitoring
33
What about biomarkers of exposure?
Passive Monitor
Breath Sampling
Described in: Yager, et al. (1993) Mut. Res. 319:155-165.
34
Styrene in Air and Breath
Reinforced-plastics workers (3 7 meas./subj.)
400
300
200
100
0 0
15
12
9
6
3
0 0
20
40
60
80
100
120
140
160
180
Data from: Rappaport,
et al. Cancer Res, 56:
20
40
60
80
100
120
140
160
180 5410-5416 (1996)
35
Styrene in Air and Breath
Reinforced-plastics workers (3 7 meas./subj.)
Unit slope
Data from: Rappaport, et al. Cancer Res, 56: 5410-5416 (1996)
36
DREAM Biomarkers of Exposure
Good idea
Highly relevant to exposure and health effects When used with exposure measurements can illuminate
important human kinetic processes Adaptable to nanosensing and LOC systems (high
throughput, multiple analytes)
But U.S. employers don't like biomarkers (even less than personal exposure measurements)
Few OSHA standards require biomonitoring (Pb, Cd) More demand in Europe, Asia, and for nonoccupational
exposures in the U.S.
37
SelfAssessment of Exposure
10000 1000
1
23
4
Terpenes (mg/m3)
100
10
1 0
10000
10 20 30
Worker
40
12 3456
1000
Styrene (mg/m3)
100
10
1
Fig. 3.3 Self assessment of benzene in air and breath during automobile refueling, as described by Egeghy et al. (2000). Left: Packaged air and breath samplers (top) and instructions for use (bottom). Right: Air sampling during refueling (top) and breath sampling after refueling (bottom).
Rappaport and Kupper (2008) Quantitative Exposure Assessment
0.1 0
10 20 30 40 50
Worker
Fig. 3.2 Exposures to terpenes in sawmills (top) and to styrene in reinforced plastics factories (bottom). Open circles represent self-measurements made by workers and closed circles represent measurements made by an occupational hygienist on different days. Numbers represent different workplaces of a particular type. [Data from Liljelind et al. (2001)].
38
Take Home Messages
Historically, exposure measurements were the holy grail that motivated breathing zone and personal sampling for studies of health effects
Early studies recognized exposure variability and included many measurements as a result
39
Take Home Messages
Hazard control trumps health surveillance for contaminant measurements in U.S. workplaces
Regulatory focus is upon acute (safety) hazards not health hazards
Only 16 new OSHA standards in 37 years Emphasis upon compliance with PELs discourages
employers from monitoring exposures Reinvigorating exposure monitoring will require a
paradigm shift (e.g., REACHtype system)
40
Take Home Messages
Exposure levels vary tremendously across groups, between workers (within groups), and within workers over time
Puts a premium on repeated exposure measurements over the long term
Sophisticated statistical models needed to characterize exposures and their determinants
41
Take Home Messages
Applications of DREAM technology for air (rather than personal) measurements are straightforward
Satisfy needs for control of acute hazards
Extending DREAM to collect large numbers of measurements of personal exposure or biomarkers will require a shift in regulatory focus and/or enforcement
42
30 Years of thinking in 183 pages
The vagaries of exposure limits
Measurement-based exposure assessment
Statistical tools for exploring exposure variability
Within- and between-worker sources of variability
Determinants of exposure
Implications for hazard control and epidemiology
Choosing between environmental measurements and biomarkers
Available from Lulu Press
http://www.lulu.com/content/1341905
43