Association between Occupational Exposure to Wood Dust and Cancer: A Systematic Review and Meta-Analysis

Objective To perform a systematic review to analyze the association between occupational exposure to wood dust and cancer. Methods A systematic literature search of entries made in the MEDLINE-PubMed database between 1957 and 2013 was conducted to identify studies that had assessed the relationship between occupational exposure to wood dust and different types of cancer. A meta-analysis of selected case-control and cohort studies was subsequently performed. Results A total of 114 studies were identified and 70 were selected for review. Of these, 42 studies focused on the relationship between wood dust and nasal cancer (n = 22), lung cancer (n = 11), and other types of cancer (n = 9). Low-to-moderate quality evidence that wood dust acts as a carcinogen was obtained, and a stronger association between wood dust and nasal adenocarcinoma was observed. A lesser association between wood dust exposure and lung cancer was also observed. Several studies suggested that there is a relationship between wood dust and the onset of other cancers, although there was no evidence to establish an association. A meta-analysis that included four case-controls studies showed that workers exposed to wood dust exhibited higher rates of nasal adenocarcinoma than other workers (odds ratio = 10.28; 95% confidence interval: 5.92 and 17.85; P<0,0001), although a large degree of heterogeneity was found. Conclusions Low-to-moderate quality evidence supports a causal association between cancer and occupational exposure to wood dust, and this association was stronger for nasal adenocarcinoma than for lung cancer. There was no evidence of an association between wood dust exposure and the other cancers examined.


Introduction
Dust generated in wood processing is one of the most common occupational and carcinogenic agents identified to date. The manipulation of wood can create fine and abundant dust with sanding, and thicker dust with milling or cutting [1,2,3]. The location and accumulation of particles has been found to depend on the size, shape, and density of the air flow available. Dust accumulates in the nose or the respiratory tract when the particles are larger or smaller than 5 microns, respectively [4,5].
Exposure to wood dust has been associated with several health problems, including pulmonary pathologies and other conditions [6,7]. In particular, cancer is a pathology that has been associated with wood dust [8][9][10][11][12][13]. Consequently, in 1995, the International Agency for Research of Cancer defined wood dust as a group I human carcinogenic substance [14].
Exposure to wood dust can vary considerably among populations, and it has not been found to be specific for a single sector or professional group, or for a single cancer. However, exposure to wood dust has been specifically linked to adenocarcinoma (ADCN). Currently, exposure to wood dust has a large impact on occupational health, and its occupational prevalence ranges from 10% to 15%. While occupational exposure to wood dust potentially contributes to an increased mortality rate for certain workers, it can also affect the mortality rate of the general population. Therefore, based on the health and social impacts of wood exposure, it is important to recognize this risk and to provide adequate professional and occupational protection. Systematic reviews and/or meta-analyses represent useful methodological tools for assessing published data, and they also provide valid and reliable evidence for hypotheses [15,16]. Over the last few years, a consensus has been established to facilitate an assessment of the different primary studies that have been conducted, and to improve the quality and homogeneity of the systematic reviews that are conducted. With this in mind, it is appropriate and necessary to perform a systematic review that offers evidence on the relationship between different kinds of cancer and occupational exposure to wood dust.
Therefore, the aim of this systematic review was to analyze the data of previously published studies in relation to work exposure to wood dust and the onset of cancer.

Study design and selection criteria
A systematic review of the literature was conducted in order to identify studies that assessed the relationship and association between occupational exposure to wood dust and cancer. Selected articles (published in English or Spanish) that were available on Medline and included primary data collected between 1975 and September 2013 were selected. The exclusion criteria for this study ruled out works which consisted of opinions and/or recommendations from experts, as well as observational and experimental research studies.

Search strategy
To perform the initial bibliographic search of the MEDLINE database, the following MeSH descriptors and keywords were used to ensure a comprehensive recovery of entries: (

Information selection and extraction
After the relevant studies were identified and selected, a standardized set of information was collected from each article including: the name of the authors, the year of publication, the journal of publication, the characteristics of the sample, the study design, and the result variables and their measures of association and/or impact. This procedure was carried out according to the recommendations of the PRISMA Statement [17,18].
The literature search was performed by an experienced documentalist and data extraction was performed independently by two of the authors. All discrepancies were solved by consensus.

Scientific evidence
All selected studies were classified according to their design type, based on the classification proposed by the US Task Force and the Centre for Evidence Based Medicine (CEBM) of Oxford [19,20]. The categories included: I) Evidence obtained from a single randomized controlled trial or a meta-analysis of randomized controlled trials; IIa) Evidence obtained from at least one well-designed controlled study without randomization; IIb) Evidence obtained from at least one well-designed quasi-experimental study; III) Evidence obtained from well-designed non-experimental descriptive studies, such as comparative studies, correlation studies, and case-control studies; and IV) Evidence obtained from expert committee reports or opinions and/or clinical experience of respected authorities. The levels according to CEBM include: level 1 (systematic review of randomized control trials), level 2 (observational study with dramatic effect), level 3 (cohort study), level 4 (case-series or case-control studies), and level 5 (mechanism-based reasoning). To improve the scientific rigor of the present systematic review, all of the published articles that were classified by each of the two reviewers as level 5 were excluded.

Meta-analysis
Case-control studies that analyzed the relationship between wood dust exposure and sinonasal ADCN were included in our meta-analysis. The following inclusion criteria were used: a) casecontrol studies published in peer-reviewed journals, b) description of occupational exposure to wood dust among cases and controls; and c) diagnosis of sinonasal ADCN by biopsy. The main purpose of the meta-analysis was to compare the presence of sinonasal ADCN among workers exposed to wood dust with non-exposed workers used as controls in case-control studies.
Data were extracted from the selected studies by three authors (A.-J.C, I.H.-G., and C.H.) and differences were solved by consensus. The odds ratio (OR), 95% confidence interval (CI), and P values are reported for the pooled results based on the use of a random effects model (DerSimonian and Laird method) [21]. A random effects model was chosen due to the low number of studies available and their observational nature. A P-value less than 0.05 was considered statistically significant. Cochran's Q-statistic was used to assess heterogeneity. A significant Q-statistic value (P < 0.10) indicated heterogeneity across the studies examined. The I 2 statistic was used to estimate inconsistency in the meta-analysis, thereby representing the percentage of the observed between-study variability due to heterogeneity rather than chance. The following suggested cut-off points were used: I 2 = 0-25%, no heterogeneity; I 2 = 25-50%, moderate heterogeneity; I 2 = 50-75%, large heterogeneity; I 2 = 75-100%, extreme heterogeneity [22].
A sensitivity analysis was performed to assess the effect of excluding individual studies in the results. The meta-analysis was performed by using the computer software package RevMan 5.0 [23].

Systematic review
Among the articles that investigated the relationship between exposure to wood dust and cancer and were published between 1975 and September 2013, 70 studies were selected for this review. Reports which only included opinions and/or recommendations from experts were excluded. Of the 70 selected studies, 42 had investigated the relationship between wood dust exposure and cancer according to the incidence of nasal ADCN (n = 22), lung cancer (n = 11), and other cancers (n = 9) (see Fig 1).
The historical evolution of the publications indexed in Medline is shown in Fig 2. A total of 114 articles were collected, and their chronological evolution is demonstrated in the vertical cylinders graph. As shown in Fig 2 scientific publications have progressively increased until they reached a peak between 2000 and 2013.
As shown in Table 1 most of the selected studies addressed the causal relationship of wood dust with ADCN, and case-control studies  and case series studies were predominant. However, there were other, less common study designs [27] and descriptive or populationbased studies [35,38,41] as well. Furthermore, most of the studies assessed the epidemiological association by obtaining a clinical history and/or occupational history for each of the patients examined. The most widely analyzed profession was that of carpenters, followed by sawmill workers and other workers who manipulate wood in their work. OR values were the main measure of impact used in the studies. Bias control and control of confounding variables were reported in eleven studies [26,27,[29][30][31][32][33]36,37,[39][40][41], and these generally included research performed over the last decade. The most common risk factor that was adjusted for was tobacco consumption [26,29,30,33,41], and it was analyzed through logistic regression. Table 2 lists eleven articles that analyzed the association between exposure to wood dust and lung cancer. Of these, seven were case-control studies  and three were population and retrospective studies [48,49,53]. OR was the main measure of impact used in these articles, and bias control and control of confounding variables were only present in four studies which were conducted over the last decade [46][47][48]50,55]. In nine of the studies, an association between lung cancer and wood dust exposure was observed. In the other two studies, a statistically significant association was not observed, potentially due to the low quality of these studies [52,53].
The nine studies that investigated the association between exposure to wood dust and other types of cancer are listed in Table 3. Two of these articles focused on the relationship between exposure to wood dust and lymphomas [56][57][58]60], while the other studies involved the following types of cancer: thyroid [61][62], mesothelioma [56], multiple myeloma [61], gastric cardia [59], glottic [63], and sarcoma [57]. Case-control studies were the most common (n = 6), followed by records and case reports (n = 2) and population-based cohort studies (n = 1). Except for a study of multiple myeloma, all of the other studies identified a statistically significant association between exposure to wood dust and cancer. Table 4 lists the reviews that have been published regarding exposure to wood dust and cancer. The list includes a meta-analysis [64][65] of larynx cancer, two systematic reviews [66][67][68][69], and five narrative reviews [33,67,68,70,71]. In the former, it was concluded that there was no association between larynx cancer and wood dust exposure.

Meta-analysis
Among the studies included in our systematic review, and according to the predefined criteria for this study, five reports were selected for a meta-analysis [26,35,43,44]. The summary OR under the random-effects model showed that subjects with wood dust exposure exhibited higher rates of sinonasal ADCN compared with non-exposed workers (see  observed between the studies (I 2 = 85%). However, in the sensitivity analysis, the exclusion of individual studies did not change this significant result (see Table 5).

Discussion
The studies that were selected and reviewed show that: 1) wood dust may act as a carcinogen, and 2) there is an association between exposure to wood dust and nasal ADCN, and to a lesser extent, with lung cancer. The results of several studies also suggest that there is a relationship between exposure to wood dust and other types of cancer, although there is currently not sufficient data or evidence to clearly establish this association. Moreover, the design of the included studies (mostly case-control studies) potentially limits the strength of the latter association.
The historical evolution of studies in this field has grown in parallel with the interest and impact of wood dust as an occupational exposure. At first, epidemiological research was mainly observational, and was based on case studies that involved occupations where exposure to wood dust was a factor. Thus, the early studies were of carpenters and sawmill workers. However later on, the studies included other situations where wood is manipulated or processed as a secondary and/or complementary activity. Analytical case-control studies also started to be conducted, thereby adding scientific rigor to the hypothesis by establishing comparisons with Most of the evidence presented in the present study was extracted from case-control analytical observational studies. This type of study design, as pointed out in two previous systematic reviews [66,69], is the most appropriate since it makes it possible to study diseases with a large latency period and to assess several types of exposure concomitantly. This study design is also economical and rapid. However, case-control studies do have limitations. For example, potential bias exists in the selection of controls, the temporal relationship between the presumed cause and the studied effect cannot be determined, and there is a need for specific biases to be controlled through adjustment techniques. These considerations have been highlighted in most of the studies and works published over the last two decades. However, advances in statistical techniques or procedures have made it possible to control confounding factors and biases, thus improving the quality of the observations. Overall, a positive evolution in the methodological quality of the studies and research on wood dust exposure has lead to the publication of evidence that has greater validity and reliability. There was also a wide geographical distribution for the studies that were considered, with the majority conducted in Europe and the United States, followed by studies conducted in Brazil and Colombia. The first study was conducted in 1975 in Australia [45]. However, very few of these studies were included in our meta-analysis.
Apart from the relationship between wood dust and nasal ADCN, the association between exposure to wood dust and other neoplasms, such as lung cancer, has also been assessed. Based on the data of the studies published to date, this association has not been established. Primarily due to the low quality of the evidence collected, the association has been weakly demonstrated in most studies, and evidence was nonexistent in two other studies [52,53]. Consequently, while a potential association exists, it remains to be validated. that was employed in most of the studies conducted. For example, most of the evidence available is from observational studies, and specifically, from analytical case-control studies. Consequently, a temporal relationship between wood dust exposure and cancer cannot be determined. We also acknowledge that the use of other databases apart from PubMed could have yielded additional results, although we hypothesize that the final result of our work would be similar.
The most important limitation of the articles that were analyzed in this systematic review is that the characterization of occupational exposure in many of the articles published, particularly up until the 1990s, has been retrospective. In addition, the characterization was assessed through the working and occupational history of the patients examined. The latter aspect may be further biased by the quality of the documents examined, the maintenance of the working and occupational history files, and the memory of the workers. Other important limitations include the low number of studies with a cohort-type or prospective monitoring design, the sample size of individual studies, and the existence of few multicentric studies; although the latter is diminished by the wide geographic distribution of the studies that have been conducted. Another limitation of the earlier studies is the lack of adjustment for occupational or environmental risk factors that can act as confounders. These should be adjusted and/or neutralized since they can be linked to both the exposure (cause) and the effect; with a specific example being tobacco.
Regarding the meta-analysis performed, there were very few studies that met the inclusion criteria due to methodological differences. In addition, large heterogeneity was observed between the included studies. Therefore, although our meta-analysis confirms that a significant relationship between wood dust and nasal ADCN exists, this result should be approached with caution.

Conclusions
The conclusion of this systematic review is that there is low-to-moderate quality evidence that supports a causal association between the incidence of cancer and occupational exposure to wood dust. However, the association between exposure to wood dust and nasal ADCN is stronger, largely because most of the causal criteria established by Bradford Hill have been assessed [72]. In regard to lung cancer, caution is still advised in establishing an association with wood dust exposure given the low number of studies that have been conducted and their poor methodological quality.
Nevertheless, it is apparent that there is a need to implement preventive measures for workers exposed to wood dust. We propose that it is appropriate and adequate to establish a series of primary and secondary preventive measures in professional and working environments in order to improve the working health, hygiene, and safety of workers exposed to wood dust. It is critical that future studies overcome the limitations that have been observed in the present systematic review, particularly by identifying the characteristics of occupational exposure, adjusting for other exposures and confounding factors, increasing the sample size, and making comparisons with better control groups.