- Email: [email protected]
Secondary Hemorrhage in Traumatic Hyphema
Secondary Hemorrhage in Traumatic Hyphema Predictive Factors for Selective Prophylaxis Lye Pheng Fong, MD Background: Treatment to decrease the rebleeding rate in traumatic hyphema remains controversial. Although antifibrinolytics recently have been shown to reduce secondary hemorrhage rates, their routine use has not been widely applied because of adverse side effects and the relatively low frequency of severe hyphema complications. Alternatively, their use may be restricted to patients at high risk, but prognostic factors for rebleeding have not been clearly identified. Methods: From a prospective ocular trauma survey, 371 patients with traumatic hyphema were identified, and Fisher's exact test was applied to test for significant differences between patients who did and did not rebleed for various characteristics. Significant factors contributing to rebleeding were fitted into a multiple logistic regression model, and odds ratios (OR) and 95% confidence intervals (95% Cl) were calculated. Results: Secondary hemorrhage occurred in 8% of patients and was significantly more frequent in those with visual acuities of 20/200 or less (OR = 3.1; 95% Cl = 1.3,7.5), initial hyphema more than one third of the anterior chamber (OR = 2.8; 95% Cl = 0.9,8.0), delayed medical attention more than 1 day after injury (OR = 2.9; 95% Cl = 1.0,8.4), and elevated intraocular pressure at time of first examination (OR = 2.9; 95% Cl = 1.1 ,7. 9). The secondary hemorrhage rate rose from 5% without any of these specified factors to 15% with at least one factor present. No statistical associations were found for age, injury-related iris abnormalities, or aspirin usage. Conclusion: Using multivariate logistic regression in populations with low rates of secondary hemorrhage, a predictive model may be used to categorize patients who have higher rebleeding rates, for whom possible benefits may outweigh the risks of prophylactic treatment, and those with lower rebleeding rates, who may not necessarily benefit from treatment. Ophthalmology 1994;101:1583-1588
The medical management of traumatic hyphema to reduce secondary hemorrhage has been controversial. Originally received: October 18, 1993. Revision accepted: March 15, 1994. From the Department of Ophthalmology, University of Melbourne, Melbourne, Australia. Presented in part at the 1992 Association for Research in Vision and Ophthalmology Annual Meeting and at the American Academy of Ophthalmology Annual Meeting, Chicago, November 1993. Supported by the Victorian Health Promotion Foundation Grant 90/ 0309, Melbourne, Victoria. Reprint requests to Lye Pheng Fong, MD, Department of Ophthalmology, University of Melbourne, Royal Victorian Eye and Ear Hospital, 32 Gisborne St, East Melbourne, Victoria, Australia 3002.
Treatment modalities that were shown not to influence the rebleeding rate included hospitalization, bed rest and sedation, unilateral or bilateral eye patching, mydriatic or miotic eyedrops, estrogens, and topical steroids. 1- 3 More recent studies have shown that antifibrinolytics and oral prednisolone potentially may reduce rebleed rates. 3- 5 However, the routine use ofantifibrinolytics and oral corticosteroids has not been accepted universally because of adverse drug side effects, inconclusive and conflicting evidence from several small studies, varying secondary hemorrhage rates, and the low frequency and severity of complications of rebleeding. 6 Recently, two North American studies have reported low rebleed rates of 5% to 7% and questioned the value of routine antifibrinolytics in these predominantly white populations, be-
1583
Ophthalmology
Volume 101, Number 9, September 1994
cause the risks and costs may outweigh the possible benefits in selected patient groups that have low rates of secondary hemorrhages. 6 · 7 An alternative approach would be to restrict their use to those patients at high risk of rebleeding. However, the prognostic factors for those at high risk have not been identified clearly. This study examined possible clinical parameters associated with secondary hemorrhage in traumatic hyphema. Emphasis was placed on those characteristics that could be recognized at the time of initial examination, which would allow early timing for clinical decisions on prophylaxis.
Patients and Methods From a prospective survey of patients with ocular trauma who were treated at the Emergency Department, Royal Victorian Eye and Ear Hospital in Melbourne, consecutive hyphema occurring after primary blunt ocular trauma was identified. Macroscopic hyphema was defined as bleeding in the anterior chamber of the eye, which may manifest as a layer of blood or clot. Patients with hyphemata due to penetrating injuries, postoperative complications, or secondary to iris neovascularization or xanthogranuloma were excluded. The following macroscopic hyphema levels were categorized: grade I, blood occupying less than one third of the anterior chamber; grade 2, blood occupying one third to one half of the anterior chamber; grade 3, blood occupying one half to three quarters of the anterior chamber; and grade 4, blood occupying three quarters to the total anterior chamber. Secondary hemorrhage was described as a definite increase in the amount of blood in the anterior chamber after presentation. At the time of presentation, each patient was invited to complete a confidential and comprehensive questionnaire that included questions about the time, place, and activity of injury; mechanisms and factors leading to the injury; usual occupation; and type of eye protection and the usual frequency of use, if relevant. Demographic and clinical information was collected by doctors on pre-coded forms detailing the age, sex, nature of injury, visual acuity, hyphema grade, associated ocular and adnexal injury, aspirin usage 2 weeks before medical attendance, management on an outpatient or inpatient basis, treatment, duration of hospitalization, complications, and surgical procedures. Additional information was obtained on follow-up forms for patients who were hospitalized. These forms were completed at 6and 12-month intervals to obtain information on visual outcome, complications, and details of any further surgery on patients examined in the outpatient department. For patients discharged to private care outside the hospital, the follow-up form was submitted to the patient's attending ophthalmologist for completion. The decision of hospital admission was made by the attending ophthalmologist; the general routine was to admit most patients with macroscopic hyphema. For patients who were hospitalized, standard treatment included
1584
bed rest with limited ambulation, unilateral or bilateral pad occlusion, avoidance of aspirin-containing products, and use of antiglaucoma medications as indicated. Topical cycloplegic and steroid were used where determined by the admitting clinician, and dosage was recorded for later analysis. Associated factors analyzed for secondary hemorrhage were age (children versus adults), initial hyphema level at presentation, visual acuity at presentation, duration of days after injury at presentation, elevated intraocular pressure (> 21 mmHg), aspirin usage 2 weeks before presentation, and iris abnormalities sustained as a result of injury. Such iris damage included traumatic mydriasis, sphincter rupture, and iris dialysis. Statistical analysis comparing the characteristics between those patients who did and those who did not rebleed was conducted using Fisher's exact test, univariate and multivariate logistic regression modeling on "Egret" (SERC, Seattle, W A). Odds ratios (OR) were calculated, with 95% confidence intervals (95% Cl).
Results A total of 375 patients (379 eyes) with traumatic macroscopic hyphema were identified during a 19-month period between November 1989 and May 1991. Only those hyphema injuries occurring after primary, blunt trauma were included. High-capture (95%), compliance (94%), and completion (90%) rates were obtained for patient questionnaires and medical information; incomplete information was supplemented by telephone interviews and reviews of hospital charts. There was a male predominance with a male:female ratio of 6: I. The majority of patients were young (mean age, 27 ± 17 years; range, 2 months-83 years) (Fig 1). There were four bilateral hyphemata. One hundred fiftythree injuries (41%) were caused by sporting activities. Thirty-two percent of all hyphemata were associated with other ocular injuries, occurring at an average of 3.3 diagnoses, other than hyphema, per patient with multiple
Number of patients
140 .---------~-------------------------,
•Female
120
D Male
100 80 60 40 20 O
LL~~~~~~~~
0-9
10-19
20-29
G 30-39
40-49
Age in years Figure L Macroscopic hyphema by age and sex.
50-59
I
Fong · Secondary Hemorrhage in Traumatic Hyphema Table 2. Initial Hyphema Level
Table 1. Macroscopic H yphema Injuries Associated with Hyphema Diagnosis Traumatic mydriasis Glaucoma Angle recession Iris Laceration Deformity Dialysis
19 13 10
Corneal abrasion Lid laceration/contusion Vitreous/retinal hemorrhage Retinal edema Retinal tear/detachment Choroidal rupture
12 8
6 8 6
7 5 2 2
* Patients may have multiple injuries.
injuries (Table 1). Eighteen patients underwent surgery, including anterior chamber paracentesis and clot removal (n = 5), repair of eyelid (n = 6) and lacrimal system (n = 2), retinal detachment prophylaxis (n = I), synechialysis (n = I), urgent canthotomy (n = 1), retinal membrane peeling (n = 1), and examination under anesthesia for an inoperable retinal detachment and associated cataract (n = 1).
Inpatients (n = 285)
%of Patients*
Of the 375 patients with macroscopic hyphemata, 90% were grade I, 6% were grade 2, 2% were grade 3, and 2% were grade 4 on initial examination. Seventy-six percent of patients with macroscopic hyphema were admitted to the hospital; only 24% with grade I were managed as outpatients, and patients with higher grades were admitted to the hospital (Table 2). Statistical analysis was based on 371 patients, omitting, by necessity, four children (8 eyes) who had missing visual acuity values. At presentation, visual acuity of 20/200 or less was associated with a significantly increased risk of rebleed when compared with visual acuity of 20/ I 00 or greater (OR= 3.1 ; 95% CI = 1.3,7.5) (Fig 2 and Table 3). The secondary rebleeding rates were 8% overall (95% CI = 5.4, ll.O), 7% for grade I, 18% for grade 2, and 29% for grade 4 hyphema. No secondary hemorrhage occurred in grade 3 (n = 6) or outpatient grade I (n = 90) hyphema (Fig 3). Hyphema with a grade 2 rating or greater had a marginally significant risk of rebleed when compared with grade I macroscopic hyphema (OR = 2.8; 95% CI = 0.9,8.0) (Table 3). The majority of patients were examined within 24 hours of injury (Fig 4); their rebleeding rate was 7%. This was significantly different from those presenting later than 24 hours after injury, who had a rebleeding rate of 18% (OR = 2.9; 95% CI = 1.0,8.4) (Table 3). Forty patients had intraocular pressures greater than 21 mmHg recorded at initial presentation. Secondary hemorrhages occurred in 8 (20%) of these 40 patients. By comparison, only 7%
Outpatients (n = 90)
Grade
No.
(o/o)
No.
(o/o)
1
249
(88)
90
(100)
2
23
(8)
3
6
(2)
4
7
(2)
of those patients with normal intraocular pressures rebled· this difference was statistically significant (OR = 2.9, 95o/; CI = 1.1 ,7.9) (Table 3). All four variables were included for the multivariate logistic regression model (Table 3). The odds ratios for delayed presentation and elevated intraocular pressure did not alter after adjusting for the other variables, indicating that these two factors were independent predictors. However, the adjusted odds ratios for visual acuity and initial hyphema level were lower, suggesting that these two were partially confounding variables. While the adjusted odds ratio for visual acuity remained significant, the borderline univariate result for initial hyphema grading was no longer significant on multivariate testing. Neither prior aspirin usage (P = 0.27) nor injury-induced iris abnormalities (P = 0.51 ), either as single or composite elements of iris damage, were associated with increased rates of secondary hemorrhage (Table 4). The lack of statistical association of prior aspirin usage to secondary hemorrhage was not surprising, in view of the low prevalence of aspirin ingestion (2.6%). There was no significant difference in the rebleed rates for children when compared to adults (P = 0.19}. One patient (0.2%) was g!ven systemic steroids, and another patient (0.2%) was gtven oral tranexamic acid, only after rebleeding had already occurred.
•Rebleed D No Rebleed
60 40
20
0
20/60-20/100
<= 20/200
Visual acuity Figure 2. Secondary hemorrhage and visual acuity at presentation.
1585
Ophthalmology
Volume 101, Number 9, September 1994
Table 3. Hyphema Secondary Hemorrhage* Multivariate Logistic Regression
Univariate Variable Level
Odds Ratio (95% Cl)
p
Odds Ratio (95% CI)
p
3.1 (1.3-7.5)
0.006
2.4 (1.0-5.8)
0.05
2.8 (0.9-8.0)
0.04
1.9 (0.6-5.7)
0.2
2.9 (1.0-8.4)
0.04
2.7 (1.0-7.5)
0.06
2.9 (1.1-7.9)
0.03
2.9 (1.1-7.6)
0.03
Visual acuity .=:;;20/100t ;:::20/200 Grade 1t :<::.2 Delay 1 dayt >1 day Intraocular pressure (mmHg) <21t ::::21 Cl
= confidence intervals. = 14.4.
• Residual deviance
t
Reference category.
The decision to use topical mydriatics and corticosteroids was made by the treating clinician. There was no statistical difference in the use of cycloplegic medication between patients who rebled and those who did not. The use of topical steroids was associated with a statistical difference for secondary hemorrhage (P = 0.0001) but this result was subject to reverse causality bias due to ( l) selection of patients with more severe hyphema for treatment and (2) initiation of steroids in some patients after rebleeding had occurred.
Discussion This series recorded an overall rebleeding rate of 8.2% (95% CI = 5.4%, ll%) (Table 5) and identified, at the time
of initial examination, four clinical markers that were predictive of secondary hemorrhage. These included presenting visual acuity of 20/200 or less, initial hyphema level of grade 2 or greater macroscopic hyphema (> 113 of the anterior chamber), delayed medical examination more than 24 hours after injury, and elevated intraocular pressure above 21 mmHg. The low rate of secondary hemorrhage in primary traumatic hyphema reported in this study (8%) agreed with rebleed rates of less than 10% previously observed in predominantly white populations in Australia, Great Britain, Canada, and Northern Europe. 6 Within the United States, rates of secondary hemorrhage have varied from as low as 5% 6 •7 to a markedly higher rate of 33%.8- 10 Higher rebleed rates have been reported in the black population. 11 •12 Although sickle cell hemoglobin-
Percent of patients
100 .---------~-------------------------,
•Rebleed
80
D No Rebleed
60 40
20
Grade 1
Grade 2
Grade 3
Grade 4
Hyphema Grade Figure 3. Secondary hemorrhage and initial hyphema level.
1586
0
<=1
2
3
4
>= 5
Days after injury Figure 4. Rebleed by days after injury for macroscopic hyphema.
Fong • Secondary Hemorrhage in Traumatic Hyphema
Table 4. Macroscopic Hyphema Factors Not Associated with Secondary Hemorrhage
Aspirin No aspirin Iris abnormalities No iris abnormalities Age< 15 yrs Age 2= 15 yrs NS
=
No.
Rebleed (%)
p
10 363 100 215 100 277
20 8 9 10 11 7
NS NS NS
not significant.
apathy trait or disease may be contributing factors, the apparent ethnic difference also may be confounded by the higher representation of the black population in lower socioeconomic classes, with higher rates and severity of injury and delayed access to medical consultation. 11 Apart from racial differences and their implications, other factors that have been related to increased rebleeding included the size of initial hemorrhage, 2 •7•13 delayed examination after injury,6 •7 and aspirin usage. 13 In the current study, the lack of association of prior aspirin ingestion to secondary hemorrhage was not surprising, in view of the low prevalence of aspirin usage (2.6% ). None of the patients received aspirin during or after hospitalization because of its recognized effect on secondary hemorrhage rates. Other studies failed to detect a statistical difference with age, sickle cell trait, and aspirin usage. 1•3• 14- 17 The failure to consistently report or identify contributing factors may be due to varying patient profiles and small patient numbers, with inferences for inadequate sample size and power calculations. To the best of our knowledge, this study of 371 eyes with macroscopic hyphema represents the largest reported series of hyphema. The associated factors for secondary hemorrhage specified in the current study can be used to categorize patient subsets into different risk levels for secondary hemorrhage (Table 5). In the absence of any of the other specified factors, grade 1 macroscopic hyphema was associated with a rebleeding rate of 4. 7% and comprised two thirds of all patients with macroscopic hyphema. The presence of any one or multiple risk factors significantly increased the rate of secondary hemorrhage to 14.5% (OR= 3.3; 95% CI = 1.6,7.2). The additive risks of multiple factors to rebleeding are depicted in a probability distribution model based on the logistic regression results (Table 6). In the baseline minimal risk situation of grade 1 hyphema, visual acuity of 20/100 or greater, normal intraocular pressure, and presenting within 24 hours, the estimated percentage of a secondary hemorrhage is 5%, which is increased twofold with one added factor of either visual acuity of 20/200 or less or elevated intraocular pressure. For more serious combinations of risk categories, this percentage ranges up to 64% in the presence of all four risk categories. The following concerns have been raised about the frequent side effects of aminocaproic acid: (1) systemic-
nausea and vomiting, syncope, hypotension, and the risk ofthromboembolic events, 18 and (2) ocular-delayed clot resolution and secondary elevations in intraocular pressure with clot dissolution. 19 Tranexamic acid, which has less gastrointestinal side effects and is more potent than aminocaproic acid, 5 has been associated with isolated case reports of thromboembolic phenomena, including two patients with central venous stasis retinopathy20 and retinal degeneration in experimental animals. 21 In the United States, aminocaproic acid, but not tranexamic acid, has been approved by the Food and Drug Administration for use in the treatment ofhyphema. Although adverse effects of oral prednisolone were not recorded in two small studies on hyphema, the Boston Collaborative Drug Surveillance reported an 11.4% incidence of acute adverse events to prednisolone. 4•13 Most studies that examined the role of antifibrinolytics were small, uncontrolled, or nonrandomized trials. Meta-analysis of six randomized clinical trials yielded a combined odds ratio for rebleeding of 0.17 (95% CI = 0.04,0.78), which suggested a protective role of antifibrinolytics against rebleeding (Elman MJ, unpublished data; presented at the 1990 AR VO Annual Meeting). However, the author cautioned that the study sample numbers were small and may not have followed a normal distribution. Most antifibrinolytic studies were conducted in populations with high rebleeding rates (range, 23%-33%) and a high proportion of black patients (range, 15%-69%). 4 Aminocaproic acid reduced the rebleed rate of 33% in placebo-treated patients to 3%, 5% and 0%, respectively. 8 - 10 More recently, Volpe et al6 reported on the lack of effectiveness of aminocaproic acid in influencing the secondary hemorrhage rate for hyphema in a predominantly white population. In their study, Volpe et al observed a secondary hemorrhage rate of 4.8% in aminocaproic acidtreated patients compared with a 5.4% rate for those not receiving treatment; these figures were quoted for those seen within 24 hours of injury. Their study suggested that the benefit of aminocaproic acid was questionable for predominantly white populations with inherently low rebleeding rates and that its routine use may not be justified for all patients with hyphema, particularly when side effects and cost effectiveness also were considered. However,
Table
5.
Risk Levels for Secondary Hemorrhage Rebleeding Rates
Total of predominantly white population Low risk, no associated factors, 64% of all macroscopic hyphema High risk, 2= 1 associated factors
8.2 (5.4-11.0)*
4.7t 14.5t
• Values in parentheses are 95% confidence intervals. t p = 0.001.
1587
Ophthalmology
Volume 101, Number 9, September 1994
Table 6. Probability of Rebleed Based on Logistic Regression Model Visual Acuity Elevated Intraocular Pressure
Grade 1
s;l day No Yes
12
>1 day No Yes
27
11
References I. Agapitos PJ, Noel L-P, Clarke WN. Traumatic hyphema in
children. Ophthalmology 1987;94: 1238-41. 2. McClellan KA, Fong F, Billson FA. Traumatic hyphema. Outcome and management. Aust Fam Physician 1989; 18: 538-46. 3. Gottsch JD. Hyphema: diagnosis and management. Retina 1990; 10 (Suppi):S65-71. 4. Farber MD, Fiscella R, Goldberg MF. Aminocaproic acid versus prednisone for the treatment of traumatic hyphema. A randomized clinical trial. Ophthalmology 1991 ;98:27986. 5. Deans R, Noel L-P, Clarke WN. Oral administration of tranexamic acid in the management of traumatic hyphema in children. Can J Ophthalmol 1992;27: 181-3. 6. Volpe NJ, Larrison WI, Hersh PS, Kim T, et al. Secondary hemorrhage in traumatic hyphema. Am J Ophthalmol 1991;112:507-13. 7. Kennedy RH, Brubaker RF. Traumatic hyphema in a defined population. Am J Ophthalmol 1988; 106: 123-30. 8. Crouch ER Jr, Frenkel M. Aminocaproic acid in the treatment of traumatic hyphema. Am J Ophthalmol 1976;81: 355-60.
(%)
Grade
5
their patients who presented more than l day after injury had a significantly elevated rate of secondary hemorrhage (38%), and Volpe et al 6 suggested that distinct patient subsets (black patients and patients with delayed examination) may potentially benefit from aminocaproic acid treatment. The current study has provided guidelines for developing future clinical trials of measures to prevent secondary hemorrhage. The identification of four predictive factors for rebleeding allows categorization of patients with hyphema into subset groups for specific prophylactic treatment at the time of initial examination. Routine hospitalization previously has been shown to have little effect on the secondary hemorrhage rates for patients with uncomplicated hyphema. 22 •23 Hospital cost containment may be effected by selectively admitting and treating those patients whose presenting clinical characteristics have identified them as being at higher risk for secondary hemorrhage.
1588
~20/100
~2
Visual Acuity s;20/200 (%)
Grade 1
Grade
8 21
26
40
20 42
24 48
40 64
11
~2
18
9. McGetrick JJ, Jampol LM, Goldberg MF, et al. Aminocaproic acid decreases secondary hemorrhage after traumatic hyphema. Arch Ophthalmol 1983; 101: 1031-3. 10. Kutner B, Fourman S, Brein K, et al. Aminocaproic acid reduces the risk of secondary hemorrhage in patients with traumatic hyphema. Arch Ophthalmol 1987;105:206-8. 11. Palmer DJ, Goldberg MF, Frenkel M, et al. A comparison of two dose regimens of epsilon aminocaproic acid in the prevention and management of secondary traumatic hyphemas. Ophthalmology 1986;93: 102-8. 12. Skalka HW. Recurrent hemorrhage in traumatic hyphema. Ann Ophthalmol 1978;10:1153-7. 13. Spoor TC, Kwitko GM, O'Grady JM, Ramocki JM. Traumatic hyphema in an urban population. Am J Ophthalmol 1990; 109:23-7. 14. Ganley JP, Geiger JM, Clement JR, et al. Aspirin andrecurrent hyphema after blunt ocular trauma. Am J Ophthalmol 1983;96:797-80 I. 15. Marcus M, Biedner B, Lifshitz T, Yassur Y. Aspirin and secondary bleeding after traumatic hyphema. Ann Ophthalmol 1988;20:157-8. 16. Cassel GH, Jeffers JB, Jaeger EA. Wills Eye Hospital traumatic hyphema study. Ophthalmic Surg 1985;16:441-3. 17. Witteman GJ, Brubaker SJ, Johnson M, Marks RG. The incidence of rebleeding in traumatic hyphema. Ann Ophthalmol 1985;17:525-9. 18. Uusitalo RJ, Ranta-Kemppainen L, Tarkkanen A. Management of traumatic hyphema in children. An analysis of 340 cases. Arch Ophthalmol 1988;106:1207-9. 19. Dieste MC, Hersh PS, Kylstra JA, et al. Intraocular pressure increase associated with epsilon-aminocaproic acid therapy for traumatic hyphema. Am J Ophthalmol 1988;106:38390. 20. Snir M, Axer-Siegel R, Buckman G, Yassur Y. Central venous stasis retinopathy following the use oftranexamic acid. Retina 1990;10:181-4. 21. Mindel JS. Problems in the use of tranexamic acid by ophthalmologists [letter]. Arch Ophthalmol 1989; 107:486-
7.
22. Spaeth GL, Levy PM. Traumatic hyphema: its clinical characteristics and failure of estrogens to alter its course. A double-blind study. Am J Ophthalmol 1966;62: 10981106. 23. Wilson FM II. Traumatic hyphema. Pathogenesis and management. Ophthalmology 1980;87 :910-19.
The medical management of traumatic hyphema to reduce secondary hemorrhage has been controversial. Originally received: October 18, 1993. Revision accepted: March 15, 1994. From the Department of Ophthalmology, University of Melbourne, Melbourne, Australia. Presented in part at the 1992 Association for Research in Vision and Ophthalmology Annual Meeting and at the American Academy of Ophthalmology Annual Meeting, Chicago, November 1993. Supported by the Victorian Health Promotion Foundation Grant 90/ 0309, Melbourne, Victoria. Reprint requests to Lye Pheng Fong, MD, Department of Ophthalmology, University of Melbourne, Royal Victorian Eye and Ear Hospital, 32 Gisborne St, East Melbourne, Victoria, Australia 3002.
Treatment modalities that were shown not to influence the rebleeding rate included hospitalization, bed rest and sedation, unilateral or bilateral eye patching, mydriatic or miotic eyedrops, estrogens, and topical steroids. 1- 3 More recent studies have shown that antifibrinolytics and oral prednisolone potentially may reduce rebleed rates. 3- 5 However, the routine use ofantifibrinolytics and oral corticosteroids has not been accepted universally because of adverse drug side effects, inconclusive and conflicting evidence from several small studies, varying secondary hemorrhage rates, and the low frequency and severity of complications of rebleeding. 6 Recently, two North American studies have reported low rebleed rates of 5% to 7% and questioned the value of routine antifibrinolytics in these predominantly white populations, be-
1583
Ophthalmology
Volume 101, Number 9, September 1994
cause the risks and costs may outweigh the possible benefits in selected patient groups that have low rates of secondary hemorrhages. 6 · 7 An alternative approach would be to restrict their use to those patients at high risk of rebleeding. However, the prognostic factors for those at high risk have not been identified clearly. This study examined possible clinical parameters associated with secondary hemorrhage in traumatic hyphema. Emphasis was placed on those characteristics that could be recognized at the time of initial examination, which would allow early timing for clinical decisions on prophylaxis.
Patients and Methods From a prospective survey of patients with ocular trauma who were treated at the Emergency Department, Royal Victorian Eye and Ear Hospital in Melbourne, consecutive hyphema occurring after primary blunt ocular trauma was identified. Macroscopic hyphema was defined as bleeding in the anterior chamber of the eye, which may manifest as a layer of blood or clot. Patients with hyphemata due to penetrating injuries, postoperative complications, or secondary to iris neovascularization or xanthogranuloma were excluded. The following macroscopic hyphema levels were categorized: grade I, blood occupying less than one third of the anterior chamber; grade 2, blood occupying one third to one half of the anterior chamber; grade 3, blood occupying one half to three quarters of the anterior chamber; and grade 4, blood occupying three quarters to the total anterior chamber. Secondary hemorrhage was described as a definite increase in the amount of blood in the anterior chamber after presentation. At the time of presentation, each patient was invited to complete a confidential and comprehensive questionnaire that included questions about the time, place, and activity of injury; mechanisms and factors leading to the injury; usual occupation; and type of eye protection and the usual frequency of use, if relevant. Demographic and clinical information was collected by doctors on pre-coded forms detailing the age, sex, nature of injury, visual acuity, hyphema grade, associated ocular and adnexal injury, aspirin usage 2 weeks before medical attendance, management on an outpatient or inpatient basis, treatment, duration of hospitalization, complications, and surgical procedures. Additional information was obtained on follow-up forms for patients who were hospitalized. These forms were completed at 6and 12-month intervals to obtain information on visual outcome, complications, and details of any further surgery on patients examined in the outpatient department. For patients discharged to private care outside the hospital, the follow-up form was submitted to the patient's attending ophthalmologist for completion. The decision of hospital admission was made by the attending ophthalmologist; the general routine was to admit most patients with macroscopic hyphema. For patients who were hospitalized, standard treatment included
1584
bed rest with limited ambulation, unilateral or bilateral pad occlusion, avoidance of aspirin-containing products, and use of antiglaucoma medications as indicated. Topical cycloplegic and steroid were used where determined by the admitting clinician, and dosage was recorded for later analysis. Associated factors analyzed for secondary hemorrhage were age (children versus adults), initial hyphema level at presentation, visual acuity at presentation, duration of days after injury at presentation, elevated intraocular pressure (> 21 mmHg), aspirin usage 2 weeks before presentation, and iris abnormalities sustained as a result of injury. Such iris damage included traumatic mydriasis, sphincter rupture, and iris dialysis. Statistical analysis comparing the characteristics between those patients who did and those who did not rebleed was conducted using Fisher's exact test, univariate and multivariate logistic regression modeling on "Egret" (SERC, Seattle, W A). Odds ratios (OR) were calculated, with 95% confidence intervals (95% Cl).
Results A total of 375 patients (379 eyes) with traumatic macroscopic hyphema were identified during a 19-month period between November 1989 and May 1991. Only those hyphema injuries occurring after primary, blunt trauma were included. High-capture (95%), compliance (94%), and completion (90%) rates were obtained for patient questionnaires and medical information; incomplete information was supplemented by telephone interviews and reviews of hospital charts. There was a male predominance with a male:female ratio of 6: I. The majority of patients were young (mean age, 27 ± 17 years; range, 2 months-83 years) (Fig 1). There were four bilateral hyphemata. One hundred fiftythree injuries (41%) were caused by sporting activities. Thirty-two percent of all hyphemata were associated with other ocular injuries, occurring at an average of 3.3 diagnoses, other than hyphema, per patient with multiple
Number of patients
140 .---------~-------------------------,
•Female
120
D Male
100 80 60 40 20 O
LL~~~~~~~~
0-9
10-19
20-29
G 30-39
40-49
Age in years Figure L Macroscopic hyphema by age and sex.
50-59
I
Fong · Secondary Hemorrhage in Traumatic Hyphema Table 2. Initial Hyphema Level
Table 1. Macroscopic H yphema Injuries Associated with Hyphema Diagnosis Traumatic mydriasis Glaucoma Angle recession Iris Laceration Deformity Dialysis
19 13 10
Corneal abrasion Lid laceration/contusion Vitreous/retinal hemorrhage Retinal edema Retinal tear/detachment Choroidal rupture
12 8
6 8 6
7 5 2 2
* Patients may have multiple injuries.
injuries (Table 1). Eighteen patients underwent surgery, including anterior chamber paracentesis and clot removal (n = 5), repair of eyelid (n = 6) and lacrimal system (n = 2), retinal detachment prophylaxis (n = I), synechialysis (n = I), urgent canthotomy (n = 1), retinal membrane peeling (n = 1), and examination under anesthesia for an inoperable retinal detachment and associated cataract (n = 1).
Inpatients (n = 285)
%of Patients*
Of the 375 patients with macroscopic hyphemata, 90% were grade I, 6% were grade 2, 2% were grade 3, and 2% were grade 4 on initial examination. Seventy-six percent of patients with macroscopic hyphema were admitted to the hospital; only 24% with grade I were managed as outpatients, and patients with higher grades were admitted to the hospital (Table 2). Statistical analysis was based on 371 patients, omitting, by necessity, four children (8 eyes) who had missing visual acuity values. At presentation, visual acuity of 20/200 or less was associated with a significantly increased risk of rebleed when compared with visual acuity of 20/ I 00 or greater (OR= 3.1 ; 95% CI = 1.3,7.5) (Fig 2 and Table 3). The secondary rebleeding rates were 8% overall (95% CI = 5.4, ll.O), 7% for grade I, 18% for grade 2, and 29% for grade 4 hyphema. No secondary hemorrhage occurred in grade 3 (n = 6) or outpatient grade I (n = 90) hyphema (Fig 3). Hyphema with a grade 2 rating or greater had a marginally significant risk of rebleed when compared with grade I macroscopic hyphema (OR = 2.8; 95% CI = 0.9,8.0) (Table 3). The majority of patients were examined within 24 hours of injury (Fig 4); their rebleeding rate was 7%. This was significantly different from those presenting later than 24 hours after injury, who had a rebleeding rate of 18% (OR = 2.9; 95% CI = 1.0,8.4) (Table 3). Forty patients had intraocular pressures greater than 21 mmHg recorded at initial presentation. Secondary hemorrhages occurred in 8 (20%) of these 40 patients. By comparison, only 7%
Outpatients (n = 90)
Grade
No.
(o/o)
No.
(o/o)
1
249
(88)
90
(100)
2
23
(8)
3
6
(2)
4
7
(2)
of those patients with normal intraocular pressures rebled· this difference was statistically significant (OR = 2.9, 95o/; CI = 1.1 ,7.9) (Table 3). All four variables were included for the multivariate logistic regression model (Table 3). The odds ratios for delayed presentation and elevated intraocular pressure did not alter after adjusting for the other variables, indicating that these two factors were independent predictors. However, the adjusted odds ratios for visual acuity and initial hyphema level were lower, suggesting that these two were partially confounding variables. While the adjusted odds ratio for visual acuity remained significant, the borderline univariate result for initial hyphema grading was no longer significant on multivariate testing. Neither prior aspirin usage (P = 0.27) nor injury-induced iris abnormalities (P = 0.51 ), either as single or composite elements of iris damage, were associated with increased rates of secondary hemorrhage (Table 4). The lack of statistical association of prior aspirin usage to secondary hemorrhage was not surprising, in view of the low prevalence of aspirin ingestion (2.6%). There was no significant difference in the rebleed rates for children when compared to adults (P = 0.19}. One patient (0.2%) was g!ven systemic steroids, and another patient (0.2%) was gtven oral tranexamic acid, only after rebleeding had already occurred.
•Rebleed D No Rebleed
60 40
20
0
20/60-20/100
<= 20/200
Visual acuity Figure 2. Secondary hemorrhage and visual acuity at presentation.
1585
Ophthalmology
Volume 101, Number 9, September 1994
Table 3. Hyphema Secondary Hemorrhage* Multivariate Logistic Regression
Univariate Variable Level
Odds Ratio (95% Cl)
p
Odds Ratio (95% CI)
p
3.1 (1.3-7.5)
0.006
2.4 (1.0-5.8)
0.05
2.8 (0.9-8.0)
0.04
1.9 (0.6-5.7)
0.2
2.9 (1.0-8.4)
0.04
2.7 (1.0-7.5)
0.06
2.9 (1.1-7.9)
0.03
2.9 (1.1-7.6)
0.03
Visual acuity .=:;;20/100t ;:::20/200 Grade 1t :<::.2 Delay 1 dayt >1 day Intraocular pressure (mmHg) <21t ::::21 Cl
= confidence intervals. = 14.4.
• Residual deviance
t
Reference category.
The decision to use topical mydriatics and corticosteroids was made by the treating clinician. There was no statistical difference in the use of cycloplegic medication between patients who rebled and those who did not. The use of topical steroids was associated with a statistical difference for secondary hemorrhage (P = 0.0001) but this result was subject to reverse causality bias due to ( l) selection of patients with more severe hyphema for treatment and (2) initiation of steroids in some patients after rebleeding had occurred.
Discussion This series recorded an overall rebleeding rate of 8.2% (95% CI = 5.4%, ll%) (Table 5) and identified, at the time
of initial examination, four clinical markers that were predictive of secondary hemorrhage. These included presenting visual acuity of 20/200 or less, initial hyphema level of grade 2 or greater macroscopic hyphema (> 113 of the anterior chamber), delayed medical examination more than 24 hours after injury, and elevated intraocular pressure above 21 mmHg. The low rate of secondary hemorrhage in primary traumatic hyphema reported in this study (8%) agreed with rebleed rates of less than 10% previously observed in predominantly white populations in Australia, Great Britain, Canada, and Northern Europe. 6 Within the United States, rates of secondary hemorrhage have varied from as low as 5% 6 •7 to a markedly higher rate of 33%.8- 10 Higher rebleed rates have been reported in the black population. 11 •12 Although sickle cell hemoglobin-
Percent of patients
100 .---------~-------------------------,
•Rebleed
80
D No Rebleed
60 40
20
Grade 1
Grade 2
Grade 3
Grade 4
Hyphema Grade Figure 3. Secondary hemorrhage and initial hyphema level.
1586
0
<=1
2
3
4
>= 5
Days after injury Figure 4. Rebleed by days after injury for macroscopic hyphema.
Fong • Secondary Hemorrhage in Traumatic Hyphema
Table 4. Macroscopic Hyphema Factors Not Associated with Secondary Hemorrhage
Aspirin No aspirin Iris abnormalities No iris abnormalities Age< 15 yrs Age 2= 15 yrs NS
=
No.
Rebleed (%)
p
10 363 100 215 100 277
20 8 9 10 11 7
NS NS NS
not significant.
apathy trait or disease may be contributing factors, the apparent ethnic difference also may be confounded by the higher representation of the black population in lower socioeconomic classes, with higher rates and severity of injury and delayed access to medical consultation. 11 Apart from racial differences and their implications, other factors that have been related to increased rebleeding included the size of initial hemorrhage, 2 •7•13 delayed examination after injury,6 •7 and aspirin usage. 13 In the current study, the lack of association of prior aspirin ingestion to secondary hemorrhage was not surprising, in view of the low prevalence of aspirin usage (2.6% ). None of the patients received aspirin during or after hospitalization because of its recognized effect on secondary hemorrhage rates. Other studies failed to detect a statistical difference with age, sickle cell trait, and aspirin usage. 1•3• 14- 17 The failure to consistently report or identify contributing factors may be due to varying patient profiles and small patient numbers, with inferences for inadequate sample size and power calculations. To the best of our knowledge, this study of 371 eyes with macroscopic hyphema represents the largest reported series of hyphema. The associated factors for secondary hemorrhage specified in the current study can be used to categorize patient subsets into different risk levels for secondary hemorrhage (Table 5). In the absence of any of the other specified factors, grade 1 macroscopic hyphema was associated with a rebleeding rate of 4. 7% and comprised two thirds of all patients with macroscopic hyphema. The presence of any one or multiple risk factors significantly increased the rate of secondary hemorrhage to 14.5% (OR= 3.3; 95% CI = 1.6,7.2). The additive risks of multiple factors to rebleeding are depicted in a probability distribution model based on the logistic regression results (Table 6). In the baseline minimal risk situation of grade 1 hyphema, visual acuity of 20/100 or greater, normal intraocular pressure, and presenting within 24 hours, the estimated percentage of a secondary hemorrhage is 5%, which is increased twofold with one added factor of either visual acuity of 20/200 or less or elevated intraocular pressure. For more serious combinations of risk categories, this percentage ranges up to 64% in the presence of all four risk categories. The following concerns have been raised about the frequent side effects of aminocaproic acid: (1) systemic-
nausea and vomiting, syncope, hypotension, and the risk ofthromboembolic events, 18 and (2) ocular-delayed clot resolution and secondary elevations in intraocular pressure with clot dissolution. 19 Tranexamic acid, which has less gastrointestinal side effects and is more potent than aminocaproic acid, 5 has been associated with isolated case reports of thromboembolic phenomena, including two patients with central venous stasis retinopathy20 and retinal degeneration in experimental animals. 21 In the United States, aminocaproic acid, but not tranexamic acid, has been approved by the Food and Drug Administration for use in the treatment ofhyphema. Although adverse effects of oral prednisolone were not recorded in two small studies on hyphema, the Boston Collaborative Drug Surveillance reported an 11.4% incidence of acute adverse events to prednisolone. 4•13 Most studies that examined the role of antifibrinolytics were small, uncontrolled, or nonrandomized trials. Meta-analysis of six randomized clinical trials yielded a combined odds ratio for rebleeding of 0.17 (95% CI = 0.04,0.78), which suggested a protective role of antifibrinolytics against rebleeding (Elman MJ, unpublished data; presented at the 1990 AR VO Annual Meeting). However, the author cautioned that the study sample numbers were small and may not have followed a normal distribution. Most antifibrinolytic studies were conducted in populations with high rebleeding rates (range, 23%-33%) and a high proportion of black patients (range, 15%-69%). 4 Aminocaproic acid reduced the rebleed rate of 33% in placebo-treated patients to 3%, 5% and 0%, respectively. 8 - 10 More recently, Volpe et al6 reported on the lack of effectiveness of aminocaproic acid in influencing the secondary hemorrhage rate for hyphema in a predominantly white population. In their study, Volpe et al observed a secondary hemorrhage rate of 4.8% in aminocaproic acidtreated patients compared with a 5.4% rate for those not receiving treatment; these figures were quoted for those seen within 24 hours of injury. Their study suggested that the benefit of aminocaproic acid was questionable for predominantly white populations with inherently low rebleeding rates and that its routine use may not be justified for all patients with hyphema, particularly when side effects and cost effectiveness also were considered. However,
Table
5.
Risk Levels for Secondary Hemorrhage Rebleeding Rates
Total of predominantly white population Low risk, no associated factors, 64% of all macroscopic hyphema High risk, 2= 1 associated factors
8.2 (5.4-11.0)*
4.7t 14.5t
• Values in parentheses are 95% confidence intervals. t p = 0.001.
1587
Ophthalmology
Volume 101, Number 9, September 1994
Table 6. Probability of Rebleed Based on Logistic Regression Model Visual Acuity Elevated Intraocular Pressure
Grade 1
s;l day No Yes
12
>1 day No Yes
27
11
References I. Agapitos PJ, Noel L-P, Clarke WN. Traumatic hyphema in
children. Ophthalmology 1987;94: 1238-41. 2. McClellan KA, Fong F, Billson FA. Traumatic hyphema. Outcome and management. Aust Fam Physician 1989; 18: 538-46. 3. Gottsch JD. Hyphema: diagnosis and management. Retina 1990; 10 (Suppi):S65-71. 4. Farber MD, Fiscella R, Goldberg MF. Aminocaproic acid versus prednisone for the treatment of traumatic hyphema. A randomized clinical trial. Ophthalmology 1991 ;98:27986. 5. Deans R, Noel L-P, Clarke WN. Oral administration of tranexamic acid in the management of traumatic hyphema in children. Can J Ophthalmol 1992;27: 181-3. 6. Volpe NJ, Larrison WI, Hersh PS, Kim T, et al. Secondary hemorrhage in traumatic hyphema. Am J Ophthalmol 1991;112:507-13. 7. Kennedy RH, Brubaker RF. Traumatic hyphema in a defined population. Am J Ophthalmol 1988; 106: 123-30. 8. Crouch ER Jr, Frenkel M. Aminocaproic acid in the treatment of traumatic hyphema. Am J Ophthalmol 1976;81: 355-60.
(%)
Grade
5
their patients who presented more than l day after injury had a significantly elevated rate of secondary hemorrhage (38%), and Volpe et al 6 suggested that distinct patient subsets (black patients and patients with delayed examination) may potentially benefit from aminocaproic acid treatment. The current study has provided guidelines for developing future clinical trials of measures to prevent secondary hemorrhage. The identification of four predictive factors for rebleeding allows categorization of patients with hyphema into subset groups for specific prophylactic treatment at the time of initial examination. Routine hospitalization previously has been shown to have little effect on the secondary hemorrhage rates for patients with uncomplicated hyphema. 22 •23 Hospital cost containment may be effected by selectively admitting and treating those patients whose presenting clinical characteristics have identified them as being at higher risk for secondary hemorrhage.
1588
~20/100
~2
Visual Acuity s;20/200 (%)
Grade 1
Grade
8 21
26
40
20 42
24 48
40 64
11
~2
18
9. McGetrick JJ, Jampol LM, Goldberg MF, et al. Aminocaproic acid decreases secondary hemorrhage after traumatic hyphema. Arch Ophthalmol 1983; 101: 1031-3. 10. Kutner B, Fourman S, Brein K, et al. Aminocaproic acid reduces the risk of secondary hemorrhage in patients with traumatic hyphema. Arch Ophthalmol 1987;105:206-8. 11. Palmer DJ, Goldberg MF, Frenkel M, et al. A comparison of two dose regimens of epsilon aminocaproic acid in the prevention and management of secondary traumatic hyphemas. Ophthalmology 1986;93: 102-8. 12. Skalka HW. Recurrent hemorrhage in traumatic hyphema. Ann Ophthalmol 1978;10:1153-7. 13. Spoor TC, Kwitko GM, O'Grady JM, Ramocki JM. Traumatic hyphema in an urban population. Am J Ophthalmol 1990; 109:23-7. 14. Ganley JP, Geiger JM, Clement JR, et al. Aspirin andrecurrent hyphema after blunt ocular trauma. Am J Ophthalmol 1983;96:797-80 I. 15. Marcus M, Biedner B, Lifshitz T, Yassur Y. Aspirin and secondary bleeding after traumatic hyphema. Ann Ophthalmol 1988;20:157-8. 16. Cassel GH, Jeffers JB, Jaeger EA. Wills Eye Hospital traumatic hyphema study. Ophthalmic Surg 1985;16:441-3. 17. Witteman GJ, Brubaker SJ, Johnson M, Marks RG. The incidence of rebleeding in traumatic hyphema. Ann Ophthalmol 1985;17:525-9. 18. Uusitalo RJ, Ranta-Kemppainen L, Tarkkanen A. Management of traumatic hyphema in children. An analysis of 340 cases. Arch Ophthalmol 1988;106:1207-9. 19. Dieste MC, Hersh PS, Kylstra JA, et al. Intraocular pressure increase associated with epsilon-aminocaproic acid therapy for traumatic hyphema. Am J Ophthalmol 1988;106:38390. 20. Snir M, Axer-Siegel R, Buckman G, Yassur Y. Central venous stasis retinopathy following the use oftranexamic acid. Retina 1990;10:181-4. 21. Mindel JS. Problems in the use of tranexamic acid by ophthalmologists [letter]. Arch Ophthalmol 1989; 107:486-
7.
22. Spaeth GL, Levy PM. Traumatic hyphema: its clinical characteristics and failure of estrogens to alter its course. A double-blind study. Am J Ophthalmol 1966;62: 10981106. 23. Wilson FM II. Traumatic hyphema. Pathogenesis and management. Ophthalmology 1980;87 :910-19.