Impact of Ambient Air Pollution on Infant Bronchiolitis in Puget Sound

1
Impact of Ambient Air Pollution on Infant
Bronchiolitis in Puget Sound

• The Border Air Quality Study Team
• University of Washington Catherine Karr, Kristin
  Miller, Jane Koenig, Tim Larson, Tim Gould
• ckarr_at_u.washington.edu

2
Study Questions

• Primary
• Do longer term (chronic) increases in ambient air
  pollution impact risk of hospitalization or
  clinic visits for bronchiolitis?
• Which pollutants and sources (traffic, woodsmoke)
  are most important?
• Secondary
• Do shorter term (acute-subchronic) increases in
  air pollution increase risk?

3
Study Design Overview

• 1. Identify all births and one year health follow
  up in Puget Sound region
• 2. Assign PM2.5 exposures to individual subjects
• ambient monitor network data (version 1)
• ?
• geospatial traffic indicators
• ?
• modeling traffic exposure and woodsmoke
  (versions 2,3)
• 3. Explore risk of increasing exposures on
  bronchiolitis
• focus of presentation today

4
Version I traditionalExposure Assessment

• Exposure windows constructed based on daily
  averages of PM2.5 at nearest monitor (within 20
  km)
• Chronic exposure Lifetime average (from date
  left hospital after birth to date of
  hospitalization for bronchiolitis)

5
Health Data Source

• WA State Dept. of Health Birth Events Registry
  Database (BERD)
• BERD birth certificate/vital statistics data
  linked to hospitalization data for all
  mother/infants
• Includes geocoded residential address at birth

6
Health Data Source

• Eligible subjects
• Singleton, livebirths for King, Snohomish,
  Pierce, Kitsap county
• Ambient PM2.5 monitor within 20 km of geocoded
  birth address
• Information on gestational age

7
Population based Cohort

• Puget Sound
• 241,136 infant livebirths available for analysis
  (1997-2002)
• 3,113 hospitalization cases for bronchiolitis
• ICD 9CM 466.11, 466.19 (RSV vs. other infectious
  etiology)

8
Puget Sound PM2.5 and bronchiolitis
hospitalization

• Analysis nested case control, conditional
  logistic regression
• All cases from original birth cohort and up to 10
  randomly selected controls matched on (/-7
  days)
• date of birth
• gestational age
• length of birth hospitalization
• Case N 2,770, Control N 24,009

9
Confounders

• Infant sex
• Mothers smoking during pregnancy
• Infant race/ethnicity
• Parity
• Mothers education
• Insurance type
• Public assistance program involvement

10
Exposure windows

• Exposure windows for case and their matched
  controls all referenced to case admission date
• Since matched on DOB, exposure windows are all
  contemporaneous
• Variability in exposure is spatial only

11
Subject Chronic PM2.5 exposure
12
Odds ratios for bronchiolitis hospitalization by
quartiles of chronic PM2.5 exposure
13
Adj OR bronchiolitis hospitalization per 10
mcg/m3 ? chronic PM2.5 exposureby age at
diagnosis
14
Adj OR bronchiolitis hospitalization per 10
mcg/m3 ? chronic PM2.5 exposureby age at
diagnosis
15
Sensitivity analysis of subject distance to
monitor
Adjusted for sex, smoking during pregnancy, mom
education, parity, family income, infant
race/ethnicity
16
Odds ratios for bronchiolitis hospitalization and
acute-subchronic PM2.5 exposures
17
Summary Effects of PM2.5 on infant bronchiolitis
in the Puget Sound

• Suggests increased risk in bronchiolitis
  hospitalization with higher PM2.5 exposure but
  not statistically significant
• Effect estimates highest for long term exposure
  window (chronic) vs acute-subacute exposure
  windows

18
Summary Effects of PM2.5 on infant bronchiolitis
in the Puget Sound

• Sensitivity analyses revealed effects more
  prominent for
• RSV Blitis compared to Blitis due to other
  infectious agents
• Youngest infants (diagnosed in first 3 months of
  life)
• Subjects with monitor within 5 km

19
Next steps

• Increase sample size by adding subjects born in
  2003
• Evaluate effects with version 2 (improved spatial
  resolution) exposure assessment
• ? non differential exposure misclassification
• More variability in exposure among subjects

20
Next steps

• Evaluate effects of PM2.5 on infant mortality
• Evaluate effects on pre term birth and low birth
  weight

21
Thanks. Questions/comments?