Seroprevalence of Coxiella burnetii in Occupational Settings: A Meta-Analysis of Italian Studies

Simple Summary: Q fever is a disease caused by the bacteria Coxiella burnetii . This pathogen usually infects some animals (i


Introduction
Q Fever (QF) is a zoonotic infectious disease with global distribution caused by the obligate intracellular bacterium Coxiella burnetii (C. burnetii) [1,2], a small (0.2 to 0.4 µm × 0.4 to 1.0 µm) gram-negative pathogen belonging to the Coxiellaceae family [3]. C. burnetii infects a wide range of domesticated and wild animals, and human cases result from contact via airborne routes, after the organism settles in dust and becomes aerosolized [3][4][5].
With a case fatality ratio ranging between 0.9% to 2.4%, QF in humans rarely represents a deadly disease, but it is usually acknowledged as a debilitating one [3,[6][7][8]. Even though up to 60% of incident cases may go unnoticed, the large majority of remaining cases evolves into mild syndromes characterized by common and unspecific symptoms such as fever, asthenia, chills, headache, myalgia, skin rashes, sweating, nausea, vomiting, and diarrhea [2,3], and in some cases may evolve into pneumonia, hepatitis, myocarditis, and even meningoencephalitis (Acute QF); moreover, QF may also develop into a chronic

Item Definition
Population of interest Workers potentially exposed to C. burnetii Investigated result Seroprevalence of biomarkers for previous exposure to C. burnetii Control Healthy individuals not occupationally exposed Outcome Seroprevalence of previous infection of C. burnetii among occupationally exposed individuals; risk of Q Fever in occupational settings.
Two conventional scientific databases (i.e., PubMed and EMBASE) and the preprint repository MedRxiv were searched through a combination of the following keywords: ("Q Fever" OR "Coxiella burnetii" OR "Coxiella") AND ("Italy" OR "Italian") AND ("epidemiology" OR "seroprevalence" OR "prevalence" OR "frequency" OR "occurrence"). No chronological restrictions were applied.
Documents eligible for review were original research publications available online or through inter-library loan, including case studies, cohort studies, case-control studies, and cross-sectional studies. Retrieved documents were excluded if: (1) full text was not available to the reviewers; (2) articles were written in a language not understood by reviewers (i.e., Italian, English, German, French, or Spanish); (3) reports lacked timeframe (i.e., the prevalence year) and geographical settings, or it was only vaguely defined; and (4) a proper definition of the occupational settings was lacking.
Retrieved entries were initially screened for their titles in terms of relevance to the subject. Titles that were considered consistent with the research outline were subsequently analyzed by their abstracts. If the content was in turn consistent with the design of the present review, full-text versions of eligible articles were independently read by two investigators (AB and FM). Disagreements were resolved by consensus between the two reviewers; where they did not reach consensus, input from a third investigator (MR) was obtained. Retrieved studies were then rated about their potential risk of bias by means of the National Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) handbook and respective risk of bias (ROB) tool [33,34]. Briefly, the ROB tool evaluates the internal validity of a given study in order to assess whether the study's design and conduct have compromised the credibility of the link between exposure and outcome. In its current version, OHAT ROB tool covers six possible sources of bias (i.e., participant selection, confounding, attrition/exclusion, detection, selective reporting, and other sources) with potential answers ranging from "definitely low", "probably low", and "probably high" to "definitely high". Interestingly, OHAT ROB tool does not apply an overall rating for each study, and OHAT handbook also recommends that even studies with "probably high" or "definitely high" ratings should not be removed from consideration of the overall body of evidence.
Characteristics of the included studies were initially summarized through descriptive analysis, with subsequent calculation of crude prevalence figures. If a study did not include raw data, either as number of prevalent cases or referent population, such figures were reverse-calculated from available data. Pooled prevalence estimates were then calculated by means of prevalent cases per 100 population.
Meta-analysis of retrieved studies was performed through a random effect model in order to cope with the presumptive heterogeneity in design of the included reports. The amount of inconsistency between studies was estimated by means of I 2 statistic (i.e., the percentage of total variation across studies that is due to heterogeneity rather than chance). For the aims of this study, I 2 values were categorized as follows: 0 to 25% low heterogeneity; 26% to 50% moderate heterogeneity; and ≥ 50% substantial heterogeneity. Contour-enhanced funnel plots and radial plots were generated in order to visually assess potential publication bias and small study bias, respectively. Funnel plot asymmetry was eventually assessed by means of the Egger test statistic. All calculations were performed in R (version 4.0.3) [35], and RStudio (version 1.4.1717; Rstudio, PBC; Boston, MA, USA) software by means of the meta package (version 4.9-9).

Results
Results of the inquiry are summarized in Figure 1. Briefly, a total pool of 313 entries (i.e., 142 from PubMed; 6 from MedRxiv; and 165 from EMBASE) were initially retrieved; of them, 107 were duplicated entries, therefore being removed. The remaining 206 records were then screened by title and abstract (65.8% of the original pool), and a total of 184 entries (58.8% of the original pool) were eventually removed from the analyses. The remaining 22 entries were assessed and reviewed by full-text. percentage of total variation across studies that is due to heterogeneity rather than chance). For the aims of this study, I 2 values were categorized as follows: 0 to 25% low heterogeneity; 26% to 50% moderate heterogeneity; and ≥ 50% substantial heterogeneity. Contour-enhanced funnel plots and radial plots were generated in order to visually assess potential publication bias and small study bias, respectively. Funnel plot asymmetry was eventually assessed by means of the Egger test statistic. All calculations were performed in R (version 4.0.3) [35], and RStudio (version 1.4.1717; Rstudio, PBC; Boston, MA, USA) software by means of the meta package (version 4.9-9).

Results
Results of the inquiry are summarized in Figure 1. Briefly, a total pool of 313 entries (i.e., 142 from PubMed; 6 from MedRxiv; and 165 from EMBASE) were initially retrieved; of them, 107 were duplicated entries, therefore being removed. The remaining 206 records were then screened by title and abstract (65.8% of the original pool), and a total of 184 entries (58.8% of the original pool) were eventually removed from the analyses. The remaining 22 entries were assessed and reviewed by full-text.
Eventually, 15 of them were excluded due to not fitting the inclusion criteria. The remaining 7 papers were eventually included in qualitative and quantitative analysis (2.2% of the initial sample). The retrieved studies (range: 2006-2022) [13,14,18,25,26,36,37] included a total of 16 estimates from occupational settings and 3 reference groups. Of the aforementioned studies, 3 were based on immunofluorescence assay (IFA), 3 on enzyme-linked immunoassay (ELISA), and one complement fixation test (CFT). Eventually, 15 of them were excluded due to not fitting the inclusion criteria. The remaining 7 papers were eventually included in qualitative and quantitative analysis (2.2% of the initial sample).
A total of 1670 specimens from individuals living and working in 6 Italian regions (i.e., Lombardy, Friuli Venezia Giulia, Tuscany, Apulia, Lombardy, and Sicily; Appendix A Figure A1) and belonging to 9 distinctive occupational groups were retrieved. Of them, 1314 were potentially exposed to C. burnetii because of their professional tasks (78.7% of the sample), and more precisely, agricultural workers operating as livestock farmers (No. 158, 9.5% of the total sample); agricultural workers having mixed tasks, including caring for animals (No. 223, 13.4%); agricultural workers not caring for animals (No. 32  Interestingly, while two studies [13,25] also estimated seroprevalence for C. burnetii in healthy blood donors from the very same geographic areas, with a total of 42 subjects over 322 samples (13.0%; actual range from 13.6% [13] to 14.3% [25]), the study from Stufano et al. included administrative workers (No. 34) not exposed to C. burnetii in occupational settings [26]. Overall, a total of 356 not-occupationally exposed individuals were reported and represented the reference group (21.3% of the total sample).
A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to the National Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) handbook and respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2: exposure assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6: other bias.

ER REVIEW 6
A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to the National Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) handbook and respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2: exposure assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6: other bias.  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2. A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to the National Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) handbook and respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2: exposure assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6: other bias.  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved stud summarized in Table 3 and Figure 2.   Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to the National Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) handbook and respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2: exposure assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6: other bias.  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to the National Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) handbook and respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2: exposure assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6: other bias.  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to the National Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) handbook and respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2: exposure assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6: other bias.  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessment on retrieved summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to th Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) han respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2 assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6:  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessment on retriev summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) h respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessm summarized in Table 3 and Figure 2. A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to the National Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) handbook and respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2: exposure assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6: other bias.  = definitively high;  = probably high;  = probably low;  = definitively low.

RISK OF BIAS
A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to the National Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) handbook and respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2: exposure assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6: other bias.  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to the National Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) handbook and respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2: exposure assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6: other bias.  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to the National Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) handbook and respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2: exposure assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6: other bias.  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessment on retrieved summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to th Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) han respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2 assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6:  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessment on retriev summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) h respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessm summarized in Table 3 and Figure 2. A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to th Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) han respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; D2 assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D6:  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessment on retriev summarized in Table 3 and Figure 2. Table 3. Tabular representation for the Risk of Bias (ROB) assessment according to Toxicology Program (NTP)'s Office of Health Assessment and Translation (OHAT) h respective risk of bias (ROB) tool [33,34]. Note: D1: possibility of selection bias; assessment; D3: outcome assessment; D4: confounding factors; D5: reporting bias; D  = definitively high;  = probably high;  = probably low;  = definitively low. A detailed description of the risk of bias (ROB) assessm summarized in Table 3 and Figure 2. A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retriev summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessm summarized in Table 3 and Figure 2. A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retriev summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessm summarized in Table 3 and Figure 2. A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retriev summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessm summarized in Table 3 and Figure 2. A detailed description of the risk of bias (ROB) assessment on retrieved studies is marized in Table 3 and Figure 2.  REVIEW 6 A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieved studies is summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retrieve summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessment on retriev summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assessm summarized in Table 3 and Figure 2.  A detailed description of the risk of bias (ROB) assess summarized in Table 3 and Figure 2.   Overall, the overall quality of retrieved studies was largely unsatisfying. First of all, in nearly all studies, how the sample was ultimately recruited remains largely unclear. Moreover, in the large majority of retrieved reports, the very same exposure is not clearly stated. For example, in the study by Aquilini et al., [18] we are only briefed that the sampled individuals (i.e., 507) were "forestry workers", and even though Authors did recall an estimate for the exposure assessment, it was not reported through the actual tasks performed by study participants, and the risk were summarized in terms of exposure to milk, ticks, and animals. Similarly, the study of Fenga et al. reports on the seroprevalence for C. burnetii among "laboratory workers" [25]. It is quite unclear whether the participants had any occupational interaction with specimens possibly contaminated by the pathogen. Moreover, more than half of the included studies were based on laboratory techniques such as IFA [13,18,37] and even CFT [14], that are far less sensitive and specific than ELISA [25,26,36]. Eventually, the analysis of the possible role of non-occupational exposures was not clearly addressed in the majority of the assessed studies, while no clear reporting bias could be identified among the pooled reports. Overall, the overall quality of retrieved studies was largely unsatisfying. First of all, in nearly all studies, how the sample was ultimately recruited remains largely unclear. Moreover, in the large majority of retrieved reports, the very same exposure is not clearly stated. For example, in the study by Aquilini et al. [18] we are only briefed that the sampled individuals (i.e., 507) were "forestry workers", and even though Authors did recall an estimate for the exposure assessment, it was not reported through the actual tasks performed by study participants, and the risk were summarized in terms of exposure to milk, ticks, and animals. Similarly, the study of Fenga et al. reports on the seroprevalence for C. burnetii among "laboratory workers" [25]. It is quite unclear whether the participants had any occupational interaction with specimens possibly contaminated by the pathogen. Moreover, more than half of the included studies were based on laboratory techniques such as IFA [13,18,37] and even CFT [14], that are far less sensitive and specific than ELISA [25,26,36]. Eventually, the analysis of the possible role of non-occupational exposures was not clearly addressed in the majority of the assessed studies, while no clear reporting bias could be identified among the pooled reports.
Interestingly, by assuming non-occupationally exposed individuals as the reference group (Figure 4) 3.703). On the contrary, a reduced risk was associated with individuals working as forestry rangers (RR 0.201, 95%CI 0.081 to 0.495), and no substantial differences with non-exposed individuals were identified among laboratory and technical professionals (RR 1.453, 95%CI 0.582 to 3.625).

Figure 4.
Forest plot representing risk ratio on main occupational groups compared to non-occupationally exposed workers (49 cases out of a total of 356 donors, 13.8%). Note: RR = Risk Ratio; 95%CI = 95% Confidence Interval.
When seroprevalence was compared by the parent Italian region, after the exclusion of non-exposed individuals, it ranged between 2.8% from Friuli Venezia Giulia to 49.0% in Apulia (Appendix A Figure A1). Assuming workers from Lombardy (the most populated Italian region, with around 10,000,000 inhabitants, that is 1/6 of the total Italian population) as the reference group, the risk ratio for reporting seroprevalence on C. burnetii ranged between RR 0.063 (95%CI 0.026 to 0.154) in the North-Eastern Region of Friuli Venezia Giulia to RR 0.568 (95%CI 0.432 to 0.746) in Tuscany and RR 0.980 (95%CI 0.751 to 1.278) in Sicily, with the greatest estimates for Basilicata (RR 1.087, 95%CI 0.767 to 1.593) and Apulia (RR 1.119, 95%CI 0.861 to 1.453) ( Figure 5). When seroprevalence was compared by the parent Italian region, after the exclusion of non-exposed individuals, it ranged between 2.8% from Friuli Venezia Giulia to 49.0% in Apulia (Appendix A Figure A1). Assuming workers from Lombardy (the most populated Italian region, with around 10,000,000 inhabitants, that is 1/6 of the total Italian population) as the reference group, the risk ratio for reporting seroprevalence on C. burnetii ranged between RR 0.063 (95%CI 0.026 to 0.154) in the North-Eastern Region of Friuli Venezia Giulia to RR 0.568 (95%CI 0.432 to 0.746) in Tuscany and RR 0.980 (95%CI 0.751 to 1.278) in Sicily, with the greatest estimates for Basilicata (RR 1.087, 95%CI 0.767 to 1.593) and Apulia (RR 1.119, 95%CI 0.861 to 1.453) ( Figure 5). When dealing with the assessment of the potential publication and small study bias, mixed results were retrieved. On the one hand, the radial plot calculated by single prevalence estimates (Figure 6), hinted towards no effects for the small size of samples included in the estimates. On the other hand, the substantial asymmetry of the corresponding funnel plot (Figure 7) suggested a residual publication bias, that was substantially rejected by the Eggers's test (t = −1.50, df = 14, intercept = 2.5879, SE = 1.7307, p-value = 0.157). When dealing with the assessment of the potential publication and small study bias, mixed results were retrieved. On the one hand, the radial plot calculated by single prevalence estimates (Figure 6), hinted towards no effects for the small size of samples included in the estimates. On the other hand, the substantial asymmetry of the corresponding funnel plot (Figure 7) suggested a residual publication bias, that was substantially rejected by the Eggers's test (t = −1.50, df = 14, intercept = 2.5879, SE = 1.7307, p-value = 0.157). When dealing with the assessment of the potential publication and small study bias, mixed results were retrieved. On the one hand, the radial plot calculated by single prevalence estimates (Figure 6), hinted towards no effects for the small size of samples included in the estimates. On the other hand, the substantial asymmetry of the corresponding funnel plot (Figure 7) suggested a residual publication bias, that was substantially rejected by the Eggers's test (t = −1.50, df = 14, intercept = 2.5879, SE = 1.7307, p-value = 0.157).

Discussion
In our systematic review and meta-analysis, a pooled prevalence of seropositivity for C. burnetii on occupationally exposed individuals was estimated at 44.0% (95%CI 27.6 to 61.8). The risk for seropositivity was greatest among veterinary professionals, followed by abattoir workers, livestock farmers, agricultural workers, geologists/agronomists, and forestry workers, while laboratory professionals had no substantially increased risk, and forestry rangers exhibited a conversely reduced seropositivity risk.
In other words, the present study is consistent with some previous reports, suggesting that seroprevalence for C. burnetii among professionals working with animals or involved in meat processing may be up to 50% [1,6,7,10,16,19,[38][39][40], largely exceeding estimates in non-occupationally exposed individuals from the index areas (where prevalence estimates ranging between 13.6% and 14.3% were retrieved) [13,25], and national figures for non-occupational exposed individuals [6,19,37,41,42]. In this regard, it should be stressed that according to official reports, between 2015 and 2021, a total of 17 cases of QF have been reported to Italian health authorities, for an annual reference rate well below 0.1 cases per 100,000 persons [23,24]. These figures are clearly inconsistent with animal records, as provided by the joint European Centre for Disease prevention and Control (ECDC) and European Food Safety Authority (EFSA) report on Zoonoses in the European Union for 2021 [24], where seroprevalence rates of 20.9% in sheep, 17.9% in goats, and 15.1% in cattle were provided. Therefore, when pooled estimates are compared to official

Discussion
In our systematic review and meta-analysis, a pooled prevalence of seropositivity for C. burnetii on occupationally exposed individuals was estimated at 44.0% (95%CI 27.6 to 61.8). The risk for seropositivity was greatest among veterinary professionals, followed by abattoir workers, livestock farmers, agricultural workers, geologists/agronomists, and forestry workers, while laboratory professionals had no substantially increased risk, and forestry rangers exhibited a conversely reduced seropositivity risk.
In other words, the present study is consistent with some previous reports, suggesting that seroprevalence for C. burnetii among professionals working with animals or involved in meat processing may be up to 50% [1,6,7,10,16,19,[38][39][40], largely exceeding estimates in non-occupationally exposed individuals from the index areas (where prevalence estimates ranging between 13.6% and 14.3% were retrieved) [13,25], and national figures for nonoccupational exposed individuals [6,19,37,41,42]. In this regard, it should be stressed that according to official reports, between 2015 and 2021, a total of 17 cases of QF have been reported to Italian health authorities, for an annual reference rate well below 0.1 cases per 100,000 persons [23,24]. These figures are clearly inconsistent with animal records, as provided by the joint European Centre for Disease prevention and Control (ECDC) and European Food Safety Authority (EFSA) report on Zoonoses in the European Union for 2021 [24], where seroprevalence rates of 20.9% in sheep, 17.9% in goats, and 15.1% in cattle were provided. Therefore, when pooled estimates are compared to official figures, QF in Italy appears as substantially underestimated, with notification rates reasonably representing the tip of a very larger burden of disease, particularly in occupational settings. Occupational physicians, i.e., the medical professionals responsible for health promotion in the workplaces [43], should therefore be made aware of the actual transmission of C. burnetii in occupational settings, promoting appropriate surveillance programs and improving the adherence to proper preventive measures among at-risk workers.
In fact, not only do pooled estimates substantially mirror several international reports on this pathogen [1,6,7,10,16,19,38,39,41,44], but they are also quite consistent with some previous Italian reports, where the relatively high occurrence of this pathogen in various agricultural settings was highlighted [20,45]. For example, by 1992, 13.1% of 99 sampled farms, 4.4% of sampled animals, and 6.5% of raw milk samples from the Apennine region of Emilia Romagna were contaminated by C. burnetii [20]. Similarly, a more recent study from the Autonomous Province of Bolzano reported an overall seroprevalence of 13.6% for cattle, 11.7% for sheep, and 7.9% for goats [45]. Moreover, in 2003 a large outbreak of QF among prison inmates did occur in the Como area following the exposure to dust contaminated by a passing flock of sheep [27,28].
As a consequence, the status of QF in Italian agricultural and forestry workers would be quite similar to other zoonotic pathogens of occupational interest, e.g., Hantaviruses, Tickborne Encephalitis virus, and Borrelia burgdoferi (the causal agent of Lyme disease) [46][47][48][49][50]. More precisely, our analyses stress the strong association of seroprevalence for C. burnetii with the meat industry (more precisely, for abattoir personnel) and farm tasks that cause the interaction between men and animals, i.e., animal breeders, but also with veterinarians [1,14,23]. On the contrary, even though C. burnetii may be acquired as a tick-borne disease [13,18], the relatively low seroprevalence in forestry rangers was not unexpected, for several reasons. For one, despite their role in maintaining the wildlife reservoir, ticks are only marginal agents of Q fever in humans. As a consequence, a marginal transmission in these specific settings was largely foreseen. Moreover, the estimates on forestry rangers were mostly obtained through the study by Cinco et al. [36], whose results should be carefully assessed as obtained through the complement fixation test, whose sensitivity and specificity are quite lower than those associated with IFA and ELISA.
Limitations. Despite the potential interest, our study is affected by several limitations. First of all, the studies that were retrieved and included in the analyses were mostly of limited quality, as summarized by the ROB tools. In this regard, two main shortcomings should be stressed. On the one hand, nearly all studies did not report how the sample was eventually recruited. In other words, the actual representativity of the assessed occupational groups, even in the targeted areas, remains unclear. On the other hand, as previously stressed for other infectious diseases affecting agricultural workers [46,47,51], the job description could fail to appreciate other exposures, possibly associated with residential and environmental factors, as agricultural settings hardly dichotomize occupational and residential environments [46,47,52,53]. Moreover, the studies were also quite heterogenous in geographical terms, sample size, and sampling strategy [13,18,25,37]. As a consequence, retrieved estimates could only be limitedly comparable.
Third, as the collected studies were performed across a very broad timespan (2000 to 2022) and laboratory techniques were quite heterogeneous (i.e., IFA, CFT, and ELISA), we cannot rule out whether differences in estimates are the result of the specificities of the diagnostic procedures or the actual seroprevalence [2]. Fourth, the pooled population was quite small, both in general and when dealing with individual occupational groups, which is not representative of Italian workforce [14].

Conclusions
In conclusion, further studies are required in order to better understand the actual occurrence of C. burnetii infections in Italian, not only in occupational settings, but also in the general population. Even though the large majority of cases clearly occur either asymptomatically or pauci-symptomatically [2,16,43,45], more accurately tailored interventions for at-risk workers should be implemented in terms of surveillance programs and informative campaigns on the appropriate preventive measures.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A
Zoonotic Dis. 2023, 3, FOR PEER REVIEW 12 Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: Raw data are available from the Study Authors.

Conflicts of Interest:
The authors declare no conflict of interest.