The study was approved by Nanjing Children’s Hospital Review Board, and the parent or guardian of each subject signed the informed consent prior to treatment.

In this study, the ages of subjects ranged from 3 to 7. The spherical equivalent refraction was between − 6.00 diopters(D) and + 1.00D. And patients diagnosed as intermittent exotropia met the following criteria: (1) intermittent exotropia or constant exotropia at distance (the distance control assessment at baseline was performed three times and the average value ≥2 points); (2) intermittent exotropia or exophoria at near (the near control assessment at baseline was performed three times, at least one time ≤ 4 points); (3) ocular deviation measured using prism and alternate cover test (PACT) was more than 15PD at distance or near, but at least 10PD at distance. Children who received nonsurgical treatment within 6 months were excluded.

Patients underwent a complete ophthalmic evaluation before enrollment. Exodeviation control ability was measured at distance (6 m) and near (33 cm) using a 6-point office control score (Table 1) [10, 11] which ranked from 0 (best control) ~ 5 (worst control). Control score was measured three times within 1 day because of its large variability. Stereoacuity at 40 cm was measured using Titmus stereo test.

Participants were randomly assigned to one of the three groups: patching, pencil push-ups, and observation. All exams were completed by a masked examiner.

Part-time patching: For intermittent exotropia with equal dominance, eyes were covered alternately for 12 weeks, and 2 hours every day. For intermittent exotropia with dominant eye, the dominant eye was covered for 4 days and non-dominant eye for 3 days per week, and 2 hours every day. In the case of changing dominancy during treatment, the patching regimen was altered accordingly.

Pencil push-up training: patients were instructed to hold a pencil at arm’s length distance along the midline, and an index card on the wall behind the pencil was used to control suppression by using physiological diplopia. Patients were instructed to look at the tip of the sharpened pencil and to try and keep the pencil point single while moving it toward their nose. When they perceived double image of the target even with maximum effort, the pencil was moved back slowly until they regained fusion. If suppression occurred and one of the physiologic diplopia images disappeared, the subjects were instructed to blink or shake the pencil as an antisuppression technique. If patients were able to regain a single vision, they were asked to continue moving the pencil closer, up to 5 cm from their nose. Patients were instructed to do this exercise for 3 sets of 20 pencil push-ups daily, 5 days per week [12].

Observation: patients in the observation group did not receive any intervention except refractive correction.

Patients were followed up for 12 weeks. The compliance for training and patching was assessed after discussing with the parents and by reviewing study calendars on which parents recorded the number of hours the child patched and the number of training sessions each day. Compliance was judged to be excellent (the percentage of the number of training sessions over the total number or covering time more than 75%), good (more than 50% to less than or equal 75%), fair (more than 25% to less than or equal 50%), or poor (less than or equal 25%). Control scores and stereopsis at 40 cm were then evaluated. Stereopsis became a continuous variable by converting the seconds of arc scores to log arc/sec values, for example, 40 (1.60), 50 (1.70), 60 (1.78), 80 (1.90), 100 (2.00), and 200 (2.30). Stereopsis threshold doubled (e.g. 100 to 200 arc/sec) with 0.3 change in log transformed value.

During the 12-week visits, patching compliance was observed to be excellent in 15 patients (54%), good in eight patients (29%), fair in four patients (14%), and poor in one patient (4%). Pencil push-up training compliance was observed to be excellent in 17 patients (57%), good in seven patients (23%), fair in three patients (10%) and poor in two patients (7%), and unknown in one patient (3%).

Throughout the follow-up period, unilateral patching was prescribed in eight participants (29%), alternate patching in 18 participants (64%), and both alternate and unilateral patching at different follow-up periods in two participants (7%).

Data was normally distributed and was analyzed using SPSS 19.0. Paired t-test was used to compare deviation control before and after the intervention. Chi-square test or Fisher’s exact test was used to compare counting data. One-way ANOVA followed by post hoc tests was used to compare measurement data. P-value less than 0.05 was considered statistically significant.