Pregnancy, Health and AIR Pollution PHAIR Exposure Study

1
Pregnancy, Health and AIR Pollution (PHAIR)
Exposure Study

• Air Pollution Model Evaluation Study

Elizabeth NetheryJune 26, 2006
2
Today

• PHAIR Study
• Methods
• Sampling Campaign Current Status
• Results
• LUR to measured
• Seasonal adjustments
• Postal Code to Address Geocoding
• Combined HomeWork
• Gas Stoves
• Activity Data Comparison to CHAPS

3
Sampling

• 62 women, 3 excluded or left study
• 59 women for analysis (to date).
• 117 sampling sessions completed

Subject Demographics

• Mean age 32 yrs
• 11 women will complete 3 samples 48, 2 samples
• 19 with at least 1 other child 39 no children
• 22 Rent/other 37 Own (housing status)
• 38 (64) working FT 16 (28) working PT 5 (9)
  Not working
• Mean home (building) age 49 years.

4
Results - June 2006

• 107 Ogawa Samples analyzed
• Equipment delays with Absorbance (47 samples) and
  PM sample analysis (85 samples)
• 110 Activity logs
• 83 GPS routes of moderate-high quality.

5
Data

• Measured personal exposure (NO/NOx/NO2,
  Absorbance, PM2.1)
• Modelled (LUR) exposure at home and work
• Measured mobility data (GPS)
• Measured Activity data (time-activity diaries)
• Measured building information, personal
  characteristics/demographics
• Comparison 1 Measured vs. Modelled Home
• Comparison 2 Measured vs. HomeWork (times from
  4.)
• Comparison 3 Measured vs. Modeled estimates
  using GPS route (to be completed)

6
Continuous Correlations between Home
Predicted(seasonally adjusted) (LUR) and Personal
Measurements
pgt.01
7
Route Data - Sample

• Red is Work Location Green is Home
• first analysis with just HOME location
• second analysis with Home Work using
• time spent Indoors_home and Indoors_work

8
Continuous Correlations between WorkHome
COMBINEDEstimate (LUR) and Personal Measurements
pgt.01
NB. Insensitive to presence of Gas Stove mean
cooking time2 hours (4.5 total time)
9
Stratified by GAS STOVE Presence Continuous
Correlations between COMBINEDEstimate (LUR) and
Personal Measurements
pgt.01
10
(No Transcript)
11
(No Transcript)
12
Postal Code vs. Address Geocoding

• Traffic data for (Home and work locations)
• Extracted using Geocoded addresses
• OR
• Extracted using Postal Code locations

13
Postal Code to Address Geocoding
14
Activity Results Summary Stats

• Mean age 32.4 (range 23-40)
• N110 (Activity Logs)

15
Activity Logs Compare to CHAPS
16
Conclusions to date..

• Important to include seasonal effect
• LUR predicted (home) and measured personal sample
  for NO and NOx show some trend Means increase
  across quartiles and group differences are
  significant.
• Improvement in the trend (for NO2) when including
  work and time-spent at work
• Postal code and address geocoding are highly
  correlated little change on overall results
  using either
• Some outliers include those with high time in
  traffic, or cooking (preliminary)

17
Future plans..

• Use GPS data to estimate exact traffic exposure
  from walking over LUR
• More analysis with other methods (mixed models)
• Compare personal measurements to GVRD Monitoring
  data (use methods similar to BAQS cohort
  approach).

18
Thank-youComments/Questions?
Acknowledgements Micheal Smith Foundation for
Health Research BAQS Group Sara Leckie, Katherine
Woolbert and all participants! Dr. Brauer, SOEH
19
Subject Demographics
20
(No Transcript)