Corneal Sensitivity and Presence of Pathogenic Organisms Among Participants Who Have Undergone Trichiasis Surgery Differ by Surgical Outcome

CLINICAL SCIENCE

Yvette Schein,* Jennifer C. Harding, MPH,†‡ Shannath L. Merbs, MD, PhD,§ Sandra D. Cassard, ScD,† Kasubi Mabula, MS,¶ Majid S. Othman,k Sheila K. West, PhD,† and Emily W. Gower, PhD**

0.001) and with reduced corneal sensitivity (trend test P , 0.0001). Purpose: To explore the potential association between reduced Symptoms of ocular irritation were not associated with previous corneal sensation and/or conjunctival bacterial colonization and trichiasis surgery, postoperative trichiasis, or eyelid postoperative trichiasis and eyelid contour abnormalities after contour abnormality. corrective eyelid surgery among participants with a history of trachomatous trichiasis. Conclusions: These findings indicate that reduced corneal sensi- Methods: As an ancillary study to the Partnership for Rapid tivity accompanies trachomatous trichiasis and suggest that reduced Elimination of Trachoma (PRET) Surgery Trial in southern Tanza- corneal sensitivity may play an important role in the harboring of nia, we collected data on 580 PRET participants who had undergone pathogenic bacteria on the ocular surface. trichiasis surgery 1 year earlier and 200 age-group–matched Key Words: trachoma, trichiasis, surgery, corneal sensitivity individuals without trichiasis. Assessments included eyelid status evaluation (presence and severity of postoperative trichiasis and/or (Cornea 2016;35:513–518) eyelid contour abnormality), corneal sensitivity by Cochet-Bonnet aesthesiometer, a questionnaire on symptoms of ocular irritation, and conjunctival microbiology. We divided PRET participants based on rachomatous trichiasis is caused by years of repeated their eyelid status and compared results across PRET groups and TChlamydia trachomatis infection. If uncorrected, trichia- versus normals. sis often leads to vision loss because of corneal opacification resulting from the chronic abrasion of eyelashes against the Results: PRET participants had reduced corneal sensitivity com- globe.1 Various surgical techniques to address trachomatous pared with age-matched normals (mean sensitivity ranged from 2.8 trichiasis (TT) have been studied, and the 2 most commonly to 3.8 cm in PRET participants vs. 5.9 cm in normals), and used procedures are bilamellar tarsal rotation and posterior increasing severity of postoperative trichiasis was associated in lamellar tarsal rotation (Trabut).2 However, these procedures a stepwise fashion with reduced corneal sensitivity (mean = 3.5 cm have been plagued by relatively high rates of postoperative for mild and 2.6 cm for severe postoperative trichiasis). Conjunctival trichiasis3–6 and eyelid contour abnormalities (ECAs).3 Pre- colonization with pathogenic bacteria was also associated with more – vious research has suggested that patients with postoperative severe postoperative trichiasis (Cochran Armitage trend test P = trichiasis are more likely to have bacteria isolated from ocular swabs taken of the lower tarsal conjunctiva than are patients who do not develop postoperative trichiasis, and patients with Received for publication September 13, 2015; revision received November severe preoperative trichiasis are more likely to have bacteria 20, 2015; accepted November 22, 2015. Published online ahead of print 7–9 February 18, 2016. isolated than patients with mild preoperative trichiasis. That From the *Brown University, Providence, RI; †Dana Center for Preventive research also hypothesized that colonization with nonchlamy- Ophthalmology, Wilmer Eye Institute, Johns Hopkins University School of dial bacterial pathogens both accompanies postoperative tri- Medicine, Baltimore, MD; ‡Helen Keller International, Dar es Salaam, chiasis and is a risk factor for it. Nonchlamydial bacterial Tanzania; §Division of Oculoplastic Surgery, Wilmer Eye Institute, Johns pathogens also may be involved in the pathogenesis of Hopkins University, Baltimore, MD; ¶School of Medicine, Baltimore, MD; 8 Muhimbili Hospital, Dar es Salaam, Tanzania; kKongwa Trachoma Project trachomatous corneal scarring. An alternative or complemen- Satellite Office, Mtwara, Tanzania; and **Departments of Epidemiology and tary hypothesis is that postoperative trichiasis and/or ECA may Ophthalmology, Wake Forest School of Medicine, Winston-Salem, NC. be associated with reduced corneal sensitivity and that this Supported by a grant from the Bill and Melinda Gates Foundation to the hypoesthesia may be the common pathway between ECAs, Johns Hopkins School of Medicine (S.K.W.) and an Ernest and Elizabeth Althouse Special Scholars Award from Research to Prevent Blindness bacterial colonization, corneal scarring, and related morbidity. (E.W.G.). S. K. West is the recipient of a Senior Scientific Investigator Reduced corneal sensitivity is well recognized as a risk factor award from Research to Prevent Blindness. for bacterial corneal infection in such settings as paresis of the The authors have no conflicts of interest to disclose. fifth cranial nerve, herpes simplex keratitis, postpenetrating Reprints: Emily W. Gower, PhD, Wake Forest Public Health Sciences, fi Medical Center Boulevard, Winston-Salem, NC 27157 (e-mail: egower@ keratoplasty, and other similar conditions. The rst step in wakehealth.edu). investigating this hypothesis is to determine the spectrum of Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved. corneal sensitivity in a representative population and to

Cornea Volume 35, Number 4, April 2016 www.corneajrnl.com | 513

Copyright Ó 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Schein et al Cornea Volume 35, Number 4, April 2016 examine cross-sectional associations between corneal sensitiv- symptoms. The questionnaire consists of 6 questions that ity, trichiasis, and bacterial colonization. ask about dryness, gritty or sandy sensation, burning, redness, The clinical consequences that may result from ECAs are and crusting and whether the participant’s eyes are stuck shut also not well understood. We hypothesize that individuals with in the morning. For each question, the participant was asked ECAs will be more likely to have bacteria isolated from their to report never, rarely, sometimes, frequently, or always. tarsal conjunctiva than those who have normal eyelid margin contours. If participants with ECAs or postoperative trichiasis have reduced corneal sensitivity, they may also have increased Corneal Scarring and Corneal Sensitivity bacterial colony formation and ultimately may be at increased Each eye was assessed for the presence, location, and risk of corneal ulceration. The participant population and study severity of corneal scarring. We measured corneal sensitivity design of the Partnership for Rapid Elimination of Trachoma using a Cochet-Bonnet aesthesiometer with a 0.08-mm- (PRET) Surgery Trial provide the opportunity to examine these diameter filament (Luneau Ophtalmologie, France), first in potential associations. the right eye and then in the left eye. The aestheisometer contains an adjustable-length filament. The shorter the fi METHODS lament length, the more rigid it becomes and the lower the threshold for sensing it touching the eye. We set the initial We conducted this project as an ancillary study to the length of the filament at 6 cm and shortened it in 1-cm PRET Surgery Trial3 in southern Tanzania. The trial was increments until the first positive and reproducible response designed to assess the relative effectiveness of the TT clamp10 for each eye. We recorded a zero for participants who were versus standard bilamellar tarsal rotation instrumentation for unable to perceive even the 1-cm filament. The person trichiasis surgery.2 All individuals enrolled in the PRET Surgery measuring corneal sensitivity was unaware of the surgical Trial and residing in either the Masasi or Mtwara districts were outcome categorization at the time of corneal invited to participate at their 1-year study visit during the sensitivity measurement. summer of 2010. Additionally, 200 individuals without active ocular infection or history of trichiasis were matched to the PRET participants by 10-year age-group and village and were Anesthetized Schirmer Test invited to participate. We refer to these participants as “age- ” We instilled 1 drop of topical anesthetic into the matched normals. All individuals who agreed to participate conjunctival sac of each eye. We then inserted 1 standard provided written consent before participation. The study Schirmer strip into the conjunctival cul-de-sac of each eye. adhered to the Tenets of Helsinki. It was approved by the Institutional Review Boards of the Johns Hopkins University School of Medicine, Wake Forest School of Medicine, and the Tanzanian National Institute for Medical Research. TABLE 1. Characteristics of Eyes With Previous TT Surgery by Eyelid Status 1 Year Postoperatively No Clinical Examination and Sample Collection Postoperative ECA and Trichiasis or Postoperative Postoperative After obtaining informed consent, the PRET Surgery ECA Trichiasis ECA Trichiasis Trial outcomes grader collected all 1-year PRET Surgery Trial data. Details of these methods have been described N 473 280 206 72 previously.3 Briefly, monocular visual acuity was obtained Baseline trichiasis severity (%) using an Early Treatment of Diabetic Retinopathy tumbling E Mild 26.4 7.9 13.1 6.9 chart. Each eyelid was evaluated for the presence and location Moderate 26.6 13.6 25.2 18.1 of trichiatic eyelashes and presence and extent of ECA. The Severe 46.9 78.6 61.7 75.0 ECAs were graded as mild, moderate, or severe depending on Follow-up the amount of vertical deviation from the natural lid contour characteristics 11 and the amount of the horizontal eyelid length affected. The (%) examination was performed with ·2.5 magnification binoc- Postoperative ular loupes according to the World Health Organization’s trichiasis simplified grading system.12 Next, the data specific to this severity ancillary study were collected, as described below, in the None 100 NA 100 NA order presented. For PRET participants, data were collected Mild NA 61.8 NA 66.6 on all study eyes: eyes that had received first-time operation Moderate NA 18.2 NA 19.4 for trichiasis 1 year earlier. For the age-matched normals, Severe NA 20.0 NA 13.9 both eyes of each participant were included. ECA None 100 100 NA NA Mild NA NA 39.3 44.5 Dry Eye Questionnaire Moderate NA NA 50.5 48.6 The examiner administered a dry eye questionnaire Severe NA NA 10.2 6.9 that had been validated and previously used in a developed- NA, not applicable. country context13,14 to evaluate participant-reported

Copyright Ó 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Cornea Volume 35, Number 4, April 2016 Corneal Sensitivity and Pathogenic Organisms

TABLE 2. Percent Who Reported Symptoms Frequently or Always, by Outcome Status Postoperative Age-Matched Without Postoperative Postoperative Trichiasis Symptom Normals Trichiasis or ECA Trichiasis ECA and ECA P* Dryness (%) 32.0 27.9 16.4 20.9 23.6 ,0.001 Gritty sensation (%) 1.5 1.5 1.1 1.9 1.4 0.94 Burning (%) 2.0 1.9 1.8 0 1.4 0.26 Redness (%) 4.5 3.0 1.1 1.0 2.8 0.035 Crusting (%) 0.5 0.4 1.8 1.0 0 0.43 Sticky eyelids (%) 1.5 1.1 1.4 0 1.4 0.40

Data analyzed at the eye level; however, responses were provided at the person level. Both eyes of a participant were assigned the same frequency of symptoms. *Fisher’s exact test P-value.

We recorded the basal tear production in millimeters on the pathogenic. The microbiologist was masked as to all clinical premarked Schirmer strips 1 minute after insertion. information regarding each participant.

Ocular Pathogen Collection Statistical Analysis We obtained conjunctival cultures of each eye by We used descriptive statistics to characterize the study everting the lower eyelid and passing a sterile Dacron swab population, including means and medians. We compared across the inferior fornix twice. Both the specimen collector differences between groups with x2 tests for categorical and her assistant wore new gloves for each specimen collected. variables, and we evaluated trends using the Cochran– Using sterile technique, we broke the tip of the culture swab off Armitage trend test. All analyses were conducted using into a tube of skim milk storage/transport medium, which was SAS 9.2 (SAS Institute, Cary, NC). promptly capped, labeled, and put on ice. Samples were frozen within 6 hours and were then transported in bulk to the microbiology laboratory at Muhimbili National Hospital in Dar RESULTS es Salaam, Tanzania. At the time of processing, samples were We examined 580 PRET participants and an additional thawed and inoculated onto both blood and chocolate agar 200 age-matched normals. Seventy-eight percent of the PRET plates and incubated at 37° for 24–48 hours. A single trained participants had previous bilateral trichiasis surgery, contrib- technician evaluated all plates for growth and identified uting 1031 eyes to this ancillary study, and each age-matched bacterial isolates using standard microbiologic techniques. normal contributed 2 eyes, bringing the total number of eyes We classified organisms using the methodology of Cevallos analyzed to 1431. Characteristics of all eyes at the time of et al.15 Specifically, viridans group Streptococci, coagulase- presentation for PRET participants are provided in Table 1. negative Staphylococcus spp., Corynebacterium spp., and By design, none of the age-matched normals had a history of Bacillus spp. were listed as commensal, and all others as trichiasis and, hence, are not included in the table. The primary ocular irritative symptom reported as occurring frequently or always was “dryness” (Table 2). TABLE 3. Schirmer Test Values by Group and Severity Age-matched normals reported the symptom of dryness more often than any of the diseased groups (P , 0.001). All the Mean (mm) % #10 mm P* groups had mean Schirmer test values for basal tear production Age-matched normals 19.3 12.3 that would be considered within normal limits (10–20 mm; Without postoperative trichiasis or ECA 15.2 28.8 NA Table 3). However, the proportion with Schirmer of 10 mm or Postoperative trichiasis (overall) 15.4 26.2 less varied considerably by group, with the age-matched Mild 14.5 29.5 0.084 normals having the lowest percentage with abnormal values, Moderate 15.3 23.5 whereas over 20% of participants with postoperative trichiasis Severe 18.4 18.2 or ECA had a reading of ,10 mm (P =0.049). ECA (overall) 15.9 22.4 Figure 1A illustrates the distribution of corneal sensi- Mild 14.4 29.6 0.049 tivity scores (possible range: 0–6 cm) by group. Although Moderate 16.9 18.5 most age-matched normals had full corneal sensitivity, Severe 16.8 14.3 reduced corneal sensitivity was common in each of the PRET ECA and post-operative trichiasis 17.7 16.9 subgroups, with the group having both postoperative tri- Seven eyes missing Schirmer test values (4 without postoperative problems, 1 with chiasis and ECA having the lowest mean (2.8 cm). Further- postoperative trichiasis, 1 with ECA, and 1 with both ECA and postoperative trichiasis). more, among those with postoperative trichiasis (Fig. 1B), *Cochran–Armitage trend test. – NA, not applicable. there was an inverse dose response relationship, with pro- gressively lower average sensitivity with increasing severity

Copyright Ó 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Schein et al Cornea Volume 35, Number 4, April 2016

FIGURE 1. Distribution of corneal sensitivity (6 = most sensitive, 0 = absent sensitivity). A, By group. B, By postoperative trichiasis severity among those with postoperative trichiasis. of postoperative trichiasis. Within each ECA category, the chiasis or ECA. The highest rate of pathogenic organisms average sensitivity was roughly 4.0 cm. (21.5%) was seen among those with severe postoperative Table 4 summarizes the microbiology results by sub- trichiasis. Within the subgroup with postoperative trichiasis, group. Age-matched normals exhibited a very low level of we saw a dose–response relationship, with higher rates of culture positivity, especially for the pathogenic organisms. pathogenic organisms with increased TT severity (Cochran– All PRET Surgery Trial groups had higher rates of pathogenic Armitage trend test P = 0.001). organisms present than the age-matched normals, and within Figure 2, which includes data from both the age- the PRET Surgery Trial groups, we saw differences in rates of matched normals and the PRET Surgery Trial groups, shows pathogenic organisms based on surgical outcomes. Both the the association between culture positivity and corneal sensi- ECA and postoperative trichiasis groups had higher rates than tivity, also demonstrating a suggestion of a dose–response the study participants who did not have postoperative tri- relationship between corneal anesthesia and bacterial colonization with pathogenic organisms (Cochran–Armitage trend test P , 0.0001).

TABLE 4. Organism Categories by Overall Groups and Subgroups DISCUSSION None Commensal Pathogenic This study provides new data regarding corneal sensi- (%) (%) (%) tivity in individuals with trichiasis. Corneal sensitivity was Age-matched normals 96.3 3.5 0.3 signiﬁcantly reduced among those with a history of trichiasis, Without postoperative trichiasis 85.8 5.0 9.2 regardless of the eyelid status 1 year postoperatively, whereas or ECA age-matched individuals without previous trichiasis did not Postoperative trichiasis (overall) 83.3 2.3 14.4 have reduced sensitivity. Although the temporal relationship Mild 85.8 2.7 11.4 between corneal sensitivity loss and trichiasis cannot be Moderate 84.2 0 15.8 determined in the setting of a cross-sectional study, this Severe 75.4 3.1 21.5 ﬁnding suggests that reduced corneal sensitivity likely ECA (overall) 84.0 2.8 13.2 predates postoperative trichiasis, most likely developing once Mild 85.5 1.2 13.3 the patient develops trichiasis initially. Those with no post- Moderate 82.4 4.6 13.0 operative trichiasis and no ECA at 1 year continued to have Severe 86.4 0 13.6 decreased corneal sensitivity, suggesting lack of recovery of Postoperative trichiasis and ECA 77.8 2.5 19.8 sensitivity even after removing the irritant.

Copyright Ó 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Cornea Volume 35, Number 4, April 2016 Corneal Sensitivity and Pathogenic Organisms

for those with postoperative trichiasis, the likelihood of bacterial colonization increased with increasing severity of trichiasis. Interestingly, the likelihood of bacterial colonization among participants with any ECA was elevated compared with trichiasis surgery participants without postoperative trichiasis or ECA (13.2% vs. 9.2%); however, we did not see the same dose–response relationship as was seen with postoperative trichiasis. Efficient movement and replenishment of tears across the ocular surface is recognized as an important mechanical component to defending the eye from infection. The presence of ECAs and trichiasis likely impedes that process. However, even those who had no postoperative trichiasis or ECAs after surgery still had much more colonization with pathogenic bacteria (9.2%) than age- matched normals (0.3%). We hypothesize that this coloniza- FIGURE 2. Percentage culture positive versus corneal sensitivity. tion may be partly explained by relative corneal hypoesthesia. A reduction in corneal sensation has both a direct effect on To our knowledge, corneal sensitivity has not pre- reflex tearing and a trophic effect on corneal epithelium.16–19 viously been reported in individuals with trachoma. No Either or both of these mechanisms would reduce the ability human tissue is more densely innervated than the cornea. In of the ocular surface to clear pathogenic organisms. addition to providing exquisite sensation as a protective We assessed symptoms of ocular irritation with a fre- function, adequate corneal sensation is necessary for the quently used dry eye questionnaire. One might expect that regulation of corneal epithelial stability and regeneration.16–18 such symptoms would have been more frequently reported by A reduction of the physiologic competence of the corneal those with ECAs and/or trichiasis than by age-matched epithelium renders the cornea more susceptible to loss of normals. However, the fact that this was not the case may clarity and infection associated with trichiasis. It has been be explained by the relative corneal hypoesthesia among the well established by confocal microscopy16 that morphologic former group. Additionally, the dry eye questionnaire itself, changes in the subbasal corneal nerves occur in such although used frequently in developed-country settings was conditions as herpes simplex and zoster keratitis, keratoconus, not independently validated for the population studied or in dry eye, corneal surgery, and in a variety of other ocular and the Swahili language. systemic conditions such as rheumatoid arthritis and diabetes. Our study is limited by its focus on a specific group of Moreover, in most of these conditions, a reduction in total participants defined by the requirements of a clinical trial on number of nerve fibers has correlated with reduced TT recurrence. Hence, the generalizability of our findings is corneal sensitivity. unknown. Second, by its nature, a cross-sectional assessment Stimulation of the sensory nerves in the cornea and cannot address directionality. Which comes first, the reduc- conjunctiva in turn stimulates the parasympathetic and tion in corneal sensitivity or the bacterial colonization? We sympathetic nerves that innervate the lacrimal gland resulting hope that assessment of corneal sensitivity can be incorpo- in tear production.19 In diabetic patients with reduced corneal rated into future prospective epidemiologic evaluations of sensitivity, diminished tear production can result from individuals with trachoma that include earlier stages of dysfunction in the neuronal circuitry regulating the lacrimal the disease. gland because of limited stimulation of the corneal surface20 or can result from a peripheral neuropathy affecting the lacrimal gland.21 Reduced tear production can also be from ACKNOWLEDGMENTS dysfunction of the tear-secreting glands themselves.19 This The authors acknowledge the trial’s surgery field team study used an anesthetized Schirmer test to measure basal tear and participants with trichiasis and healthy age-matched production. Basal tears are produced by the accessory normals in Mtwara, Tanzania, without whom this work would lacrimal glands of Krause and Wolfring in the conjunctiva not have been possible. and by the main lacrimal gland.22 In addition to the reduced corneal sensitivity we observed in trichiasis patients that REFERENCES likely leads to decreased reflex tearing, damage to the 1. Taylor HR, Burton MJ, Haddad D, et al. Trachoma. Lancet. 2014;384: accessory lacrimal glands by the trachomatous conjunctival 2142–2152. scarring is likely responsible for the observed reduction in 2. Merbs SL, Resnikoff S, Kello AB, et al. Trichiasis Surgery for Trachoma. 2nd ed. Geneva, Switzerland: World Health Organization; basal tear production we measured by the anesthetized 2013. Schirmer test. 3. Gower EW, West SK, Harding JC, et al. Trachomatous trichiasis clamp This study reconfirms previous research describing the vs standard bilamellar tarsal rotation instrumentation for trichiasis association between TT and conjunctival colonization with surgery: results of a randomized clinical trial. JAMA Ophthalmol. 7–9 2013;131:294–301. pathogenic bacteria. All groups of participants with pre- 4. Burton MJ, Kinteh F, Jallow O, et al. A randomised controlled trial of operative trichiasis had significantly greater colonization with azithromycin following surgery for trachomatous trichiasis in the pathogenic bacteria than did the age-matched normals, and Gambia. Br J Ophthalmol. 2005;89:1282–1288.

Copyright Ó 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Schein et al Cornea Volume 35, Number 4, April 2016

5. Khandekar R, Mohammed AJ, Courtright P. Recurrence of trichiasis: 14. Bandeen-Roche K, Munoz B, Tielsch JM, et al. Self-reported assessment a long-term follow-up study in the Sultanate of Oman. Ophthalmic of dry eye in a population-based setting. Invest Ophthalmol Vis Sci. 1997; Epidemiol. 2001;8:155–161. 38:2469–2475. 6. Zhang H, Kandel RP, Atakari HK, et al. Impact of oral azithromycin on 15. Cevallos V, Whitcher JP, Melese M, et al. Association of conjunctival recurrence of trachomatous trichiasis in Nepal over 1 year. Br J bacterial infection and female sex in cicatricial trachoma. Invest Ophthalmol. 2006;90:943–948. Ophthalmol Vis Sci. 2012;53:5208–5212. 7. Burton MJ, Hu VH, Massae P, et al. What is causing active trachoma? 16. Cruzat A, Pavan-Langston D, Hamrah P. In vivo confocal microscopy of The role of nonchlamydial bacterial pathogens in a low prevalence corneal nerves: analysis and clinical correlation. Semin Ophthalmol. setting. Invest Ophthalmol Vis Sci. 2011;52:6012–6017. 2010;25:171–177. 8. Hu VH, Massae P, Weiss HA, et al. Bacterial infection in scarring 17. Oliveira-Soto L, Efron N. Morphology of corneal nerves using confocal trachoma. Invest Ophthalmol Vis Sci. 2011;52:2181–2186. microscopy. Cornea. 2001;20:374–384. 9. Burton MJ, Adegbola RA, Kinteh F, et al. Bacterial infection and 18. Guthoff RF, Wienss H, Hahnel C, et al Epithelial innervation of human trachoma in the Gambia: a case control study. Invest Ophthalmol Vis Sci. cornea: a three-dimensional study using confocal laser scanning fluores- 2007;48:4440–4444. cence microscopy. Cornea. 2005;24:608–613. 10. Merbs SL, Kello AB, Gelema H, et al. The trachomatous trichiasis 19. Dartt DA. Neural regulation of lacrimal gland secretory processes: clamp: a surgical instrument designed to improve bilamellar tarsal relevance in dry eye diseases. Prog Retin Eye Res. 2009;28: rotation procedure outcomes. Arch Ophthalmol. 2012;130:220–223. 155–177. 11. Gower EW, West SK, Cassard SD, et al. Definitions and standardization 20. Saito J, Enoki M, Hara M, et al. Correlation of corneal sensation, but not of a new grading scheme for eyelid contour abnormalities after trichiasis of basal or reflex tear secretion, with the stage of diabetic retinopathy. surgery. PLoS Negl Trop Dis. 2012;6:e1713. Cornea. 2003;22:15–18. 12. Thylefors B, Dawson CR, Jones BR, et al. A simple system for the 21. Cousen P, Cackett P, Bennett H, et al. Tear production and corneal assessment of trachoma and its complications. Bull World Health Organ. sensitivity in diabetes. J Diabetes Complications. 2007;21:371–373. 1987;65:477–483. 22. Stern ME, Beuerman RW, Fox RI, et al. The pathology of dry eye: the 13. Schein OD, Munoz B, Tielsch JM, et al. Prevalence of dry eye among the interaction between the ocular surface and lacrimal glands. Cornea. elderly. Am J Ophthalmol. 1997;124:723–728. 1998;17:584–589.