Article Text

Surgical intervention in childhood intermittent exotropia: current practice and clinical outcomes from an observational cohort study
  1. Deborah Buck1,
  2. Christine J Powell2,
  3. John J Sloper3,
  4. Robert Taylor4,
  5. Peter Tiffin5,
  6. Michael P Clarke1,2, for the Improving Outcomes in Intermittent Exotropia (IOXT) Study group
  1. 1Institute of Neuroscience, Newcastle University, Newcastle upon Tyne, UK
  2. 2Royal Victoria Infirmary Eye Department, Newcastle upon Tyne Hospitals NHS Trust, Newcastle upon Tyne, UK
  3. 3Department of Strabismus and Paediatric, Moorfields Eye Hospital, London, UK
  4. 4Department of Ophthalmology, York Hospitals NHS Trust, York, UK
  5. 5Ophthalmology Unit, Sunderland Eye Infirmary, Sunderland, UK
  1. Correspondence to Dr Deborah Buck, Institute of Neuroscience, c/o Clinical Trials Unit, 4th Floor William Leech Building, Medical School, Newcastle University, Newcastle upon Tyne NE2 4HH, UK; deborah.buck{at}


Purpose To describe surgical outcomes in intermittent exotropia (X(T)), and to relate these to preoperative and surgical characteristics.

Methods 87 children (aged <11 years) underwent surgery in 18 UK centres; review data (mean 21 months post-surgery) were available for 72. The primary outcome measure was motor/sensory outcome (angle and stereoacuity). The secondary outcome measure was satisfactory control assessed by Newcastle Control Score (NCS).

Results 35% of patients had excellent, 28% had fair and 37% had poor primary outcome. Preoperative and surgical characteristics did not influence primary outcome. Satisfactory control was achieved in 65% of patients, while X(T) remained/recurred in 20%. Persistent over-correction occurred in 15% of children. There was no relationship between over-correction and preoperative characteristics or surgical dose/type. Median angle improved by 12 prism dioptres (PD) at near and 19 PD at distance (p<0.001). Median NCS improved by 5 (p<0.001). 40% of those initially over-corrected remained so by last postoperative assessment; no relationship was found between an initial over-correction and good outcome.

Conclusions Whilst excellent motor/sensory outcome was achieved in one-third and satisfactory control in two-thirds of patients, the 37% poor outcome and 15% persistent over-correction rate is of concern. Surgical dose was similar in those under- and over-corrected, suggesting that over-corrections cannot be avoided merely by getting the dosage right: a randomised controlled trial (RCT) would shed light on this issue. Initial over-correction did not improve the chance of a good outcome, supporting the growing literature on this topic and further highlighting the need for randomised controlled trials of X(T) surgery.

  • Treatment Surgery
  • Child health (paediatrics)
View Full Text

Statistics from


Intermittent exotropia (X(T)) is a form of childhood strabismus affecting approximately 30 per 100 000 children aged under 19 years1 and with a prevalence of around 1% in children under 11.2 It is characterised by periodical divergent misalignment which is present on distance fixation, or during periods of tiredness or inattention; this may become more frequent and be present on near fixation, and can progress to constant exotropia.2 ,3 There is no consensus regarding the optimum management of X(T). One option is surgery to improve control for aesthetic considerations, to prevent deterioration to constant exotropia, and/or to improve distance stereoacuity.4 However, the surgical approach comes with a risk of persistent overcorrection, although an intentional small initial overcorrection has been thought to provide the best results because of a tendency toward postoperative exotropic drift.5 ,6 Whilst there is some evidence to support this7 ,8 others report that X(T) can recur in corrected and over-corrected patients.9 ,10 Consequently there have been calls for further studies to establish whether or not initial postoperative alignment is associated with stability of the alignment.9 ,11

A range of criteria has been used in determining surgical outcome, including motor criteria (size/direction of the postoperative deviation) and measures of control of X(T). There is no consensus regarding which parameters are most important and, consequently, definitions of surgical success have varied. Few have included sensory criteria (eg, stereoacuity) in their evaluations but there is increasing recognition of the importance of this in X(T) outcomes.12–15

The Improving Outcomes in Intermittent Exotropia (IOXT) study16 ,17 is an observational study which recruited 460 children from an inception cohort: 87 underwent surgery. This paper reports the surgical management, preoperative and surgical characteristics, and postoperative outcomes for these 87 children, and explores whether any characteristics, particularly initial postoperative alignment, were associated with outcome.



The IOXT study16 ,17 recruited a total of 460 patients from 26 paediatric eye clinics across the UK between May 2005 and December 2006. Children under 12 years, diagnosed with X(T) within the prior 12 months and previously untreated, were eligible. Children with convergence insufficiency type of X(T) (near deviation ≥10 prism dioptres (PD) more than distance), constant exotropia or significant coexisting ocular pathology, were excluded. Children who underwent surgery between enrolment and December 2008 were followed-up prospectively until December 2009. This was an observational study. Participating clinicians managed children according to their normal clinical criteria. Informed consent was obtained from parents/guardians. The study received a favourable opinion from UK North West Multi-Centre Research Ethics Committee.

Data collection

Near and distance angle were assessed with the alternate prism cover test, near stereoacuity with Frisby Near Stereoacuity Test (FNS), and control with the Newcastle Control Score (NCS).18 The NCS combines an estimate of observed frequency of exotropia by parents (home control) with assessment of the child's ability to realign the eye following a cover test to induce misalignment (clinic control): possible total NCS scores range from 0–9 (0–3 for home, 0–6 for clinic control) with higher scores indicative of poorer control. Near stereoacuity data are reported for those aged ≥4 years at their preoperative assessment because younger children were unable to consistently complete testing.16 Distance stereoacuity testing was possible in a limited number of centres and is therefore not reported.

Measurement outcomes

Outcomes are reported ≥6 months postoperatively. The primary outcome measure is based on the motor/sensory success criteria used by Pineles et al14 as follows:

  • Excellent outcome: 0–8Δ X(T) at distance; stable near stereoacuity

  • Fair outcome: ≤4Δ esotropia (ET) or 9–15Δ X(T) at distance; stable near stereoacuity

  • Poor outcome: >4Δ ET or >15Δ X(T) at distance; worsening or loss of near stereoacuity

The rate of persistent over-correction (presence of any manifest ET at 1/3 and/or 6 m) is reported.

A secondary outcome measure was the proportion of patients with satisfactory NCS control—defined as total NCS ≤2 or improvement by ≥3 at final outcome. This was assessed with a categorical variable indicating patient status as:

  • NCS ≤2 and/or improved by ≥3

  • Stable, that is, still X(T) with no major improvement/deterioration

  • deteriorated by ≥3

  • ET

Mean/median changes in angle, stereoacuity, NCS and visual acuity are reported.

Given the uncertainty regarding the relationship between initial postoperative alignment and stability of the alignment longer-term, we report the subsequent status of patients who were initially over-corrected. We also report (Web Only file) recurrence in those with initially excellent alignment or initially corrected (NCS 0), and the subsequent status of patients who were initially under-corrected. Development of amblyopia in those over-corrected is examined.

Outcomes are presented for patients with available data from both initial and final postoperative (≥6 months) assessment. We report total amount of unilateral surgery, calculated by summing lateral rectus recession (LRR)+medial rectus resection (MRR); and total amount of bilateral surgery, calculated by summing left and right eye LRR. Surgical dose per PD was calculated by dividing amount of surgery by preoperative distance angle; median change in angle per mm of surgery performed was calculated by dividing the change in angle at last postoperative assessment by amount of surgery. Changes over time were assessed with Wilcoxon tests for non-parametric and paired-samples t-tests for parametric data. To avoid over-estimation of improvement in angle, those with persistent over-corrections are excluded from analyses of median change in angle of deviation. Additional information on Methods is available in the online supplementary file.


Surgical details

18/26 centres performed surgery on 87/460 (19%) children. Of these 87, 52 (60%) had unilateral surgery (all LRR plus MRR), 33 (38%) had bilateral LRR, and two (2%) had 3-muscle surgery (horizontal). None had single-muscle surgery. Mean (SD) amount of surgery was 10.9 mm (1.3), range 7–14 mm for unilateral and 13.2 mm (2.7), range 8–22 mm for bilateral. Four bilateral surgery patients had inferior oblique procedures (two recessions, one myectomy, one disinsertion). Mean (SD) surgical dose per PD was 0.33 mm (0.06) for unilateral (range 0.23–0.48) and 0.50 mm (0.18) for bilateral (range 0.31–1.22).

Demographic and preoperative characteristics

Patient characteristics and intervals between surgery and clinical assessments are illustrated in table 1.

Table 1

Study and patient characteristics

Data from initial and last postoperative visits are available for 72 (83%) patients, to which the remainder of these results pertain. No significant differences were found between the 72 included and 15 excluded from the analyses in terms of preoperative angle or NCS. Included patients were significantly younger at surgery than those without data (mean 58 compared to 77 months, p=0.002).

Median preoperative angle was 20Δ at near and 30Δ in the distance: details of these and other preoperative measurements are presented in table 2 (first row).

Table 2

Comparison of angle, stereoacuity, control and visual acuity at pre- and last postoperative assessments

Further procedures

Nine patients (12.5%) underwent additional procedures because of over-correction. (Four had botulinum toxin injection within 3–4 months of surgery, four underwent one additional operation and one had two further operations). A further three patients (4%) had a second operation for under-correction. The timing of further surgery varied from 9–12 months to 3 years after the first procedure, although for the patient who underwent two additional operations the timing was earlier (four and 7 months following the initial procedure).

Outcomes at last postoperative assessment

Surgical success

Based on motor criteria alone, 36% of patients had excellent alignment, 28% had fair, and 36% had poor alignment (30% X(T) >15 and 6% ET >4). Almost identical results were found when motor and sensory criteria were used together: 35% fell into the excellent category, 28% achieved fair results and 37% had a poor outcome. No patients experienced significant deterioration in near stereoacuity of ≥0.6 log seconds of arc.

Persistent over-correction

ET was present in 11 children at last postoperative assessment (2Δ–18Δ ET). Three remained over-corrected despite further procedures (two additional surgery and one botulinum toxin injection), and one remained so following treatment with glasses post-surgery. The remaining seven children received no further intervention for ET. Comparisons of surgical dose found that mean amount of unilateral surgery per PD was 0.31 mm in corrected, 0.34 mm in under-corrected and 0.33 mm in over-corrected patients (p=0.44), and that mean bilateral dose was 0.46 mm, 0.49 mm and 0.45 mm respectively (p=0.68). No significant relationship was found between surgery type: 17% of those who had bilateral surgery were over-corrected compared to 14% of those who had unilateral surgery; 50% and 54% respectively were under-corrected; and 33% and 32% respectively were corrected (χ2 test p=0.91).

Secondary outcomes

Table 2 shows significant improvements following surgery in median near and distance angle and median NCS, but not in stereoacuity or visual acuity. Median (IQR) change in angle per mm of surgery was 2.5 PD (1.2–3.1) for unilateral and 1.5 PD (0.9–2.2) for bilateral surgery. Patient characteristics within the excellent and poor groups are provided in table 3: no statistically significant differences were seen on any parameter.

Table 3

Comparison of patient characteristics with excellent or poor final surgical outcome


33% of patients were corrected (NCS 0) at their last assessment. Satisfactory control was achieved in 65% of patients, while X(T) remained/recurred in 20%. No patients deteriorated by NCS ≥3. The remaining 15% were over-corrected. Of those with an excellent result based on the motor and sensory criteria all but one also achieved satisfactory control. The analyses undertaken for the primary success criteria (table 3) were replicated for this secondary outcome. There were no significant differences except those over-corrected at last postoperative assessment were more likely than those with satisfactory outcome or under-corrected to have been over-corrected initially (90% compared to 30% and 7% respectively, p<0.001). Two patients developed a constant exotropia (although each had deteriorated by only one point on the NCS clinic component).

Stability in those initially over-corrected

Of 25 patients who were initially over-corrected, 9 (36%) were corrected by last assessment, that is, NCS=0, 6 (24%) were under-corrected whilst 10 (40%) remained over-corrected. Even a small initial overcorrection did not predict good outcome. For example, of seven patients with an initial angle of ≤10 PD ET, three were corrected by last assessment but two were under- and two remained overcorrected.

Supplementary data on: recurrence in those with initially excellent alignment or initially corrected; subsequent status of patients initially under-corrected; relationship between preoperative NCS and the magnitude of change in NCS; and postoperative conservative treatment are available in ‘Web Only’ Results file.

Two children may have developed amblyopia as a result of over-correction. LogMAR acuity at last postoperative (0.500, 0.375 and 0.400, 0.200) was worse than at their preoperative assessment (0.200, 0.200 and 0.100, 0.100); however acuity had been assessed on uncrowded vision tests preoperatively and on crowded tests post-operatively.


This was a multicentre study of X(T) involving 460 children in 26 centres throughout the UK of which 18 performed surgery in 87 children. A problem for practitioners is the lack of consensus regarding how and when to treat X(T), and what constitutes a good outcome. Thus, comparisons with earlier studies often have to be made with some caveat regarding different age groups, inclusion criteria and/or indicators of success. In this study we reported motor and sensory outcome, and level of NCS control, in children with intermittent distance exotropia who were aged between 1.5 to 10 years at the time of surgery.

From a total cohort of 460 children, 19% were operated on over a 3-year period. 16% of surgery patients subsequently received botulinum toxin injection or underwent a second operation for over- or under-correction. Primary surgical outcome was excellent, fair and poor in approximately one-third of patients respectively. This is almost identical to the rates reported by Pineles et al14 although they included distance stereoacuity in their formulation (discrepancy between distance and near postoperative stereo)— which was not feasible in the present study. Almost no difference in success rates were found in this study when using the combined criteria compared to the motor criteria alone. Pineles, on the other hand, found higher success rates than we did when applying the motor criteria alone (68% compared to 36%) albeit over a much longer follow-up period and with higher re-operation rates. Other studies which have used motor criteria alone have reported success rates ranging from 56% to 79%.19–22 These defined success as <8 or <10 PD of deviation and thus some children with comparatively larger ETs and slightly bigger X(T)s fell into their successful category. Another important difference is that mean age at surgery in these studies was relatively higher (range approximately 7–9 compared to 4.8 years in the current study) and the mean age in Pineles’ study was considerably higher at 14 years.

Satisfactory control was achieved in two-thirds of patients in this study. All but one of those with excellent primary outcome likewise achieved satisfactory control post-operatively. Overall there were significant improvements in median angle of deviation and NCS. The risk of developing amblyopia, losing stereoacuity, or deteriorating to constant exotropia after surgery was slight. However, the 37% poor surgical outcome and 15% over-correction rate is of concern. Some authors have found over-correction rates as low as 023 and 2.5%9 but these studies were less stringent in their classification of an over-correction (>5 and >10 PD ET respectively). Conversely, others have reported rates comparable to those found in the current study. For example 12% in Ruttum et al6 although they also categorised over-correction as >5 PD; moreover the average age was 14.7 years. Ing et al,24 ,25 who defined successful alignment as the absence of any postoperative tropia, recorded an over-correction rate of 17% in children of an average age akin to our study.

We are unable to identify any particular causes of long-term over-correction as we found no relationship between this and any characteristics including age, surgery type or dose. This implies that one of the main risks of X(T) surgery, over-correction, cannot be avoided merely by getting the dosage right: a randomised controlled trial (RCT) would shed light on this. Our data do suggest that initial over-correction is not necessarily beneficial to longer-term outcome, adding to the growing literature that this is not always the case.9 ,10 ,22 Indeed, 40% of those initially over-corrected remained so at their last postoperative visit and even a small initial over-correction did not predict success.

In terms of limitations, the average postoperative follow-up period was around 2 years and while this is not particularly prolonged the study was not set up to determine outcomes beyond this. Surgery was not performed in one-third of participating hospitals throughout the course of the study, indicating variation in the criteria for surgical intervention, as does the relatively low rate of further intervention in those who remained overcorrected. There is little evidence regarding the natural course of childhood X(T)26 and we cannot be certain that spontaneous improvement over time is any less likely than a good outcome following surgery. In the absence of such knowledge, surgery will continue to be offered by clinicians and requested by parents and although the chances of a good result are reasonable the cause or causes of over-correction remain unclear. This highlights the need for RCTs with tightly-defined criteria regarding indications for surgery, surgical dose, management of over-corrections, and definition of surgical success. This task is far from straightforward, particularly if the expectations of parents are considered: it has been shown that surgical success rates do not necessarily parallel, and can be lower than, levels of patient satisfaction.12 Nevertheless we believe any future studies should incorporate sensory and control outcomes alongside the more customary motor criteria as benchmarks for success, and we encourage clinicians to take into account patient/parental opinion regarding what constitutes a desirable outcome.


This article was written on behalf of the Improving Outcomes in Intermittent Exotropia (IOXT) Study group. We wish to thank the orthoptists and ophthalmologists at all IOXT collaborating centres, and the parents who agreed to take part in the study.


View Abstract

Supplementary materials


  • Contributors DB: study design, obtaining funding, data analysis and interpretation, statistical analysis, literature search and manuscript preparation, revising manuscript, final approval of the manuscript. CJP: acquisition of data, data analysis and interpretation, revising manuscript, final approval of the manuscript. JS: study conception and design, obtaining funding, data analysis and interpretation, revising manuscript, final approval of manuscript. RT: study conception and design, acquisition of data, data analysis and interpretation, drafting and revising manuscript, final approval of the manuscript. PT: acquisition of data, data analysis and interpretation, drafting and revising manuscript, final approval of manuscript. MPC: study conception and design, obtaining funding, acquisition of data, data analysis and interpretation, drafting and revising manuscript, final approval of manuscript, overall supervision.

  • Funding This work was funded by grants from the Guide Dogs for the Blind Association, UK and The BUPA Foundation, UK. The funders were not involved in the study design; the collection, analysis or interpretation data; in writing the report; or in the decision to submit the paper for publication.

  • Competing interests None.

  • Ethics approval North West Multi-Centre Research Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • At a glance
    Harminder Singh Dua Arun D Singh