BJO Online First, published on October 28, 2016 as 10.
1136/bjophthalmol-2016-309736
▸ Additional material is
published online only. To view
please visit the journal online
(http://dx.doi.org/10.1136/
bjophthalmol-2016-309736).
Department of Ophthalmology,
University of Cape Town/
Groote Schuur Hospital, Cape
Town, Western Cape, South
Africa
Correspondence to
A/Prof N Du Toit, Department
of Ophthalmology, University
of Cape Town/Groote
Schuur
Hospital, Ward D4, Anzio
Road, Observatory, Cape
Town, 7806, Western Cape,
South Africa; nagib.dutoit@
uct.ac.za; ndutoit@mweb.co.za
Received 27 September 2016
Accepted 10 October 2016
To cite: Du Toit N,
Mustak S, Cook C. Br J
Ophthalmol Published
Online First: [ please include
Day Month Year]
doi:10.1136/bjophthalmol-
2016-309736
Clinical science
Randomised controlled trial of prophylactic
antibiotic treatment for the prevention of
endophthalmitis after open globe injury at
Groote Schuur Hospital
N Du Toit, S Mustak, C Cook
ABSTRACT the time of admission. This protocol is based on
Background/aims Most post-traumatic acute unpublished data from an incomplete study. Using
infectious endophthalmitis occur within a week of this prophylaxis, we estimated an incidence of
open globe trauma, necessitating early antibiotic 1.36% (19 out of 1393 open globe cases) for acute
7
prophylaxis.
There are few randomised studies that demonstrate the PTE. This estimate came from data collected in a
benefits of prophylactic antibiotics. This randomised Schuur Hospital (GSH) each year and they receive
controlled non-inferiority trial was aimed at determining both oral ciprofloxacin and intravenous cefazolin
the incidence of post-traumatic endophthalmitis using prophylactically for three consecutive days from
established intravenous/oral prophylaxis and
comparing this to the incidence using oral antibiotics
only.
Methods All adult patients admitted with open globe
injury were included. Those with proven endophthalmitis,
high-risk features, who underwent primary evisceration
and those allergic to the trial antibiotics were excluded.
Patients were randomised to receive either intravenous
cefazolin and oral ciprofloxacin or oral ciprofloxacin and
oral cefuroxime for 3 days from admission. Acute
endophthalmitis was the primary outcome. Patients
completed the study if they were followed up for 6 weeks
post injury.
Results Three hundred patients were enrolled, with
150 in each arm. There were 99 exclusions. Seven
patients developed endophthalmitis despite prophylaxis
—2.0% (three cases) in the intravenous and oral arm,
compared
with 2.7% (four cases) in the oral-only arm—this
difference was not statistically significant ( p=0.703).
Conclusions The incidence of endophthalmitis with
prophylaxis was 2–3%. Selected patients with open
globe injuries (without high-risk features) may receive
either intravenous cefazolin and oral ciprofloxacin, or oral
cefuroxime and oral ciprofloxacin as prophylaxis against
acute endophthalmitis—the latter regimen has the
advantage of shortening patients’ hospital stays and
reducing costs. Non-inferiority study-design limitations
should be taken into account, however.
INTRODUCTION
Post-traumatic acute infectious endophthalmitis is a
potentially devastating complication of an open
globe injury (OGI). In most cases, it occurs within
the 1st week of trauma.1–3 It has been reported,
however, to occur months and even years later,4
but these cases are more likely to be caused by
fungi.5 Since the infection in post-traumatic
endophthalmitis (PTE) often worsens despite
prompt treatment and the outcomes are generally
poor,3 5 6 antibiotic prophylaxis after an OGI
should be instituted as early as possible.
Many patients with OGI are admitted to Groote
 Clinical science
retrospective review of dardised prophylactic antibiotic protocol.9 The 3 days, thereby alleviating the strain on hospital
all OGIs over a 10- use of systemic and topical antibiotics is beds and financial
year period (1995– considered prudent. There is little agreement as resources. Lorch and Sobrin9 have also promoted
2004) however, which to the most appropriate route of administration this idea, but stated that, “There are no studies that
raises some doubts over for prophy- laxis, but systemic therapy, usually use oral antibiotics for endophthalmitis prophylaxis
the accuracy of this intravenously, is generally used.4 Systemically as a standard and report infection outcomes”.
figure, since cases may administered antibio- tics are known to This randomised controlled clinical trial was
have been missed. penetrate the vitreous at low levels, but an OGI aimed at determining the incidence of PTE in adult
There are few large, may cause a temporary disrup- tion of the patients admitted to GSH with OGIs, using an
randomised studies that blood-retinal barrier allowing them to more established intravenous/oral prophylactic antibiotic
demonstrate the benefit easily enter the eye.10 regimen and comparing this to the incidence
of using prophylactic It should be determined whether our routine achieved using oral-only antibiotic prophylaxis, so
anti- biotics after prophylactic treatment does indeed result in a as to establish how the latter regimen (experimental
OGI.8 Authors of a low incidence of PTE. Then, a new regimen, which therapy) measured up to the former (active control)
recent review article includes oral antibiotic therapy only, can be on a non-inferiority basis. The primary outcome
noted that despite com- pared with the routine (oral and intravenous) measure was thus the occurrence of acute PTE.
extensive reports on therapy—used as an active control. If the
OGIs and their oral regimen is proven to be effective, patients
outcomes, there is no
MATERIALS AND METHODS
may be discharged on oral therapy only. Their All adult patients admitted to GSH (a tertiary uni-
recommended stan- hospital stays might thus be shortened to less than versity hospital) between April 2012 and October
Du Toit N, et al. Br J Ophthalmol 2016;0:1–6. doi:10.1136/bjophthalmol-2016-309736 1
Copyright Article author (or their employer) 2016. Produced by BMJ Publishing Group Ltd under licence.
 for 2 weeks. All patients were
2014 with an OGI were included. Patients with clinically sus-
pected or proven PTE were immediately treated for endophthal-
mitis and were thus excluded. All patients who presented with
features that were regarded as high risk for infection were
excluded. This group comprised those with intra-ocular foreign
bodies (IOFBs) (visualised clinically or with imaging) and
patients with injuries that were ‘dirty’ that is, a wound that was
caused by a contaminated object or those injured in a rural
setting with organic matter (making fungal infection more
likely). All high-risk patients were treated with prophylactic
intravitreal antibiotics. Primary eviscerations were performed in
those OGIs who displayed the following features: visual acuity
(VA) of no light perception, a total afferent pupil defect (APD),
a normal fellow eye, irreparable wounds and informed patient
consent. Eviscerated cases were thus excluded as well. Patients
known to be allergic to any of the trial antibiotics or to have
had anaphylactic reactions to penicillin were excluded—these
patients were given prophylactic vancomycin.
‘Intravenous and oral’ versus ‘oral only’ treatment arms were
formed on the basis of providing antibiotic prophylaxis against
the various common pathogens in acute PTE. Patients were ran-
domised to receive either ‘protocol 1’ (intravenous cefazolin 1 g
8 hourly and oral ciprofloxacin 750 mg 12 hourly for 3 days) or
‘protocol 2’ (750 mg of oral ciprofloxacin and 250 mg of oral
cefuroxime 12 hourly each for 3 days). Treatment was com-
menced on admission.
A review article reported an incidence for PTE (without anti-
biotic prophylaxis) ranging from 7% to 17%.4 We aimed to
reduce the incidence of PTE to around 2% ( previously calcu-
lated to be the approximate PTE incidence with prophylaxis in
our department) in each arm of the study, using either of the
two prophylactic regimens, on a non-inferiority basis. Using a
published power calculator (Blackwelder’s approach),11 with set-
tings of 80% power and 5% significance, a 2% success rate for
both our standard and experimental treatment arms and a non-
inferiority limit of 4%, the sample size required was calculated
to be 152 in each arm.12 Randomisation was performed by the
author (who did not participate in the treatment of patients)
and was achieved by randomly placing an equal number of slips
marked as either ‘protocol 1’ or ‘protocol 2’ into opaque, sealed
envelopes. These envelopes were mixed, then consecutively
numbered from ‘1’ to ‘300’ (by the author) and were available
for selection in sequential order for each patient admitted (by
the residents). We aimed to enter 150 patients into each treat-
ment arm.
A history, including the details of the injury, and
examination via slit lamp biomicroscopy was performed in all
cases on admission. The extent of the wound in terms of zone,
type and grade of injury was documented as per the Ocular
Trauma Score.13 The presence of APD, iris prolapse, lens
disruption, vit- reous in the wound and hyphaema were also
documented on the patient charts. The treatment envelope was
then selected by the resident and the patient treated according
to the protocol
on the randomised slip. The eye injury was repaired and find-
ings rechecked intraoperatively. General ophthalmological
prin- ciples of open globe repair were followed. Uveal tissue
was reposited when found to be viable and free of debris,
but excised when contaminated or necrotic. The timing of
presenta- tion and of primary repair was recorded in hours
post injury. Any delays in initiating prophylactic therapy or
primary surgery after the trauma were also recorded.
Postoperatively, the treat- ment regimen included topical
betamethasone/chloramphenicol combination drops administered
six times daily for 6 weeks and atropine drops twice daily
2 Du Toit N, et al. Br J Ophthalmol 2016;0:1–6. doi:10.1136/bjophthalmol-2016-309736
hospitalised for at least 3 days. On a daily basis the patients
were checked for symptoms and signs of endophthalmitis, that
is, increasing pain, periocular swelling, further loss of vision,
signs of severe intraocular inflammation (including hypopyon),
vitreal haze and debris, retinitis and vasculitis. This was done
for the entire duration of the hospital stay, at discharge and at
the first follow-up visit (usually a week or two after discharge).
Any patients diagnosed with acute PTE during their hospital
stay had the diagnosis of endophthalmitis confirmed by two
other observers. These patients were then managed with vitreal
taps, intravitreal injection of antibiotics and therapeutic pars
plana vitrectomy (PPV) on an individual case basis. In all
patients, acuity at the last visit post injury (at least 6 weeks) was
recorded and any further signs of endophthalmitis documented.
Data were collected prospectively via a standardised data
sheet. Patients were deemed to have completed the study if they
had been followed up for at least 6 weeks post injury. Data
were analysed using the statistical program Stata V.9.0.
Variables were described using means, medians and
proportions, as appropri- ate. The main analysis focused on
determining the incidence of PTE after OGIs using our
standard prophylactic therapy and describing the differences (if
any) in the effectiveness of the two prophylactic regimens.
Categorical variables were analysed using
the χ2 test. ORs and 95% CIs were estimated using logistic
regression analysis. Factors were considered statistically signifi-
cant if p<0.05. Informed consent was obtained from all
patients. Patient confidentiality was strictly maintained. Ethical
approval was obtained.

RESULTS
Three hundred patients were enrolled in the study—see figure 1.
The two groups were equivalent in terms of demographics
and clinical findings, showing no significant differences with
regards to any parameters (see tables 1 and 2).
The cases of endophthalmitis in each arm are shown in table 3.
A total of seven patients developed PTE despite prophylaxis. The
frequency of endophthalmitis was thus 2.0% (3 out of 150) with
protocol 1 and 2.7% (4 out of 150) with protocol 2—this differ-
ence was not statistically significant (OR 1.34; 95% CI 0.2953 to
6.1037, p=0.703). The details of these PTE cases are seen in
table 4.
There were 99 cases excluded from the study—their break-
down is illustrated in table 5.

DISCUSSION
Despite important advances in medical and surgical manage-
ment, PTE continues to have a poor prognosis; so with this in
mind, ‘prevention remains the best cure’.

Choice of antibiotics and the need for prophylaxis

Endophthalmitis following OGI is caused by a specific range of
microorganisms, of which coagulase-negative Staphylococcus
and Bacillus spp are the most frequent.14 15 Infections with
more than one organism are also more likely.16 The precise
route of administration of prophylactic antibiotics remains
unresolved.
Antibiotics have been used intravitreally, subconjunctivally,
topically and/or systemically.17 Systemic prophylaxis is hindered
by the poor ocular penetration of most drugs, except for imipe-
nem, ofloxacin and ciprofloxacin.18 Fluoroquinolones may be
used in the systemic prophylaxis of OGIs because of the good
intraocular penetration and bactericidal activity against many
organisms which are known to cause PTE.4 Cephalosporins,
Du Toit N, et al. Br J Ophthalmol 2016;0:1–6. doi:10.1136/bjophthalmol-2016-309736 3
 Clinical science

Figure 1 Subject flow chart

(according to Consort Statement
guidelines).

4 Du Toit N, et al. Br J Ophthalmol 2016;0:1–6. doi:10.1136/bjophthalmol-2016-309736

 Clinical science

when given intravenously, developing and is used as a there is contamination with f

achieve therapeutic last treatment resort for organic matter.17 19 Such
intravitreal concentrations.17 endophthalmitis in cases IOFB cases (21 in total) were P
Intravitreal injection of where therapy has failed. In therefore excluded. Certain T
antibiotics may be a retrospective review of risk factors were not selected E
important in the prevention 675 as exclusion criteria—these The incidence of
of acute PTE.19 Routine use OGIs over a 7-year period, included delays in case endophthalmitis with either
of intravitreal antibiotics in the authors excluded 5.8% presentation and delays in method of prophylaxis at
all OGIs has been advocated of cases due to globe the initiation of prophylaxis our institution was between
by certain authors,2 but this removal.3 In the current of more than 24 hours.2 21 22 2% and 3%. The incidence
sug- gestion has not been study, about 15% (61 out of Essex et al21 noted that there of PTE in the literature, based
universally accepted since 401) of all OGIs were was no safe delay and that it largely on retrospective
there are definite risks to excluded due to evisceration was conceivable that the risk studies, varies widely. In
intravitreal injections and it is conceivable that this of infection increased China, a total of 4795
which therefore should be may have been a confounding steadily with each hour. OGIs were
reserved for high-risk cases factor in cre- ating the Therefore any delay may be
only.4 8 20 impression of a low incidence detrimental and time periods
of endophthalmitis with used as cut-off points in
prophylaxis. different studies may be
D The risk factors for PTE,
i arbitrary. Andreoli et al,3
including IOFB, have been suggested that neither a delay
s identi- fied in several in presentation nor a delay
c studies.4 14 17 19
in the repair of an OGI
u Prophylactic intravitreal increased the risk of
s anti- biotic injections should infection. The mean delay
s be used when there is an from injury to antibiotic
i IOFB or when prophylaxis was around 2
o days and the delay from
n injury to surgery around 2–3
days in the current study.
o Had we adopted the cut-off
f of 24 hours to treatment as
an exclusion cri- terion,
e almost all of our cases
x would have been ineligible.
c Another risk factor for
l endophthalmitis after OGI, is
u lens involvement,1 21 but
d this also was not selected as
e an exclusion criterion
d because it has been suggested
that the perceived associ-
ation between IOFB and
c
lens rupture may have led
a to the impression of an
s increased risk of PTE with
e lens rupture and there is also
s the possibility that a sterile
An important group lens antigen-related
excluded was made up of inflammatory response may
those in whom PTE was
masquerade as PTE.21
diagnosed at first
presentation (16 cases).
Early com- prehensive I
treatment in the form of PPV n
and injection of intravi- treal c
antibiotics may significantly i
improve visual outcomes in d
PTE following OGI.14 e
Patients who underwent n
primary eviscerations c
totalled 61 cases. e
Evisceration prevents
endophthalmitis from o

Du Toit N, et al. Br J Ophthalmol 2016;0:1–6. doi:10.1136/bjophthalmol-2016-309736 5

 Clinical science

6 Du Toit N, et al. Br J Ophthalmol 2016;0:1–6. doi:10.1136/bjophthalmol-2016-309736

 Clinical science

Table 1 Comparison of our routine versus oral treatment arms Table 2 Comparison of the two arms in terms of clinical findings
in
terms of demographics
Intravenous/oral Oral only Significance
Intravenous/oral Oral only Significance (n=150) (n=150) testing
(n=150) (n=150) testing
Affected Eye
Age, years Left 63 57
2
χ p=0.450
Min 21 16 t-test p=0.236 Right 86 93
Max 64 68 Both 1 0
Mean 38.2 36.3 VA affected eye (LogMAR)
SD 12.6 14.8 Min 0 0 t-test p=0.189
Gender Max 4 4
2
Male 112 118 χ p=0.413 Mean 2.9 2.7
Female 38 32 SD 1.3 1.2
Ethnicity Type of injury
2
Black 89 76 χ p=0.231 Rupture 34 35
2
χ p=0.816
White 1 0 Penetrating 116 112
Coloured 60 73 IOFB 0 0
Asian 0 1 Perforating 0 2
Nature of injury Mixed 0 1
2
Assault 108 105 χ p=0.121 Grade
Accidental 30 41 A 8 8
2
χ p=0.498
Work related 8 0 B 6 10
MVA 4 3 C 5 3
Sports 0 1 D 126 119
Site E 5 10
2
Home 69 73 χ p=0.469 Zone
Recreation 41 37 I 41 45
2
χ p=0.341
Transport 8 15 II 51 59
Other 23 24 III 58 46
Work 9 5 OTS Score
Object Min 27 27 t-test p=0.603
2
Knife 32 30 χ p=0.812 Max 100 100
Bottle 26 30 Mean 62.4 63.6
Glass 22 26 SD 18.6 17.9
Other 70 64 VA at discharge (LogMAR)
Education Min 0.2 0.3 t-test p=0.594
2
Primary 66 72 χ p=0.708 Max 4.0 4.0
Secondary 80 73 Mean 2.8 2.8
Tertiary 4 5 SD 1.1 1.1
Polytrauma VA at completion (LogMAR)
2
Yes 21 19 χ p=0.905 Min 0.1 0.2 t-test p=0.383
No 139 131 Max 4 4
Employed Mean 2.1 2.2
2
Yes 58 64 χ p=0.481 SD 1.6 1.4
No 92 86 LogMAR, logarithm of the minimum angle of resolution; OTS, Ocular Trauma
MVA, motor vehicle accident. Score; VA, visual acuity.

studied with a PTE incidence of 11.91%—eviscerated cases phenomenon of ‘culture-negative endophthalmitis’ is not
were excluded and the authors do not mention the routine use uncommon, while culture positivity does not always correlate
of antibiotic prophylaxis.22 In Iran, 117 cases of PTE out of
2340 OGIs were reported (an incidence of 5.1%) when patients
were given intravenous antibiotics (cefazolin and gentamicin)
for a period of 3 days.23 In a review of 558 patients with OGI
from the USA,3 an incidence of 0.9% was found when intraven-
ous vancomycin and ceftazidime prophylaxis was used. The
wide variation in PTE rates in patients with OGI may be attribu-
ted to different risk factors in the different patient populations,
as well as variations in antibiotic prophylactic regimens.8
In 6 (out of 7) PTE cases in the current study, there were no
organisms isolated or cultured, that is, culture-negative. The

Du Toit N, et al. Br J Ophthalmol 2016;0:1–6. doi:10.1136/bjophthalmol-2016-309736 7

 Clinical science

with infection.8 Reports on the prevalence of culture-negative

cases of PTE range from 17% to 42%.19 21 24 Several authors
have emphasised the importance of using a clinical diagnosis of
endophthalmitis and thus have included culture-negative cases
in their studies.19 21 23

Strengths and limitations

The main strength of this study lies in the fact that it is pro-
spective and randomised. Most other studies on this subject are
retrospective. A major limitation lies in the fact that 99 patients
fulfilled the criteria for exclusion and were thus not included in
the study. Another limitation of this study is the short follow-up
period, although the primary outcome was stated to be acute
PTE which tends to occur early after OGI. A further limitation
lies in the non-inferiority design of the study. These types of

8 Du Toit N, et al. Br J Ophthalmol 2016;0:1–6. doi:10.1136/bjophthalmol-2016-309736

 Table 3 Comparison of the two arms in terms of endophthalmitis Table 5 Numbers in the groups that were excluded from the
cases and time periods study (n=99)
Intravenous/oral Oral only Significance Exclusion criteria Number
(n=150) (n=150) testing
Primary evisceration performed 61
Endophthalmitis Intraocular foreign body 21
Yes 3 4 χ2 p=0.702 Endophthalmitis diagnosed at presentation 16
No 147 146 Allergy to antibiotics used in study 1
Time from injury to surgery (hours) Wound contaminated with organic matter 0
Min 12 9 t-test p=0.475
Max 326 432
Mean 70.5 76.5 coverage with oral fluoroquinolones for Gram-negative coverage.
SD 63.2 80.5 The first published randomised trial (2016) using oral
Time from injury to antibiotic prophylaxis (hours) antibiotics included 1255 patients with OGIs in a hospital
Min 3.5 2 t-test p=0.054 setting in Iran (2011–2013). The patients received intravenous or
Max 283 408 oral (ciproflox- acin only) systemic antibiotics to prevent PTE.
Mean 40.6 55.2 The authors found a PTE rate of 2.1% in their intravenous
SD 47.0 79.5 group and 2.2% in their oral group.26 In a retrospective study,
Andreoli et al3 found an incidence of PTE of just under 1.0%
(among the lowest reported in the literature) when intravenous
prophylactic treatment was combined with early primary wound
studies are prone to difficulties, but a placebo-controlled trial is
closure. Having determined a rate of 2.0% for intravenous
only ethical if no standard treatment exists and there are no
combined with oral prophylaxis and 2.7% for oral antibiotics
risks associated with delaying treatment.25 This is clearly not only, the current study (which is the
the case with the prophylactic treatment of OGIs, where a delay first randomised trial to combine oral penicillins with oral fluoro-
increases the risk of PTE. In this situation, active-controlled quinolones for more comprehensive cover) may have
trials on a non-inferiority basis are generally appropriate.12 In provided more treatment options.
theory, a control group could have had prophylaxis with topical Patients with OGIs should be treated for PTE if already
or periocular antibiotics at the time of and following surgery. infected; be eviscerated if warranted by severe injuries; and
Also, the use of outpatient surgery may have shortened the time receive intravi- treal injections of prophylactic antibiotics if risk
between injury and surgical repair, since the long duration factors for infec- tion are present. The rest may receive
between injury and presentation may also have been a con- prophylaxis either in the form of 3 days of intravenous
founding factor. In theory, however, these delays should cefazolin and oral ciprofloxacin, or
possibly have led to higher infection rates. It is, however, 3 days of oral cefuroxime and oral ciprofloxacin, provided
recommended that larger randomised trials be performed to that they are not allergic. The latter regimen has the advantage of
confirm that oral prophylactic therapy alone is sufficient. short- ening patients’ hospital stays, freeing up beds and reducing
costs.
CONCLUSION
Contributors All the authors contributed towards: The design of the work and the
The findings of this study are significant and are relevant to interpretation of data; the drafting of the work; the final approval of the version for
current practice. Lorch and Sobrin9 have stated that there is publication; the agreement to be accountable for all aspects of the work in terms
no consensus on the best method of antibiotic administration, as of accuracy. NDT was the main author who did the write-up and submission. SM
there have been few randomised controlled trials on this also contributed towards the acquisition of data and analysis. CC helped with
revising
subject and there are no studies combining oral penicillins for
the work critically.
Gram-positive
Competing interests None declared.

Table 4 Some important features of the endophthalmitis cases Ethics approval Human Research Ethics Committee of University of cape Town.
in the study (n=7) Provenance and peer review Not commissioned; externally peer reviewed.

REFERENCES
1 Danis RP. Endophthalmitis. Ophthalmol Clin N Am 2002;15:243–8.
2 Narang S, Gupta V, Gupta A, et al. Role of prophylactic intravitreal antibiotics
in
Oral Intravenous open globe injuries. Indian J Ophthalmol 2003;51:39–44.
only and oral 3 Andreoli CM, Andreoli MT, Kloek CE, et al. Low rate of endophthalmitis in a
large series of open globe injuries. Am J Ophthalmol 2009;147:601–8.
Number 4 3 4 Bhagat N, Nagori S, Zarbin M. Post-traumatic infectious endophthalmitis. Surv
Mechanism of injury—assault 2 3 Ophthamol 2011;56:214–51.
OTS Score (range) 60–80 70–90 5 Meredith TA. Post-traumatic endophthalmitis. Arch Ophthalmol 1999;117:320–1.
Time from injury to antibiotic prophylaxis 12–144 14–130 6 Das T, Choudhury K, Sharma S, et al. Endophthalmitis Research Group. Clinical
(hours) profile and outcome in Bacillus endophthalmitis. Ophthalmology
2001;108:1819–25.
Time from injury to surgery (hours) 22–192 22–144
7 Du Toit N, Motala MI, Richards J, et al. The risk of sympathetic ophthalmia
Days from injury to diagnosis of 4–10 2–15 following evisceration for penetrating eye injuries at Groote Schuur Hospital.
endophthalmitis Br J Ophthalmol 2008;92:61–3.
Microbial culture or isolate—positive 0 1 Pain at presentation 1 2
Lid swelling present 0 0 8 Ahmed Y, Schimel AM, Pathengay A, et al. Endophthalmitis following open-globe
Hypopyon present 2 3 injuries. Eye (Lond) 2012;26:212–17.
9 Lorch A, Sobrin L. Prophylactic Antibiotics in Posttraumatic Infectious
Vitritis visible No No
Endophthalmitis. Int Ophthalmol Clin 2013;53:167–76.
OTS, Ocular Trauma Score. 10 Bunt-Milam AH, Black RA, Bensinger RE. Breakdown of the outer blood-retinal
barrier in experimental commotio retinae. Exp Eye Res 1986;43:397–412.
11 Blackwelder WC. “Proving the Null Hypothesis” in Clinical Trials. Control Clin Trials
1982;3:345–53.
12 Sealed Envelope Ltd. Power calculator for binary outcome non-inferiority trial. 2012. 20 Cebulla CM, Flynn HW Jr. Endophthalmitis after open globe injuries.
https://www.sealedenvelope.com/power/binary-noninferior/ Am J Ophthalmol 2009;147:567–8.
13 Kuhn F, Maisiak R, Mann L, et al. The Ocular Trauma Score (OTS). Ophthalmol Clin 21 Essex RW, Yi Q, Charles PG, et al. Post-traumatic endophthalmitis. Ophthalmol
North Am 2002;15:163–5. 2004;111:2015–22.
14 Duch-Samper AM, Chaqués-Alepuz V, Menezo JL, et al. Endophthalmitis following 22 Zhang Y, Zhang MN, Jiang CH, et al. Endophthalmitis following open globe injury.
open-globe injuries. Curr Opin Ophthalmol 1998;9:59–65. Br J Ophthalmol 2010;94:111–14.
15 Davey RT, Tauber WB. Posttraumatic endophthalmitis: the emerging role of Bacillus 23 Faghihi H, Hajizadeh F, Esfahani MR, et al. Posttraumatic endophthalmitis: report
cereus infection. (Abstract) Rev Infect Dis 1987;9:110–23. No. 2. Retina 2012;32:146–51.
16 Alfaro DV, Roth D, Liggett PE. Posttraumatic endophthalmitis—causative organisms, 24 Long C, Liu B, Xu C, et al. Causative organisms of post-traumatic endophthalmitis:
treatment, and prevention. Retina (Philadelphia, Pa) 1994;14:206–11. a 20-year retrospective study. BMC Ophthalmology 2014;14:34.
17 Mittra RA, Mieler WF. Controversies in the management of open-globe injuries 25 D’Agostino RB Sr, Massaro JM, Sullivan LM. Non-inferiority trials: design concepts
involving the posterior segment. Surv Ophthalmol 1999;44:215–25. and issues—the encounters of academic consultants in statistics. Statist Med
18 Peyman GA, Daun M. Prophylaxis of endophthalmitis. Ophthalmic Surg 2003;22:169–86.
1994;25:671–4. 26 Tabatabaei SA, Soleimani M, Behrooz MJ, et al. Systemic oral antibiotics as a
19 Soheilian M, Rafati N, Mohebbi MR, et al. Prophylaxis of acute posttraumatic prophylactic measure to prevent endophthalmitis in patients with open globe
bacterial endophthalmitis: a multicenter, randomized clinical trial of intraocular injuries in comparison with intravenous antibiotics. Retina (Philadelphia, Pa)
antibiotic injection, report 2. Arch Ophthalmol 2007;125:460–5. 2016;36:360–5.