Risk of COPD From Exposure to Biomass Smoke
Risk of COPD From Exposure to Biomass Smoke
Papers published in the MEDLINE database, the Latin American and Caribbean Literature in Health Sciences Database, and EMBASE were searched (up to January 2009) with key words including "COPD," "chronic bronchitis," "emphysema," "chronic obstructive pulmonary disease," "biomass fuel," " biofuel," "organic fuel," "wood," and "indoor air pollution." Articles about relevant studies in the references were also obtained. Only studies published in the English language were included in the analysis. We communicated with some of the authors for additional data that did not appear in the text, and also communicated with the Chinese COPD Alliance and e-mailed researchers on COPD outside China for data from unpublished or ongoing studies.
All potentially relevant manuscripts were reviewed independently by two investigators. Areas of disagreement or uncertainty were adjudicated by other investigators. For studies to be included in the metaanalysis, they had to meet the following criteria: (1) they had to contain a case-control or cross-sectional study design; (2) they had to have compared at least two groups (COPD vs control, or biomass smoke vs control); (3) they had to show odds ratios (ORs) to estimate the association between COPD and biomass smoke with corresponding 95% CIs, or with sufficient data for calculation; and (4) they had to be independent from other studies. Studies with the same data sets as already published studies were not deemed to be independent. As such, only studies with large sample sizes and sufficient information for data extraction were metaanalyzed. No limitations were set for participants' ages or definition of exposure to biomass smoke as used in individual studies. Two major phenotypes of COPD, emphysema and chronic bronchitis, were included, although some cases are not characterized by airflow limitation that is not fully reversible. Chronic bronchitis was clinically diagnosed as chronic productive cough for 3 months in each of 2 successive years with no known causative factors. Because epidemiologic surveys have shown undiagnosed COPD in about two-thirds of subjects, studies in which the diagnostic criterion was a history of ever-diagnosed COPD, chronic bronchitis, or emphysema were excluded from the primary metaanalysis but used in a sensitivity analysis. This is because if some undiagnosed COPD were included in the control group, it could result in a differential diagnostic bias between biomass-exposed and nonexposed subjects. Case-control studies with a demonstrated source and matching control, criteria of exposure, and definition of COPD, or cross-sectional studies with demonstrated criteria of exposure and definition of COPD, were considered high-quality studies.
All data were extracted independently by two investigators using a standardized protocol and data-collection forms. Disagreements were resolved by discussion. The studies were recorded as follows: first author, year of publication, study design (case-control study or cross-sectional study), characteristics of the study subjects (definition of exposure and nonexposure to biomass smoke, diagnosis criteria of COPD, sample size, age, gender, ethnicity, status of cigarette smoking, and duration of exposure to biomass smoke), measures of outcome and exposure, the ORs of COPD associated with biomass smoke, and standard errors (overall and in each subgroup, according to gender and smoking status).
Metaanalysis was performed using Stata, version 7.0, statistical software (Stata Corporation; College Station, TX). The heterogeneity among studies was examined with the χ-based Q statistic test. Depending on the presence of heterogeneity between studies, either a random effect model or a fixed effect model was used. The ORs of COPD associated with biomass smoke were estimated using nonexposure to biomass smoke as the reference. Subgroup analyses were performed with stratifications by cigarette-smoking status, sex, study design, ethnicity, duration of biomass smoke, and the phenotypes (lung function test-diagnosed COPD, emphysema, or chronic bronchitis). We further examined the relationship between exposure to biomass smoke and COPD by including the studies in which the diagnostic criterion was a history of having been given a diagnosis of COPD, chronic bronchitis, or emphysema. The significance of pooled ORs was determined by z test. All statistical tests were rendered two tailed, and P < .05 was considered significant. Potential for publication bias was assessed using the Egger test and funnel plots. Another method of identifying publication bias was the number of unpublished studies that would have to exist to negate the results of the metaanalysis.
Materials and Methods
Search of the Literature
Papers published in the MEDLINE database, the Latin American and Caribbean Literature in Health Sciences Database, and EMBASE were searched (up to January 2009) with key words including "COPD," "chronic bronchitis," "emphysema," "chronic obstructive pulmonary disease," "biomass fuel," " biofuel," "organic fuel," "wood," and "indoor air pollution." Articles about relevant studies in the references were also obtained. Only studies published in the English language were included in the analysis. We communicated with some of the authors for additional data that did not appear in the text, and also communicated with the Chinese COPD Alliance and e-mailed researchers on COPD outside China for data from unpublished or ongoing studies.
Study Selection
All potentially relevant manuscripts were reviewed independently by two investigators. Areas of disagreement or uncertainty were adjudicated by other investigators. For studies to be included in the metaanalysis, they had to meet the following criteria: (1) they had to contain a case-control or cross-sectional study design; (2) they had to have compared at least two groups (COPD vs control, or biomass smoke vs control); (3) they had to show odds ratios (ORs) to estimate the association between COPD and biomass smoke with corresponding 95% CIs, or with sufficient data for calculation; and (4) they had to be independent from other studies. Studies with the same data sets as already published studies were not deemed to be independent. As such, only studies with large sample sizes and sufficient information for data extraction were metaanalyzed. No limitations were set for participants' ages or definition of exposure to biomass smoke as used in individual studies. Two major phenotypes of COPD, emphysema and chronic bronchitis, were included, although some cases are not characterized by airflow limitation that is not fully reversible. Chronic bronchitis was clinically diagnosed as chronic productive cough for 3 months in each of 2 successive years with no known causative factors. Because epidemiologic surveys have shown undiagnosed COPD in about two-thirds of subjects, studies in which the diagnostic criterion was a history of ever-diagnosed COPD, chronic bronchitis, or emphysema were excluded from the primary metaanalysis but used in a sensitivity analysis. This is because if some undiagnosed COPD were included in the control group, it could result in a differential diagnostic bias between biomass-exposed and nonexposed subjects. Case-control studies with a demonstrated source and matching control, criteria of exposure, and definition of COPD, or cross-sectional studies with demonstrated criteria of exposure and definition of COPD, were considered high-quality studies.
Data Extraction
All data were extracted independently by two investigators using a standardized protocol and data-collection forms. Disagreements were resolved by discussion. The studies were recorded as follows: first author, year of publication, study design (case-control study or cross-sectional study), characteristics of the study subjects (definition of exposure and nonexposure to biomass smoke, diagnosis criteria of COPD, sample size, age, gender, ethnicity, status of cigarette smoking, and duration of exposure to biomass smoke), measures of outcome and exposure, the ORs of COPD associated with biomass smoke, and standard errors (overall and in each subgroup, according to gender and smoking status).
Statistical Analysis
Metaanalysis was performed using Stata, version 7.0, statistical software (Stata Corporation; College Station, TX). The heterogeneity among studies was examined with the χ-based Q statistic test. Depending on the presence of heterogeneity between studies, either a random effect model or a fixed effect model was used. The ORs of COPD associated with biomass smoke were estimated using nonexposure to biomass smoke as the reference. Subgroup analyses were performed with stratifications by cigarette-smoking status, sex, study design, ethnicity, duration of biomass smoke, and the phenotypes (lung function test-diagnosed COPD, emphysema, or chronic bronchitis). We further examined the relationship between exposure to biomass smoke and COPD by including the studies in which the diagnostic criterion was a history of having been given a diagnosis of COPD, chronic bronchitis, or emphysema. The significance of pooled ORs was determined by z test. All statistical tests were rendered two tailed, and P < .05 was considered significant. Potential for publication bias was assessed using the Egger test and funnel plots. Another method of identifying publication bias was the number of unpublished studies that would have to exist to negate the results of the metaanalysis.