Health education decreases incidence of hand eczema in metal work apprentices: Results of a controlled intervention study

Metal work apprentices (MWAs) frequently develop work‐related hand eczema (HE).

HE in MWAs were smoking cigarettes (P < .01) and FLG mutations (P < .001). No significant associations were found regarding epidermal cytokine levels and GWAS.

Conclusions: Health education is effective in primary prevention of HE in MWAs.
Individual factors should be considered in targeted counseling.

| INTRODUCTION
Hand eczema (HE) is the most common work-related skin disease. 1 Due to its often chronic course, it is associated with impaired healthrelated quality of life and a considerable socioeconomic burden related mainly to work absenteeism, loss-of-job, or retraining. [2][3][4] Metal workers have a particularly high share of work-related HE, with a lifetime prevalence of 30% to 63.1%. [5][6][7] In metal processing, this is due mainly to mechanical strain, wet work, and exposure to metalworking fluids with substantial irritant and allergenic potential. The onset of work-related HE occurs frequently already during apprenticeship. 7 However, data on the incidence of HE in metal work apprentices (MWAs) are scarce. [8][9][10] Considering the low adherence to skin protection recommendations among metalworkers, 5 raising awareness and health education on use of personal protective equipment is required.
Although there is scant evidence for the effectiveness of health education in primary prevention of work-related HE due to a lack of standardized trials, 2,11 a few controlled prospective studies in apprentices of other professions suggest that health education effectively reduces the incidence of HE by improving skin protection behavior. [12][13][14] To our knowledge, with the exception of a single study assessing any skin changes in apprentices from various occupations including a small number of MWAs, 15 no such studies have been performed in MWAs.
Individual factors may modulate the risk of work-related HE and interfere with preventive measures. 16 Loss-of-function mutations in the filaggrin gene (FLG) cause skin barrier defects by reducing levels of epidermal filaggrin and its degradation products, which contribute to the natural moisturizing factor (NMF). FLG mutations are present in 8% to 10% of the general European population and have been linked to atopic dermatitis (AD). 17 Carriers have an increased risk of work-related irritant contact dermatitis, particularly when in combination with AD, 16,[18][19][20] a more unfavorable course of the disease, and a lower probability of remaining in the workforce. 21 Genome-wide association studies (GWAS) may help to discover other genetic variants linked to workrelated HE. They have already been applied in AD and identified susceptibility loci, mostly implicated in immune dysregulation. 22 Moreover, phenotypic variations in epidermal cytokine levels may explain individual predispositions toward skin inflammation and thus susceptibility to develop work-related HE. 23 Data on the association between tobacco smoking and prevalence of HE have been inconsistent, also in occupational settings. 24 However, recent studies suggest that work-related HE is more severe and persistent in smokers than in nonsmokers. 25,26 Here, a controlled prospective health education intervention study was conducted in a large cohort of MWAs and control cohorts.
The primary outcome was the effect of the intervention on the incidence of HE during apprenticeship. Secondary outcomes were the effect of the intervention on the disease-specific knowledge and protective behavior as well as associations between the incidence of HE and individual factors.

| Study population and study design
We conducted a controlled, prospective intervention study in MWAs from nine vocational schools (one in Osnabrück and eight in the city and catchment area of Göttingen, Germany) involving apprentices who were receiving training in a range of professions involving metal processing. An additional control cohort of office work apprentices from the same schools was recruited to assess the prevalence and incidence of HE in young adults of similar age, and ethnic and social background without exposure to major occupational skin hazards. The study protocol was independently approved by the ethics committees at the University of Osnabrück and the University Medical Center Göttingen. After obtaining permission from the regional school authorities and the vocational schools, the apprentices were recruited on a voluntary basis within the first weeks of apprenticeship in autumn of 2013 and 2014. All participating apprentices provided written informed consent. In minors (<18 years of age), informed written consent was given by their legal representatives. Assessments were carried out at baseline (T0) and at the end of the first (T1), the second (T2), and the third (T3) year of training (May/June of each respective year).

| Intervention
The MWAs from the vocational school in Osnabrück were assigned to the metalwork intervention group (MW-int). In the first weeks of apprenticeship, they received a structured 90-minute training on causes and prevention of work-related HE consisting of an oral, interactive, and dialogue-oriented presentation by an experienced health educator, combined with hands-on practical exercises on correct use of protective gloves, barrier creams, and skin care products. This training was based on seminars regularly conducted by the University of Osnabrück for patients with work-related HE and previous studies 27,28 with adjustments for MWAs. 29 The maximum number of MWAs receiving the training at a time was 25. One year after enrollment (T1), a leaflet summarizing the key information of the training program and free samples of an unscented barrier cream and emollient were handed out to the participants in the MW-int. The MWAs from the other vocational schools were assigned to the metalwork control group (MW-ctrl) and did not receive training on prevention of HE by us, similar to the apprentices in the office work control group (OW-ctrl).

| Questionnaires
At baseline, the participants filled out a paper-based questionnaire to provide basic sociodemographic data and information on preexisting

| Clinical examination and atopy score
At each time point, a trained dermatologist used the validated Osnabrueck Hand Eczema Severity Index (OHSI) 31,32 to assess the apprentices' hands for presence of six morphological features (erythema, scaling, papules, vesicles, infiltration, and fissures) and the extent of the affected skin area. An Erlangen atopy score ≥10 at baseline was defined as atopic skin diathesis. 33  Data-Analysis software. Sample and marker quality control was performed with PLINK (v1.9, https://www.cog-genomics.org/plink/1. 9). 34 Individuals with a person-wise call rate <92%, with excess of heterozygosity (inbreeding coefficient |F| > .1), duplicated and related samples (PI HAT > .1875), as well as outliers from principal component analysis were excluded. Variants with a call rate <95%, Hardy-Weinberg disequilibrium (P < 10 −8 ), minor allele frequency <1%, and a differential missingness between cases and controls (P < 10 −50 ) were also excluded. After quality control, 477 263 variants and 328 individuals were eligible for analysis.

| Sequential tape stripping of stratum corneum and analysis
Eight round adhesive tape discs (3.8 cm 2 , D-Squame, CuDerm, Dallas, Texas) were consecutively attached to the skin at the flexural aspect of the lower arm. 35 Each tape was pressed on for 10 seconds with standardized force using a disc pressure applicator (CuDerm).
NMF was defined as the sum of the concentrations of histidine, 2-pyrrolidone-5-carboxylic acid, and trans-and cis-isomers of urocanic acid. NMF was extracted from tape strips number 6 and subsequently analyzed by high-performance liquid chromatography (HPLC) as published before. 35,36 The samples for cytokine analyses were extracted from tape strips number 8 by 0.6 mL of phosphate-buffered saline (PBS) con-

| Data analysis
Statistical analyses were conducted with R (3.5.1) (www.r-project.org, RRID:SCR_001905) using the car and ez packages. Differences between groups for continuous variables were tested either by the Welch test or analysis of variance (ANOVA) with White adjustment.
The Wilcoxon test was applied for categorical variables. In case of comparison of repeated measures, the Wilcoxon signed-rank test for categorical variables, and for dichotomous variables the McNemar test was used. Odds were analyzed by logistic regression using the generalized linear model procedure. As specified in the text, variables such as age, gender, or smoking behavior were either included in the regression equation or were used to check the effect of confounding variables in the analysis for effect control. Differences were considered significant if the P-value obtained was less than .05. For GWAS, allele frequencies were compared between different groups (MWAint, MWA-ctrl, OWA-ctrl) using logistic regression models with adjustment for sex and population stratification. All analyses were conducted with PLINK (v1.9, https://www.cog-genomics.org/plink/1.9). 34 3 | RESULTS

| Demographic data
The 303 MWAs enrolled in this study received vocational training as industrial mechanic, cutting machine operator/precision mechanic, model builder/construction mechanic, toolmaker, or process mechanic (Table S1).
While 131 MWAs were allocated to the intervention group (MW-int), the others (n = 172) were assigned to the metalwork control group (MW-ctrl).
The office work control group (OW-ctrl) consisted of 118 apprentices in vocational programs for information technology or bank professions. The sociodemographic characteristics of the two groups of MWA did not differ in main aspects at baseline ( Table 1). The apprentices in OW-ctrl had a slightly higher mean age of nearly 21 years, a higher share of females (14.4%), and on average a higher educational level than the MWAs in both groups. About one third of MWAs in both groups were self-reported tobacco smokers, with a significantly higher daily cigarette consumption in smokers of the MW-ctrl compared to the MW-int (P < .01).

| Hand eczema
HE was defined as the presence of (a) vesicles or (b) erythema in combination with at least one of the symptoms papules, scaling, or fissures on the hands, either observed during clinical examination (T0-T3) or selfreported for the time before baseline (T0) or within the year before the respective follow-up visits (T1-T3). Incident HE was defined as HE newly developed after recruitment, that is, in those without HE according to the above criteria during baseline examination and who denied having had HE prior to T0. Data of self-reported HE were included in the analysis. However, most HE was detected by clinical examination, in particular at T2 and T3 (data not shown). As shown in   (Figure 3). There were no significant differences in the 1-year, 2-year or 3-year incidence or prevalence of HE between MW-int and OW-ctrl over the years.

| Exposure to skin hazards and skin protection behavior
The percentage of apprentices exposed to metalworking fluids and/or oil as well as the number of days per week in contact with metalworking fluids and/or oil was similar in the two groups of MWAs at baseline and at all follow-ups (Table S2). However, the number of individuals who reported wet work for at least 2 hours a day was higher in MW-ctrl at follow-up than in MW-int. This difference was significant at T3 (37.8% vs 19.1%, P < .05). The rate had increased significantly from T0 (19.6%) to T3 (37.8%) in MW-ctrl (P = .004) and was nearly unchanged in MWint. Similarly, a daily hand washing frequency of ≥5 was more common in MW-ctrl than in MW-int at follow-ups, particularly pronounced at T3 (79.6% vs 68.1%), although not statistically significant. This is possibly related to a higher use of protective gloves in the MW-int compared to the MW-ctrl (T3: 93% vs 84%). In line with this, abrasive detergents were more often used in MW-ctrl than in MW-int (difference not significant). The self-reported use of skin barrier cream and emollients was similar in both groups at all time points and nearly unchanged over time.

| Knowledge on occupational skin diseases and skin protection measures
The mean score (number of correct answers) of the OSD-KQ-short is given in Table S2 (maximum possible score: 30). At all follow-ups, the mean score was significantly higher in MW-int than in MW-ctrl Apprentices with a history of flexural eczema were more likely to have an FLG mutation than those without (12.5% vs 6.2%, n.s.). Epidermal levels of NMF were significantly lower in individuals with FLG mutation (n = 18, 0.583 ± 0.188 nmol/μg protein) than in those without (n = 287, 0.851 ± 0.268 nmol/μg protein, P < .00001).

| Factors associated with incident hand eczema
The number of apprentices with incident HE during vocational training (cases) was much higher in the MW-ctrl (n = 53) than in the MWint (n = 16) and the OW-ctrl (n = 14). In a univariate analysis, MWAs with incident HE at any follow-up (n = 69) were compared with MWAs without incident HE (n = 234) ( low score in the knowledge test and not using skin barrier creams at T2 were additional risk factors for incident HE. A lower age, the selfreported amount of cigarettes smoked, and extent of exposure to metalworking fluids were not independent risk factors.

| DISCUSSION
To our knowledge, this is the first controlled prospective intervention study encompassing the entire period of apprenticeship that uses structured health education to prevent work-related HE in MWAs.
The results indicate that the intervention effectively improved the disease-specific knowledge and reduced the incidence and prevalence of HE during apprenticeship. Factors significantly associated with developing HE in MWAs were smoking cigarettes and FLG mutations, irrespective of the intervention. Within the intervention group, not using skin barrier creams and lower knowledge scores were additionally associated with incident HE.
The incidence and prevalence of HE in the second and third years of apprenticeship in the metalwork intervention group (MW-int) were similar to the rates in the nonoccupationally exposed office work control group (OW-ctrl) and significantly lower than in the metalwork control group (MW-ctrl), indicating a positive effect of the intervention. Considering the high incidence rates of HE in the MW-ctrl compared to the other two groups and the exclusion of preexisting HE, it is likely that most of them were work-related. Berndt  Abbreviations: OSD-KQ, occupational skin diseases knowledge questionnaire; SD, standard deviation. *P < .05.; **P < .01.; a Erlangen atopy score ≥10. b According to the German school grading system (1 = very good, 2 = good, 3 = satisfactory, 4 = sufficient, 5 = poor, 6 = very poor).
prevalence of about 10% reported for young people from the general population in Denmark and Sweden. 39,40 In contrast to the increasing prevalence of HE in the MW-ctrl group, this rate was nearly unchanged in the following years in the MW-int and in the OW-ctrl groups.
Our educational intervention led to improvements of knowledge regarding prevention of work-related HE as seen in other studies. 30,41 As expected, the knowledge gain was faster and higher in the inter- This was probably due to the comparatively small study size. Another possible explanation is that exposure-related factors have outweighed endogenous effects.
Around 30% of the MWA were self-reported tobacco smokers.
This is in line with current data demonstrating that among the 18-to 25-year-old young adults in Germany, 22.5% of women and 29.4% of men are self-reported smokers. 47 Apart from male gender, blue-collar work and a lower level of education are associated with a higher prevalence of tobacco smoking. 48 Accordingly, the rate of self-reported tobacco smokers was higher among MWAs than among office work apprentices, who had on average a higher educational level and a higher share of females. Data on associations between tobacco smoking and work-related HE are contradictory. 24 In contrast to Berndt et al, 8 we found a significant association between tobacco smoking and incident HE in MWAs and similarly in office work apprentices. This effect was independent from the daily number of cigarettes smoked. From previous studies we know that work-related HE is more severe and persistent in smokers than in nonsmokers and that tobacco smoking is associated with work absenteeism and leaving the workforce. 25,26 Tobacco smoking is considered to induce proinflammatory mechanisms and tissue damage in the skin, 49  Although this is a prospective, controlled intervention study in a real-world setting with regular follow-ups covering the whole apprenticeship, a limitation of this study was the randomization on school level and the lack of blinding for practical reasons. However, the two groups of MWAs were considered appropriate for comparison, as they did not differ significantly with respect to basic demographic data at baseline and occupational exposure to skin hazards, such as metalworking fluids. Occurrence of HE was detected mainly by clinical examination, but was also assessed by self-administered questionnaires. Self-reporting may be less accurate, but it allowed retrieval of information on the presence of HE before recruitment and inbetween school visits.
In conclusion, the intervention based on health education was effective in primary prevention of HE in MWAs, suggesting that its implementation in regular vocational training may reduce the burden of the disease. Moreover, tobacco smoking and FLG mutations should be considered as individual risk factors and addressed in moretargeted approaches.