Exposure Assessment: What are we all about? 2003 Weselowski Award Presentation Mike Lebowitz • The scientific discipline of EA needs & deserves growth - conceptually & in technology. • We need to educate – insure that EA is used appropriately in bio-medical environmental sciences and hazard assessment, in risk assessment, and in policy making • We need to learn how best to utilize good EA in policy making.
59
Embed
Exposure Assessment: What are we all about? 2003 Weselowski Award Presentation Mike Lebowitz
Exposure Assessment: What are we all about? 2003 Weselowski Award Presentation Mike Lebowitz. The scientific discipline of EA needs & deserves growth - conceptually & in technology. - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Exposure Assessment: What are we all about? 2003 Weselowski Award Presentation
Mike Lebowitz• The scientific discipline of EA needs & deserves
growth - conceptually & in technology.• We need to educate – insure that EA is used
appropriately in bio-medical environmental sciences and hazard assessment, in risk assessment, and in policy making
• We need to learn how best to utilize good EA in policy making.
Interactions between
Host, Agent &
Environment
Environment
HostAgent
Biological Physical
Social
Exposure Assessment & the Environmental Health Paradigm
Sexton (1992)
Sources (s) • PropertiesAmount Released/UsedLocation/Setting
Concentrations AirWaterSoil/DustFoodSurfaces
Human ExposuresRouteMagnitudeDurationFrequency
Internal DoseAbsorbed DoseTarget DoseBiomarkers
Health Effect (s)CancerNon Cancero Damage/Diseaseo Signs/Symptoms
Exposure Assessment
Effects Assessment
Steps in Exposure estimation
1.1. Environmental Sources & Pathways in Multiple Environmental Sources & Pathways in Multiple MediaMedia
2.2. Concentration Measurements (Monitoring) & Concentration Measurements (Monitoring) & Exposure Info Exposure Info
3.3. Population Distributions of Concentrations in Population Distributions of Concentrations in MediaMedia Measurements of Time-Activity & Other Measurements of Time-Activity & Other Exposure FactorsExposure Factors
4.4. Statistical Modeling of Exposure Factors on Statistical Modeling of Exposure Factors on Exposures Exposure Assessment ModelingExposures Exposure Assessment Modeling
Lioy & Pellizzari, 1995
Lioy & Pellizzari, 1995
WHO - Biomarkers and Risk Assessment: Concepts and Principles
(ICPS, EHC 155, 1993)(ICPS, EHC 155, 1993)
WHO - Biomarkers and Risk Assessment: Concepts and Principles
(ICPS, (ICPS, EHC 155, 1993)EHC 155, 1993)
Exposure & Health Hazard Assessments
Exposure - Dose Estimates
Health Hazards Assessments
Risk Assessments
& Proposed Risk Management
Policies & Logistics:
Avoidance, Prevention, & Intervention Programs
Lioy & Pellizzari, 1995
Percentile
98877665544231205.88BDLBDL
Val
ue (
ug/m
^3)
140
120
100
80
60
40
20
0 DL
Air – PM10-Outdoor
Percentile
928068554330188BDL
Val
ue (
ug/m
^3)
300
250
200
150
100
50
0
GL
DL
Air – PM10-Indoor
Comparison of Indoor Air Arsenic
Mann-Whitney U Test: p<.001
Normalized T-Test: p<.001
Percentile
1009080706050BDL
Va
lue
(n
g/m
3̂)
30
25
20
15
10
5
0
STUDY
Border
NHEXAS
Pb, As, and Cd ComparisonsAmong Border Counties
Analyte Media Lead Arsenic Cadmium Fooda NS NS NS Airb Indoor p<.001 p<.001 p<.001 Outdoor p=.034 p<.001 p<.001 Soilb p<.001 p<.001 p<.001 Floor Dustc NS NS NS a NS: no significant differences between counties (Pb, p=.08; As, p=.92; Cd, p=.53) b significant differences between counties using Kruskal-Wallis Test c NS: no significant differences between counties (Pb, p=.25; As, p=.39; Cd, p=.57)
Mean Percent Total Ingestion of Metals from Different Sources
Specific food types associated with elevated metal concentrations
• Hispanicity itself was a significant predictor of lead and cadmium when included in the models (p<.001); retained significance even after adjusting for food types
• Significant differences in arsenic and lead; primarily driven by region & outliers• Food from Border Hispanics contains significantly
more lead than the food from Hispanics living elsewhere in the State
• Most outliers occur in Hispanics vs. non-Hispanics• Significant differences in cadmium - Higher levels among
non-Hispanics and in AZ v.s. Border
Indoor / Outdoor Air VOC Ratio at the 90th Percentile
•PM10, air and dust Cd concentrations are significantly greater (p < 0.05) in homes with smokers when compared to non-smoking Households.
•Individuals who smoke have significantly greater Pb & Cd in blood and Cd in Urine (p < 0.05).
•Indoor PM10 correlated with blood Pb:
•rs .385 (p=.001) all Households
•rs .610 (p=.001) Smoking Households
•Food, Beverage and Water are the primary contributors to metal exposures.
Lead Blood ConcentrationsF
ractio
n
log-blood concentration Pb-1.20397 2.89037
0
.136646
Lead Total Daily DoseF
ractio
n
log-total daily dose Pb2.35758 4.67571
0
.142045
Pb-Blood vs. Total Daily Doselo
g-b
loo
d c
on
ce
ntr
atio
n P
b
log-total daily dose Pb2.35758 4.67571
-1.20397
2.89037
Cadmium Urine ConcentrationsF
requency
log-biomarker concentration Cd-2.99573 1.72277
0
22
Cadmium Total Daily DoseF
ractio
n
log-total daily dose Cd, ug2.11048 4.24839
0
.142045
Cadmium Blood by Total Daily Dose Conc.
log
-blo
od
co
nce
ntr
atio
n C
d
log-total daily dose Cd2.11048 4.24839
-1.89712
1.06471
Bio-marker Distributions for Select Metals (g/dL)
*Hispanics and non-Hispanics significantly different; # Significantly different by age; see regression results for Ni & Pb; ** No significant differences The two significant regressions of biomarkers on total daily dose estimates were for Ni (involving age and
hispanicity as well); and Pb (involving sex and hispanicity,after which neither age nor the total daily dose were significant).
25% 50% 75% 90% Max
Cadmium-Urine* 0.200 0.500 0.900 1.500 5.600
Cadmium-blood** 0.150 0.600 1.000 1.500 2.900
Chromium# 0.200 0.400 0.800 1.400 18.400
Nickel 2.800 4.200 6.400 8.100 21.100
Lead 1.100 1.700 2.600 4.000 18.000
Lead Dose Distribution by Media
Contribution to Lead Dose for an Adult Male at the 90th Percentile of exposure for each Media
Weighted and un-weighted mean metals intake for Hispanic and Non-Hispanic Whites, AZ NHEXAS
Weighted Exposures Unweighted Exposures
Hispanic Non-Hispanic Hispanic Non-Hispanic
N = 1284049 3652221 53 126
Inhalation
As 0.032 0.036 0.049 0.043**
Mn 0.475 0.399* 0.427 0.379*
Ingestion
As 24.71 47.299 18.405 45.122
Mn 1805.621 2166.714 2324.044 2850.442*
Total
As 24.743 47.335 18.455 45.165
Mn 1806.096 2167.114 2324.471 2850.821*
Descriptive Statistics for Drinking Water Chemical Residue Concentrations in Tap and Non-tap Water
Residue Median 90th% Median 90th% Median 90th%* Median 90th%
Arsenic 4.76 16 0.14 4.23 5.09 10.7 0.23 4.67
Chromium 1.09 11.3 0.28 1.74 1 4.34 0.36 4.08
Lead** 0.39 1.59 0.05 0.94 0.28 0.6 0.17 0.95
Nickel** 0.32 9.62 3.9 5.89 2.27 5.62 0.33 4.27
1,2-DCE 0.29 1.94 0.3 2.28 0.29 0.77 0.08 0.72
1,3-Butadiene - - - - 0.048 0.11 0.031 0.11
DCM 0.074 0.57 0.067 0.64 - - - -
Chloroform** 0.03 2.04 0.05 2 0.11 1.19 0.15 0.87
Toluene 0.22 4.51 0.57 6.78 0.49 5.71 0.5 2.67
*Sample size is not adequate for distribution fitting, percentile value is calculated from empirical cumulative distribution** Median tap water concentrations of Arizona and Border are significantly different with Mann Whitney test
Uncertainty Analysis of Probabilistic ADE Estimates
ADE (ng/kg/day) - Arizona
Chemical Median ADE 90th Percentile ADE
Residue Mean SD Range Mean SD Range
Arsenic 22.2 2 19.7-24.5 172.3 11.1 159-187
Chromium 18.1 1.1 16.7-19.3 148.6 12.5 121-147
Lead 6.4 0.3 6.1-6.7 26.8 1.6 24.8-29.0
Nickel 15.3 1.3 13.6-17.1 138.5 11.6 123-152
1,2-DCE 7.6 0.4 7.1-8.0 34.8 2.2 31.7-37.5
1,3-Butadiene - - - - - -
DCM 0.6 0.1 0.53-0.65 7.6 0.8 6.5-8.6
Chloroform 1.2 0.1 1.0-1.3 36.2 5 30-43
Toluene 4 0.4 3.6-4.5 65.8 7.3 58-77
Uncertainty was estimated in a two-step process using the bootstrap technique. 1- selects randomly a set of input to the model variable values, and estimates an exposure value. This step is repeated 1,000 times to formulate one exposure distribution. 2 - the process is repeated 200 times, resulting in 200 exposure distribution curves.
Uncertainty Analysis of Probabilistic ADE Estimates ADE (ng/kg/day) - Border
Chemical Median ADE 90th Percentile ADE
Residue Mean SD Range Mean SD Range
Arsenic 19.65 1.76 17.4-21.6 186 14 168-205
Chromium 19.41 1.15 17.9-20.8 143 12 129-157
Lead 7.49 0.61 6.8-8.4 104 12 89-118
Nickel 25.22 1.63 23.3-27.3 177 14 159-194
1,2-DCE 3.54 0.21 3.3-3.8 17 1 15.9-18.4
1,3-Butadiene 0.37 0.02 0.35-0.39 1.4 0.1 1.3-1.5
DCM - - - - - -
Chloroform 4.9 0.29 4.6-5.3 32 3 29-35
Toluene 6.97 0.39 6.5-7.5 39.7 2.7 36-43
Uncertainty was estimated in a two-step process using the bootstrap technique. 1- selects randomly a set of input to the model variable values, and estimates an exposure value. This step is repeated 1,000 times to formulate one exposure distribution. 2 - the process is repeated 200 times, resulting in 200 exposure distribution curves.
Relationship of Diseases to VOC Indoor Air Concentrations
VOCs* Odds Ratio 95% CI #Cases
Toluene
gastrititis (9.1%) 2.15 1.20-3.88 16
Formaldehyde gastrititis 2.64 1.15-6.09 16
* No diseases were significant for benzene. Adjusted for hispanicity, age, and sex.
Other significant diseases include indigestion (9.1%), colitis (3.4%), intestinal /bowl trouble (7.4%) and other liver trouble (2.9%)
Relationship of Diseases to VOC Biomarkers
VOCs* Odds Ratio 95% CI#Cases
Benzene
kidney disease (3.4-5.7%) 2.20 1.07-4.56 10
PCE Indigestion (9.1%) 2.83
* No diseases were significant for chloroform, toluene, p-dichlorobenzene. Adjusted for hispanicity, age, and sex.
Relationship of Diseases to Metal Daily Dose
Metals* Odds Ratio 95% CI #Cases
Cadmium liver trouble 10.83 1.04-112.83 5
Lead yellow jaundice 9.74 2.09-45.26 8
* No diseases were directly related with chromium, arsenic, and nickel. Adjusted for all significant covariates; hispanicity, age, sex, individual and household smoking
Relationship of Diseases to Metal Biomarkers
Metals* Prev. Odds Ratio 95% CI #
Cadmium
Cd urine-stomach 7.4% 2.60 1.17-5.79 167
Cd-blood-hepatitis 6.9% 4.02 1.48-10.94 161
Nickel (urine)
jaundice 4.6% 4.71 1.11-19.96 165
Lead (blood)
bowl trouble 7.4% 3.47 1.27-9.47 160
* No diseases were directly related with chromium and arsenic. Adjusted for all significant covariates; hispanicity, age, sex, individual and household smoking
Other significant diseases include colitis (3.4%) and bowl trouble
Other significant diseases included fatty liver (2.2%)
PM10 Concentration
NHEXAS P5
n 163 271
Mean (µ g/m3) 40.7 34.8
Median (µ g/m3) 31.3 27.3
Range (µ g/m3) BDL - 211.7 BDL-179.4
1717179179.
43
BDL 2 (1.2%) 6 (2.2%)
Ln(Detection Limit) = 1.7 µ g/m3
Prediction Model
Variable P Intercept 2.8 < 0.01
(Cigarettes/day smoked by each HH member) ½ 0.1 < 0.01 Monsoon -0.4 < 0.01 People 0.1 < 0.01 Winter 0.3 < 0.01 Remodeled (yes/no) 0.2 < 0.01 (# Smokers)½ 0.2 < 0.01 Pet (yes/no) 0.2 0.02 Fall 0.2 0.03 Vacuumed (yes/no) -0.2 0.03 Adj. R2 = 0.376 N = 244
A
Association of Prevalent Respiratory Disease with Measured PM10
# Cases
Adj odds ratio*
95% ConfInterval p-value
Asthma 125 1.5 0.83, 2.72 0.183Chronic
Bronchitis 89 1.45 1.00, 2.11 0.051
Emphysema 7 2.37 0.70, 8.07 0.168
Bronchiectasis 32 1.5 0.83, 2.72 0.183
* Adjusted for age, gender, smoking
Association of Incident Cardiovascular Disease with Measured PM10
Odds 95% Conf #cases Ratio Interval p-value
HighBlood Pressure 10 1.32 0.60, 2.90 0.493
Other HeartProblems 5 4.60 1.30, 16.31 0.018
Some Conclusions• We have come a long way and have a long way
to go – Scientifically & Politically – We need to be proactive
• We need better objectives, study designs, screening instruments, & collaborations with social scientists
• We should learn about policy-making and work with politicians to insure use of our science & should participate more out there