air pollution and public health limited time, we will discuss: asthma ischemic stroke otitis media...
TRANSCRIPT
Air Pollution and Public Health
• Limited time, we will discuss:
• Asthma
• Ischemic Stroke
• Otitis Media
• Myocardial Infarction
Co-Authors for many projects
• Leonard Bielory, MD
• Van Dunn, MD MPH
• Susan Meehan RPh
• William Gauff, EMT
• Yu-Feng Chan, MD
• Hosseinali Shahidi, MD MPH
• Ronald B. Low MD
Most Data from New York City
• Hospital data available to me now from NYC Health and Hospitals Corporation (HHC)
• Pollution data from EPA
• Weather data from NWS
• Pollen data from Dr. Bielory
• NJ, NYC Issues pretty much the same
• Inferred NJ calculations
Similar atmospheric Conditions: NJ & NYC
• Weather
• Pollution
• Pollens: Only measured in NJ
NOx: NY and NJ
• Not as close as temperature, but clearly related (p<.0001)
• The closer to NYC, the tighter the relationship
• Generally, NJ levels only slightly lower than NYC levels: both 0.01-0.20 ppm
Asthma
• One of the most studied diseases related to air pollution
• Our model is conservative, ascribing changes in asthma rates to:
• Time (seasons)
• Then weather and airborne allergens
• Last pollution effects
Basic Statistics
• NJ: 16,390 admissions in 2003
• HHC: 15,914 admission in 2003
• HHC: 59,865 ED Visits for asthma in 2003
Other effects
• Confounded with temperature, hard to show graphically in 3 or more dimensions
Model
• Time modeled first: Autoregressive effects of 1,2 and 7 days earlier; moving average effects of 6 and 365 days earlier
• 4 visits/day increase with weed pollen count increase of 1000
• 100 more URI visits1 more asthma
• 10μg/m3 particles<10μ, 2 more visits
Rough estimate of the effect small particles on NJ asthma
• Assuming causality
• Conservative Model
• 900 additional admissions in 2003
• 3000 additional ED Visits in 2003
Ischemic Stroke
• Not as seasonal/time dependent as asthma
• Weekly effect and holiday effects. Conservatively, we adjust for them before looking for a pollution effect.
Ischemic Stroke
• NJ 2003:– 21,899 Admissions for all ischemic events
• NYC study average 9.34 strokes/day*2556 days. In 2003:– 2615 Admissions for all ischemic events– 1338 Ischemic Strokes
• Entire 8 year NYC: 23,888 ischemic strokes
Weekend, Holiday Effects
• Holiday=New Years, Martin Luther King, President’s Day, Easter, Memorial Day, July 4, Labor Day, Thanksgiving, Christmas
• Average 0.9 fewer strokes (0.7-1.3) on weekends (p<0.0001)
• Average 1.1 fewer strokes (0.5-1.6) on holidays, p=.0002
White Stroke and Pollution
• Probably a real NOx effect, p=0.0455.
• Best modeled as a logarithmic effect (Normality)
• Average effect 0.47 strokes/day, 2118 strokes during study, adjusted for weekends, holidays and temperature
• Assuming causality, ESTIMATED 2003 NJ effect: 1,900 strokes; 95th % 112 strokes
Otitis Media
• Like asthma, seasonal and weekly effects
• We looked at clinic visits as well as ED visits
• Again, we model conservatively
Otitis visits: Basic Statistics
• HHC:– 809,252 visits during study– Average 181/day (lots of seasonal variability)– 2003: 55,533 visits
• NJ: We do not have outpatient data
Otitis Model
• Except for NOx, I will not discuss coefficients: Log transform makes interpretation difficult– Temperature: lower is worse, p=0.0144– Weekends better than weekdays, p<.0001– Holidays better than workdays, p<.0001– 365 day seasonal pattern, p<.0001– URIs make otitis worse, p<.0001
Effect of NOx on NJ Otitis
• The log*log effect means that high levels are more problematic than low levels
• Assuming Causality: Lowering NOx to low levels should reduce OMV visits by between 2% to 8%, depending on starting levels and other assumptions.
Model Works Prospectively
Myocardial Infarction
• NJ in 2003: 22,464
• HHC: – in 2003: 2,623– Entire study: 22,371 (very close to NJ 2003)
MI Model
• No significant seasonal effect• Worse on weekdays (p<.0001)• Worse as temperature drops, p<.0001• No significant snow effect• Curvilinear exacerbating effect of NOx,
worse at highest levels, p=0.0051. The effect is statistically small except until around the top 5th% (>.14ppm). Difficult to accurately estimate effect.
Reservations
• Observational Studies– Limited to temperature and pollutant ranges
we observed– Diagnostic accuracy dependent upon
clinicians and coders– Pollutants co-correlated with each other, with
weather, seasons, ?weekday traffic– These studies do not prove causality– Pollutant effects may be underestimated, they
were always added last to model
Overall Conclusions
• Air pollution, at current levels, has some measurable relationship to asthma, otitis media, ischemic strokes and MI.
• NOx and suspended particles have the most widespread associations.
• If you assume causality, the health effects are significant
Questions?