INT J TUBERC LUNG DIS 19(5):596–602 Q 2015 The Union http://dx.doi.org/10.5588/ijtld.14.0686
Childhood tuberculosis and exposure to indoor air pollution: a systematic review and meta-analysis N. Jafta,* P. M. Jeena,† L. Barregard,‡ R. N. Naidoo* *Discipline of Occupational and Environmental Health, School of Nursing and Public Health, and †Discipline of Paediatrics and Child Health, School of Clinical Medicine, College of Health Sciences, University of KwaZulu-Natal, Durban, South Africa; ‡Department of Occupational and Environmental Medicine, Sahlgrenska University Hospital and University of Gothenburg, Gothenburg, Sweden SUMMARY B A C K G R O U N D : Indoor air pollution (IAP) from environmental tobacco smoke (ETS) and biomass fuel smoke (BMS) poses respiratory health risks, with children and women bearing the major burden. O B J E C T I V E S : We used a systematic review and metaanalysis to investigate the relation between childhood tuberculosis (TB) and exposure to ETS and BMS. M E T H O D S : We searched three databases for epidemiological studies that investigated the association of childhood TB with exposure to ETS and BMS. We calculated pooled estimates and heterogeneity for studies eligible for inclusion in the meta-analysis and stratified studies on ETS by outcome. R E S U LT S : Five case-control and three cross-sectional studies were eligible for inclusion in the meta-analysis
and quality assessment. Pooled effect estimates showed that exposure to ETS is associated with tuberculous infection and TB disease (OR 1.9, 95%CI 1.4–2.9) among exposed compared to non-exposed children. TB disease in ETS studies produced a pooled OR of 2.8 (95%CI 0.9–4.8), which was higher than the OR for tuberculous infection (OR 1.9, 95%CI 0.9–2.9) for children exposed to ETS compared to non-exposed children. Studies on BMS exposure were too few and too small to permit a conclusion. C O N C L U S I O N : Exposure to ETS increases the risk of childhood TB disease or tuberculous infection. K E Y W O R D S : pulmonary tuberculosis; cooking fuel; passive smoke; risk factors; indoor air pollution
INDOOR AIR POLLUTION (IAP) has been identified as a public health problem in developing countries, where the majority of people are still dependent on biomass and fossil fuels for cooking, heating and lighting; more than 10% of the global respiratory disease burden is attributable to this exposure.1,2 Exposure to biomass fuel smoke (BMS) is not the only type of indoor exposure of concern. Exposure to environmental tobacco smoke (ETS) is recognised as a major risk factor for adverse respiratory health outcomes.3–13 The most vulnerable groups in terms of exposure to ETS and BMS are children, the elderly and the chronically ill, all of whom spend a considerable amount of their time indoors and are likely to be immunocompromised.1,14 In a recent systematic review and meta-analysis, Po et al. reported that BMS exposure increased the risk of respiratory diseases due to bacterial and viral infections in both children and women.15 Indoor tobacco smoke also posed an increased risk for tuberculosis (TB) disease.5,7 Most studies investigat-
ing the relationship between exposure to IAP and TB have either studied women alone or have combined women, men and children as the population of interest. These studies have shown an increased risk of TB in children compared to adults. Children’s studies of the natural history of TB disease (without chemotherapy) have shown that the risk of developing extra-pulmonary TB (tuberculous meningitis and miliary TB) is much higher in young children aged ,2 years than in older children or adults.16 A systematic review and meta-analysis on IAP and TB by Lin et al. assessed exposure to ETS and risk of TB in five studies (two adult and three child participant trials), and showed that ETS is a risk factor for childhood tuberculous infection and TB disease.5 Since the review by Lin et al.,5 several studies on IAP and childhood TB have been published.4,8,10,13 Children aged 615 years are less likely to be smokers, and therefore provide the best opportunity to understand IAP-related respiratory effects. We therefore conducted an updated systematic review to establish the relation between exposure to ETS and
Correspondence to: Nkosana Jafta, Discipline of Occupational and Environmental Health, School of Nursing and Public Health, University of KwaZulu-Natal, 321 George Campbell Building, Howard College Campus, Durban, 4041, South Africa. Tel: (þ27) 31 269 4528. Fax: (þ27) 31 260 4663. e-mail: [email protected] Article submitted 9 September 2014. Final version accepted 6 January 2015.
Childhood TB and indoor air pollution
Table Search strategy and terms used to identify studies on biomass fuel smoke (BMS), fossil fuel smoke, and environmental tobacco smoke (ETS) exposures Medical 1 2 3 4 5 6 7
subject headings (MeSH) term search: ‘tuberculosis’ ‘tobacco smoke pollution’ ‘biomass’ ‘fossil fuels’ ‘fuel oils’ ‘air pollution, indoor’ ‘(1) AND (2)’ OR ‘(1) AND (3)’ OR ‘(1) AND (4)’ OR ‘(1) AND (5)’ OR ‘(1) AND (6)’
Direct keyword search: 8 ‘childhood tuberculosis’ 9 ‘passive smoke’ 10 ‘environmental tobacco smoke’ 11 ‘secondhand smoke’ 12 ‘cooking fuels’ 13 ‘predictors’ 14 ‘risk factors’ 15 ‘(8) AND (9)’ OR ‘(8) AND (10)’ OR ‘(8) AND (11)’OR ‘(8) AND (12)’ OR ‘(8) AND (13)’ OR ‘(8) AND (14)’ 16 ‘(7) OR (15)’
BMS and tuberculous infection and TB disease among children using both quantitative and qualitative methods.
METHODS Data source and search strategy We conducted a systemic search of the PubMed (including Medline), Web of Science and CAB abstracts online databases, with the aim of identifying original research articles published between 1953 and April 2014 covering the association between exposure to IAP and/or ETS and tuberculous infection and/or TB disease. We also conducted a manual search based on the reference lists of the review articles and reports yielded by the electronic databases. Our search strategy is described in the Table. All results from the systematic search were entered into EndNote X7 (Thomson Reuters Scientific Inc, Carlsbad, CA, USA) and duplicates removed. NJ and RN reviewed titles and abstracts with the aim of removing publications that were unlikely to meet the inclusion criteria. They then examined the full texts of the remaining articles, and excluded any that did not meet the inclusion criteria. Those publications considered eligible were included in a systematic review and meta-analysis. Figure 1 is a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow chart of the selection process for this study.17 Selection criteria We reviewed all potentially eligible studies written in or translated into English and published in peerreviewed journals to ensure that they met the inclusion and exclusion criteria. The inclusion criteria were as follows: 1) the study population consisted of children aged 615 years with either tuberculous
597
infection or TB disease and control groups at risk of acquiring tuberculous infection or TB disease; 2) tuberculous infection and/or TB disease diagnoses were confirmed by laboratory diagnosis (acid-fast bacilli [AFB], culture and/or molecular tests), clinical diagnosis, radiograph, tuberculin skin test (TST) or a combination of these; 3) the study was an epidemiological study of primary or secondary data analysis design; and 4) the study reported adjusted effect estimates with 95% confidence intervals (95%CIs) for the association between TB and IAP (i.e., fuel smoke and/or ETS exposure). All studies of children aged .15 years in whom the above criteria could not be met were excluded. Data extraction and analysis Using a data extraction form, we extracted the following data from publications that were eligible for inclusion in the final meta-analysis: 1) study authors, 2) year of publication, 3) setting (country), 4) associations studied, 5) participants, 6) study type, 7) sample size, 8) exposures studied, 9) outcome(s) measured, 10) covariates adjusted for, 11) effect size estimates and 12) 95%CIs of effect estimates (Appendix Table A.1).* The effect estimates reported by Jubulis et al. were for different definitions of TB disease, namely probable TB only, confirmed TB only, and both probable TB and confirmed TB, according to the broader definition of TB.13 For this meta-analysis, we extracted data using the definition for confirmed TB. Although Jubulis et al. define IAP as a combination of ETS and BMS and report combined adjusted effect estimates,13 we used only the disaggregated reported estimates on ETS and BMS in the analysis. After extracting the relevant data on all studies judged eligible for the meta-analysis, we assessed the association between exposure and outcome using a forest plot of the adjusted odds ratio (OR) for TB (tuberculous infection and TB disease) for each study and exposure effect estimate. We pooled all studies on ETS exposure (Figure 2), and then stratified them by health outcome (latent tuberculous infection [LTBI] and TB disease) (Figure 3) and obtained pooled overall random effect estimates (ORs and 95%CIs) for the different strata21,22 using the DerSimonian-Laird method.21,22 We did not pool estimates of association between BMS and TB, as there were only two studies; however, we reported their individual estimates and analysed them for quality. We assessed heterogeneity between the studies on ETS by calculating the I2 statistic for each of the strata and for all the studies on ETS. The area of the square of each study plot reflects the Mantel-Haenszel weight that the study contributed to the final pooled effect estimate. We also performed a qualitative analysis of the risk of bias on all eight studies included using a * The appendix is available in the online version of this article, at http://www.ingentaconnect.com/content/iuatld/ijtld/2015/ 00000019/00000005/art00018
598
The International Journal of Tuberculosis and Lung Disease
ETS and/or BMS exposure and TB, or were nonhuman studies, while a further 633 were not related to IAP and TB, leaving 125 publications eligible and relevant for detailed review. Following this full-text review, we excluded 114 of these, as children were not the population of interest, the analysis did not include a stratum for children (aged 615 years) or the exposure variable was personal smoking rather than ETS. We excluded a further three studies because they did not adjust for covariates and reported only crude ORs and 95%CIs. The remaining eight publications, all of which were either case-control or cross-sectional studies, were eligible for further detailed review and meta-analysis. Appendix Table A.1 lists the studies that were included in the final analysis, along with the main variables that were considered in the assessment and extraction of data. Exposure measures All eight studies reported an association between ETS exposure and TB. Five reported ETS exposure only;4,8,18–20 the remaining three reported an association between both ETS and BMS exposure with TB in children. Of these three studies reporting BMS exposure, two reported adjusted risk estimates separately for both ETS and BMS,9,13 and the other reported adjusted risk estimates for ETS and only crude estimates for BMS.10 Figure 1 Flow chart of the literature search for studies investigating IAP and TB in children. IAP ¼ indoor air pollution; TB ¼ tuberculosis; ETS ¼ environmental tobacco smoke; HAP ¼ hazardous air pollutant; BMS ¼ biomass fuel smoke.
RESULTS
Outcome measures Five studies9,10,13,19,20 looked at TB disease as an outcome based on clinical diagnosis, with two of these studies13,19 also adding laboratory identification (smear and/or culture) of Mycobacterium tuberculosis in the diagnostic method. Three studies included children with both pulmonary and extrapulmonary disease,9,10,20 whereas the other two included children with only pulmonary TB.13,19 All three studies that had infection as an outcome variable used the TST diameter as indicator; Du Preez et al. used three cut-off points, with 10 mm as the primary measure and 5 mm and 15 mm as secondary measures or cut-offs for infection,8 whereas Singh et al.18 and Den Boon et al.4 used only the 10 mm cutoff. For this review, the 10 mm TST cut-off risk estimate was used in the analysis. The study by Altet et al., which used TB disease as outcome variable, also used TST as part of the definition of disease, with a 5 mm diameter as the cut-off for diagnosing active pulmonary TB in children.19 None of the studies indicated whether a positive TST was the result of bacille Calmette-Gu´erin (BCG) vaccination.
Search and selection of studies Figure 1 summarises the process used to identify the studies eligible for the systematic review and metaanalysis according to the PRISMA reporting process.17 Of the 1088 publications identified from electronic databases and manual search, 224 were excluded because they were duplicates, not related to
Association of exposure to environmental tobacco smoke and biomass fuel smoke with tuberculosis in children Six of the eight studies showed a statistically significantly increased risk of tuberculous infection or TB disease when children were exposed to ETS, with adjusted ORs ranging from 1.8 (95%CI 1.2–2.7)
combination of elements of the Critical Appraisal Skills Programme (CASP, Oxford, UK, 2014; http:// www.caspinternational.org/?o¼1012) for case control studies and a method used by Ijaz et al.23 The six elements included were 1) a definition of outcome and its assessment or measure, 2) the definition for exposure used, 3) the source and method for measuring the exposure, 4) reliability, 5) controlling for confounding covariates in the analysis, and 6) the analysis performed in the study (Appendix Table A.2). We tested for bias on studies on ETS and study design strata using a method described by Egger et al.24 to identify asymmetry in a funnel plot. All statistical analyses were performed using Stata IC 13 (StataCorp, College Station, TX, USA).
Childhood TB and indoor air pollution
599
Figure 2 Forest plot showing studies on ETS exposure and childhood TB, with pooled ORs. ETS ¼ environmental tobacco smoke; OR ¼ odds ratio; CI ¼ confidence interval; TB ¼ tuberculosis.
to 9.3 (95%CI 3.1–27.6), compared to non-exposed children (Figure 2). Two studies found no statistically significant association between ETS exposure and TB disease and tuberculous infection (OR 1.4, 95%CI 0.9–2.1, and OR 2.9, 95%CI 0.7–12.4, respectively).4,13 The studies by Ramachandran et al.9 and Jubulis et al.,13 which reported an association of BMS exposure and TB, had significant estimates (ORs) of 6.9 (95%CI 2.5–18.9) and 7.2 (95%CI 1.4–44.5), respectively.9,13 We estimated random effects by pooling the effect estimates of the eight studies 1) for all the studies on ETS (Figure 2), and 2) for each stratum of outcome for the ETS exposure studies (tuberculous infection or TB disease; Figure 3). Two studies, by Den Boon et al.4 and Patra et al.,10 had risk estimates that were not statistically significant and that weighted .70% to the pooled estimate. The pooled forest plot of effect estimates (ORs) of the eight studies reporting an association between tuberculous infection and TB disease in children and exposure to ETS was 1.9 (95%CI 1.3–2.5) (Figure 2). Studies with TB disease as an outcome (defined as
clinical symptoms and radiological abnormality consistent with TB, plus a TST induration above the cut-off diameter) showed a slightly stronger association (OR 2.8, 95%CI 0.86–4.78) than those with tuberculous infection as an endpoint (OR 1.9, 95%CI 0.9–2.9) when exposed to ETS (Figure 3). The degree of heterogeneity between all ETS studies (I2 ¼ 20.6%, P ¼ 0.266) showed that although the studies made different contributions to the final, pooled estimate, the differences between the strata were not significant when random effect estimates were computed (Figures 2 and 3). Although the method employed by Egger et al. to assess publication bias is not sensitive when the number of studies is small,24 we nevertheless computed a funnel plot of logOR estimates of the eight studies on ETS against their standard errors (selogOR). The asymmetric results, with three of the studies falling outside the funnel plot, suggest that there may have been a reporting bias or chance variation in the studies selected for this review (Figure 4). The eight studies showed varying results when assessed for quality using CASP in combination with
Figure 3 Forest plot showing studies on ETS and childhood TB, stratified by outcome variable and the calculated pooled effect estimate. ETS ¼ environmental tobacco smoke; OR ¼ odds ratio; CI ¼ confidence interval; TB ¼ tuberculosis; LTBI ¼ latent tuberculous infection.
600
The International Journal of Tuberculosis and Lung Disease
Figure 4 Funnel plot with pseudo 95% CIs assessing publication bias of the eight studies on environmental tobacco smoke included in the meta-analysis. CI ¼ confidence interval; SE logOR ¼ standard errors logOR; logOR ¼ log odds ratio.
the qualitative method employed by Ijaz et al. (Appendix Table A.2).23 Five of the eight studies had low risk rating in at least four of the six elements of bias assessed, and only one of the eight studies had a high risk of bias rating for one of the elements assessed. Most of the elements were rated ‘unclear’ because insufficient detail was provided in the publications.
DISCUSSION The pooled effect estimates of this meta-analysis showed that children exposed to ETS had a two-fold increased risk of acquiring tuberculous infection and TB disease compared to non-exposed children. This was higher than the estimates reported in adult studies or in studies that included all age groups.5,11,19,25 Stratification of ETS exposure by outcome showed that the OR for TB disease was higher (OR 2.8) than that for tuberculous infection (OR 1.9) in children exposed to ETS than in nonexposed children (Figure 3). However, as all of the studies with tuberculous infection as an outcome had a cross-sectional design (Appendix Table A.1), the influence of study design on the effect estimate could not be disentangled from the effect of how outcome was defined (disease or infection). The literature provides evidence of biological mechanisms of exposure to air pollution and respiratory illnesses, with indoor air constituents such as nicotine, particulate matter (PM2.5), nitrogen dioxide and carbon monoxide implicated.26,27 The biological mechanisms involved in the relationship between IAP exposure and tuberculous infection and TB disease can be extrapolated from the work that has investigated the mechanism(s) involved in IAP constituents and other respiratory infections.26–30 A failure of the involved immune system components, such as cellmediated immunity to activate response that is optimal for M. tuberculosis clearance, can lead to the pathogen causing tuberculous infection or TB
disease. The four major immunological systems that may be implicated are 1) particle or bacterial clearance,28,29 2) recognition and immune signalling to the pathogen,31 3) intracellular antibacterial response,30 and 4) recruitment of adaptive immune components in elimination and containment of the M. tuberculosis pathogen.32 Two studies, by Den Boon et al.4 and Patra et al.,10 contributed .70% to the weighting of the pooled overall OR for the eight studies, and neither was statistically significant. Furthermore, when both studies were assessed for confounding or adjustment for covariates, they were rated respectively as high risk and unclear for bias, as they were either vague in their description or had excluded important covariates. The results from the pooling of studies on ETS which used tuberculous infection as outcome or as a measure contributing to the outcome definition of disease should be interpreted with caution, as the standard induration of the TST weal for positive infection is not the same across different populations. In populations with a high prevalence of BCG immunisation against TB and high human immunodeficiency virus (HIV) prevalence in children, the diagnosis of tuberculous infection using the TST diameter is variable. The choice of a 10-mm cut-off for three of the four studies4,8,19,20 that used TST could be due to the high BCG vaccination coverage or TB prevalence in the two populations studied, namely India and South Africa. On the other hand, in the study by Altet et al.,19 the use of a 5-mm cut-off could be related to the low BCG vaccination coverage and low TB prevalence in the Spanish population at the time of the study. In all the studies included in this review, exposure to IAP was assessed by asking the participating care givers about energy use and/or presence of smokers in their homes. This measure of pollutant exposure could have varied from study to study, depending on the details of the assessment. Most of these studies had exposure time as a factor in their assessment,
Childhood TB and indoor air pollution
especially for ETS. In some of the studies, children living in the same house as a smoker for at least 6 or 12 months were considered to be exposed to ETS,8,10,19 but the time the child spent each day in that environment was not taken into account. On the other hand, behaviours that could influence exposure of young children to BMS, such as the child being carried on the back of a woman during cooking, were usually not adequately assessed or described in most studies. Air pollution emissions from indoor cooking are affected not only by the number of hours, but also by the quality of the fuel and the technique used in burning fuel. Moreover, exposure to IAP (ETS or BMS) depends not only on the emissions, but also on the ventilation in the house (chimney, open doors and windows) and on the climate. Objective methods of exposure assessment are therefore necessary, such as measurements of air pollution in the homes. Exposures reported by care givers represent a relatively crude measure of IAP, and are usually not comparable. ETS and BMS exposures are a result of different indoor pollution sources, and a detailed assessment is needed to characterise them, especially for children, as these exposures could result in effects that are additive or synergistic. The age and sex of the children33 are also determinants for IAP exposure. In the studies included in our meta-analysis, this was addressed either by matching or by adjustment in the analysis. All of the studies included in this meta-analysis have been adjusted for one or more of the important covariates implicated in tuberculous infection or TB disease: household TB contact, socio-economic status (SES), malnutrition, age, HIV status, vaccination status, and crowding. Although HIV status of children is known to be the major driver of acquisition (tuberculous infection) and progression (TB disease),34 none of the studies quantified the effect of HIV when modelling the association between TB and IAP exposure. Six of the studies included in this meta-analysis were from high HIV burden countries, India and South Africa, where .20% of TB patients are HIV-positive.12 The covariates adjusted for across the studies were not consistent, and this could have caused some of the heterogeneity observed between them. Some covariates, such as BCG vaccination status and HIV status, are particularly important when studying TB in certain populations. This can be observed in the differences between the two studies from similar communities in South Africa published 4 years apart.4,8 In South Africa, antiretroviral treatment had been rolled out publicly only to a limited extent during the earlier study period (2003–2007),4 whereas it was widely available when the later study was conducted (from 2009 onwards).8 Neither study assessed the HIV status of the participants. The study by Den Boon et al. was conducted in populations with higher HIV prevalence,4 and may have underestimated the occurrence of tuberculous infection, as these authors used an induration of 10 mm as cut-off, and
601
there was no stratification or control for HIV in the analysis.4,16 Studies investigating the relation between IAP exposure and childhood TB have used proxy measures for estimating exposure, such as reported exposures obtained from care giver interviews and medical records. In this review, sources of bias for four of the eight studies were classified as ‘unclear’ or ‘high risk’ due to the definition of exposure, and for two of these studies also due to the measurement of exposure. A standardised methodology using a questionnaire or walkthrough instrument for assessing exposure to IAP needs to be developed for future studies. Studies that objectively quantify exposure are needed for estimating the dose-response relation of exposure to disease outcome. The limitations of this review include reporting bias that could result from a higher likelihood of studies with positive and significant findings being published than studies with negative or non-significant findings. In this systematic review, we excluded three studies because they did not report adjusted effect estimates and documented only positive significant results regarding associations between IAP and childhood TB. Furthermore, only two of the three studies separately reported the adjusted effect estimate for exposure to BMS. These studies on BMS exposure had very wide CIs, which could be the result of the small number of observations.
CONCLUSIONS Despite the limitations, pooled effect estimate indicates an increased risk of childhood TB in children exposed to ETS compared to those not exposed. The limited number of studies on the association of BMS exposure and childhood TB and the small sizes of the studies do not permit any conclusions to be drawn on causality. Although some of the reviewed studies may suffer from bias, programmes addressing IAP should be considered in the comprehensive management of the TB epidemic because of the known high incidence and severity of TB disease in children. Future studies should consider BCG and HIV status, molecular TB tests and/or isolation for better clinical definition of tuberculous infection or TB disease and more accurate evaluation of IAP exposure to confirm these findings. Acknowledgements NJ is a recipient of the National Institutes for Health-Fogarty International (Bethesda, MD, USA) and South African TB/AIDS Research Training (SATBAT, Johannesburg, South Africa) Grant: 5U2RTW00773 and 5U2RTW007373. Conflicts of interest: none declared.
References 1 Rehfuess E A, Bruce N G, Smith K R. Solid fuel use: health effect. In: Nriagu J O, ed. Encyclopedia of environmental health. Vol 5. Burlington, VT, USA: Elsevier, 2011; pp 150– 161.
602
The International Journal of Tuberculosis and Lung Disease
2 Torres-Duque C, Maldonado D, P´erez-Padilla R, Ezzati M, Viegi G. Biomass fuels and respiratory diseases. Proc Am Thorac Soc 2008; 5: 577–590. 3 Shetty N, Shemko M, Vaz M, D’Souza G. An epidemiological evaluation of risk factors for tuberculosis in South India: a matched case control study. Int J Tuberc Lung Dis 2006; 10: 80– 86. 4 Den Boon S, Verver S, Marais B J, et al. Association between passive smoking and infection with Mycobacterium tuberculosis in children. Pediatrics 2007; 119: 734–739. 5 Lin H-H, Ezzati M, Murray M. Tobacco smoke, indoor air pollution and tuberculosis: a systematic review and metaanalysis. PLOS MED 2007; 4: e20. 6 Lin H-H, Murray M, Cohen T, Colijn C, Ezzati M. 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: 1473–1483. 7 Leung C C, Lam T H, Ho K S, et al. Passive smoking and tuberculosis. Arch Intern Med 2010; 170: 287–292. 8 Du Preez K, Mandalakas A M, Kirchner H L, et al. Environmental tobacco smoke exposure increases Mycobacterium tuberculosis infection risk in children. Int J Tuberc Lung Dis 2011; 15: 1490–1496. 9 Ramachandran R, Indu P S, Anish T S, Nair S, Lawrence T, Rajasi R S. 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: 204–208. 10 Patra S, Sharma S, Behera D. Passive smoking, indoor air pollution and childhood tuberculosis: a case control study. Indian J Tuberc 2012; 59: 151–155. 11 Sumpter C, Chandramohan D. Systematic review and metaanalysis of the associations between indoor air pollution and tuberculosis. Trop Med Int Health 2013; 18: 101–108. 12 World Health Organization. Global tuberculosis report, 2013. WHO/HTM/TB/2013.11. Geneva, Switzerland: WHO, 2013. 13 Jubulis J, Kinikar A, Ithape M, et al. Modifiable risk factors associated with tuberculosis disease in children in Pune. India. Int J Tuberc Lung Dis 2014; 18: 198–204. 14 Baumgartner J, Schauer J J, Ezzati M, et al. Patterns and predictors of personal exposure to indoor air pollution from biomass combustion among women and children in rural China. Indoor Air 2011; 21: 479–488. 15 Po J Y T, FitzGerald J M, Carlsten C. Respiratory disease associated with solid biomass fuel exposure in rural women and children: systematic review and meta-analysis. Thorax 2011; 66: 232–239. 16 Marais B J, Gie R P, Schaaf H S, et al. The natural history of childhood intra-thoracic tuberculosis: a critical review of literature from the pre-chemotherapy era. Int J Tuberc Lung Dis 2004; 8: 392–402. 17 Moher D, Liberati A, Tetzlaff J, Altman D G; PRISMA Group. Reprint—preferred reporting meta-analyses: the PRISMA statement. Phys Ther 2009; 89: 873–880. 18 Singh M, Mynak M L, Kumar L, Mathew J L, Jindal S K. Prevalence and risk factors for transmission of infection among children in household contact with adults having pulmonary tuberculosis. Arch Dis Child 2005; 90: 624–628.
19 Altet M N, Alcaide J, Plans P, et al. Passive smoking and risk of pulmonary tuberculosis in children immediately following infection: a case–control study. Tubercle Lung Dis 1996; 77: 537–544. 20 Tipayamongkholgul M, Podhipak A, Chearskul S, Sunakorn P. Factors associated with the development of tuberculosis in BCG immunised children. Southeast Asian J Trop Med Public Health 2005; 36: 145–150. 21 DerSimonian R, Kacker R. Random-effects model for metaanalysis of clinical trials: an update. Contemp Clin Trials 2007; 28: 105–114. 22 DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986; 7: 177–188. 23 Ijaz S, Verbeek J, Seidler A, et al. Night-shift work and breast cancer – a systematic review and meta-analysis. Scand J Work Environ Health 2013; 39: 431–447. 24 Egger M, Smith G D, Schneider M, Minder C. Bias in metaanalysis detected by a simple, graphical test. BMJ 1997; 315: 629–634. 25 Lin H-H, Suk C-W, Lo H-L, Huang R-Y, Enarson D, Chiang CY. Indoor air pollution from solid fuel and tuberculosis: a systematic review and meta-analysis. Int J Tuberc Lung Dis 2014; 18: 613–621. 26 Gardner D E. Tobacco smoke: In: Cohen M D, Zelikoff J T, Schlesinger R B, eds. Pulmonary immunotoxicology. New York, NY, USA: Kluwer Academic Publishers, 2000: pp 387– 409. 27 Schlesinger R B, Chen L-C, Zelikoff J T. Sulfur and nitrogen oxides. In: Cohen M D, Zelikoff J T, Schlesinger R B, eds. Pulmonary immunotoxicology. New York, NY, USA: Kluwer Academic Publishers, 2000: pp 337–352. 28 Haswell L E, Hewitt K, Thorne D, Richter A, Ga¸ca M D. Cigarette smoke total particulate matter increases mucous secreting cell numbers in vitro: a potential model of goblet cell hyperplasia. Toxicol In Vitro 2010; 24: 981–987. 29 Lillehoj E P, Kato K, Lu W, Kim K C. Cellular and molecular biology of airway mucins. Int Rev Cell Mol Biol 2013; 303: 139–202. 30 Yang H-M, Antonini J M, Barger M W, et al. Diesel exhaust particles suppress macrophage function and slow the pulmonary clearance of Listeria monocytogenes in rats. Environ Health Perspect 2001; 109: 515–521. 31 Hagiwara E, Takahashi K, Okubo T, et al. Cigarette smoking depletes spontaneously secreting Th-1 cytokines in the human airway. Cytokine 2001; 14: 121–126. 32 Miranda M S, Breiman A, Allain S, Deknuydt F, Altare F. The tuberculous granuloma: an unsuccessful host defence mechanism providing a safety shelter for the bacteria? Clin Dev Immunol 2012; 2012: 139 127. 33 Kim S, Wipfli H, Navas-Acien A, et al. Determinants of hair nicotine concentrations in nonsmoking women and children: a multicountry study of secondhand smoke exposure in homes. Cancer Epidemiol Biomarkers Prev 2009; 18: 3407–3414. 34 Jeena P M, Pillay P, Pillay T, Coovadia H M. Impact of HIV-1 co-infection on presentation and hospital-related mortality in children with culture proven pulmonary tuberculosis in Durban, South Africa. Int J Tuberc Lung Dis 2002; 6: 672– 678.
Childhood tuberculosis and exposure to indoor air pollution: a systematic review and meta-analysis N. Jafta,* P. M. Jeena,† L. Barregard,‡ R. N. Naidoo* *Discipline of Occupational and Environmental Health, School of Nursing and Public Health, and †Discipline of Paediatrics and Child Health, School of Clinical Medicine, College of Health Sciences, University of KwaZulu-Natal, Durban, South Africa; ‡Department of Occupational and Environmental Medicine, Sahlgrenska University Hospital and University of Gothenburg, Gothenburg, Sweden SUMMARY B A C K G R O U N D : Indoor air pollution (IAP) from environmental tobacco smoke (ETS) and biomass fuel smoke (BMS) poses respiratory health risks, with children and women bearing the major burden. O B J E C T I V E S : We used a systematic review and metaanalysis to investigate the relation between childhood tuberculosis (TB) and exposure to ETS and BMS. M E T H O D S : We searched three databases for epidemiological studies that investigated the association of childhood TB with exposure to ETS and BMS. We calculated pooled estimates and heterogeneity for studies eligible for inclusion in the meta-analysis and stratified studies on ETS by outcome. R E S U LT S : Five case-control and three cross-sectional studies were eligible for inclusion in the meta-analysis
and quality assessment. Pooled effect estimates showed that exposure to ETS is associated with tuberculous infection and TB disease (OR 1.9, 95%CI 1.4–2.9) among exposed compared to non-exposed children. TB disease in ETS studies produced a pooled OR of 2.8 (95%CI 0.9–4.8), which was higher than the OR for tuberculous infection (OR 1.9, 95%CI 0.9–2.9) for children exposed to ETS compared to non-exposed children. Studies on BMS exposure were too few and too small to permit a conclusion. C O N C L U S I O N : Exposure to ETS increases the risk of childhood TB disease or tuberculous infection. K E Y W O R D S : pulmonary tuberculosis; cooking fuel; passive smoke; risk factors; indoor air pollution
INDOOR AIR POLLUTION (IAP) has been identified as a public health problem in developing countries, where the majority of people are still dependent on biomass and fossil fuels for cooking, heating and lighting; more than 10% of the global respiratory disease burden is attributable to this exposure.1,2 Exposure to biomass fuel smoke (BMS) is not the only type of indoor exposure of concern. Exposure to environmental tobacco smoke (ETS) is recognised as a major risk factor for adverse respiratory health outcomes.3–13 The most vulnerable groups in terms of exposure to ETS and BMS are children, the elderly and the chronically ill, all of whom spend a considerable amount of their time indoors and are likely to be immunocompromised.1,14 In a recent systematic review and meta-analysis, Po et al. reported that BMS exposure increased the risk of respiratory diseases due to bacterial and viral infections in both children and women.15 Indoor tobacco smoke also posed an increased risk for tuberculosis (TB) disease.5,7 Most studies investigat-
ing the relationship between exposure to IAP and TB have either studied women alone or have combined women, men and children as the population of interest. These studies have shown an increased risk of TB in children compared to adults. Children’s studies of the natural history of TB disease (without chemotherapy) have shown that the risk of developing extra-pulmonary TB (tuberculous meningitis and miliary TB) is much higher in young children aged ,2 years than in older children or adults.16 A systematic review and meta-analysis on IAP and TB by Lin et al. assessed exposure to ETS and risk of TB in five studies (two adult and three child participant trials), and showed that ETS is a risk factor for childhood tuberculous infection and TB disease.5 Since the review by Lin et al.,5 several studies on IAP and childhood TB have been published.4,8,10,13 Children aged 615 years are less likely to be smokers, and therefore provide the best opportunity to understand IAP-related respiratory effects. We therefore conducted an updated systematic review to establish the relation between exposure to ETS and
Correspondence to: Nkosana Jafta, Discipline of Occupational and Environmental Health, School of Nursing and Public Health, University of KwaZulu-Natal, 321 George Campbell Building, Howard College Campus, Durban, 4041, South Africa. Tel: (þ27) 31 269 4528. Fax: (þ27) 31 260 4663. e-mail: [email protected] Article submitted 9 September 2014. Final version accepted 6 January 2015.
Childhood TB and indoor air pollution
Table Search strategy and terms used to identify studies on biomass fuel smoke (BMS), fossil fuel smoke, and environmental tobacco smoke (ETS) exposures Medical 1 2 3 4 5 6 7
subject headings (MeSH) term search: ‘tuberculosis’ ‘tobacco smoke pollution’ ‘biomass’ ‘fossil fuels’ ‘fuel oils’ ‘air pollution, indoor’ ‘(1) AND (2)’ OR ‘(1) AND (3)’ OR ‘(1) AND (4)’ OR ‘(1) AND (5)’ OR ‘(1) AND (6)’
Direct keyword search: 8 ‘childhood tuberculosis’ 9 ‘passive smoke’ 10 ‘environmental tobacco smoke’ 11 ‘secondhand smoke’ 12 ‘cooking fuels’ 13 ‘predictors’ 14 ‘risk factors’ 15 ‘(8) AND (9)’ OR ‘(8) AND (10)’ OR ‘(8) AND (11)’OR ‘(8) AND (12)’ OR ‘(8) AND (13)’ OR ‘(8) AND (14)’ 16 ‘(7) OR (15)’
BMS and tuberculous infection and TB disease among children using both quantitative and qualitative methods.
METHODS Data source and search strategy We conducted a systemic search of the PubMed (including Medline), Web of Science and CAB abstracts online databases, with the aim of identifying original research articles published between 1953 and April 2014 covering the association between exposure to IAP and/or ETS and tuberculous infection and/or TB disease. We also conducted a manual search based on the reference lists of the review articles and reports yielded by the electronic databases. Our search strategy is described in the Table. All results from the systematic search were entered into EndNote X7 (Thomson Reuters Scientific Inc, Carlsbad, CA, USA) and duplicates removed. NJ and RN reviewed titles and abstracts with the aim of removing publications that were unlikely to meet the inclusion criteria. They then examined the full texts of the remaining articles, and excluded any that did not meet the inclusion criteria. Those publications considered eligible were included in a systematic review and meta-analysis. Figure 1 is a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow chart of the selection process for this study.17 Selection criteria We reviewed all potentially eligible studies written in or translated into English and published in peerreviewed journals to ensure that they met the inclusion and exclusion criteria. The inclusion criteria were as follows: 1) the study population consisted of children aged 615 years with either tuberculous
597
infection or TB disease and control groups at risk of acquiring tuberculous infection or TB disease; 2) tuberculous infection and/or TB disease diagnoses were confirmed by laboratory diagnosis (acid-fast bacilli [AFB], culture and/or molecular tests), clinical diagnosis, radiograph, tuberculin skin test (TST) or a combination of these; 3) the study was an epidemiological study of primary or secondary data analysis design; and 4) the study reported adjusted effect estimates with 95% confidence intervals (95%CIs) for the association between TB and IAP (i.e., fuel smoke and/or ETS exposure). All studies of children aged .15 years in whom the above criteria could not be met were excluded. Data extraction and analysis Using a data extraction form, we extracted the following data from publications that were eligible for inclusion in the final meta-analysis: 1) study authors, 2) year of publication, 3) setting (country), 4) associations studied, 5) participants, 6) study type, 7) sample size, 8) exposures studied, 9) outcome(s) measured, 10) covariates adjusted for, 11) effect size estimates and 12) 95%CIs of effect estimates (Appendix Table A.1).* The effect estimates reported by Jubulis et al. were for different definitions of TB disease, namely probable TB only, confirmed TB only, and both probable TB and confirmed TB, according to the broader definition of TB.13 For this meta-analysis, we extracted data using the definition for confirmed TB. Although Jubulis et al. define IAP as a combination of ETS and BMS and report combined adjusted effect estimates,13 we used only the disaggregated reported estimates on ETS and BMS in the analysis. After extracting the relevant data on all studies judged eligible for the meta-analysis, we assessed the association between exposure and outcome using a forest plot of the adjusted odds ratio (OR) for TB (tuberculous infection and TB disease) for each study and exposure effect estimate. We pooled all studies on ETS exposure (Figure 2), and then stratified them by health outcome (latent tuberculous infection [LTBI] and TB disease) (Figure 3) and obtained pooled overall random effect estimates (ORs and 95%CIs) for the different strata21,22 using the DerSimonian-Laird method.21,22 We did not pool estimates of association between BMS and TB, as there were only two studies; however, we reported their individual estimates and analysed them for quality. We assessed heterogeneity between the studies on ETS by calculating the I2 statistic for each of the strata and for all the studies on ETS. The area of the square of each study plot reflects the Mantel-Haenszel weight that the study contributed to the final pooled effect estimate. We also performed a qualitative analysis of the risk of bias on all eight studies included using a * The appendix is available in the online version of this article, at http://www.ingentaconnect.com/content/iuatld/ijtld/2015/ 00000019/00000005/art00018
598
The International Journal of Tuberculosis and Lung Disease
ETS and/or BMS exposure and TB, or were nonhuman studies, while a further 633 were not related to IAP and TB, leaving 125 publications eligible and relevant for detailed review. Following this full-text review, we excluded 114 of these, as children were not the population of interest, the analysis did not include a stratum for children (aged 615 years) or the exposure variable was personal smoking rather than ETS. We excluded a further three studies because they did not adjust for covariates and reported only crude ORs and 95%CIs. The remaining eight publications, all of which were either case-control or cross-sectional studies, were eligible for further detailed review and meta-analysis. Appendix Table A.1 lists the studies that were included in the final analysis, along with the main variables that were considered in the assessment and extraction of data. Exposure measures All eight studies reported an association between ETS exposure and TB. Five reported ETS exposure only;4,8,18–20 the remaining three reported an association between both ETS and BMS exposure with TB in children. Of these three studies reporting BMS exposure, two reported adjusted risk estimates separately for both ETS and BMS,9,13 and the other reported adjusted risk estimates for ETS and only crude estimates for BMS.10 Figure 1 Flow chart of the literature search for studies investigating IAP and TB in children. IAP ¼ indoor air pollution; TB ¼ tuberculosis; ETS ¼ environmental tobacco smoke; HAP ¼ hazardous air pollutant; BMS ¼ biomass fuel smoke.
RESULTS
Outcome measures Five studies9,10,13,19,20 looked at TB disease as an outcome based on clinical diagnosis, with two of these studies13,19 also adding laboratory identification (smear and/or culture) of Mycobacterium tuberculosis in the diagnostic method. Three studies included children with both pulmonary and extrapulmonary disease,9,10,20 whereas the other two included children with only pulmonary TB.13,19 All three studies that had infection as an outcome variable used the TST diameter as indicator; Du Preez et al. used three cut-off points, with 10 mm as the primary measure and 5 mm and 15 mm as secondary measures or cut-offs for infection,8 whereas Singh et al.18 and Den Boon et al.4 used only the 10 mm cutoff. For this review, the 10 mm TST cut-off risk estimate was used in the analysis. The study by Altet et al., which used TB disease as outcome variable, also used TST as part of the definition of disease, with a 5 mm diameter as the cut-off for diagnosing active pulmonary TB in children.19 None of the studies indicated whether a positive TST was the result of bacille Calmette-Gu´erin (BCG) vaccination.
Search and selection of studies Figure 1 summarises the process used to identify the studies eligible for the systematic review and metaanalysis according to the PRISMA reporting process.17 Of the 1088 publications identified from electronic databases and manual search, 224 were excluded because they were duplicates, not related to
Association of exposure to environmental tobacco smoke and biomass fuel smoke with tuberculosis in children Six of the eight studies showed a statistically significantly increased risk of tuberculous infection or TB disease when children were exposed to ETS, with adjusted ORs ranging from 1.8 (95%CI 1.2–2.7)
combination of elements of the Critical Appraisal Skills Programme (CASP, Oxford, UK, 2014; http:// www.caspinternational.org/?o¼1012) for case control studies and a method used by Ijaz et al.23 The six elements included were 1) a definition of outcome and its assessment or measure, 2) the definition for exposure used, 3) the source and method for measuring the exposure, 4) reliability, 5) controlling for confounding covariates in the analysis, and 6) the analysis performed in the study (Appendix Table A.2). We tested for bias on studies on ETS and study design strata using a method described by Egger et al.24 to identify asymmetry in a funnel plot. All statistical analyses were performed using Stata IC 13 (StataCorp, College Station, TX, USA).
Childhood TB and indoor air pollution
599
Figure 2 Forest plot showing studies on ETS exposure and childhood TB, with pooled ORs. ETS ¼ environmental tobacco smoke; OR ¼ odds ratio; CI ¼ confidence interval; TB ¼ tuberculosis.
to 9.3 (95%CI 3.1–27.6), compared to non-exposed children (Figure 2). Two studies found no statistically significant association between ETS exposure and TB disease and tuberculous infection (OR 1.4, 95%CI 0.9–2.1, and OR 2.9, 95%CI 0.7–12.4, respectively).4,13 The studies by Ramachandran et al.9 and Jubulis et al.,13 which reported an association of BMS exposure and TB, had significant estimates (ORs) of 6.9 (95%CI 2.5–18.9) and 7.2 (95%CI 1.4–44.5), respectively.9,13 We estimated random effects by pooling the effect estimates of the eight studies 1) for all the studies on ETS (Figure 2), and 2) for each stratum of outcome for the ETS exposure studies (tuberculous infection or TB disease; Figure 3). Two studies, by Den Boon et al.4 and Patra et al.,10 had risk estimates that were not statistically significant and that weighted .70% to the pooled estimate. The pooled forest plot of effect estimates (ORs) of the eight studies reporting an association between tuberculous infection and TB disease in children and exposure to ETS was 1.9 (95%CI 1.3–2.5) (Figure 2). Studies with TB disease as an outcome (defined as
clinical symptoms and radiological abnormality consistent with TB, plus a TST induration above the cut-off diameter) showed a slightly stronger association (OR 2.8, 95%CI 0.86–4.78) than those with tuberculous infection as an endpoint (OR 1.9, 95%CI 0.9–2.9) when exposed to ETS (Figure 3). The degree of heterogeneity between all ETS studies (I2 ¼ 20.6%, P ¼ 0.266) showed that although the studies made different contributions to the final, pooled estimate, the differences between the strata were not significant when random effect estimates were computed (Figures 2 and 3). Although the method employed by Egger et al. to assess publication bias is not sensitive when the number of studies is small,24 we nevertheless computed a funnel plot of logOR estimates of the eight studies on ETS against their standard errors (selogOR). The asymmetric results, with three of the studies falling outside the funnel plot, suggest that there may have been a reporting bias or chance variation in the studies selected for this review (Figure 4). The eight studies showed varying results when assessed for quality using CASP in combination with
Figure 3 Forest plot showing studies on ETS and childhood TB, stratified by outcome variable and the calculated pooled effect estimate. ETS ¼ environmental tobacco smoke; OR ¼ odds ratio; CI ¼ confidence interval; TB ¼ tuberculosis; LTBI ¼ latent tuberculous infection.
600
The International Journal of Tuberculosis and Lung Disease
Figure 4 Funnel plot with pseudo 95% CIs assessing publication bias of the eight studies on environmental tobacco smoke included in the meta-analysis. CI ¼ confidence interval; SE logOR ¼ standard errors logOR; logOR ¼ log odds ratio.
the qualitative method employed by Ijaz et al. (Appendix Table A.2).23 Five of the eight studies had low risk rating in at least four of the six elements of bias assessed, and only one of the eight studies had a high risk of bias rating for one of the elements assessed. Most of the elements were rated ‘unclear’ because insufficient detail was provided in the publications.
DISCUSSION The pooled effect estimates of this meta-analysis showed that children exposed to ETS had a two-fold increased risk of acquiring tuberculous infection and TB disease compared to non-exposed children. This was higher than the estimates reported in adult studies or in studies that included all age groups.5,11,19,25 Stratification of ETS exposure by outcome showed that the OR for TB disease was higher (OR 2.8) than that for tuberculous infection (OR 1.9) in children exposed to ETS than in nonexposed children (Figure 3). However, as all of the studies with tuberculous infection as an outcome had a cross-sectional design (Appendix Table A.1), the influence of study design on the effect estimate could not be disentangled from the effect of how outcome was defined (disease or infection). The literature provides evidence of biological mechanisms of exposure to air pollution and respiratory illnesses, with indoor air constituents such as nicotine, particulate matter (PM2.5), nitrogen dioxide and carbon monoxide implicated.26,27 The biological mechanisms involved in the relationship between IAP exposure and tuberculous infection and TB disease can be extrapolated from the work that has investigated the mechanism(s) involved in IAP constituents and other respiratory infections.26–30 A failure of the involved immune system components, such as cellmediated immunity to activate response that is optimal for M. tuberculosis clearance, can lead to the pathogen causing tuberculous infection or TB
disease. The four major immunological systems that may be implicated are 1) particle or bacterial clearance,28,29 2) recognition and immune signalling to the pathogen,31 3) intracellular antibacterial response,30 and 4) recruitment of adaptive immune components in elimination and containment of the M. tuberculosis pathogen.32 Two studies, by Den Boon et al.4 and Patra et al.,10 contributed .70% to the weighting of the pooled overall OR for the eight studies, and neither was statistically significant. Furthermore, when both studies were assessed for confounding or adjustment for covariates, they were rated respectively as high risk and unclear for bias, as they were either vague in their description or had excluded important covariates. The results from the pooling of studies on ETS which used tuberculous infection as outcome or as a measure contributing to the outcome definition of disease should be interpreted with caution, as the standard induration of the TST weal for positive infection is not the same across different populations. In populations with a high prevalence of BCG immunisation against TB and high human immunodeficiency virus (HIV) prevalence in children, the diagnosis of tuberculous infection using the TST diameter is variable. The choice of a 10-mm cut-off for three of the four studies4,8,19,20 that used TST could be due to the high BCG vaccination coverage or TB prevalence in the two populations studied, namely India and South Africa. On the other hand, in the study by Altet et al.,19 the use of a 5-mm cut-off could be related to the low BCG vaccination coverage and low TB prevalence in the Spanish population at the time of the study. In all the studies included in this review, exposure to IAP was assessed by asking the participating care givers about energy use and/or presence of smokers in their homes. This measure of pollutant exposure could have varied from study to study, depending on the details of the assessment. Most of these studies had exposure time as a factor in their assessment,
Childhood TB and indoor air pollution
especially for ETS. In some of the studies, children living in the same house as a smoker for at least 6 or 12 months were considered to be exposed to ETS,8,10,19 but the time the child spent each day in that environment was not taken into account. On the other hand, behaviours that could influence exposure of young children to BMS, such as the child being carried on the back of a woman during cooking, were usually not adequately assessed or described in most studies. Air pollution emissions from indoor cooking are affected not only by the number of hours, but also by the quality of the fuel and the technique used in burning fuel. Moreover, exposure to IAP (ETS or BMS) depends not only on the emissions, but also on the ventilation in the house (chimney, open doors and windows) and on the climate. Objective methods of exposure assessment are therefore necessary, such as measurements of air pollution in the homes. Exposures reported by care givers represent a relatively crude measure of IAP, and are usually not comparable. ETS and BMS exposures are a result of different indoor pollution sources, and a detailed assessment is needed to characterise them, especially for children, as these exposures could result in effects that are additive or synergistic. The age and sex of the children33 are also determinants for IAP exposure. In the studies included in our meta-analysis, this was addressed either by matching or by adjustment in the analysis. All of the studies included in this meta-analysis have been adjusted for one or more of the important covariates implicated in tuberculous infection or TB disease: household TB contact, socio-economic status (SES), malnutrition, age, HIV status, vaccination status, and crowding. Although HIV status of children is known to be the major driver of acquisition (tuberculous infection) and progression (TB disease),34 none of the studies quantified the effect of HIV when modelling the association between TB and IAP exposure. Six of the studies included in this meta-analysis were from high HIV burden countries, India and South Africa, where .20% of TB patients are HIV-positive.12 The covariates adjusted for across the studies were not consistent, and this could have caused some of the heterogeneity observed between them. Some covariates, such as BCG vaccination status and HIV status, are particularly important when studying TB in certain populations. This can be observed in the differences between the two studies from similar communities in South Africa published 4 years apart.4,8 In South Africa, antiretroviral treatment had been rolled out publicly only to a limited extent during the earlier study period (2003–2007),4 whereas it was widely available when the later study was conducted (from 2009 onwards).8 Neither study assessed the HIV status of the participants. The study by Den Boon et al. was conducted in populations with higher HIV prevalence,4 and may have underestimated the occurrence of tuberculous infection, as these authors used an induration of 10 mm as cut-off, and
601
there was no stratification or control for HIV in the analysis.4,16 Studies investigating the relation between IAP exposure and childhood TB have used proxy measures for estimating exposure, such as reported exposures obtained from care giver interviews and medical records. In this review, sources of bias for four of the eight studies were classified as ‘unclear’ or ‘high risk’ due to the definition of exposure, and for two of these studies also due to the measurement of exposure. A standardised methodology using a questionnaire or walkthrough instrument for assessing exposure to IAP needs to be developed for future studies. Studies that objectively quantify exposure are needed for estimating the dose-response relation of exposure to disease outcome. The limitations of this review include reporting bias that could result from a higher likelihood of studies with positive and significant findings being published than studies with negative or non-significant findings. In this systematic review, we excluded three studies because they did not report adjusted effect estimates and documented only positive significant results regarding associations between IAP and childhood TB. Furthermore, only two of the three studies separately reported the adjusted effect estimate for exposure to BMS. These studies on BMS exposure had very wide CIs, which could be the result of the small number of observations.
CONCLUSIONS Despite the limitations, pooled effect estimate indicates an increased risk of childhood TB in children exposed to ETS compared to those not exposed. The limited number of studies on the association of BMS exposure and childhood TB and the small sizes of the studies do not permit any conclusions to be drawn on causality. Although some of the reviewed studies may suffer from bias, programmes addressing IAP should be considered in the comprehensive management of the TB epidemic because of the known high incidence and severity of TB disease in children. Future studies should consider BCG and HIV status, molecular TB tests and/or isolation for better clinical definition of tuberculous infection or TB disease and more accurate evaluation of IAP exposure to confirm these findings. Acknowledgements NJ is a recipient of the National Institutes for Health-Fogarty International (Bethesda, MD, USA) and South African TB/AIDS Research Training (SATBAT, Johannesburg, South Africa) Grant: 5U2RTW00773 and 5U2RTW007373. Conflicts of interest: none declared.
References 1 Rehfuess E A, Bruce N G, Smith K R. Solid fuel use: health effect. In: Nriagu J O, ed. Encyclopedia of environmental health. Vol 5. Burlington, VT, USA: Elsevier, 2011; pp 150– 161.
602
The International Journal of Tuberculosis and Lung Disease
2 Torres-Duque C, Maldonado D, P´erez-Padilla R, Ezzati M, Viegi G. Biomass fuels and respiratory diseases. Proc Am Thorac Soc 2008; 5: 577–590. 3 Shetty N, Shemko M, Vaz M, D’Souza G. An epidemiological evaluation of risk factors for tuberculosis in South India: a matched case control study. Int J Tuberc Lung Dis 2006; 10: 80– 86. 4 Den Boon S, Verver S, Marais B J, et al. Association between passive smoking and infection with Mycobacterium tuberculosis in children. Pediatrics 2007; 119: 734–739. 5 Lin H-H, Ezzati M, Murray M. Tobacco smoke, indoor air pollution and tuberculosis: a systematic review and metaanalysis. PLOS MED 2007; 4: e20. 6 Lin H-H, Murray M, Cohen T, Colijn C, Ezzati M. 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: 1473–1483. 7 Leung C C, Lam T H, Ho K S, et al. Passive smoking and tuberculosis. Arch Intern Med 2010; 170: 287–292. 8 Du Preez K, Mandalakas A M, Kirchner H L, et al. Environmental tobacco smoke exposure increases Mycobacterium tuberculosis infection risk in children. Int J Tuberc Lung Dis 2011; 15: 1490–1496. 9 Ramachandran R, Indu P S, Anish T S, Nair S, Lawrence T, Rajasi R S. 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: 204–208. 10 Patra S, Sharma S, Behera D. Passive smoking, indoor air pollution and childhood tuberculosis: a case control study. Indian J Tuberc 2012; 59: 151–155. 11 Sumpter C, Chandramohan D. Systematic review and metaanalysis of the associations between indoor air pollution and tuberculosis. Trop Med Int Health 2013; 18: 101–108. 12 World Health Organization. Global tuberculosis report, 2013. WHO/HTM/TB/2013.11. Geneva, Switzerland: WHO, 2013. 13 Jubulis J, Kinikar A, Ithape M, et al. Modifiable risk factors associated with tuberculosis disease in children in Pune. India. Int J Tuberc Lung Dis 2014; 18: 198–204. 14 Baumgartner J, Schauer J J, Ezzati M, et al. Patterns and predictors of personal exposure to indoor air pollution from biomass combustion among women and children in rural China. Indoor Air 2011; 21: 479–488. 15 Po J Y T, FitzGerald J M, Carlsten C. Respiratory disease associated with solid biomass fuel exposure in rural women and children: systematic review and meta-analysis. Thorax 2011; 66: 232–239. 16 Marais B J, Gie R P, Schaaf H S, et al. The natural history of childhood intra-thoracic tuberculosis: a critical review of literature from the pre-chemotherapy era. Int J Tuberc Lung Dis 2004; 8: 392–402. 17 Moher D, Liberati A, Tetzlaff J, Altman D G; PRISMA Group. Reprint—preferred reporting meta-analyses: the PRISMA statement. Phys Ther 2009; 89: 873–880. 18 Singh M, Mynak M L, Kumar L, Mathew J L, Jindal S K. Prevalence and risk factors for transmission of infection among children in household contact with adults having pulmonary tuberculosis. Arch Dis Child 2005; 90: 624–628.
19 Altet M N, Alcaide J, Plans P, et al. Passive smoking and risk of pulmonary tuberculosis in children immediately following infection: a case–control study. Tubercle Lung Dis 1996; 77: 537–544. 20 Tipayamongkholgul M, Podhipak A, Chearskul S, Sunakorn P. Factors associated with the development of tuberculosis in BCG immunised children. Southeast Asian J Trop Med Public Health 2005; 36: 145–150. 21 DerSimonian R, Kacker R. Random-effects model for metaanalysis of clinical trials: an update. Contemp Clin Trials 2007; 28: 105–114. 22 DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986; 7: 177–188. 23 Ijaz S, Verbeek J, Seidler A, et al. Night-shift work and breast cancer – a systematic review and meta-analysis. Scand J Work Environ Health 2013; 39: 431–447. 24 Egger M, Smith G D, Schneider M, Minder C. Bias in metaanalysis detected by a simple, graphical test. BMJ 1997; 315: 629–634. 25 Lin H-H, Suk C-W, Lo H-L, Huang R-Y, Enarson D, Chiang CY. Indoor air pollution from solid fuel and tuberculosis: a systematic review and meta-analysis. Int J Tuberc Lung Dis 2014; 18: 613–621. 26 Gardner D E. Tobacco smoke: In: Cohen M D, Zelikoff J T, Schlesinger R B, eds. Pulmonary immunotoxicology. New York, NY, USA: Kluwer Academic Publishers, 2000: pp 387– 409. 27 Schlesinger R B, Chen L-C, Zelikoff J T. Sulfur and nitrogen oxides. In: Cohen M D, Zelikoff J T, Schlesinger R B, eds. Pulmonary immunotoxicology. New York, NY, USA: Kluwer Academic Publishers, 2000: pp 337–352. 28 Haswell L E, Hewitt K, Thorne D, Richter A, Ga¸ca M D. Cigarette smoke total particulate matter increases mucous secreting cell numbers in vitro: a potential model of goblet cell hyperplasia. Toxicol In Vitro 2010; 24: 981–987. 29 Lillehoj E P, Kato K, Lu W, Kim K C. Cellular and molecular biology of airway mucins. Int Rev Cell Mol Biol 2013; 303: 139–202. 30 Yang H-M, Antonini J M, Barger M W, et al. Diesel exhaust particles suppress macrophage function and slow the pulmonary clearance of Listeria monocytogenes in rats. Environ Health Perspect 2001; 109: 515–521. 31 Hagiwara E, Takahashi K, Okubo T, et al. Cigarette smoking depletes spontaneously secreting Th-1 cytokines in the human airway. Cytokine 2001; 14: 121–126. 32 Miranda M S, Breiman A, Allain S, Deknuydt F, Altare F. The tuberculous granuloma: an unsuccessful host defence mechanism providing a safety shelter for the bacteria? Clin Dev Immunol 2012; 2012: 139 127. 33 Kim S, Wipfli H, Navas-Acien A, et al. Determinants of hair nicotine concentrations in nonsmoking women and children: a multicountry study of secondhand smoke exposure in homes. Cancer Epidemiol Biomarkers Prev 2009; 18: 3407–3414. 34 Jeena P M, Pillay P, Pillay T, Coovadia H M. Impact of HIV-1 co-infection on presentation and hospital-related mortality in children with culture proven pulmonary tuberculosis in Durban, South Africa. Int J Tuberc Lung Dis 2002; 6: 672– 678.
