Aqueous tear deficiency (commonly described as dry eye) is a well-recognized disorder in dogs that can lead to serious ocular complications if undiagnosed and untreated. In contrast, little is known about tear deficiency in cats.1 In both species, lacrimation is most commonly evaluated by use of the STT; however, a PubMed search in September 2018 with the keywords “Schirmer tear test” and “dog” returned a total of 102 pertinent articles, whereas use of the term “cat” instead of “dog” led to identification of only 17 articles. Some authors have suggested that STT is not reliable in cats, as low measurements could result from stress-induced stimulation of sympathetic tone, causing a temporary reduction in tear production during testing.2 Further, cats may be perceived as less tolerant than dogs to the standard STT procedure, which requires placement of the end of the strip in the conjunctival fornix for a full minute.
In a recent study,3 investigators established reference values for STT-1 measurements in healthy cats. However, those findings may not have reflected potential changes in sympathetic tone experienced by the general population of cats in various settings, as the study sample comprised laboratory-housed cats that were used to human contact.
The purpose of the study reported here was to evaluate whether the STT-1 can be relied on as a diagnostic tool for use in cats in various environmental settings, determine whether sympathetic stimulation impacts STT-1 measurements, and determine whether meaningful conclusions regarding lacrimation in cats can be drawn from STT-1 measurements obtained with strip placement for < 1 minute. We hypothesized that STT values would not differ among cats examined in environments of variable perceived stress levels and that measurements obtained in < 60 seconds would correlate positively with those obtained at 60 seconds.
Materials and Methods
Cats
Cats were examined in 3 different settings (a private veterinary practice in Iowa with feline patients only [private practice; n = 100], a spay-neuter program clinic for feral cats at the Iowa State University Lloyd Veterinary Medical Center [feral cat clinic; 56], and the small animal hospital of the Iowa State University Lloyd Veterinary Medical Center [teaching hospital; 20]). All cats at the private practice were client owned and were seen for nonocular reasons, including general wellness examination, vaccination, dermatologic problems, and gastrointestinal disorders. Cats brought to the feral cat clinic were barn or shelter cats that were considered to have little or no human contact prior to the study. Cats evaluated at the teaching hospital were all staff owned or student owned and were used to frequent human contact. Both eyes were tested for every cat in the study (n = 352 eyes). Prior to inclusion, cats were deemed free of ocular surface disease with no noticeable ocular discharge, conjunctivitis, or keratitis. The study was approved by the Iowa State University Institutional Animal Care and Use Committee, and informed consent was obtained from owners of privately owned cats.
STT-1
The order of testing (right or left eye first) was randomized for each cat by use of commercially available software.a A single lot number of STT stripsb was used throughout the study to reduce variability in absorption.4 The STT-1 was performed in each eye by placing the designated portion of the strip in the lateral third of the ventral conjunctival fornix. An assistant handled cats on the examination table during testing. A stopwatch was used to time the tests, and the wet portion of the strip was measured in millimeters every 10 seconds (at the feral cat clinic and teaching hospital) or every 30 seconds (at the private practice) for 1 minute. Owners were present for tests of privately owned cats.
Sympathetic stimulation—For the 20 cats evaluated at the teaching hospital, the STT-1 was performed in both eyes twice, with 30 minutes between tests. One test was performed in a quiet examination room (nonstimulated conditions), and the other test was performed in the same room but in the presence of prerecorded noise intended to cause brief sympathetic nervous system stimulation (stimulated conditions). The stimulus consisted of a standardized loud barking dog sound that was initiated 30 seconds before the STT-1 was started and continued throughout the full 1-minute test period, augmented periodically with a loud, sharp noise generated by hand cymbals. The order of the 2 tests was randomized for each cat with commercially available software.a For these tests, each cat's heart rate was evaluated with a stethoscope before and after STT-1 was performed.
Statistical analysis
Normality of the data was evaluated with the Shapiro-Wilk test. Data were normally distributed (P ≥ 0.184) and reported as mean ± SD and range (minimum to maximum). Paired Student t tests were used to compare STT-1 measurements of right and left eyes for all cats and to compare STT-1 measurements and mean heart rates (before and after STT-1) of cats under nonstimulated and stimulated conditions. Aside from the aforementioned analysis, STT-1 measurements from cats under stimulated conditions were excluded from statistical analyses. One-way ANOVA was performed to compare STT-1 measurements obtained in cats at the private practice, the feral cat clinic, and the teaching hospital. Correlation between STT-1 values obtained at 30 seconds and 60 seconds for each cat was assessed by Pearson correlation analysis. To classify the strength of correlation, an absolute Pearson r value of 0 to 0.19 was regarded as very weak, 0.2 to 0.39 as weak, 0.40 to 0.59 as moderate, 0.6 to 0.79 as strong, and 0.8 to 1 as very strong.5 Statistical analysis was performed with commercially available software,c and values of P ≤ 0.05 were considered significant.
For modeling purposes, two-thirds of the data were partitioned out as a training set, with the remaining third used as a verification set and right eye and left eye measurements treated as separate sample occasions paired for each cat (ie, occasion 1 for left eye and occasion 2 for right eye). The stimulated groups were included in nonlinear mixed-effects models but were excluded from the verification set. Establishment and validation of a statistical model for STT-1 measurements in cats required several steps that are described in detail elsewhere (Supplementary Appendix S1, available at avmajournals.avma.org/doi/suppl/10.2460/javma.256.6.681). Briefly, data were visually examined with a freely available statistical tool,d local regression analysis (locally estimated scatterplot smoothing) was performed,e and mathematical modeling and simulation were performed as previously described.6–10 Testing of covariates such as age, body weight, sex, breed, eye, and experimental condition (stimulated vs nonstimulated) was performed with commercial softwaref; the model was validated by assessment of the relative SDs of parameter estimates, visual predictive checks from 1,000 Monte Carlo simulations, normality tests of random effects, and other standard goodness-of-fit diagnostic tests.
The resulting model was implemented with a statistical toold and graphed against the verification set to determine the quality of data fit and to rule out overfitting as part of an external validation step for the model. For visual predictive checks, 1,000 Monte Carlo simulations were used to derive a distribution of predicted STT-1 time courses from which a median and 95% (2.5th to 97.5th percentile) PI (comprising 95% of the simulated STT-1 profiles) could be generated.
Results
The STT-1 measurements were obtained from both eyes in each cat for all predetermined time points. No significant (P ≥ 0.348) difference in STT-1 measurements was noted between right and left eyes for any group of cats 30 or 60 seconds after test initiation, so mean data from both eyes was considered for further analysis. At 30 seconds, STT-1 measurements were not significantly (P = 0.432) different among cats examined at the private practice (mean ± SD, 10.1 ± 4.0 mm; range, 1.5 to 20 mm), feral cat clinic (9.9 ± 2.6 mm; range, 4.5 to 16.5 mm), and teaching hospital (9.0 ± 2.8 mm; range, 4 to 13 mm). Similarly, STT-1 measurements at 60 seconds did not differ significantly (P = 0.350) among cats evaluated at the private practice (mean ± SD, 15.0 ± 5.4 mm; range, 3 to 29.5 mm), feral cat clinic (15.5 ± 3.7 mm; range, 8 to 24 mm), and teaching hospital (13.7 ± 4.6 mm; range, 7 to 23 mm). The STT-1 measurements at 30 seconds and 60 seconds were significantly (P < 0.001) and very strongly correlated (r = 0.941; Figure 1). Stimulated conditions at the teaching hospital were associated with significantly (P < 0.001) increased heart rate (mathematical average of measurements obtained before and after STT-1) in cats (mean ± SD, 226 ± 12 beats/min; range, 204 to 243 beats/min), compared with the same measure under nonstimulated conditions (185 ± 15 beats/min; range, 153 to 210 beats/min), but STT-1 measurements did not differ significantly between stimulated and nonstimulated conditions at any of the 6 time points evaluated (10-second intervals up to 60 seconds; Figure 2).
Results of Pearson correlation analysis for STT-1 measurements obtained 30 and 60 seconds after test initiation in 176 cats without signs of ocular disease evaluated in 3 different environments (a private practice [n = 100], a feral cat clinic [56], or a veterinary teaching hospital [20]). Data points represent the mean measurement for left and right eyes in each cat. A very strong positive correlation is evident (r = 0.941; P < 0.001).
Citation: Journal of the American Veterinary Medical Association 256, 6; 10.2460/javma.256.6.681
Results of Pearson correlation analysis for STT-1 measurements obtained 30 and 60 seconds after test initiation in 176 cats without signs of ocular disease evaluated in 3 different environments (a private practice [n = 100], a feral cat clinic [56], or a veterinary teaching hospital [20]). Data points represent the mean measurement for left and right eyes in each cat. A very strong positive correlation is evident (r = 0.941; P < 0.001).
Citation: Journal of the American Veterinary Medical Association 256, 6; 10.2460/javma.256.6.681
Results of Pearson correlation analysis for STT-1 measurements obtained 30 and 60 seconds after test initiation in 176 cats without signs of ocular disease evaluated in 3 different environments (a private practice [n = 100], a feral cat clinic [56], or a veterinary teaching hospital [20]). Data points represent the mean measurement for left and right eyes in each cat. A very strong positive correlation is evident (r = 0.941; P < 0.001).
Citation: Journal of the American Veterinary Medical Association 256, 6; 10.2460/javma.256.6.681
Mean ± SD measurements of tear production by STT-1 over 1 minute for 20 of the cats in Figure 1 evaluated at a veterinary teaching hospital under nonstimulated (in a quiet examination room; solid line with black circles) and stimulated (in a room with loud prerecorded noises designed to cause brief sympathetic nervous system stimulation; dashed line with white triangles) conditions. Despite a significantly (P < 0.001) higher mean heart rate under stimulated conditions indicating sympathetic nervous system stimulation, no significant (P ≥ 0.335) differences in STT-1 values were detected at any time point.
Citation: Journal of the American Veterinary Medical Association 256, 6; 10.2460/javma.256.6.681
Mean ± SD measurements of tear production by STT-1 over 1 minute for 20 of the cats in Figure 1 evaluated at a veterinary teaching hospital under nonstimulated (in a quiet examination room; solid line with black circles) and stimulated (in a room with loud prerecorded noises designed to cause brief sympathetic nervous system stimulation; dashed line with white triangles) conditions. Despite a significantly (P < 0.001) higher mean heart rate under stimulated conditions indicating sympathetic nervous system stimulation, no significant (P ≥ 0.335) differences in STT-1 values were detected at any time point.
Citation: Journal of the American Veterinary Medical Association 256, 6; 10.2460/javma.256.6.681
Mean ± SD measurements of tear production by STT-1 over 1 minute for 20 of the cats in Figure 1 evaluated at a veterinary teaching hospital under nonstimulated (in a quiet examination room; solid line with black circles) and stimulated (in a room with loud prerecorded noises designed to cause brief sympathetic nervous system stimulation; dashed line with white triangles) conditions. Despite a significantly (P < 0.001) higher mean heart rate under stimulated conditions indicating sympathetic nervous system stimulation, no significant (P ≥ 0.335) differences in STT-1 values were detected at any time point.
Citation: Journal of the American Veterinary Medical Association 256, 6; 10.2460/javma.256.6.681
Data modeling
The resulting model fit of the STT-1 data was a hyperbolic function adapted to saturation similar to that used to describe a drug concentration-effect relationship (ie, the Emax model) as follows:
where Sij is the STT-1 measurement (in mm) for the ith animal at time tj, Smax is the maximum value of the STT (fixed at 35 mm), and S50i is the time at which half of the maximum STT value is reached for the ith animal. This model was tested against a reduced (ie, linear) model and a more complex full sigmoid Emax model.11 The hyperbolic Emax model strongly outperformed the linear model and performed similarly to the more complex full sigmoid Emax model while being more parsimonious (Figure 3).
Graph depicting comparison of population model performances for prediction of STT-1 measurements in cats. Hyperbolic, linear, and sigmoidal models (dashed green, red, and purple lines, respectively) were compared; the hyperbolic model most closely matched the line created by plotting the geometric means at the 10−, 20−, 30−, 40−, 50−, and 60-second time points (solid black line) and performed in a similar manner to the more complex sigmoid model. All the observations are depicted as circles with varying shades of gray (darker gray indicates greater overlap of observations).
Citation: Journal of the American Veterinary Medical Association 256, 6; 10.2460/javma.256.6.681
Graph depicting comparison of population model performances for prediction of STT-1 measurements in cats. Hyperbolic, linear, and sigmoidal models (dashed green, red, and purple lines, respectively) were compared; the hyperbolic model most closely matched the line created by plotting the geometric means at the 10−, 20−, 30−, 40−, 50−, and 60-second time points (solid black line) and performed in a similar manner to the more complex sigmoid model. All the observations are depicted as circles with varying shades of gray (darker gray indicates greater overlap of observations).
Citation: Journal of the American Veterinary Medical Association 256, 6; 10.2460/javma.256.6.681
Graph depicting comparison of population model performances for prediction of STT-1 measurements in cats. Hyperbolic, linear, and sigmoidal models (dashed green, red, and purple lines, respectively) were compared; the hyperbolic model most closely matched the line created by plotting the geometric means at the 10−, 20−, 30−, 40−, 50−, and 60-second time points (solid black line) and performed in a similar manner to the more complex sigmoid model. All the observations are depicted as circles with varying shades of gray (darker gray indicates greater overlap of observations).
Citation: Journal of the American Veterinary Medical Association 256, 6; 10.2460/javma.256.6.681
None of the investigated covariates (including non-stimulated vs stimulated conditions) had a significant impact (P > 0.05) on the Akaike or Bayesian information criteria. Materials supporting the validity and robustness of the model are provided (Supplementary Tables S1–S3 and Supplementary Figures S1 and S2, available at avmajournals.avma.org/doi/suppl/10.2460/javma.256.6.681).
A comparison of observed STT measurements versus measurements predicted by the model is shown for a randomly selected sample of study cats (Figure 4). The median and 95% (2.5th to 97.5th percentile) PIs for STT-1 values of cats at 10-second intervals after strip placement generated with the final model were summarized (Table 1). An online toolg was created, allowing for individual predictions of STT-1 values at 1 minute for shorter test durations. For example, 7 mm of strip wetness at 20 seconds corresponded to 15 mm at 1 minute (predicted by Monte Carlo simulation), a value that is considered normal in cats.3
Comparison of model-predicted STT-1 measurements (gray line) with observed data points (black circles) for a randomly selected subsample (n = 16) of the cats in Figure 1. Each graphic represents data for an individual cat.
Citation: Journal of the American Veterinary Medical Association 256, 6; 10.2460/javma.256.6.681
Comparison of model-predicted STT-1 measurements (gray line) with observed data points (black circles) for a randomly selected subsample (n = 16) of the cats in Figure 1. Each graphic represents data for an individual cat.
Citation: Journal of the American Veterinary Medical Association 256, 6; 10.2460/javma.256.6.681
Comparison of model-predicted STT-1 measurements (gray line) with observed data points (black circles) for a randomly selected subsample (n = 16) of the cats in Figure 1. Each graphic represents data for an individual cat.
Citation: Journal of the American Veterinary Medical Association 256, 6; 10.2460/javma.256.6.681
Median and 95% (2.5th to 97.5th percentile) PIs of STT-1 values in cats without ocular disease 10, 20, 30, 40, 50, and 60 seconds after test initiation.
| Time (s) | Median (95% PI) |
|---|---|
| 10 | 3.7 (1.8–7.1) |
| 20 | 6.6 (3.4–12.0) |
| 30 | 9.1 (4.8–15.6) |
| 40 | 11.1 (6.1–18.2) |
| 50 | 12.8 (7.3–20.4) |
| 60 | 14.3 (8.2–22.3) |
The results were generated from a validated model of STT-1 data on the basis of measurements for 176 cats of diverse backgrounds evaluated in various testing environments.
Discussion
Results of the present study clearly indicated that the STT-1 is a reliable diagnostic tool for assessment of tear production in cats. No differences in STT-1 values were noted among cats of different backgrounds that were tested in various clinical settings or between cats that were evaluated under stimulated and nonstimulated conditions. These findings disproved a long-standing belief that stress falsely decreases tear production in cats, presumably by increasing sympathetic tone to the lacrimal gland2—an idea widely spread in the veterinary community despite the lack of peer-reviewed literature to support it.
Sympathetic and parasympathetic nerves reach the lacrimal glands in cats,12 dogs,13 and other species.14,15 Although the effects of these autonomous systems are generally antagonistic (eg, increased or decreased heart rate or dilation or constriction of the pupil), the 2 systems work synergistically in some tissues such as the salivary16 and lacrimal glands.15 The parasympathetic nervous system dominates anatomically and functionally in the lacrimal gland, but sympathetic nerves also promote tear secretion by releasing the stimulatory neurotransmitters norepinephrine and neuropeptide Y.15 An investigation by Whitwell14 showed that sympathetic nerves play a role in normal continuous tear secretion in people, and in that same study, the STT measurements for 1 cat were higher when the lacrimal sympathetic innervation was electrically stimulated in one eye, compared with results for the contralateral, unstimulated eye. The mechanism by which sympathetic nerves regulate tear secretion is unclear and warrants further investigation.
Importantly, the results of our study showed that application of the test strip for < 60 seconds when performing STT-1 for cats can provide results of diagnostic value, a finding that can be particularly useful for evaluation of select feline patients that are either uncooperative or intolerant to strip placement for an entire minute. First, a very strong correlation (r = 0.941) was found between STT-1 measurements obtained 30 seconds and 60 seconds after strip placement. Second, data modeled from values recorded every 10 seconds provided a solid method to extrapolate STT-1 measurements for shorter testing durations (10 to 50 seconds) into what results would have been at the standard 1-minute time point.
Nevertheless, there are benefits to testing cats for an entire minute whenever possible, as it facilitates comparisons with established normative data3 and previous reports of feline ocular diseases such as dry eye1,17 and conjunctivitis.18 Further, testing for a full minute provides insights into the kinetics of strip wetting, whereby the initial rapid rise originates from tears in the lacrimal lake, while the subsequent slower rise originates from the steady-state tear production.19 This temporal examination of STT-1 results has revealed differences in dogs with healthy eyes, dry eye, or nasolacrimal blockage,19 and the same may be true in cats.
The present study provided an excellent framework to establish 95% PIs for STT-1 measurements in cats given the large number of animals examined, their diverse backgrounds, and the different testing environments. The median STT-1 measurement at 60 seconds was 14.3 mm (95% PI, 8.2 to 22.3 mm). The lower limit of the 95% PI for STT-1 values in cats without ocular disease in this study was comparable to the lower limit of the 95% central range reported in 135 healthy domestic cats in 2015 (9 mm/min)3 but higher than the lower limit (calculated as 2 SDs below the mean) suggested by Veith et al20 in 1970 (6 mm/min). Importantly, collection of STT-1 data from additional cats, whether healthy or affected with ocular surface disease, can be used to further refine the model predictions and prediction intervals identified in the present study.
Another strength of the study was the complementary use of conventional statistical tests (eg, paired t tests, ANOVA, and Pearson correlation analysis) along with a nonlinear mixed-effect modeling approach. Modeling of the STT-1 data provided several advantages: first, the model used measurements of strip saturation across multiple time points (from 10 to 60 seconds) rather than only 1 time point, thus increasing the power and robustness of the results; second, the model that was created allows for prediction of STT-1 measurement in cats for any test duration, thus providing insight about the clinical relevance of STT-1 measurements obtained in < 60 seconds; and third, the model simulated the distribution of STT-1 measurements for different percentiles of the population, thus providing PIs (ie, a reference range) for the general feline population and for any given time point.
The authors encourage clinicians to perform STT-1 as part of an ophthalmic examination in cats, especially ones with evidence of ocular surface disease, as this diagnostic tool is reliable and its results could improve case management. Indeed, aqueous tear deficiency has detrimental effects on the ocular surface that lead to pain and inflammation if left untreated.21 It is important to note that a single low STT-1 reading in a cat should not justify lifelong treatment for presumed dry eye. Instead, STT-1 results should take into account the clinical signs observed, and if in doubt, the assessment should be repeated at subsequent visits to confirm the suspicion of aqueous tear deficiency.
Acknowledgments
No third-party funding or support was received in connection with this study or the writing or publication of the manuscript. The authors declare that there were no conflicts of interest.
Presented orally at the American College of Veterinary Ophthalmologists Annual Conference, Minneapolis, September 2018.
ABBREVIATIONS
| Emax | Maximum effect |
| PI | Predictive interval |
| STT | Schirmer tear test |
Footnotes
Excel 2016, Microsoft Corp, Redmond, Wash.
STT strips, Eye Care Product Manufacturing LLC, Tucson, Ariz.
SigmaPlot, version 14.0, Systat Software Inc, San Jose, Calif.
R, version 3.4.4, The R Foundation for Statistical Computing, Vienna, Austria.
ggplot2 2.2.1, The R Foundation for Statistical Computing, Vienna, Austria.
Monolix Suite 2018R1, Lixoft, Paris, France.
Experimental STT-1 Calculator, Ames, Iowa. Available at: benjamin-pkpd.shinyapps.io/stt-calculator/. Accessed Feb 4, 2020.
References
1. Sebbag L, Pesavento PA, Carrasco SE, et al. Feline dry eye syndrome of presumed neurogenic origin: a case report. JFMS Open Rep 2018;4:2055116917746786.
2. Lim CC, Reilly CM, Thomasy SM, et al. Effects of feline herpesvirus type 1 on tear film break-up time, Schirmer tear test results, and conjunctival goblet cell density in experimentally infected cats. Am J Vet Res 2009;70:394–403.
3. Sebbag L, Kass PH, Maggs DJ. Reference values, intertest correlations, and test-retest repeatability of selected tear film tests in healthy cats. J Am Vet Med Assoc 2015;246:426–435.
4. García-Porta N, Mann A, Sáez-Martínez V, et al. The potential influence of Schirmer strip variables on dry eye disease characterisation, and on tear collection and analysis. Cont Lens Anterior Eye 2018;41:47–53.
5. Swinscow T, Campbell M. Correlation and regression. In: Campbell MJ, Swinscow TDV, eds. Statistics at square one. 11th ed. Hoboken, NJ: Wiley-Blackwell, 2009;119–132.
6. Mochel JP, Gabrielsson J, Collard W, et al. Animal Health Modeling & Simulation Society: a new society promoting model-based approaches in veterinary pharmacology. J Vet Pharmacol Ther 2013;36:417–419.
7. Mochel JP, Fink M, Peyrou M, et al. Pharmacokinetic/pharmacodynamic modeling of renin-angiotensin aldosterone biomarkers following angiotensin-converting enzyme (ACE) inhibition therapy with benazepril in fogs. Pharm Res 2015;32:1931–1946.
8. Riviere JE, Gabrielsson J, Fink M, et al. Mathematical modeling and simulation in animal health. Part I: moving beyond pharmacokinetics. J Vet Pharmacol Ther 2016;39:213–223.
9. Bon C, Toutain PL, Concordet D, et al. Mathematical modeling and simulation in animal health. Part III: using nonlinear mixed-effects to characterize and quantify variability in drug pharmacokinetics. J Vet Pharmacol Ther 2018;41:171–183.
10. Pelligand L, Soubret A, King JN, et al. Modeling of large pharmacokinetic data using nonlinear mixed-effects: a paradigm shift in veterinary pharmacology. A case study with robenacoxib in cats. CPT Pharmacometrics Syst Pharmacol 2016;5:625–635.
11. Macdougall J. Analysis of dose-response studies—Emax model. In: Ting N, ed. Dose finding in drug development. New York: Springer, 2006;127–145.
12. Botelho SY, Hisada M, Fuenmayor N. Functional innervation of the lacrimal gland in the cat. Origin of secretomotor fibers in the lacrimal nerve. Arch Ophthalmol 1966;76:581–588.
13. Powell CC, Martin CL. Distribution of cholinergic and adrenergic nerve fibers in the lacrimal glands of dogs. Am J Vet Res 1989;50:2084–2088.
14. Whitwell J. Role of the sympathetic in lacrimal secretion. Br J Ophthalmol 1961;45:439–445.
15. Dartt DA. Neural regulation of lacrimal gland secretory processes: relevance in dry eye diseases. Prog Retin Eye Res 2009;28:155–177.
16. Proctor GB, Carpenter GH. Salivary secretion: mechanism and neural regulation. Monogr Oral Sci 2014;24:14–29.
17. Uhl LK, Saito A, Iwashita H, et al. Clinical features of cats with aqueous tear deficiency: a retrospective case series of 10 patients (17 eyes). J Feline Med Surg 2019;21:944–950.
18. Davis K, Townsend W. Tear-film osmolarity in normal cats and cats with conjunctivitis. Vet Ophthalmol 2011;14(suppl 1):54–59.
19. Williams DL. Analysis of tear uptake by the Schirmer tear test strip in the canine eye. Vet Ophthalmol 2005;8:325–330.
20. Veith LA, Cure TH, Gelatt KN. The Schirmer tear test in cats. Mod Vet Pract 1970;51:48–49.
21. Craig JP, Nelson JD, Azar DT, et al. TFOS DEWS II Report executive summary. Ocul Surf 2017;15:802–812.
