Methods

Individually randomised controlled trial

Combination of parallel and split‐body design, 2x2 arms

Duration: 12 months

3 follow‐ups at 3, 6, and 12 months

Participants

Final number evaluable after 12 months: N = 96 healthy metal workers exposed to cutting fluids

• Group 1 (N = 46): skin protection/no product on randomly allocated hand

• Group 2 (N = 50): skin protection + skin care/skin care only on randomly allocated hand

Number of participants randomised: N = 100 (group 1: N = 50; group 2: N = 50)

Lost to follow‐up: 4 in group 1

Mean age in years (participants evaluable after 12 months): group 1: 35.5 (range 17 to 59); group 2: 41.5 (range 17 to 59)

Sex: male

Inclusion criteria

• Male

• 17‐65 years

• Exposure to cutting fluids

• Forced to mostly work without hand gloves

• Written consent

Exclusion criteria

• Female

• Fitzpatrick skin type IV, V , VI

• Hand eczema during recruitment

• Local therapy of the hands

• Intake of cortisone or immunosuppressives 3 weeks before recruitment

• Absence from factory for 3 months or more

Setting: one medium‐sized German factory

Interventions

Comparison of the effectiveness of skin protection, skin care, or both, versus no intervention (4 study arms, all relevant to the review). The metal workers' hands were randomised to:

• Travabon or Stoko Protect (skin protection / barrier cream);

• Estolan (skin care/moisturiser);

• both;

• no intervention.

No details on application described, but participants received a flyer and the product. One‐sided application using a glove for the other hand. Participants were randomised to use product on either left or right hand.

Outcomes

1. Proportion of OIHD: after 12 months, abnormal morphology ('klinischer Hautbefund') was assessed by study physician. Conspicuous and minor conspicuous findings were extracted as signs and symptoms of OIHD.

2. Treatment discontinuation due to adverse effects: all 4 withdrawals were due to personal reasons. We deduce that there were no cases of treatment discontinuation due to adverse effects.

3. TEWL and skin hydration were extracted (only quartiles, no mean or SD given)

4. Microtopographic parameters (not extracted for this review)

All outcomes were measured at baseline, after 3, 6, and 12 months. For the visits at 3 and 6 months, abnormal morphology was not reported while the other outcomes were reported in diagrams of the medians only.

Diagnostic criteria for OIHD (degree of certainty for the diagnosis of OIHD)

Strengths

• Clinical examination by dermatologists was performed. Good method to assess prevalence of hand eczema.

Limitations

• No exclusion of endogenous/atopic hand eczema or allergic contact dermatitis of the hands (no patch test performed)

• No exclusion of non‐occupational hand eczema

Comparability

• Broad outcome definition ‐> comparatively many cases expected;

• Time frame: point prevalence

Scoring system for the severity of OIHD: the study investigators applied a score which consisted of 3 categories: 'unauffällige klinische Befundung' (inconspicuous findings), 'geringfügig auffällige klinische Befundung' (minor conspicuous findings) and 'auffällige klinische Befundung' (conspicuous findings).

Strengths

• Blinded: the dermatologist did not know which group the participant belonged to

• Double assessment: a second dermatologist independently assessed a photograph of the skin

• Already minor symptoms were noted

Limitations

• Score not validated for inter‐ or intra‐observer reliability

• Not based on different symptoms or extent of affected area (objective parameters)

• Very subjective

Quality of bioengineering methods (Pinnagoda 1990): the measurements were performed on four defined skin areas (back of the left and right hand, distal part of the volar site of the lower left and right arm) after a 20‐minute adjustment phase at a standardised temperature (20°C) and relative humidity (50%). The participants were asked not to take a shower, eat or drink anything and to refrain from smoking within the hour before the measurement. TEWL was measured according to international recommendations (Pinnagoda 1990, Rogiers 2001). For the corneometry, electrical measurements of conductibility (κ), impedance (Ω) and capacity (F) were performed.

Strengths

• Information on the test room is given (temperatures, relative humidity)

• Measurements were taken in a standardized environment (temperature, relative humidity)

• Potentially interfering factors were excluded (shower, food, drinks, smoking)

• An adjustment phase of 20 minutes was allowed before measurements

• Method is well‐described

Limitations

• No information is given on the instruments

• Measurements were taken on the back of the hand and the distal part of the volar site of the lower arm. Rather vague information. Does not ensure that measurements were performed always in the exact same area.

Notes

The left/right hand design may have led to failures when applying the products.

Study funding sources: Vereinigung der Metall‐Berufsgenossenschaften (VMBG) (financial, content‐related and organisational support)

Possible conflicts of interest not described.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Method not described. Quote: 'Während der Erstuntersuchung wurden die Probanden zufällig in zwei Gruppen aufgeteilt. Danach erfolgte in jeder Probandengruppe die zufällige Aufteilung der Probanden in jeweils zwei Interventionsarme (jeweils einer pro Hand).' (p. 7) [At baseline, probands were randomly assigned to two groups. Subsequently in each group of probands, the probands were randomly assigned to two intervention arms each (one for each hand).]

Allocation concealment (selection bias)

Unclear risk

Block randomisation was performed, no details given. Unclear selection bias.

Blinding of participants and personnel (performance bias)
All outcomes

Unclear risk

No blinding of participants, participants could have influenced performance by intentionally applying products to the wrong hand. Possible influence on proportion of OIHD.

Blinding of outcome assessment (detection bias)
Subjective outcomes

Low risk

Physician was blinded.

Blinding of outcome assessment (detection bias)
TEWL and corneometry

Low risk

No influence of blinding is expected for these objective measurements.

Incomplete outcome data (attrition bias)
Hand‐eczema‐related outcomes

Low risk

Only 4 withdrawals, due to personal reasons, in group 1 (control/skin protection).

Incomplete outcome data (attrition bias)
Consumer satisfaction

Unclear risk

The study did not address this outcome.

Selective reporting (reporting bias)

Unclear risk

No protocol available. All outcomes mentioned in the report were reported at least for first and last examination.

Other bias

Unclear risk

Design: more than one intervention arm. This was accounted for in meta‐analysis. The control group was not double‐counted within one meta‐analysis.

The split‐body design may have introduced contamination, especially concerning the control arm. (unclear risk)

Baseline imbalances

• Median age 35.5 in group 1 (skin protection, control) vs 41.5 in group 2 (skin protection + care, skin care)

• 2.2% left‐handed in group 1 vs 10% in group 2

• Slightly more smokers in group 2

• 13% with known HE in group 1 vs 24% in group 2

Blocked randomisation in unblinded trials: block randomisation was performed, no details given. Unclear selection bias.

Differential diagnostic activity: no different diagnostic activities across study arms.

Methods

Individually randomised controlled trial
Parallel groups, 2 arms

Duration: 1 year

3 follow‐ups at 4, 8, and 12 months

Participants

Final number evaluable after 12 months: N = 497 initially healthy dye and print factory workers (intervention: N = 248; control: N = 249)

Number of participants randomised: N = 868 (intervention: N = 428; control: N = 440)

Lost to follow‐up: N = 211 (intervention: N = 102; control: N = 109)

Excluded from review due to OIHD at beginning of study: N = 160 (intervention: N = 78; control: N = 82)

Mean age in years (all included participants): intervention: 32.6; control group: 32.1

Sex: 91% male, balanced between groups

Mean duration of employment (all 868 participants): barrier cream and control group, 7.9 years

Inclusion criteria: dye and print industry workers
Exclusion criteria: other dermatological disease on the hands

Setting: field study in 13 dye and print factories, North Italy

Interventions

Comparison of the effectiveness of:

• barrier creams 2 x/day (silicone or hydrocarbon containing formulations) versus;

• no intervention.

The workers were randomised to receive either the barrier creams (silicone or hydrocarbon containing barrier creams), which they used twice a day for 1 year, or no intervention.

Outcomes

1. Proportion of OIHD: positive objective exam for at least one follow‐up. Presence of the following skin changes: erythema, edema, exudation, vesicles, blisters, desquamation, hyperkeratosis, rhagades, dryness, atrophy, lichenisation. ('Positività all'esame obiettivo in almeno un controlo.' p. 235; 'In particolare l'esame obiettivo consisteva nel rilevare la presenza di lesioni dermatologiche a mani e avambracci distinguendo: eritema, edema, essudazione, vescicole, bolle, desquamazione, ipercheratosi, ragadi, secchezza, atrofia, lichenificazione.' p. 233)

Assessments took place after 4, 8, and 12 months.

Diagnostic criteria for OIHD (degree of certainty for the diagnosis of OIHD): objective skin lesions assessed by study personnel (erythema, edema, exudation, vesicles, blisters, desquamation, hyperkeratosis, rhagades, dryness, atrophy, lichenisation)

Strengths

• Clinical examination by physicians who were trained by a dermatologist was performed. Good method to assess prevalence of hand eczema.

Limitations

• No exclusion of endogenous/atopic hand eczema or allergic contact dermatitis of the hands (no patch test performed)

• No exclusion of non‐occupational hand eczema

Broad outcome definition ‐> comparatively many cases expected

Time frame: combined incidence at 4, 8, or 12 months

Scoring system for the severity of OIHD: no score assessed
Quality of bioengineering methods (Pinnagoda 1990): not relevant

Notes

Study funding sources: Not described.

Possible conflicts of interest: Not described.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: 'La procedura di randomizzazione.....base di una sequenza di numeri casuali' (p. 233) [The randomisation procedure ... based on a sequence of random numbers]

Matched pairs were used in order to achieve balanced groups.

Allocation concealment (selection bias)

Unclear risk

Inadequate concealment of the allocation sequence may have resulted from pairwise assigning the first person of the pair to group A if the random number was even, to group B if odd. The second person of the pair was assigned to the alternative group. Efforts to conceal the allocation were not described.

Blinding of participants and personnel (performance bias)
All outcomes

Unclear risk

Blinding of participants remains unclear.

Blinding of outcome assessment (detection bias)
Subjective outcomes

Low risk

Quote: 'Controllo clinici periodici ....all`oscuro del gruppo' [Periodical clinical examination ... blinded to group] ‐ Outcome assessors were blinded.

Blinding of outcome assessment (detection bias)
TEWL and corneometry

Low risk

TEWL and corneometry were not measured.

Incomplete outcome data (attrition bias)
Hand‐eczema‐related outcomes

Unclear risk

12 months follow‐up: 102 missing from the barrier cream group, 109 missing from the control group due to dropout and unavailability at the examination time points. The study investigators did mention ITT analysis but this merely meant that compliance was ignored. Missing values were not estimated.

Dropout was due to the long duration of recruitment and due to workers leaving the factories, but there was no detailed analysis.

Incomplete outcome data (attrition bias)
Consumer satisfaction

Unclear risk

The study did not address this outcome.

Selective reporting (reporting bias)

Unclear risk

Insufficient information.

Other bias

Low risk

Design: no sources of bias were identified.

Baseline imbalances: demographic data of participants were only available for the entire study population (N = 868). It remains unclear whether the demographic characteristics of the initially healthy workers from intervention and control group (N = 708) differed significantly in this study (low risk)

Blocked randomisation in unblinded trials: not enough information was given to decide whether or not allocation concealment was broken by the pairwise allocation. (unclear allocation concealment bias)

Differential diagnostic activity: no different diagnostic activities across study arms.

Methods

Cluster‐randomised controlled trial; 18 clusters (departments)

Parallel design, 2 arms

Duration: 1 year

1 follow‐up at 1 year

Participants

Final number evaluable after 12 months: N = 212 initially healthy Danish gut cleaners (intervention: N = 59; control: N = 153)

Number of participants randomised: N = 644 baseline responders (intervention: N = 205; control: N = 439)

Lost to follow‐up: N = 228 were lost to follow‐up or stopped working as gut cleaners (intervention: N = 69; control: N = 159)

Excluded from review due to OIHD at beginning of study: N = 204 (intervention: N = 77; control: N = 127)

Mean age in years (all follow‐up respondents including those with OIHD at baseline and those not available at baseline): intervention: 36.1 (range 17‐62 years); control: 37.8 (range 17‐66)

Sex: 66.3% male (intervention), 64.1% male (control)

Inclusion criteria: not described

Exclusion criteria: not described

Setting: field study in 18 swine slaughterhouses in different Danish cities, all departments belonging to one company

Interventions

Comparison of the effectiveness of a:

• prevention programme, versus

• comparison group (probably no intervention).

The prevention strategy consisted of a two part concept, with an evidence based prevention programme giving recommendations for prevention of work related skin problems in wet work occupations, and a documented method for implementation. The recommendations were aimed at the management and at the employees.

• Avoid or reduce wet and dirty manual working procedures.

• Use protective gloves for wet and dirty working procedures, if possible.

• Protective gloves must be intact, clean, and dry inside and must be worn on clean, dry, and well cared for skin.

• Use fabric gloves, e.g. cotton gloves underneath the protective gloves.

• Use a skin care product when needed during the working day and always after work.

• Use a skin care product before wet and dirty working procedures if you do not use protective gloves.

• The skin care product should have a high content of petrolatum and a low content of water.

• Do not wear finger rings, jewellery, or wristwatch on hands or forearms during work.

• Wash your hands in cool water, rinse off the soap thoroughly, and dry your hands carefully with a soft material afterwards.

• When there is no visible contamination of the hands, hand washing can with advantage be substituted by an alcohol based hand disinfectant.

• Protective gloves, hand soaps, skin care products, and hand disinfectants should be without known irritant and allergic substances or with lowest possible content of them (p. 643).

The local project group included 2‐5 gut cleaners who acted as role models.

Outcomes

1. Proportion of OIHD (hand eczema): The case definition for eczema was: eczema on hands or forearms within the past three months based on questions D1, D2, and D5 of NOSQ 2002 (telephone interview). Assessed at 12 months.

2. Use of gloves, use of cotton gloves underneath plastic/rubber gloves, and use of skin care products were not extracted for this review.

Diagnostic criteria for OIHD (degree of certainty for the diagnosis of OIHD)

Strengths

• The questions used to assess the prevalence of hand eczema belong to a validated questionnaire (NOSQ 2002).

Limitations

• Self‐reporting of hand eczema underestimates the true prevalence of hand eczema. Especially mild changes are often not reported.

• No differentiation between hand eczema and eczema on the forearms

• No exclusion of endogenous/atopic hand eczema or allergic contact dermatitis of the hands (no patch test performed)

• No exclusion of non‐occupational hand eczema

Comparability

• relatively narrow outcome definition

• time frame: period prevalence: last 3 months ‐> comparatively many cases expected

Scoring system for the severity of OIHD: no score assessed
Quality of bioengineering methods (Pinnagoda 1990): not relevant

Notes

79 participants had stopped working as gut cleaners and were analysed as a separate group in the report regardless of the intervention. For this review, they were not considered and treated as dropouts. Demographic data were only available for participants who were not lost to follow‐up, including those with OIHD at baseline.

There was a mistake in the paper concerning primary outcome 1, which MA Flyvholm corrected in an email (05/052015): 'For the intervention group (corrected: comparison group), 67% of those with eczema at baseline reported eczema at follow up and 33% no eczema; 37% with no eczema at baseline reported eczema at follow up.' (p. 646)

Study funding sources: 'The project was financially supported by an appropriation for prevention of asthma and allergy, administered by the Danish Ministry of Health.' (p. 648)

Possible conflicts of interest: none declared.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Method not described ('randomly allocated' p. 643).

Allocation concealment (selection bias)

Unclear risk

Method not described.

Blinding of participants and personnel (performance bias)
All outcomes

Unclear risk

No blinding, possible influence on proportion of OIHD.

Blinding of outcome assessment (detection bias)
Subjective outcomes

High risk

No blinding likely to influence reporting of OIHD.

Blinding of outcome assessment (detection bias)
TEWL and corneometry

Unclear risk

TEWL and corneometry were not measured.

Incomplete outcome data (attrition bias)
Hand‐eczema‐related outcomes

Unclear risk

149 participants (23,1%) lost to follow‐up, an unspecified number of which (max. 51) dropped out without stating reasons.

Incomplete outcome data (attrition bias)
Consumer satisfaction

Unclear risk

The study did not address this outcome.

Selective reporting (reporting bias)

Unclear risk

Insufficient information.

Other bias

High risk

Design: no correction for cluster randomisation ‐> no bias, but over‐precise results and limited comparability.

Baseline imbalances

• Significantly less OIHD (eczema on hands/forearms in the previous 3 months) in comparison group, study investigators explain this with the more frequent use of gloves in one comparison department ‐> high risk

• Those who stopped as gut cleaners had worked as such for a shorter time ‐> probably no bias

Blocked randomisation in unblinded trials: no block randomisation reported.

Differential diagnostic activity: no different diagnostic activities across study arms.

Methods

Individually randomised controlled trial
Parallel groups, 3 arms

Duration: 6 months (also reported as 30 weeks) for each participant, 3 years altogether

3‐weekly follow‐ups

Participants

Final number evaluable after 6 months: N = 54 initially healthy, newly employed metal workers exposed to cutting fluids (barrier cream: N = 17; moisturiser: N = 14; control: N = 23)

Number of participants randomised: N = 54

Lost to follow‐up: N = 0

Mean age in years: barrier cream group: 22 (range 16 to 37); moisturiser group: 22 (range 16 to 36); control: 22 (range 17 to 35)

Sex: 50 out of 54 male

Exclusion criteria: 'Only machinists who had not handled cutting fluids previously were included in the study. Machinists who had already worked for more than I week were excluded.' (p. 177)

Setting: field study in the grinding and turning sections of one large ball‐bearing manufacturing factory, Singapore

Interventions

Comparison of the effectiveness of:

• a barrier cream Arretil (Stockhausen, Germany); 3 to 4 x /day, and

• an after work moisturiser Keri Lotion (Westwood Pharmaceuticals, USA); 1 x after work, versus

• no intervention.

The participants were randomised to receive either a barrier cream (60 g tube every 6 weeks used on the hands before work and after each meal break), or a moisturiser used daily as an after work emollient, or no intervention. Arretil is a water‐soluble, non‐oily, silicone‐free barrier cream. Keri lotion is a liquid paraffin lotion (16%) and contains lanolin oil. The participants were followed up every 3 weeks for 6 months.

Outcomes

1. Proportion of OIHD (as defined by the study investigators): 'All cases of cutting fluid dermatitis were diagnosed clinically as irritant contact dermatitis' (p. 178). Dermatitis was assessed 'arbitrarily' (p. 177) and 'cutting fluid dermatitis was diagnosed on the history and clinical findings' (p. 177) (which probably meant the arbitrary assessment of dermatitis). The study investigators probably counted a subject as a case when irritant contact dermatitis was diagnosed at any of the 3‐weekly follow‐ups (combined point prevalence): 'Machinists who developed cutting fluid dermatitis (cases) and those who did not (non‐cases)' (p. 178). This was measured from week 1 to 12 and subsequent 18 weeks and extracted only for the last 18 weeks.

2. The secondary outcome measures was the amount of skin barrier impairment (TEWL) in the groups (mean and SD).

3. Frequency of treatment discontinuation due to adverse effects: No dropouts reported.

Diagnostic criteria for OIHD (degree of certainty for the diagnosis of OIHD): classification of dermatitis severity: mild = less than 25% of the total surface area of either hand involved (up to wrist line), moderate = more than 25% of the total surface area of either hand involved (up to wrist line).

Strengths

• Clinical examination was performed. Good method to assess prevalence of hand eczema.

Limitations

• No exclusion of endogenous/atopic hand eczema or allergic contact dermatitis of the hands (no patch test performed)

• No exclusion of non‐occupational hand eczema

Comparability

• Relatively narrow outcome definition

• Time frame: probably combined incidence at any weekly follow‐up ‐> comparatively many cases expected

Scoring system for the severity of OIHD: no score assessed

Quality of bioengineering methods (Pinnagoda 1990): 'Baseline TEWL measurement was made on the skin overlying the dorsal 3rd metarcarpophalangeal (MCP) joint of both hands with an Evaporimeter (Servomed, Vallingby, Sweden). We chose the 3rd MCP joint because of the ease of identifying the exact spot for repeat measurements. We conducted a pilot measurement on mid‐dorsal hand, 1st, 3rd, 5th dorsal MCP joints, mid‐ventral forearms and mid‐dorsal forearms on the 1st 30 volunteers. All showed fairly consistent recordings. We found the 3rd dorsal MCP joints to have the highest TEWL recordings and that these were fairly consistently reproducible. The method of TEWL measurement was as follows. The machinist rested in the examination room for 10 to 15 min. The TEWL on the dorsal 3rd MCP joint of each hand was then measured. TEWL values on the left hand followed by those on the right hand were recorded. The procedure was repeated once. 2 recordings of TEWL values for each hand were thus obtained; the average was recorded.' (p. 177)

Strengths

• Information is given on instrument (Evaporimeter, Servomed, Vallingby, Sweden) and the test room (temperatures, humidity)

• TEWL measurements were taken in a standardized environment (draft‐free, air‐conditioned)

• Measurements were always taken from the same skin area

• Method is well‐described

Notes

Groups were comparable at baseline.

No dropouts

Study funding sources not described. ('A research grant was obtained.' p. 176)

Possible conflicts of interest not described.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: 'Machinists were randomly assigned.' (p. 177). No further information provided.

Allocation concealment (selection bias)

Unclear risk

No information provided.

Blinding of participants and personnel (performance bias)
All outcomes

Unclear risk

Participants were not blinded to the intervention because original samples of the products were supplied. Possible influence on proportion of OIHD.

Blinding of outcome assessment (detection bias)
Subjective outcomes

Unclear risk

No information provided on blinding of outcome assessors.

Blinding of outcome assessment (detection bias)
TEWL and corneometry

Low risk

No influence of blinding is expected for the objective measurement of TEWL. Corneometry was not measured.

Incomplete outcome data (attrition bias)
Hand‐eczema‐related outcomes

Low risk

No dropouts reported in either group.

Incomplete outcome data (attrition bias)
Consumer satisfaction

Unclear risk

The study did not address this outcome.

Selective reporting (reporting bias)

Unclear risk

All outcomes reported. No protocol available.

Other bias

Unclear risk

Design: participants were recruited consecutively (when they started work) over a period of 3 years, in which the work conditions and exposures might have been changed ‐> unclear bias

More than one intervention group. This was accounted for in meta‐analysis. The control group was not double‐counted within one meta‐analysis.

Baseline imbalances: baseline imbalance regarding turning vs grinding section: controls ‐ 14 vs 9; barrier cream ‐ 8 vs 9; emollient cream ‐ 4 vs 10 ‐> unclear bias

Blocked randomisation in unblinded trials: no block randomisation reported

Differential diagnostic activity: no different diagnostic activities across study arms

Methods

Individually randomised controlled trial
Cross‐over design, 2 arms

Duration: 2x2 weeks

2 follow‐ups (at 2 and 4 weeks or at time of dropout)

Participants

Final number completing the 2x2 weeks: N = 70 initially healthy cleaners and kitchen workers

Number of participants randomised: N = 111 (N = 56 started with moisturiser; N = 55 started with no treatment)

Lost to follow‐up: N = 41

Mean age in years (groups I, II, III): 41 (range 19 to 67)

Mean duration of employment (groups I, II, III): 10 (range 1 to 35) years

Sex: 110 out of 111 included participants were female.

Setting: field study, Denmark

Interventions

Comparison of the effectiveness of:

• a moisturiser Locobase (Ferndale Laboratories, USA);

• versus no treatment

in a 2x2 weeks cross‐over design. Application requirements were not described. Locobase, manufactured by Ferndale Laboratories, USA, is a wound and skin emulsion formulation intended for topical application. No details were given about the composition of the cream.

Outcomes

The study investigators reported the results after grouping all participants in 4 groups according to their ability to complete the 4 week study. Finally, only results of groups I and II were reported.

Comparisons of intervention vs no treatment were only reported for corneometry, TEWL, and sum score in groups I and II

• Group I, N = 70, completed both periods Locobase (L) and Control (C).

• Group II, N = 23, completed L, but dropped out of in period C after developing severe dryness of the skin.

• Group III, N = 12, violated the protocol or declined to continue the study because they developed or feared to develop hand dermatitis during period C.

• Group IV, N = 6, turned up only once, did not find Locobase acceptable or went on vacation.

1. Proportion of skin changes (dryness only, dryness and scaling) and OIHD (dry eczema)

2. Biometric methods: corneometry values, TEWL (only figures and P values)

3. Subjective opinion

4. Mild adverse effects

5. A sum score was assessed, but not reported separately for incident cases of OIHD. This outcome does therefore not fit the review's outcomes (not extracted).

Diagnostic criteria for OIHD (degree of certainty for the diagnosis of OIHD): the degree of certainty could not be assessed because no diagnostic criteria for OIHD were described.

Scoring system for the severity of OIHD

• Clinical signs and symptoms assessed: Itching, dryness, scaling, fissuring, erythema, vesiculation

• Localisation of eczema assessed: fingers, hands, wrists, arms

• Severity rating: 1 = mild, 2 = moderate, 3 = severe

• Sum Score: multiplication of the severity ratings of all symptoms and sites (highest possible score: 8 (sites) x 7 (symptoms) x 3 (rating for severity) = 168

Strengths

• Contains different symptoms and severity rankings

Limitations

• Score not validated for inter‐ or intra‐observer reliability

• Contains a parameter (itching) which is self‐reported by the participant

• Extent of affected area is not assessed for each site

Quality of bioengineering methods (Pinnagoda 1990): 'Skin surface temperature (digital thermometer, Ellab type TRD, probe diameter 12 mm), transepidermal water loss (TEWL) (Evaporimeter EPI, Stockholm, Servomed, Sweden), and electrical capacitance (Corneometer CM420, Schwarzhaupt, W. Germany) were measured on both the right and left side and on the volar and dorsal aspect of the distal part of the 3rd finger, middle of the hands, and on the forearms (12 measurement points). The results are presented as overall mean values (the mean values for all measurements on all sites). All probes were hand‐held. The measurements were performed in a separate room with minor draught and protection shields. The mean room temperature was 23°C (20‐26°C) and the mean relative humidity (RH) 34% (25‐42%).' (p. 267).

Strengths

• Information is given on instruments and the test room (temperatures, humidity)

• Measurements were taken in a standardized environment (minor draught and protection shields)

• Measurements were always taken from the same skin areas

• Method is well‐described

Notes

The study investigators did not report the results according to the groups randomised, but regrouped all participants in 4 groups according their ability to complete the 4 week study period.

Therefore not included in meta‐analysis.

Study funding sources: 'The study was supported by the Danish Insurance Association, Copenhagen, L. P. Hansen's fund,
Odense, and Danfoss A/S, Nordborg, Denmark.' (p. 270)

Possible conflicts of interest not described.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: 'They were randomised into 2 groups' (p. 267). No further information provided.

Allocation concealment (selection bias)

Unclear risk

No information provided.

Blinding of participants and personnel (performance bias)
All outcomes

Unclear risk

Participants were not blinded due to study design.

Blinding of outcome assessment (detection bias)
Subjective outcomes

Unclear risk

Blinding of outcome assessors was probably not done. Quote: 'Clinical examination and skin physiological measurements were performed on entry...' (p. 267).

Blinding of outcome assessment (detection bias)
TEWL and corneometry

Unclear risk

No influence of blinding is expected for these objective measurements.

Incomplete outcome data (attrition bias)
Hand‐eczema‐related outcomes

High risk

N = 23 dropped out in period C after developing severe dryness of the skin, N = 12 violated the protocol or declined to continue the study because they developed or feared to develop hand dermatitis, N = 6 turned up only once, did not find Locobase acceptable, or went on vacation.

Incomplete outcome data (attrition bias)
Consumer satisfaction

Unclear risk

Dropout analysis did not address this outcome conclusively.

Selective reporting (reporting bias)

High risk

Results were not reported according to the original randomisation, but according to the ability of the participants to complete the study period. This was an attempt to deal with the high risk of attrition bias.

Consequently, the results were not reported in a way appropriate for an RCT.

Other bias

High risk

In‐study use of moisturiser was twice as high in participants who dropped out in the no‐treatment period (N = 23) compared to the participants who completed both parts of the study (2.3 g/person/day versus 1.2 g/person/day).

Design: cross‐over ‐> unclear bias

Baseline imbalances: pre‐study use of moisturiser differed significantly between the groups despite randomisation (96% versus 82%, P < 0.01). ‐> high risk

Blocked randomisation in unblinded trials: no block randomisation reported.

Differential diagnostic activity: not applicable because study arms were not evaluated separately.

Methods

Cluster‐randomised controlled trial; 18 clusters (enterprises)

Parallel design, 4 arms

Duration: 12 months

2 follow‐ups at 6 and 12 months

Participants

Final number evaluable after 12 months: N = 800 initially healthy metal workers exposed to cutting fluids (group 1: N = 217 in 6 enterprises; group 2: N = 209 in 5 enterprises; group 3: N = 213 in 4 enterprises; control: N = 161 in 3 enterprises)

Number of participants randomised: N = 1020 (group 1: N = 263; group 2: N = 253; group 3: N = 258; control: N = 246)

Lost to follow‐up: N = 220 (21.6%) withdrawal/exclusion at 2nd follow‐up (group 1: N = 46; group 2: N = 44; group 3: N = 45; control: N = 85)

Median age (all included participants): group 1: 40; group 2: 40; group 3: 44,5; control: 42

Sex: male

Inclusion criteria

• Male

• Age >= 18 y

• Regular exposure to cutting fluids

• Working contract for one further year at minimum

• Fitness for work at randomisation

• Willingness to comply

Exclusion criteria:

• Manifest hand eczema

• Intake of immunosuppressive drugs or topical application of corticosteroids or of other immunosuppressive agents on the hands

Setting: German factories mainly of small or medium‐size

Interventions

Comparison of the effectiveness of skin protection and/or skin care creams vs no recommendation in a 4‐armed trial

• Group 1: 'skin protection programme as generally recommended (i.e. use of skin protection and skin care)' (p. 363). Timing not explicitly stated, but obviously skin protection (barrier cream) before or during work and skin care (moisturiser) after work.

• Group 2: 'use of skin protection creams before or during working hours but complete avoidance of postexposure skin care' (p. 363)

• Group 3: 'use of skin care products solely after work' (p. 363)

• Group 4: 'The control group did not receive any recommendations concerning the use of skin protection or skin care.' (p. 363) (Only 31.4% used neither skin protection nor skin care after 6 months, 25.5% after 12 months. However, the participants were not instructed to avoid these measures.)

Pragmatic approach: 'All participants used the skin care and protection products provided by the employer.' (p. 363).

Outcomes

1. Proportion of OIHD: Dermatological history (hand eczema) in the last 6 months as reported by the participant in a standardised personal interview. This was also reported for the first 6 months, but not extracted for this review.

2. Severity was rated in a skin score, but not reported separately for incident cases of OIHD. This outcome does therefore not fit the review's outcomes (not extracted).

Diagnostic criteria for OIHD (degree of certainty for the diagnosis of OIHD)

Strengths

• Standardised interview

Limitations

• Self‐reporting of hand eczema underestimates the true prevalence of hand eczema. Especially mild changes are often not reported.

• No exclusion of endogenous/atopic hand eczema or allergic contact dermatitis of the hands (no patch test performed)

• No exclusion of non‐occupational hand eczema

Comparability

• Relatively narrow outcome definition

• Time frame: period prevalence: last 6 months ‐> comparatively many cases expected

Scoring system for the severity of OIHD: percentage change of skin score as primary outcome and relative change from baseline as secondary outcome: 'A quantitative skin score was used. In brief, the score comprised all morphological criteria and physiological abnormalities (e.g. dryness) characteristic of hand eczema. Extent, intensity and anatomical site of each type of skin lesion were also recorded.' (pp. 363‐4).

'All three physicians underwent standardized training before using the score. Moreover, during the study period digital photographs of the hands of randomly chosen participants were regularly evaluated and results discussed in order to maintain similarity of visual assessments between observers.' (p. 366).

Strengths

• Dermatologist were trained in using the score before start of study

• Good to excellent inter‐ and intra‐observer reliability

• Quantification of minimal skin lesions

Quality of bioengineering methods (Pinnagoda 1990): not relevant

Notes

Primary intervention: no inclusion of workers with manifest hand eczema (despite high baseline values on their very sensitive score). This was confirmed through correspondence with H Drexler.

Study funding sources: 'The study was funded by the German Statutory Accident Insurance (DGUV) and the Franz‐Koelsch‐Stiftung e.V.' (p. 369).

Possible conflicts of interest: none declared

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: 'Randomization was performed on the basis of the 19 included enterprises, assigning each enterprise randomly to one of four study arms.' (p. 363)

Apparently, an undescribed method was applied to ensure approximately equal size of the groups despite varying numbers of workers per enterprise.

Allocation concealment (selection bias)

Unclear risk

It cannot be excluded that allocation was biased by the results of the baseline screening. Potential confounders differed significantly at baseline. Even though the study investigators do not report significant associations with their outcomes, they did not test the association with review outcome 'proportion of OIHD'. Baseline skin condition was better in the control group, thus allowing less improvement after 1 year compared to other groups.

Blinding of participants and personnel (performance bias)
All outcomes

Unclear risk

No blinding reported; possible to influence proportion of OIHD.

Blinding of outcome assessment (detection bias)
Subjective outcomes

Unclear risk

Blinding of participants (= outcome assessors) was probably not done. This would be likely to influence reporting of OIHD, but clear information on blinding is lacking.

Blinding of outcome assessment (detection bias)
TEWL and corneometry

Unclear risk

No influence of blinding is expected for these objective measurements.

Incomplete outcome data (attrition bias)
Hand‐eczema‐related outcomes

Unclear risk

78.4% of recruited participants were available at last follow‐up (1 year), no analysis of dropouts.

Attrition in control was 2x as high as in other groups. 'High risk' judgement was therefore considered but discarded because dropout reasons are unknown.

Incomplete outcome data (attrition bias)
Consumer satisfaction

Unclear risk

The study did not address this outcome.

Selective reporting (reporting bias)

Unclear risk

Skin score was presented in various ways: percentage change of skin score as primary outcome, categorized percentage change as secondary outcome; further outcomes including absolute change of skin score were reported. This is questionable, but at least the study investigators did not omit their less significant calculations.

No protocol available

Other bias

High risk

Design: no correction for cluster randomisation ‐> no bias, but over‐precise results and limited comparability.

More than one intervention group. This was accounted for in meta‐analysis. The control group was not double‐counted within one meta‐analysis.

Baseline imbalances: significant differences of potential confounders and dyshidrotic eczema, but not of other dermatological disorders. It remains unclear whether this was due to lack of allocation concealment or happened by chance and was worsened by the cluster design ‐> high risk of bias

Blocked randomisation in unblinded trials: no block randomisation reported.

Differential diagnostic activity: no different diagnostic activities across study arms.

Methods

Cluster‐randomised controlled trial; 14 clusters (nursing schools)

Parallel groups, 2 arms

Duration: 3 years

2 follow‐ups at 1.5 and 3 years

Participants

Final number evaluable after 3 years: N = 250 initially healthy 1st‐year nurse apprentices (intervention: N = 121; control N = 129)

Number of participants randomised: N = 521 (numbers of participants allocated to groups were not reported)

Lost to follow‐up: N = 196 (37.6%; dropout per group was not reported)

Excluded from review due to OIHD at beginning of study: only the number of healthy participants who were evaluable were provided (N = 250). This included participants who had OIHD at least at 1 follow‐up before they dropped out.

Mean age in years (all included participants): intervention: 20.9 (SD 4.9); control: 23.6 (SD 7.7)

Sex: 13% male (intervention), 12% male (control)

Mean duration of employment: 0 years, baseline examination was before beginning of the nurse training

Inclusion criteria: not described
Exclusion criteria: not described

Setting: field study in 14 nursing schools (general nurses, paediatric and geriatric nurses, midwifes), Central Germany

Interventions

Comparison of the effectiveness of a:

• training program, versus

• no intervention

The intervention group received regular training (educational lecture with practical parts), and skin care cream (Asche Basis Creme). The lectures took place 3 times in the first year, 2 times in the second and third year. The nurse apprentices were encouraged to use alcoholic hand disinfection instead of hand washing or scrubbing. The effect of applying skin care cream was assessed by the use of a fluorescence technique with the Dermalux‐system as part of the training. Participants in the control group also received the skin care cream.

Outcomes

1. Proportion of OIHD (abnormal morphology = irritant skin changes during apprenticeship) defined as the occurrence of at least 1 morphology category at 1.5 or 3 years, or both

2. Frequency of treatment discontinuation due to adverse effects: study investigators state that dropouts were not due to skin problems

Diagnostic criteria for OIHD (degree of certainty for the diagnosis of OIHD): Irritant Skin changes were recorded using the operational definitions of Uter 1998b.

Strengths

• Clinical examination by physician. Good method to assess prevalence of hand eczema.

• Differentiation between mild changes and hand eczema (‘moderate’ or ‘severe’)

Limitations

• No exclusion of endogenous/atopic hand eczema or allergic contact dermatitis of the hands (no patch test performed)

• No exclusion of non‐occupational hand eczema

Comparability

• Broad outcome definition ‐> comparatively many cases expected

• Time frame: combined incidence at 1.5 years, 3 years (or both). No information on hand eczema in between.

Scoring system for the severity of OIHD: no scores assessed
Quality of bioengineering methods (Pinnagoda 1990): not relevant

Notes

Demographic data and dropout rates were only available for the entire study population of 521 nurse trainees.

Study funding sources: 'This work was generously supported by a grant from Asche Chiesi GmbH, Hamburg, Germany.' (p. 207)

Possible conflicts of interest: not described

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: 'Randomly divided' (p. 203); details not described

Allocation concealment (selection bias)

Unclear risk

Method not described.

Blinding of participants and personnel (performance bias)
All outcomes

Unclear risk

Blinding of participants unclear, possible influence on proportion of OIHD.

Blinding of outcome assessment (detection bias)
Subjective outcomes

Low risk

Evaluation of skin changes was performed by 2 blinded physicians.

Blinding of outcome assessment (detection bias)
TEWL and corneometry

Unclear risk

No influence of blinding is expected for these objective measurements.

Incomplete outcome data (attrition bias)
Hand‐eczema‐related outcomes

Low risk

Quote: 'The dropout rate was 37.6% which was not due to skin problems, but due to their absence because of illness at the date of evaluation.' (p. 204)

Quote: 'For follow up of dropouts, we performed a telephone interview with the trainee or (if he could not be reached) with his teacher. The aim of this interview was to evaluate, whether or not skin changes were responsible for the dropout.' (p. 204)

Low bias because the study investigators convincingly explain that they ensured no dropout was related to hand eczema

Incomplete outcome data (attrition bias)
Consumer satisfaction

Unclear risk

The study did not address this outcome.

Selective reporting (reporting bias)

Unclear risk

No protocol available.

Other bias

Low risk

Design: no correction for cluster randomisation ‐> no bias, but over‐precise results and limited comparability.

Baseline imbalances: no imbalances

Blocked randomisation in unblinded trials: no block randomisation reported

Differential diagnostic activity: no different diagnostic activities across study arms

Methods

Cluster‐randomised controlled trial; 48 clusters (hospital departments)

Parallel design with different recruitment dates, 2 arms

Duration: 12 months

4 follow‐ups at 3, 6, 9, and 12 months

Participants

Final number evaluable after 12 months: N = 981 initially healthy hospital employees handling irritants during work (intervention: N = 559; control: N = 422)

After correction for ICC = 0.005: N = 893 initially healthy hospital employees handling irritants during work (intervention: N = 509; control: N = 384)

Number of participants randomised: N = 1649 (intervention: N = 876; control: N = 773)

Excluded from review due to OIHD at beginning of study: N = 144 (intervention: N = 64; control: N = 80)

Lost to follow‐up: N = 524 (intervention: N = 253; control: N = 271) out of the 1505 initially healthy participants

Mean age in years (all included participants): intervention: 40.07 years, SD 11.5; controls: 40.8 years, SD 11.3

Sex: 78.4% female (intervention), 78.3% female (control)

Inclusion criteria

• Being employed at one of the participating hospitals

• Being able to fill out Dutch questionnaires

• Being aged between 18 and 64 years

• Working for at least 8 hr weekly

Exclusion criteria: not handling irritants during work

Setting: field study in 48 hospital departments (different cities in the Netherlands)

Interventions

Comparison of the effectiveness of:

• a multifaceted implementation strategy, versus

• leaflet only

The multifaceted implementation strategy included participatory working groups, role modes, educational programme including reminders, and a leaflet. The main recommendations were:

'1. When there is no visible contamination of the hands, use an alcohol‐based hand disinfectant instead of water and soap to
disinfect the hands*
2. Wear gloves when performing wet work
3. Wear cotton undergloves when you wear gloves for longer than 10 min
4. Use a moisturizer on a daily basis to nurse the skin and do not use a body lotion
5. Do not wear jewellery at work
6. Perform as little wet work as possible

* This means that the use of disinfectant should be increased and the use of water and soap should be decreased.' (p. 3)

All workers who were present at the educational session received a bag with one moisturiser, a pair of cotton undergloves, and two disinfectants (no product names or details reported). The intervention went on for 4 months. The comparison group received only the leaflet.

Outcomes

1. Proportion of OIHD (hand eczema in the past 3 months) as measured with the Nordic Occupational Skin Questionnaire – 2002 (NOSQ 2002): Defined as answering ‘yes’ to question D1 (ever had HE) or D2 (ever had eczema on wrists or forearms), and choosing one of the following answer categories for question D5: ‘I have it just now’ or ‘Not just now, but within the past 3 months’. Assessed at 3, 6, 9, and 12 months

2. A healthy skin score was assessed, but not reported separately for incident cases of OIHD. This outcome does therefore not fit the review's outcomes (not extracted).

Study investigators did not evaluate healthy skin score separately for participants in incident cases (as defined as review outcome).

Diagnostic criteria for OIHD (degree of certainty for the diagnosis of OIHD)

Strengths

• The questions used to assess the prevalence of hand eczema belong to a validated questionnaire (NOSQ 2002).

• Questions cover 3‐months‐period

Limitations

• Self‐reporting of hand eczema underestimates the true prevalence of hand eczema. Especially mild changes are often not reported.

• No differentiation between hand eczema and eczema on the forearms

• No exclusion of endogenous/atopic hand eczema or allergic contact dermatitis of the hands (no patch test performed)

• No exclusion of non‐occupational hand eczema

Comparability

• Relatively narrow outcome definition

• Time frame: period prevalence (last 3 months) ‐> comparatively many cases expected

Scoring system for the severity of OIHD: 'Workers assessed the health of the skin on their hands by means of the following question: ‘How would you judge the health of your hands/forearms at the moment on a scale from 0 to 10?’ (0, unhealthy skin; 10, healthy skin). This question was based on question D12 of the NOSQ‐2002.' (p. 4)

Strengths

• Standardised questions that belong to a validated questionnaire (NOSQ 2002)

Limitations

• Self‐reporting is less accurate than clinical examination by a physician

• Very subjective, no objective parameter

Quality of bioengineering methods (Pinnagoda 1990): not relevant

Notes

Demographic data and only reported for the study population including those with hand eczema at baseline.

Contingency table and dropouts were calculated based on the raw data received from the study investigators.

Study funding sources: 'This study was supported by a grant from the Netherlands Organization for Health Research and Development (ZONMW).'

Possible conflicts of interest: none declared

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: 'Based on the sequence of inclusion, randomisation was performed in strata of two by an independent researcher.' (Meer 2011, p. 3)

Allocation concealment (selection bias)

Unclear risk

Not described. Randomisation took place before baseline measurements by an independent researcher and was stratified by cluster‐based criteria. The study investigators excluded 17 out of 1666 baseline responders after randomisation.

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: 'Workers were not informed about the design of the study and outcome of randomisation.' (pp. 2‐3)

No blinding of personnel and department managers is expected to introduce only a low risk.

Blinding of outcome assessment (detection bias)
Subjective outcomes

Low risk

OIHD was self‐reported and participants were blinded.

Blinding of outcome assessment (detection bias)
TEWL and corneometry

Unclear risk

No influence of blinding is expected for these objective measurements.

Incomplete outcome data (attrition bias)
Hand‐eczema‐related outcomes

Unclear risk

Only superficial analysis of dropouts, reasons for dropout unknown. Quote: ‘Another limitation was that there was a non‐response rate of >30% at the final follow‐up measurement. However, the differences between the baseline values of the non‐responders and the baseline values of the total population were minimal.’ (p. 11)

Incomplete outcome data (attrition bias)
Consumer satisfaction

Unclear risk

The study did not address this outcome.

Selective reporting (reporting bias)

Low risk

Primary study outcome predefined in strategy and reported; secondary study outcomes predefined in strategy.

Other bias

Unclear risk

Design: no correction for cluster randomisation ‐> no bias, but over‐precise results and limited comparability.

Baseline imbalances (whole study population including participants with HE at baseline)

• In the intervention group 7.3% had HE at baseline compared to 10.3% in control group.

• In the intervention group, 69.4% performed patient related tasks compared to 81.2% in control group.

• In the intervention group, 57.4% had high education compared to 51.7% in control group.

• ‐> probably no strong impact on HE

Blocked randomisation in unblinded trials: no block randomisation reported

Differential diagnostic activity: study investigators speculate that participants of the intervention group are more aware of HE and were therefore more likely to report them: 'The educational session given during the intervention period might have increased awareness among the participants in the intervention group concerning their hand eczema symptoms. They may have evaluated their symptoms — which might have already been present at baseline — differently after the educational session. This might have led to an increase in hand eczema reports at follow‐up as compared with the control group that may be independent of a change in clinical status.' (p. 10)

Methods

Individually randomised controlled trial
Cross‐over design, 2 arms

Duration: 2 x 2 weeks

Follow‐ups: day 12, day 15, day 26, day 29

Participants

Final number evaluable after 2x2 weeks: N = 16 initially healthy 2nd year apprentice hairdressers (numbers allocated to groups not reported)

Number of participants randomised: N = 21 (numbers allocated to groups not reported)

Lost to follow‐up: N = 5 (numbers allocated to groups not reported)

Sex: 20 female out of 21 included participants

Median age (all included participants): 18 (range 16 to 30)

Atopy: 1 had a probable and 3 had a possible atopic disposition

Inclusion criteria: at least 5 times shampooing per day without gloves
Exclusion criteria: hand dermatitis

Setting: field study with apprentices from the main regional occupational training centre, Geneva, Switzerland

Interventions

Comparison of the effectiveness of:

• a barrier cream Excipial protect, and

• its vehicle

The participants were randomised to start either with barrier cream (Excipial protect) or with its vehicle. Excipial protect contains aluminium hydroxychloride and glycerine. Glycerine promotes water retention in the skin and the aluminium salt reduces excess sweating. The vehicle was designed specifically for skin care for occupational users. The first cream was applied 5 days a week for 2 weeks with a washout period of 2 days followed by another 2‐week treatment period with the second cream.

Outcomes

1. Proportion of irritant skin changes and OIHD (but as scores for dryness, redness, breaks in the skin; no quantitative data reported)

2. Biometric: corneometry; chromometry and TEWL; no quantitative values given, only figures and P values (test and comparison not described sufficiently)

3. Subjective opinion

4. Mild adverse effects

5. Frequency of treatment discontinuation due to adverse effects: The study investigators stated that the dropouts were 'for reasons not related to the study.'

6. A clinical scores was applied, but without reporting any quantitative data. This outcome does therefore not fit the review's outcomes (not extracted).

Participants were assessed at the beginning at day 12, day 15, day 26, and day 29. The primary outcome measures were the proportion of irritant skin changes (score defined by study investigators). The secondary outcome measures were the amount of skin barrier impairment (TEWL), skin hydration (corneometry), and skin colour (chromometry) in the groups. A further secondary outcome measure was the subjective opinion of the participants concerning different features of the creams (ease of use, consistency, oiliness, protective effect, tolerance, general aspects).

Diagnostic criteria for OIHD (degree of certainty for the diagnosis of OIHD): not relevant

Scoring system for the severity of OIHD: three scores were assessed: Dryness, redness, breaks in the skin. Scale of 0 to 3 (0 = none, 1 = mild, 2 = strong, 3 = maximum).

Strengths

• Clinical examination

Limitations

• Only a very limited number of symptoms were documented, many other symptoms of hand eczema are missing (e.g. vesicles, hyperkeratosis).

• No exclusion of endogenous/atopic hand eczema or allergic contact dermatitis of the hands (no patch test performed)

• No exclusion of non‐occupational hand eczema

Quality of bioengineering methods (Pinnagoda 1990)

'The biometric measurements were taken on the back of the dominant hand. The following instruments

• The Evaporimeter EP2 (ServoMed, Kinna, Sweden) was used to measure the transepidermal water loss (TEWL). The measurement represents the average over a 15 s period following a 30 s stabilization period.

• The Corneometer 820 PC (Courage and Khazaka, Köln, Germany) measured skin capacitance as a reflection of moisture. 3 measurements were taken and the average was calculated.' (p. 135)

Strengths

• Information is given on instruments and the test room (temperatures, relative humidity)

• Measurements were taken in a standardized environment (temperature, relative humidity)

• Measurements always on the same days of the week

• Method is well‐described

Limitations

• Measurements were taken on the back of the dominant hand. Rather vague information. Does not ensure that always the same area was measured.

Notes

Not included in meta‐analysis because no quantitative data reported.

Study funding sources: not described

Possible conflicts of interest: not described

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: 'The subjects were randomly assigned.' No further information provided.

Allocation concealment (selection bias)

Unclear risk

No information provided.

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

The participants were blinded: They received 3x 50 g tubes with identical markings at the beginning of the study and another 3 tubes after 2 weeks.

Comment: Probably done

Blinding of outcome assessment (detection bias)
Subjective outcomes

Low risk

Quote: 'Double‐blind' (p. 135). Participants, and outcome assessors were blinded.

Comment: Probably done

Blinding of outcome assessment (detection bias)
TEWL and corneometry

Low risk

No influence of blinding is expected for these objective measurements.

Incomplete outcome data (attrition bias)
Hand‐eczema‐related outcomes

Unclear risk

5 out of 21 apprentices dropped out during only 4 weeks. The study investigators stated that the dropouts were 'for reasons not related to the study. Their withdrawal did not effect the balance between the 2 groups to which they had been assigned' (p. 134). No further details provided. Dropout reasons unknown although it is doubtful they were truly unrelated to the study or outcome.

Incomplete outcome data (attrition bias)
Consumer satisfaction

Unclear risk

The study did not address this outcome.

Selective reporting (reporting bias)

Unclear risk

Results concerning primary outcome measure was only summarised briefly. Quote: 'Clinical scores were generally very low... under either cream' (p. 136), but not reported separately for verum and vehicle.

Other bias

Unclear risk

Application frequency of barrier cream and vehicle unclear.

Design: cross‐over ‐> no impact because no quantitative data for proportion of HE provided

Baseline imbalances: comparability of groups at baseline unclear, no details given.

Blocked randomisation in unblinded trials: no block randomisation reported

Differential diagnostic activity: no different diagnostic activities across study arms