Prognostic preoperative factors for successful outcome of surgery in horizontal strabismus.

CONTEXT: Surgery for horizontal strabismus reportedly has a success rate of 60%-80%. However, which preoperative factors are predictive of this success is not clear. AIMS: To identify prognostic factors those are predictive of successful outcome in horizontal strabismus surgery. SETTINGS AND
DESIGN: Observational analytical study using multiple logistic regression (MLR). SUBJECTS AND METHODS: We assessed the medical records of patients who had undergone first-time horizontal muscle strabismus surgery between 2002 and 2013, where complete follow-up data were available for ≥6 weeks, and also, we collected data prospectively on patients operated between January 2014 and September 2015. Successful outcome was defined as a postoperative angle of deviation within 10 prism diopter of orthophoria at ≥6 weeks postoperatively. Independent variables considered were age at onset, age at surgery, duration, gender, deviation - type and amount, logMAR visual acuity (VA) - mean and of the poorer eye, mean refractive error, amount of anisometropia, and presence of dense amblyopia. Only those with P < 0.2 on univariate analyses (UAs) were included in the MLR, with significance set at P ≤ 0.05. STATISTICAL ANALYSES: UA (Chi-square for categorical variables and t-tests for continuous variables), followed by logistic regression analysis.
RESULTS: Of 113 patients, on UA, type of deviation (P = 0.01), age at surgery (P = 0.16), absence of dense amblyopia (P = 0.002), and logMAR VA of the poorer eye (P = 0.005) qualified for the inclusion in MLR. On MLR, esotropia (ET) (odds ratio [OR]: 4.46) and absence of dense amblyopia (OR: 5.90) were associated with success.
CONCLUSIONS: With an overall success rate of 83%, ET and absence of dense amblyopia were significantly predictive of surgical success.

Introduction

The prevalence of strabismus in general varies between 2% and 5%.[1234] In children, esotropia (ET) is 3–5 times more common than exotropia (XT).[345] Strabismus usually begins before 6 years of age and peaks at 3 years.[6]

Apart from improving the quality of life and psychosocial aspects, surgery offers the possibility of recovery of binocularity. In addition, it serves to eliminate diplopia, increase visual fields, correct abnormal head posture, and improve employment prospects. Successful alignment is considered when postoperative deviation is within 10 prism diopter (PD) of orthophoria.[7891011] Surgery for horizontal strabismus has a success rate between 60% and 80%.[11121314] By its very nature, it is difficult to predict successful outcome with certainty. Moreover, which preoperative factors are predictive of this success is unclear. Some studies suggest that a smaller preoperative deviation (POD)[101115] and earlier age at surgery[811] are associated with success, while others imply that amblyopia may adversely affect the outcome.[816] Few attempts have been made to identify prognostic factors associated with a successful surgical outcome, following strabismus. Due to lack of studies from India, we decided to undertake this exercise, using a logistic regression model.

We planned to identify prognostic factors predictive of successful outcome, following surgery in horizontal strabismus.

Subjects and Methods

After clearance from the Board of Studies and the Ethical Committee of the Faculty of medicine, we retrospectively assessed data of the patients who had undergone strabismus surgery during 2002–2013 and prospectively collected data from 2014 to 2015. We included cases of concomitant horizontal strabismus, with vertical deviation ≤5 PD, needing surgery only on the horizontal recti. Successful surgery was considered if final postoperative deviation (FPD) at ≥6 weeks was within 10 PD of orthophoria. We excluded patients with nystagmus, paralytic or restrictive strabismus, neurological problems, previously operated and those who underwent adjustable muscle surgeries. An exhaustive history including details about age at onset, duration, diminution of vision, diplopia, and family history, or any prior treatment was taken. A complete general, ocular, and orthoptic workup was done including logMAR visual acuity (VA) and dry and wet retinoscopy. The interocular difference between spherical equivalents constituted the degree of anisometropia. Ocular workup included biomicroscopy, ophthalmoscopy, sensory testing, and clinical photography. A difference of ≥2 lines in best-corrected VA (BCVA) between normal and amblyopic eye was considered as amblyopia, which graded into mild (<0.30 logMAR), moderate (0.30–0.60 logMAR), and dense (>0.60 logMAR). Bilateral amblyopia implied a BCVA >0.30 logMAR. Ocular deviation was measured by prism and cover test or prism bar reflex test (PBRT/Krimsky test). Deviations were recorded preoperatively (POD) and finally at ≥6 weeks postoperative (FPD). Strabismus correction was done by a single surgeon, under peribulbar block or general anesthesia, using standard approaches. Surgical outcome was considered successful if the FPD measured ≤10 PD. For patients with large angle strabismus who underwent bilateral recession/resection procedures in two stages, FPDs were recorded after the second surgery. We listed 11 possible predictor variables: age at onset; age at surgery; duration; gender; deviation, type and amount; logMAR VA: mean and of the poorer eye; mean refractive error; amount of anisometropia; and presence of dense amblyopia.

Statistical analysis

We assessed each predictor variables using univariate analysis (UA) (Chi-square for counts/proportions and t-tests for parametric data), followed by multiple logistic regression (MLR) on selected ones. On UA, those variables with P < 0.2 were considered for the MLR model,[17] and the latter considered statistically significant at P < 0.05. Data analyses were performed with commercial software, Statistical Package for Social Sciences (SPSS IBM SPSS Statistics for Windows, version 20.0. Armonk, NY: IBM Corp.) version 20. We checked the assumptions for MLR by looking for linearity, independence of errors, and absence of multicollinearity.[18]

Results

We had data on 113 patients, with an overall success rate of 83%, having a mean age of 16.84 ± 7.25 years ; with 65 (57.5%) males to 48 (42.5%) females; 49 (43.4%) being exotropes and 64 (56.6%) esotropes; while overall 28 (24.8%) had dense amblyopia [Table 1]. On UA five of the 11 listed predictor variables (7 continuous [Table 2] and 4 categorical [Table 3] achieved – poorer-eye logMAR VA, mean logMAR VA, age at surgery, type of deviation, and absence of dense amblyopia.

Table 1

Demographic and clinical characteristics of patients (n=113)

Table 2

Univariate analysis of continuous predictor variables between failure and success groups using t-test: Difference of mean with 95% confidence interval

Table 3

Comparison of categorical predictor variables between success and failure groups using Chi-square test

Linearity assumption was met since all interaction terms for continuous predictor variables in the MLR did not achieve significance: age at surgery (P = 0.29), mean VA on logMAR (P = 0.20), VA of the poorer eye (P = 0.18)

Independence of errors: No overdispersion was found since goodness-of-fit/degree of freedom (0.025) was well under 2

Multicollinearity: Poorer VA and mean VA showed collinearity (tolerance <0.1 and variance inflation factor >10), suggesting interaction, and after consensus, the latter was dropped [Table 4].

Table 4

Collinearity statistics

We therefore had four predictors for MLR: age at surgery, type of deviation, poorer eye logMAR VA, and absence of dense amblyopia. We selected a forward likelihood ratio for variable entry [Table 5]. Our model showed that there was a significantly greater likelihood of having a successful outcome, following horizontal strabismus surgery in patients with ET and those lacking dense amblyopia [Table 6]. These two factors explain 22% of the variation in surgical outcome: As a corollary then, a sizeable 78% is still unknown.

Table 5

Summary of multiple logistic regression analysis on factors for surgical success

Table 6

Logistic regression model – 95% confidence interval for odds ratios

Discussion

In this study, we evaluated the effect of 11 candidate preoperative factors that we envisaged could affect the outcome of strabismus surgery. Our success rate of 83% for horizontal strabismus was similar to others.[11121314] However, comparisons may not be very appropriate because of variability in populations; surgical procedures and definition of success. One study, quite like ours, by Kampanartsanyakorn et al. reported a success rate of 60.2%.[11] Although two studies reported that differing surgeons have similar success rates with strabismus outcomes,[1219] we took care that all patients in our study were operated by a single surgeon so as to preclude the surgeon factor from playing any influential role.

Our study revealed that ET (as opposed to XT) and the absence of dense amblyopia were the only factors significantly predictive of surgical success. Like us, Paulita, in a recent study on children <12 years (n = 352), also found that surgical success was significantly greater with ET as compared to XT although to a lesser extent comparing with ours (their odds ratio [OR]: 1.9; 95% confidence interval [CI] 1.2–3, P = 0.008 vs. our OR: 4.46; 95% CI: 1.42–14.01, P = 0.010).[20]

One possible explanation for poorer success rate with XT compared to ET is the reported tendency of postoperative drift in the former: This is also the reason why clinicians often prefer to target slight ET postoperatively, in the former, especially in a setting of moderate to dense amblyopia.[21] Interestingly, a retrospective study by Pineles et al. on 63 patients of XT did not find that targeting initial postoperative ET was associated with long-term surgical success.[22] In contrast to Pineles et al., Choi et al. in nonrandomized retrospective case series study of 81 XT patients suggested that initial overcorrection had lower probability of recurrence within 2 years of surgery, but long-term outcome could not be predicted.[23] Randomized controlled trials with longer follow-up period may help clarify the issue.

We also found that the absence of dense amblyopia was associated with surgical success (OR: 5.90; 95% CI: 1.92–18.13, P = 0.002). Surprisingly, there seems to be little-published research clearly discussing any association of amblyopia with successful outcome.

Yurdakul and Ugurlu in a retrospective analysis on 101 patients of risk factors for consecutive XT found that amblyopia was significantly (P = 0.004) associated with development of consecutive XT.[16]

Kampanartsanyakorn et al. in their study found that age <6 years (P = 0.04) and POD <30 PD (P < 0.001) were significantly associated with successful surgery.[11] Since most of our patients were of an older age group, we have no age comparative data; moreover, we did not find any association with the size of the POD. Similar to Kampanartsanyakorn, Trigler and Siatkowski in their retrospective study also found that deviations of <30 PD were associated with lesser recurrence of horizontal strabismus (16% vs. 31%, P = 0.047).[15] On the contrary, Umazume in a retrospective multiple linear regression analysis of 179 patients paradoxically found that higher value of preoperative distance deviation was associated with surgical success (measured by degree of change in deviation per mm of surgery) in horizontal strabismus.[24] He also found that average spherical equivalent toward myopia and presences of dissociated vertical deviation (DVD) were associated with surgical success in exotropes and esotropes, respectively, at 1 and 6 months. Although we did not get any association with POD, since we had no patient with DVD, we could not assess this aspect.

Lim in a retrospective multivariate logistic regression analysis of 511 XT patients who underwent bilateral lateral rectus recessions, with minimum follow-up of a year, found that the recurrence rates decreased with increased age at surgery (OR: 0.733; 95% CI: 0.626–0.858; P < 0.001); however, on the other hand, it increased with larger deviation during the immediate postoperative period (OR: 1.106; 95% CI: 1.045–1.171; P < 0.001).[25] We did not find any association with age at surgery. Although immediate postoperative deviation was not mentioned as a possible predictor in our protocol, we conducted an analysis for any possible association and found that odds of being successful would decrease with increased immediate postoperative deviation (OR: 0.850; 95% CI: 0.781–0.925; P < 0.0001): This appears a reasonable outcome, with larger immediate postoperative deviations precluding success.

In a recent retrospective multivariate logistic regression analysis, Mireskandari et al., on 353 patients (222 with XT, 117 with ET, and 14 with hypertropia/hypotropia) of age group >12 years, who underwent recession, resection, advancement of muscles, or a combination of these and using adjustable or nonadjustable sutures with minimum follow-up of 6 months, found that achieving the target angle within 1 week postoperatively was the only significant factor for higher success (83.60% vs. 63.70%, P < 0.0001, OR: 2.9; 95% CI 1.8–4.9).[26] Overall surgical success rate was significantly higher in patients who underwent adjustable suture surgery (75.7% vs. 54%; P < 0.0001). We have excluded the patients who underwent adjustable suture surgery since it is only carried out for special indications, in our setup like a repeat surgery or in restrictive strabismus, and not as a routine.

Choi et al. in a retrospective study tried to find out the risk factors for additional surgery in 52 patients having consecutive ET. They found that only postoperative deviation >5 PD at 1 month was significantly associated with additional surgery (P = 0.035). In our study on horizontal strabismus cases, we also found that odds of being successful would decrease with larger immediate postoperative deviation.[27] Wang studied 100 XT patients (measuring 15–35 PD, age group <15 years) who underwent unilateral recession and resection with minimum follow-up of 6 months. POD was a significant factor associated with success at final visit (P = 0.041); however, it was not significant for success at 6 months (P = 0.52).[28] Other factors such as age at surgery and initial refractive error were not associated with success at any time; however, successful alignment at 6 months was found to be highly correlated with success at final visit (P = 0.0001), with a mean of 3 years and 2 months following surgery. Unfortunately, amblyopic patients were excluded in this study, so its role could not be assessed.

Our study is not without some limitations: a rather short follow-up of 6 weeks and an age group (mean age 17 years) considerably older than that of children. In our defense for the former issue, many studies have found that there is a significant correlation between success measured at 6 weeks and ≥6 months.[25272829]

Conclusion

Numerous studies and ours show that surgical success in strabismus surgery remains largely unpredictive. We could only identify a significant odd of success with ET and absence of dense amblyopia.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.