PREreview del Exposure to Respirable Dust, Fine Particulates and Crystalline Silica and Comparative Respiratory Health Patterns Among Non-Smoking Workers in the Ceramic Industry

This report summarizes the comments arising from a preprint journal club session in which we assessed the manuscript “Exposure to Respirable Dust, Fine Particulates and Crystalline Silica and Comparative Respiratory Health Patterns Among Non-Smoking Workers in the Ceramic Industry” (https://doi.org/10.21203/rs.3.rs-8487804/v1). During approximately 45-minute discussion, the group identified reporting aspects which could be strengthened. It is worth highlighting that, since this was not intended as an exhaustive peer review, it is possible that not all relevant aspects of the manuscript were discussed during the meeting. The comments below are organized by manuscript section and are offered as constructive suggestions for improvement.

Introduction

1. Clarification of the research question. The current phrasing of the study objective does not make explicit whether the research question is descriptive, predictive, or aimed at causal inference. Since the language used later in the manuscript implies causal relationships (see Discussion comments below), it would be advisable to clarify the nature of the research question upfront; this distinction would help align the study design, analysis approach, and the interpretation of findings accordingly.

2. Approach to confounding through exclusion. The exclusion of smokers and individuals with a history of respiratory disease is presented as a strategy to minimize confounding. However, whether exclusion is the most appropriate approach for addressing these factors, as opposed to statistical adjustment, deserves further consideration. Excluding these groups reduces the generalizability of the findings to the broader ceramic industry workforce, where smoking and pre-existing disease are likely prevalent; furthermore, the extent of this limitation cannot currently be judged from the data presented. Thus, it would be advisable to include a participant flow diagram following STROBE recommendations, so that readers can assess the number of individuals screened, excluded at each step, and ultimately included. This information would help quantify the impact of the exclusion criteria on the final study population.

3. Balance of introduction and discussion content. Some of the background material could be better organized. The rationale for the exposure–response analysis and its interpretation would benefit from being discussed more explicitly alongside the study findings in the Discussion, whereas the Introduction could focus on establishing the knowledge gap and the study objective.

Methods

Study Design, Setting, and Population

1. Number of industries and participant distribution. The manuscript does not report the total number of ceramic factories included in the study. It would be advisable to clearly state this number and to provide the distribution of participants by job title within each factory, since this information is essential for evaluating the representativeness of the sample and for understanding potential clustering across facilities.

2. Exclusion of workers with fewer than two years of employment. Workers with fewer than two years of employment were excluded to ensure chronic exposure assessment. Based on the distribution of duration in the industry reported, it is not clear that this exclusion criterion alone achieved assessment of chronic exposure health effects (see comment #19). It would be advisable to discuss if this may have affected the findings and to weigh any risks of selection bias against the intended analytical advantages.

3. Relationship between zones and job titles. Work zones (ball mill, spray dryer, kiln, reception) are used as the primary exposure classification; however, the manuscript does not clearly describe the job titles assigned to each zone or whether workers rotate across zones. Clarifying this would be advisable, since zone-based exposure estimates can only be meaningfully linked to health outcomes if the correspondence between zones and job roles is well-defined and stable.

4. Control group definition and potential effect underestimation. The use of administrative staff working within the same factories as a comparison group has the advantage of sharing similar socioeconomic and demographic backgrounds with production workers. However, the results indicate that administrative areas are not free of occupational exposure, with measurable levels of respirable dust, PM2.5, and a relatively high proportion of crystalline silica. This means that the comparison group was also exposed to some degree, a reason which could lead to underestimation of the true effect of occupational dust exposure on respiratory health. Accordingly, it would be advisable to discuss this potential source of bias and to consider whether a different, truly unexposed comparison group (e.g., a community-based sample) could have been more appropriate. Furthermore, it should be clarified whether “administrative staff” and “reception” staff (used as a zone in Table 1 and Table 5) refer to the same group, since the exposure profile of the control group depends on this distinction.

5. Non-smokers restriction. Whereas restricting the study to lifelong non-smokers addresses the role of smoking as a confounder, it also limits the generalizability of the findings to the broader ceramic workforce, which likely includes a substantial proportion of smokers. It would be advisable to acknowledge this limitation and to discuss whether the observed effects may underestimate the true burden of respiratory disease in this industry, since smoking and occupational dust exposure may act synergistically.

6. Sex distribution. The manuscript does not describe the sex composition of the study sample. Since sex is a well-established determinant of lung function and spirometric predicted values are sex-specific, it would be advisable to report the sex breakdown of participants and discuss its implications for the analyses.

Exposure Assessment

1. Number of personal measurements. The total number of personal air samples collected is not clearly stated. It would be advisable to specify how many measurements were taken per zone, per shift, and in total, since without this information it is difficult to judge the reliability of the reported exposure concentrations.

2. Variability of exposure. Table 1 presents only single-point TWA values for each zone with no measure of variability. Occupational exposure data are typically log-normally distributed and exhibit substantial day-to-day and between-worker variability. Thus, it would be advisable to report the geometric mean and geometric standard deviation for each zone and, where possible, to present exposure distributions stratified by job title within each zone. Without these data, it is not possible to assess how representative the reported values are of typical worker exposure.

Statistical Analysis

1. Treatment of continuous variables. Several continuous variables (e.g., age, years of experience) are categorized in the descriptive analyses and, in some cases, in the models. It would be advisable to analyze continuous variables as such whenever possible, since categorization leads to loss of information and reduces statistical power. At a minimum, continuous distributions should be reported for descriptive purposes. Where categorization is deemed necessary, greater justification for the chosen cut-offs and their impact on the analysis should be provided.

2. Model for respiratory symptoms. The multivariable regression models presented in Table 4 address spirometric outcomes but do not include a model for the primary outcome of respiratory symptoms (WRURS and WRLRS). Since respiratory symptoms are the main dependent variable used for the sample size estimation, it would be advisable to present a model adjusted for relevant confounders for these outcomes as well.

3. Clarification of Table 4. In Table 4, the variables “job profile (worker vs. admin)” and “duration of exposure (months)” would benefit from clearer definitions. Specifically, it is unclear whether “duration of exposure” refers to total time worked in the ceramic industry or to time spent in the current job or zone. This distinction is important since, in a cross-sectional study, duration of employment does not necessarily reflect the intensity or consistency of exposure over time (see also comment 19 below).

4. Adjustment for confounders. The regression models adjust for age, exposure duration, and job role, whereas other established predictors of lung function are omitted. Important examples of such predictors are sex and height, both of which are strong determinants of spirometric values. Their omission may result in residual confounding and biased estimates; thus, it would be advisable to include these variables in all lung function models or to justify their exclusion. Furthermore, if job is included as a covariate while the model simultaneously adjusts for exposure duration, it would be helpful to discuss the potential collinearity between these variables.

5. Outcome specification in regression models. It is unclear whether the spirometric outcomes used in the regression models (Table 4) are absolute values (e.g., FEV₁ in liters) or percent predicted values (e.g., FEV₁%). This distinction is critical since absolute values require adjustment for age, sex, and height, whereas percent predicted values already account for these factors. Additionally, beta coefficients cannot be properly interpreted in the absence of this information. It would be advisable to clearly specify the outcome used and to ensure the modelling approach is consistent with that choice.

6. Tables 4 and 5. A footnote or legend clearly listing all variables included in the regression models, not just the predictors displayed, would be a good addition. Table 5, on the other hand, presents descriptive exposure and spirometry data by zone but includes no statistical testing. The absence of any formal comparison weakens the claim of an “exposure–response gradient.” If statistical testing was not performed on Table 5, it would be advisable to state this explicitly and to temper interpretive language accordingly.

Results

1. Crystalline silica in administrative areas. One noteworthy finding is that the proportion of crystalline silica in respirable dust was highest in the reception/administrative area (1.29%), despite this zone having the lowest total dust concentration. This observation is only briefly mentioned and deserves more substantive discussion. Several possible explanations could be explored: particle size fractionation during dispersion from production zones, whereby coarser particles settle more readily and leave a higher relative proportion of fine silica-containing particles in distant areas; ventilation patterns carrying fine silica-rich aerosols from production into administrative spaces; resuspension of settled silica-containing dust through foot traffic or material transport between zones; and the physical proximity and spatial layout of administrative areas relative to production zones, which in the ceramic cluster of Morbi may involve shared buildings or open layouts with limited physical barriers. The implications for the health of administrative workers, who may be assumed to be unexposed, should also be discussed.

2. Workforce experience and exposure classification. Table 2 shows that nearly half of the workers (47.1%) had only 1–3 years of experience and only 10% had more than 7 years. This means the study sample does not adequately represent workers with long occupational histories, which limits the ability to detect chronic or cumulative effects such as restrictive lung disease. Furthermore, the categorization of years of experience into broad groups (1–3, 3–7, >7 years) results in a loss of information about the underlying distribution. The study also does not distinguish between a newly hired worker in a high-exposure zone and an experienced worker in the same zone; both would be classified identically in the current framework, despite markedly different cumulative exposure. This limitation should be discussed.

3. Duration of exposure vs. current exposure intensity. The manuscript treats “duration of exposure” as a proxy for cumulative dose; however, in a cross-sectional design, duration of employment does not necessarily reflect the intensity or consistency of exposure over time. Workers may have changed zones, tasks, or factories during their career. Thus, it would be advisable to acknowledge that duration and current cross-sectional exposure levels are not interchangeable.

4. Exposure–response analysis. The authors describe an “exposure–response relationship” based on Table 5; however, this table only presents mean spirometry values alongside mean exposure levels by zone, without any formal analysis. Whereas individual-level exposure–response relationships cannot be established in a cross-sectional design, a group-level analysis could strengthen the manuscript. For instance, the authors could use zone-level mean exposure estimates as an exposure variable and model lung function outcomes at the group level, or use trend tests across ordered exposure categories.

5. Cumulative exposure estimation. Since occupational history data were collected, it would be advisable to explore the possibility of constructing a cumulative exposure index (e.g., combining current zone-specific exposure levels with years of service in each zone).

Discussion

1. Causal language. Throughout the discussion and conclusion, the manuscript uses causal language or language implying causality. There are two concerns which should be addressed. First, the cross-sectional design inherently limits the ability to establish temporal or causal directionality. Second, the manuscript does not employ a formal causal inference framework. No directed acyclic graph (DAG) or equivalent tool is used to articulate the assumed causal structure, identify confounders, mediators, or colliders, or justify the set of adjustment variables. Without such a framework, claims about causal or etiological relationships are not well-supported regardless of the study design. Accordingly, it would be advisable to revise these statements to reflect the associational nature of the findings, or alternatively, to incorporate a causal inference framework if causal claims are intended.

Competing interests

The authors declare that they have no competing interests.

Use of Artificial Intelligence (AI)

The authors declare that they did not use generative AI to come up with new ideas for their review.