Lightmasks that prevent dark adaptation for non-central diabetic macular oedema: the CLEOPATRA RCT

Article history

The research reported in this issue of the journal was funded by the EME programme as project number 11/30/02. The contractual start date was in January 2014. The final report began editorial review in April 2018 and was accepted for publication in August 2018. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The EME editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Sobha Sivaprasad has received research grants, personal fees and advisory board honoraria from Bayer AG (Leverkusen, Germany), Novartis International AG (Basel, Switzerland) and Allergan plc (Dublin, Republic of Ireland), research grants and advisory board honoraria from F. Hoffman-La Roche AG (Basel, Switzerland) and Boehringer Ingelheim GmbH (Ingelheim am Rhein, Germany) and advisory board honoraria from Heidelberg Engineering GmbH (Heidelberg, Germany). Sobha Sivaprasad is also on the National Institute for Health Research (NIHR) Health Technology Assessment Commissioning Committee. Philip Hykin reports grants, personal fees and non-financial support from Novartis International AG; grants, personal fees, non-financial support and other (including travel expenses and advisory board honoraria) from Bayer; and grants and personal fees from Allergan AG outside the submitted work. Andrew Toby Prevost is a member of the NIHR Public Health Research Funding Committee.

Copyright statement

© Queen’s Printer and Controller of HMSO 2019. This work was produced by Sivaprasad et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2019 Queen’s Printer and Controller of HMSO

Background

Rationale for a randomised clinical trial and mechanistic evaluation

Arden hypothesised that if dark-adaptation can be prevented, the rod dark current would never become maximal and DMO and diabetic retinopathy could be alleviated by decreasing the oxygen demand. 16 As people only dark adapt at night during sleep, sleeping in an illuminated environment should prevent or reverse the condition. Sufficient light is transmitted by the eyelids to reduce the dark current. Therefore, wearing a lightmask of 500–505 nm to suppress the rod-induced dark adaptation may alleviate DMO and prevent the progression of diabetic retinopathy. The lightmasks ensure the uniform illumination of weak light, to which the eyes adapts quickly owing to the Troxler Effect. However, it is important that the masks are comfortable to wear and do not disturb sleep. A few short trials have supported the proof-of-concept that prevention of dark adaptation by using lightmasks may be safe and effective in reducing the retinal thickness in non-central DMO. 19,20

The design must permit the mask to be worn by people with different head shapes and deliver rod excitation efficiently. The spectral output is important and should be matched as closely as possible to the response spectrum of the rod cells. This has been tested in two clinical trials. The first was a proof-of-concept study in which 12 participants slept in a mask containing a chemiluminescent source that exposed one eye only to light. 19 The trial lasted 3 months and all participants found the masks comfortable and the method of treatment acceptable. There were no reports of adverse effects. Measurements of colour contrast sensitivity and examination of standard fundus photographs showed that in the 10 participants for whom complete records were available, colour vision improved and the number of red dots (microaneurysms and small dot haemorrhages) decreased. These results were significant even though the trial was short and the numbers treated were small. A second study was carried out by the chief investigator of this grant at King’s College Hospital using electronic sources of 505-nm light-emitting diodes to illuminate one eye. 20 The electrical power of the system was < 3 mW. A total of 40 participants were recruited and follow-up visits were at 3 and 6 months. All participants had early bilateral DMO and the eye with more swollen retinal thickness was chosen as the study eye. Thirty-four out of 40 participants completed the study and the rest did not attend their exit visit appointments and so we are not aware of their mask wear-time. Mean baseline OCT macular cube thickness (Cirrus, Carl Zeiss Meditec AG, Jena, Germany) was equivalent for study and fellow eyes. But study eyes had a greater mean thickness in the central subfield zone 1 (282 ± 53 µm) than the fellow eyes (256 ± 19 µm). At baseline, 28 study eyes showed intraretinal cysts compared with nine in the fellow eyes. At 6 months, only 19 study eyes had cysts compared with 20 cysts in fellow eyes. After 6 months, the worst-affected Early Treatment Diabetic Retinopathy Study (ETDRS) zone and the central subfield zone 1 reduced in thickness in study eyes by 12 µm only (p = 0.01). The secondary outcomes of change in visual acuity, achromatic contrast sensitivity, and microperimetric thresholds improved significantly in study eyes and deteriorated in fellow eyes.

Based on this study, Polyphotonix Medical Ltd (Durham, UK) manufactured the Noctura 400 Sleep Masks designed to suppress rod-related dark adaptation. The acceptability of this lightmask has been reported in normal eyes as well as eyes with DMO. This report was published after the CLEOPATRA (clinical efficacy and safety of light-masks at preventing dark-adaptation in the treatment of early diabetic macular oedema) study was initiated21 and showed no serious adverse events (SAEs) from the masks, although 27% of the participants had withdrawn by the end of the study at 3 months. A further 6-month study conducted in 2012 evaluating the lightmask in participants with diabetic retinopathy with central macular thickness of at least 220 µm were included. 22 Thirty-five out of the 45 recruited participants completed the study. The primary outcomes were safety and mask wear-time. The average mask wear-time was 4.96 [standard deviation (SD) 2.97] hours and showed interindividual variations. A total of 7% of participants experienced mask-related adverse events (AEs). However, the effect of this treatment option needs to be investigated in a randomised controlled trial (RCT) powered to answer the research question of whether or not the lightmask has a superior effect to a placebo. In this project, our main study was a multicentre randomised clinical trial to study the effects of wearing a lightmask to suppress dark adaptation during sleep at night versus a non-lightmask arm in participants with non-central DMO over 24 months. 23

There are only very small observational studies that suggest that hypoxia contributes to DMO and diabetic retinopathy. 15 In the mechanistic evaluation, we explored the concept of hypoxia as a contributing factor in early DMO. We studied the effect of inhalation of 100% oxygen and lightmasks on the retinal morphology measured by OCT and visual function in a small sample of 28 participants with non-central DMO to establish whether or not local regions of retinal oedema correspond to areas of hypoxia, and whether or not functional defects of outer and inner retinal layers and corresponding anatomical changes improved with 100% oxygen or the lightmasks.

Hypotheses

The working hypothesis was that offering lightmasks to wear during sleep at night, compared with not doing so, reduces the thickness in the zone identified at baseline to have maximal thickness (caused by DMO, of which there is clinical evidence) in participants with non-centre-involving DMO at 24 months’ follow-up from baseline, as measured by OCT.

The statistical null hypothesis was that wearing lightmasks does not alter the absolute thickness at the zone of maximum thickness as determined by OCT at 24 months when compared with standard care (non-lightmask).

Statistical alternative hypothesis was that wearing lightmasks alters the absolute thickness at the zone of maximum thickness, as determined by OCT at 24 months, when compared with standard care (non-lightmask). If there was a significant reduction in absolute thickness in the zone of maximum thickness in participants wearing the lightmasks compared with the control arm (from the trial sample by two-sided 5% significance level for the test of zero difference in mean thickness between arms), then the null hypothesis would be rejected.

Trial objectives

The primary objective was to explore whether or not wearing lightmasks during sleep at night reduces, relative to the non-lightmask arm, the maximal zone thickness as measured by OCT in the study eye of participants with non-centre-involving DMO at 24 months.

The secondary objectives were:

• exploration of whether or not wearing lightmasks during sleep at night reduces, relative to the non-lightmask arm, the baseline maximal zone thickness as measured by OCT in the study eye of participants with non-centre-involving DMO at 12 months

• assessment of the safety profile of the lightmasks, including sleep disturbance, using validated questionnaires at 12 and 24 months

• evaluation of the compliance of wearing these lightmasks over 24 months

• evaluation of change in macular thickness in both the parafoveal zones as well as the perifoveal zones by OCT, macular volume, progress to centre-involving DMO and change in morphological characteristics of DMO, proportion of participants requiring treatment for DMO, best corrected visual acuity (BCVA), and the levels of diabetic retinopathy at 12 and 24 months.

The objectives of the mechanistic evaluation were exploration of:

• the changes induced by supplemental oxygen on multifocal electroretinography (mfERG) and scotopic microperimetry as a function of retinal location at 12 months

• whether or not supplemental oxygen affects tests of function in regions of anatomical change and in those of adjacent regions without apparent change to a different extent at 12 months

• whether outer (microperimetry) or inner retinal (mfERG) functional loss is closely associated with structural changes at 12 months

• whether or not the use of lightmasks affects retinal hypoxia in regions of anatomical disturbance, and whether or not a distinction can be made regarding inner and outer retinal functional changes at 12 months

• whether or not the long-term changes induced by lightmasks are similar to changes induced by oxygen supplementation at 12 months.

Trial design

This is a Phase III, randomised controlled, single-masked, clinical trial that evaluated the clinical effectiveness and safety of lightmasks in treating and preventing the progression of non-centre-involving DMO. A mechanistic evaluation studied the role of hypoxia in early DMO.

Target population

The target population, to which inferences from the end of this trial are intended to generalise, is the population of participants with diabetes mellitus who have non-central, non-clinically significant macular oedema (early DMO).

Trial population

The trial population, from which the study sample was drawn, was defined as participants aged ≥ 18 years who attended the ophthalmology referral centres in the UK (n = 15 clinical sites). In subjects for whom both eyes met the inclusion criteria, the eye with the worse visual acuity was included in the study to be the study eye. The fellow eye (non-study eye) was treated in accordance with the NHS standard of care. In subjects for whom only one eye meets the inclusion criteria, that eye was the study eye and the fellow eye (non-study eye) was also monitored and treated in accordance with the NHS standard of care.

Selection and withdrawal of subjects

Selection of participants

Participants were identified from diabetic retinopathy screening programmes and medical retina clinics of the trial sites and their satellite clinics. In addition, participants were referred by other medical retina consultants from other hospitals to the principal investigators (Table 1).

Recruitment strategy

Description of the investigational treatment

The investigational treatment evaluated in the study was the use of lightmasks during sleep at night for 24 months in treating and preventing the progression of non-centre-involving DMO.

The lightmask is manufactured by Polyphotonix Medical Ltd, to deliver 500- to 505-nm light through closed eyelids. The lightmask contains a light-emitting unit of two organic light-emitting diodes (OLEDs) placed within a fabric mask and this lightmask was placed over both of the patient’s eyes.

The fabric mask is made of nylon, polyurethane and polyester. The pod and fabric mask are designed to be thin and flexible and contoured to complement the face and improve comfort for the wearer. The OLEDs are powered by two 3V (CR2450) batteries, which power the device without the need for an external power source or recharging. At the end of the mask’s lifetime of 83 days, a replacement lightmask is required. A new fabric mask was provided with each mask to minimise contamination resulting from continued use.

The mask is time, date and touch sensitive and would only ‘work’ between pre-determined operational windows – typically 20.00 to 08.00 during the lifetime of the lightmask. Within these times the mask can be activated by a light touch. If worn within 3 minutes of activation, sensors on the pod keep the mask illuminated for the night’s therapy. The times at which the mask was worn were logged for compliance analysis. The returned mask contained anonymised data on wear-time that was downloaded to the manufacturer’s portal for producing data for the mask compliance on an individual level.

The lightmask has Conformite Européenne (CE) certification as a 2a medical device and its design and manufacture met the standards of ISO13485. The intensity of the light was approximately six orders of magnitude less than for threshold toxicity and two orders below that which causes a 1% change in the melatonin cycle that drives circadian rhythms. We tested sleep disturbance and daytime drowsiness by the use of validated sleep questionnaires. Compliance with the lightmasks was assumed to be an issue and so all site personnel were informed to stress optimal compliance with all participants and each mask was equipped with a capacitive sensor and memory chip capable of sensing when the mask was worn. These data from the sensor were downloaded and analysed by the company to provide us with an accurate measure of compliance in this trial.

Dosing regimen

The participants were advised to wear the lightmask each night, receiving a maximum of 8 hours’ therapy per night. The optical output of the mask was tuned by the manufacturer to optimise scotopic intensity while minimising photopic intensity. The masks regulate the light output to a constant luminosity x, 60 cd/m² ≤ × ≤ 100 cd/m², which is well below toxic levels of luminosity but of sufficient scotopic intensity to prevent dark adaption. Emission of < 470 nm is < 3% of total output posing little or no risk of harm.

If patient compliance was deemed to be low (< 70% of the maximum therapeutic dose), the patient received telephone reminders and/or counselling from the research team at each site.

Concomitant therapy

All concomitant and current and past therapies in the last 12 months were recorded at screening. Any change in concomitant medications were recorded at each visit. If the clinician observed disease progression and opted to treat the DMO, a variety of options were allowed.

Laser photocoagulation could be performed as indicated for clinically significant macular oedema if the investigator decided that the oedema had deteriorated to require laser after considering the risks and benefits of laser therapy. Laser therapy was arranged to be done at the same visit or deferred to the next visit. Laser treatment was avoided between study visits unless a detrimental effect was anticipated if laser was deferred to the next visit. Regardless of laser treatment, the participant continued to wear the mask until the end of the study. Repeat laser treatment was also done at any scheduled visit but the interval between two laser treatment sessions had to be not less than 4 months apart.

As intravitreal anti-VEGF is the first line option for treatment of centre-involving DMO of > 400 µm, this treatment could be offered to participants if the oedema deteriorated in order to meet the eligibility for current standard of care. Intravitreal steroids could also be given to these individuals as per investigator discretion but the masks had to be worn until the end of the study.

The fellow eye could be treated according to standard of care and this included laser photocoagulation, intravitreal anti-VEGF therapy or steroids.

Pan-retinal photocoagulation to either eye was permitted if high-risk retinal or disc neovascularisation was observed in any visit. The patient had to be seen at 2-weekly intervals until sufficient pan-retinal photocoagulation was applied. The participants continued to use the masks until the end of the study.

Changes in medications related to diabetes mellitus were recorded within concomitant medications.

Cataract surgery was avoided during the period of the study and was recorded as a protocol deviation if the surgery was deemed necessary and performed during the study period.

Randomisation procedure

A patient identification number (PIN) was generated by registering the patient on the MACRO electronic case report form (eCRF) system (InferMed MACRO, Elsevier, Oxford, UK) after consent had been signed. This unique PIN was recorded on all source data worksheets (SDWs) and used to identify the patient throughout the study. Randomisation was via a bespoke web-based randomisation system hosted at the King’s Clinical Trials Unit (KCTU).

Authorised site staff were allocated a username and password for the randomisation system. Once a patient was consented, all baseline data collected and eligibility confirmed, the staff member logged into the randomisation system (www.ctu.co.uk) and clicked ‘randomisation – advanced’ and selected CLEOPATRA and entered the participants details using the unique PIN. The ‘help’ section of the system had video demonstrations to aid new staff in using the system. Once randomised, the system automatically generated confirmation e-mails to key staff, with or without treatment allocation information depending on their role in the study.

Each participant was randomised to one of two arms: lightmasks (treatment) or non-lightmasks (control). Participants were randomised at the level of the individual, using the method of minimisation incorporating a random element. The minimisation factors were glycated haemoglobin (HbA_1c) level (< 8% or ≥ 8%), baseline thickness in excess of 320 µm in parafoveal zones (2 to 5), and study site. As different types of OCT machines were used across sites, having site as a minimisation factor overcame systematic differences between machines.

Participants were randomised into the study only by an authorised member of staff at the study research site, as detailed on the delegation log. Participants could be randomised into the study only once.

Blinding

Control participants were provided with identical non-lightmasks that contain no active light. Primary outcome assessors (optometrists and OCT technicians) remained masked to treatment allocation. The optometrists were the visual acuity examiners and OCT technicians did the OCT scans at all visits and both were masked to the participant study arm. The visual acuity examiners received the participants into the visual acuity lanes with a visual acuity case report form (CRF), study number and detail of study eye and non-study eye to be refracted, but with no previous subject records or CRF by which the subject treatment arm could be identified. Similarly, the OCT technicians received the subjects into the OCT room on a specific CRF that provides details of subject study number and eye to be examined. The subjects were advised at enrolment that they must not discuss the study arm they are in with the OCT or visual acuity examiner. Retinal photographs were graded by masked graders in the independent reading centre at Gloucestershire Eye Unit. This avoided performance and detection bias. We described the completeness of outcome data for each outcome, including reasons for attrition and exclusions from the analysis.

The decision of the Data Monitoring Committee (DMC) was that the trial/senior statisticians would not be blinded to study arm.

Summary of assessments and trial flow chart23

Table 2 and Figure 1 show the summary of assessments and the trial flow chart.

FIGURE 1.

Trial pathway showing flow of participants in the trial. Reproduced with permission from Sivaprasad et al. 24 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Reproduced with permission from Sivaprasad et al. 24 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Colour photographs and OCT had to be carried out before commencing on first laser or anti-VEGF therapy. These tests were allowed to be repeated at the investigator’s discretion during the period of the study. The visual acuity tests with new refraction were repeated before dilating the pupils at baseline to measure intertest variability. The baseline tests that include mechanistic tests could be done over 8 days. The last procedure was the issuance of the mask. The week 1 visit could be either in person at the recruiting site or by telephone. If on the telephone, Pittsburgh Insomnia Rating Score and Epworth Sleepiness Scale (ESS) questionnaires were provided to the patient at screening to be completed at home and returned to the recruiting site in a prepaid envelope.

Trial procedures: informed consent

We designed the patient information leaflets and consent forms with service user involvement. We supplied individuals with as much information as they required to make an informed decision about participation in the study. A copy of the informed consent was given to the participant for review before consent was obtained by the investigator. All participants were required to read, sign and date a consent form before participating in the study. All information about participants was collected during the course of the study and was not derived from existing databases.

Optical coherence tomography

The primary outcome was change in the zone of maximal retinal thickness on OCT. OCT is a well-established tool used as an assessment and monitoring tool for DMO. The OCT thickness map is divided into nine zones with the central zone representing the centremost area of the retina (the fovea). The DMO trials to date have used this central zone as the outcome measure because the trials are on centre-involving DMO. However, in non-central DMO the central zone is not affected and disease progresses to the centre over time. Therefore, we used the retinal thickness of the baseline zone of maximal retinal thickness as a measure. We also measured, as a secondary outcome, the thickness of parafoveal zones (zones 2–5) and perifoveal zones (zones 6–9) and macular volume to provide further evidence about changes in other zones at 24 months or at the point of withdrawal. For participants with the same maximum baseline retinal thickness in two zones, the zone located in the parafoveal zone was chosen. When these two zones were in the parafoveal zone, the average retinal thickness was taken in subsequent follow-up measurements. The OCT macular thickness protocol was carried out twice at 12 and 24 months each, to assess test–retest variability. The OCT technicians were masked of the treatment arm. When research appointments had been missed, OCTs performed as part of routine clinical care within the visit time frame were allowed to be used.

Visual acuity tests

The visual acuity tests were carried out using the validated ETDRS vision charts using standard operating procedure (SOPs) for trials in DMO. Refracted visual acuity tests were carried out at baseline and at the point of withdrawal. At baseline, following recording of refracted BCVA, the patient could complete another assessment that did not require pupil dilatation and then return to repeat visual acuity recording with the new refraction to ensure that we accounted for intertest variability. Refraction was not repeated the second time. The optometrists were masked of the treatment arm. At baseline, the optometrist was not allowed to have the recorded visual acuity from the first BCVA test when conducting the second test. The second test could also be done by another visual acuity assessor.

Retinal colour photographs

Colour fundus photographs were carried out at baseline, 12 and 24 months or at the point of withdrawal to explore progression of diabetic retinopathy. The photographs were read by masked graders at the Independent Reading Centre in Gloucestershire Eye Unit.

Sleep and insomnia rating scales

We had additional tests to explore sleep disturbance in this study. We used the validated Pittsburgh Insomnia Score Index questionnaire to assess insomnia. 25 Daytime sleepiness was measured by the ESS, which is another validated self-administered questionnaire. 26 Both questionnaires were administered at baseline, week 1, 12 and 24 months or at the point of withdrawal.

Mechanistic tests

A total of 28 participants who consented to the mechanistic evaluation underwent further tests in the baseline visit. Oximetry, mfERG and scotopic microperimetry were carried out at baseline while breathing either air or oxygen through a face mask. The test began 10 seconds after the gas flows and continued for the length of the test. The tests were repeated at 12 months. Half of the participants used the lightmasks and the other half used the non-lightmasks. Blood pressure and intraocular pressure were also measured in these visits. A within-visit flexibility of 14 days was allowed for these participants to complete all the tests in these visits.

Laboratory tests

Laboratory tests for levels of HbA_1c for all participants were carried out at the local laboratories at each site. Results of HbA_1c levels from within 3 months of the visit date were also accepted.

Statistics

Justifications for the sample size calculation

Determination of the primary outcome variability

The SD of the change from baseline in retinal thickness was anticipated to be 35.68 µm, based on the previous pilot trial of this intervention. 19

Clustering of outcomes from eyes within subjects effects

Only one eye per subject was selected for the study despite the fact that the lightmasks covered both eyes. Therefore, clustering did not need to be accounted for in the sample size calculation.

Power to detect effects

The power was chosen to be 90% and was based on a two-sided, unadjusted, unpaired t-test at the 5% level of significance.

Determination of the sample size based on the primary outcome

The sample size was set to be 300 subjects (150 per arm). The target of 240 followed up with data for analysis having the primary outcome allowed for 20% attrition. This was based on the previous study in the same population.

The statistical package used for the calculated sample size was nQuery Advisor (version 4.0; Statistical Solutions, Saugus, MA, USA).

Detectable effects sizes expressed in general standardised form

For secondary outcomes, such as change in visual acuity, a medium effect size difference in means (of size 0.42 of a SD) between the arms, based on the unpaired t-test at the 5% significance level could be detected with 90% power if 240 participants were followed up with data for analysis.

Factors contributing to improved power and precision

It was anticipated that, relative to the methods specified in the sample size calculation, there would be an improvement in power and in the precision of estimated treatment effects on each outcome because the statistical analysis approach would incorporate repeated measures data of each outcome and with adjustment for the baseline corresponding to each outcome.

Analysis

A detailed statistical analysis plan (SAP) was set out to test the study objectives and hypotheses, describe the analytical approaches and procedures necessary to address these for the main trial paper and to provide guidance for further research reported in other papers, promoting consistent approaches and methods. The SAP was updated whenever the DMC recommended any changes based on accruing data or new literature being made available, and the changes were made after approval by the Trial Steering Committee (TSC).

As there can typically be more than one analytical approach to address a hypothesis, there is the potential for different results to be produced from using alternative approaches, alternative methods, alternative outcome definitions and the alternative data that may be involved. These differences can be influential, for example when results are of borderline statistical significance. Therefore, the SAP transparently recorded these decisions that were made about study hypotheses, outcome definitions and statistical procedures, along with their basis and the appropriateness of the assumptions required for their use, in advance of the main trial analysis. Changes within subsequent versions of the SAP prior to analysis were dated, with the basis for the changes reasoned, and recorded as an updated version and reasons justified.

It was also not intended that the strategy set out in the SAP should prohibit sensible practices. However, the principles established in the SAP were planned to be followed as closely as possible when analysing and reporting the trial.

Trial sample

Primary outcome

The purpose of the CLEOPATRA trial was to evaluate whether or not prevention of dark adaptation using lightmasks is safe and prevents the progression of early DMO. The principal research question was as follows: does wearing lightmasks during sleep at night decrease or slow the progression of early (non-central, non-clinically significant) DMO? Therefore, the primary efficacy measure was the change from baseline to 24 months in retinal thickness at the zone of maximum thickness at baseline in the study eye. It was measured using standard OCT at baseline and 24 months. For the 12- (secondary) and 24- (primary) month measurements, OCT tests were each done twice to obtain a more precise measurement of retinal thickness.

The trial had two arms, with equal allocation of participants in a 1 : 1 ratio to lightmask treatment or no treatment. After baseline, assessments were carried out at months 4, 8, 12, 16, 20 and 24 (when the last follow-up visit was carried out).

Trial secondary measures

Secondary efficacy outcome measures

The secondary efficacy outcome measures are listed below according to the type of the variable.

Continuous outcome variables

The change in retinal thickness of the baseline zone of maximum thickness at 12 months from baseline.

The change in BCVA at 12 and 24 months was measured using validated ETDRS vision charts. The BCVA test was carried out twice at baseline to reduce the intertest variability and the average BCVA was used in the analysis.

Three secondary continuous outcomes that complement the primary outcome at 12 and 24 months were as follows:

• Retinal thickness in the central zone (zone 1) of the macula (Figure 2).

• Retinal thickness in centre and parafoveal zones as measured by the sum of the five regions (1–5) of the macula (see Figure 2). This outcome captured thickness not just in the important central zone but also across the other four neighbouring parafoveal zones.

• Retinal thickness as measured by the sum of the nine regions (1–9) of the macula. This outcome captured thickness across the macula.

FIGURE 2.

Retinal thickness zones numbered 1–9 in the right and left eye. Reproduced with permission from Sivaprasad et al. 24 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Reproduced with permission from Sivaprasad et al. 24 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Time-to-event outcome variables

Time in months to occurrence of centre-involving macular oedema as defined by > 300 µm on OCT over the 24 months of the study.

Categorical outcome variables

The categorical outcome variables included:

• Progression to clinically significant centre-involving macular oedema as defined as thickness of > 300 µm on OCT occurring within 12 and 24 months.

• Requirement for macular laser treatment, anti-VEGF or steroids treatment at 12 and 24 months.

• Progression of retinopathy as assessed by the independent reading centre as changes in severity of ETDRS scale at 12 and 24 months.

• Microaneurysm turnover of six or more in a ratio of appearance versus disappearance of microaneurysms, using computerised software at 12 and 24 months. The software was not available for this analysis.

• Change in status (yes/no) of each morphological characteristic of macular thickening on OCT from baseline to follow-up at 12 and 24 months.

Other secondary outcome measures

1. Change in HbA_1c level at 12 and 24 months.

   As those developing high levels of HbA_1c may have an increase in retinal thickness, it was important to make sure that there was not an unequal number of participants with high HbA_1c levels between the arms. Some participants with high levels of HbA_1c were treated by their general practitioners, and these medications and other concomitant medications were recorded in each arm.

2. Compliance measures.

   Compliance with the lightmasks was determined at 12 and 24 months using an electronic device embedded in the mask, which quantified the time (in minutes) the masks were used each night. As the number of sleep hours generally varied quite considerably from person to person and within person, and the duration of light varied by season, a pragmatic decision of 6 hours (360 minutes) in a day was taken to be sufficient to represent the 100% compliance within a single day and, therefore, also represented the level at, and above which, maximal benefit is derived. An observed duration of > 6 hours (> 360 minutes) on a single day was taken to be 100% compliance for that day.

   For each participant, compliance was measured primarily over the 2 years that the mask was offered to the patient provided the masks were returned at their next 4-monthly visit and the anonymised data in each mask was sent to Polyphotonix Medical Ltd to convert into the number of minutes worn per day. A compliant participant was defined for this study (and accepted in the grant application) as masks that were worn 70% of the time. In practice, this means that over 2 years, the mean of the daily percentages (as defined above, being truncated to the 6-hour maximum of 100% on a day where necessary) needs to be ≥ 70%.

   This pragmatic approach (i.e. an upper truncated value for 100% and the mean across days) has been adopted in smoking cessation trials where there is daily measurement of compliance to prescribed nicotine replacement therapy. 28

   As an illustration, if a participant wore the mask for at least 6 hours on ≥ 70% of the days, then this participant would be compliant. An alternative participant could wear the mask for as little as 70% of 6 hours (i.e. 4.2 hours, every day) and would also be compliant. More specifically and realistically, a participant who accumulates ≥ 70% (in whatever way over time) of the truncated (or available to be counted) 6 hours per day across the 2-year period, was defined to be compliant. Six hours per day over 2 years (i.e. 730 days) is 4380 hours, and 70% of this is 3066 hours.

   For the purpose of DMC monitoring, compliance was also calculated in each of the six successive 4-month periods in the trial to match the timing of the outcome measures, and using the same approach, based on the mean of the daily (truncated) percentages within the period, and applying a 70% threshold. If compliance data were missing on a day, then it was assumed that there had been no compliance.

Safety monitoring measures

Safety outcome measures (ocular and systemic AEs) were graded and reported by arm to the DMC as follows:

• Presence of AEs, ADEs, SAEs, SADEs, SPRAEs and USADEs. All AEs were also graded by intensity (mild, moderate, severe) and by relationship to treatment (none, remote, possible, probable and definite).

Timing of measures

Retinal thickness was measured by OCT at baseline, 4, 8, 12, 16, 20 and 24 months.

Refracted BCVA as well as retinal colour photographs were measured at baseline, 12 and 24 months. The measures of sleep disturbance in terms of daytime sleepiness and insomnia were to be evaluated at baseline, 12 and 24 months.

Participant duration in the study

Each subject was to participate in the trial for 24 months.

Final assessment

The final study assessment was done after the last study subject completed their 24-month assessment.

Subgroup variables

Subgroup analysis was carried out by baseline HbA_1c level (< 8% or ≥ 8%).

Outcomes requiring derivation

The source of derivation codes for the list of outcomes on sleep-related questionnaires are shown in the following sections.

Missing items in scale and subscales

The number (%) of participants with complete data for each scale was to be reported. The missing value guidance given by ESS and PIRS_20 scales was used.

Use of data transformation

It was not anticipated that consideration for transformation of any continuous outcomes would be needed because of the reasonably large sample size for group comparisons in the main trial analyses. Assumptions of normality and constant variance required by the models were examined using residual and other diagnostic plots. If relevant, and necessary, a log-transformation would be considered because this retains a sensible interpretation for inferences between arms. This was anticipated to be more relevant in the mechanistic evaluation, in which sample size was smaller, and for measures of areas and volumes. However, if an absolute interpretation was important, then rather than considering data transformation, a non-parametric bootstrap method for obtaining CIs was considered. 29 For the mechanistic evaluation, the analysis plan included consideration of non-parametric methods.

Defining outliers

Outliers are observations that have extreme values relative to other observations observed under the same conditions. An outlier was defined here as a data point being at least four SDs from the mean of its distribution of values observed across other participants. This definition would apply to the transformed scale for those outcomes that may have been log-transformed.

Handling outliers

Sensitivity analysis would be undertaken to check whether or not the outlier is influential by obtaining results with and then without inclusion of the outlier. If the conclusions are changed, then this will be noted. Outliers in key outcomes were reported back to the study team after DMC analysis reporting.

Comparison of rates of adherence and follow-up

Compliance was assessed in the lightmask arm.

Non-attendance at any visit and withdrawal rates were to be calculated per arm and compared between arms using Fisher’s exact test.

Analysis covariates

Baseline

The corresponding baseline measure for a continuous outcome is also often predictive of the outcome at follow-up, whereas standard errors (SEs) of statistics derived from binary outcomes vary little with the prevalence to offer gain in precision while expending degrees of freedom. Therefore, ‘baseline’ was included as an additional covariate when modelling only continuous outcomes wherever data were available (e.g. for retinal thickness and visual acuity measures).

Participants with missing data in the baseline of a continuous outcome were accommodated in the analysis using the missing indicator method of White and Thompson. 31

Primary outcome analysis

Missing data assumptions required to be made to interpret the model

This analysis involved fitting a suitable statistical model to all the data that had been observed over the repeated (4-monthly) time points recorded for the outcome, and the valid interpretation of the results from this model implicitly assumes that outcome data that are missing from participants are so-called ‘missing at random’ (MAR). As the outcome was measured repeatedly over time, this means that participants with ‘missing data’ were assumed to have outcomes equal in distribution to participants with ‘observed data’, conditional on the (i.e. for the same given) baseline and 4- to 24-month follow-up measurements of the outcome variable (and controlling for the same impacts over time of minimisation factors and outcome correlation) estimated from the participant data that were included in the analysis. 33

Reasoning behind the chosen model versus alternatives

Sensitivity analysis for missing data

An expert missing-data group concluded that rather than statisticians reacting to missing data at the end of a trial, there should be comprehensive, proactive planning for handling missing data at the stage of designing trials. 34 The group recommended that there should be consideration of missing data mechanisms (e.g. MAR) and, if the missing data may be informative, that appropriate sensitivity analyses should be undertaken to investigate the robustness of the inferences to the different assumptions made by the main analysis. It has also been recommended that analyses allowing for non-response and low intervention uptake (or compliance) are best specified in advance and included in the analysis plan. 35 In this section we consider the handling of missing data, and in Sensitivity analysis for non-compliance we consider handling low compliance.

In the funded grant application for our trial, a sensitivity analysis was specified, which would be additional to, and nested within, the multiple imputation. Given the change from multiple imputation to the LME model, the sensitivity analysis was instead applied to examine robustness of the LME results to the MAR assumption, with the aim of adequately exploring the impact of departures from the MAR assumption33 on the primary outcome results.

For the sensitivity analysis, we prespecified a range for retinal thickness from –20 µm to +20 µm over which the mean of the ‘unobserved outcome data’ might depart (or be different) from the mean of the ‘observed outcome data’. 33 In other words, this range could be defined as how much a typical subject with missing data may, on average, have had a different estimated treatment effect compared with the corresponding subject with the outcome data observed (given the same baseline covariates and follow-up data in the LME model). The range (–20 µm to + 20 µm) is chosen to represent both negative and positive departures that could potentially arise as the ‘net effect’ of alternative reasons, which may be unknown, such as dropout caused by no anticipated further improvement or dropout caused by no improvement so far, together with no anticipated achievable improvement.

This range of 40 µm (from –20 µm to +20 µm) is generously wide for exploring sensitivity of the main results to departures from the MAR assumption, because 20 µm (as the maximum departure in either direction) is larger than the detectable between-arm treatment effect (difference in means) of 15 µm that is set in the sample size calculation, and because it represents over half of the SD of 35.68 µm, which is a sizeable shift in the mean of the distribution for dropouts compared with completers. The generous width would also cover the situation that there may be some bias in the estimated treatment effect from having chosen this LME model for the main analysis.

At the end of the trial, the fractions of individuals with missing data for retinal thickness at 24 months was available in each arm f_i (for intervention) and f_c (for control). The parameter representing excess retinal thickness in those missing compared with those observed, δ, included values across the range –20 µm to +20 µm. Three scenarios were undertaken within the sensitivity analysis. 33,35 These reflect whether departures from the MAR assumption apply within the intervention arm only (lightmasks), within the control arm only (non-lightmasks) or within both arms equally and in the same direction (thereby potentially cancelling out across the sensitivity range, if the dropout rate were to be the same in both arms).

Scenario 1: the treatment effect from the LME model will be increased by f_iδ.

Scenario 2: the treatment effect from the LME model will be increased by –f_cδ.

Scenario 3: the treatment effect from the LME model will be increased by (f_i–f_c)δ.

Sensitivity analysis for non-compliance

Incomplete uptake of trial interventions often means that randomised groups have more similar experience than the investigators had intended, which usually causes the difference in outcomes to be smaller than it would have been with better uptake.

White et al. 35

For this trial, this means that low compliance with the active lightmasks would bring about two groups that have more similar intervention experience to each other than was planned (i.e. intervention more similar to control through low compliance).

We could use compliance to make the MAR assumption more plausible by jointly modelling the continuous (percentage) compliance variable as an additional outcome variable in the LME model alongside the post-baseline 4-monthly retinal thickness outcome measures, where compliance has a natural zero (equivalent to intervention uptake in the control arm) and would need to be standardised to have the same SD as the 24-month retinal thickness outcome. The model was extended with an extra parameter such that the residual SD in the model was allowed to be different in each arm of the trial. 35

However, the purpose of incorporating compliance is to estimate a causal effect of treatment among compliers. This purpose arises from the CLEOPATRA trial being funded by the Efficacy and Mechanism Evaluation programme of the National Institute for Health Research (NIHR), where there is a need to understand what the treatment effect might be if compliance were adequate. ‘Efficacy’ is of interest from this additional analysis, rather than ‘effectiveness incorporating low compliance’. In the application, a complier average causal effect (CACE) of treatment analysis, under a MAR assumption, was specified, as described in Dunn et al. 36

Further to the previous sensitivity analysis, we conducted an analysis estimating the effect of lightmasks versus non-lightmasks on the primary outcome (retinal thickness) in a complier population, while respecting randomisation. This approach provided a better estimate of the true effect without suffering from potential biases seen in a per-protocol analysis.

A CACE analysis was carried out as recommended and outlined by Dunn et al. 36 The CACE estimate is the comparison of the average outcome of the compliers in the lightmasks arm with the average outcome of the comparable group of ‘would-be compliers’ in the non-lightmasks arm (not requiring measurement of compliance in the control arm). A participant was assumed to be a complier, if he/she wore the mask for at least 70% of the time over the 2-year period, where a maximum of 6 hours is taken to be 100% on a single night. A ‘would-be complier’ is then a person in the control arm who would have complied with their treatment allocation had they been allocated to the intervention arm. To determine the CACE estimate, the ‘exclusion restriction’ criteria was anticipated, which assumes that the mean retinal thickness for non-compliers in the control arm is the same as that for those in the treatment arm (i.e. there is no effect of the ‘label’ of one’s group on outcome among non-compliers).

Sensitivity analysis to the requirement for treatment of centre-involving macular oedema

Some participants were anticipated, during the 2-year study period, to develop centre-involving macular oedema at the level of 300 µm. If this reached 400 µm they would become eligible for treatment according to National Institute for Health and Care Excellence (NICE) technology appraisal guidance although some may be treated at a higher level depending on clinician discretion. After treatment, the retinal thickness and the primary outcome may be reduced for this reason rather than being due to the randomised treatment alone.

Therefore, we undertook two sensitivity analyses to examine the robustness of the results of the outcome findings. In these sensitivity analyses, we used 400 µm as a treatment-related value because it is not dependent on the clinician’s decision but on an external fixed value from NICE.

First, for a patient receiving treatment before the 24-month end point, the retinal thickness measurement taken just after the patient first reached 400 µm was considered in the analysis to carry forward to be their final measurement in this sensitivity analysis. For all other participants, the retinal thickness reached at the 24-month visit were retained.

Second, we undertook a time-to-event analysis for the time to requirement of treatment (> 400 µm) using stratified Cox proportional hazards regression incorporating the minimisation factor HbA_1c. ‘Study site’ was not included because there would be too many strata for the number of expected events to be reliably modelled.

Sensitivity analysis for large changes (i.e. > 50 µm) in the zone of maximum baseline retinal thickness

Given the observation that a subject’s retinal thickness may suffer drops or increases of > 50 µm, the variability in retinal thickness could be different across time, and potentially between treatment arms, and so the non-parametric Mann–Whitney U-test was made available in this case to be used as a sensitivity analysis to compare retinal thickness between arms at 24 months. This was applied to the primary outcome in the SAP version 5.2.

Interim analysis

There was planned to be no formal interim analysis.

Secondary outcomes analysis

Analysis of binary outcomes

For the binary outcomes, such as the proportion of participants progressing to clinically significant centre-involving macular oedema at 24 months, a difference in proportions was used to compare arms, with 95% CIs and Pearson’s chi-squared tests, or Fisher’s exact tests and Wilson’s exact CIs when any expected table counts were < 5. When required (e.g. small overall numerator), exact unadjusted CIs were estimated for proportions and differences between proportions.

Analysis methods for secondary outcomes

For the secondary outcomes, the analyses in Table 5 were planned.

Substudy on mechanistic evaluation

The concept of hypoxia as a contributing factor in early DMO was explored in this substudy. The main objective of this substudy was to evaluate the effect of inhalation of oxygen and lightmasks on the retinal morphology and visual function. It was originally planned for this substudy to be carried out with 30 participants of the main study in Moorfields Eye Hospital. However, as the tests were time-consuming, it was difficult to consent participants for the substudy. Therefore, the study was carried out by conducting these time-consuming tests on 28 participants to establish whether or not local regions of retinal oedema corresponded to hypoxia, and whether or not functional defects of outer and inner retinal layers are associated with corresponding anatomical changes. The additional tests included oximetry, mfERG and scotopic microperimetry at baseline and at 12 months. At baseline, these tests were carried out twice: either with breathing air (normal situation) first, followed by participants having their measurement taken while breathing 100% oxygen through a face mask, or vice versa to ensure that the tests were completely done in random order.

Of the 28 participants, half were from the group randomised to use the lightmasks, and the other half from those randomised to use the non-lightmasks.

Comparison of the two baseline tests in all of the 28 participants, were prior to use of any masks, and addressed questions relating to the effect of inhalation of oxygen relative to normal conditions.

Mechanistic evaluation outcomes and correspondent analysis

For each continuous outcome assessed in the mechanistic evaluation substudy, means (SDs) or medians [interquartile ranges (IQRs)] of changes from baseline to 12 months were used to describe all participants. Table 6 shows the outcomes and corresponding analysis.

Handling multiple comparisons

All study analyses were based on tests that are two-sided, including the two-sided 95% CIs.

Significance tests were used sparingly and restricted when possible to addressing stated hypotheses. Secondary outcomes, as well as the primary outcome, were summarised using an effect size with a 95% CI. Interpretation for those secondary outcomes that did not directly address the stated study hypotheses were more cautious.

Software

The principal software package was IBM SPSS statistics 23 (IBM Corporation, Armonk, NY, USA) with the R software (The R Foundation for Statistical Computing, Vienna, Austria) planned to be available.

Data Monitoring Committee monitoring

The DMC monitored study power and the study statisticians regularly provided the DMC members (see Appendix 2) with information such as non-compliance, withdrawal and variability of the primary outcome as an increasing proportion of the participants pass each of the 4-monthly measurement points.

Data handling

The chief investigator is the custodian for the trial data. Personal data were regarded as strictly confidential. To preserve anonymity, any data leaving the site identified participants by their initials and a unique study identification code only. No identifiable patient data left the study site. The study complied with the Data Protection Act 1998. 38 All study records and investigator site files (ISFs) were kept at site in a locked filing cabinet with restricted access. Any breach of confidentiality was minimised by adherence to the Data Protection Act 1998,38 with reassurance stated on the consent form to minimise any potential distress.

Data management

Data management was consistent with Medical Research Council Guidelines for Good Clinical Practice in Clinical Trials39 and the Data Protection Act. 38 Centre principal investigators ensured that all personnel were familiar with and complied with these guidelines. Data management procedures for the trial were developed and overseen by KCTU.

Electronic case report form

All baseline and follow-up data were entered on the online InferMed MACRO electronic data capture system (www.infermed.com). This system is regulatory compliant [Good Clinical Practice (GCP) and the European Commission Clinical Trial Directive]. An eCRF using the MACRO electronic data capture was programmed by the Clinical Trials Unit in collaboration with the trial manager and trial statistician, and was hosted on a dedicated secure server within King’s College London. The eCRF system had a full audit trail, data discrepancy functionality, database lock functionality and supported real-time data cleaning and reporting. The Clinical Trials Unit provided training, essential documentation and user support to the study centres, as well as on-site audit and monitoring. A detailed SOP covered data recording, online entry, checking, central backup and storage. A regularly updated coding manual was developed to accompany the study database. Each research worker and centre principal investigator had a unique username and password provided by the Clinical Trials Unit for the eCRF. The trial manager provided usernames and passwords to any new researchers. Only those authorised by the trial manager were able to use the system.

Data collection and recording

Baseline data were collected and entered by researchers in each study site prior to randomisation. Each participant was assigned a unique trial PIN at the start of the assessment process. This number was written on all clinical assessment forms, data sheets and databases used to record participant data. Trial data were first entered on to paper source data sheets provided to each centre during the preparation phase. We endeavoured to minimise the use of paper at all times. The data sheets were immediately checked for completeness and accuracy. If data queries arose, these were logged and followed up locally before data were entered online. A hard copy of a record sheet linking patient identity, contact details and trial PIN for all participants was kept at each site. These were placed securely in a locked filing cabinet separate from data sheets. All data were kept secure at all times and maintained in accordance with the requirements of the Data Protection Act 199838 and archived locally according to clinical trial GCP regulations and the host institutions additional procedures.

Quality assurance

The study incorporated a range of data management quality assurance functions. As the data were entered online, the data manager logged any queries generated and fed back to the centre research workers in a timely manner. Maintaining a single point of contact between each centre and the Clinical Trials Unit, the trial manager conducted regular monitoring visits at each centre, checking 10% of entered data for consistency with the written data worksheet. Any necessary alterations to entered data were indicated clearly with an audit trail from the original point of data entry, to ensure that any such amendments, and the reasons for them, could be inspected and tracked.

The study was managed through the KCTU. The Trial Management Group included the chief investigator, operational director, trial manager, data management strategic lead, sponsor representative and other members of the trial team when applicable.

The KCTU provided day-to-day support for the site and provide training through investigator meetings, site initiation visit and routine monitoring visits. The principal investigator was responsible for the day-to-day study conduct at site. Quality control was maintained through adherence to Clinical Trials Unit SOPs, study SOPs, study protocol, the principles of GCP, research governance and clinical trial regulations.

Database lock

After written recording, each research worker transcribed data onto the eCRF within 1 working week of a participant assessment. After completion of all follow-ups and prompt entry of data, the trial manager reviewed the data and issued queries. The research worker answered these queries before the participants’ data were ‘frozen’ within the database. After that time, changes were not made to the database by the centres unless specifically requested by the study office in response to statistician data checks. At the end of the trial, the centre principal investigator reviewed all the data for each participant and provided electronic sign off to verify that all the data are complete and correct. At this point, all data were formally locked for analysis. At the end of the trial, each centre was supplied with a compact disc read-only memory (CD-ROM) containing the eCRF data for their centre. This was filed locally for any future regulatory or internal audit.

Amendments to statistical analysis plan versions

Version 1 was written by the trial statistician on 31 July 2013 and was verified by the senior statistician on 23 August 2013, leading to version 2.

Version 3 was produced on 24 October 2013 after comments from the chief investigator and the trial team.

Version 4 was produced on 31 March 2014 and accounts for comments made at a meeting on 14 February 2014 with the DMC statistician. The main changes were to record that the statisticians would be unblinded in reporting at DMC meetings, that two measurements would be taken of ETDRS at baseline and two of retinal thickness at 12 and 24 months, and that the correlation structure of the analysis of repeated measures continuous outcomes would have an unstructured covariance matrix with covariate contrasts interacting at each follow-up time point. In addition, the principal investigator added a stratifier (whether or not retinal thickness is 320 µm already in any of the inner zones 2–5 at baseline).

Version 5 was produced on 22 May 2014 and this was the first SAP version formally signed off by the DMC chairperson and statistician and TSC chairperson. The method of stratified randomisation was replaced for minimisation because there were too many strata for a reliable stratification.

Version 5.2 was produced in August 2016 after discussions with the trial management team, DMC and TSC in June 2016.

• A per-protocol analysis was added as a secondary analysis because some subjects take steroids, anti-VEGF or laser treatments, which makes retinal thickness improve substantially.

• The minimisation stratifier ‘Baseline thickness in excess of 320 µm in zones 2–5 or not’ was to be removed from the planned analysis model and as a subgroup analysis variable because all but a handful of participants had baseline retinal thickness > 320 µm in zones 2–5.

• A sensitivity analysis to large retinal thickness departures (≥ 50 µm) was added, which incorporates a non-parametric test to the primary outcome.

• The sensitivity analysis for non-compliance would also take into account the 60% and 50% levels of compliance.

• The use of anti-VEGF, steroids and laser at 12 and 24 months was added as a secondary outcome.

• SPSS statistics 23 would be used instead of version 21.

Summary of changes made to protocol

When CLEOPATRA protocol version 1.0, dated 1 February 2012, was first submitted to the REC, the device that was being used was non-CE marked from Keepsight Ltd and, therefore, the trial required MHRA approval. This was applied for in version 2.0, dated 14 February 2013, but the company could not provide the required documents.

We then purchased a device from Polyphotonix Medical Ltd. This device was CE marked and, therefore, no longer required MHRA approval. The protocol was amended to version 3.0, dated 1 August 2013, (substantial amendment 1) and was submitted and approved by the REC because of this and notice was given to the MHRA that the submission for this study was withdrawn.

The changes in protocol included the following.

The lightmasks for this study were changed to be supplied by Polyphotonix Medical Ltd. These masks were already CE marked as a class 2a medical device. These masks have inbuilt software that assessed compliance so a separate compliance data logger was not required. The protocol was amended to incorporate the input from the company. The compliance section was changed to be in line with how compliance was measured for these masks.

The representative of Keepsight Ltd was removed from the protocol and the new suppliers were inserted. The patient information sheet showed the photograph of the new mask and details of the new mask and compliance and the name of the new supplier of the mask. The consent form was amended to show the new version number of the patient information sheet.

A second substantial amendment (substantial amendment 2) was then submitted to the REC for changes in the protocol version 4.0, dated 1 November 2013, that included the addition of five new sites, clarification of the secondary outcomes and altering the participants’ assessments. A DMC member was also removed because she became a principal investigator at an additional site. REC approval was requested and granted for this amendment.

We had received a non-invasive CE-approved oximetry machine for measuring oxygenation within retinal vessels. It gave a more direct measure of oxygenation of the retinal vessels than our originally planned indirect estimation from fluorescein angiography. Therefore, we replaced the original invasive fluorescein angiography test with oximetry. The outcome measures in the mechanistic part were changed to reflect these new measurements. As part of oximetry measurement we also added measurement of blood pressure and ocular examination in the table on summary of assessments. We also replaced the word ‘early’ DMO with ‘non-central’ DMO as ‘early’ is a subjective term while ‘non-central’ is an objective measure of the oedema. This ensured that all sites recruited similar participants into the trial and we could maintain an audit trail on the inclusion criteria. Different sites were using different OCT machines for the outcome measure, so we added ‘macular volume’ as a secondary outcome measure to substantiate or better qualify our primary outcome. We also included patient identification centre sites to enable identification of participants from these diabetic retinopathy screening units. The junior statistician and the research fellow for the mechanistic studies were added. We also formatted the protocol and removed all the typographical errors. These changes were also added to the participant information sheet and informed consent form so the versions were amended to version 4.0 for these documents.

A non-substantial amendment was then submitted to the REC (amendment 3) to remove a sentence from the protocol version 5.0, dated 7 February 2014, that should have been removed previously, to accurately address the TSC (see Appendix 2), to include the company’s video demonstration for participants and instructions for use of the masks provided by the company. A correction of the contents page and typographical errors led to a non-substantial change in protocol to version 5.1, dated 27 February 2014.

A non-substantial amendment (amendment 4) was then submitted to the REC for changes in the protocol version 6.0, dated 16 July 2014, that included the correction of typographical errors, inclusion of an invitation letter and poster to help with patient recruitment.

A final substantial amendment of the protocol version 7.0, dated 3 August 2016 (amendment 5), was submitted and approved by REC and included a change to the address of statistics team, and the addition of per-protocol analysis as a secondary analysis to exclude subjects from the ITT sample who were treated with steroids, anti-VEGF or laser treatments, which makes retinal thickness improve substantially. The baseline stratifier on baseline thickness in excess of 320 µm in zones 2–5 was removed, a sensitivity analysis to large retinal thickness departures (i.e. ≥ 50 µm) was added and a further sensitivity analysis for non-compliance also took into account the 60% and 50% levels of compliance to understand the dose effect of the lightmasks. The use of anti-VEGF, steroids and laser at 12 and 24 months was added as a secondary outcome. The version of the statistical software was amended to SPSS statistics 23.

Setting and locations

The study was conducted at 15 NHS clinical sites (Table 7). The sites were chosen based on previous clinical trial experience or by the estimated volume of potentially eligible participants. Interested site investigators completed a site feasibility questionnaire.

The actual recruitment period was on target to the planned original target of 12 months. Figure 3 shows the actual monthly recruitment compared with the original target.

FIGURE 3.

Cumulative monthly accrual of participants into the CLEOPATRA trial.

Recruitment was completed in 13 months. Table 8 shows the total number of participants recruited from each site.

Study set-up

Key activities

The SIV meeting consisted of training on the protocol, AE reporting, MACRO and randomisation and data entry, checking and monitoring training and an outline of GCP. Sites were given a practical session (when possible) on the information technology (IT) systems being used in the study. The ISF was reviewed with the research team during the SIV. Attention was drawn to key sections including SDWs completion, mask accountability forms and mark order forms. All relevant documentation was also provided electronically to all sites [either on compact disc (CD) or via e-mail]. Training was conducted with staff members who had been identified by the principal investigator as being responsible for programming the masks, database entry and randomisation.

The staff were also shown screen shots of the randomisation system and given guidance on how to use this system. They were given an explanation of the randomisation e-mails and given information on how to effectively randomise and minimise errors.

A demonstration of the MACRO V4 system was also given. The staff members were shown the training site and steps on how to generate a patient PIN. The MACRO V4 user guide was provided in the site file and this document was highlighted to staff during this section of the training.

Site staff were also given access to the dummy database for practice while waiting for the green light.

Mask programming

A demonstration of how to programme the lightmasks was given. The local team was provided with a demo lightmask, which was programmed during this training session. A training version of the PPX works website (Polyphotonix, Durham, UK) was reviewed and the demonstration pod was programmed. Instructions were provided in the site file and these documents were highlighted during this section of the training. Training was also provided on the study laptop, using the dongle provided to gain internet access, on how the compliance data of each returned mask was collected and downloaded. It was highlighted to the site staff that the compliance data are dependent on the mask being returned so that the site staff could be extra vigilant about requesting the return of the used masks.

At the end of the SIV, the following documents were copied for holding in the trial master file (TMF):

• completed delegation of authority log

• curricula vitae (CVs) signed and dated by all staff, evidence of GCP training, if not previously sent (site may send these after the SIV if necessary)

• SIV attendance record.

All clinical procedures were also detailed in an operation manual. The optometrists were certified by the clinical research trial optometrists at Moorfields Eye Hospital. The trial photographers were certified by Moorfields reading centre. These were done outside the SIV meeting.

Each site received a report on the SIV produced by the trial manager. The report was distributed to local staff alongside the greenlight letter. Principal investigators reviewed and signed the report. Copies of the signed report were stored in the TMF and ISF. The trial manager ensured that all relevant documentation and certifications were complete before a greenlight letter was sent to each site to initiate the study access to the IT systems for the trial.

Baseline characteristics

The clinical study results have been published in Sivaprasad et al. 24

When we consider the whole study population, the average age of the participants was 57 years (SD 11 years) and 63% (194/308) were males. Mean baseline maximum retinal thickness was 348 µm (SD 28 µm). A total of 286 (93%) participants had baseline maximal retinal thickness in zones 2–5 while 22 (7%) had maximal retinal thickness in the parafoveal zones 6–9. A total of 183 out of 308 (59%) participants were measured on Heidelberg Spectralis (Heidelberg Engineering GmbH, Heidelberg, Germany), 54 (18%) were measured on Cirrus (Carl Zeiss Meditec, Cambridge, UK), 61 (20%) were measured on Topcon 2000 (Topcon Great Britain Medical Ltd, Newbury, UK) and 10 (3%) were measured on RS3000 (Nidek Co. Ltd, Aichi, Japan). Mean baseline macular volume was 8.89 µm³ (SD 0.86 µm³). The first measurement of BCVA was 84.1 (7.5) and second was 84.4 (SD 7.5) and the average of the two means was 84.3 (SD 7.3) ETDRS letters, which was equivalent to 6/6 Snellen. The mean HbA_1c level was 68.0 mmol/mol (SD 18.8 mmol/mol), and a total of 154 participants (50%) had a HbA_1c concentration of < 8% or 63.9 mmol/l at baseline.

Baseline characteristics were well balanced between treatment arms (Table 9).

There were 62 participants without primary outcome data at 24 months and Table 10 shows that there were no significant differences in baseline characteristics between those participants who dropped out and those that did not except for site, which was already adjusted for in the analysis. This was largely because one site had a high drop-out rate.

Withdrawals

Participants who withdrew from the trial but were still reviewed in clinics with OCT were also included in the primary outcome analysis. Table 11 shows the number of participants included in the primary outcome data.

The flow of patients are shown in the Consolidated Standards of Reporting Trials (CONSORT) flow diagram (Figure 4). 24

FIGURE 4.

The CONSORT flow diagram for the CLEOPATRA trial. Reproduced with permission from Sivaprasad et al. 24 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Reproduced with permission from Sivaprasad et al. 24 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Derivation of the intention-to-treat model population

Participants included in the prespecified ITT LME model were derived as follows: (1) the OCT data were available for 246 participants of 308 randomly assigned participants (lightmask arm, n = 127; and non-lightmask arm, n = 119) at 24 months, and (2) these data included OCT data from routine clinical care obtained from five participants in the lightmask arm and four participants in the non-lightmask arm who attended their clinic appointment but did not attend an intervening research visit. A further 17 participants in the lightmask arm and 14 participants in the non-lightmask arm were included in the model with data from previous time points. Therefore, a total of 277 participants were included in the primary outcome model, 133 in the non-lightmask arm and 144 in the lightmask arm (see Table 11).

Sensitivity analysis

Sensitivity analyses for missing data were carried out to represent three possible scenarios to reflect whether departures from the MAR assumption apply within the lightmask arm only, within the non-lightmask arm only and within both arms equally and in the same direction. All three sensitivity analysis scenarios for departures from MAR assumption showed a non-significant effect in the change in maximal retinal thickness of the lightmask arm over the non-lightmask arm (Table 17).

TABLE 17 - Sensitivity analysis for missing data

Values are adjusted difference between arms in mean maximal retinal thickness (95% CI).

Reproduced with permission from Sivaprasad et al. 24 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Three scenarios reflecting whether or not departures from the MAR assumption apply within:	Retinal thickness range over which the mean of the ‘unobserved outcome data’ might depart from the mean of the ‘observed outcome data’
	–20 µm	+ 20 µm
Scenario 1. The intervention arm only (lightmasks)	–4.3 (–10.5 to 2.0)	3.0 (–3.3 to 9.2)
Scenario 2. The control arm only (non-lightmasks)	–5.1 (–11.3 to 1.1)	3.8 (–2.5 to 10.0)
Scenario 3. Both arms equally and in the same direction	0.2 (–6.1 to 6.4)	–1.5 (–7.7 to 4.8)

Assuming those with unobserved outcome data in one, or other, or both arms would take values as much as a prespecified 20 µm either side of the adjusted observed effect, all of the 95% CIs included 0 and excluded –15 µm, thus confirming that the finding of an absence of a clinically important lightmask effect is robust to missing data (Figure 5).

FIGURE 5.

Figure showing the robustness of the lightmask effect to missing data. Reproduced with permission from Sivaprasad et al. 24 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Reproduced with permission from Sivaprasad et al. 24 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

We also performed sensitivity analysis for non-compliance by CACE analysis. The CACE estimate of the treatment effect for compliers defined by 70% compliance was –4.2 (95% CI –44.6 to 36.1), by 60% compliance was –3.1 (95% CI –32.4 to 26.3) and by 50% compliance was –2.5 (95% CI –26.7 to 21.7). This was based on 277 participants (lightmask arm, n = 144; non-lightmask arm, n = 133). Across these three definitions of compliers, the results were consistent in estimating a small non-significant intervention effect, which was not close to the detectable effect of 15 µm retinal thickness.

For the sensitivity analysis on participants who met the requirement for treatment of centre-involving DMO because the retinal thickness reached 400 µm before the 24-month end point, the retinal thickness measurement taken just after the participant first reached 400 µm was considered in the analysis and carried forward to be their final measurement. There were nine participants achieving 400 µm in the central macula: three in the non-lightmask arm who achieved this point at 24 months, and six in the lightmask arm (two achieving it at 24 months, two at 20 months and two at 12 months). A total of 161 participants were included in this LME model and the adjusted difference between arms was –0.22 µm (95% CI –8.36 to 7.92 µm; p = 0.96). Using stratified Cox proportional hazards regression stratified by the minimisation factor HbA_1c level, the time-to-event analysis (time to requirement of treatment for retinal thickness to be 400 µm or before the 24-month end point) included 259 participants. The hazard ratio was 2.0 (95% CI 0.5 to 8.0; p = 0.33). For the sensitivity analysis for potential differential variability between arms over time in the zone of maximum baseline retinal thickness, there was no significant difference between arms in the change in retinal thickness from baseline to 24 months using the Mann–Whitney U-test; p = 0.38 (n = 246). The sensitivity analysis on participants that had their OCT scans on the Spectralis OCT at baseline also showed no significant difference between arms (Table 18).

TABLE 18 - Sensitivity analysis including only those participants who had the spectralis OCT measurements

A total of 161 participants are included in the LME models involving 12 and 24 months.

Reproduced with permission from Sivaprasad et al. 24 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Time point	Primary outcome: maximum retinal thickness using OCT (µm)					p-value
	Treatment arm, mean (SD); n		Treatment arm, change from baseline, mean (SE)		Adjusted difference between arms (95% CI)
	Non-lightmask	Lightmask	Non-lightmask	Lightmask
Baseline	353.8 (26.0); 92	349.7 (20.6); 91	–	–	–	–
12 months	344.2 (35.2); 73	345.1 (32.7); 74	–9.4 (4.1)	–5.0 (3.8)	0.15 (–9.83 to 10.13)	0.98
24 months	340.3 (28.4);70	339.8 (23.3); 72	–13.9 (4.0)	–9.3 (2.9)	–0.22 (–8.36 to 7.92)	0.96

Summary of findings

The CLEOPATRA trial is the first Phase III RCT that evaluated the Noctura 400 Sleep Mask as an intervention to treat and prevent non-central DMO in a multicentre setting. The results of this study show that offering a Noctura 400 Sleep Mask to wear at night during sleep is not an effective option in the treatment or prevention of progression of non-centre-involving DMO. Although objective assessment of reduction of maximal retinal thickness was our primary outcome, we have made our conclusion based on the primary outcome, per-protocol secondary analysis and five prespecified sensitivity analyses of the primary outcome, as none of these showed any significant therapeutic benefit to wearing the lightmask. In addition, because of the dynamic nature of DMO, we considered several secondary outcomes including reduction in total retinal thickness, macular volume, progression of central subfield thickness to ≥ 300 µm, the proportion of participants requiring treatment for new-onset centre-involving DMO, and the number of participants who were treated with standard therapy during the trial because of worsening DMO. None of the changes in these parameters was significant between arms substantiating the results of the primary outcome. There was also no effect of the lightmask on diabetic retinopathy severity levels.

Both arms showed a gradual mean reduction in the zone of maximal retinal thickness over 24 months. The reduction is within the SD that we used for the sample size calculation. Participants do pay more attention to their diabetic control when they participate in clinical trials. The event rate of progression to centre-involving DMO was also similar to previous reports.

Mechanistic evaluation

This study is the first study to evaluate the effect of oxygen supplementation on retinal function in participants with early DMO. Mean pulse oximetry increased by 2% during 100% oxygen breathing. However, mean retinal arteriolar oxygen saturation was unchanged. This finding suggests that the higher systemic oxygen saturation was not conveyed to the retinal arterioles because of a countercurrent exchange between the central retinal artery and vein, a possible calibration error of retinal oximetry or the arteries are already carrying maximum possible oxygen saturation. 40 Palkovits et al. 41 measured the effect of hyperoxia with a different type of oximeter (from Imedos Systems GmbH, Jena, Germany) and similarly found that the retinal arterioles did not reach 100% oxygen saturation during hyperoxia.

Retinal oxygenation is dependent on the rate of retinal blood flow and the oxygen concentrations in arterial and venous blood. Blood flow, in turn, is dependent on the diameter of the retinal arteries and veins. 41 The diameter of both the arteries and veins did not show a significant change with supplemental oxygen at baseline; however, there was a difference in diameter change of the arteries and veins at 12 months. The mean arterial and venous diameters increased in eyes that used the lightmask, while they decreased in those not wearing the lightmask. Loss of autoregulation of retinal blood vessels is an early change in participants with diabetes mellitus. These study results are in the opposite direction to that predicted, as an increasing blood vessel diameter is typically a sign of increasing severity of diabetic retinopathy and loss of autoregulation. Although these findings may be attributed to a small sample size and technical difficulties with performing these tests, they may also suggest that the inner retinal tissue oxygenation may not be compromised in early DMO. We have not measured the choroidal blood flow or the choroidal oxygen saturation but the fact that there was no change in retinal function with oxygen supplementation suggests that the outer retina may not be significantly affected by hypoxia in early DMO. However, animal experiments have shown significant oxidative stress at the photoreceptor layer,42 although a recent study by Lau and Linsenmeier showed that hypoxia is not a contributory factor for diabetic retinopathy in rodents. 43,44 Therefore, our study results together with the current evidence from laboratory work indicate that further work is required to understand the role of photoreceptors and hypoxia in the development and progression of DMO.

Strengths and limitations

The main strength of this study was that we ensured that the primary outcome was corroborated by predefined sensitivity analysis and secondary analysis to reduce potential systemic biases in this dynamic condition. Other strengths included clear definition of objective end points, publication of the protocol, significant public and patient involvement throughout the study, and strict assessment assured high-quality data and pre-planned analysis for expected non-compliance. The baseline characteristics of the trial population were typical for the intended patient population. The representative multiethnic patient population, together with the multicentre trial design, permit wide generalisability of our results. The primary outcome of retinal thickness was measured twice at the 12 and 24 months time points and the average taken to analyse the outcomes. Similarly, we took the average of two BCVA measurements at baseline to ensure that test–retest variability did not influence the result as the expected differences between arms were small. We took several prespecified steps to increase the robustness of our analysis. We buttressed our primary outcome analysis with several sensitivity analyses. In a study in which we are reliant on asymptomatic participants using an intervention, missing data, suboptimal compliance and drop-out rates can significantly affect the study outcomes. We expected compliance with lightmasks to be an issue based on the phase II study19 and because non-centre-involving DMO is asymptomatic. Therefore, we made multiple efforts to tackle compliance-related issues in this study. We calculated the sample size with a 20% attrition rate, which is higher than most ophthalmic trials. In addition, we allowed for OCTs from clinic appointments to be used in cases for which participants attended a clinic visit and not a clinical trial visit appointment. We had also carefully considered the influence of non-compliance on the potential therapeutic effect of the lightmasks by incorporating a predefined CACE analysis for non-compliance at three levels: –70%, 60% and 50% compliance. The non-compliance was observed as early as 4 months into the trial and across all three definitions of compliers highlighting that offering these lightmasks to wear during sleep is not an effective option for this condition. This study illustrates the importance of the study statisticians planning for all possible outcomes and likely challenges with the analysis. The statisticians were also expertly advised by a very experienced DMC statistician ensuring the robustness of the study outcomes. A full explanation of how the statistical model was designed is provided in Chapter 2. We expect more interventions to be developed to prevent the progression of diabetic retinopathy and DMO and we recommend that considerable planning of anticipated challenges be considered in this area as was done in this study.

We are also grateful to the reviewers of our grant application as it strengthened our study by defining our primary outcome at 24 months rather than 12 months. This 24-month study has clearly shown several outcomes that may not have come to light with a shorter study.

Nevertheless, some limitations should be considered when interpreting the results. The study only shows that offering Noctura 400 Sleep Masks to wear during sleep at night to suppress rod function is not an effective option to treat DMO. It is possible that the retinal illumination achieved with these devices as used in this protocol did not reduce the dark current sufficiently to alter the hypoxic state. The study also showed that offering the lightmasks to participants to use during sleep is associated with suboptimal compliance by 24 months. The electronic recording of compliance was possible only if the participants returned the lightmasks at each 4-monthly clinic visit, and so compliance may have been better than the recorded compliance data suggests.

The study also questions whether or not non-central DMO is the correct patient group in which to test this device. The role of hypoxia is better understood in severe retinopathy. Recent literature on oxygen metabolism in early stages of retinopathy is conflicting. 42,43,45 Oxygen metabolism in different stages of diabetic retinopathy has to be studied further in both animals and humans. Currently, this is challenging because of the technical difficulties of measuring regional variations in oxygen saturation in the retina.

Comparison with existing literature

The studies on the lightmask have all been short-term studies. 19–22 In diabetic retinopathy trials, all interventions should be evaluated for at least 2 years to understand the holistic effect of the intervention on the patient. On average, diabetic retinopathy affects participants for > 20 years and most eyes with mild to moderate retinopathy or non-central DMO deteriorate very slowly. Therefore, clinical trials in this area should be long-acting, with clearly predefined objective end points and should have a control arm. This is the first RCT that has evaluated lightmasks over 2 years, to our knowledge, and we have highlighted that the lightmasks did not show any clinically meaningful benefit at any time point in the study compared with the non-lightmask.

Implications for practice and research recommendations

The CLEOPATRA study shows that the lightmask to prevent dark adaptation is not recommended in clinical practice for DMO or diabetic retinopathy. As recent laboratory experiments have shown conflicting results on the role of oxygen metabolism on microvascular changes in diabetic retinopathy, we do not recommend more clinical trials on lightmasks suppressing rod function until there is very convincing laboratory experiments to suggest that the retinal oxygen consumption for dark adaptation has a detrimental effect on microvascular changes in the retina. We also do not recommend the use of the lightmask to prevent dark adaptation in participants for diabetic retinopathy based on our study results, in which we found no change in diabetic retinopathy severity scale between the study arms. However, our study results encourage more research on interventions for non-central DMO. Currently, several treatment options for early or non-central oedema have not shown clinical benefit. Research in this area should accelerate to tackle the increasing numbers of participants presenting with non-central diabetic oedema given the exponential increase in the worldwide diabetic population. The research should focus on easy to use and cheap options to global benefit. Until new preventative options are made available, control of systemic risk factors remain crucially important as the only effective option for this cohort of participants.

Support

The study was sponsored by NIHR Moorfields Biomedical Research Centre and supported by the UK Clinical Research Network. The research was supported by the NIHR Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and University College London Institute of Ophthalmology, the NIHR Moorfields Clinical Research Facility and the UK Clinical Research Collaboration-registered KCTU at King’s Health Partners, which is part funded by the NIHR Biomedical Research Centre for Mental Health at South London and Maudsley, NHS Foundation Trust and King’s College London. We would also like to thank Maggie Shergill, the Research Manager (Monitoring) and the other team members of the NIHR Evaluation Trials and Studies Coordinating Centre.

Contributions of authors

Sobha Sivaprasad (Professor, Consultant Ophthalmologist) was the chief investigator.

Joana Vasconcelos (Research Fellow, Medical Statistics) was involved in the design and statistical analysis of the study.

Helen Holmes (Trial Manager) was involved in all aspects of the study’s management and monitoring.

Caroline Murphy (King’s Clinical Trial Unit) was a co-applicant and was involved in the design and trial management.

Joanna Kelly (King’s Clinical Trial Unit) was a co-applicant and was involved in the design and data management.

Philip Hykin (Consultant Ophthalmic Surgeon) was a co-applicant and was involved in the design and recruitment of the study.

Andrew Toby Prevost (Professor, Medical Statistics) was a co-applicant and was involved in the design and statistical analysis of the study.

All authors contributed to the first draft, and reviewed, revised and approved the final version of the manuscript.

Publications

Sivaprasad S, Arden G, Prevost AT, Crosby-Nwaobi R, Holmes H, Kelly J, et al. A multicentre Phase III randomised controlled single-masked clinical trial evaluating the clinical efficacy and safety of light-masks at preventing dark-adaptation in the treatment of early diabetic macular oedema (CLEOPATRA): study protocol for a randomised controlled trial. Trials 2014;15:458.

Sivaprasad S, Vasconcelos JC, Prevost AT, Holmes H, Hykin P, George S, et al. Clinical efficacy and safety of a lightmask for prevention of dark adaptation in treating and preventing progression of early diabetic macular oedema at 24 months (CLEOPATRA): a multicentre, Phase III, randomised controlled trial. Lancet Diabetes Endocrinol 2018;6:382–91.

Data-sharing statement

Data are archived at Moorfield Eye Hospital. All data requests should be submitted to the corresponding author for consideration. Access to anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research. The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, the MRC, NETSCC, the EME programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the EME programme or the Department of Health and Social Care.

Resource centres

We would also like to thank the following resource centres for their support of the trial:

• King’s College Clinical Trial team: Gill Lambert, Beverley White-Alao, Oliver Pressey, Negin Sarafraz-Shekary, Evangelos Georgiou and Janice Jimenez.

• The Royal Free London NHS Foundation Trust Pharmacists team: Nicky Heath and Sabina Melander.

• The North East Diabetes Research Network’s Lay Panel.

• Patient-reported outcomes consultant to the study: Clare Bradley, Royal Holloway, University of London.

• Mechanistic evaluation team: Luke Nicholson, Roxanne Crosby-Nwaobi and Lauren Leitch-Devlin at the NIHR Moorfields Clinical Research Facility.

• Certifications for visual acuity and contrast sensitivity: Catherine Grigg and Katherine Binsted at Moorfields Eye Hospital, London.

• Members of the Networc UK: Usha Chakravarthy, Tunde Peto, Simon Harding, Pauline Lenfestey, Savitha Madhusudhan, Clare Newell, Michelle McGaughey, Vittorio Silvestri, Karleigh Kelso, Barbra Hamill, Graham Young, Irene Leung, Peter Blows, Frank Picton, David Parry and Sophie Leach.

Trial Steering Committee

Gillian Hood (independent chairperson; Queen Mary University of London, London, UK); Graham A Hitman (Barts and The London School of Medicine and Dentistry, London, UK); David Crabb (City University, London, UK); Alaistair Denniston (Queen Elizabeth Hospital, Birmingham, UK); Reverend Douglas Lewin (lay member); and Ian Grierson (University of Liverpool, Liverpool, and Polyphotonix Medical Ltd).

Data Monitoring Committee

Sarah Walker (chairperson; Medical Research Council, Clinical Trials Unit, University College London, London, UK), Jackie Sturt (King’s College London, London, UK) and Debendra Sahu (Southampton NHS Trust, Southampton, UK).

Trial Management Group members

Sobha Sivaprasad, Philip Hykin, Andrew Toby Prevost, Joana Vasconcelos, Helen Holmes, Beverley White-Alao, Caroline Murphy, Joanna Kelly and the sponsor representative (Moorfields Eye Hospital, London, UK).