Design

The first study, involving 78 people, was a double‐blind RCT comparing two steroid creams of different potencies (Roenigk 1980). The creams were prepared in the same cream base and identical tubes, and were compared in a 12‐week cross‐over trial. A single active lesion on each person was chosen for monitoring. One of the creams was applied for 6 weeks, then the other for a further 6 weeks. Only the outcome prior to cross‐over was included in the analysis for this review.

The second trial was a double‐blind RCT, over 8 weeks (Ruzicka 1992).

The third trial, was a double‐blind RCT, over 8 weeks (Barikbin 2009).

In Jemec 2009, an active agent was compared with placebo, in a double‐blind RCT. The creams were identical.

A 2011 trial was a double‐blind placebo‐controlled RCT, over 12 weeks (Kuhn 2011). The vehicle was used as the placebo, to ensure blinding.

Sample sizes

There were a total of 197 participants in the five trials: 78 in Roenigk 1980; 58 in Ruzicka 1992; 10 in Barikbin 2009; 37 in Jemec 2009; and 14 in the study by Kuhn 2011.

Setting

The setting of Roenigk 1980 was not stated. Ruzicka 1992, Barikbin 2009, Jemec 2009 and Kuhn 2011 were set in specialised lupus clinics, as most people with DLE were attending such clinics.

Participants

Roenigk 1980 randomised for inclusion 32 men and 61 women, aged 17 to 82 years, from the following ethnic groups: 23 black American, and 25 white American in one group; and 23 black American, and 22 white American in the other. The duration of their disease was very variable (1 to 240 months). There were no significant differences between people in the two arms, in terms of demographics or duration of disease. All participants had chronic DLE. However final number of participants was 78, (others excluded at onset) and ethnic and gender breakdown of participating group was not supplied.

In Ruzicka 1992, there were 22 men and 36 women, with a mean age of 43.3 years. The ethnic group was not mentioned. All participants had cutaneous lupus, 39 discoid and 19 subacute LE. People with subacute LE were more highly represented in the hydroxychloroquine arm.

In Barikbin 2009 , participants were Iranian, aged 20 to 53, and 7 of the 10 were women. Participants in Jemec 2009 included 28 women and 9 men, aged 34 to 72. There was no ethnicity provided. In the Kuhn 2011 trial, there were six women and eight men with DLE, with a mean age of 49.8 ± 5.9. Ethnicity was not supplied.

Interventions

Roenigk 1980 compared two topical steroids, fluocinonide 0.05% (potent steroid) and hydrocortisone 1% (mild steroid) in the same cream base.

Ruzicka 1992 compared two oral agents, acitretin (a synthetic retinoid) and hydroxychloroquine (an anti‐malarial).

Barikbin 2009 compared pimecrolimus 1% cream with betamethasone 17‐valerate 0.1% cream.

Jemec 2009 compared salbutamol 0.5% cream with placebo (same vehicle, so indistinguishable).

Kuhn 2011 compared tacrolimus 0.1% cream with placebo (vehicle).

Outcomes

In Roenigk 1980, the outcome was based on the authors' own 5‐point scale, using a single test site. The investigators used a scale, providing a score of 1 if the lesion became worse, 2 if there was no improvement, 3 if there was a little improvement, 4 if improvement was marked, and 5 if the lesion cleared or showed excellent improvement. There was no assessment by the participants.

Assessment of severity in Ruzicka 1992 was based on an ordinal scale using erythema (redness), infiltration (thickness), and scaling with complete clearing graded as 0, improvement as 1, and no change or deterioration as 2. Extent of skin disease and the occurrence of adverse reactions were also documented.

Barikbin 2009 used digital photographs, which were examined by three blinded dermatologists, using a severity score.

Jemec 2009 reported outcomes using the CLASI score, the clinician's global assessment and the participant's global assessment.

Kuhn 2011 used digital photography and a clinical assessment, using a clinical score. They reported adverse events.

Quality‐of‐life measures or participant global evaluation were not recorded in any of the studies.

Comparability of the two arms

The risk of bias was rated as low in four studies. In the Roenigk 1980 trial, the groups were comparable at baseline. The participants in the two arms in the Jemec 2009 study had similar active lesions at baseline (activity scores 1.32 ± 0.48 vs 1.11 ± 0.47). The mean disease severity score was similar between the two groups at baseline in the Barikbin 2009 study. Kuhn 2011 also had comparable arms at baseline.

Ruzicka 1992 included 19 people with subacute cutaneous lupus erythematosus. As people with this form of cutaneous lupus were more strongly represented in the chloroquine group, this might have influenced the outcome of treatment, as people with subacute lupus may have a better response rate. The response rates for discoid lupus and subacute lupus were not described separately. Hence, this study was judged to be at high risk of bias.

We are of the opinion that none of the five studies included in this review are affected by other potential threats to validity, including:

• interim results analysis;

• deviation from study protocol not reflecting clinical practice;

• prerandomisation of the intervention which could affect the result;

• contamination of drugs (pooling of drugs by participants);

• overly wide inclusion criteria for participants;

• use of insensitive instruments to measure outcomes;

• selective reporting of subgroups;

• inappropriate influence of funders.

Percentage of people with complete resolution of skin lesions (i.e. return to normal skin appearance)

Note that we accepted the following terms to describe complete resolution: complete clearing, clearing, marked improvement and excellent improvement.

In the Roenigk 1980 study fluocinonide 0.05% cream appeared to be superior to hydrocortisone 1% cream (low‐potency steroid) in the treatment of discoid lupus erythematosus (summary of findings Table for the main comparison). At the 6‐week cross‐over point, clearing or excellent improvement was found in 10 out of 37 people (27%) using fluocinonide and in four out of 41 people (10%) using hydrocortisone (RR 2.77, 95% CI 0.95 to 8.08), 1 study, n = 78, low‐quality evidence) (Analysis 1.1). This shows an absolute benefit of potent steroid of 17% (95% CI 0 to 34%). The quality of evidence for this outcome was judged to be low.

In the Ruzicka 1992 study, the overall outcome appeared similar in the hydroxychloroquine 400 mg and acitretin 50 mg groups (summary of findings Table 2). There was marked improvement or complete clearing in 13 out of 28 (46%) participants given acitretin and in 15 out of 30 (50%) of those on hydroxychloroquine; i.e. there was no significant difference between the two groups (RR 0.93, 95% CI 0.54 to 1.59) (Analysis 2.1).The quality of evidence for this outcome was judged to be low.

In the Barikbin 2009 study, those treated with pimecrolimus 1% had an 86% reduction in severity score compared to 73% in the betamethasone 17‐valerate 0.1% cream group (P = 0.1). The number of participants with complete resolution could not be obtained from the paper (summary of findings Table 3).

In the study by Jemec 2009, the number of participants with complete resolution could not be obtained from the paper (summary of findings Table 4).

In the Kuhn 2011 study looking at tacrolimus 0.1% cream versus placebo (vehicle), no participants with DLE were reported as having experienced complete clearing (summary of findings Table 5). The quality of evidence for this outcome was judged to be low.

Percentage of people with clearing of erythema in at least 50% of lesions (postinflammatory pigmentation could persist)

In none of the studies was there a measurement of the percentage of participants with clearing of erythema in at least 50% of lesions.

Roenigk 1980 used a score that included erythema to measure outcome, but did not report the erythema separately. In the Ruzicka 1992 study the effect on erythema was assessed using an ordinal scale where complete clearing was graded as 0 and improvement as 1, so we take this to mean there was clearing of erythema in at least 50% of lesions. Erythema showed more marked improvement in the hydroxychloroquine 400 mg group and more people in this group showed complete clearing compared to those taking acitretin 50 mg, although the result was not statistically significant. There was complete clearing or marked improvement of erythema in 10/24 (42%) of participants in the acitretin group compared to 17/25 (68%) in the hydroxychloroquine group (RR 0.61, 95% CI 0.36 to 1.06) (Analysis 2.2.) The quality of evidence for this outcome was judged to be low.

While Barikbin 2009 recorded erythema in treated sites as part of the clinical score, results for erythema were not reported separately.

In the Jemec 2009 study, there was no significant improvement in erythema, but the actual data were not provided.

In the Kuhn 2011 study, erythema was included in the clinical score, but actual data were not provided.

Improvement in participant satisfaction/quality of life measures

There was no record of any measurements of participant satisfaction or quality of life in any of the included studies. Jemec 2009 and Kuhn 2011 recorded itch and pain as reported by participants, but no global score or assessment of quality of life.

Relapse rate when medication stopped or reduced

There was no explicit record of relapse rates when medication was stopped in three of the studies. Barikbin 2009 reported that there was no relapse at eight weeks, after stopping the trial medication at four weeks. The trial of Kuhn 2011 reported improvement in skin lesions at four and eight weeks but the benefit was not maintained at 12 weeks. Thus the skin condition had relapsed while the participants were still using the trial cream, tacrolimus 0.1%.

Prevention of new lesions

In none of the studies was this outcome directly measured.

Adverse effects of medication, leading to discontinuation or significant morbidity

In the Roenigk 1980 study, adverse effects were minor and were seen in both groups (moderate‐quality evidence). Skin irritation occurred in three people using hydrocortisone cream 1% and burning occurred in two people using fluocinonide cream 0.05%. One person in the hydrocortisone group developed acne. The trial was too short to assess other side‐effects, such as skin atrophy (thinning) or capillary damage.

In the Ruzicka 1992 study, compared with those taking hydroxychloroquine 400 mg adverse effects were more frequent in the acitretin group 50 mg and were more severe, with four participants having to discontinue treatment because of adverse effects (moderate‐quality evidence). The main clinical adverse effects were dry lips/cheilitis (93% of the acitretin and 20% of the hydroxychloroquine group) and gastrointestinal disturbance (11% of the acitretin and 17% of the hydroxychloroquine group, but no specific information was provided). Laboratory studies revealed changes in triglycerides in both groups, the levels rising from a mean of 1.66 to 2.10 mmol/l in the acitretin group (P = 0.01) and falling from 1.94 to 1.50 mmol/l in the hydroxychloroquine group (P = 0.02). The difference between the two groups was highly significant (P = 0.006).

In the Barikbin 2009 study, adverse events were not reported.

In the Jemec 2009 trial there were 39 reported adverse events, 15 in the placebo group and 24 in the salbutamol 0.5% topical cream group (moderate‐quality evidence). None of the adverse events were considered serious and the adverse events did not appear to be related to the trial medication.

Kuhn 2011 reported that five people experienced itching or burning at the site of tacrolimus 0.1% cream application, sometimes lasting up to six weeks, and one participant developed herpes simplex infection of the lip during treatment (moderate‐quality evidence). There were no serious adverse events. No adverse events were experienced by the participants using the placebo cream.

Implications for healthcare costs

There was no measurement of this outcome in any of the studies.