Microbial Keratitis Isolates at a Midwestern Tertiary Eye Care Center

Abstract

Purpose:

To describe the pathogen, antimicrobial susceptibility, and trends over time of microbial keratitis (MK) at a Midwestern tertiary eye center.

Methods:

MK patients were identified in the electronic health record from August 2012 to December 2021. Diagnostic laboratory tests with an MK diagnosis were identified and classified as lab-positive or lab-negative. Lab-positive infections were categorized as bacterial (gram-positive, gram-negative, acid-fast bacilli [AFB]), fungal, viral, Acanthamoeba, or polymicrobial. Antimicrobial susceptibilities were obtained. Trends over time were assessed using linear regression.

Results:

Of 3288 MK patients identified, 1012 (30.8%) had labs performed. Lab-positive infections (n=499, 49.3%) were bacterial in 73.5% (n=367) of cases, fungal in 7.8% (n=39), viral in 1.6% (n=8), Acanthamoeba in 1.4% (n=7), and polymicrobial in 15.6% (n=78). Of bacterial infections, 70% (n=257) were gram-positive with coagulase-negative Staphylococcus (CoNS; 31%) and Staphylococcus aureus ( S. aureus; 23%) as the most common pathogens. Bacteria were AFB in 1.9% (n=7) of cases and gram-negative in 28.1% (n=103) with Pseudomonas aeruginosa as the predominant pathogen (47.7%). S. aureus showed antibiotic resistance from 0% (vancomycin, gentamicin) to 50% (erythromycin); CoNS from 0% (vancomycin, gentamicin, moxifloxacin) to 64% (erythromycin). The rate of lab-negative MK significantly increased over time (slope estimate=2.1% per year, p=0.034). Rates of bacterial, fungal, viral, Acanthamoeba, and polymicrobial infections were stable over time (all slope p>0.05).

Conclusion:

Bacterial keratitis accounted for most MK cases. Gram-positive bacteria were the most common isolates. CoNS and S. aureus were universally susceptible to vancomycin. Rates of MK infection types were stable over time.

INTRODUCTION

Microbial keratitis (MK) is a vision threatening infection of the cornea that is one of the leading causes of blindness in the world.¹ While the global incidence of MK is estimated at 1.5–2 million cases annually, this is likely an underestimate of the true incidence.¹ A Center for Disease Control report from 2010 estimated that 930,000 outpatient visits and 58,000 emergency department visits occurred annually in the United States (US) for keratitis.² Studies have estimated a MK incidence rate between 11 and 20 per 100,000 people and up to 200 per 100,000 in contact lens wearers; However, these studies were limited to specific geographic locations in the US.^3–5

MK can be caused by several pathogens including bacteria, fungi, viruses, or amebic agents (namely Acanthamoeba species). The lack of epidemiological data and geographic variability creates uncertainty for clinicians as to the likely organisms causing MK infections in regional communities. The primary pathogens causing MK can vary depending on geographic region, environment, climate, and other factors such as contact lens wear.^6,7 MK caused by gram-positive bacteria has been shown to be more prevalent in temperate climates, whereas gram-negative bacteria, fungi, and Acanthamoeba are more prevalent in tropical climates or warmer temperatures.^7–11 Data on rates of MK infections by pathogen within geographic regions serve to inform clinicians and researchers of initial medication management, to compare populations from different regions, or to detect shifts in infection types within a community.

There is a need to understand the current pathogens causing MK in the areas of the US with limited epidemiologic data. Historical etiologic data from the Midwestern US exists for Michigan from 1975 to 1981¹² and Missouri from 1999 to 2013.¹³ Studies have also reported single species outbreaks for fungal organisms, Acanthamoeba, and herpes simplex virus occurring in the Midwest.^14–16 However, the lack of data from the past ten years represents a gap in knowledge. This study will allow an opportunity to compare trends over time and between regions. Also, none of these Midwestern studies have analyzed differences between infection types related to patient demographics, which helps researchers understand if patient factors, such as age, race, and ethnicity, play an important role in disease manifestation. This study describes the causative organisms of MK infections, etiologic trends over time, antimicrobial susceptibility patterns, and the relationship between MK organisms and patient demographics in the last decade at a tertiary medical center in the Midwestern United States.

METHODS

This study was approved by the Institutional Review Board of the University of Michigan (UM) and adheres to the tenets of the Declaration of Helsinki. A retrospective review of the electronic health record (EHR) of patients with MK at the UM, a tertiary care hospital with a referral-based outpatient ophthalmology department, was conducted from August 2012 to December 2021. Patients were identified from the UM EHR by International Classification of Diseases 9^th or 10^th revision codes (ICD-9 or ICD-10) related to MK that were associated with an office visit encounter in the UM ophthalmology department (see Supplemental Table 1 for full diagnostic codes). The first date of MK diagnosis was identified for each patient, and only the first infection in patients with multiple infections over time or those with infections in both eyes was included in analysis.

Laboratory tests for patients with MK were also obtained from the EHR. To ensure lab tests were from the targeted encounter, lab tests were limited to those from 7 days before to 90 days after the first date of MK diagnosis. If multiple cultures were taken within the period, all positive cultures were aggregated to determine the final lab result. For example, if 3 cultures were taken and a particular strain of Staphylococcus aureus (S. aureus) grew twice and acanthamoeba grew once, we reported one isolate of S. aureus and one of acanthamoeba. Tests with specimens that were not collected from various eye or eye-related sources were excluded. Specimens included samples collected directly from the cornea and from foreign bodies such as contact lenses, cases, solution, corneal sutures, glues, filaments, grafts, etc. (see Supplemental Table 2 for full specimen types). Specimen collection from 2012 to June 2018 was performed by techniques at the clinicians’ discretion, but typically involved use of spatulas or blades. Starting in July 2018, specimens were collected and transported to the laboratory using COPAN Diagnostics ESwab’s (COPAN Diagnostics, Murrieta, CA, USA). The lab tests for bacterial infections used Gram stain and Acid Fast stain from the collected patient specimen, followed by culture protocols specific for bacterial or mycobacterial organisms. Anaerobic cultures were only performed when ordered specifically by the managing clinician. Test for fungal infections used calcofluor fungal stain from the collected patient specimen, followed by culture protocols specific for fungal organisms. Testing for viral pathogens (HSV,VZV) was performed by polymerase chain reaction in house. Of note, viral testing was infrequently performed only when the clinician had a supposition for viral etiology in complex cases. Testing for Acanthamoeba was performed by polymerase chain reaction as a send-out test.

Laboratory results were determined by searching within the comments for terms associated with any quantity of any isolates. Negated terms and non-pathogenic findings such as polymorphonuclear leukocytes, mononuclear leukocytes, squamous epithelial cells, and skin flora were excluded. Laboratory testing was considered positive when any isolates were found. A chart review of 135 randomly selected patients showed this method had 99% accuracy in determining lab results using this method.

Lab-positive infections from any specimen type related to the eye culture were categorized into five organism classes: bacterial, fungal, viral, Acanthamoeba, or polymicrobial. Bacterial infections were further classified as gram-positive, gram-negative, or acid-fast bacilli (AFB). Polymicrobial infections were defined as infections caused by any combination of bacterial, fungal, viral, or Acanthamoeba organisms or by two or more organisms of different bacterial classes (e.g., gram-positive and gram-negative). Bacterial infections with two or more organisms of the same bacterial class were classified as that respective category (e.g., infections with two different gram-positive isolates were considered gram-positive, not polymicrobial). Fungal and viral infections with two or more isolates of the same class were also categorized as single class infections, not polymicrobial. This decision was based on a treatment perspective, as organisms of the same class are often treated with the same antimicrobials. Any lab test returning no growth or inconclusive results was classified as a lab-negative infection.

Antibiotic susceptibilities were obtained through chart review for the three most common isolates, Coagulase-negative Staphylococcus (CoNS), Staphylococcus aureus (S. aureus), and Pseudomonas Aeruginosa (P. aeruginosa). Susceptibility testing for S. aureus and P. aeruginosa is performed for any isolate recovered; susceptibility testing for CoNS is only performed when requested by the clinician or when a highly pathogenic organism, as determined by the UM laboratory, is isolated. Per protocol, CoNS and S. aureus were tested for susceptibility to methicillin, trimethoprim/sulfamethoxazole, vancomycin, erythromycin, ciprofloxacin, moxifloxacin, and gentamicin. P. aeruginosa was tested for susceptibility to tobramycin, gentamicin, ciprofloxacin, amikacin, and ceftazidime. Susceptibility results were reported as treatment susceptible, moderate, or resistant per the UM Clinical Microbiology Laboratory protocol.

Other data obtained from the UM EHR included patient demographics (age, gender, race, ethnicity) and Area Deprivation Index (ADI). ADI is a measure of socioeconomic disadvantage within a neighborhood and is ranked nationally to provide percentiles from 1 to 100, with higher values indicating higher levels of disadvantage or deprivation.¹⁷ ADI was obtained for each patient by linking their 9-digit zip code to ADI values from 5-year American Community Survey estimates (2009–2013).

Statistical Methods

This analysis focuses on the subset of MK patients identified who also had laboratory testing performed. Patient demographics and MK isolates were summarized with descriptive statistics including mean and standard deviation (SD) for continuous measures and frequencies and percentages for categorical measures. Demographics were compared between lab-positive versus lab-negative patients and between organism classes for lab-positive patients using two-sample t-tests or Kruskal-Wallis tests for continuous measures and Chi-square or Fisher exact tests for categorical measures. A p-value <0.05 was considered statistically significant. Linear regression and line plots were used to examine the trends of patients with laboratory testing over time from 2013 to 2021. Data from 2012 were excluded due to incomplete data as the EHR started mid-year. Bar charts were used to display the antibiotic susceptibility patterns for S. aureus, CoNS, and P. aeruginosa. All analysis was performed using the statistical programming language R (R version 4.1.1; Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Patient Demographics

A total of 3,288 MK patients were identified, including 1,012 (30.8%) who had labs performed. Descriptive statistics of the sample of MK patients who had labs performed are summarized in Table 1. Patients were on average 56.5 years old at diagnosis (SD=20.2), 55.7% (n=564) were female, 44.3% (n=448) male, 88.1% were White, 7.3% Black, 2.4% Asian, and 2.1% other races, and 2.7% were Hispanic. Approximately half of patients who had labs performed were lab-positive (n=499, 49.3%). Patients who were lab-positive were slightly older than those who were lab-negative (mean=57.8 years, SD=20.4 versus mean=55.2, SD=20.0, respectively; p=0.04). There were no significant differences between lab-positive and lab-negative patients with respect to gender, ethnicity, race, and ADI.

Table 1.

Patient demographics of those who received labs.

	Overall (n=1012)		Lab-positive (n=499)		Lab-negative (n=513)
Continuous Variable	Mean (SD)	Min, Max	Mean (SD)	Min, Max	Mean (SD)	Min, Max	P-value*
Age (years)	56.5 (20.2)	0.1, 99.0	57.8 (20.4)	0.1, 95.5	55.2 (20.0)	2.5, 99.0	0.040
Area Deprivation Index	54.4 (24.0)	2.0, 100.0	55.7 (23.9)	2.0, 100.0	53.1 (24.0)	2.0, 100.0	0.092
Categorical Variable	Frequency (%)		Frequency (%)		Frequency (%)		P-value*
Gender
Female	564 (55.7)		273 (54.7)		291 (56.7)		0.519
Male	448 (44.3)		226 (45.3)		222 (43.3)
Ethnicity
Hispanic	26 (2.7)		8 (1.7)		18 (3.7)		0.054
Non-Hispanic	941 (97.3)		470 (98.3)		471 (96.3)
Race
White	869 (88.1)		426 (87.1)		443 (89.1)		0.479
Black	72 (7.3)		42 (8.6)		30 (6.0)
Asian	24 (2.4)		11 (2.2)		13 (2.6)
Other	21 (2.1)		10 (2.0)		11 (2.2)

^*T-test for continuous variables and Chi-square test for categorical variables

SD, standard deviation; Min, minimum; Max, maximum

Specimen Isolates

Of the 499 MK patients with laboratory positive results, 73.5% (n=367) of infections were bacterial, 7.8% (n=39) were fungal, 1.4% (n=7) were Acanthamoeba, 1.6% (n=8) were viral, and 15.6% (n=78) were polymicrobial (Table 2). There were no significant differences between MK infection types with respect to age (p=0.26), gender (p=0.85), ethnicity (p=0.54), race (p=0.73), and ADI (p=0.97) (Table 3). Of the patients with bacterial infections, gram-positive bacteria accounted for 70% (n=257) of cases, gram-negative bacteria for 28.1% (n=103), and AFB for 1.9% (n=7).

Table 2.

Organisms identified for patients that were lab-positive (n=499).

Type of keratitis	Frequency (%)
	Overall (Column %)	Subgroup (Column %)
Bacterial	367 (73.5)
Gram Positive		257 (70.0)
Gram Negative		103 (28.1)
Acid Fast Bacilli		7 (1.9)
Fungal	39 (7.8)
Acanthamoeba	7 (1.4)
Viral	8 (1.6)
Polymicrobial	78 (15.6)
Bacterial+		45 (57.7)
Gram Positive + Gram Negative		40 (51.3)
Gram Positive + Acid Fast Bacilli		4 (5.1)
Gram Negative + Acid Fast Bacilli		1 (1.3)
Bacterial and Fungal		28 (35.9)
Gram Positive + Gram Negative + Fungal		7 (9.0)
Gram Positive + Fungal		14 (17.9)
Gram Negative + Fungal		5 (6.4)
Gram Positive + Acid Fast Bacilli + Fungal		2 (2.6)
Bacterial and Acanthamoeba		3 (3.9)
Gram Positive + Gram Negative + Acanthamoeba		1 (1.3)
Gram Positive + Acanthamoeba		2 (2.6)
Bacterial (Gram Positive) and Viral		1 (1.3)
Fungal and Acanthamoeba		1 (1.3)

Table 3.

Patient demographics for those with lab-positive results, stratified by organism classes.

	Bacterial (n=367)	Fungal (n=39)	Acanthamoeba (n=7)	Viral (n=8)	Polymicrobial (n=78)
Continuous Variable	Mean (SD)					P-value*
Age (years)	58.8 (20.7)	54.4 (14.6)	46.7 (16.6)	49.4 (18.1)	56.4 (21.7)	0.256
Area Deprivation Index	55.5 (24.1)	56.4 (22.0)	57.3 (23.6)	52.6 (24.9)	56.4 (24.3)	0.974
Categorical Variable	Frequency (%)					P-value**
Gender
Female	198 (54.0)	23 (59.0)	4 (57.1)	5 (62.5)	43 (55.1)	0.853
Male	169 (46.0)	16 (41.0)	3 (42.9)	3 (37.5)	35 (44.9)
Ethnicity
Hispanic	8 (2.3)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0.537
Non-Hispanic	345 (97.7)	38 (100.0)	6 (100.0)	7 (100.0)	74 (100.0)
Race
White	312 (86.2)	37 (97.4)	6 (100.0)	4 (57.1)	67 (88.2)	0.729
Black	34 (9.4)	1 (2.6)	0 (0.0)	2 (28.6)	5 (6.6)
Asian	9 (2.5)	0 (0.0)	0 (0.0)	0 (0.0)	2 (2.6)
Other	7 (1.9)	0 (0.0)	0 (0.0)	1 (14.3)	2 (2.6)

^*Kruskal-Wallis Rank Sum Test (excluding Acanthamoeba and viral category because of small sample size)

^**Fisher exact test with Monte Carlo simulation (excluding Acanthamoeba and viral because of small sample size)

SD, standard deviation.

Of the patients with polymicrobial infections, 57.7% (n=45) had two or more different classes of bacterial isolates, where the most common combination was gram-positive and gram-negative bacteria (n=40). For the remaining polymicrobial cases, 35.9% (n=28) had bacterial and fungal isolates identified, 3.9% (n=3) had bacterial and Acanthamoeba, 1.3% (n=1) had bacterial and viral, and 1.3% (n=1) had fungal and Acanthamoeba. (See Table 2 for a full list of polymicrobial infection combinations). When reclassifying the MK types to include polymicrobial infections into their respective main infection types, 89.0% (n=444) of the total 499 patients had bacterial infections, 13.6% (n=68) had fungal infections, 2.2% (n=11) had Acanthamoeba infections, and 1.8% (n=9) had viral infections.

The most common gram-positive isolates were CoNS (30.9%, n=109), S. aureus (23.2%, n=82), Cutibacterium (Propionibacterium) acnes (16.4%, n=58), Corynebacterium species (spp.) (6.5%, n=23), and Streptococcus pneumoniae (5.4%, n=19). The most common gram-negative isolates were P. aeruginosa (47.7%, n=82), Moraxella spp. (11%, n=19), Serratia spp. (7.6%, n=13), Klebsiella spp. (6.4%, n=11), and Stenotrophomonas maltophilia complex (4.7%, n=8). The only AFB isolates were Mycobacterium chelonae (92.9%, n=13) and Mycobacterium avium (7.1%, n=1). Of the fungal isolates, the most common organisms identified were Candida spp. (27%, n=17), Fusarium spp. (17.5%, n=11), Aspergillus spp. (14.3%, n=9), Alternaria spp. (12.7%, n=8), and Paecilomyces spp. (7.9%, n=5). All viral isolates identified were in the herpetic family. Herpes simplex virus was the most common isolate (90%, n=9) followed by varicella zoster virus (10%, n=1). The full listing of the MK isolates is presented in Supplemental Table 3.

Microbial Isolate Trends Over Time

Among those MK patients who had lab testing, Figure 1 displays the trends in frequency and percentage of MK types over time from 2013 to 2021. On average, the frequency of patients receiving lab tests increased by 8.50 patients per year (95% Confidence Interval, CI=6.33 to 10.67; p<0.001), from 70 patients in 2013 to 139 in 2021. Patients with negative lab results increased by 6.85 patients per year (95% CI=4.75 to 8.95; p<0.001) from 35 patients in 2013 to 86 in 2021. Patients with bacterial, polymicrobial, viral, and Acanthamoeba infections also showed non-statistically significant increase over time (bacterial: slope estimate=1.28 patients/year, 95% CI=−0.82 to 3.38, p=0.055; Polymicrobial: estimate=0.3, 95% CI=−1.8 to 2.4, p=0.647; Viral: estimate=0.08, 95 %CI=−2.02 to 2.18, p=0.899; Acanthamoeba: estimate=0.27, 95% CI=−1.83 to 2.37, p=0.684). Patients with fungal infections had a non-significant decrease over time (estimate=−0.28 patients/year, 95% CI=−2.38 to 1.82, p=0.665). However, trends were stable over time for specific MK classes. Percentage wise, the rate of patients receiving lab tests increased by 2.45% patients per year (95% CI=1.64 to 3.25; p<0.001). The rate of patients with lab negative results increased by 2.12% per year (95% CI=0.21 to 4.02; p=0.034). The rates of patients with bacterial, polymicrobial, or fungal infections were stable over time with a non-statistically significant decreases by 1.44% per year (95% CI=−3.45 to 0.57; p=0.133), 0.3% (95% CI=−0.97 to 0.38; p=0.335), and 0.58% (95% CI=−1.27 to 0.11; p=0.086), respectively. The rates of patients with viral or Acanthamoeba infections also were stable over time (viral: estimate=0.01%/year, 95% CI=−0.33 to 0.36, p=0.934; Acanthamoeba: estimate=0.19%/year, 95% CI=−0.02 to 0.39, p=0.067).

Figure 1.

Frequency (A) and percentage (B) of microbial keratitis patients with laboratory positive and negative infections over time (2013–2021).

Antibiotic Susceptibility

Antibiotic susceptibility and resistance profiles of the three most common bacterial isolates S. aureus, CoNS, and P. aeruginosa are shown in Figure 2. Susceptibilities were obtained for 100% (n=82) of S. aureus isolates, 26.6% (n=29 out of 109) of CoNS isolates, and 100% (n=82) of P. aeruginosa isolates. Of the S. aureus isolates, 32% were methicillin resistant. S. aureus exhibited high susceptibility to vancomycin (100%), gentamicin (100%), and trimethoprim/sulfamethoxazole (96%), and reduced susceptibility to moxifloxacin (80%), ciprofloxacin (63%), and erythromycin (50%). Of CoNS isolates, 52% were methicillin resistant. CoNS was universally susceptible to vancomycin, gentamicin, moxifloxacin (all 100%), and had reduced susceptibilities to ciprofloxacin (69% susceptible, 8% moderate), trimethoprim/sulfamethoxazole (69%), and erythromycin (36%). P. aeruginosa exhibited high susceptibility to all tested antibiotics, including ceftazidime (100%), amikacin (98%), tobramycin (98%), gentamicin (95%), and ciprofloxacin (94%).

Figure 2.

Antibiotic susceptibility and resistance profiles for S. aureus, CoNS, and P. aeruginosa. Abbreviation: Trimethoprim/sulfamethoxazole, Trimethoprim/Sulfa.

DISCUSSION

In this study of 1,012 MK patients with laboratory data over the last decade, we analyzed differences between infection types, trends over time, and antibiotic susceptibilities. Of the 499 patients with positive lab results, bacteria accounted for most cases (73.5%), followed by polymicrobial cases (15.6%), fungi (7.8%), viruses (1.6%), and Acanthamoeba (1.4%). The bacterial cases were predominantly caused by gram-positive bacteria (70%), followed by gram-negative bacteria (28.1%) and AFB (1.9%).

Approximately 31% of MK patients identified in the University of Michigan EHR underwent diagnostic lab testing. First, MK billing codes could have been used improperly, potentially over-identifying cases that were not true MK. Second, clinicians may not culture MK if the infection is small or if the process to use cultures is inconvenient or unknown to the clinician. Third, as a tertiary medical center, a significant portion of MK patients are referred from outside clinics. In these cases, diagnostic lab testing may have already been completed, leading to the decision to not repeat testing. Fourth, laboratory data not linked to an ophthalmology office visit that included a diagnostic code of keratitis were not extracted from the EHR.

More patients with MK were identified and more laboratory tests were ordered with each consecutive year of study, except for less testing during the COVID-19 pandemic in 2020, as shown in Figure 1. The significant increase in the frequency of MK patients seen and positive labs likely reflects the growth of our department and subsequent increase in our patient population. The percentage of the different organism classes causing MK infections has remained relatively stable.

Overall, the diagnostic lab testing positivity rate was 49%, consistent with the positivity range of 32–79% reported in other studies of MK.¹⁸ As a tertiary eye care center, patients may come pre-treated with antimicrobials from their local ophthalmologist, primary care physician, urgent care centers, and emergency departments lowering the likelihood of a positive lab result.

Of note, there was a concerning significant increase in the percentage of lab-negative infections that occurred around the time of switching from corneal scrapings via spatulas or blades to using ESwab’s in July 2018. We investigated this decline in positivity rates in more detail. We confirmed the proper procedures for transport were used (and the same as other published reports). We checked the time to transportation and found a slight increase in the transportation times from 4–5 hours from 2015–2018 and 5–6 hours from 2018–2021 to the laboratory. Once the proper procedures were confirmed, there are a few hypotheses to be explore. First, it is possible that patients’ characteristics changed – there were more patients pretreated with antibiotics in more recent years or more wore contact lenses. Second, provider practice patterns could have changed with the ESwab creating greater availability for all faculty to culture any eye with keratitis, even smaller or less suspicious infections. This steep rise in the rate of lab negative results is concerning so it is being explored in a current follow-up study that will include chart reviews of all cases reported in this study for relevant factors that may be associated with lab negative results, such as contact lens wear and prior treatment with antibiotics.

The current study complements other MK epidemiologic and microbiologic studies conducted. A previous study at our institution from 1975 to 1981 reviewed 224 keratitis cases. This study reported bacterial keratitis as the most common type (35%), with gram-positive bacteria isolated in 55% of bacterial cases.¹² Our current data reports higher percentage of gram-positive cases (70%) within the bacterial keratitis lab positive results, compared to the prior cases from our institution. MK etiological data in the US has been reported in Missouri¹³, California^3,19–22 (bacterial only²³), Florida (bacterial only,²⁴ fungal only^25,26), Pennsylvania^27,28 (bacterial only,²⁹ fungal only³⁰), Texas,^31–35 New York³⁶ (fungal only³⁷) North Carolina (bacterial only,³⁸ fungal only³⁹), and Maryland.⁴⁰ The pathogenic distribution of Michigan cases (51.5% gram-positive, 20.6% gram-negative, 15.6% polymicrobial, 7.8% fungal) is within the range from other states: Missouri (48% gram-positive, 34% gram-negative, 16% fungal), the five California studies (59–67% gram-positive, 22–32% gram-negative, 7–9% fungal), the five Texas studies (46–58% gram-positive, 31–39% gram-negative, 5–15% fungal), New York (55–83% gram-positive, 17–45% gram-negative, 1% fungal), and Eastern Pennsylvania (77% bacterial, 14% fungal). CoNS, S. aureus, and P. aeruginosa were the most common bacterial isolates, like the Missouri study, however that study reported higher rates of P. aeruginosa. In our study, some cases were concomitant bacterial and fungal or Acanthamoeba infections so rates of fungal and Acanthamoeba organisms are in fact higher than shown by our classification scheme.

Only 27% of CoNS infections had antimicrobial susceptibility performed as CoNS susceptibilities are not routinely tested by the UM microbiology lab. Only species with high pathogenic risk, as determined by the UM lab, receive susceptibility testing; but ocular pathogenic risk is not considered. Our study shows that 32% of S. aureus and 52% of CoNS were resistant to methicillin, which is similar to the findings in the ARMOR study, a nationwide study surveilling the etiologic and antibiotic resistance patterns of ocular infections (34.9% methicillin resistance in S. aureus and 49.3% in CoNS infections).⁴⁵ Our report demonstrated that S. aureus and CoNS were 100% susceptible to vancomycin, consistent with findings in both the ARMOR and Missouri reports (100% susceptibility in both studies), demonstrating no increase in resistance in the last 4 years in this region. Similarly, our study reported high susceptibility of P. aeruginosa infections to gentamicin, tobramycin, and ciprofloxacin (range 94%–99%), consistent with findings in both the ARMOR (range 93–97%) and Missouri (100%).

In our study, MK patients with positive lab testing were statistically significantly older than those with negative labs. One potential explanation is a weakened immune system for older patients with less clearance of pathogens.⁴¹ Another explanation could be that older patients more readily seek eye care⁴² or have established care with ophthalmologists, thereby streamlining the process to directly see an eye clinician for eye problems and reducing the likelihood of presenting pre-treated with antimicrobials by non-ophthalmic providers. Younger patients, on the other hand, tend to visit their primary care physicians, urgent care clinics, or emergency departments^43,44 before being referred to an ophthalmologist. This would lead to pre-treatment with antimicrobials and therefore a higher likelihood of negative culture results when they present to eye clinics. This study also showed no significant differences in area deprivation (a surrogate for poverty), gender, ethnicity, or race between receipt or non-receipt of a lab and also no difference in these characteristics for lab-positive and lab-negative patients. This is a promising sign that disparities in care did not affect performance of testing or lab positivity in our population.

Our study has limitations. Only lab data from seven days before and within 90 days of the date diagnosis were included in this study. Any lab data after 90 days of diagnosis were missed, and as a result, our findings may be an underestimate of the true etiologic data. The patients at UM are predominantly White, decreasing generalizability of results. We did not include pathology analysis of corneal biopsies, potentially leading to underreporting of infectious etiologies. Further, as etiology and antimicrobial susceptibility patterns differ across different geographic regions, clinicians should be cautioned against the generalizability of these results.

Our study demonstrates that while percentage of organisms causing MK has changed since the prior study at our institution from 1975–1981, the ratio of organism types have remained grossly stable over the past decade. Bacterial keratitis continues to be the most prevalent type with CoNS and S. aureus as the most common isolates. Both isolates were susceptible to vancomycin and gentamicin with high rates of susceptibility to moxifloxacin. P. aeruginosa had high susceptibility to tobramycin, gentamicin, and ciprofloxacin, demonstrating that the commonly used first-line agents are still effective choices in this specific region.

Supplementary Material

Supplemental Table 1

Supplemental Table 1. ICD-9 and ICD-10 codes used to identify patients with microbial keratitis in the electronic health record.

Supplemental Table 2

Supplemental Table 2. Specimens used to identify patients with microbial keratitis in the electronic health record.

Supplemental Table 3

Supplemental Table 3. Organisms isolated from laboratory testing of microbial keratitis patients.

Acronyms:

AFB

Acid-Fast Bacilli

ADI

Area Deprivation Index

CoNS

Coagulase-Negative Staphylococcus

EHR

Electronic Health Record

MK

Microbial Keratitis

Spp.

Species

US

United States

UM

University of Michigan