Case-control Study of Endophthalmitis After

Cataract Surgery Comparing Scleral Tunnel

and Clear Corneal Wounds
BLAKE A. COOPER, MD, NANCY M. HOLEKAMP, MD, GEORGE BOHIGIAN, MD,
AND PAUL A. THOMPSON, PHD

PURPOSE: To study the possible association between

clear corneal incision with or without placement of a
suture during cataract extraction and postoperative endophthalmitis.
DESIGN: Retrospective, comparative, case-controlled
study.
METHODS: Thirty-eight patients treated for culturepositive, acute postcataract surgery endophthalmitis, and
371 randomly selected control patients who underwent
uncomplicated cataract surgery in the referring community were studied. Incision type and use of suture during
cataract surgery of endophthalmitis patients were compared with the controls. The data were analyzed using
logistic regression methods.
RESULTS: Of the 38 patients with endophthalmitis, 17
patients (45%) had clear corneal incisions and 21 patients (55%) had a scleral tunnel incision. In 371
controls, 76 patients (20%) had clear corneal incisions
and 295 patients (80%) had scleral tunnel incisions.
Clear corneal incision was associated with a threefold
greater risk of endophthalmitis than was scleral tunnel
incision (Odds Ratio, 3.36, 95% Confidence Interval
1.67 to 6.78). The type of incision was significant (2
11.53, P .0007); a clear corneal incision was more
frequently associated with endophthalmitis. A subgroup
analysis revealed that the presence or absence of a suture
was not significant (2 1.31, P .2524).
CONCLUSIONS: In this retrospective, case-controlled
study, clear corneal incisions were found to be a statistically significant risk factor for acute postcataract
surgery endophthalmitis when compared with scleral
tunnel incisions. (Am J Ophthalmol 2003;136:
300 305. 2003 by Elsevier Inc. All rights reserved.)
Accepted for publication Feb 10, 2003.
InternetAdvance publication at ajo.com Feb 13, 2003.
From the Barnes Retina Institute (N.M.H.), and Washington University School of Medicine (B.A.C., N.M.H., G.B., P.A.T.), St. Louis,
Missouri.
Inquiries to Nancy M. Holekamp, MD, Barnes Retina Institute, 1600
South Brentwood Boulevard, 8th Floor, St. Louis, MO 63144; fax: (314)
367-1503; e-mail: nholekamp@pol.net

300

2003 BY

HE EVOLUTION OF MODERN CATARACT SURGERY IS

a series of remarkable technical refinements. Many

of these advances involved changing the type of
surgical incision. Examples are as follows: a transition from
intracapsular cataract extraction to extracapsular cataract
extraction; a transition from extracapsular technique to
small-incision scleral tunnel with phacoemulsification;
and, finally, a transition from scleral tunnel incision to
clear corneal incision.
The driving forces behind many of these refinements
have included decreased time of operation, simplified
postoperative care, and faster visual recovery. At the same
time, much of this evolution has also been justified by
reducing the risk of complications. Specifically regarding
endophthalmitis, the transition from intracapsular surgery
to extracapsular surgery was associated with a lower rate of
postoperative infection,1 a finding that was subsequently
confirmed in Denmark.2 Transitioning from extracapsular
technique to small incision scleral tunnel with phacoemulsification has been demonstrated to have an equivalent
rate of endophthalmitis.1 A recently published report
examined the transition from scleral tunnel incision to
clear corneal incision.3 In that study the incidence of
endophthalmitis was higher with clear corneal incision but
the difference was not found to be statistically significant.
It is important to note that this conclusion arose from only
8 cases of culture-positive endophthalmitis.
The objective of this case-controlled study was to
answer the following two questions: (1) is clear corneal
incision associated with greater odds of endophthalmitis
compared with a scleral tunnel incision?; and (2) does the
presence of a suture modify the risk of endophthalmitis for
either clear corneal or scleral tunnel incisions?

DESIGN
THIS WAS A RETROSPECTIVE, COMPARATIVE, CASE-CON-

trolled study.

ELSEVIER INC. ALL

RIGHTS RESERVED.

0002-9394/03/$30.00
doi:10.1016/S0002-9394(03)00202-2

METHODS
WE PERFORMED A RETROSPECTIVE REVIEW OF ALL PA-

tients with culture-positive acute postoperative endophthalmitis following cataract surgery from a tertiary referral
center, the Barnes Retina Institute, for a 4-year period
from January 1, 1997 through December 31, 2000. Acute
postoperative endophthalmitis was defined as culturepositive infection occurring within 6 weeks of surgery.
Patients who underwent secondary lens implantation,
intraocular lens (IOL) exchange, or cataract surgery combined with filtering procedures or corneal transplantation
were excluded. Operative reports from the referring physicians were reviewed to determine the type of incision and
closure that was used during the cataract surgery. When
operative reports were not available, the records of the
Barnes Retina Institute were reviewed to determine the
type of incision and closure.
Patients in whom the type of incision and closure could
not be determined were excluded. In addition, the operative note of the treating retina specialist was reviewed to
find comments on cataract wound abnormality and need
for revision.
Three major centers for cataract surgery in the community serve as the referral base for the Barnes Retina
Institute. Each center is comprised of approximately 15
active cataract surgeons. Each center performs approximately 1,000 to 1,500 cataract surgeries per year. A control
group was established by randomly selecting 400 operative
reports coded for cataract extraction with lens implantation over the same 4-year period (100/year) equally from
each of the centers. Because this was a randomly selected
retrospective control group, the control group was not
matched to the study group on any factors such as age,
gender, or ocular characteristics. Patients where the incision type and or closure could not be determined from the
operative reports were excluded. Secondary lens implantation, IOL exchange, cataract surgery combined with other
procedures, and patients with complications were excluded. Data from these operative reports were used to
establish a baseline for comparing the type of cataract
incision and closure that was being used in the referring
community during that time period.
The data were analyzed using logistic regression methods. Incision type and suture presence were considered
independent variables. The presence/absence of endophthalmitis was the dependent variable. Thus, the analysis
examined the influence of incision type (clear corneal,
scleral tunnel) and suture (suture present, suture absent)
on risk of endophthalmitis. The analysis first examined the
effects of suture and incision type with their possible
interaction. Following this examination, the analysis examined the effects of suture and incision type factors
without consideration of a possible interaction. More
specific analyses (effect of suture for scleral tunnel cases,
effect of suture for clear corneal cases, effect of incision
VOL. 136, NO. 2

type for cases without suture, effect of incision type for

cases with suture) were conducted to specifically identify
what factors or combination of factors were most important. To determine if the findings in this report would be
altered substantially by differences in the composition of
the control group, Monte Carlo statistical analyses were
performed.4 We wanted to rule out the possibility that
sampling variability associated with the retrospectively
selected control group could have overly affected the
findings. All records were de-identified to maintain confidentiality of the medical record, and to maintain anonymity of patient and surgeon.

RESULTS
SIXTY-FIVE PATIENTS WITH POSTCATARACT SURGERY EN-

dophthalmitis were identified over the 4-year period of this

study. Twenty-seven patients were excluded from analysis
for the following reasons: 16 patients were culture-negative; 5 patients had complications during the cataract
extraction or within the postoperative period; and 6
patients had incomplete records. Thus, 38 patients of
culture-positive, acute postcataract surgery endophthalmitis met criteria for inclusion into this study. In 15 of 38
patients treated for endophthalmitis, the treating retina
surgeon made a note in the operative report on the
integrity of the cataract incision. In 8 patients there was a
wound abnormality (poor closure, wound gape, wound
leak) requiring one or more sutures.
Of the 38 patients, 17 patients (45%) had clear corneal
incisions and 21 patients (55%) had a scleral tunnel
incision. Of the 17 patients with clear corneal incisions, 12
(71%) were sutureless and the remaining 5 patients
(29%) had at least one suture. Of the 21 patients that had
scleral tunnel incisions, 6 patients (29%) were sutureless
and the remaining 15 patients (71%) had at least one
suture (Figure 1).
Of the 400 randomly selected operative reports coded as
cataract extraction with lens implantation from the referring community, 29 were excluded from the study because
the incision type or closure could not be determined from
the operative report, and thus 371 control patients were
obtained. In the 371 control patients, 76 patients (20%)
had clear cornea incisions and the remaining 295 (80%)
patients had scleral tunnel incisions. Of the 76 clear
corneal incisions, 48 patients (63%) were sutureless and
28 patients (37%) had at least one suture. Of the 295
scleral tunnel incisions, 149 patients (51%) were sutureless and 146 patients (49%) had at least one suture
(Figure 1).
The data were analyzed using logistic regression methods
with incision type and suture presence and their possible
interaction as predictors of endophthalmitis in PROC

ENDOPHTHALMITIS AFTER CATARACT SURGERY

301

FIGURE 1. Incision and closure type in 38 patients with endophthalmitis over the 4-year period of the study and incision and
closure type in the controls from the referring community over the 4-year period of the study is illustrated.

LOGISTIC.5,6 The interaction between incision type and

suture existence was not significantly associated with
endophthalmitis (2 2.76, df 1, P .0965). Using a
model with only the incision type and suture presence as
predictors, the type of incision was significant (2 11.53,
df 1, P .0007, OR 3.36, 95% CI 1.67 6.78); a
clear corneal incision was more frequently associated with
endophthalmitis. Suture presence (presence or absence)
was not significantly associated with endophthalmitis (2
1.31, df 1, P .2524, Odds Ratio [OR] 0.67, 95%
Confidence Interval [CI] 0.34 1.33). To more fully
understand the results, more specific questions were asked.
For clear corneal incision patients, suture presence (presence or absence) was not significant (2 0.33, df 1, P
.5638, OR 1.4, 95% CI 0.45 4.4). For scleral
tunnel incision patients, the suture presence (presence,
absence) was not significant (2 3.55, df 1, P .0594,
OR 0.39, 95% CI 0.151.04). Although not statistically significant, the value is sufficiently close to significance, given the small sample size, to warrant future
investigation. For patients with no suture, incision type
(clear corneal vs scleral tunnel) was significant, with clear
corneal methods more likely to be associated with endophthalmitis (2 12.01, df 1, P .0005, OR 6.21, 95%
CI 2.2217.44). For patients with a suture, incision type
(clear corneal vs scleral tunnel) was not significant
(2 0.99, df 1, P .3202, OR 1.74, 95% CI
0.58 5.17).
A secular trend was noted that may affect the interpretation of this study. Specifically, the use of clear corneal
incisions increased throughout the first 3 years in this study
(Table 1). Therefore, it is important to ascertain that the
controls, even though they were randomly selected, remained representative of cataract surgeries done over the
302

AMERICAN JOURNAL

TABLE 1. Number of Eyes in Control Group With Clear

Corneal Incisions Listed by Year

Year

1997
1998
1999
2000

Control Group With Clear

Corneal Incisions

9
11
28
28
Total: 76

time period of this study. This was done using a random

selection method.
The control sample composition may be very important
in the observed results. Because it was obtained using a
random selection process, unknown factors may affect its
composition, and in turn, affect the observed rate of
endophthalmitis. To examine the potential impact of the
control sample, Monte Carlo simulations were performed
with 3,600 different simulated control samples generated
randomly and used in an analysis (the Monte Carlo
technique). These simulated control samples were generated by varying the control sample characteristics. The
proportion of clear corneal incisions was varied (10%,
20%, 30%, and 40%). The proportion of sutures for clear
corneal cases was varied (33%, 50%, 67%). The proportion of sutures for scleral tunnel cases was varied (33%,
50%, 67%). These three factors are sufficient to fully
specify the control sample composition. The final step
involved randomly changing the sample compositions,
using a random number and the standard deviation of
proportions. Each randomly generated data set was then
OF

OPHTHALMOLOGY

AUGUST 2003

analyzed using the same logistic analysis methods as were

used on the actual data.5,6 For each data set, the different
2 values (for incision type, the suture presence, and the
interaction between these factors) were accumulated and
classified as either significant (P .05) or not significant
(P 0.05). Thus, at the end of this portion of the
evaluation, 3,600 P values were generated for each of the
effects being examined. To evaluate the control sample
factors, these classified P values were examined using
logistic regression to see which factors affect the likelihood
of a significant result.
The Monte Carlo simulations support the finding that
the important factor in the risk of endophthalmitis was the
clear corneal incision rate in the control sample. When
the clear corneal rate was low, incision type was almost
always significant (99% of the low-frequency clear corneal
samples had a significant incision type effect). When the
clear corneal rate was higher, incision type was less
frequently significant (45% of the high frequency clear
corneal samples had a significant incision type effect). The
difference between the low and high rate of clear corneal
incision was significant (2 541.8, df 1, P .0001),
indicating that clear corneal frequency in the control
sample was important. Neither of the other two factors was
important (proportion of sutures for clear corneal patients:
2 0.7, df 1, P .4011; proportion of sutures for
scleral tunnel patients: 2 0.3, df 1, P .5915).
The Monte Carlo evaluation answers the question:
How much would the composition of this sample need to
change to affect the results? The conclusion is that the
sample would need to change greatly. Were the frequency
of clear corneal incisions to increase substantially (from
the observed 19% to 40%), then the incision type might
not be significant. This is a very considerable change,
however. The simulations suggest that the findings of this
study would probably not be changed if the composition of
the control group data changed in a small way.

DISCUSSION
IN THIS RETROSPECTIVE, CASE-CONTROLLED STUDY, CLEAR

corneal incisions were found to be associated with threefold greater odds of acute postcataract surgery endophthalmitis, compared with scleral tunnel incisions. The gold
standard in any clinical research endeavor is a prospective,
randomized, masked study. However, endophthalmitis is
such a rare event that such a clinical trial would need to
include nearly 100,000 patients to have sufficient study
power to detect even a 50% difference in the risk of
endophthalmitis between two groups of patients. Moreover, because surgeons tend to have preferences for one
surgical approach versus another and may be more skilled
in one approach as opposed to the other, such a clinical
trial is logistically challenging. Therefore, a case-controlled study such as this can play an important role in the
VOL. 136, NO. 2

scientific method by identifying risk factors that warrant

further discussion and examination.
In the recent ophthalmic literature, wound abnormality
has been identified as a risk factor for endophthalmitis. In
a 10-year review of endophthalmitis at the Bascom Palmer
Eye Institute, the authors concluded that postoperative
wound defects are a risk factor for the development of
endophthalmitis.7 In a case-control study retrospectively
reviewing 22,091 cataract operations performed at a single
institution, wound abnormality was found to be a statistically significant risk factor for postcataract surgery endophthalmitis.8 Also, Maxwell and Diamond9 reported an
association between surgical wound defects and endophthalmitis. In that report of 25 patients with culture-proven,
postsurgical endophthalmitis, more than half demonstrated wound abnormalities including frank gape, wound
leak, necrosis, loose apposition, suture abscess, and dehiscence. Our findings are consistent with these reports. Of 15
patients in which the integrity of the cataract wound was
checked, 8 patients were treated for a wound abnormality.
We now recommend that the treating retina surgeon
carefully inspect the cataract incision in all suspected cases
of endophthalmitis. If a wound abnormality is found, a tap
and inject in the office should be deferred in favor of
returning to the operating room for wound revision and
management of endophthalmitis.
We propose two possible explanations why clear corneal
incisions may be more prone to wound abnormality and
subsequently endophthalmitis. First, whenever a new surgical technique is introduced, an associated learning curve
is experienced. It is important to note that this community
of cataract surgeons appeared to be undergoing a transition
from scleral tunnel incision to clear corneal incision during
cataract surgery (Table 1). The use of a clear corneal
incision in the control group increased through the first 3
years of the study. It is possible that there is a steep
learning curve to the construction of a watertight clear
corneal incision. Therefore, a period of transition from a
scleral tunnel incision to a clear corneal incision may be
accompanied by a transient bump in the incidence of acute
postoperative endophthalmitis. A similar situation would
be the transition from an extracapsular technique to
phacoemulsification being accompanied by a transient
bump in the incidence of dropped nuclei. However, once
the technique is mastered it is certainly plausible that the
incidence of the complication would return to normally
low levels. This hypothesis is supported by the Monte
Carlo Analysis, which found that if the percentage of clear
corneal incisions used by cataract surgeons increased substantially (say to 40%) the incision type may not be a
significant risk factor for endophthalmitis. Second, a stable, self-sealing incision may be technically more difficult
in the cornea than in the sclera. The key feature of
sutureless cataract surgery is the anterior corneal lip
entrance into the anterior chamber. It is constructed so
that the inner corneal surface seals against the outer

ENDOPHTHALMITIS AFTER CATARACT SURGERY

303

corneal surface once the eye has been repressurized. Consequently, the sutureless scleral tunnel incision is also
known as the scleral tunnel incision with internal corneal
lip. A clear corneal incision creates an internal corneal lip
without the scleral component. Two studies in cadaver
models of cataract surgery offer information on the relative
strength of these two incision types. A study by Ernest and
associates10 demonstrated that a square (3.2 3.2 mm)
scleral corneal incision appears to offer greater stability and
safety than the conventional rectangular (3.2 2.0 mm)
clear corneal incision. Mackool and Russell11 reported that
clear corneal incisions at least 2.0 mm in length had
substantially greater resistance to incision failure than
shorter incision lengths. This suggests that the integrity of
a self-sealing incision depends to some extent on length.
This may be more difficult in a clear corneal incision. If the
incision is too short, the cataract wound may be susceptible to a postoperative perturbation (such as rubbing of the
eye) and wound abnormality.
We did not find the presence or absence of a suture to
have any effect on the risk for endophthalmitis. A previously reported study on 28 cadaver eyes concluded that
both sutured and unsutured cataract incisions resist bacterial growth equally and that a properly constructed wound
is not a significant valve for bacterial inoculation in an eye
pressurized to physiologic conditions.12
This small, retrospective case-control study has many
limitations. Cataract surgery is varied and continually
changing among cataract surgeons in any community, and
the surgeons in our community may or may not be
representative of the universe of surgeons. As is the case
with any retrospective study, we do not know what factors
influenced surgeons to choose on incision type over the
other for specific patients, nor do we know the extent to
which these factors (confounders) may predispose a patient
to endophthalmitis, independent of incision type. Preoperative, intraoperative, and postoperative management is
different among surgeons and was not evaluated in this
study. The use of Monte Carlo modeling to assess the
potential influence of control group selection provides
some indication that our conclusions are not based solely
on the selection of a particular control group, but does not
attenuate any of the other concerns inherent to a retrospective, case-control study. For these reasons and many
others that are well known to the clinical research community, results and conclusions from retrospective casecontrol studies must be interpreted cautiously, and the
finding that a clear corneal incision is a risk factor for
endophthalmitis is not necessarily an indictment of this
particular technique in cataract surgery. However, this
study is strengthened by the fact that researchers in other
communities are noticing a similar trend. Similar findings
from an entirely different population were presented at the
2001 American Academy of Ophthalmology meeting by
Mary Lawrence, MD (personal communication).
304

AMERICAN JOURNAL

There is an increasing preference for clear corneal

incisions over scleral tunnel incisions. In a survey published by the American Society of Cataract and Refractive
Surgery in June 2001, 47% of respondents used clear
corneal incisions. A review of the same survey for previous
years reveals the following trend: 1.5% in 1992, 12.4% in
1995, 23% in 1996, 30% in 1998, and 40% in 1999.13 We
would welcome a large series from experienced surgeons
reporting the incidence of postcataract endophthalmitis
after clear corneal incision as a valuable addition to the
literature and important to establishing the safety profile of
this technique. Surgeons promoting this technique are
duty-bound to not only report positive results in terms of
rapid visual rehabilitation and astigmatism control, but
also complications such as endophthalmitis. In addition,
cataract surgery centers should document the incidence of
and risk factors for endophthalmitis as part of quality
control. Such baseline data are important not only for
comparison with other studies but also for comparison with
future changes in the continuing evolution of cataract
surgery.14
In conclusion, with the development of new surgical
techniques the risk:benefit ratio with regard to complications and eventual visual outcome is always of utmost
importance. The surgeon should consider the advantages
and disadvantages of where the incision is placed and if a
suture should be used to close the wound. Regardless of
incision type, great care and skill must be taken because a
poorly constructed wound may be the strongest determinant of risk for postcataract surgery endophthalmitis.
ACKNOWLEDGMENTS

The authors would like to acknowledge Mae Gordon for

her singular expertise and assistance with the statistical
analysis.

REFERENCES
1. Javitt JC, Vitale S, Canner JK, et al. National outcomes of
cataract extraction: endophthalmitis following inpatient surgery. Arch Ophthalmol 1991;109:10851089.
2. Norregard JC, Thoning H, Bernth-Pettersen P, et al. Risk of
endophthalmitis after cataract extraction: results from the
International Cataract Surgery Outcomes Study. Br J Ophthalmol 1997;81:102106.
3. Colleaux KM, Hamilton WK. Effect of prophylactic antibiotics and incision type on the incidence of endophthalmitis
after cataract surgery. Can J Ophthalmol 2000;35:373378.
4. Rubenstein RY. Simulation and the Monte Carlo method.
New York: Wiley, 1981.
5. Hosmer DW Jr, Lemeshow S. Applied logistic regression.
New York: John Wiley & Sons, Inc, 1989.
6. SAS Institute, Inc. PROC LOGISTIC. In SAS OnLine Doc.
Version 8. Cary, NC: SAS Institute, Inc., 1999.
7. Aaberg TM, Flynn HW, Schiffman J, Newton J. Nosocomial
acute-onset postoperative endophthalmitis survey. A 10-year
review of incidence and outcomes. Ophthalmology 1998;
105:1004 1010.
OF

OPHTHALMOLOGY

AUGUST 2003

8. Montan PG, Koranyi G, Setterquist HE, Stridh A, Philipson

BT, Wiklund K. Endophthalmitis after cataract surgery: risk
factors relative to technique and events of the operation and
patient history: a retrospective case-control study. Ophthalmology 1998;105:21712177.
9. Maxwell DP, Diamond JG. Infectious endophthalmitis following sutureless cataract surgery [Letter]. Arch Ophthalmol
1992;110:915.
10. Ernest PH, Lavery KT, Kiessling LA. Relative strength of
scleral corneal and clear corneal incisions constructed in
cadaver eyes. J Cataract Refract Surg 1994;20:626 629.

VOL. 136, NO. 2

11. Mackool RJ, Russel RS. Strength of clear corneal incisions in

cadaver eyes. J Cataract Refract Surg 1996;22:721725.
12. Turkalj JW, Carlson AN, Moanos JP, Apple DJ. Is the
sutureless cataract incision a valve for bacterial inoculation?
J Cataract Refract Surg 1995;21:472476.
13. Leaming DW. Practice styles and preferences of ASCRS
members2000 survey. J Cataract Refract Surg 2001;27:956
960.
14. Bohigian GM. A study of the incidence of culture-positive
endophthalmitis after cataract surgery in an ambulatory care
center. Ophthalmic Surg Lasers 1999;30:295298.

ENDOPHTHALMITIS AFTER CATARACT SURGERY

305