Risks for respiratory NCDs from exposure to HAP: an overview of the evidence Dr Nigel Bruce, University of Liverpool, UK
Overview Life course perspective useful Disease risks How may inter-relate over lifespan Pre-conception Pregnancy Young child Older child Adult Mother s health; other (?) Epigenetic (?); birth weight ALRI and lung growth Asthma and wheeze Asthma COPD Lung cancer Other cancers Asthma Older person COPD and exacerbations Lung cancer Other cancers Epidemiological evidence (GDB-2010 etc.)
Birth weight Pope et al (2010); Bruce et al (2013) 7 studies (9 estimates): 1.40 (1.26, 1.54) 5 studies with adequate adjustment: 1.57 (1.33, 1.86) HAP effect: -97 gm (5 studies) Supportive analogous evidence from studies of active (larger effect) and second-hand smoking (smaller effect) Evidence from OAP less conclusive
Impact of (HAP) on lung growth Lung growth: First 3 years new alveolae developed Later childhood lung growth by expansion 12 Community California study: Most vs. least polluted communities School-age (mean 10 years) Assessed difference in annual lung growth: Poll FVC (ml) FEV 1 (ml) PM 2.5-60 (NS) -80 (0.04) EC -78 (0.08) -88 (0.007) Conclude AP results in clinically significant FEV 1 deficits as children reach adulthood: Kulkani et al, NEJM (2006)
Asthma: ISAAC study Wong et al (2013) 108 centres; 48 countries >500,000 school=age children: 6-7 years 13-14 years ISAAC questionnaire: Wheeze in past year Samples have relatively low use of solid fuels (but large numbers) Systematic review (Wong et al): cooking and asthma ; biomass fuel and asthma 2 studies found Inconsistent results % Fuel use Latin America Indian S/C Africa 0 5 10 15 20 25 Any OF OF only Age group Fuels OR 95% CI 6-7 years OF only 2.17 1.64 2.87 Any OF 1.51 1.25 1.81 13-14 years OF only 1.35 1.11 1.64 Any OF 1.35 1.15 1.58
COPD Systematic reviews: Stern-Nezer and Balmes (GBD) 3 others report similar findings 24 studies (26 estimates) Clinical diagnosis (ATS/CB) = 13 Prior physician diagnosis (COPD) = 2 Spirometry (obstruction) = 6 Spirometry + clinical = 3 Headline results: Women OR= 2.30 (95% CI: 1.73, 3.06) Men OR= 1.90 (95% CI: 1.15, 3.13) Some evidence of larger effect for biomass vs. coal Intervention studies Analogous evidence: tobacco Cohort study from Xianwei: Coal using area Chimney stoves 10+ years of use Interview-based recall of diagnosed CB an COPD (Minimal) HAP or exposure data available OR Chapman (2005) 95% CI Women 0.75 0.62, 0.92 Men 0.58 0.49, 0.70
Risk of COPD with solid fuel
Intervention-based cohort study in Guangzhou Sample: Rural biomass users No ventilation Cohort: N = 996; 9 year follow-up Around 30% smokers Interventions: Chimney/extractor Biogas Progressive adoption HAP/exposure: Breath CO 5 pollutants while cooking Outcomes: Spirometry COPD (FEV1/FVC <70% post bronchodilator) NB: during cooking Zhou et al (2014)
Results Reduced annual decline in FEV 1 Group ml 95% CI Ventilation 12 4-20 Clean fuel* 13 4-23 Both 16 9-23 * Partial finance for biogas provided, but includes biogas, LPG and electricity
Results Reduced annual decline in FEV 1 Group ml 95% CI Ventilation 12 4-20 Clean fuel* 13 4-23 Both 16 9-23 Duration-effect trend (COPD) OR (95% CI) Years Ventilation Clean fuel* 0 1.00 1.00 1 4.9 0.52 (0.23, 1.19) 5-9 0.39 (0.15, 0.99) 0.64 (0.30, 1.36) 0.33 (0.10, 1.03 P-value 0.053 0.017 * Partial finance for biogas provided, but includes biogas, LPG and electricity
IER curve: COPD PM 2.5 µg/m 3 ) Burnett et al, EHP 2014
Restriction or Obstruction? Key findings from the BOLD study (Burney et al 2014) Suggests that high mortality from COPD unlikely to be due to obstruction important implications for assessing impacts of HAP Ecological analysis of national and 22 BOLD site data But findings for restriction (60-70% in LIC) seem questionable? Negative (statistically significant) associations between COPD mortality (WHO) and GNI consistent with increased risk from solid fuel use? Negative association with smoking prevalence needs explaining Centre Men Women P/years Obstruct. Restrict. P/years Obstruct. Restrict. Pune 1.2 5.7 63.1 0.1 6.8 70.5 Srinagar 24.2 17.3 25.3 1.3 14.8 31.2 Manila 18.7 15.1 62.4 2.7 4.2 62.9 Philippines 20.6 16.9 52.7 3.3 13.5 61.0
COPD mortality vs. poverty & smoking: (by country ecological analysis) Women Men COPD mortality (WHO) GNI (poverty) Smoking prevalence
Lung cancer IARC Monograph 95 (2010) Coal SRMA (Hosgood et al IJE) Biomass SRMA (Bruce - submitted) IER (Burnett et al 2014) Evidence Coal Biomass Human Sufficient Limited Animal Sufficient Sufficient Mechanism +++ +++ Group 1 2(a)
HH coal use and lung cancer IARC: Emissions are Group 1 carcinogen Summary of results (GBD-2010): Women: 1.98 (1.16, 3.36) Men: 1.31 (1.05, 1.76)
Biomass fuel use and lung cancer
Headline results Men (n=4): OR=1.15 (1.01, 1.31) Women (n=12): OR=1.20 (0.95, 1.52) Men and women (n=1): OR=2.50 (1.00, 6.25) Overall (n=13): OR=1.17 (1.01, 1.37)
Biomass: results for studies of men
Results Men Fixed effects meta-analysis Analysis group N studies I 2 OR (95% CI) All studies 4 0% 1.15 (1.01, 1.31) Moderate/good adjustment; clean fuel comparison Excluding kerosene in comparison group 2 0% 1.21 (1.05, 1.39) 1* N/A 1.19 (1.02, 1.39) *Pooled result from 4 studies in high/middle income countries
Biomass: results for studies of women
Results - Women Group Sensitivity analysis N Heterogeneity OR (95% CI) # Sub-group studies (I 2 ; p-value) FE=Fixed Effects All 1 All studies 12 51% (p=0.008) 1.20 (0.95, 1.52) 2 Clean fuel comparison 5 56% (p=0.04) 1.95 (1.16, 3.27) Design 3 Hospital 9 57% (p=0.005) 1.31 (0.95, 1.80) 4 Population 2 0 % 0.89 (0.58, 1.37) (FE) 5 Mixed 1 N/A 1.19 (0.94, 1.51) Strong or 6 All 11 51% (p=0.01) 1.23 (0.98, 1.55) moderate 7 Clean fuel comparison 5 56% (p=0.04) 1.95 (1.16, 3.27) adjustment 8 Clean fuel comparison, excluding kerosene in comparison group 4 64% (p=0.03) 2.33 (1.23, 4.42) Asia and Mexico 9 All 11 53% (p=0.006) 1.22 (0.93, 1.61) 10 Clean fuel comparison 4 21% (p=0.28) 2.33 (1.46, 3.72) (FE) 11 Clean fuel comparison, excluding kerosene in comparison group 3 0% 3.26 (1.88, 5.65) (FE) Europe, North America, Brazil 12 All (all clean fuel) 1 N/A 1.19 (0.94, 1.51) Non-smokers 13 All 4 66% (p=0.007) 1.14 (0.78, 1.67) only 14 Clean fuel comparison 1 N/A 2.75 (0.85, 8.86)
Exposure-response evidence 5 published studies reported results by duration of exposure, but inconsistent analysis and findings Re-analysis of Lissowska et al data by IARC (wood fuel for heating and/or cooking): Duration of use* Men Women OR** 95% CI OR** 95% CI None 1.00-1.00 - <25% 1.06 0.81 1.39 0.99 0.55 1.79 25-50% 1.13 0.90 1.43 1.05 0.64 1.73 >50% 1.37 1.03 1.81 1.02 0.59 1.79 P-value (trend) 0.03 0.87 *Percentage of lifetime exposed **Adjusted for centre, age, education, tobacco years and coal fuel use
IER Lung cancer Consistent with China cohort study results for impact of long term use of chimney coal stoves: Men: 0.59 (0.49 to 0.71) Women: 0.54 (0.44 to 0.65) Lan et al 2002 PM 2.5 µg/m 3 )
Upper aero-digestive tract cancers Cancers of the nasopharynx: 5 eligible studies; high heterogeneity (I 2 = 89%) Pooled OR = 1.10 (0.98 1.24); weak on quality One study used clean fuel comparison: 1.60 (0.39, 6.50) Cancer of larynx, oropharynx and hypopharynx: Study or Subgroup log[odds Ratio] SE Weight Odds Ratio IV, Random, 95% CI Year Odds Ratio IV, Random, 95% CI Franco - Wood stove Dietz - Fossil fuel (FE) Pintos - Wood stove Sapkota - Wood (FE) Sapkota - Coal (FE) 0.916291 0.48858 0.896088 0.254642 0.779325 0.227289 0.192289 0.145909 0.12929 0.359484 18.5% 20.8% 23.9% 25.0% 11.7% 2.50 [1.60, 3.90] 1.63 [1.12, 2.38] 2.45 [1.84, 3.26] 1.29 [1.00, 1.66] 2.18 [1.08, 4.41] 1989 1995 1998 2008 2008 Total (95% CI) 100.0% 1.90 [1.39, 2.59] Heterogeneity: Tau² = 0.08; Chi² = 13.74, df = 4 (P = 0.008); I² = 71% Test for overall effect: Z = 4.06 (P < 0.0001) 0.2 0.5 1 2 5 FE: studies with multiple estimates pooled using fixed effects meta-analysis
Exposure to solid fuel not going away
Summary and conclusions Life course approach valuable: Evidence on range of important steps and outcomes for COPD Further study needed to understand overall impacts across lifespan Focus on key periods (disease markers) and longer-term cohorts Intervention studies starting to show effects but long follow-up Other new evidence: Stronger for childhood asthma but SR(MA) needed? Stronger case of causal link between biomass fuels and Ca lung Issue of obstructive vs. restrictive needs clarifying Policy: HAP not going away Reduction of HAP needs to be central to NCD strategy Link to control of active and second-hand smoking, and other sources of air pollution
Thanks!
Geographical variation in risk: China Hosgood et al, IJE 2011
No evidence of publication bias Test results (Egger s): Women: p=0.59 Men: p=0.59 All studies: p=0.48
Evidence for causal inference: Summary of assessment with Bradford Hill viewpoints Viewpoint Assessment of evidence 1 Strength of association Women, clean fuel, adjusted: 1.6 to >2.0 2 Consistency across populations, study designs Men: no statistical heterogeneity Women: high I 2 ; less with geog. stratification 3 Specificity Biomass linked to multiple outcomes; N/A 4 Temporality (exposure precedes outcome) 5 Biological gradient (doseresponse) Although retrospective, exposure lifelong from early life (incl. in utero) Significant exposure-risk trend in Lissowska (men) Higher vs. lower income regions consistent 6 Biological plausibility Strong evidence (IARC assessment) 7 Coherence with natural history, animal studies Higher risk in countries where exposures higher IARC (animal): sufficient for carcinogenicity 8 Experiment No experimental evidence in humans 9 Analogy Other combustion: outdoor AP, SHS and AS
Prospects for clean fuels: Africa IEA (2014)
Summary Pre-conception Pregnancy Young child Older child Mother s health; other (?) Epigenetic (?); birth weight ALRI and lung growth Asthma and wheeze Asthma Adult COPD; ILD (?) Lung cancer: coal/biomass Asthma UADT cancers Older person COPD and exacerbations Lung cancer: coal/biomass UADT cancers