Epidemiology of spider mite sensitivity: a meta-analysis and systematic review

Background Spider mites, including Tetranychus urticae, Panonychus citri, and Panonychus ulmi, are common pests in gardens, greenhouses, and orchards. Exposure, particularly occupational exposure, to these organisms may lead to the development of respiratory or contact allergies. However, the prevalence of sensitivity to spider mites is unclear. Methods We examined the literature to generate an estimate of the global prevalence of allergies to spider mites. Results Electronic databases were searched and twenty-three studies reporting the prevalence of sensitivity to spider mites (based on skin prick tests or IgE-based detection systems) in an aggregate total of 40,908 subjects were selected for analysis. The estimated overall rate of spider mite sensitivity was 22.9% (95% CI 19–26.8%). Heterogeneity was high and meta-regression analysis considering variables such as published year, country, number of study subjects, methods for allergen detection (skin prick test, ImmunoCAP, RAST testing, or intradermal test), and mite species revealed no single significant source. Twelve of the 23 studies reported rates of monosensitization (i.e., patients responsive to spider mites but no other tested allergen), yielding a global average of 7% (95% CI 5–9%), hence spider mites represent a unique source of allergens. Conclusions Spider mites are an important cause of allergic symptoms. However, the publication bias and heterogeneity evident in this study indicate that further trials using standardized detection methods are needed to determine the association of exposure and symptoms as well as the specific patient characteristics that influence developing spider mite sensitivity.


Background
The allergenic role of mites of the genus Dermatophagoides in indoor floor and mattress dust was discovered in 1967 [1,2]. Since then, numerous species have been described as the source of allergens capable of sensitizing and inducing allergic symptoms in susceptible and genetically predisposed individuals [3]. The major mites in indoor house dust, D. pteronyssinus, D. farinae, Blomia tropicalis, and Euroglyphus maynei, account for 80% of the total allergenic species, with storage mites making up from those generated by indoor mites [9]. The aim of our present study was to analyze existing information on the prevalence of spider mite sensitization.

Search strategy
We have used a search and analysis strategy based on the PRISMA system [10]. To identify related studies published through June 1st, 2017, we performed systematic literature searches of electronic databases including Pub-Med, the Cochrane Library, EMBASE, Medion, and Web of Science. Search terms were applied by various combinations of Medical Subject Headings (MeSH) and non-MeSH terms as follows: [(spider mite or Tetranychus or Panonychus) AND (sensitization or allergy or hypersensitivity or specific IgE positive or skin test positive or RAST positive)]. Titles and abstracts identified by electronic searches were examined independently and on screen by two researchers to select potentially relevant studies. Eligibility criteria are given below. Differences were resolved by consensus. A full text paper was obtained wherever possible.

Eligibility criteria
Studies that investigated the prevalence of sensitivity to spider mites (family Tetranychidae) in full journal articles were selected for review, including cross-sectional, cohort studies, controlled clinical trials and other types. Studies published in conference proceedings, books, book chapters, or research not published in English were excluded.
Eligible studies focused on individuals with allergic disorders defined by in vivo or in vitro tests with mite extract made from Tetranychus or Panonychus mites. Thus, inclusion into the meta-analysis was restricted to those studies that reported prevalence data for sensitivity to spider mites.

Data extraction
The following specific information relating to data collection and results was extracted individually from each identified article and entered into a pre-designed Excel spread sheet: data and geographical location, study design, participant inclusion and exclusion criteria, recruitment procedures, number of investigated subjects, age and gender of investigated subjects, occupations or characteristics of the patients, number sensitized to spider mites, detection methods, and mite species. To ensure accuracy, two researchers extracted the data and then compared the results of their extractions.

Meta-analysis according to the studied population groups
For meta-analysis, the prevalence rates of spider mite sensitization were pooled using the random effects model [11]. Heterogeneity was calculated via Cochran's Q and τ 2 tests, and inconsistency is presented as I 2 , which describes the percentage of variability that is due to heterogeneity rather than chance [11].

Meta-regression analysis
To identify the sources of heterogeneity among studies, meta-regression analysis was carried out [12]. Possible sources of heterogeneity, including published year, country, number of study subjects, methods for allergen detection (skin prick test, ImmunoCAP, RAST testing, or intradermal test), and mite species (Tetranychus urticae, Panonychus ulmi, or Panonychus citri), were included in the analysis.

Publication bias and meta-analysis
The possibility of publication bias was assessed by graphical analysis of funnel plots. Deeks' funnel plot asymmetry analysis was performed to identify publication bias [13]. In Deeks' funnel plots, each data point represents a study, its effect size or prevalence, and the standard error. The meta-analysis was conducted using the Stata v12 software package (Stata Corporation, College Station, TX, USA) and the graphical representation was conducted using forest plots.

Characteristics of included studies
Our searches initially retrieved 48 journal article references from electronic databases. Twenty-four of these were subsequently removed due to either duplication or a failure to meet the inclusion criteria. The remaining twenty-four full text articles were then retrieved and critically appraised [8,9,. Of these, the Gargano study [30] was subsequently deleted from the analysis, because this study selected only patients that were SPT+ and tested them to see what percentage had spider mite reactive IgEs. This does not represent an unbiased patient population (since all patients were known to be SPT+). The remaining 23 studies were found to be eligible and were entered into our review and meta-analysis (Fig. 1). Among the 23 included papers, 13 were conducted in Korea, three were conducted in Italy, one was conducted in Japan, two were conducted in Spain, two were conducted in South Africa, and two were conducted in Sweden ( Table 1). The sample sizes of the studies entered into the review varied widely from 10 [33] to 8595 [22] with the median sample size being 308. In total, the 23 studies examined 40,908 subjects. Among these 23 papers, Kim et al. [22] reported the prevalence for sensitivity to both T. urticae and P. citri using separate patient populations. Kim and Lee et al. [25] reported the sensitivity prevalence for both T. urticae and P. ulmi in the same patient population. For the purposes of meta-analysis, the different mite species were considered separately. Hence, these two studies contributed to two data points.

Publication bias, sensitivity, and meta-regression analysis
Deeks' funnel plot (Fig. 3) was applied to assess publication bias. In Fig. 3, which shows the prevalence among the cases, the prevalence of the analyzed studies is presented on the x-axis and the standard error of each study is shown on the y-axis. Visual evaluation revealed that the plot was an asymmetric funnel shape, indicating that publication bias was likely present. and upper (0.27) 95% confidence values to the left and right, respectively. Each circle represents the new mean obtained when the indicated study is removed from the pool. These means all fell within the 95% confidence interval of the total data set, indicating that no individual study had a disproportionate effect on the mean.
As displayed by the forest plot in Fig. 2 and in Table 2, the heterogeneity was significant for T. urticae (I 2 = 99.4%) and P. citri (I 2 = 99.3%). One possible source of heterogeneity was the study population. Eight studies [15,20,24,29,[31][32][33][34] enrolled only symptomatic patients (i.e., patients with airway allergy symptoms including asthma and rhinitis or patients with dermatitis) whereas the remaining studies enrolled a mixture of  symptomatic and asymptomatic patients. If all symptomonly studies are removed from the sensitivity analysis, the estimated prevalence dropped to 15.43% outside the 95% confidence interval for the total data set (data not shown). From this, we conclude that these studies inflated the mean. However, it is difficult to conclude whether this is due to the patient populations or some other factor. The symptomatic studies typically enrolled fewer patients, so study size might have had an influence. Additionally, when subgroup analysis was performed, the heterogeneity of both the symptomatic and mixed studies was still extremely high (Table 3), indicating that patient populations alone did not contribute much to the overall heterogeneity of the included studies. To examine other sources of heterogeneity, a meta-regression analysis considering the publication year, country, number of study subjects, methods, and mite species analysis was performed, and the results showed that no single analyzed factor could account for the large variability in the reported prevalences. It is likely that a combination of factors makes these studies extremely diverse.

Discussion
This review provides the first comprehensive search and synthesis of the international literature on the prevalence of spider mite sensitization. The result of our synthesis of all prevalence estimates was 22.9% (95% CI 19.0-26.8%) but may be higher when only symptomatic patients are considered [43.9% (95% CI 35.1-52.9%)]. Our pooled estimate indicates that spider mite sensitivity is moderately common in farming populations. Mite subgroup prevalence estimates were 27% (95% CI 20.5-33.5%) for T. urticae sensitivity and 18.2% (95% CI 12.4-24.0%) for P. citri sensitivity. Only one paper reported the prevalence of P. ulmi sensitivity. Therefore, agricultural workers dealing with fruit trees or working in greenhouses as well as in the surrounding rural population are at risk for developing sensitivity to T. urticae and P. citri. Further studies are needed to confirm the prevalence of P. ulmi   sensitivity and to examine if sensitivity to spider mites is a cause of occupational allergies and/or general allergies in rural populations. The overall sensitivity estimates include patients reactive to spider mite species who may also be sensitized to other environmental allergens. For such polysensitized individuals, a positive skin reaction to spider mites could indicate a primary allergic response or cross-reactivity. To address this, some studies reported the rates of monosensitization (defined as reactivity to spider mites but no other tested allergen). Our pooled prevalence estimate for monosensitization to spider mites was 7% (95% CI 5-9.0%), and subgroup prevalence estimates were 7% (95% CI 5-10%) for T. urticae sensitivity and 7% (95% CI 1-12.0%) for P. citri sensitivity. This indicates that spider mites are the primary sensitizing agent for a moderate number of individuals living primarily in rural settings. Jee et al. used competitive ELISAs and found that D. pteronyssinus extracts could not compete with IgE binding to T. urticae proteins in serum from a mono-sensitized patient but could compete in serum from polysensitized patients. Unfortunately, little progress has been made in identifying spider mite-specific antigens. Studies have used SDS-PAGE and IgE-immunoblotting to identify 20 [31], 24 [27] and 10 [36] IgE-reactive bands in spider mite extracts, but which of these components are species specific has yet to be determined. Additionally, it should be noted that patients sensitized to other allergens (including domestic mites and/or non-taxonomically related species) are more likely to also be reactive to spider mites [9,16,19,25,35]. This generalized atopy is known to be true for a variety of high molecular weight allergens and is believed to indicate a hyper-reactive IgE response in certain sensitive patients [37].
The authors believe that the searches conducted were comprehensive and the sensitivity analysis demonstrates that the calculated mean was not unduly influenced by a single study, and thus our findings are generally robust. However, publication bias is present based on the asymmetric funnel plot, and the heterogeneity of the studies was quite large. The heterogeneity observed could come from the different study settings and populations. The estimated prevalence of spider mite sensitization in symptomatic patients was 43.9% (95% CI 35.1-52.9%) which was 2.9 times higher than that found in mixed populations [15.4% (95% CI 11.6-19.7%)]. Heterogeneity was, however, still very high within the subgroups, hence these results should be interpreted cautiously. Some studies reporting data from mixed populations did detect an association between spider mite reactivity and symptoms, but others did not. For example, using patient subgroup information published by Kim and Son et al. [9] revealed that, in this study, the prevalence of spider mite allergies in symptomatic patients was 4 times higher than that in asymptomatic patients. However, in Lee et al. [19], prevalence estimates were similar in symptomatic and non-symptomatic patients. Of note, several studies tested the onset of symptoms in response to a T. urticae challenge. Astarita et al. [34] examined the onset of allergic symptoms and tracked the peak expiratory flow rate in spider mite-sensitive patients exposed to an infested green-house environment, and two studies [29,31] performed a bronchial challenge with T. urticae extracts and observed responses in the majority of T. urticae-sensitive patients. This indicates that spider mite sensitivity has clinical relevance, but this may vary based on the location and population being considered.
We investigated other possible sources of heterogeneity with meta-regression analyses but could not identify a single factor responsible for the variation. Two factors that may be relevant based on individual studies are patient age and site of residence. Kim et al. [8] reported that the sensitization rate to T. urticae increased with age, and Kim et al. [22] reported that the prevalence of spider mite allergies in rural areas was higher than the prevalence in urban settings. In regards to age, few studies of T. urticae sensitivity included children (Table 1), whereas the majority of the P. citri studies only enrolled children. This could account for the lower prevalence of sensitivity in the P. citri studies, or it could indicate that P. citri is a weaker sensitizing agent.

Conclusions
In brief, spider mites are important sensitizing agents particularly in farming populations where contact is the most likely. In some of the reviewed studies, the prevalence of spider mite sensitivity was reported to be higher in patients with allergic symptoms (particularly occupational allergies), and thus exposure may correlate with disease. The moderate prevalence of spider mite monosensitization indicates that these organisms produce unique allergens, and thus specific diagnostic tests and treatment regimens for spider mite sensitization are likely warranted. These conclusions should, however, be interpreted cautiously. Publication bias was present, the heterogeneity of the analyzed studies was extremely high, and the sources contributing to this heterogeneity were unclear. Additional cross-sectional studies using more standardized protocols are needed to assess how specific patient characteristics influence the acquisition of spider mite sensitization and whether and how this progresses to allergic disease.