Keywords
perioperative corneal injury - corneal abrasion - exposure keratopathy - quality improvement
initiative
Corneal injury is the most common perioperative ocular injury.[1] While data on perioperative corneal injury awareness is highly published in the
anesthesia literature, there are few reports in the ophthalmology literature. The
American Society of Anesthesiologists (ASA) Closed Claims Project reported that, prior
to 1990, eye injuries were responsible for 3% of all claims and, of these, 35% were
corneal injuries.[2] A 2014 ASA publication reports that the incidence of perioperative corneal injury
declined from 31% before 1995 to 18% in the period 1995 to 2011.[2]
[3]
The most common mechanisms of perioperative corneal injury include direct trauma to
the corneal epithelium (corneal abrasion), corneal drying (exposure keratitis), and
chemical exposure (toxic keratopathy).[2] The eye may be injured by direct trauma either from tape used to close the eyelids,
the face mask, the anesthesia provider's watch strap, name badge, or laryngoscope
during intubation, drapes during surgical preparation, or the pulse oximeter probe
upon patient emergence from anesthesia.[2]
[4]
[5] Alcohol and chlorhexidine-containing surgical preparation solutions have been shown
to cause corneal epithelial, as well as endothelial, disruption.[6] Anesthesia itself also poses a direct threat to the cornea, as it decreases the
production and stability of tears, abolishes the normal blink reflex, and abolishes
the normal Bell's phenomenon that innately protects the cornea during sleep.[3] Patients with anatomical variations such as lagophthalmos or exophthalmos are at
increased risk of perioperative corneal injuries.[1]
[3] Additional reported risk factors for perioperative corneal injury include prone
and lateral positioning, head and neck surgery, prolonged surgery (> 90 minutes),
early practitioner training level, presence of diabetes mellitus or thyroid disease,
intraoperative sustained hypotension, and anemia.[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
A computerized search of the PubMed database reveals four academic institutions that
have published in the anesthesia literature the incidence of perioperative corneal
injury at their sites. The incidence of perioperative corneal injury was reported
to be 0.15% in 2005 and 0.33% between 2011 and 2017 at the Mayo Clinic (of note, the
reporting strategy differed between the two studies),[1]
[14] 0.034% between 1988 and 1992 at the University of Chicago,[15] 0.12% in 2012 at the University of Alabama (UAB),[16] and 0.07% between 2011 and 2013 at New York University.[13] The Mayo Clinic and UAB anesthesia departments each performed a quality improvement
(QI) corneal abrasion prevention program; incidence of perioperative corneal injury
decreased from 0.15 to 0.047% at Mayo and from 0.12 to 0.009% at UAB.[16] No such reports exist in the ophthalmology literature, with primarily anesthesiologists
taking the lead to develop their own corneal injury prevention and management plans.
The purpose of the current study was to investigate the impact of an ophthalmology
resident-led QI initiative aimed to decrease the incidence of perioperative corneal
injury at an academic medical center.
Methods
The study was granted an exemption from the Institutional Review Board of the Penn
State Health Milton S. Hershey Medical Center. A retrospective chart review was conducted
of all surgical cases performed during 6 months prior to, and 6 months after, implementation
of an ophthalmology resident-led QI initiative at Penn State Milton S. Hershey Medical
Center. The QI initiative consisted of a lecture and distribution of educational materials
to anesthesia providers based on a Mayo Clinic Model.[1] The lecture focused on perioperative corneal injury awareness, understanding of
risk factors, and presentation of an algorithm designed to prevent perioperative corneal
injury. Anesthesia providers were instructed by the ophthalmology chief resident how
and when to tape the eyes shut during anesthesia induction, with emphasis on ensuring
full eyelid closure over the globe to avoid corneal drying. Use of eye ointment was
recommended for those cases involving previously reported risk factors for corneal
injury, including operative times > 90 minutes, head and neck cases, prone or lateral
patient positioning, or the presence of proptosis or lagophthalmos. Data collected
through the chart review included type of surgical case, presence of diabetes mellitus
or thyroid disease (which are risk factors for preexisting corneal pathology), patient
age, gender, patient positioning (supine, prone, or lateral), level of primary anesthesia
provider training (all cases were supervised by an attending anesthesiologist) (resident
in postgraduate year [PGY] 1, 2, 3, or 4, certified registered nurse anesthetist [CRNA]),
or fellow), length of surgical case in minutes, surgical service, type of anesthesia,
and type (if any) of perioperative eye injury.
Statistical Methods
All variables were summarized prior to analysis to check for errors and assess their
distributions. Duration of surgery was grouped into durations of < 90, 90 to 180,
or > 180 minutes based on the model by Batra and Bali.[11] Logistic regression was performed to evaluate potential risk factors for perioperative
corneal injury including several patient and surgical factors: age (< 25, 25–< 50,
50–< 65, and ≥ 65 years), gender, type of anesthesia (general anesthesia vs. other),
type of surgery (head and neck surgery vs. other), surgical service (general surgery
vs. other subspecialty services), diabetes mellitus, thyroid disease, patient positioning
(supine, prone, or lateral), level of anesthesia provider training (PGY 1, 2, 3, or
4, CRNA, fellow), length of case in minutes, and type (if any) of perioperative corneal
injury (abrasion, exposure, vs. toxic). This analysis of potential predictors of perioperative
corneal injury was adjusted for the phase of initiative by including it as a covariate
in the logistic regression model. The incidence of perioperative corneal injury was
compared pre- versus post-QI initiative using logistic regression and was adjusted
for age and gender in addition to the significant variables from the analysis of potential
risk factors: general surgery, patient position, and duration of surgery. Odds ratios
(ORs) and 95% confidence intervals (CIs) are used to quantify the magnitude and direction
of significant associations for the logistic regression analyses. Statistical significance
was set at 0.05. All analyses were performed using SAS software version 9.4 (SAS Institute,
Cary, NC).
Results
Effectiveness of QI Initiative
The incidence of perioperative corneal injury decreased from 0.37% (36 of 9,745 cases)
pre-initiative to 0.19% (19 of 9,991 cases) post-initiative (p = 0.012). There was no significant difference pre- vs. post-initiative in the proportion
of corneal injury cases resulting from abrasion (vs. exposure) (p = 0.909; [Table 1]). Of note, there were no reported cases of toxic keratopathy in either the pre-
or post-initiative phases.
Table 1
Pre- and post-initiative corneal injury data comparison
|
Pre-initiative
|
Post-initiative
|
Odds ratio (95% CI)
|
p-Value
|
|
Rate of perioperative corneal injury after nonocular surgery[a]
|
0.37%
|
0.19%
|
0.49 (0.28–0.85)
|
0.012
|
|
Proportion of perioperative corneal injury cases resulting from abrasion (vs. exposure)
|
53.66%
|
52.17%
|
0.94 (0.34–2.62)
|
0.909
|
Abbreviation: CI, confidence interval.
Note: Odds ratios and p-values from logistic regression.
a Adjusted for age, gender, general surgery, patient position, and duration of surgery.
Assessment of Risk Factors for Perioperative Corneal Injury
Analysis of risk factors, with consideration of initiative phase as a covariate, demonstrated
an increased risk of perioperative corneal injury associated with non-head and neck
cases (OR = 0.32, 95% CI 0.11–0.87; p = 0.026), supine compared with lateral patient positioning (OR = 0.13, 95% CI 0.07–0.23;
p = 0.006), prone compared with lateral positioning (OR = 0.25, 95% CI 0.09–0.67; p < 0.001), surgery performed by the general surgery service (OR 6.50, 95% CI 2.39–24.76,
p < 0.001), and longer surgical time. A significantly higher incidence of perioperative
corneal injury was observed when comparing cases of 90 to 180 minutes' duration to
cases of < 90 minutes' duration (OR = 4.18, 95% CI 1.43–12.18; p = 0.009), > 180 minutes to < 90 minutes (OR = 8.56, 95% CI 3.01–24.32; p < 0.001), and > 180 minutes to 90 to 180 minutes (OR = 2.05, 95% CI 1.17–3.58; p < 0.001; [Table 2]). There was no increased risk of corneal injury with regard to the following factors:
age, female versus male gender (OR = 0.94, 95% CI 0.55–1.59, p = 0.809), general anesthesia versus regional/conscious sedation/other (OR = 0.77,
95% CI 0.35–2.02, p = 0.639), diagnosis of diabetes mellitus (OR = 0.98, 95% CI 0.42–2.30; p = 0.967) or thyroid disease (OR = 1.72, 95% CI 0.78–3.81; p = 0.182), and level of the anesthesia provider ([Table 2]).
Table 2
Risk factor analysis for pre- and post-initiative cases of perioperative corneal injury
after nonocular surgery
|
Control cases (n = 20,187)
|
Injury cases (n = 55)
|
Odds ratio (95% CI)
|
p-Value
|
|
Age (y)
|
|
< 25
|
4,921 (99.9%)
|
7 (0.1%)
|
0.55 (0.18–1.50)
|
0.286
|
|
25–< 50
|
5,334 (99.7%)
|
18 (0.3%)
|
1.33 (0.61–3.03)
|
0.561
|
|
50–< 65
|
4,991 (99.6%)
|
18 (0.4%)
|
1.41 (0.64–3.21)
|
0.460
|
|
≥ 65
|
4,691 (99.7%)
|
12 (0.3%)
|
1.0
|
|
|
Gender
|
|
Female
|
10,289 (99.7%)
|
27 (0.3%)
|
0.94 (0.55–1.59)
|
0.809
|
|
Male
|
9,898 (99.7%)
|
28 (0.3%)
|
1.0
|
|
|
Diabetes mellitus
|
|
Yes
|
2,255 (99.7%)
|
6 (0.3%)
|
0.98 (0.42–2.30)
|
0.967
|
|
No
|
17,932 (99.7%)
|
49 (0.3%)
|
1.0
|
|
|
Thyroid disease
|
|
Yes
|
1,579 (99.6%)
|
7 (0.4%)
|
1.72 (0.78–3. 81)
|
0.182
|
|
No
|
18,608 (99.7%)
|
48 (0.3%)
|
1.0
|
|
|
General surgery
|
|
Yes
|
13,403 (99.6%)
|
51 (0.4%)
|
6.50 (2.39–24.76)
|
< 0.001
|
|
No
|
6,784(99.9%)
|
4 (0.1%)
|
1.0
|
|
|
General anesthesia
|
|
Yes
|
17,905 (99.8%)
|
48 (0.3%)
|
0.77 (0.35–2.02)
|
0.639
|
|
No
|
2,023 (99.7%)
|
7 (0.3%)
|
1.0
|
|
|
Head and neck case
|
|
Yes
|
4,008 (99.9%)
|
4 (0.1%)
|
0.32 (0.11–0.87)
|
0.026
|
|
No
|
16,179 (99.7%)
|
51 (0.3%)
|
1.0
|
|
|
Positioning[a]
|
|
Lateral
|
1,331 (98.6%)
|
19 (1.4%)
|
1.0
|
|
|
Prone
|
1,393 (99.6%)
|
5 (0.4%)
|
0.25 (0.09–0.67)
|
0.006
|
|
Supine
|
16,998 (99.8%)
|
31 (0.2%)
|
0.13 (0.07–0.23)
|
< 0.001
|
|
Anesthesia provider[b]
|
|
PGY1
|
2,606 (99.6%)
|
11 (0.4%)
|
1.67 (0.86–3.23)
|
0.131
|
|
PGY2
|
5,423 (99.8%)
|
9 (0.2%)
|
0.54 (0.26–1.10)
|
0.090
|
|
PGY3
|
4,392 (99.7%)
|
15 (0.3%)
|
1.32 (0.73–2.40)
|
0.357
|
|
PGY4
|
2,185 (99.8%)
|
5 (0.2%)
|
0.81 (0.32–2.02)
|
0.644
|
|
CRNA
|
5,213 (99.7%)
|
15 (0.3%)
|
1.13 (0.63–2.05)
|
0.681
|
|
Fellow
|
458 (100.0%)
|
0 (0.0%)
|
N/A
|
0.283
|
|
Duration[c]
|
|
< 90 min
|
6,579 (99.9%)
|
4 (0.1%)
|
1.0
|
|
|
90–180 min
|
8,043 (99.7%)
|
21 (0.3%)
|
4.18 (1.43–12.18)
|
0.009
|
|
> 180 min
|
5,523 (99.5%)
|
30 (0.5%)
|
8.56 (3.01–24.32)
|
< 0.001
|
Abbreviations: CI, confidence interval; CRNA, certified registered nurse anesthetist;
OR, odds ratio; PGY, postgraduate year.
Note: Odds ratios and p-values from a logistic regression adjusted for the phase of the initiative. Exact
logistic regression used as needed.
a OR for prone versus supine is 1.95 (0.77, 5.02), p = 0.167.
b Comparison of each provider level individually to all other levels combined to determine
if there was a significant difference between groups.
c OR for > 180 minutes versus 90–180 minutes is 2.05 (1.17–3.58), p < 0.001.
The general surgery service was then reviewed for the incidence of corneal injury
in each separate general surgery category (colorectal surgery, cardiac surgery, emergency
general surgery, general surgery, neurosurgery, orthopedics, plastic surgery, minimally
invasive surgery, surgical oncology, transplant surgery, trauma surgery, urology,
vascular surgery, and pediatric general surgery); there was a significantly higher
incidence and odds of corneal injury in the vascular surgery group versus the other
groups even after adjustment for other significant risk factors for perioperative
corneal eye injury such as surgical time and positioning (1.10% vs. 0.31%, OR = 3.70,
CI 1.57–8.73, p = 0.003).
Discussion
In the current study, an ophthalmology resident-led educational initiative for anesthesia
providers on perioperative corneal injury awareness, risk factors of perioperative
corneal injury, and strategies to prevent perioperative corneal injury was associated
with a significant decrease in the incidence of perioperative corneal injury ([Fig. 1], [Table 1]). Similar to previous studies that evaluated risk factors for perioperative corneal
injury, we found that longer duration of surgery poses an increased risk for corneal
injury.[4]
[17] This is likely due to the additional time during which the patient is exposed to
such effects of anesthesia as decreased tear film production and absence of the Bell's
phenomenon.[4]
[5] We did not find a time point at which that risk plateaued; rather, with each 90-minute
unit increase in length of surgery, the risk of injury also increased. In our study,
there was no significant difference in injury rates among the various levels of anesthesia
providers. A previous study showed an increased incidence of perioperative corneal
injury with student nurse anesthetists compared with residents or CRNAs[1]; however, our institution does not have such trainees and we found no increased
risk associated with earlier trainee level of resident physician.
Fig. 1 Decreased incidence of perioperative corneal injury after ophthalmology resident-led
quality improvement initiative.
As in previous studies, we found an increased risk of perioperative corneal injury
in patients who were placed in the lateral or prone positions compared with the supine
position.[4]
[12]
[14] To place a patient in a lateral or prone position often requires rolling the patient
after anesthetics have been administered. The process of moving the patient and location
of the face in a compromising position may be why lateral and prone positioning demonstrate
an increased risk. In contrast to other published reports,[1]
[15] our study identified an increased risk of perioperative corneal injury in nonhead
and neck cases. A potential explanation for this increased risk may be a decreased
awareness of contact with the patient's face, including the eye, if the eye is not
in direct view during the procedure. Also, in contrast to a previous report,[1] our study did not identify thyroid disease as a significant predictor of perioperative
corneal injury. This may be due, at least in part, to the fact that the previously
published study[1] isolated Graves' disease as a specific diagnosis. Due to the nature of our electronic
medical records database, we were unable to distinguish Graves' disease from other
thyroid diagnoses.
To our knowledge, and based on a computerized search of the PubMed database, one other
study looked for a difference in risk of corneal injury among surgical services. The
latter study, conducted at New York University, did not find a significant difference.[13] In contrast, our study identified a significantly higher rate of perioperative corneal
injury associated with vascular surgery compared with other surgical services. Although
the vascular surgery group had a significantly higher proportion of patients with
such previously identified risk factors for corneal injury such as supine position
and longer surgical time, vascular surgery had an independently significant effect
on corneal injury risk even after adjustment for the previously identified risk factors.
Further study of this finding is warranted.
One limitation of the current study is the retrospective nature of the chart review.
However, this applied equally to the pre- and post-initiative phases. Another limitation
of our study is the relatively short (6 months) follow-up period of post-initiative
data collection. The study from Mayo Clinic initially included 16 months of follow-up
post-initiative and then increased follow-up to 31 months, and UAB included a follow-up
of 45 months post-initiative.[1]
[16] Our study was designed to fit within one training year to allow for a resident-run
initiative. Further research is warranted to investigate longer-term outcomes of an
ophthalmology resident-led program designed to decrease the incidence of perioperative
corneal injury.
In sum, our findings show that an ophthalmology resident-led educational initiative
to anesthesiology providers was associated with a significant decrease in the incidence
of perioperative corneal injury. We hope that this QI initiative may inspire other
ophthalmology practices to take the lead in perioperative corneal injury awareness
and prevention.
Conclusion
An ophthalmology resident-led educational initiative to anesthesiology providers was
associated with a significant decrease in the incidence of perioperative corneal injury.
