Three-Year Observation of Children 3 to 10 Years of Age with Untreated Intermittent Exotropia

Three-Year Observation of Children 3 to 10 Years of Age with Untreated Intermittent Exotropia

Three-Year Observation of Children 3 to 10 Years of Age with Untreated Intermittent Exotropia The Pediatric Eye Disease Investigator Group Writing Com...

861KB Sizes 0 Downloads 3 Views

Three-Year Observation of Children 3 to 10 Years of Age with Untreated Intermittent Exotropia The Pediatric Eye Disease Investigator Group Writing Committee: Brian G. Mohney, MD,1 Susan A. Cotter, OD, MS,2 Danielle L. Chandler, MSPH,3 Jonathan M. Holmes, BM, BCh,1 David K. Wallace, MD, MPH,4 Tomohiko Yamada, OD,1 David B. Petersen, MD,5 Raymond T. Kraker, MSPH,3 Christie L. Morse, MD,6 B. Michele Melia, ScM,3 Rui Wu, MS3 Purpose: To describe the course of intermittent exotropia (IXT) in children followed up without treatment for 3 years. Design: Observation arm from randomized trial of short-term occlusion versus observation. Participants: One hundred eighty-three children 3 to 10 years of age with previously untreated IXT and 400 seconds of arc (arcsec) or better near stereoacuity. Methods: Participants were to receive no treatment unless deterioration criteria were met at a follow-up visit occurring at 3 months, 6 months, or 6-month intervals thereafter for 3 years. Main Outcome Measures: The primary outcome was deterioration by 3 years, defined as meeting motor criterion (constant exotropia 10 prism diopters [D] at distance and near) or near stereoacuity criterion (2-octave decrease from best previous measure). For the primary analysis, participants also were considered to have deteriorated if treatment was prescribed without meeting either deterioration criterion. Results: The cumulative probability of protocol-specified deterioration by 3 years was 15% (95% confidence interval, 10%e22%), but that was likely an overestimate, partly because of misclassification. Among 25 deteriorations, 2 met motor deterioration, 11 met stereoacuity deterioration, and 12 started treatment without meeting either criteria (7 for social concern, 1 for diplopia, 4 for other reasons). Among the 132 participants who completed the 3-year visit and had not been treated during the study, only 1 (<1%) met motor or stereoacuity deterioration criteria at 3 years. Of the 4 participants completing the 3-year visit who met deterioration criteria previously and had not started treatment, none still met deterioration criteria. Between the baseline and 3-year examination for these 132 patients, improvement occurred in distance and near stereoacuity (mean improvement, 0.14 and 0.14 logarithm of arcsec; P  0.001 and P  0.001, respectively), distance exotropia control (mean improvement, 0.6 points; P  0.001), and distance exodeviation magnitude (mean improvement, 2.2 D; P ¼ 0.002). Conclusions: Among children 3 to 10 years of age with IXT for whom surgery was not considered to be the immediately necessary treatment, stereoacuity deterioration or progression to constant exotropia over 3 years was uncommon, and exotropia control, stereoacuity, and magnitude of deviation remained stable or improved slightly. Ophthalmology 2019;-:1e12 ª 2019 by the American Academy of Ophthalmology Supplemental material available at www.aaojournal.org.

Intermittent exotropia (IXT) is a disorder of ocular alignment characterized by an intermittent outward deviation of one or both eyes. This deviation, which occurs in nearly 1% of children in the United States1 and up to 4% of children in Asia,2 generally manifests more frequently with distance viewing, illness, or fatigue. Although common, there are minimal data regarding the natural history of this disorder. Published reports almost exclusively comprise retrospective studies from single institutions3e6 with varying definitions for progression and duration of follow-up. Moreover, the results of these investigations are disparate, with some reports suggesting that IXT worsens over time,7,8 others reporting ª 2019 by the American Academy of Ophthalmology Published by Elsevier Inc.

improvement,3,5 and still others demonstrating no change at all.4,6 The present report provides 3-year follow-up data on a cohort of previously untreated, prospectively identified children 3 to younger than 11 years of age with IXT who were assigned randomly to an observation group in a previously reported randomized clinical trial.9

Methods The study was supported through a cooperative agreement with the National Eye Institute of the National Institutes of Health and was conducted according to the tenets of the Declaration of Helsinki by https://doi.org/10.1016/j.ophtha.2019.01.015 ISSN 0161-6420/19

1

Ophthalmology Volume -, Number -, Month 2019 the Pediatric Eye Disease Investigator Group (PEDIG) at 60 academic and community-based clinical sites. The protocol and Health Insurance Portability and Accountability Actecompliant informed consent forms were approved by each site’s respective institutional review board, and a parent or guardian of each study participant gave written informed consent. An independent data and safety monitoring committee provided study oversight. The study is registered at www.clinicaltrials.gov (identifier, NCT01032330; accessed October 6, 2017). The full study protocol is available on the PEDIG web site (www.pedig.net; accessed October 6, 2017).

Eligibility The study comprised children 3 to younger than 11 years of age with previously untreated IXT (other than refractive correction) and near stereoacuity of 400 seconds of arc (arcsec) or better on the Preschool Randot Stereotest (Stereo Optical Co, Chicago, IL). For eligibility, the child’s exodeviation had to meet the following criteria: (1) intermittent or constant exotropia at distance, and either IXT or exophoria at near; (2) exodeviation magnitude of 15 prism diopters (D) or more at distance or near measured by the prism and alternate cover test (PACT); and (3) exodeviation of 10 D or more at distance measured by the PACT. In addition, the investigator and the child’s family had to be willing to observe the IXT without treatment for 3 years unless specific criteria for deterioration were met. Additional eligibility criteria have been published previously.9 Whether to offer enrollment into the present study to a given patient with IXT was at the investigator’s discretion. Twenty-five of the 60 clinical sites also enrolled patients into a concurrent PEDIG randomized clinical trial comparing 2 surgical approaches for IXT (clinicaltrials.gov identifier, NCT01032330). At these sites, some patients were eligible only for the current study (e.g., patients with IXT that was a convergence insufficiency type, true divergence excess type, or high accommodative convergence or accommodation type), some patients were eligible only for the surgery study (e.g., patients with prior nonsurgical treatment), and some patients were eligible for both studies. We did not collect data on which study or studies were offered to each patient or on patients who declined to participate.

Treatment In the initial randomized clinical trial, each child (hereafter referred to as a participant) was assigned randomly either to observation (n ¼ 183 reported herein) or to 5 months of part-time daily patching (n ¼ 175 not reported herein). Participants randomly assigned to observation received no treatment other than a standard refractive correction9 (if needed) unless protocol-specified deterioration criteria (Table 1) were met at a masked examination, after which surgical or nonsurgical treatment was provided at the investigator’s discretion. The only exceptions for allowing IXT treatment, aside from meeting deterioration criteria, included overwhelming social concern, debilitating diplopia, or failure to maintain stereoacuity age norms,10 which was permitted only after a discussion with the protocol chair.

Testing Procedures and Follow-up Visits Follow-up visits occurred 3 months after randomization (2 weeks), 6 months after randomization (1 month), and every 6 months thereafter until 3 years. At each visit, a study-certified examiner (pediatric ophthalmologist, pediatric optometrist, or certified orthoptist) masked to the participant’s randomized treatment group measured stereoacuity (Distance Randot Stereotest11,12 and Randot Preschool Stereoacuity Test at near), performed cover testing at 6 m and 33 cm, and measured exodeviation control at 6 m and 33 cm using the IXT office control score13 that categorizes control from 0 (phoria, best control) to 5 (constant exotropia,

2

worst control) at both distance and near fixation. Control levels 3 through 5 were assigned based on the proportion of time that the exotropia was manifest during a 30-second period before any dissociation. If no exotropia was observed during this period, control levels 0 through 2 were assigned based on the longest time it took for fusion to be reestablished after 3 consecutive 10-second periods of dissociation. Details of testing procedures have been published previously.9

Statistical Methods The primary outcome measure was deterioration at a masked examination during the 3 years of follow-up, defined as meeting motor or stereoacuity criteria, or both, at any follow-up visit (Table 1). Participants also were considered to have deteriorated for the primary analysis if they underwent surgery for IXT or if nonsurgical treatment for IXT was prescribed without first meeting motor or stereo deterioration criteria, or both. The cumulative probability of deterioration by 3 years and the 95% confidence interval (CI) were estimated using the Kaplan-Meier method, including all participants and censoring participants who either did not meet deterioration criteria or who had begun IXT treatment at their last completed visit (i.e., giving partial credit when calculating the event proportion based on the amount of follow-up completed). Three cause-specific deterioration outcomes by 3 years were identified post hoc: motor deterioration only, stereo deterioration only, or both motor and stereo deterioration. These cause-specific outcomes differed from the primary outcome in 2 ways: (1) the primary outcome refers to the first occurrence of any deterioration criteria being met, whereas the cause-specific outcomes refer to the first occurrence of a particular deterioration criteria being met; and (2) starting IXT treatment without meeting the criteria for the outcome was considered a deterioration at the time treatment was started in the primary outcome, but was censored as nondeterioration at the time treatment was started in the cause-specific outcomes. These cause-specific deterioration outcomes were analyzed using the aforementioned Kaplan-Meier method used for the primary analysis. In addition, because participants starting treatment after meeting the cause-specific deterioration outcome (e.g., motor criteria) that occurred first were censored at the time treatment was started, starting treatment was a competing risk for experiencing the other cause-specific deterioration outcome (e.g., stereo criteria). Because the Kaplan-Meier method is known to overestimate the event rate when competing risks are present, a cumulative incidence analysis also was conducted.14 The results from the cumulative incidence analyses were similar to those found with the Kaplan-Meier analyses for these cause-specific deterioration outcomes (data not shown). Secondary outcomes of 3-year stereoacuity, exotropia control, and IXT magnitude were analyzed in participants who completed the 3-year visit and had not been prescribed treatment. The hypothesis of no change between baseline and 3 years was tested using the paired-sample t test for all secondary outcomes that were continuous measures; comparison of exotropia control scores using the nonparametric Wilcoxon rank-sum test yielded similar results (data not shown). To help address possible informative censoring (i.e., whether loss to follow-up was related to outcome) in the continuous secondary outcome measures of stereoacuity, exotropia control, and PACT results magnitude, sensitivity analyses were conducted using multiple imputation with the Monte Carlo Markov chain method15 to impute data for participants who did not complete the full 3 years of follow-up (data not shown). The imputation model for each 3-year outcome (e.g., control, stereo deterioration) was based on baseline and interim follow-up data for that assessment. Dichotomized improvement and worsening versions of these outcomes were defined using the threshold likely to represent real change based on the coefficient of repeatability for

Mohney et al



Observation of Children with IXT

Table 1. Definition of Deterioration by 3 Years Deterioration by 3 Years (Primary Outcome) The participant’s IXT was considered to have deteriorated if any of the following 3 criteria were met during masked examiner testing occurring at any followup visit within 3 years of randomization: 1. Motor criteria: constant exotropia 10 D at distance and near (throughout examination) by SPCT, confirmed by a retest.*  Constant tropia was defined as manifest tropia that was present 100% of the time during the examination, determined by at least 3 covereuncover tests (one before any dissociation).  Because any amount of near stereoacuity may be inconsistent with a constant near tropia of 6 D or more, if the child seemed to have a constant tropia and near stereoacuity on the Randot preschool stereotest, the masked examiner was instructed to look over the child’s Polaroid glasses while the child viewed the 800-arcsec stereogram while performing a covereuncover test to determine if the child was tropic at the time he or she was reporting stereoacuity. If the child was not tropic at the time he or she was reporting stereoacuity, the near tropia was not considered to be constant. 2. Stereoacuity criteria: drop in near stereoacuity by Randot preschool stereotest of at least 2 octaves (at least 0.6 log arcsec) from the best previous stereoacuity test from any study visit, confirmed by a retest* (see below). Randot Preschool Near Stereoacuity Best Previous Stereoacuity (at Any Study Visit; arcsec) Stereoacuity Level Needed at Follow-up Visit to Meet Deterioration Criteria (arcsec) 40” 200” or worse 60” 400” or worse 100” 400” or worse 200” 800” or worse 400” Nil 3. Surgical or nonsurgical treatment for IXT has been received without first meeting either of the above deterioration criteria. arcsec ¼ seconds of arc; IXT ¼ intermittent exotropia; log arcsec ¼ logarithm of seconds of arc; SPCT ¼ simultaneous prism and cover test; D ¼ prism diopter. *If the participant seemed to meet 1 or both of the motor or stereo deterioration criteria, the masked examiner retested after a 10-minute break to confirm or refute deterioration.

each measurement16: 2 log octaves for stereoacuity,17 3 points for control,18 and 8 D and 13 D for PACT measurements at distance and near19; the proportion of participants with each outcome and an exact Clopper-Pearson 95% CI were calculated. Each was analyzed including only participants whose baseline level allowed for potential improvement or worsening of the specified threshold amount. The proportion and exact Clopper-Pearson 95% CI also were calculated for a post hoc outcome of meeting motor or stereo deterioration at the 3-year visit, which included participants completing the 3-year visit who had not been prescribed treatment, regardless of whether deterioration criteria had been met at any previous visit. All analyses were conducted using SAS software version 9.4 (SAS Institute, Inc, Cary, NC).

Results Baseline Characteristics In this observational study of previously untreated IXT, the mean age at enrollment of the 183 participants was 6.12.0 years, 115 (63%) were female, and 112 (61%) were white. Baseline clinical characteristics are shown in Table 2.

Visit Completion The 1-year visit was completed by 159 participants (87%), the 2-year visit was completed by 146 participants (80%), and the 3-year visit was completed by 146 participants (80%). Comparing the 152 participants (83%) who either met the primary deterioration outcome by 3 years or who completed 3 years of follow-up without ever meeting deterioration criteria (including 6 who withdrew early after meeting the deterioration criteria) with the 31 participants (17%) who withdrew early without having met deterioration criteria, 61% versus 71% were female, the mean age was 6.2 years versus 5.6 years, and the median distance simultaneous prism and cover test magnitude was 17 D versus 12 D, respectively (Table 2). The reasons for early withdrawal are listed in Table S3 (available at www.aaojournal.org).

Deterioration at Any Time over 3 Years The cumulative probability of protocol-defined estimate of deterioration by 3 years was 15% (95% CI, 10%e22%; Tables 4 and 5; Fig 1). Of the 25 participants meeting the primary outcome criteria for protocol-defined deterioration, 12 started treatment without meeting the motor or stereoacuity deterioration criteria (7 for overwhelming social concern, 1 for debilitating diplopia, 3 for exotropia worsening that did not meet deterioration criteria, and 1 for headache), 11 demonstrated a near stereoacuity decrease of 2 octaves or more from the best previous measure, and 2 demonstrated a constant exotropia of 10 D or more at distance and near (although both participants with constant XT demonstrated random dot stereoacuity at near, one with 40 arcsec and the other with 100 arcsec, which suggests that they did not have a moderate- or largemagnitude tropia at near during stereoacuity testing; Table 4). The cumulative probabilities of meeting various cause-specific deterioration by 3 years are shown in Table 5.

Motor or Stereo Deterioration or Both at the 3Year Visit Of the 132 participants who completed the 3-year visit and had not been treated during the study, 1 (<1%; 95% CI, 0%e4%) met motor or stereo deterioration criteria at the 3-year visit, regardless of whether motor or stereo deterioration criteria had been met at an earlier visit. Of the 4 participants who previously met deterioration criteria (all for stereoacuity), had not started treatment, and had completed the 3-year visit, none of these participants still met the protocol-specified definition of deterioration at the 3-year visit.

Near Stereoacuity, Exotropia Control, and Deviation Magnitude at the 3-Year Visit Near stereoacuity, control, and deviation magnitude by the PACT for the 132 participants who completed the 3-year visit and had not

3

Ophthalmology Volume -, Number -, Month 2019 Table 2. Baseline Demographics and Clinical Characteristics* Study Completion Status* All Participants (n [ 183)

Withdrawn (n ¼ 31)

Completed (n ¼ 152)

Characteristic

No.

%

No.

%

No.

%

All Female gender Race or ethnicity White Black/African American Hispanic or Latino Other Unknown/not reported Age at randomization (yrs) 3e<5 5e<7 7e<9 9e<11 Mean (SD) Range Average best-corrected visual acuity between 2 eyes 20/12 or 20/16 20/20 20/25 20/32 20/40 Mean (SD), logMAR Range (logMAR) Interocular difference in best-corrected visual acuity 0 lines >0 to <1 line 1 line >1 to <2 lines 2 linesy Mean (SD), logMAR Range (logMAR) Spherical equivalent refractive error (average between 2 eyes), D e6.00 to
183 115

100 63

31 22

100 71

152 93

100 61

112 23 32 14 2

61 13 17 8 1

14 3 12 2 0

45 10 39 6 0

98 20 20 12 2

64 13 13 8 1

61 67 33 22

33 37 18 12

11 16 3 1

35 52 10 3

50 51 30 21

33 34 20 14

4

6.1 (2.0) 3.0e11.0

5.6 (1.5) 3.2e9.5

6.2 (2.1) 3.0e11.0

11 36 40 9 4

2 6 10 32 15 48 1 3 3 10 0.05 (0.10) e0.10 to 0.30

19 13 55 36 58 38 16 11 4 3 0.04 (0.09) e0.27 to 0.30

47 25 23 2 3

16 52 3 10 11 35 0 0 1 3 0.05 (0.07) 0.00e0.30

70 46 42 28 31 20 4 3 5 3 0.04 (0.05) 0.00e0.20

3 4 4 16 42 28 4

0.27 (1.27) e5.63 to 2.63 42 23

1 3 0 0 2 6 5 16 11 35 9 29 3 10 0.46 (1.33) e4.38 to 2.63 6 19

4 3 7 5 6 4 24 16 65 43 42 28 4 3 0.23 (1.26) e5.63 to 2.56 36 24

75 39 43 19 7

13 6 8 2 2

62 33 35 17 5 60” (1.78) 69” (1.84) 0.27

21 65 73 17 7 0.04 (0.09) e0.27 to 0.30 86 45 42 4 6 0.05 (0.06) 0.00e0.30 5 7 8 29 76 51 7

41 21 23 10 4 60” (1.78) 69” (1.84) 0.27

64 39 30 18 28

60” (1.78) 69” (1.85) 0.29 36 22 17 10 16

100” (2.00) 148” (2.17) 0.41

42 19 26 6 6

12 40 4 13 44 13 5 17 5 17 100” (2.00) 158” (2.20) 0.44

41 22 23 11 3

52 35 35 23 26 17 13 9 23 15 60” (1.78) 148” (2.17) 0.40

Mohney et al



Observation of Children with IXT Table 2. (Continued.) Study Completion Status* All Participants (n [ 183)

Characteristic

Withdrawn (n ¼ 31)

Completed (n ¼ 152)

No.

%

No.

%

No.

%

127 5 4 39 7

70 3 2 21 4

22 2 0 6 1

71 6 0 19 3

105 3 4 33 6

70 2 3 22 4

0 5 34 71 40 21 12

0 3 19 39 22 11 7

0 0 8 15 6 1 1

0 0 26 48 19 3 3

0 5 26 56 34 20 11

0 3 17 37 22 13 7

x

Exotropia type Basic Convergence insufficiency High AC/A Pseudodivergence excess True divergence excess Exotropia control score at distance13 No exodeviation (0) No exotropia unless dissociated, recovers (1) No exotropia unless dissociated, recovers (2) No exotropia unless dissociated, recovers (3) Exotropia <50% of 30 sec (4) Exotropia >50% of 30 sec (5) Constant exotropia Mean (SD) Exotropia control score at near13 No exodeviation (0) No exotropia unless dissociated, recovers (1) No exotropia unless dissociated, recovers (2) No exotropia unless dissociated, recovers (3) Exotropia <50% of 30 sec (4) Exotropia >50% of 30 sec (5) Constant exotropia Mean (SD) Deviation type at distance Constant exotropia Intermittent exotropia Deviation type at near Constant exotropia Intermittent exotropia Exophoria No exodeviation Exotropia (D) by SPCT at distance 0 (no measurable tropia)k 1e9 10e14 16e18 20e25 30e35 40e50 Median Range Exotropia (D) by SPCT at near 0 (no measurable tropia) 1e9 10e14 16e18 20e25 30e35 Median Range Exodeviation (D) by PACT{ at distance No exodeviation (orthophoria) 1e9 10e14 16e18 20e25 30e35 40e45 50 Mean (SD) Range

<1 sec 1e5 sec >5 sec

2.4 (1.2) <1 sec 1e5 sec >5 sec

5 58 71 28 18 2 1

2.1 (0.9) 3 32 39 15 10 1 1

2 9 11 4 3 1 1

1.1 (1.0)

2.5 (1.2) 6 29 35 13 10 3 3

3 49 60 24 15 1 0

1.2 (1.3)

2 32 39 16 10 1 0 1.0 (1.0)

9 174

5 95

1 30

3 97

8 144

5 95

1 129 49 4

1 70 27 2

1 20 9 1

3 65 29 3

0 109 40 3

0 72 26 2

48 15 18 30 57 13 2

26 8 10 16 31 7 1

11 3 2 5 9 1 0

35 10 6 16 29 3 0

37 12 16 25 48 12 2

24 8 11 16 32 8 1

16 0e45 108 23 21 13 14 4

12 0e30 59 13 11 7 8 2

19 3 3 3 3 0

0 0e30 d d 7 34 101 35 5 1

61 10 10 10 10 0

89 20 18 10 11 4

0 0e25 d d 4 19 55 19 3 1

23 (7) 10e50

17 0e45

— — 3 5 16 7 0 0

0 0e30 d d 10 16 52 23 0 0

22 (6) 10e35

59 13 12 7 7 3

d d 4 29 85 28 5 1

d d 3 19 56 18 3 1 24 (7) 10e50 (Continued)

5

Ophthalmology Volume -, Number -, Month 2019 Table 2. (Continued.) Study Completion Status* All Participants (n [ 183) Characteristic

Withdrawn (n ¼ 31)

Completed (n ¼ 152)

No.

%

No.

%

No.

%

6 19 49 32 52 21 4

3 10 27 17 28 11 2

1 3 6 7 10 3 1

3 10 19 23 32 10 3

5 16 43 25 42 18 3

3 11 28 16 28 12 2

Exodeviation (D) by PACT at near No exodeviation (orthophoria) 1e9 10e14 16e18 20e25 30e35 40e45 Mean (SD) Range {

18 (9) 0e45

19 (9) 0e40

18 (9) 0e45

AC/A ¼ accommodative convergence/accommodation; arcsec ¼ seconds of arc; D ¼ diopter; logMAR ¼ logarithm of the minimum angle of resolution; log arcsec ¼ logarithm of seconds of arc; PACT ¼ prism and alternate cover test; SD ¼ standard deviation; SPCT ¼ simultaneous prism and cover test; D ¼ prism diopter; d ¼ not applicable. *Completed refers to participants who either met the primary deterioration outcome by 3 years or completed 3 years of follow-up without ever meeting deterioration criteria; withdrawn refers to participants who withdrew early without having met deterioration. y One participant had >2-line interocular visual acuity difference and was ineligible. z The distance stereoacuity measurement is missing for 4 participants (2%). Note that “nil” distance stereoacuity is assigned as 2.9 log arcsec, which is 1 octave (0.3 log arcsec) more than the highest disparity measured by the test. x See protocol at www.pedig.net for details of exotropia classification. One participant (<1%) was not classified. k “No measurable tropia” includes participants who met any of the following: (1) did not demonstrate a tropia during the examination, (2) demonstrated an exotropia not detectable by the covereuncover test, and (3) demonstrated an exotropia that was not measurable because it was too brief or too small or the participant was not cooperative enough to allow a SPCT measurement. { Distance exodeviation was required to be 10 D by PACT for eligibility. No esodeviations were present at distance or near at either baseline or 3 years.

started any treatment are listed in Table 6. Statistically significant improvement between baseline and the 3-year visit was observed in stereoacuity at distance (mean improvement, 0.14 logarithm of arcsec [approximately one-half octave]; P < 0.001) and near (mean

improvement, 0.14 logarithm of arcsec; P  0.001), distance exotropia control (mean improvement, 0.6 points; P  0.001), and distance PACT magnitude (mean improvement, 2.2 D; P ¼ 0.002); no statistically significant change was observed for near exotropia

Table 4. Deterioration at Any Time during 3 Years by Reason (Primary Outcome; n ¼ 183)* Month 0

3

6

12

18

24

30

36

No. at risk 183 173 165 158 147 138 132 128 10 2 3 6 3 3 4 0 No. with early withdrawaly z d 1 1 No. with motor deterioration d 3 3 3 1 1 No. with stereoacuity deteriorationk No. treated without meeting motor d 2 1 1 5 3 0 0 or stereoacuity deterioration criteria Cumulative no. (%) with motor or d 6 (3%) 10 (6%) 15 (9%) 21 (13%) 24 (14%) 24 (14%) 25 (15%) stereoacuity deterioration or treatment without meeting criteria for either type of deterioration

Cumulative % (95% Confidence Total Interval) d 31 2x 11 12{

d d 1 (0e5) 7 (4e12) d

25

15 (10e22)

d ¼ not applicable. *Deterioration, the primary outcome, comprised motor deterioration, stereo deterioration, or receiving treatment without meeting motor or stereoacuity deterioration criteria. y Number of participants who withdrew early without having met deterioration criteria or starting treatment; an additional 6 participants withdrew early after having met deterioration criteria or starting treatment. z Constant exotropia of 10 D by simultaneous prism and cover test at distance and near, confirmed by a retest. x One participant with 40” stereoacuity and near and constancy during stereoacuity testing not verified; 1 participant with 100” stereoacuity and uncertain if constancy during stereoacuity testing was verified. k Decrease in Randot preschool near stereoacuity of 2 octaves from best previous measure, confirmed by a retest. { Of the 12 participants who were considered to have deteriorated in the primary analysis because treatment was prescribed without meeting motor or stereoacuity deterioration criteria: 3 participants (2%) underwent strabismus surgery and 9 participants (5%) were prescribed nonsurgical treatment (patching for 7 participants, vision therapy for 1 participant, patching plus vision therapy for 1 participant); 8 participants met protocol-approved exceptions for starting treatment (7 participants for overwhelming social concern, 1 participant for debilitating diplopia) and 4 participants did not meet protocolapproved exceptions.

6

Mohney et al



Observation of Children with IXT

Table 5. Deterioration Outcomes by 3 Years (n ¼ 183) Outcome

No. with Outcome

Deterioration* by 1 yr Deterioration* by 2 yrs Deterioration* by 3 yrs Motor or stereoacuity deterioration by 3 yrsy Stereoacuity deterioration by 3 yrsy Motor deterioration by 3 yrsy

15 24 25 13 11 2

Kaplan-Meier Cumulative Probability (%; 95% Confidence Interval) 9 14 15 8 7 1

(5e14) (10e21) (10e22) (5e13) (4e12) (0e5)

*Deterioration because of motor or stereo deterioration or by starting nonrandomized treatment. y For the 3 cause-specific outcomes, participants who met criteria other than that particular criteria of interest remained at risk for the criterion of interest unless they were prescribed treatment. For example, participants who meet the stereo loss outcome remained at risk for the motor outcome until they either met that outcome or were prescribed treatment.

control or near exodeviation magnitude (Table 6; Figs 2e4). For 3year continuous measures of stereoacuity, control, and exodeviation magnitude by PACT, the imputed data for the full cohort (data not shown) were similar to the observed data for the 132 participants who completed the 3-year visit and had not been treated.

Treatment during Study Eighteen participants (10%) underwent IXT treatment during the study; these included 2 of the 2 participants who met motor deterioration criteria, 4 of the 11 participants who met stereo deterioration criteria, and 12 participants who were treated without meeting either set of criteria. Six participants (3%) underwent strabismus surgery (including 1 who also received patching), 10 participants (5%) were prescribed patching (including 1 who was also prescribed other nonsurgical treatment), and 2 participants (1%) were prescribed other nonsurgical treatment (1 overminus lenses, 1 vision therapy).

Discussion Progression to constant exotropia or loss of stereoacuity was uncommon in this study of children 3 to 10 years of age with IXT who were followed up without treatment for 3 years. There was, in fact, a small mean improvement in distance stereoacuity, near stereoacuity, distance control, and magnitude of distance exodeviation over the 3-year period.

Figure 1. Graph showing cumulative proportion with deterioration at any time during 3 years of follow-up (primary outcome). Parentheticals indicate 95% confidence intervals.

Our protocol-specified estimate of deterioration by 3 years (15%; 95% CI, 10%e22%) likely was overestimated for several reasons. First, approximately half (12 of 25) of those who were considered to have “deteriorated” did not meet the study definition for motor or stereoacuity deterioration, or both. Instead, they were treated for social and other reasons, and therefore counted as having deteriorated. Second, both of the participants classified as having motor deterioration showed measurable random dot stereoacuity at near (one demonstrated 40 arcsec and the other demonstrated 100 arcsec). Given that a constant moderate angle of exotropia is inconsistent with these levels of random dot stereoacuity, we suspect that these participants were not tropic during the stereoacuity assessment, and therefore may not have met the constancy requirement for motor deterioration. Finally, deterioration on the basis of stereoacuity similarly seems to have been overestimated because of the high degree of testeretest variability when measuring stereoacuity in children with IXT,17 which is partly attributable to the characteristic intermittency of the disease. We defined deterioration of stereoacuity as a 2-octave decrease, whereas others have reported a threshold of 3 octaves for real change in IXT.17,20 The 2-octave definition may have biased our study to finding “deterioration” when the reduction in stereoacuity actually was within normal variability. In an effort to mitigate the effect of testeretest variability for stereoacuity testing, we required a same-day confirmatory retest. Nevertheless, despite the same-day retest requirement, none of the 4 untreated patients who initially met the stereoacuity deterioration criteria still met that deterioration criteria at the 3-year final examination. A similar phenomenon was observed in a retrospective study20 with stereoacuity deterioration criteria similar to ours. That study reported a very low rate of near stereoacuity loss and found that 4 of 6 children with IXT meeting stereoacuity deterioration criteria subsequently demonstrated baseline results again, despite not being treated. We found small but statistically significant mean improvements in distance and near stereoacuity, distance control, and distance exodeviation magnitude over the 3-year period in participants who were not treated during the study. The significance of these findings are unknown and in part may indicate, at least for the stereopsis measures, increasing maturity among the participants younger than 6 years of age

7

Ophthalmology Volume -, Number -, Month 2019 Table 6. Three-Year Stereoacuity, Exotropia Control, and Ocular Alignment in Participants Who Completed the 3-Year Visit and Had Not Been Prescribed Treatment Anytime during the Study (n ¼ 132) Results Preschool Randot stereoacuity at near After 3 yrs of observation Mean arcsec (log arcsec) 95% CI of mean arcsec (log arcsec) SD (log arcsec) Range (log arcsec) Change between baseline and 3 yrs* Mean (SD) (log arcsec) 95% CI of mean Range (log arcsec) P value No. with baseline near stereoacuity 200” or worsey No. (%) improved 2 octaves (0.6 log arcsec) from baseline 95% CI of percentage No. with baseline near stereoacuity 400” or bettery No. (%) worsened 2 octaves (0.6 log arcsec) from baseline 95% CI of percentage Distance Randot stereoacuity After 3 yrs of observation Mean arcsec (log arcsec) 95% CI of mean arcsec (log arcsec) SD (log arcsec) Range (log arcsec) Change between baseline and 3 years* Mean (SD) (log arcsec) 95% CI of mean Range (log arcsec) P value No. with baseline distance stereoacuity 400” or worsey No. (%) improved 2 octaves (0.6 log arcsec) from baseline 95% CI of percentage No. with baseline distance stereoacuity 200” or bettery No. (%) worsened 2 octaves (0.6 log arcsec) from baseline 95% CI of percentage Exotropia control at distance After 3 yrs of observation Mean (SD) 95% CI of mean Range Change between baseline and 3 yrsz Mean (SD) 95% CI of mean Range P value No. with baseline control 3 pointsy No. (%) improved 3 points from baseline 95% CI of percentage No. with baseline control 2 pointsy No. (%) worsened 3 points from baseline 95% CI of percentage Exotropia control at near After 3 yrs of observation Mean (SD) 95% CI of mean Range Change between baseline and 3 yrsz Mean (SD) 95% CI of mean Range P value No. with baseline control 3 pointsy No. (%) improved 3 points from baseline 95% CI of percentage

8

Participants (n [ 132)

50” (1.70) 46”e54” (1.66e1.73) (0.21) 40”e1600” (1.6e3.2) 0.14 (0.29) 0.09e0.19 e1.20 to 1.00 <0.001 18 12 (67) 41e87 132 1 (<1) 0 to 4

107” (2.03) 93”, 123”(1.97, 2.09) (0.36) 1.78 to 2.90 0.14 (0.40) 0.07e0.21 e1.12 to 1.12 <0.001 32 19 (59) 41e76 100 7 (7) 3e14

1.8 (1.4) 1.6e2.1 0e5 0.6 (1.5) 0.3e0.9 e4 to 5 <0.001 55 16 (29) 18e43 77 5 (6) 2e15

0.9 (1.1) 0.7e1.1 0e5 0.1 (1.2) e0.1 to 0.3 e4 to 3 0.42 13 6 (46) 19e75

Mohney et al



Observation of Children with IXT Table 6. (Continued.)

Results No. with baseline control 2 pointsy No. (%) worsened 3 points from baseline 95% CI of percentage PACT exodeviation at distance (D) After 3 yrs of observation Mean (SD) 95% CI of mean Range Change between distance baseline and 3 yrsx Mean (SD) 95% CI of mean Range P value No. with baseline PACT at distance 10Dy No. (%) decreased 8 D from baseline 95% CI of percentage No. with baseline PACT at distancey No. (%) increased 8 D from baseline 95% CI of percentage PACT exodeviation at near (D) After 3 yrs of observation Mean (SD) 95% CI of mean Range Change between baseline and 3 yrsx Mean (SD) 95% CI of mean Range P value No. with baseline PACT near 13 Dy No. (%) decreased 13D from baseline 95% CI of percentage No. with baseline PACT at neary No. (%) increased 13 D from baseline 95% CI of percentage

Participants (n [ 132) 119 4 (3) 1e8

21.0 (8.7) 19.5e22.5 0e45 2.2 (8.3) 0.8e3.7 e20 to 35 0.002 132 22 (17) 11e24 132 13 (10) 5e16

17.0 (10.1) 15.2e18.7 0e45 0.4 (9.3) e1.2 to 2.0 e24 to 29 0.60 86 10 (12) 6e20 132 12 (9) 5e15

arcsec ¼ seconds of arc; CI ¼ confidence interval; log arcsec ¼ logarithm of seconds of arc; PACT ¼ prism and alternate cover test; SD ¼ standard deviation; D ¼ prism diopters. To help address possible informative censoring (i.e., whether loss to follow-up was related to outcome) in the continuous outcomes, sensitivity analyses were conducted using multiple imputation with the Monte Carlo Markov chain method15 to impute data for participants who did not complete the full 3 years of follow-up. The imputation model for each 3-year outcome (e.g., control, stereo) was based on baseline and interim follow-up data for that assessment. The imputed data for the full cohort (data not shown) was similar to the observed data for 132 participants who completed the 3-year visit and had not undergone treatment. *Change in stereoacuity is calculated as baseline level minus 3-year level, so positive change indicates improvement. y Note that binary improvement and worsening outcomes were limited to participants whose baseline data allowed them the room to improve (or worsen) by the specified amount. z Change in control is calculated as baseline level minus 3-year level, so positive change indicates improvement. Improvement in control was defined as an improvement of 3 points based on the 3-point threshold for real change.18 x Change in PACT results is calculated as baseline level minus 3-year level, so positive change indicates improvement (i.e., decrease in magnitude over time). Improvement in PACT results at distance and near were defined as a decrease of 8 D and 13 D, respectively, because these amounts exceed the repeatability coefficients of 7.2 D and 12.8 D for PACT angles larger than 20 D at distance and near.19 No esodeviations were present at distance or near at either baseline or 3 years.

at enrollment.10 However, improvement in control and a smaller distance deviation magnitude would not be expected to be associated with increasing age, and thus likely reflect a slight overall improvement in the underlying condition. Additionally, regression to the mean also may explain a portion of these apparent improvements. Near control and the near exodeviation magnitude remained stable, but these measurements had less room to improve than their corresponding measurements at distance.

There are few published reports with which to compare these findings. Prior reports are almost exclusively retrospective studies from single institutions with various durations of follow-up and definitions of deterioration. Moreover, these reports focus primarily on a change in the magnitude of the exodeviation as a sign of worsening, and some included patients who received medical treatment for IXT.3,5,21 Reported outcomes from these studies range from overall improvement,3,5 to overall worsening,7,8 to no change at all.4,6 Buck

9

Ophthalmology Volume -, Number -, Month 2019

Figure 2. Graphs showing baseline versus 3-year stereoacuity for participants who completed the 3-year visit and had not been prescribed treatment anytime during the study (n ¼ 132). A, Distance stereoacuity. B, Near stereoacuity. Center diagonal line represents the same score for baseline and 3 years. The upper diagonal line indicates worsening of 2 octaves (0.6 logarithm of seconds of arc); the lower diagonal line indicates improvement of 2 octaves (0.6 logarithm of seconds of arc).17

Figure 3. Graphs showing baseline control versus 3-year control for participants who completed the 3-year visit and had not been prescribed treatment anytime during the study (n ¼ 132). A, Distance control. B, Near control. Center diagonal line represents the same control score for baseline and 3 years. The upper diagonal line indicates worsening of 3 points on a scale between 0 (phoria) and 5 (constant) based on the 3-point threshold for real change18; the lower diagonal line indicates improvement of 3 points.

et al,22 who prospectively observed 195 children with untreated IXT over a 2-year period, reported that only 0.5% of participants (1 of 191 completing follow-up) showed deterioration to a constant exotropia. For most of the 195 children with untreated IXT (85%), mean control of the IXT (assessed using the Newcastle control scale23) remained stable. Similar to our study, they found that the mean magnitude of exodeviation at both distance and near did not worsen. There are several limitations to this study. It is possible that our participants did not represent the full clinical spectrum of IXT because those with poorer or worsening control or whose parents or eye care providers, or both, were concerned about the child being followed up without treatment for 3 years (unless deterioration criteria were met) may have been enrolled into the concurrent PEDIG surgical trial of IXT (clinicaltrials.gov identifier, NCT01032330) at participating sites or may not have been enrolled into either study. We did not collect data on which study or studies were offered to each patient or on those patients who declined participation. Nevertheless, the IXT participants in this study showed a wide range of exotropia

control and magnitude of exodeviation at baseline. Additionally, although we do not know the outcome for the 17% of participants who did not show deterioration and were lost to follow-up, we have no reason to believe that these participants, who included a higher proportion of males, were 6 months older on average, and showed a slightly larger baseline exotropia by simultaneous prism and cover test, would affect our results substantively. Moreover, some participants were treated without meeting formal deterioration criteria, making it impossible to know their true outcomes and further lessening the pool of participants observed for the full duration of the study. Classifying each of these participants as having deteriorated by 3 years almost certainly resulted in an overestimation of the rate of deterioration. Omitting all participants who started treatment (either after meeting formal deterioration criteria or without meeting this criteria) from motor or stereo deterioration at the 3-year time point could have excluded patients who might have had worse outcomes at 3 years had they not been treated. Our criteria for both stereoacuity and motor deterioration also were subject to misclassification because of

10

Mohney et al



Observation of Children with IXT or deterioration of stereoacuity over a 3-year period was uncommon. We believe the actual rate is likely lower than the observed 15%, given that a substantial number of participants underwent treatment without meeting deterioration criteria or showed an apparent reduction in stereoacuity that later resolved without treatment. Although treatment may be appropriate for psychosocial or symptomatic concerns, treatment is not necessary to prevent motor deterioration or loss of stereoacuity over a 3-year period for most young children with IXT.

References

Figure 4. Baseline versus 3-year prism and alternate cover test (PACT) results for participants who completed the 3-year visit and had not been prescribed treatment anytime during the study (n ¼ 132). A, Distance PACT results. B, Near PACT results. Center diagonal line represents the same PACT results for baseline and 3 years. The upper diagonal line indicates worsening that exceeds published repeatability coefficients based on testeretest data (8 D for distance PACT and 13 D for near PACT)19; the lower diagonal line indicates improvement of those amounts.

testeretest variability; if we had required a confirmation on a subsequent day, we might have avoided classifying “deteriorations” that resulted from a young child’s tiredness or inattentiveness on a given day. In addition, stereoacuity deterioration was defined as worsening from the best previous stereoacuity measure rather than from the baseline measure. This allowed for a release to treatment in case of worsening stereopsis after any age-related or other improvement in stereoacuity during the study; however, it also resulted in some best stereoacuity measurements being based on a random high measurement from a series of visits,17 causing the “worsening” criterion to be met more easily. Finally, we did not directly evaluate suppression or depth of suppression in this study, and it is possible that the development of suppression may be detrimental in and of itself or may have an adverse effect in the long term or on future treatment effectiveness. In summary, for children 3 to 10 years of age with IXT for whom surgery was not considered as the immediately needed initial treatment, progression to a constant exotropia

1. Govindan M, Mohney BG, Diehl NN, Burke JP. Incidence and types of childhood exotropia: a population-based study. Ophthalmology. 2005;112:104e108. 2. Fu J, Li SM, Liu LR, et al. Prevalence of amblyopia and strabismus in a population of 7th-grade junior high school students in Central China: the Anyang Childhood Eye Study (ACES). Ophthalmic Epidemiol. 2014;21: 197e203. 3. Hiles DA, Davies GT, Costenbader FD. Long-term observations on unoperated intermittent exotropia. Arch Ophthalmol. 1968;80:436e442. 4. Chia A, Seenyen L, Long QB. A retrospective review of 287 consecutive children in Singapore presenting with intermittent exotropia. J AAPOS. 2005;9:257e263. 5. Rutstein RP, Corliss DA. The clinical course of intermittent exotropia. Optom Vis Sci. 2003;80:644e649. 6. Romanchuk KG, Dotchin SA, Zurevinsky J. The natural history of surgically untreated intermittent exotropia-looking into the distant future. J AAPOS. 2006;10:225e231. 7. Nusz KJ, Mohney BG, Diehl NN. The course of intermittent exotropia in a population-based cohort. Ophthalmology. 2006;113:1154e1158. 8. von Noorden GK, Campos EC. Exodeviations. In: Lampert R, Cox K, Burke D, eds. Binocular Vision and Ocular Motility: Theory and Management of Strabismus. St. Louis: CV Mosby; 2002:356e376. 9. Pediatric Eye Disease Investigator Group. A randomized trial comparing part-time patching with observation for children 3e10 years of age with intermittent exotropia. Ophthalmology. 2014;121:2299e2301. 10. Birch E, Williams C, Drover J, et al. Randot Preschool Stereoacuity Test: normative data and validity. J AAPOS. 2008;12: 23e26. 11. Fu VL, Birch EE, Holmes JM. Assessment of a new distance Randot stereoacuity test. J AAPOS. 2006;10:419e423. 12. Holmes JM, Birch EE, Leske DA, et al. New tests of distance stereoacuity and their role in evaluating intermittent exotropia. Ophthalmology. 2007;114:1215e1220. 13. Mohney BG, Holmes JM. An office-based scale for assessing control in intermittent exotropia. Strabismus. 2006;14: 147e150. 14. Pintile M. Competing Risks: A Practical Perspective. First ed. Chichester, England: John Wiley & Sons, Ltd; 2006. 15. Little RJA, Rubin DB. Statistical Analysis with Missing Data. New York: Wiley; 1987. 16. Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res. 1999;8:135e160. 17. Adams WE, Leske DA, Hatt SR, Holmes JM. Defining real change in measures of stereoacuity. Ophthalmology. 2009;116:281e285.

11

Ophthalmology Volume -, Number -, Month 2019 18. Hatt SR, Leske DA, Liebermann L, Holmes JM. Quantifying variability in the measurement of control in intermittent exotropia. J AAPOS. 2015;19:33e37. 19. Hatt SR, Leske DA, Liebermann L, et al. Variability of angle of deviation measurements in children with intermittent exotropia. J AAPOS. 2012;16:120e124. 20. Holmes JM, Leske DA, Hatt SR, et al. Stability of near stereoacuity in childhood intermittent exotropia. J AAPOS. 2011;15:462e467.

21. Friendly D. Surgical and nonsurgical management of intermittent exotropia. Ophthalmol Clin North Am. 1992;5:23e30. 22. Buck D, Powell CJ, Rahi J, et al. The improving outcomes in intermittent exotropia study: outcomes at 2 years after diagnosis in an observational cohort. BMC Ophthalmol. 2012;12: 1e7. 23. Buck D, Clarke MP, Haggerty H, et al. Grading the severity of intermittent distance exotropia: the revised Newcastle control score. Br J Ophthalmol. 2008;92:577.

Footnotes and Financial Disclosures Originally received: October 24, 2018. Final revision: January 17, 2019. Accepted: January 18, 2019. Available online: ---. 1

Manuscript no. 2018-2401.

Department of Ophthalmology, Mayo Clinic, Rochester, Minnesota.

2

Southern California College of Optometry at Marshall B. Ketchum University, Fullerton, California.

3

Jaeb Center for Health Research, Tampa, Florida.

4

Department of Ophthalmology, Indiana University, Indianapolis, Indiana.

5

Rocky Mountain Eye Care Associates, Salt Lake City, Utah.

6

Concord Eye Center, Concord, New Hampshire.

HUMAN SUBJECTS: Human subjects were included in this study. The study was supported through a cooperative agreement with the National Eye Institute of the National Institutes of Health and was conducted according to the tenets of the Declaration of Helsinki by the Pediatric Eye Disease Investigator Group (PEDIG) at 60 academic and community-based clinical sites. The protocol and Health Insurance Portability and Accountability Act compliant informed consent forms were approved by each respective institutional review board. A parent or guardian of each study participant gave written informed consent. No animal subjects were included in this study. Author Contributions: Conception and design: Mohney, Cotter, Chandler, Holmes, Wallace, Yamada, Kraker, Morse, Melia

Presented at: Association of Research in Vision and Ophthalmology Annual Meeting, May 2016, Seattle, Washington; and European Strabismological Association/American Association for Pediatric Ophthalmology and Strabismus (ESA/AAPOS) Joint Meeting, September 2017, Porto, Portugal.

Analysis and interpretation: Mohney, Cotter, Chandler, Holmes, Wallace, Yamada, Kraker, Morse, Melia, Wu

Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article.

Obtained funding: Holmes, Kraker, Wallace

Supported by the National Eye Institute, National Institutes of Health, Bethesda, Maryland (grant nos.: EY011751, EY023198, and EY018810). The funding organization had no role in the design or conduct of this research. A complete list of participating members of the Pediatric Eye Disease Investigator Group (PEDIG) can be found in Acknowledgments (available at www.aaojournal.org).

12

Data collection: Mohney, Cotter, Chandler, Holmes, Wallace, Yamada, Petersen, Kraker, Morse Overall responsibility: Mohney, Cotter, Chandler, Holmes, Wallace, Yamada, Petersen, Kraker, Morse, Melia, Wu Abbreviations and Acronyms: arcsec ¼ seconds of arc; CI ¼ confidence interval; IXT ¼ intermittent exotropia; PACT ¼ prism and alternate cover test; PEDIG ¼ Pediatric Eye Disease Investigator Group; D ¼ prism diopter. Correspondence: Brian G. Mohney, MD, Jaeb Center for Health Research, 15310 Amberly Drive, Suite 350, Tampa, FL 33647. E-mail: [email protected]