Randomized Controlled Trial of Topical Corticosteroid and Home-based Narrowband Ultraviolet B for Active and Limited Vitiligo

Results of the Hi-Light Vitiligo Trial

K.S. Thomas; J.M. Batchelor; P. Akram; J.R. Chalmers; R.H. Haines; G.D. Meakin; L. Duley; J.C. Ravenscroft; A. Rogers; T.H. Sach; M. Santer; W. Tan; J. White; M.E. Whitton; H.C. Williams; S.T. Cheung; H. Hamad; A. Wright; J.R. Ingram; N.J. Levell; J.M.R. Goulding; A. Makrygeorgou; A. Bewley; M. Ogboli; J. Stainforth; A. Ferguson; B. Laguda; S. Wahie; R. Ellis; J. Azad; A. Rajasekaran; V. Eleftheriadou; A.A. Montgomery


The British Journal of Dermatology. 2021;184(5):828-839. 

In This Article

Patients and Methods

The trial protocol has been published previously.[29,30] No changes were made to the eligibility criteria or outcome measures after trial commencement. The study was approved by the Health Research Authority East Midlands (Derby) Research Ethics Committee (14/EM/1173) and the Medicines and Healthcare products Regulatory Agency (EudraCT 2014-003473-42). Participants or their parents or carers gave written informed consent. The trial was informed by a pilot trial,[31] and was registered prior to the start of recruitment (ISRCTN 17160087; 8 January 2015). A full trial report is available through the NIHR Journal series.[32]

Study Design and Setting

The study was a multicentre, three-arm, parallel-group, pragmatic, placebo-controlled randomized controlled trial, with nested health economics and process evaluation studies (reported separately).[30]

Trial interventions were delivered in secondary care across 16 UK hospitals. Participants were identified through secondary care dermatology clinics and general practice mailouts, and by self-referral. Participants were enrolled for up to 21 months (9 months of treatment, 12 months of follow-up) and attended hospital clinics on two consecutive days at baseline for recruitment and training, and then at 3, 6 and 9 months to assess outcomes. Follow-up thereafter was by 3-monthly questionnaires, by post or email.


The study's objectives were as follows. (1) To evaluate the comparative effectiveness and safety of home-based interventions for the management of active, limited vitiligo in adults and children, comparing firstly, handheld NB-UVB vs. potent TCS (mometasone furoate 0·1% ointment); and secondly, a combination of handheld NB-UVB plus potent TCS vs. potent TCS alone. (2) To assess whether treatment response (if any) is maintained once the interventions are stopped. (3) To compare the cost-effectiveness of the interventions from a UK National Health Service (NHS) perspective. (4) To understand the barriers and facilitators to adoption of these interventions within the UK NHS. Objectives 3 and 4 are reported elsewhere.[30]


Participants were aged > 5 years, with nonsegmental vitiligo limited to approximately 10% or less of body surface area, and at least one vitiligo patch that had been active in the last 12 months (reported by the participant, or parent or carer). Full eligibility criteria are listed in the protocol.[29]


All participants received an NB-UVB light unit (active or dummy; used on alternate days) and either a TCS (mometasone furoate 0·1% ointment; Elocon®; Merck Sharp & Dohme, Kenilworth, NJ, USA) or vehicle (placebo), applied daily on alternate weeks. Any device found to have an output that had more than ± 20% of the expected mean output, or a dummy device testing positive for any NB-UVB emission, was returned to the manufacturer. Treatments were continued for up to 9 months, and concomitant medications were logged. Further details of the interventions are provided in the protocol and full trial report.[29,32] Dummy devices were identical to active devices but used special covers that blocked transmission of NB-UVB. Placebo ointment was identical in appearance to active ointment.

Participants selected up to three patches of vitiligo for assessment: one on each of three anatomical regions (head and neck, hands and feet, and rest of body). One patch was selected as the target for primary outcome assessment and was reported as active (new or changed) within the last 12 months.


The outcomes examined were the core outcome domains for vitiligo.[33,34]

Primary Outcome

The primary outcome was participant-reported treatment success at the target patch of vitiligo after 9 months of treatment. This was measured using the Vitiligo Noticeability Scale (VNS),[35,36] with treatment success defined as 'a lot less noticeable' or 'no longer noticeable' compared with before treatment. Participants used digital images of the target patch before treatment to help inform their assessment.

Secondary Outcomes

Secondary outcomes were as follows. (i) Blinded assessment of treatment success (VNS) at the target patch assessed by a panel of three people with vitiligo, using digital images. (ii) Participant-reported treatment success for each of the three anatomical regions (all assessed patches) using VNS, assessed at 9 months. (iii) Onset of treatment response at the target patch, assessed by investigators. (iv) Percentage repigmentation at the target patch at 9 months, using blinded clinician assessment of digital images (0–24%, 25–49%, 50–74%, 75–100%). Investigator assessments were used if images at 9 months were unavailable. (v) Quality of life at baseline, end of treatment (9 months) and end of follow-up (21 months). Disease-specific quality-of-life (VitiQoL, Skindex 29) and generic quality-of-life (EuroQol 5 Dimensions 5 Levels; EQ-5D-5L) instruments were completed by adults aged > 18 years. Children aged 5–17 years completed the Child Health Utility 9D (generic) and children aged > 11 years also completed the EQ-5D-5L (generic). (vi) Maintenance of treatment response assessed by participants for the target patch at 12, 15, 18 and 21 months. (vii) Safety: adverse device effects, and adverse reactions during the treatment phase. (viii) Time burden of treatment: time per session for active NB-UVB treatment and participant-reported treatment burden for active TCS and NB-UVB treatments at 3, 6 and 9 months.

Adherence with treatments was recorded using treatment diaries and was collated at 3-monthly clinic visits.

Randomization and Blinding of Allocation and Outcome Assessment

Participants were randomized 1 : 1 : 1 to receive TCS plus dummy NB-UVB (TCS group), vehicle ointment plus NB-UVB (NB-UVB group), or TCS ointment plus NB-UVB (combination group). Allocation was minimized by recruiting centre, body region of target patch and age, weighted towards minimizing the imbalance in trial arms with probability 0·8. The randomization sequence was accessed by staff at the recruiting hospital, using a secure web server created and maintained by the Nottingham Clinical Trials Unit (NCTU) to ensure concealment. A central pharmacy (Mawdsleys, Doncaster, UK) distributed the interventions. The pharmacy was notified of the allocation for randomized participants via the web-based system and trial treatments were sent directly to participants' homes. Only the NCTU programmer, the pharmacy staff and the NCTU Quality Assurance staff had access to treatment allocations. Additional blinded outcome assessments were performed by a panel of three people with vitiligo (for the primary analysis) and a clinician for the secondary outcome of percentage repigmentation, using digital images taken at baseline and at 9 months.

Statistical Methods

Sample Size. Assuming that 15% of participants allocated to receive TCS would achieve treatment success,[37] 372 participants were required to detect a clinically significant absolute difference between groups of 20%, with 2·5% two-sided alpha and 90% power. Allowing for up to 15% noncollection of primary outcome data at 9 months, the target sample size was 440 participants. A planned sample-size review by the Data Monitoring Committee after 18 months of recruitment resulted in a recommended increase in sample size to 516 participants.

Analysis. All analyses were prespecified in a statistical analysis plan, which was finalized prior to database lock.[29] Amendments to the analysis plan compared with the protocol are summarized in Table S1 (see Supporting Information).

The primary analysis included all participants, regardless of adherence, and with multiple imputation of missing outcome data. Estimates of the analyses were obtained from 30 multiply imputed datasets by applying the combination rules developed by Rubin.[38] Prior to primary analysis, baseline characteristics were summarized by treatment arms and the availability of primary outcome at 9 months, in order to check the missing-at-random assumption of multiple imputation.

For the primary outcome, the number and percentage of participants achieving 'treatment success' were reported for each treatment group at 9 months. Randomized groups were compared using a mixed effects model for binary outcome adjusted by minimization variables. The primary effectiveness parameter for the two comparisons of NB-UVB alone and combination treatment, each vs. TCS alone, was the difference in the proportion of participants achieving treatment success at 9 months, presented with the 95% confidence interval (CI) and P-value. By default, risk differences are reported, because these estimates are more clinically intuitive for binary outcomes. However, where models estimating risk difference did not converge, odds ratios are reported instead of risk differences.

Sensitivity analyses were conducted: (i) to adjust for any variables with imbalance at baseline, (ii) to repeat the primary analysis based on participants whose primary outcome was available at 9 months and (iii) to investigate the effects of treatment adherence. Complier-average causal effect analysis[39] was conducted where taking ≥ 75% of expected treatments was considered a complier. Planned subgroup analyses were: (i) children vs. adults, (ii) by body region of the target vitiligo patch, (iii) by activity of the target patch (hypomelanotic patch: definitely vs. maybe or no) and (iv) ≥ 4 years duration of vitiligo vs. < 4 years. These analyses were conducted by inclusion of appropriate interaction terms in the regression model and were exploratory.

Secondary outcomes were analysed by a similar approach, using appropriate regression modelling depending on the outcome type. An additional post hoc subgroup analysis explored the impact of skin type (types I–III vs. types IV–VI).

Patient and Public Involvement

People with vitiligo were involved in all aspects of the trial, including prioritization of the research questions, study design, oversight, and conduct and interpretation of the results.[32]

Data Sharing

Anonymized patient-level data are available from the corresponding author upon reasonable request.