We used the MEDLINE (via Ovid) and Embase databases and the Cochrane Central Register of Controlled Trials (CENTRAL) to identify articles through the seventh week of 2013 (Embase) and through February 18, 2013 (MEDLINE and CENTRAL). The search terms related to patient type and interventions were Medical Subject Headings (MeSH) terms and free terms. The search terms related to design type were based on the Cochrane Highly Sensitive Search Strategy.5 The results were limited to studies conducted in humans and published in English, French, or Spanish. Before carrying out the online searches, we identified 2 articles that met the inclusion criteria and were expected to be retrieved by the searches in all of the databases.6,7 The search strategies will be provided on request by the corresponding author.

Additionally, we carried out a search in PubMed Clinical Queries using the narrow search filter and a manual search of potentially relevant references cited in the included studies. We also asked Dr José Manuel Carrascosa, an expert on the treatment of AD with phototherapy, to review our results in order to detect the possible absence of relevant studies.

Finally, we also searched for clinical trials registered with Current Controlled Trials (www.controlled-trials.com) and with the World Health Organization's International Clinical Trials Registry Platform (http://apps.who.int/trialsearch/).

We considered all randomized clinical trials (RCTs) performed in patients clinically diagnosed with atopic dermatitis, without any age limit. We accepted as interventions all types of phototherapy as well as phototherapy in combination with psoralens (photochemotherapy). We excluded animal studies, studies in which the intervention was applied in localized areas (hands or feet), and studies in which the use of topical and/or systemic corticosteroids and immunosuppressants was not systematized or controlled. We accepted all outcome measures, although measures of disease improvement or quality of life were preferred over economic or laboratory measures.

The results obtained by the searches were downloaded to a reference management software package (EndNote, Thomson Reuters, 2011), which allowed us to filter the articles by title and abstract. Potentially relevant articles were independently evaluated in their entirety by pairs of reviewers. During data extraction, we used a questionnaire that included a study quality assessment form based on the Cochrane Collaboration's risk of bias tool.5 Throughout the process, all duplicate, rejected, and selected references were recorded in a PRISMA flow diagram (Fig. 1).

Figure 1.

Flow of articles during the review process.

a Kowalzick et al32

b Kowalzick et al,32 David,33 Der-Petrossian et al,34 Granlund Het,35 Niedner and Iliev,36 Potekaev et al,37 Raap et al,38 Schiener et al,39 Sowden et al40

c Heinlin et al,6 Der-Petrossian et al,28 Reynolds et al.30

(0.39MB).

Thirty-four articles were excluded from the analysis because they had a non-RCT design and/or they did not meet the patient-related inclusion criteria (Fig. 1).

Twenty-four records corresponding to 21 RCTs (961 patients) were included in the analysis. Three of the studies8–10 were also published as conference proceedings.11–13 Two studies14,15 included children and adolescents (32 patients). Two studies14,16 were carried out in Asia and the rest were carried out in Europe. Three were multicenter studies.6,17,18 Most of the studies compared different types of phototherapy, including high-dose (HD) UV-A1, medium-dose (MD) UV-A1, UV-B, UV-A and UV-B combination therapy (UV-AB), NB UV-B, and PUVA. Other modalities were excimer laser (EL),8 full-spectrum-light phototherapy (FSL),16 and synchronous balneophototherapy (sBPT).6 One study evaluated the utility of a skin-reflectance-guided UV-B regimen as a means of reducing the cumulative dose of radiation.19 Three studies compared phototherapy with other treatments, namely ciclosporin,17 topical pimecrolimus,14 and topical corticosteroid therapy combined with phototherapy.15 The shortest follow-up period was 4 weeks9,14,20 and the longest periods were 6 months6,21 and 12 months.7,17 Several studies had no follow-up period.10,15,18,19,22–27 In most studies, the outcome measures were changes in scores on clinical scales. The most frequently used scales were SCORing Atopic Dermatitis (SCORAD) or a modified form of SCORAD,6,7,9,16,17,19,21,28 Costa's Simple Scoring System (SSS),10,15,18,27 the Leicester scale,15,27 and the severity score29 of Hanifin and Rajka.23–26 Some studies also used quality-of-life scales (Eczema Disability Index,17 Sickness Impact Profile6) and assessments of pruritus.8,14,20,22,27 Cumulative dosage was an outcome measure in 1 study19 and length of remission was used in 2 studies.7,21

In general, the studies reviewed had a high risk of bias and relevant information was frequently missing (Tables 1–6, Fig. 2).

Figure 2.

Risk of bias.

+ indicates low risk of bias; ?, unknown risk of bias; –, high risk of bias; 1, generation of random allocation sequence (selection bias); 2, intervention allocation (selection bias); 3, masking of participants and personnel (performance bias); 4, masking of assessors (detection bias); 5, incomplete outcome data (attrition bias); 6, selective reporting (reporting bias); 7, other biases; BJD, British Journal of Dermatology.

(0.49MB).

This is the first systematic review on phototherapy in AD to be published in Spanish. We analyzed 21 studies that met the inclusion criteria. These studies included a total of 961 patients, of whom 32 were younger than 18 years of age.

The use of the various phototherapy modalities had a chronological distribution. In the 1980s and 1990s, studies of UV-A, UV-B, and UV-AB phototherapy were predominant. PUVA, NB UV-B, and UV-A1 emerged later, followed by modalities such as EL and FSL.

In some studies, there were statistically significant differences between combined UV-AB phototherapy and UV-A or UV-B phototherapy (Table 1). HD UV-A1 was subsequently found to be superior to UV-AB.10,18 Later studies showed that MD UV-A1 had similar efficacy and fewer side effects than HD UV-A19,21 (Table 2).

The most homogeneous studies are those that compare UV-A1 to NB UV-B (Table 3). Two studies27,30 found NB UV-B to be superior to UV-A1. In an RCT with a large number of participants and low risk of selection bias, the superiority of NB UV-B was maintained after 3 months of follow-up.30 The other 2 studies20,22 that compared UV-A1 to NB UV-B found no important differences between the treatment modalities. It should be noted that an unconventional UV-A1 regimen was used in all 4 studies (2 or 3 weekly sessions instead of 5); the effectiveness of UV-A1 could, therefore, be greater.

A RCT28 with no follow-up period compared 8-MOP PUVA bath therapy to NB UV-B. The results were consistent with clinically relevant differences between the treatments, but the statistical power of the study was low. Another study7 found that oral 5-MOP PUVA therapy yielded a significantly longer remission time than UV-A1, although differences in treatment duration and frequency of application may have influenced the results.

As for the more novel techniques, FSL16 and EL8 have each been evaluated in an RCT and found to have optimal results, although publication bias may have played a role. Similarly, there is scant evidence on combined therapies. One study6 found sBPT to be superior to NB UV-B. In another study,15 the addition of topical corticosteroids to UV-AB therapy was found to reduce UV-B dose and treatment duration. In an RCT14 carried out in children, concomitant use of topical pimecrolimus during NB UV-B therapy did not yield any added benefit. Only 1 RCT17 has compared an immunosuppressive treatment—ciclosporin A—to UV-AB. Ciclosporin A had significantly better results, but the UV-AB doses were suboptimal and the study had a high risk of bias.

All phototherapy modalities were generally described as well-tolerated, although adverse effects were recorded systematically in only a few RCTs. The absence of standardized protocols for the application of these techniques—determination of initial dose according to skin phototype or MED, incremental dose increases, frequency of sessions, etc.—makes it difficult to interpret and compare the results. Given the heterogeneity of the treatments and the reporting deficiencies, we opted not to conduct a meta-analysis.

In general, the studies had a high or uncertain risk of bias5 and omitted relevant information. In many studies, the number of participants was small, the randomization method was unclear, and patients lost to follow-up were excluded from statistical analysis. The follow-up period was short or nonexistent in most of the studies. In RCTs that used within-patient comparisons, there is a possibility of bias resulting from the possible systemic effect of phototherapy applied to each side of the body.

Patient inclusion and exclusion criteria varied, and the criteria of Hanifin and Rajka were not used in all of the RCTs. In one RCT,14 the inclusion and exclusion criteria were not mentioned explicitly.

To ensure that we compiled as many studies as possible, we reviewed the Global Resource of Eczema Trials database (Centre of Evidence Based Dermatology, http://www.greatdatabase.org.uk/) and found no RCTs that had not been included in our results. However, we restricted our search to studies in English, French, and Spanish. Dr José Manuel Carrascosa, an expert in phototherapy treatment, reviewed our results and detected only the absence of studies in German. Another recent systematic review4 included only 1 article31 in German.

Finally, publication bias is possible, given that most of the studies reported a positive result for the tested treatment modality; however, this bias is less likely because none of the studies were placebo-controlled.

There have been 2 previous systematic reviews on phototherapy in AD. The first review3 did not include studies of PUVA therapy. This review concluded that UV-A1 phototherapy, if available, should be used to treat acute forms of AD and that NB UV-B should be used to treat chronic forms of AD. In our opinion, these conclusions are difficult to justify because these clinical differences are not well established in the cited studies. As for the more recent systematic review,4 there were 4 discrepancies between the RCTs we selected and those selected by the other authors. We included 2 studies8,14 comparing topical therapy to phototherapy that the other review omitted because the authors considered the diagnosis of AD to be doubtful. We also included a study19 that evaluated the utility of reflectance-guided UV-B for reducing UV-B dose. Finally, we omitted 1 study31 included in the other review because it was published in German.

• •

  There is evidence to support the use of NB UV-B and UV-A1 phototherapy in moderate to severe forms of AD. There is scant evidence to support the use of PUVA.

• •

  It may be possible to find indications for modalities such as EL in the prurigo form of AD, FSL, and sBPT, but further studies are needed.

• •

  Data on the use of phototherapy in childhood AD are limited, and therefore caution must be exercised when this technique is used in children. There is no evidence to support the use of phototherapy in pregnant women with AD.

• •

  There are few data on the long-term effects of phototherapy in AD, including possible carcinogenic effects.

• •

  We found no RCTs comparing the use of phototherapy to the use of oral corticosteroids.

• •

  Only 1 RCT has compared systemic immunosuppressive therapies to phototherapy, and it did not include the modalities for which the strongest evidence is available (NB UV-B and UV-A1).

• •

  AD severity assessment criteria, irradiation techniques, and assessment scales and other outcome measures should be standardized; this is the objective of the Harmonizing Outcome Measures for Eczema initiative (http://homeforeczema.org/).

• •

  Given the impact of AD, it is surprising that only 2 studies included quality-of-life measures (Skindex and the Eczema Disability Index) and that only 2 other studies included subjective assessment scales.

• •

  A minimum follow-up period should be included and tolerability parameters and adverse effects should be recorded.

• •

  Results should be reported in accordance with the Consolidated Standards of Reporting Trials statement.

The authors declare that no experiments were performed on humans or animals for this investigation.

The authors declare that no private patient data are disclosed in this article.

The authors declare that no private patient data are disclosed in this article.