Vitreous Hemorrhage in Pediatric Age Group

Purpose. To identify and study causes of vitreous hemorrhage (VH) in pediatric age group and to investigate factors predicting visual and anatomical outcomes. Procedure. A retrospective review of patients aged 16 years or less with the diagnosis of vitreous hemorrhage from January 2005 until December 2010. Results. A total number of 230 patients (240 eyes) were identified. Traumatic vitreous hemorrhage accounted for 82.5%. In cases of accidental trauma, final visual acuity of 20/200 was significantly associated with visual acuity of ≥20/200 at presentation and the absence of retinal detachment at last follow-up. Patients with nontraumatic vitreous hemorrhage were significantly younger with higher rates of enucleation/evisceration/exenteration and retinal detachment at last follow-up compared to traumatic cases. Conclusion. Trauma is the most common cause of VH in pediatric age group. In this group, initial visual acuity was the most important predictor for visual outcome, and the presence of retinal detachment is a negative predictor for final good visual outcome. The outcome is significantly worse in nontraumatic cases compared to traumatic cases.


Introduction
Vitreous hemorrhage has been well investigated in the adult age group [1][2][3][4][5]. In children, however, vitreous hemorrhage is uncommon, and as a result, it is not studied as well as in the adult age group and little is known about it. On literature review, different causes have been published, but most of these publications are either case reports or a small series describing particular group of patients or different modalities of treatment . To the best of our knowledge, there are only two studies that investigated vitreous hemorrhage as a separate entity in children in English literature and these studies showed different causes than that published in adults [27,28]. In these two studies trauma was the most common cause of vitreous hemorrhage in children accounting for 73.1% [28] and 68.5% [27]. Ocular trauma is the leading cause of monocular visual disability and monocular blindness in children. Vitreous hemorrhage has been reported in different studies (either in children alone or in both adult and children) as one of poor visual prognostic signs in both open globe injury [29][30][31][32][33] and closed globe injury [34]. In a large series of ocular trauma in children, vitreous hemorrhage was documented in 6.7% of the eyes [35]. Another study reported VH in 20% of eyes after open globe injury in children [36]. Traumatic vitreous hemorrhage in children can lead to tractional retinal detachment and occlusion amblyopia [34,37]. Spirn et al. reported that the most common cause of nontraumatic vitreous hemorrhage in children was regressed retinopathy of prematurity (ROP) accounting for 5.9% [28]. In a recent study from India, Rishi et al. demonstrated that idiopathic retinal periphlebitis was the most common cause in eyes with nontraumatic vitreous hemorrhage in children accounting for 12.6% [27]. Because little is known about vitreous hemorrhage in children, we 2 Journal of Ophthalmology conducted this retrospective study to investigate etiology and visual and anatomic outcomes of vitreous hemorrhage in pediatric age group.

Patient and Methods
Institutional board approval was obtained for this project. We retrospectively reviewed all charts of patients aged 16 years or less and diagnosed with vitreous hemorrhage, seen at King Khaled Eye Specialist Hospital, Riyadh, Saudi Arabia, from of January 2005 until December 2010. Charts were identified through the coding system. Our inclusion criteria included any patient diagnosed with vitreous hemorrhage either in the outpatient clinic note or in the operative note, with the age of sixteen years or less. Exclusion criteria included followup duration of less than one month and the development of endophthalmitis at presentation or during the followup. A specialized data collection sheet was designed to  [38]. The mechanism of trauma was identified (sharp, blunt, fireworks, or missile). In addition the following information was collected: vision at presentation, the presence of strabismus or nystagmus, the presence of an afferent pupillary defect, cornea status, extent of the wound (≤10 mm or >10 mm), zone of Injury (zone I: cornea only, zone II: corneoscleral within 5 mm of the limbus, and zone III: corneoscleral extending >5 mm from the limbus) [33], presence of hyphema, lens status, density of vitreous hemorrhage based on indirect ophthalmoscopy (mild to moderate if there was a view to the fovea and blood vessels or severe if there was no view to the fovea and blood vessels), condition of the retina either flat or detached, B-scan findings, management that was done, follow-up duration, and clinical findings at last follow-up. The diagnosis of the cause of nontraumatic vitreous hemorrhage was based on the clinical diagnosis written in the chart of the patient. Data were analyzed first to identify the causes of vitreous hemorrhage. Univariate analysis was used to identify factors that predicted the visual and anatomical outcomes in traumatic cases by studying the following variables: age of the patient, initial vitreous hemorrhage, mechanism of injury, wound length, wound location, lens injury at presentation, iris injury, hyphema, density of vitreous hemorrhage, retinal detachment at presentation, optic nerve avulsion, vitrectomy performed, lens status on last follow-up (clear, cataract, pseudophakic, or aphakic), and the presence of retinal detachment. The same factors were used to determine  factors affecting the anatomical outcome. Then we compared traumatic and nontraumatic cases of vitreous hemorrhage.

Statistical Methods
Comparisons between two categorical variables were made using the Chi-square test or Fisher's exact test, as appropriate. Student's -test was used to compare the difference between two proportions from the same sample. A value less than 0.05 indicated statistical significance. Stepwise logistic regression analysis was conducted to identify the variables that tended to influence the achievement of final visual acuity of 20/200 or better after adjustment for all other factors. A 95% confidence interval (CI) around the odds ratio that did not include a value of 1.0 indicated statistical significance. The statistical software used for the analyses included SPSS Version 15.0, Stats Direct Version 1.9, and BMDP 2007.

Results
During the study period 230 patients (240 eyes) met our inclusion criteria and were included in the study. Causes of vitreous hemorrhage were classified into traumatic and nontraumatic (Table 1). Traumatic vitreous hemorrhage accounted for 198 out of 240 (82.5%) eyes. Of these 198 eyes, 11 developed vitreous hemorrhage after surgical intervention (nonaccidental trauma) (7 eyes after glaucoma surgeries, 2 eyes after secondary intraocular lens implantation, and 2 eyes after retinal detachment repair). Nontraumatic cases accounted for 42 out of 240 (17.5%) eyes. Table 2 shows the demography of these patients.

Nontraumatic Vitreous Hemorrhage.
Causes of nontraumatic vitreous hemorrhage are illustrated in Table 1. This group was divided into four subgroups according to the age at presentation, (Table 3). In the group of patients aged less than one year, spontaneous idiopathic cause was the most common cause (Figure 1(a)). In the group of patients aged more than one year up to 5 years, spontaneously regressed retinopathy of prematurity (ROP), retinoblastoma (RB), and Terson's syndrome were the most common causes (Figure 1(b)). In the group of patients aged more than 5 years up to 10 years, acute lymphoblastic leukemia and intermediate uveitis were the most common causes (Figure 1(c)). In the group of patients aged more than 10 years up to 16 years, familial exudative vitreoretinopathy (FEVR) and exudative retinitis pigmentosa were the most common causes (Figure 1(d)). All patients with Terson's syndrome were related to head trauma and had a history of intensive care unit admission. Three patients had history of subdural hematoma and one patient had history of cerebral hematoma. The mean follow-up for nontraumatic VH cases was 25.8 ± 22.1 months, the median was 16 months, and the range was 2-75 months. Table 4 is a summary of the clinical findings at presentation, management, and the outcome of this group of patients.
Observation was the modality of management for the nontraumatic cases in 19 out of 42 (45.2%) eyes. Incisional management was the modality of treatment done for 17 out of 42 (40.5%) eyes that included vitrectomy with or without tamponade or retinal detachment surgery and enucleation/exenteration. Vitrectomy was done for 12 out of 42 (28.6%) eyes and enucleation/exenteration was done for 5 out of 42 (11.9%) eyes. In addition, laser therapy was the only modality of treatment done for 6 out of 42 (14.3%) eyes.  (1) LP (1) Enucleation (4) Exenteration (1) Prosthesis (5) (2) Terson's syndrome Observation (3) Prosthesis (1) 20/100 (2) Prosthesis (1) No F& F, inoperable TRD (1) Laser therapy (2) 20  (2) Observation (2) No F&F, both inoperable TRD + RRD (2)       [39,40], Marfan syndrome in one eye, Stickler syndrome in one eye, Retinitis pigmentosa (RP) with Coats-like disease in one eye, and nanophthalmia in one eye. At last follow-up 11 out of 42 (26.2%) eyes had persistent retinal detachment. Two eyes with Terson's syndrome had inoperable RD, one eye with persistent fetal vasculature had inoperable TRD, one eye with FEVR had recurrent RD, one eye with FRAM had inoperable TRD, two eyes with neonatal meningitis had inoperable RD, one eye with Coats disease had inoperable RD, one eye with Marfan syndrome had recurrent RD, one eye with nanophthalmos had inoperable RD, and one eye with  The mean follow-up for accidental traumatic VH cases was 23.8 ± 18.9 months, the median was 20 months, and the range was 1-75 months. Vitrectomy was done for 104 out of 187 (55.6%) eyes, and evisceration/enucleation was done for 3 out of 187 (1.6%) eyes. Observation was the modality of treatment for 64 out of 187 (34.2%) eyes. Prophylactic laser photocoagulation was performed in 6 out of 187 (3.2%) eyes that presented with retinal break without RD (Figure 2). Student's test for comparing two percentages from the same sample). Univariate analysis was performed to investigate factors predicting the visual outcome. It demonstrated a significant association between final visual acuity of 20/200 or better and the following factors: initial visual acuity of 20/200 or better ( < 0.001), mild to moderate degree of VH at presentation (partial view to the retina) ( = 0.0057), and presence of clear lens or pseudophakia at last follow-up ( < 0.001) ( Table 8).

Visual Outcome of Traumatic Vitreous
Univariate analysis was performed to identify factors associated with poor final visual outcome (hand motions or less). Poor final visual outcome was significantly associated with poor initial visual acuity (hand motions or less) ( < 0.001), age at presentation of 3 years or younger ( = 0.0033), severe vitreous hemorrhage at presentation (no view to the fundus) ( = 0.0039), and presence of cataract or aphakia at last follow-up ( = 0.0012) ( Table 8).
Multivariate analysis based on stepwise logistic regression analysis was conducted to identify the variables that tended to influence the attainment of final visual acuity of 20/200 or better after adjustment for all the other factors. In this analysis, all the variables that were investigated for association with final visual acuity in univariate analysis were included as the predictor or independent variables. Visual acuity of 20/200 or better at presentation was significantly positively associated with the attainment of final visual acuity of 20/200 or better at last follow-up (odds ratio = 9.08; 95% confidence interval [CI] = 2.46-33.5). On the other hand, presence of retinal detachment at last follow-up was hindrance to achieving final visual acuity of 20/200 or better (odds ratio = 0.24; 95% CI = 0.09-0.68).    (Table 10). Multivariate analysis to identify the variables that tended to influence the anatomical outcome could not be done due to small number of the affected eyes.

Comparison between Nontraumatic Vitreous Hemorrhage and Vitreous Hemorrhage Caused by Accidental Trauma.
The mean age of accidental traumatic vitreous hemorrhage cases was significantly higher than in nontraumatic cases (7.7 years versus 4.9 years, resp.; = 0.0004). Males were the most affected in both traumatic 134 out of 187 (71.7%) patients and 24 nontraumatic cases out of 42 (57.1%) patients. The rate of performance of vitrectomy was significantly higher in eyes with accidental trauma (55.6%) than in nontraumatic (28.6%) ( = 0.0027). The rate of enucleation/evisceration/exenteration was significantly higher in nontraumatic eyes than in eyes with accidental trauma (11.9% versus 1.6%; = 0.0067). The incidence of retinal detachment at last follow-up was significantly higher in nontraumatic (42.9%) than in eyes with accidental trauma (2.1%) ( = 0.0159) ( Table 11).

Discussion
In this retrospective study of pediatric vitreous hemorrhage that presented to our institute, we analyzed 240 eyes of 230 children diagnosed with vitreous hemorrhage. The causes of vitreous hemorrhage were classified as traumatic and nontraumatic causes. Trauma was the most common cause of vitreous hemorrhage in children accounting for 82.5%. Nontraumatic cases accounted for 17.5%. These findings are in agreement with previous studies by Spirn et al. [28] and Rishi et al. [27] who demonstrated that trauma was the most common cause of vitreous hemorrhage in pediatric age group (73.1% and 68.5%, resp.).
In our study, the etiology on nontraumatic vitreous hemorrhage was in agreement with other reported series [27,28]. However, in the present study, retinoblastoma was more frequent compared to other studies. Eyes with retinoblastoma had the worst prognosis compared with other nontraumatic causes. Previous studies showed that vitreous hemorrhage in eyes with retinoblastoma is rare and its presence is a predictor of poor prognosis [41,42]. Shields et al. [43] recommend that the clinician should seriously consider the possibility of retinoblastoma in children who present with signs of unexplained vitreous hemorrhage or endophthalmitis, even if they are older than 5 years of age.
The novel finding in our study is that nontraumatic vitreous hemorrhage occurred in younger age group and had poor prognosis in the form of high incidence of retinal detachment and enucleation/exenteration as compared with traumatic cases. This can be explained by the fact that many of nontraumatic causes like hereditary diseases or retinoblastoma tend to occur in younger age group, which are usually associated with retinal detachment, while trauma tends to occur in a physically active child.
In the present study, in traumatic cases, the most common mechanism of trauma was trauma by a sharp object followed by blunt trauma. Our findings are in contrast to other studies that reported that blunt trauma was the most common  mechanism of traumatic vitreous hemorrhage in pediatric age group [27,28]. Vitreous hemorrhage due to shaken-baby syndrome was reported in previous studies with an incidence of 8.6% [28]. However, in the present retrospective study we did not encounter any case of shaken-baby syndrome. This can be attributed to the fact that our hospital is a specialized eye hospital and patients who are referred to us should have stable medical condition. It is known that babies with shaken-baby syndrome usually presented with unstable medical condition [19]. In this study, univariate analysis of cases that presented after accidental trauma demonstrated that the initial visual acuity of 20/200 or better was a significant important prognostic indicator of final visual outcome of 20/200 or more that remained a significant predictor by using logistic regression analysis. On the other hand, we found that worse initial visual acuity of hand motions or less was associated with a poor visual outcome (hand motions or less). Our observation is consistent with previous studies that included children alone [36] or both children and adults [29,33,[44][45][46][47]. In addition, our study showed a significant association between younger age at presentation (3 years or less) and poor visual outcome and this finding was similar to other published studies [36,48]. Amblyopia is expected to be the cause of poor vision in this younger age group.
Our study showed that density of vitreous hemorrhage at presentation significantly affects the visual outcome. Eyes with dense vitreous hemorrhage with no view to the fundus had a poor visual outcome. On the other hand, eyes with mild to moderate vitreous hemorrhage (partial view to the retina) had a significant association with visual outcome of 20/200 or better. This is similar to Matthews and Das finding in infants with shaken-baby syndrome who developed vitreous hemorrhage [19]. Our analysis demonstrated that retinal status significantly affects the visual outcome. In addition, we found that lens status at last follow-up significantly affected the visual outcome. Presence of clear lens or pseudophakia was associated with visual outcome of 20/200 or more and the presence of aphakia or cataract was associated with poor visual outcome. This finding is consistent with previous reports [29,33,36].
In this study, we also investigated factors that predict final anatomical outcome. Our analysis showed that the development of retinal detachment in eyes with vitreous hemorrhage caused by accidental trauma was significantly associated with younger age at presentation, open globe injury, performance of vitrectomy, and aphakia at last follow-up. Our findings are consistent with other reports that demonstrated that the risk of retinal detachment is high in aphakic eyes [49] and eyes that had vitrectomy [29].
A major limitation of our study is its retrospective nature. Therefore, in most of the patients included in this study, factors related to amblyopia were not investigated. The visual outcome in children with vitreous hemorrhage especially those aged 8 years or less is highly dependent on the duration of vitreous hemorrhage, early initiation of amblyopia treatment, clarity of visual axis, and also the family compliance with amblyopia treatment and wearing of glasses.
In conclusion, etiology of vitreous hemorrhage in children is unique. Trauma is the most common cause of vitreous hemorrhage in pediatric age group. In this group, initial visual acuity was the most important predictor for visual outcome, and the presence of retinal detachment at the final followup was a negative predictor for final good visual outcome. Patients with nontraumatic vitreous hemorrhage were significantly younger and the visual outcome was significantly worse in comparison to patients with traumatic vitreous hemorrhage.

Conflict of Interests
There are no conflicting interests for the authors or financial support.