An external pilot study to test the feasibility of a randomised controlled trial comparing eye muscle surgery against active monitoring for childhood intermittent exotropia [X(T)]

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 09/01/20. The contractual start date was in January 2011. The draft report began editorial review in July 2013 and was accepted for publication in January 2014. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Elaine McColl received grants from Newcastle University, fees and expenses from the NIHR Journals Library Editorial Board and expenses for meeting attendance from the National Institute for Health Research (NIHR) Programme Grants for Applied Research (PGfAR) panel during the course of the study. Luke Vale is a member of the NIHR PGfAR and the NIHR Health Technology Assessment (HTA) Clinical Evaluation and Trials Board.

Copyright statement

© Queen’s Printer and Controller of HMSO 2015. This work was produced by Clarke et al. under the terms of a commissioning contract issued by the Secretary of State for Health. 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.

Background

Objectives

The specific objectives of the SamExo study (Surgery vs. Active Monitoring in Intermittent Exotropia), as stated in the original project description, were to:

1. determine whether or not participating centres are likely to recruit a sufficient number of patients to deliver the trial

2. determine whether or not recruited patients will stay within their allocated groups and complete follow-up in sufficient numbers to deliver the trial

3. develop a web-based trial management system to centralise and automate trial processes such as invitation, logging of replies, scheduling of appointments, confirmation of eligibility, randomisation and printing of letters

4. pilot the procedures involved in the trial, including recruitment (giving information and obtaining consent), randomisation, intervention (surgery), masking, and baseline and follow-up data collection

5. monitor potential bias by comparing the demographic and clinical status of patients retained with any withdrawing from the trial, and by comparing those who consent with those eligible but refusing to participate

6. identify through questionnaires and qualitative interviews, where possible, reasons why parents decline permission to participate

7. prepare a detailed protocol and application for funding for such a RCT (if findings from the pilot study indicate that a full-scale RCT is feasible).

Systematic review

The opportunity to apply for further funding for an associated project was offered once the trial was under way. Given the difficulty in synthesising the literature on X(T), and the lack of RCT data highlighted in the Cochrane review, a systematic review of non-randomised studies was proposed and funded, and the output of this work is included in this report.

Background

Intermittent exotropia [X(T)] is a common form of childhood strabismus (squint) affecting approximately two out of every 100 children before the age of 3 years. 35 This particular ocular misalignment is characterised by an outwards deviation of the eye, which is not constant but is usually present initially on distance fixation or when the child is tired. 20 The natural history of X(T) is poorly understood: the ocular misalignment may worsen or deteriorate into constant exotropia, which adversely affects stereo vision; conversely the misalignment may resolve over time. 36 X(T) is also of concern for psychosocial reasons, as the cosmetic appearance might cause the child to develop social or psychological problems,2 which can impact into adult life, with effects on self-image, work and personal relationships. 37

A range of both conservative and surgical treatment options are available and include observation (watchful waiting), orthoptic exercises/vision therapy, occlusion therapy (patching), minus lens therapy (glasses) and surgery. 38 However, surgery is associated with possible adverse effects, including a risk of overcorrection, which may also adversely impact on stereoacuity. Evidence for the comparative effectiveness of treatment options is limited by the absence of RCT data,27 but there is a much larger literature of observational studies for the various interventions. As a consequence of the absence of robust and reliable effectiveness data on treatment options, and of uncertainty about the natural history of the condition, wide service variation exists (both nationally and internationally) in management of the condition.

Aim

The main aim of the review was to determine the effectiveness of surgical and non-surgical approaches as a means for managing X(T) in childhood. Secondary objectives were to (1) understand the circumstances under which particular interventions are most effective; (2) determine adverse effects associated with particular interventions; and (3) better understand the natural history of X(T).

Methods

The review was conducted following best practice guidelines for the design, conduct and reporting of systematic reviews. 39–41

Data synthesis

We used the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidance and flow chart to support reporting41 (Figure 1). Studies can be pooled only if a number of studies are identified reporting the same primary outcome measure and there is sufficient homogeneity in terms of study design, intervention type and population. We were unable to synthesise any data quantitatively in the form of a meta-analysis due to high levels of study heterogeneity. Differences in study design, population, intervention and outcome measures precluded meta-analysis. In particular, there was a great deal of divergence in defining what constitutes a positive outcome. For example, change in angle of deviation versus with stereoacuity versus degree of control and how success was defined [e.g. deviation of > 20 prism dioptres (PD)] compared with deviation of < 10 PD, which is clearly likely to impact on comparable effectiveness between studies) and at what time point measurements were undertaken. Equally, there were differences in the study populations, for example some studies included only divergence excess type, whereas others grouped basic, true and simulated divergence excess types together.

FIGURE 1.

Flow chart of search process (based on PRISMA guidelines).

Because heterogeneity between studies precluded meta-analysis, we performed a narrative synthesis and report study findings separately by intervention type. In the narrative synthesis, we describe the main characteristics of the studies included, along with information about study quality and estimates of effect with relevant statistics. To avoid the introduction of bias into the narrative synthesis we have aimed to report the findings of each study judiciously and have made efforts to avoid inappropriate emphasis on any one particular study or author.

Findings

Discussion

Conclusions

Although being mindful of the limitations of observational data to compare the effectiveness of different treatment options, the findings suggest that in some circumstances R&R surgery can produce better outcomes than BLR surgery; however, there is a question about the sustainability of success with the possibility of regression over time. The extent and quality of the evidence have clear limitations that future studies, ideally RCTs or well-designed prospective controlled intervention studies, should address. We look forward to the results of the North American RCTs on efficacy of type of surgery, BLR compared with R&R,29 and observation therapy compared with occlusion therapy for treatment of X(T). 30 The implications of this review for research and practice are summarised in Box 3. To echo the conclusions of Hatt and Gnanaraj,27 consensus is needed as to what constitutes the ideal measure of success, as well as agreement on how this should be measured. Related to this point is the time at which outcomes should be measured post intervention (e.g. 6 months, 1 year, 2 years) as we recognise that surgical outcomes in particular are, to some extent, plastic, with possible drift over time.

Acknowledgements

We are very grateful to Fiona Shrive, Omar Hadid, Alan Connor and Umear Ahmed who were involved in the early phases of the review, including contributing to design of the search strategies and screening. Many thanks to Meng Fan, Svetlana Glinianaia, Dominika Kwasnicka, Ibrahim Masri, Lavinia Miceli, Christian Walter and Susie Wong for assistance with translation of non-English-language papers. We would also like to thank Alison Etherington, Ann Payne and Carolyn Wake for administrative support.

Trial design

The SamExo trial was a rehearsal pilot RCT to assess the feasibility of a full RCT of the effectiveness of surgical treatment against active monitoring in X(T). The trial was conducted according to recommendations for good practice in pilot studies. 64 Fuller detail about the protocol can be found in our publication. 65

Changes to trial design

Participants

Children aged between 6 months and 16 years, who were referred to the clinics with suspected X(T) from community screening, general practice or other health-care professionals, and subsequently diagnosed with X(T), as well as existing patients fulfilling the eligibility criteria, were eligible for the SamExo study. The parent/guardian provided written informed consent for participation in the study prior to any trial-specific procedures.

Data were anonymised by use of a unique trial identification number (ID) assigned by the online trial data management database post consent.

Inclusion criteria

• Age between ≥ 6 months and ≤ 16 years.

• Evidence of X(T) on the basis of parental history and clinical examination.

• No ongoing or planned amblyopia treatment.

• VA of 0.500 or better on an age-appropriate logMAR (logarithm of minimum angle of resolution) test or, where uniocular testing is not possible, central steady maintained fixation when one eye is occluded.

• NCS61 of ≥ 3.

• Minimum of 15 PD misalignment in the distance.

• Presence of near stereopsis documented using the preschool Randot test if ≥ 3 years of age.

• If < 3 years old must be able to overcome a base-out prism (10, 15 or 20 PD).

Exclusion criteria

• Age > 16 years.

• Previous treatment for X(T).

• XT (other than microtropia).

• VA of > 0.500 logMAR in either eye.

• X(T), where near misalignment is > 10 PDs more than the distance misalignment.

• Structural ocular pathology.

• Significant neurodevelopmental delay.

• Families planning to move out of area.

Study setting

Four secondary ophthalmology care facilities at The Newcastle upon Tyne Hospitals NHS Foundation Trust (co-ordinating site), City Hospitals Sunderland NHS Foundation Trust, Moorfields Eye Hospital NHS Foundation Trust and York Hospitals NHS Foundation Trust, each of which are large centres with specialist paediatric ophthalmology clinics.

The sites (York, Sunderland and Moorfields) were chosen on the basis of their willingness to collaborate with the study and previous experience of collaboration on similar studies.

The patient demographics were significantly different at Moorfields, with a higher variation in ethnicity of the patient population than in the other three sites.

Initial site visits were carried out to each site in order to introduce the study protocol and test procedures. Planned recruitment figures were discussed and any possible issues flagged up early on in the set-up phase of the study.

Feedback from the initial meetings was positive; however, the York and Sunderland sites did express reservations regarding their ability to recruit to target.

Interventions

Clinic appointments

Figure 2 illustrates the schedule of study visits and corresponding assessments for the active monitoring and surgery arm. The assessments involved routine clinical measurements together with the evaluation of QoL using the IXTQ66 (see Appendix 5), a Health Services Use Questionnaire (HSUQ) and a Time and Travel Costs Questionnaire (TTQ) (see Appendix 7). Children in the active monitoring group were offered surgery if a constant strabismus (XT) appeared to be developing or parents requested surgery and the responsible clinical team agreed that this was appropriate. Constant exotropia was defined as NCS 9 with no demonstrable binocular single vision.

FIGURE 2.

Summary of study visits, assessments and interventions. HSUQ, Health Services Use Questionnaire; RO, research orthoptist; TTQ, Time and Travel Costs Questionnaire.

Eye muscle surgery

Surgery was performed by the local PI or delegated deputy, in accordance with agreed surgical formulae tailored to the clinical characteristics of the strabismus and the usual practice of the surgeon. Principles involved in the surgical treatment of children in the study were agreed as follows:

• general anaesthesia

• BLR surgery to be performed for true-distance exotropia

• unilateral recession/resection surgery to be performed for other types of exotropia

• standard sterile preparation of the operative site

• conjunctival incisions

• standard isolation and cleaning of muscle to be operated

• muscle secured with 6/0 VICRYL^® (polyglactin 910) suture (Johnson & Johnson, New Brunswick, NJ, USA)

• amount of R&R assessed on the basis of the maximum distance angle according to table, modified according to standard practice of surgeon

• measurement of amount of muscle adjustment to be checked post placement of scleral sutures

• conjunctival incisions closed with VICRYL sutures

• topical anaesthetic and antibiotic drops given at end of procedure.

A surgical table was used with modification as appropriate to determine the amount of eye muscle movement to be performed depending upon the size of the angle of exotropia. Surgical technique was carefully recorded and monitored during the pilot, with a view to standardising surgical technique – as far as it was possible to do so – in a full trial and provide a clear description of the intervention in subsequent reports.

Primary outcome visit

The final assessment (9-month outcome) was conducted by a research orthoptist (RO) who was masked to the allocation of the child and was not otherwise in contact with children enrolled in the study. The parent and child were requested not to reveal the group allocation of the child to the RO prior to the assessment. Although children have noticeably red eyes immediately following eye muscle surgery, it is recognised that this redness resolves within 6 weeks when this is a first procedure. Residual scarring of the conjunctiva following eye muscle surgery will be inconspicuous by 6 months.

Parent interviews

We gathered qualitative data (primarily through telephone interviews) from parents to explore their reasons for either accepting or declining participation in the pilot study, their thoughts on randomisation and the study information received, and any ideas they may have had for improving the trial. These data would inform the design of a full RCT. Although the principles of informed consent meant that individuals were not obliged to give a reason for their decision if they did not want to, we invited parents to take part in a telephone interview if they agreed to this level of involvement. Without this information, it is difficult to see how research design can be improved to make it more acceptable. These interviews were conducted by a university researcher who was entirely separate from the clinical team, emphasising that parents’ decisions not to participate have been respected and reassuring that no attempt was being made to change their minds. A total of 48 parents who consented to the telephone call were interviewed, 34 of whom had declined and 14 of whom had agreed participation in the trial. Most (40) were mothers and eight were fathers. Interviews took on average 10 minutes (minimum 5 minutes, maximum 30 minutes). These parents had been approached about the trial between September 2011 and May 2012.

Planned outcome measures

There are three classes of clinical measure that could be used as outcome measures for the treatment of X(T).

1. Stereoacuity This is normally preserved for near fixation in affected children and will generally be unchanged by treatment unless it deteriorates as a result of surgical overcorrection of the development of a constant squint. Stereoacuity at distance may be affected but is not easily measurable and has high test/test variability.

2. Ocular alignment This is measured in PDs for near and distance fixation. It may be measured by a simultaneous prism cover test, which attempts to capture the alignment before binocular vision is disrupted, or on an alternating prism cover test, which is the total misalignment demonstrated by disrupting binocular vision by covering each eye in turn. In practice, binocular vision is easily disrupted in children with the condition and the usual reported measure is the total misalignment on an alternating prism cover test. The limitation of this measure is that alignment may be improved at the expense of deterioration in stereoacuity, if a constant, but small, misalignment is a consequence of treatment. Hence, as an outcome measure, this may overestimate the benefit of treatment by concentrating on the cosmetic outcome at the expense of the functional outcome. Neither does it attempt to measure the frequency with which misalignment occurs, which is a key factor in assessing the success of intervention in a condition in which ocular misalignment is intermittent.

3. Control This is measured using either the NCS61 or the Mayo scale19 for the condition. Both are suited to the assessment of an intermittent misalignment as is seen in the condition, and measure the frequency with which the misalignment occurs and the ease with which realignment occurs. Possible scores range from 0 to 9 and from 0 to 10, respectively; higher scores denote poorer control.

We feel that, because of the limitations of measures of alignment, a combination of measures of control and stereoacuity represent the most appropriate clinical outcome measures for treatment of the condition.

For children with X(T), and their parents, the most relevant outcome from intervention is the restoration of normal eye alignment, with associated cosmetic and functional benefits. The primary outcome in a full-scale trial would be the difference in the cure rate of X(T) between the surgical and actively monitored group; this was also the primary outcome for which data were collected in the current pilot in order to inform sample size calculations for a definitive RCT. Cure was defined as:

• a control score (NCS61) of 0 (misalignment never noticed by parents, no observable deviation on cover test)

• demonstrable near stereoacuity in children over 3 years of age.

Secondary outcomes included age-specific QoL assessments, satisfactory control of exotropia assessed by parental report and clinical components of the NCS61 and the Mayo score,19 rates of amblyopia, use of health-care resources, NHS costs, costs to families accessing the treatments being evaluated and incremental cost per cured patient (with cure as defined by the primary outcome) and a cost–consequences analysis based on incremental cost with respect to changes in all relevant outcomes where possible.

With respect to the choice of QoL measure, we know from our previous work with the Pediatric Quality of Life Inventory20 that there is little or no effect of the condition on the scores obtained, so we sought an alternative generic instrument that would capture any visual or psychosocial effect of the condition. We proposed the use of the Health Utilities Index Mark 3 (HUI3) in this study but found it unsuitable for the age of the population involved. In a longer-duration study, where the views of older children and their parents could be captured at outcome, the HUI3 may prove to be an acceptable generic instrument. We therefore used the newly developed condition-specific IXTQ. 66

The key outcomes of this pilot study were:

• data on the variability of the primary and secondary outcome measures

• rates of participant recruitment and randomisation

• nature and extent of participation bias

• rates of crossover and retention of recruited participants

• nature and extent of biases arising from crossover or loss to follow-up.

An initial recruitment rate of > 60% and a retention rate of > 70% was considered necessary to indicate feasibility of a full-scale RCT.

Our intention in the full trial would be to conduct a cost-effectiveness analysis based on incremental cost per cured patient (as defined by the primary outcome) and a cost–consequences analysis based on incremental cost with respect to changes in all relevant outcomes where possible, including the QoL measure and different clinical measures. In this study we assessed the ease of collecting information on outcome and costs needed for the health economics analysis.

With respect to collecting costs data, we piloted the TTQ and HSUQ that captured patient costs and the NHS costs. We assessed the response and completion rates of these instruments. Patient costs included travel costs for accessing NHS primary and secondary care; time costs of travelling and attending NHS primary and secondary care; and self-purchased health-care and related management costs. NHS costs comprised use of health-care resources in both primary and secondary care. Total costs consisted of patient costs and the NHS costs compared with the randomised interventions.

Sample size

A formal power calculation was not performed for this feasibility study. It was not powered to detect a clinically or economically meaningful difference in the primary outcome between the surgical and active monitoring groups. Rather, the aim was to provide robust estimates of the likely rates of recruitment and retention, and to yield estimates of the variability of the primary and secondary outcomes to inform power calculations for a subsequent full-scale RCT. We originally estimated that over a recruitment period of 6 months (subsequently extended to 9 months) across four centres we would be able to approach 240 patients who met the entry criteria. From their responses we would be able to determine whether or not the study is acceptable to parents and consequently whether or not it is possible to recruit patients and follow them up. We would also be able to estimate attrition rates. By approaching 240 children/parents we would be able to estimate the recruitment rate with a standard error no larger than 3.3%. Assuming that half of these children were actually recruited we would be able to estimate the 6-monthly attrition rate with a standard error of ≤ 4.3%.

Randomisation

Randomisation was in permuted blocks, stratified by collaborating centre, age and severity of X(T), as measured by the NCS. 61 A blocked allocation (permuted random blocks of variable length) system was used to allocate patients to the two groups in a 1 : 1 ratio to intervention (surgery) and control (active monitoring) groups. Randomisation was administered using a centralised, password-protected, web-based system that was managed by the Newcastle Clinical Trials Unit (NCTU).

As participants in the study were children of ≤ 16 years, consent was taken from the parent or legal guardian; however, every effort was made to include the child in the consent process.

Once consent was obtained, the PI at site, or individual with delegated authority, entered the patient ID, initials and the stratifying variables that then returned the allocation status. Participants were informed of their group allocation and given the appropriate Group Allocation Information Sheet.

Children in the surgery group proceeded to surgical preassessment before undergoing standard eye muscle surgery for X(T). They were then reviewed within 2 weeks of surgery and reviewed as was clinically appropriate, depending on the result of the surgery, before further review at 6 months post recruitment and final review at 9 months post recruitment.

Children in the active monitoring group proceeded to a clinic appointment arranged for 3 months’ time. Further review took place at 6 months post recruitment and then final review at 9 months post recruitment, as for the surgery group. At the final review at 9 months post recruitment (6 months post surgery), clinical outcome measures were obtained for subsequent analysis.

Masking

As surgery was the intervention of interest, it was not possible to mask participants or parents to their group allocation. Masking of investigators was achieved by the designation of a TO and a RO at each site. TOs could not be masked to the group allocation of participants, as they conducted all assessments other than the outcome assessment, and dealt with queries from parents/children during the course of the trial. Clinical examination at the primary outcome visit was carried out by the ROs who were unaware of treatment group allocation. At this final study visit, the success of masking was assessed by asking the outcome assessor: ‘Do you think the patient has had surgery or not? Why do you think this?’ Their responses were recorded on a separate form.

Statistical methods

For primary outcomes Student’s t-tests and Mann–Whitney U-tests were used to compare age and severity of X(T) of those participants retained with patients withdrawing from the trial, and to compare those who consented with those eligible but refused to participate. A Fisher’s exact test was carried out to assess variability of the primary outcome (difference in cure rate at the final assessment between the surgery and active monitoring arm); effect size [95% confidence intervals (CIs)] is also reported.

For secondary outcomes A chi-squared test was used to compare rates of satisfactory NCS61 outcome between groups; odds ratios and 95% CIs are also reported. Rates of development of amblyopia in the two groups were determined by monitoring VA at the 3-, 6- and 9-month assessments. For the QoL measurements, we were primarily concerned with response and completion rates for those instruments in both groups; in addition, their validity was assessed by matching individual pre- and post-treatment scores to the post-treatment primary outcome, using Student’s t-tests.

Statistical significance was set at p < 0.05. Data were analysed with the SPSS statistical package [SPSS (Statistical Package for the Social Sciences) Inc., Chicago, IL, USA], version 19.

Adverse event reporting

Adverse event reporting was undertaken in accordance with the National Research Ethics Service guidelines for adverse event reporting in trials that do not involve investigational medical products.

Definition of serious adverse event

A serious adverse event (SAE) (Table 4) was defined as an untoward occurrence that:

1. resulted in death,

2. is life-threatening

3. required hospitalisation or prolongation of existing hospitalisation

4. resulted in persistent or significant disability or incapacity,

5. was otherwise considered medically significant by the investigator.

Trial management processes

The Trial Management Meeting Schedule for the SamExo Trial included:

• Trial Management Group Meetings Monthly study management meetings were held between members of the NCTU (Senior Trial Manager and Trial Manager), the Chief Investigator and the core study staff at Newcastle. This meeting reviewed the progress at all sites.

• Newcastle Study Group Meeting A further monthly study meeting was held between the Trial Manager, the Chief Investigator and the research staff at Newcastle. This meeting reviewed the progress at Newcastle. Both meetings were minuted.

• Monthly minuted conference calls were held between all PIs (Robert Taylor, York; John Sloper, Moorfields; Peter Tiffin; Sunderland) and the Chief Investigator. TOs could also dial in if they had any issues to discuss.

• Four Trial Steering Committee (TSC) Meetings were held, chaired by Professor Charlotte Wright.

Good clinical practice training

All staff involved in the study were required to complete Good Clinical Practice (GCP) training. As this study was portfolio adopted, GCP training was provided free of charge through the NIHR.

Site initiation visits

Each site received the Investigator Site File (ISF) with a CD-ROM containing all of the documentation for the Pilot Rehearsal Trial and had a site initiation visit from the Trial Manager and the Co-Investigator, Christine Powell, in advance of patient recruitment. This visit covered training in use of the various logs within the ISF (screening, delegation, etc.), briefing on which paperwork was relevant at each stage and safety reporting using the Serious Adverse Event form and issues such as version control of documents. A ‘typical’ patient route through the study was included, allowing site staff to talk through the practicalities of recruitment. Training was also provided on the electronic Case Report Forms and the web-based randomisation system.

Throughout the Pilot Rehearsal Trial, PIs and TOs were encouraged to contact the Trial team with any queries they had.

Several communication strategies were adopted to encourage retention and fidelity to protocol. These were mainly personal contact through telephone, e-mail and face-to-face meetings. Other methods of communication included the SamExo website (http://research.ncl.ac.uk/samexo/). The website served a dual role of being a point of reference for the study staff and a means of sharing information with the general public. Feedback from the orthoptists, children and parents involved in the trial was very positive.

Facebook was also used as a more novel method of communication and a way to share non-sensitive information with the general public (www.facebook.com/pages/The-SamExo-research-study/279725438765959).

The experiences of the participating sites were captured during a final face-to-face SamExo team meeting after the end of follow-up.

Site recruitment and retention

All sites that began recruitment of patients were retained throughout the Pilot Rehearsal Trial and all have expressed an interest in continuing with a full trial. A web-based survey of other ophthalmology units in the UK was conducted (via e-mail discussion group PAED-OPHTH-STRABISMUS list: 157 recipients) and responses were obtained from 37 units treating X(T) (see Appendix 6). Of these 37 units, 35 (95%) indicated that they would be willing to participate in a future RCT.

Participant screening and recruitment

As of 29 February 2012, the proposed end of the recruitment period, 183 children had been screened out of an expected 240. Of those screened, 117 (64%) were eligible (expected 228 or 95%) and 29 (25% of eligible) children were recruited [expected 144 (64% of eligible)].

At the TSC meeting on 18 October 2011, the slow pace of recruitment was identified, and it was agreed to make an application for a no-cost extension, to extend recruitment to 31 May 2012, in order to see if strategies to improve recruitment had been effective. This request was granted.

Between 29 February 2012 and 31 May 2012, a further 48 children were screened, of whom 24 (50%) were eligible and 20 (83% of eligible) were recruited. Although this population includes a number of children who were either rescreened for eligibility, or whose parents had expressed a provisional interest in the trial, and was therefore an enriched population in terms of potential recruits, it does indicate that the strategies adopted by the study team to improve recruitment had a significant effect.

In total therefore, by 31 May 2012, 231 children had been screened (expected 240), 138 (60%) of whom were eligible (expected 228: 95%) and 49 (35% of eligible children) were recruited (expected 144: 64% of eligible: Figure 3).

FIGURE 3.

SamExo recruitment graph: expected vs. actual recruitment.

Eligibility

Many more children than predicted did not fulfil the eligibility criteria for the study (10/240 predicted vs. 93/231 observed). Reasons for non-eligibility were determined for 87/93 (94%) and can be summarised as:

• severity of X(T) insufficient to meet eligibility criteria, 40%

• inability to demonstrate binocular vision at near fixation, 23%

• did not have target condition, 19.5%

• ocular comorbidity, 7%

• could not be adequately assessed, 4.5%

• received previous treatment, 3.5%

• systemic comorbidity, 2%.

The most common reason for children not to meet the eligibility criteria was that their strabismus was not sufficiently severe. This was either a reflection of the size of the misalignment when present (a misalignment of 15 PDs was chosen to reflect usual criteria for surgical intervention) or that the control of the strabismus (the ease with which the eyes were realigned following dissociation) was too good.

Consent to participation in the trial

Brief information on reasons why eligible parents refused participation was available from the screening logs in 80/89 (90%) cases.

These can be summarised as:

• parents did not feel strabismus severe enough to warrant surgery, 65%

• parents felt surgery was necessary and did not want to wait, 15%

• objections to trial processes, for example randomisation, 12.5%

• other reasons, 7.5%.

Reasons for agreeing to, or declining, participation in the trial were explored in more detail in the qualitative interviews.

Participation bias

Consent was obtained from 56/89 (63%) ‘eligible not recruited’ (ENR) patients to record baseline and 9-month follow-up clinical data. The demographic and clinical characteristics of the ENR group were compared with those who agreed participation in the SamExo trial (Table 5). Those who agreed to take part had poorer control of their exotropia (as assessed by the NCS61) than those who declined; although statistically significant this difference is unlikely to be clinically significant. The mean age of those who agreed was slightly older but not significantly so (see Table 5). The proportion of males and females was almost identical in each arm (34–37% male).

Randomisation

The randomisation process was generally smooth, although it was intermittently difficult to access the website. This problem was generally solved through contact with the database team, although occasionally the result of randomisation had to be given to the parents after they had left the clinic.

Randomisation appeared to work well, with 25 participants allocated to surgery and 24 to active monitoring. One parent immediately declined the allocation to surgery, but the remaining 48 participants remained in their allocated group.

Seven children in the active monitoring group, and four in the surgery group, were prescribed glasses at baseline to correct refractive error. No glasses were issued subsequently.

Baseline demographic and clinical characteristics by treatment arm are provided in Table 6; there were no significant differences on any parameter.

Adherence to protocol and retention

Two patients received surgery after the proposed 3-month window: one who was awaiting the results of sickle cell investigations and one whose operation was arranged but then cancelled due to chickenpox.

The outcome results for these patients were included in the analysis even although only 4 months’ postoperative follow-up was possible, as opposed to the desired 6 months.

The parent of one patient allocated to surgery immediately declined the allocation. The participant was retained in the trial and the results were analysed on an intention-to-treat basis.

Two patients in the active monitoring group were lost to follow-up despite repeated postal and telephone reminders.

A total of 47/49 (96%) recruited participants attended for the final appointment, scheduled at 9 months post randomisation.

The flow of participants through the trial is shown in the CONSORT (Consolidated Standards of Reporting Trials) diagram (Figure 4).

FIGURE 4.

Consolidated Standards of Reporting Trials diagram.

Adverse events

Masking

Masking was successfully maintained throughout, in that there were no overt breaches. In the majority of cases, however, the orthoptist assessing outcome correctly guessed which treatment arm the participant was in, except for one case in the surgery arm and two in the active monitoring arm.

Quality-of-life outcomes: Intermittent Exotropia Quality of Life Questionnaires

No parent or child refused to complete this instrument. There were 49 out of 49 expected parent IXTQs at baseline, 45 out of 46 expected proxy IXTQs (one was not administered, in error; the child in the other three cases was too young, i.e. < 2 years) and 23 out of 23 expected child IXTQs (the remainder were not old enough to complete an IXTQ themselves). At follow-up there were 45 out of 47 expected parent IXTQs (two were not administered in error; the remaining two withdrew/were lost to follow-up), 44 out of 47 expected proxy IXTQs (three were not administered, in error; two withdrew/were lost to follow-up) and 29 out of 30 expected child IXTQs (one not administered, in error, and 17 were too young; the remaining two withdrew/were lost to follow-up).

Missing response rates to individual items within the IXTQ were no > 7%. The highest percentage of missing responses was for item 7 on the child-rated IXTQ (‘Does it bother you that you have to shut one eye when it is sunny?’): this was unanswered by 1 out of 23 (4%) of children at baseline and 2 out of 29 (7%) at follow-up. All other items on the child-rated IXTQ were completed 100%. Items 1, 4, 9 and 11 on the proxy-rated IXTQ were missed in 1 out of 45 (2%) cases at baseline; at follow-up, items 3, 5, 7, 9 and 12 were each unanswered in 1 out of 44 (2%) cases. No items were skipped on the parent IXTQ at baseline, and at follow-up items 4, 5 and 15 were missed in only 1 out of 45 (2%) cases.

Higher scores denote better perceived quality of life. At baseline, there were no differences between treatment arm and IXTQ scores on any scale or subscale. There were no statistically significant changes over time in the active monitoring group on any parent IXTQ scales. Significant improvements were found within the surgery arm on all Parent IXTQ scales (total score, and function, psychosocial and surgery subscales, Tables 7 and 8). Mean score on the Proxy IXTQ deteriorated for those in the active monitoring arm and improved for those in the surgery group, but the differences were small and not or only just statistically significant (Table 9). A small, but not statistically significant, improvement was seen in the Child IXTQ scores for each arm (also note the very small numbers) but there was no significant difference between groups (Table 10).

Clinical outcomes

Details of the clinical outcomes are provided in Table 11a–d, below.

In a pilot trial of this nature, which was not set up to provide definitive clinical information about the effectiveness of treatment and was conducted over a short period of follow-up, it would be inappropriate to draw definitive conclusions from a comparison of the clinical outcomes between the intervention and active monitoring arms. Nevertheless, these data are of interest, and one of the primary outcomes of the pilot trial was stated to be a comparison of children achieving a control score (NCS61) of zero between the two groups.

Following discussion at the TSC on 19 June 2012, it was agreed that children who had well controlled overcorrections (esophorias or intermittent esotropias), and in whom measurements of stereoscopic vision had not deteriorated, would be classified as having a clinic control score of zero. Children with constant esotropias following surgery cannot be classified using the NCS61 [which is designed only for the assessment of X(T)] and are referred to as overcorrections.

The management of, and handling of data in relation to, children who developed overcorrections following surgery requires further thought in a future study, with improved standardisation of the management of overcorrections. We think it is reasonable to classify children who have intermittent overcorrections, but have retained near stereo, as a good result but this approach needs to be considered alongside QoL data.

Achievement of a score of zero on the NCS61 requires both that the parents never notice the strabismus (NCS61 home = 0), and that on clinical testing, any deviation is immediately controlled (NCS61 clinic = 0). The strabismus could therefore be regarded as ‘cured’ at this point, however this might not remain the case over the longer term.

Furthermore, this criterion ignores any improvement that might be generated clinically, or in terms of QoL, by improvement, rather than cure, of X(T). As we have seen, however, QoL gains were seen only in relation to parental, rather than perceived or actual child QoL. Although numbers were small, analysis of the QoL data by outcome group indicates that QoL gains were actually greatest in the group for which the control score was improved but not zero (‘cured’), suggesting that ‘cure’, as defined by NCS61 = 0, may not necessarily be the most appropriate outcome measure for a full trial.

It is also known that ‘control’ of strabismus varies from day to day, and this may also be the case for children with a clinic control score of zero. For these reasons, other potential outcome measures were assessed to judge their applicability for further studies.

Measures of control of intermittent exotropia

Measures of ocular alignment

Ocular misalignment is measured clinically by determining the size of prism which, when placed in front of a misaligned eye, abolishes the corrective fixation movement that would otherwise result when the fixing eye is covered. Differences of ≤ 10 PD are generally regarded as within the limits of measurement error, and not considered clinically significant. 67

Measurements of distance misalignment showed that, in the active monitoring arm, four children had improvements of > 10 PD between baseline and follow-up, none had deteriorated, and 18 had measurements at outcome within 10 PD of the baseline measurement.

Measures of distance alignment in the surgery arm showed that 13 children had improvements of > 10 PD, four had intermittent overcorrections, two had constant overcorrections, none had deteriorated and six had measurements within 10 PD of baseline.

In summary, there were significant improvements in distance alignment in 18% of the active monitoring arm and 52% of the surgery arm.

Measurements of near misalignment showed that, in the active monitoring group, three children had improvements of > 10 PD, one child deteriorated by > 10 PD and 18 had measurements at outcome within 10 PD of baseline.

Measures of near misalignment in the surgery arm showed five children had improvements of > 10 PD, two children deteriorated by > 10 PD, 10 had measurements within 10 PD of baseline, five were intermittently overcorrected and three had a constant overcorrections.

In summary, there were significant improvements in near alignment in 14% of the active monitoring arm and 20% of the surgery arm. Of the surgery arm, 32% showed either intermittent or constant overcorrections at near fixation at 6 months postoperatively.

Binocular vision and stereoacuity

A change of two octaves of stereoacuity is generally considered to be clinically significant.

At outcome in the active monitoring arm, of 19 children who could complete near stereoacuity testing at near fixation, two had deteriorated by two or more octaves, nine were within two octaves and eight had improved by two octaves. Of the remaining three, who were too young to perform stereoacuity testing and whose binocular vision was assessed using their ability to demonstrate a fixation movement in response to a prism placed before one eye, two were stable and one had deteriorated.

In summary, the binocular vision of 86% of children in the active monitoring arm was either stable or improved over the follow-up period, and significantly deteriorated in 14%.

In the surgery arm, near stereoacuity deteriorated by more than two octaves in four children, remained within two octaves in 15, and improved by more than two octaves in four. Of two children assessed using prisms, one was stable and one had deteriorated.

In summary, the binocular vision of 80% of children in the surgery arm was either stable or improved, and significantly deteriorated in 20% over the follow-up period.

Eligible not recruited outcome data

Follow-up data were obtained for 49 out of 56 (88%) of the ENR patients who consented to data collection and storage. Mean duration between screening and follow-up was 9.4 ± 2 months, range 4–15 months. In most cases (n = 34), the follow-up data were collected within 10 months from screening. In 12 cases, review was between 11 to 12 months and in 3 cases it was beyond a year from screening (2 at 13 and one at 15 months). However, any treatment reported was received within 9 months from screening; a few were already having treatment for vision prior to screening as indicated in Table 12.

Details of primary and secondary clinical outcomes for ENR patients are provided in Tables 13 and 14.

Collection of data

Data were collected on effects and costs for both study arms. Sources of data collection are shown in the tables in Appendix 8.

Data collection results

As described in the previous section, there are 49 patients in total recruited to this trial, with 24 randomised to the control arm and 25 in the treatment arm; however, one of the patients in the treatment arm changed their mind and asked to be switched to the control arm, so as a result 24 patients in the treatment arm received surgery and 25 patients in the control arm were monitored only of their progress without surgery.

Identification of data collection and entry issues with the questionnaire

Summary

This section of the pilot study sought to investigate the feasibility of conducting an economic evaluation as part of a RCT comparing eye muscle surgery against active monitoring for childhood intermittent distance exotropia. Specifically, the health economic component of the study aims to assess the ease of health economics data collection.

Costs data are collected from a number of sources. Information regarding the surgery is recorded on the CRF, and services use following the intervention is collected through the Participant Costs Questionnaire, which also gathers information on patients’ out-of-pocket expenses. Assessing both sources of data collection, a > 80% completion rate was achieved. The high response rate of Participant Costs Questionnaire may be due to the high level of commitment from the participants and the fact that the questionnaires were completed on patients’ study visit. There are, however, some issues regarding questionnaire design, which will require the tool to be refined for the definitive trial.

The use of health services appears to be concentrated within the first 6 months of the trial follow-up for both trial arms. Consideration will be given for a definitive study streamlining the data collection and hence reducing the burden to both the participants and the research team, and potentially also reducing the costs of research to the funder. Overall, the work conducted has shown that it is feasible to collect meaningful health economics data in a definitive trial, subject to adjustment of data collection points and refinement of the data collection tool.

Trial management

The structures that were put in place to manage the trial appeared to be valuable in maintaining regular contact between key stakeholders and ensuring efficiency. Telephone conferences were a useful supplement to face-to-face meetings.

Screening and recruitment

Although screening rates were initially high, as patients had been identified in readiness for the opening of recruitment, there was a lag in recruitment due to the delay in the subsequent appointment for the recruitment clinic.

New referral letters often did not contain sufficient information to make a judgement about whether or not the child had X(T), and many children referred with suspected X(T) did not have X(T) at all. This wasted appointments in the screening clinics and meant that it would have been confusing to provide parents with information about a trial for which their child might not be eligible, so it was decided to reserve providing parents with information leaflets until the initial clinic visit.

Given the age of the children involved, it was not always possible to confirm eligibility for the trial at the initial screening visit, and many children had to be reviewed before eligibility was confirmed. This generated further appointments in the screening clinics, which blocked slots for other potentially eligible children.

Subsequent blockage of appointment slots by children who needed rescreening for eligibility, and parents who wished to take more time to consider recruitment when their child was eligible, contributed to a failure to recruit to target. This was compounded by an unexpectedly high proportion of children who failed to meet the eligibility criteria.

The most common reason for children failing to meet the eligibility criteria was that their strabismus was not sufficiently severe: either the angle of misalignment was too small or their X(T) was too well controlled to meet conventional criteria for surgery for X(T). The NCS61 grading for trial entry was defined as ≥ 3. However, this score could be achieved by a parent who noticed X(T) all the time without any objective evidence that X(T) was present. Although this was not the case for any trial participant, the criteria for trial entry were on the limit of what would be considered an acceptable degree of severity of X(T) to justify intervention, particularly in younger children.

Most surgeons would not be comfortable operating on a child whose X(T) was objectively well controlled. Conventional surgical criteria suggest that a NCS61 of ≥ 2 should be present on distance fixation in order for eye muscle surgery to be appropriate. 18,20 This was not specified in the current study and should be in a future trial.

On the other hand, many children in the active monitoring arm more than fulfilled conventional criteria for surgical treatment, and, by and large, did not deteriorate significantly, with many showing some evidence of spontaneous improvement, particularly in stereoacuity, over the follow-up period.

In a longer duration study, it is uncertain what proportion of parents would be happy to defer intervention for a prolonged period.

Recruitment of children who had failed school vision screening because of X(T) enriched the population of children referred with X(T), but also targeted a group of children whose parents had not presented their children for treatment and who, by implication, were less concerned to have the strabismus treated. It also introduced an age bias to the participants, as those who were recruited from school screening were between 4 and 5 years of age.

For parents and clinicians, the initial screening appointment presented a challenge, in that it had to encompass a diagnosis of the child’s eye condition, an assessment of eligibility for the trial, an explanation of possible treatment options, an explanation of the lack of robust evidence underpinning the timing and effectiveness of these interventions, and an explanation of the trial. It was rarely possible to cover all of these points in a single consultation, not least because of the amount of new information that had to be assimilated by the parents, and the questions they had about treatment and the trial, were often not answerable in the subsequent telephone call. There were also difficulties in contacting many parents by telephone during the working day, and attempts had to be made to contact them outside of normal working hours, which were not always successful.

Furthermore, the explanation of the lack of evidence underlying the effectiveness and timing of intervention served, in many cases, to undermine the parent’s confidence in the treating clinician, and by extension, the trial.

In Newcastle, a research nurse was recruited during the screening and recruitment phase of the trial, and she subsequently gave most of the information about the trial itself. This separation of the role of the treating clinician from the main recruiter to the trial proved extremely beneficial in aiding the process of recruitment. Such use of research nurses in all centres should be considered in a future study.

It became apparent during the process of recruitment that the initial two visits, for screening and recruitment, and the intervening telephone call, often gave insufficient time for parents to fully consider participation in the trial. Many parents wished to observe the progress of the condition before considering treatment, and many were concerned at the prospect of surgery. This often generated further appointments in the recruitment clinics, which blocked further slots for potential recruits.

Some parents had concerns about trial processes, particularly randomisation.

During the screening and recruitment phase, it was decided to not only restrict entry to the trial to newly diagnosed children, but also to provide study information to children who had a recent confirmed diagnosis of X(T) and had not received any treatment, at all sites. These children were then allocated appointments for recruitment clinics if their parents expressed an interest in participating in the trial.

Despite the difficulties outlined above, it is important to state that the rate of refusal of eligible children to participate was similar to that predicted in the initial submission. It was predicted that the refusal to participate rate would be around 33% – in fact, 64% (89/138) of eligible children identified refused to participate.

In summary, the strategies adopted to improve recruitment included improving the efficiency of the utilisation of screening clinic appointments by screening out children who did not have the condition, recruitment of prevalent cases and cases from preschool screening, and the separation of recruitment from clinical care by the use of a research nurse in one centre.

Participation bias

For ethical reasons, it was possible to collect limited information only on children who were eligible for, but declined participation in, the trial. On the basis of the information available to us, there appeared to be differences in age and strabismus control between those participating and those declining participation, with those participating being older and having poorer control. Such differences, if truly present, might affect the generalisability of a longer-term study. QoL was not measured in the group eligible but declining participation, but, given the effect of intervention on parental anxiety, it is tempting to speculate that those parents entering the trial were, on the whole, more anxious about their child’s condition than those who declined participation.

Randomisation

No significant issues were identified in relation to randomisation, which appeared to provide groups that were matched for baseline demographics and clinical severity.

Adherence to group allocation and protocol

There were minor issues in relation to the timing of surgery due to comorbidity.

Adherence to group allocation was excellent, with only one parent declining the allocated group. Adherence to group allocation might be a more significant issue in a longer-term study. In particular, parents who were in favour of surgery for their child’s X(T) would be unlikely to remain in an active monitoring arm in a long-duration study.

Retention of participants for the duration of the study was generally excellent, and is a tribute to the intensive attention paid to this by the study team at all sites.

Adverse events

The major identified risk for participants in the active monitoring arm of the study was that the condition might have a deterioration in their clinical condition. It is challenging to distinguish a permanent deterioration from fluctuation of the condition in this patient group. 62,71,72 However, in spite of some apparent deterioration in the clinical picture in two children (10301, 46701), none developed a constant strabismus.

At outcome, two children in the surgery arm had constant esotropia at near and distance and were classified as over corrected; a further child had surgery for an overcorrection within the follow-up period. In addition, there were four children who had eso deviations for near viewing only, three of which were intermittent and one that was constant. This is in line with previous studies25,26 and permanent overcorrection with loss of stereoacuity remains the most significant complication of surgery. The management of surgical overcorrection should be standardised in a future trial.

Maintaining masking

Although no breaches of the masking to group allocation of the orthoptists assessing outcome at each centre were reported, in almost every case the orthoptist, when asked, correctly guessed the group allocation. It is unlikely that, 6 months after surgery, traces of the surgical procedure itself were still visible, as redness of the eyes following strabismus surgery generally lasts for a maximum of only 6 weeks. In cases of overcorrection, previous surgery would be assumed, as this is not an outcome that occurs spontaneously. Similarly, there may well have been an assumption that cases which had changed little in clinical appearance over the follow-up period were more likely to have been actively monitored. This effect is likely to be lessened, but not abolished, in a study of longer duration.

Qualitative interviews

These indicated that a trial with a preference arm would be more acceptable to parents. In general, the information provided to parents appeared to be of an acceptable standard, and was not a major barrier to trial participation. The way in which the diagnosis and information about the trial was communicated did appear to be an issue, indicating the need for a greater use of research nurses and training in future studies. Some parents were keen to have more details of surgical success rates, indicating the need for a definitive trial.

Quality of life

The IXTQ questionnaires showed no differences between groups for the child-reported questionnaire and the proxy questionnaire, where the parents indicated the impact that they felt that X(T) had on their child. There was a significant difference between the surgery group and active monitoring group for the parental questionnaire, which was spread across all subscales. This portion of the IXTQ mainly enquires about parental anxiety (see Appendix 5) – in fact, all of the questions in this section commence with either the phrase ‘I worry . . .’ or ‘It worries me . . .’.

This suggests that the improvement seen on this instrument mainly reflects the effect of the intervention on parental anxiety, either because ‘something has been done’ or because the effect of the intervention on the condition has reduced parental anxiety, rather than that the child’s QoL has improved, or even the parent’s estimation of the child’s QoL.

This does reflect current clinician experience, where, in the absence of robust evidence of the long-term effectiveness of interventions for X(T), treatment is often driven by parental anxiety and anecdotal or personal experience.

It was not possible to collect IXTQ data on parents of eligible children who did not consent to participation, but it is interesting to speculate that there may have been a bias, with those parents not consenting because they did not wish their child to have surgery being less anxious, and vice versa. It would be interesting to investigate generic measures of anxiety in parents seeking surgery for their child’s X(T) and those wishing to avoid it.

Clinical outcomes

There appeared to be variability in outcomes between the four centres, although the numbers involved were too small for formal analysis. One potential improvement that could be made in a further trial would be greater standardisation of the surgical procedure; however, it is likely that this would not completely eliminate the variation in clinical outcomes. Furthermore, where standardisation – for example photographic documentation with review by a reading centre – has been performed, variations in surgical technique have persisted.

Ocular alignment

In parallel with the benefits on control, there were clear benefits on ocular alignment from surgery. This needs to be balanced by six participants who were permanently or intermittently ovecorrected at distance, and one only overcorrected at near, at outcome, plus another participant who had further surgery for an overcorrection within the follow-up period.

Four of these seven patients had retained or improved stereoacuity; however, some of these overcorrections were cosmetically significant. Although initial overcorrection has been regarded as a good prognostic indicator of long-term alignment after surgery for X(T)74,75 (although this has been disputed26,76), overcorrection at 6 months following surgery may not resolve, and may require further treatment if cosmetically significant or associated with a reduction in binocular vision or amblyopia.

In summary, there were significant improvements in near alignment in 14% of the active monitoring arm and 20% of the surgery arm; 32% of the surgery arm showed either intermittent or constant overcorrections at near fixation at 6 months postoperatively.

Binocular vision and stereoacuity

Although there were three participants whose binocular vision deteriorated in the active monitoring arm, given the variability in current clinical testing of binocular vision in young children, the clinical significance of this is unclear. The eight participants in whom stereoacuity improved could either be demonstrating a maturational effect or a spontaneous improvement in their condition.

In the surgery arm, two of the participants who were overcorrected at both distance and near, and the one who was overcorrected only at near, had a reduction in binocular vision and stereoacuity, and can be classed as functional overcorrections. The binocular vision or stereoacuity of a further two children deteriorated. These children were not overcorrected, but did show some evidence of deterioration of their X(T). The other four children with intermittent overcorrections did not show any deterioration in their binocular vision.

Hypothesis generation

The SamExo study was a stand-alone pilot study to assess the feasibility of a full trial of surgery compared with active monitoring for X(T). Although not powered to assess the effectiveness of surgery as an intervention, the clinical outcomes do indicate agreement with previous research, which suggests that, over a short follow-up period, the majority of patients who are actively monitored do not significantly improve or deteriorate, whereas most patients who undergo surgery have, in the short term, improved alignment, albeit with a rate of between 10% and 20% of overcorrection with a deterioration in stereoacuity. The proportion of patients who are ‘cured’, i.e. demonstrate no significant strabismus while having preserved stereoacuity is low, at 36%.

There are two questions that then arise:

1. What are the long-term outcomes of surgery and active monitoring? Most studies indicate that there is a significant rate of recurrence of strabismus in children who have undergone surgery for X(T). 50,77,78 As discussed in the systematic review, one study50 showed that recurrences were most common within 6 months from surgery; however, after that, recurrences occurred continuously in the R&R group and rarely in the BLR group. In a further study, the Kaplan–Meier rate of developing ≥ 10 (delta) of misalignment after the first surgery was 54% by 5 years, 76% by 10 years and 86% by 15 years. 78 This indicates that outcomes from surgery may not be stable, which needs to be taken into account in the design of a future trial.

The data on spontaneous cure rates in children who have not undergone surgery are less clear. The lack of randomised studies means that children who have not undergone surgery will tend to have a less severe strabismus than those who have. Nevertheless, there are studies that suggest that a proportion of children with X(T) will show spontaneous improvement36 or at least no deterioration. 79

This suggests the possibility that the long-term outcomes following strabismus surgery for X(T) may be no better than a strategy of awaiting spontaneous improvement.

We would propose that a future trial would follow up children for at least 5 years from recruitment. Although this might seem onerous, an annual assessment of outcomes might be sufficient, reducing the burden on participants.

Furthermore, a longer-term study would be better able to assess QoL outcomes from a participant perspective, as most of the participants in the pilot study were too young to give their view.

This suggests the need for a formal RCT of surgery compared with active monitoring, which incorporates long-term follow-up of participants in the trial.

2. How can the cure rate of surgery be improved? There are sound theoretical reasons for the view that outcomes from strabismus surgery in children are improved if the surgery is undertaken soon after the onset of strabismus. 80–82 There is evidence of improved outcomes from early surgery in infantile esotropia (convergent strabismus). 83,84 The evidence supporting early surgical intervention in X(T) is less clear. A widely cited study by Pratt-Johnson et al. 85 showed evidence of improved outcomes from surgery for X(T) undertaken in children aged < 4 years.

We have shown in a previous study20 that rates of surgery for X(T) within 2 years of presentation are low in the UK, indicating a reluctance among clinicians in the NHS to adopt a practice of early surgery. Qualitative interviews undertaken by our group (Lecouturier et al. , submitted to BMC Ophthalmology) indicated that the reasons behind this reluctance included:

• difficulty in assessment of young children, particularly those under 4 years, in whom measurements of stereoacuity and strabismus angle are inaccurate

• the need for repeat assessment owing to variability of X(T) in order to plan the need for and the amount of surgery

• concerns about the more severe functional impact of overcorrection of X(T) in younger children, particularly the development of amblyopia in cases where a constant esotropia develops.

Although these concerns are valid, given the low short-term cure rates following surgery undertaken at an older age, and the high rates of recurrence following surgery, there is clearly a need to consider whether or not an alternative treatment strategy would be more effective than deferring surgery until the child is old enough to obtain accurate measurements, by which time the X(T) has usually been present for several years. This suggests the need for a formal RCT that incorporates early surgery as an intervention.

Contribution of authors

Mr Michael Clarke (Chief Investigator) conceived the study and contributed to the study design and funding acquisition; was PI at the Newcastle site and oversaw delivery of the intervention; and drafted the report.

Dr Vanessa Hogan (Clinical Trial Manager) co-ordinated local approvals for collaborating sites, and site set-up; monitored conduct of study and site and study close down; and co-ordinated and assisted in drafting the final report.

Dr Deborah Buck (Co-investigator) was data manager and analyst for the SamExo study and final report, with overall responsibility for data quality and analysis; provided expertise in clinical and psychosocial outcomes assessment and qualitative interview techniques, and interviewed parents regarding the acceptability of the study design and conduct; provided a narrative summary of those findings.

Dr Jing Shen (Health Economist) developed the health economics protocol; designed and produced the health economics questionnaires; analysed data; drafted the economic section; and contributed to the final report.

Ms Christine Powell (Co-Investigator) contributed to the study design, protocol development, creation of case report forms and standard operating procedures, ethics committee and sponsor approvals and adoption of the study onto the UKCRN study portfolio.

Mr Chris Speed (Senior Clinical Trial Manager) oversaw study management.

Mr Peter Tiffin (Consultant Ophthalmologist, Sunderland Eye Infirmary) conceived the study; contributed to the study design and funding acquisition; and oversaw delivery of the intervention.

Mr John Sloper (Consultant Ophthalmologist, Moorfields Eye Hospital NHS Foundation Trust) – conceived the study; contributed to the study design and funding acquisition; and oversaw delivery of the intervention.

Mr Robert Taylor (Consultant Ophthalmologist, York Hospitals NHS Foundation Trust) conceived the study, contributed to the study design and funding acquisition, and oversaw delivery of the intervention.

Dr Mahmoud Nassar (Data analyst) analysed the Mayo Control Score data.

Dr Kerry Joyce (Senior Research Associate) contributed to data extraction, quality assessment and synthesis for the systematic review, and wrote up the report.

Mrs Fiona Beyer (Research Associate) devised and carried out the literature searches for the systematic review; wrote up the search methods; contributed to data extraction, quality assessment and synthesis; and commented on the report.

Professor Richard Thomson (Professor of Epidemiology and Public Health) conceived the systematic review (with Michael Clarke), supervised the researchers, participated in discussions about eligibility and quality of the included studies, and critically reviewed and approved the final version of this section of the report.

Professor Luke Vale (Health Foundation Chair in Health Economics) oversaw the health economics element of the study and contributed to the final report.

Professor Elaine McColl (Director Newcastle Clinical Trials Unit, Institute of Health and Society) contributed to the study design and funding acquisition, and supervised the involvement of the Newcastle Clinical Trials Unit.

Dr Nick Steen (Statistician) provided guidance on statistical issue of study design and data analysis.

Publications

Joyce KE, Beyer F, Thomson RG, Clarke MP. A systematic review of the effectiveness of treatments in altering the natural history of intermittent exotropia. Br J Ophthalmol 2014;Epub ahead of print.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). 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, NETSCC, the HTA programme or the Department of Health. 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 HTA programme or the Department of Health.

MEDLINE (via Ovid)

1. Exotropia/

2. (divergen$ adj10 (excess$ or strabismus or squint$)).tw.

3. (exotrop$ or IDEX).tw.

4. or/1-3

5. limit 4 to humans

Cochrane risk of bias tool for randomised controlled trials (http://bmg.cochrane.org/research-projectscochrane-risk-bias-tool)

Quality Appraisal Tool for Cohort Studies (adapted from Critical Appraisal Skills Programme)

1. Was the cohort recruited in an acceptable way?

2. Was the study prospective?

3. Was the intervention conducted in an explicit and standardised manner?

4. Was the outcome appropriately measured to minimise bias?

5a. Did they identify all potential important confounding factors?

5b. Did they adjust for confounding factors in the design and/or analysis?

6a. Were they followed up for at least 12 months?

6b. Were at least 80% of participants followed up (for prospective studies)?

7. Are the authors’ conclusions substantiated by the reported data?

Reasons for being unwilling

1. ‘Probably unethical.’ (This was the one who did not operate but still answered this question.)

2. ‘Probably not rather than absolutely not, 2 reasons: 1. parents probably won’t accept it. 2. I would be concerned about loss of BSV.’

3. ‘The decision for surgery is one made by the parents and the surgeon. Hence to randomise to no treatment after this decision I feel will be a major stumbling block.’

Potential barriers

Potential barriers to such a trial succeeding broadly fell into three main themes: loss to follow-up (mentioned by four people), dropout rates (four people) and the unwillingness of parents to wait for surgery (eight people). Other issues included definitions of squint severity and outcomes:

1. Loss to follow-up:

   1. ‘Long follow-up would increase the DNA rate especially in the surgical cases that were successful.’

   2. ‘Losing people under follow-up.’

   3. ‘Loss of patients to follow up.’

   4. ‘Failure to continue follow-ups.’

2. Dropout rates:

   1. ‘Parents motivation to continue lengthy follow-up period. There will be more dropouts in the observation group in favour of surgery.’

   2. ‘Parents may decide to go to another centre for surgical treatment.’

   3. ‘Change of mind by parents.’

   4. ‘Dropout because of child being teased at school.’

3. The desire for surgery/unwillingness to wait:

   1. ‘Many parents are keen for surgical intervention and it may be difficult to monitor them for 5 years.’

   2. ‘In case of severe intermittent XT, may be difficult to convince parents for observation alone.’

   3. ‘Parents desire for surgery.’

   4. ‘Pressure from parents wanting surgery.’

   5. ‘Patient/parental pressure for surgery with worsening control.’

   6. ‘Difficulty convincing parents that it is satisfactory to monitor especially if their child/teacher complains of difficulty seeing the smart board.’

   7. ‘Once you mention surgery – parents commonly don’t want to wait.’

   8. ‘Parents may not want to be randomly assigned into one group especially those not wanting surgery.’

4. Criteria/definitions:

   1. ‘Defining the outcome and what constitutes success/failure.’

   2. ‘Investigators agreeing the definition of “moderate to severe” as the point at which surgery is recommended is so variable across the UK currently.’

   3. ‘I think the real difficulty would be in trying to rationalise not only the type of surgery, but also the aim of the surgery. Do you leave them XT or aim to prevent outdrift by making them ET and risk amblyopia?’

5. Random:

   1. ‘Culture amongst some orthoptic colleagues of advising parents that child should have operation based on angle or stereo or age. (“I’ll send you through to the consultant so they can add you to the waiting list,” etc.). This may affect parental attitudes to being randomised.’

   2. ‘Ethics management pressures on workload.’

   3. ‘Will need to be clear when patients can withdraw if binocularity threatened.’

Question: How important do you think it is to conduct a randomised controlled trial of the clinical effectiveness of surgery versus active monitoring in this condition?

Seventeen (46%) felt that it was very important, 17 (46%) quite important and 3 (8%) not very important. (One skipped.)

Question: How important do you think it is to conduct a randomised controlled trial of the cost effectiveness of surgery versus active monitoring in this condition?

Eight (22%) felt it was very important, 20 (54%) quite important and 9 (24%) not very important. (One skipped.)

Other comments

Interesting idea but personally I would prefer to await outcome of such a trial before delaying surgery in children for whom I currently offer surgery.

5 years seems like a long period to monitor. During period or monitoring would you suggest over minus glasses or convergence exercises. What criteria would move a child from monitoring to surgery?

5 years is ambitious – but probably necessary – and I guess with necessary checks (i.e. break code if obviously doing worse)

I think the first trial should be on observation versus surgery in moderate control, as this is more ethical and management for this varies between departments you may even compare with minus lenses.

Are there any plans to include a third arm for BTxA?

Having trained in the UK and now practising in my home university hospital (Thessaloniki, Greece), I wonder if you’d be interested in non-UK participants in your current and future studies. If so, what would be minimum requirements?

This respondent was based in Wales and indicated he would not be willing:

Mike- this is what I feel though I feel for the establishing a natural history it is extremely important but will be difficult to recruit in my practice – but I’m willing to give it a go.

Good luck Mike.

Go for it guys. Answers not found yet!

Health Service Use Questionnaire (Part A)

1a. Has your child seen or contacted a GP because of his or her eyes during the last 3 months?

YES

NO

1b. If YES to Question 1a, how many appointments did your child attend with a GP?

1c. If YES to Question 1a, how many times did a GP visit your child at home?

1d. If YES to Question 1a, how many times did you or your child have a telephone conversation with a GP?

2a. Has your child see a practice nurse because of his or her eyes during the last 3 months?

YES

NO

2b. If YES to Question 2a, how many times in total?

3a. Has your child seen a community optician or optometrist during the last 3 months?

YES

NO

3b. If YES to Question 3a, how many times in total?

4a. Has your child seen a hospital specialist (consultant or one of his/her team) because of his or her eyes other than attending planned follow-up visits during the last 3 months?

YES

NO

4b. If YES to Question 4a, how many times in total?

5a. Has your child been admitted to hospital because of his or her eyes during the last 3 months?

YES

NO

5b. If YES to Question 5a, how many nights was your child in hospital (if admitted as a day case, please enter 0)?

6a. Has your child had prescription medicine for his or her eyes during the last 3 months?

YES

NO

6b. If YES to Question 6a, what type of medication has your child been prescribed?

7a. Have you purchased over the counter medicine (e.g. eye drops) for your child’s eyes during the last 3 months?

YES

NO

7b. If YES to Question 7a, how much did you pay in total?

8a. Have you paid for any other private health care for your child because of his or her eyes during the last 3 months?

YES

NO

8b. If YES to Question 8a, what type of care did you pay for?

8c. If YES to Question 8a, how much in total did it cost?

9a. Have you taken time off from work to look after your child because of his or her eye condition other than attending planned follow up visits during the last 3 months?

YES

NO

9b. If YES to Question 9a, how many days in total were you absent from work?

Time and Travel Questionnaire (Part B)

Please note that this questionnaire assumes that for every form of health care your child attended, you have travelled with and accompanied your child. If your child travelled alone, please only include the costs that incurred to your child.