1. “Indoor Air Pollution” Framing Issues Risk Factor Exposure Outcomes Background Disease Counterfactual

2. Risk Factor Household solid fuel use –Clearly polluting in nearly all settings –Available in household surveys –Most common metric in epi studies Expands risk boundary –Portion of outdoor air pollution –Burns –Other non-air-pollution pathways, e.g. risks of gathering biomass fuel and coal mining -? –Even a portion of climate change due to household fuels-? –Include high-income countries where much wood is used in households, but not for cooking-?

3. Difficulties of Household Solid Fuel Use as RF Exposures are not uniform because –Coal and biomass are different –Amount of usage usually not known –Difficulty of separating cooking from the heating often done in non-tropical settings –Chimneys dominate in richer countries and are found in poor ones – effectiveness variable –Location of cooking in household not uniform, which affects exposure –A set of the major interventions being explored are low-emissions biomass stoves

4. Exposure Few studies used direct exposure assessment, i.e., average air pollution concentration affecting people, which roughly corresponds to annual means used in outdoor air pollution studies Reviews of health effects need to capture a wide range of indicators of exposures, incl: –Solid fuel vs. another (cleaner) fuel –Time spent near stove –Duration of exposure (years) to solid fuels –Separate kitchen from main house –Improved vs. Traditional stove How to combine? And do so in a way consistent with available household fuel inventories

5. Outcome Measures A growing number of outcomes are associated with exposures –Some have dozens of epi studies, i.e. ALRI, COPD, lung cancer from coal – outcomes from CRA-2000 –Some have less than 10 with consistent results, i.e., LBW and cataracts –Some have less than 10 with some inconsistency, i.e., TB, lung cancer from biomass –Some have few studies but would be expected from other pollution exposures (outdoor air, ETS, and smoking), i.e., heart disease, adult pneumonia, and other cancers –Some have no studies but would be expected from animal studies, i.e., birth defects and cognitive impacts Need to decide cut off, i.e., which to include in quantification

6. What to do about Heart Disease? CVD is probably the principal outcome from all other forms of smoke exposure –Active smoking –Passive smoking (ETS) –Outdoor air pollution Considerable toxicological/physiological (sub- clinical) evidence Although case-control studies underway, only one related study published: blood pressure in Guatemalan RCT The inconsistency was side-stepped in the CRA- 2000, but must be addressed this time since solid fuel portion of outdoor air pollution is being assessed.

7. Chronic Cardiovascular Effects of Ambient Air Pollutants Brooks et al. Circulation 2004

8. RESPIRE and Blood Pressure (BP) RESPIRE: a randomized control trial monitoring the health impacts in women and children of introducing a woodburning stove with chimney in ~260 Guatemalan households compared to ~260 control households still using open woodfires for cooking

9. RESPIRE Results “Control w/ stoves” values are the average reduction seen in BP of the control group after receiving the new stoves *Adjusted for age, BMI, asset index, smoking, secondhand tobacco smoke, apparent temperature, season, day of week, time of day, random subject intercept SBP (95% CI) DBP (95% CI) Intervention Group -3.7 mmHg (- 8.1, --0.6) -3.0 mmHg (-5.7, -0.4) Control w/ stoves -3.1 mmHg (- 5.3, --0.8) -1.9 mmHg (-3.5, -0.4) McCracken et al., 2007

10. Global Deaths from BP due to IAP Estimated cardiovascular deaths in 2004 attributable to elevated BP caused by solid fuel smoke exposure. Number of deaths preventable by a 50% reduction in solid fuel smoke exposure by WHO sub-region and disease category. McCracken and Schwartz, 2009

11. Heart Disease? Options to fill gap –Use BP results Only one study Not pathway thought to dominate CVD effect –Use exposure estimates to apply against outdoor and ETS results Exposure response probably not linear Do not have exposure estimates for most of world, although hoping to have for India –Use COPD to calibrate, i.e., apply COPD/CVD ratio in outdoor and ETS studies Ratio probably different at different exposure levels Some evidence that biomass smoke less dangerous for CVD than fossil fuel smoke, although similar for respiratory outcomes

12. Background Disease For applying PAF values –Separate rates for rural and urban pops? –Separate rates for smoking and non-smoking pops? How to deal with inability to separate LBW into IUGR and prematurity?

13. Counterfactual In CRA-2000, non-solid fuel use, i.e., electricity, gas, or kerosene –Roughly fit available epi and HH survey data –Clearly feasible – half world already there Problems: physical –Kerosene not clean in many circumstances –Even gas stoves have been associated with ill- health in dozens of first world studies Problems: methodological –Clear that there is much exposure misclassification with use of binary variables –True low-exposure control groups rare or non- existent in current studies due to high levels of confounding with SE status

14. Simple Measurements Mixed Model Predictions

15. MD-diagnosed ALRI RESPIRE- Guatemala Approximate PM2.5 exposure in 100s of ug/m3

16. Severe (hypoxic) MD-ALRI RESPIRE- Guatemala Approximate PM2.5 exposure in 100s of ug/m3

17. Many thanks Solid-Fuel-Use CRA Website: http://ehs.sph.berkeley.edu/krsmith/page.asp?id=25