You are here

Article Text

Article menu

Download PDFPDF

Environmental exposures and health
Tuberculosis risk from exposure to solid fuel smoke: a systematic review and meta-analysis
  1. Om P Kurmi1,
  2. Charandeep S Sadhra2,
  3. Jon G Ayres3,
  4. Steven S Sadhra3
  1. 1Clinical Trial Service Unit & Epidemiological Services Unit, University of Oxford, Oxford, UK
  2. 2College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK
  3. 3Institute of Occupational and Environmental Medicine, University of Birmingham, Birmingham, UK;
  1. Correspondence to Dr Om P Kurmi, Clinical Trial Service Unit and Epidemiological Studies Unit, Nuffield Department of Population Health, University of Oxford, Richard Doll Building, Old Road Campus, Roosevelt Drive, Oxford, OX3 7LF, UK; om.kurmi{at}ndph.ox.ac.uk

Abstract

Background Studies, particularly from low-income and middle-income countries, suggest that exposure to smoke from household air pollution (HAP) may be a risk factor for tuberculosis. The primary aim of this study was to quantify the risk of tuberculosis from HAP and explore bias and identify possible causes for heterogeneity in reported effect sizes.

Methods A systematic review was conducted from original studies. Meta-analysis was performed using a random effects model, with results presented as a pooled effect estimate (EE) with 95% CI. Heterogeneity between studies was assessed.

Results Twelve studies that considered active tuberculosis and reported adjusted effect sizes were included in the meta-analyses. The overall pooled EE (OR, 95% CI) showed a significant adverse effect (1.43, 1.07 to 1.91) and with significant heterogeneity between studies (I2=70.8%, p<0.001). When considering studies of cases diagnosed microbiologically, the pooled EE approached significance (1.26, 0.95 to 1.68). The pooled EE (OR, 95% CI) was significantly higher among those exposed only to biomass smoke (1.49, 1.08 to 2.05) when compared with the use of kerosene only (0.70, 0.13 to 3.87). Similarly, the pooled EE among women (1.61, 0.73 to 3.57) was greater than when both genders were combined (1.39, 1.01 to 1.92). There was no publication bias (Egger plot, p=0.136). Significant heterogeneity was observed in the diagnostic criteria for tuberculosis (coefficient=0.38, p=0.042).

Conclusions Biomass smoke is a significant risk factor for active tuberculosis. Most of the studies were small with limited information on measures of HAP.

  • TUBERCULOSIS
  • Environmental epidemiology
  • AIR POLLUTION

https://doi.org/10.1136/jech-2014-204120

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Approximately half of the world's population is exposed to household air pollution (HAP) through daily burning of domestic fuels for cooking and heating, particularly those from low-income and middle-income countries (LMIC). HAP is an important pollutant which was responsible for 3.5 million deaths in 2010.1 Although there has been a 46% decline in the percentage of age-adjusted tuberculosis (TB) deaths over the past 20 years, the total number of TB deaths in 2010 was 1.2 million globally.1 While there is consistent evidence for an association between tobacco smoking and TB, the evidence for the link between TB and HAP warrants investigation, in particular the role of household smoke generated by burning solid fuels.

    The prevalence of TB infection is highest among sub-Saharan Africa and Southeast, Central and South Asian populations where the majority of people use solid fuel for domestic purposes.1 Smoking is a risk factor of tuberculosis2 but in areas such as South and South East Asia, where the prevalence of smoking in women is much lower than that in men, the gender prevalence of TB remains almost equal (4.6% vs 5.1%).1

    It has been suggested that exposure to solid fuel might be a risk factor for TB.3 However, systematic reviews with meta-analysis4–6 to date report differing levels of evidence to support such an association, perhaps due to the small number of studies available and different fuel types used by households in different parts of the world. Fine particulate matter (PM2.5) generated by incomplete combustion of solid fuel has been associated with bronchial irritation, inflammation, increased reactivity, reduced mucociliary clearance, reduced macrophase response and reduced local immunity that leads to poor lung health6–8 and, consequently, potential tuberculosis infection.

    A number of additional studies in this area have recently been published, so we carried out a systematic review and meta-analysis of peer-reviewed journal articles to determine the association between exposure to smoke from solid fuel burning and risk of tuberculosis. We also tested for bias and explored possible causes of heterogeneity in reported effect sizes. Heterogeneity has been recognised previously in studies reviewing the evidence of the link between TB and HAP but has not been fully investigated.

    The primary aims of this study were to (1) review the existing literature that quantifies the risk of tuberculous disease from exposure to smoke from solid fuel burning; (2) quantify the pooled effect estimate (EE) for active tuberculous disease among adults.

    Methods

    A systematic review and meta-analysis of peer-reviewed publications reporting the association between HAP and TB was carried out adhering to the guidelines laid out in Preferred Reporting Items of Systematic reviews and Meta-Analysis (PRISMA).9

    Data source

    We performed searches using three databases, PubMed, EMBASE and LILIACS, from 1947 to January 2014 to identify all epidemiological studies that investigated the association between HAP and TB. Searches were also conducted using the Cochrane database for meta-analysis studies and other original studies so as to identify any additional studies. Bibliographies of peer-reviewed papers were also screened to ensure that our searches had not missed any relevant peer-reviewed publications.

    View this table:

    Search terms and strategy

    Study selection

    The searches were limited to human studies only. No language restrictions were imposed during the searches, with papers written in languages other than English being read and interpreted by fellow researchers with a good understanding of the relevant languages. Published full-length peer-reviewed studies (cohort, case–control or cross-sectional) were included. Published conference abstracts, government reports and letters to journals were excluded.

    In conducting the systematic review, we included studies for both adults and children with active TB defined primarily by microbiological criteria (participant with one or more sputum smear alcohol-fast bacilli-positive) but also by doctor diagnosed active TB, where microbiological proof may have been found but was not reported. Meta-analysis included case–control studies which defined cases as active or cases from nationally recognised TB registers or cross-sectional studies with self-reported but doctor diagnosed active TB.

    Data extraction and quality assessment

    A detailed assessment was conducted for each of the selected studies on the effect of HAP on TB. For each study, information was extracted on the following: year and country of study, study design, study population (matched or unmatched), types of pollution assessed, confounder(s) adjusted for, quantitative or qualitative measurement of exposure, type of tuberculosis (pulmonary or extra pulmonary; active or latent) and type of risk estimate reported. This information was extracted independently by CSS and SSS using a predefined template. Any disagreement in reporting between CSS and SSS was reviewed by OPK. Data on ORs and 95% CIs, both before and after adjustment for confounders, were extracted. Both quantitative and qualitative data on exposure to tobacco smoke and HAP were recorded.

    Statistical analysis

    All studies were pooled and sensitivity analysis conducted to assess the impact of methodological approaches (due to sample selection, design of study, adjusted for possible confounding factors, sample size and publication year) by grouping them into different subgroups.10 We used natural logarithms of ORs and the associated SEs to estimate the pooled effect size of all studies and subgroups and assessed the within-group heterogeneity using Q-tests and/or I2 statistics (I2>50% was used as a threshold indicating significant heterogeneity).11 Meta-regression was used to explore the sources of heterogeneity. We report both the random and fixed pooled EEs (using forest plots), but only random effects estimates are reported in the text (results section). Funnel plots and Egger's regression were used to assess publication bias.12 All analyses were performed in STATA (V.12, STATA, College Station, Texas, USA).

    Results

    A total of 2853 articles were identified, of which 23 were duplicates. A further 2803 papers were considered irrelevant and were removed after screening the titles and abstracts. An additional six papers were removed after detailed assessment of each paper against the inclusion criteria. Data were extracted from the remaining 21 papers (17 on adults and 4 on children). All the four studies on children were from India, of which two reported a positive association between TB infection and exposure to solid fuel smoke. Of the 17 peer-reviewed studies in adults, 12 studies considering active TB reported adjusted (at least adjusted for smoking or the study conducted among non-smokers only) risk estimates (see online supplement table S1). Out of the 12 with adjusted risk estimates, 10 studies were case–control studies with cases defined by microbiological criteria, whereas the remaining two studies were cross-sectional with doctor diagnosed active TB with no reported microbiological proof of infection (figure 1). Half of the case–control studies selected control groups from hospitals and the remaining from the community. Information on household exposures to solid fuels used was obtained by questionnaire, but no quantitative exposure data were available.

    Figure 1
    Figure 1

    Flow chart showing studies related to tuberculosis and exposure to smoke from household air pollution.

    Effect estimates

    The adjusted pooled EE (OR, 95% CI) for all types of solid fuel (1.43, 1.07 to 1.91) was greater than for those using kerosene only (0.70, 0.13 to 3.87) and mixed fuel (kerosene and biomass) (1.30, 0.20 to 8.63; figure 2). The pooled EE for all solid fuel types with cases defined by microbiological criteria fell to 1.26 (0.95 to 1.68) when the two studies which had higher pooled EE (2.38, 1.88 to 3.00) with physician diagnosed active TB were removed from the analysis (figure 3). The unadjusted pooled effect size for each fuel type was greater than the corresponding adjusted estimates that controlled for age, gender and smoking (table 1).

    View this table:
    Table 1

    Subgroup analyses of Tuberculosis risk associated with the use of different fuel types

    Figure 2
    Figure 2

    Forest plot of studies reporting tuberculosis associated with exposure to burning of different types of fuel for domestic purposes.

    Figure 3
    Figure 3

    Forest plot of studies reporting tuberculosis associated with exposure to solid fuels only and stratified by diagnosis criteria and study design.

    Subgroup analysis was carried out for types of fuel used, gender, TB diagnosis criteria, study design, sample size, country of data collection and year of publication. The pooled effects estimate for women was greater (1.61, 0.73 to 3.57) than for men and women combined (1.39, 1.01 to 1.92) but not statistically significantly. There was greater heterogeneity among studies that reported exposure to biomass, kerosene and mixed fuel (kerosene/biomass; table 1).

    There were moderate heterogeneities among studies which used microbiological diagnostic criteria for diagnosis of TB, women only studies, case–control study designs, controls selected from hospitals, studies carried out post-2005 and those with sample size less than the median for all studies in the meta-analysis. Higher levels of heterogeneity were observed in studies which combined results for men and women, studies conducted in India, studies conducted pre-2005 and those with a study sample size greater than the median.

    Pooled EEs reported by cross-sectional studies and those with self-reported but doctor diagnosis of TB were significantly higher (p<0.001) compared with case–control studies using clinical and microbiological diagnosis criteria. Studies which used a national demographic health survey had the highest pooled effect size followed by case–control studies with controls selected from a population/community with lowest effect sizes where controls were taken from a hospital/clinic (table 1).

    No significant difference was found in the pooled EEs by gender when considering studies with less than or equal to median sample size compared to those of greater than median sample size and studies conducted before or after 2005. The majority (50%) of the studies were from India with significant heterogeneity (I2=68.4%, p=0.007) among the eight studies which reported an effect size adjusted at least for smoking.

    Publication bias

    For all studies with adjusted EEs, the funnel plot and the Egger test (bias=−1.94, p=0.136) suggested no potential publication bias (figure 4) unlike the case for the unadjusted EE (bias=−2.33, p=0.048) which showed significant publication bias. Studies with negative or insignificant findings were published twice as quickly after the end of data collection compared to those with positive findings (2.5 vs 5.7 years).

    Figure 4
    Figure 4

    Funnel plot of studies reporting tuberculosis associated with exposure to different types of cooking fuel.

    Heterogeneity by meta-regression

    The contribution of the following factors for heterogeneity was assessed using meta-regression, fuel types, gender, study design, TB diagnostic criteria used, location from where cases and control were selected, country in which the study was conducted, study period, sample size and publication year. Galbraith plots suggest that studies with unadjusted EEs were more heterogeneous. In studies with an adjusted EE, there was significant heterogeneity in the diagnostic criteria for TB (coefficient=0.359, p=0.019). When the meta-regression was rerun for those studies with use of solid fuel including biomass only, significant heterogeneity was noticed in the study design (coefficient=0.476, p=0.025), diagnostic criteria (coefficient=0.303, p=0.024), types of control (coefficient=0.380, p=0.004) and country in which the study was conducted (coefficient=−0.203, p=0.001).

    Similarly, studies that reported unadjusted EEs showed significant heterogeneity for the study period (coefficient=0.490, p=0.005), country in which the study was conducted (coefficient=−0.114, p=0.007) and publication year (coefficient=−0.048, p=0.001).

    Discussion

    The results of this meta-analysis suggest that individuals exposed to smoke from biomass burning during cooking have a 43% increased risk of having active TB compared to those using relatively clean fuel, but the risk reduced to 26% for case–control studies with cases diagnosed by microbiological criteria. However, no significant increased risk of TB was found for those using kerosene and biomass fuels for cooking. This lack of association with fuels other than biomass may be due to the small number of studies included in the pooled EE when examining the association with TB. Six of eight studies which reported an increased risk estimate of TB from exposure to fuel smoke and which adjusted for at least smoking showed a statistically significant effect.13–20

    All studies which investigated the association between TB and fuels were based on self-reported household exposure to smoke from solid fuels such as coal, wood, dung or charcoal. Only 5 of 18 selected studies were based on populations exposed to a combination of solid and liquefied fuels such as kerosene and LPG.15 ,18 ,19 ,21 ,22 Of these, only two studies attempted to categorise exposure to different fuel types.15 ,21 It is important to recognise that the use of different fuels and their combinations is likely to vary in different parts of the world, which will impact on risk estimates of TB. The relative importance of using different cooking fuel types and risk of TB is an important finding which should be considered when developing interventions for reducing risk of TB from exposure to HAP.

    There are a number of limitations when considering the studies selected for meta-analysis. First, most of the studies included in this meta-analysis are small-scale observational studies with the exception of three large household health surveys.16 ,17 ,22 These smaller scale observational studies are more prone to recruitment bias, and the majority only include information on a limited number of variables while others did not adjust for potential confounding factors, resulting in less reliable risk estimates. Most studies did not collect information on occupational risk factors for TB, which may be important in some countries where workplace risks are high and not adequately controlled. None of the 21 selected studies provided quantitative household airborne fuel smoke levels.

    The present study extends the previous meta-analyses,4 ,6 one of the important features being the inclusion of separate pooled EEs for studies with and without adjustment for smoking. This is particularly important as smoking is recognised as an important risk factor for TB. Furthermore, we also investigated other possible confounders such as socioeconomic status, household crowding and age. The pooled EEs for studies with adjustments for possible confounders including smoking were lower than those which did not adjust for these confounders.

    All studies included smokers except one study21 of female non-smokers. Of the 17 studies, four16 ,23–25 did not adjust for smoking and only two studies19 ,20 adjusted for passive smoking. Information on smoking in most studies was limited to individuals reporting ‘ever smoked’. Few studies18 ,19 ,26 attempted to quantify cumulative cigarette smoking as pack years.

    Heterogeneity among different studies was mostly due to the diagnostic criteria used for TB, selection of sampling population and the country in which the study was performed. The diagnostic criteria used in studies varied from self-reported symptoms to cases defined bacteriologically with or without chest X-rays. Studies which defined cases based on self-reported symptoms16 ,17 ,25 ,27 showed a significantly higher pooled effects estimate, which was also the case for hospital-based studies compared to community-based studies. Although a large number of studies are available from India, the diagnosis of latent TB is not a common practice in this country; hence, the actual risk in studies from India may have been underestimated. However, individuals from rural areas and small cities are likely to be diagnosed at local hospitals, usually based on radiography only, which might overestimate the risk.

    The larger scale cross-sectional household health surveys16 ,17 ,22 showed a higher effect size compared to case–control studies, although in these studies health outcome was assessed based on self-reported information where confounding due to recall bias is likely to be high. More than 60% of the studies were case–control in design, of which approximately 50% were hospital based with controls selected from hospital patients and not matched for age and neighbourhood, which may have introduced some bias in the effect size estimate.

    The majority of the studies (59%) used in calculating the adjusted pooled estimates were conducted in India with relatively few studies available from other high-TB-prevalent parts of the world where different household fuel types are used.6

    Many studies reported combined EEs for men and women but when we stratified the result based on gender, the differences in the pooled effect sizes for women and combined for men and women were statistically not significant (1.39 vs 1.61, p=0.732), with the pooled effect size for women being higher compared to that for combined men and women. The finding is inconsistent with the fact that household cooking in the majority of the LMIC is largely conducted by women if HAP is a risk factor for TB. However, a higher proportion of men compared with women are smokers in most middle-income and low-income countries and as the majority of studies did not adjust for secondhand smoking, adjusting for active smoking only may not have completely removed the smoking effect. This may have impacted on the lack of gender difference for HAP and the risk of TB in this analysis. Equally, this could reflect a relatively low threshold for exposure to solid fuel smoke above which the risk of TB increases. This suggests that other confounding factors, such as burning of incense sticks and mosquito coils, a common practice in countries with a higher prevalence of TB or misclassification of exposures, might explain the lack of gender difference.

    Assessment of indoor exposure to smoke from household fuels in all studies was based on interviewer delivered questionnaires, which are likely to be subject to recall bias and also possible misclassification if more than one fuel is used for different purposes, for example, heating and cooking. Different fuel types may also be used over time based on local availability and household affordability. Assessment of exposure, therefore, in future studies needs to be based on more reliable estimates of patterns of fuel usage and their variability. Quantitative exposure estimates would also allow derivation of dose–response relationships.

    In recent years, a number of studies have evaluated various interventions for reducing smoke from fuel such as improved stoves with improved burning efficiencies.28 These interventions when implemented will help to reduce airborne exposures, which in turn might impact on the contribution of HAP to the burden of TB. A modelling study in China3 has predicted reductions in TB for different household fuel use and DOTS (directly observed treatment, short course) scenarios. The model estimates a 14–52% decline in TB incidence by 2033 based on complete cessation of smoking and solid fuel use if 80% DOTS coverage is sustained.

    In conclusion, this systematic review and meta-analysis shows that the risk of active TB is dependent on fuel type with the highest risk being associated with biomass burning, which confers a 43% increased risk. The review also demonstrates the heterogeneity in risk estimates between reported studies, which needs to be understood in conjunction with overall risk estimates and also to inform future research. Future studies of HAP and TB risk would benefit from reliable quantitative measured exposure data as well as investigation of specific fuels such as kerosene and the use of defined combined fuels. There are few good quality studies in LMIC which include sufficient information on exposure assessment, smoking (age when started smoking, quantity of cigarettes consumed per day as well as passive smoking) and relevant domestic and occupational risk factors for TB. Future studies should also include information on nutritional status and also HIV infection as many studies, particularly from the African subcontinent and India, suggest a close association between HIV infection and the spread of TB.

    What is already known on this subject?

    • Approximately half of the world’s population, particularly those from low-income and middle-income countries, is exposed to household air pollution through daily burning of domestic fuels. There has been an increase in the number of publications in the last few years related to the association between tuberculosis and exposure to household air pollution, but the findings between different studies have not been consistent.

    What this study adds?

    • The majority of published studies have been small and some have reported their findings without taking account of the confounding factors during their analysis. This study reports the pooled effect estimate of similar published studies and looks into the cause of heterogeneities and publication bias.

    References

    1. ↵
      1. Lim SS,
      2. Vos T,
      3. Flaxman AD,
      4. et al
      . A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990?2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012;380:2224–60.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Kolappan C,
      2. Gopi PG
      . Tobacco smoking and pulmonary tuberculosis. Thorax 2002;57:964–6.
      OpenUrlAbstract/FREE Full Text
    3. ↵
      1. Lin H-H,
      2. Murray M,
      3. Cohen T,
      4. et al
      . Effects of smoking and solid-fuel use on COPD, lung cancer, and tuberculosis in China: a time-based, multiple risk factor, modelling study. Lancet 2008;372:1473–83.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Sumpter C,
      2. Chandramohan D
      . Systematic review and meta-analysis of the associations between indoor air pollution and tuberculosis. Trop Med Int Health 2013;18:101–8.
      OpenUrlCrossRefPubMed
    5. ↵
      1. Slama K,
      2. Chiang CY,
      3. Hinderaker SG,
      4. et al
      . Indoor solid fuel combustion and tuberculosis: is there an association? Int J Tuberc Lung Dis 2010;14:6–14.
      OpenUrlPubMedWeb of Science
    6. ↵
      1. Lin HH,
      2. Ezzati M,
      3. Murray M
      . Tobacco smoke, indoor air pollution and tuberculosis: a systematic review and meta-analysis. PLoS Med 2007;4:e20.
      OpenUrlCrossRefPubMed
    7. ↵
      1. Jubulis J,
      2. Kinikar A,
      3. Ithape M,
      4. et al
      . Modifiable risk factors associated with tuberculosis disease in children in Pune, India. Int J Tuberc Lung Dis 2014;18:198–204.
      OpenUrlCrossRefPubMed
    8. ↵
      1. Kurmi OP,
      2. Dunster C,
      3. Ayres JG,
      4. et al
      . Oxidative potential of smoke from burning wood and mixed biomass fuels. Free Radic Res 2013;47:829–35.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Liberati A,
      2. Altman DG,
      3. Tetzlaff J,
      4. et al
      . The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. BMJ 2009;339:b2700.
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Kurmi OP,
      2. Semple S,
      3. Simkhada P,
      4. et al
      . COPD and chronic bronchitis risk of indoor air pollution from solid fuel: a systematic review and meta-analysis. Thorax 2010;65:221–8.
      OpenUrlAbstract/FREE Full Text
    11. ↵
      1. Higgins JP,
      2. Thompson SG,
      3. Deeks JJ,
      4. et al
      . Measuring inconsistency in meta-analyses. BMJ 2003;327:557–60.
      OpenUrlFREE Full Text
    12. ↵
      1. Borenstein M,
      2. Hedges LV,
      3. Higgins JPT,
      4. et al
      . Introduction to meta-analysis. Oxford: Wiley, 2009. xxviii, 421p.
    13. ↵
      1. Garcia-Sancho MC,
      2. Garcia-Garcia L,
      3. Baez-Saldana R,
      4. et al
      . Indoor pollution as an occupational risk factor for tuberculosis among women: a population-based, gender oriented, case-control study in Southern Mexico. Rev Invest Clin 2009;61:392–8.
      OpenUrlPubMed
    14. ↵
      1. Kolappan C,
      2. Subramani R
      . Association between biomass fuel and pulmonary tuberculosis: a nested case-control study. Thorax 2009;64:705–8.
      OpenUrlAbstract/FREE Full Text
    15. ↵
      1. Lakshmi PV,
      2. Virdi NK,
      3. Thakur JS,
      4. et al
      . Biomass fuel and risk of tuberculosis: a case-control study from Northern India. J Epidemiol Community Health 2012;66:457–61.
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Mishra VK,
      2. Retherford RD,
      3. Smith KR
      . Cooking with biomass fuels increases the risk of tuberculosis. Natl Fam Health Surv Bull 1999;13:1–4.
      OpenUrlPubMed
    17. ↵
      1. Oxlade O,
      2. Murray M
      . Tuberculosis and poverty: why are the poor at greater risk in India? PLoS ONE 2012;7:e47533.
      OpenUrlCrossRefPubMed
    18. ↵
      1. Perez-Padilla R,
      2. Perez-Guzman C,
      3. Baez-Saldana R,
      4. et al
      . Cooking with biomass stoves and tuberculosis: a case control study. Int J Tuberc Lung Dis 2001;5:441–7.
      OpenUrlPubMedWeb of Science
    19. ↵
      1. Pokhrel AK,
      2. Bates MN,
      3. Verma SC,
      4. et al
      . Tuberculosis and indoor biomass and kerosene use in Nepal: a case-control study. Environ Health Perspect 2010;118:558–64.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Ramachandran R,
      2. Indu PS,
      3. Anish TS,
      4. et al
      . Determinants of childhood tuberculosis—a case control study among children registered under revised National Tuberculosis Control Programme in a district of South India. Indian J Tuberc 2011;58:204–7.
    21. ↵
      1. Behera D,
      2. Aggarwal G
      . Domestic cooking fuel exposure and tuberculosis in Indian women. Indian J Chest Dis Allied Sci 2010;52:139–43.
      OpenUrlPubMed
    22. ↵
      1. Singh A,
      2. Jamal S
      . Unhealthy cooking and prevalence of tuberculosis in Indian women: a case study. J Environ Prot 2012;3:648–56.
      OpenUrlCrossRef
    23. ↵
      1. Crampin AC,
      2. Glynn JR,
      3. Floyd S,
      4. et al
      . Tuberculosis and gender: exploring the patterns in a case control study in Malawi. Int J Tuberc Lung Dis 2004;8:194–203.
      OpenUrlPubMedWeb of Science
    24. ↵
      1. Gupta B,
      2. Mathur N,
      3. Mahendra P,
      4. et al
      . A study of household environmental risk factors pertaining to respiratory diseases. Energy Environ Monit 1997;13:61–6.
      OpenUrl
    25. ↵
      1. Saha A,
      2. Sharma YK,
      3. Kulkarni PK,
      4. et al
      . Risk of tuberculosis and fuel use: a population study. Occup Environ Med 2011;68:934.
      OpenUrlFREE Full Text
    26. ↵
      1. Perez-Padilla R,
      2. Regalado J,
      3. Vedal S,
      4. et al
      . Exposure to biomass smoke and chronic airway disease in Mexican women. A case-control study. Am J Respir Crit Care Med 1996;154:701–6.
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Patra S,
      2. Sharma S,
      3. Behera D
      . Passive smoking, indoor air pollution and childhood tuberculosis: a case control study. Indian J Tuberc 2012;59:151–5.
      OpenUrlPubMed
    28. ↵
      1. Smith KR,
      2. McCracken JP,
      3. Weber MW,
      4. et al
      . Effect of reduction in household air pollution on childhood pneumonia in Guatemala (RESPIRE): a randomised controlled trial. Lancet 2011;378:1717–26.
      OpenUrlCrossRefPubMedWeb of Science

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:

    Footnotes

    • Contributors CSS carried out all the searches. CSS and SSS went through the selection process of papers and extraction of data from the papers. OPK carried out all the analyses and wrote the first draft of the paper. All authors planned the study, interpreted the results and revised the paper.

    • Competing interests None.

    • Provenance and peer review Not commissioned; externally peer reviewed.