Appendix

Appendix

Contents

Data tables

Table 1: Overview of the Caffery et al. (2014) study

Table 2: Overview of the Jacobi et al. (2011) study

Table 3: Overview of the Messmer et al. (2010) study

Table 4: Overview of the Lemp et al. (2011) study

Table 5: Overview of the Sullivan et al. (2012) study

Table 6: Overview of the Tomlinson et al. (2010) study

Table 1 Overview of the Caffery et al. (2014) study

Study component

Description

Objectives/hypotheses

To assess the correlation between tear osmolarity readings (captured with the TearLab osmolarity system) and symptoms of DED, and to determine how well these correlate with the self‑assessment of DED.

Study design

Prospective, cross‑sectional, single centre study. No follow‑up period was reported.

Setting

USA, 3‑day study (2012) at optometric conference.

Inclusion/exclusion criteria

Inclusion criteria: none stated.

Exclusion criterion: attendees who had worn contact lenses in the past 2 weeks.

Primary outcomes

The correlation between tear osmolarity and symptoms of DED.

Statistical methods

Statistical significance was assessed using α=0.05 and two‑sided hypothesis tests. Pearson correlations were used to assess the relationship between interval‑scaled measurements such as age, DEQ‑5 score, and osmolarity. Linear regression analyses were used to further characterise these relationships.

Intraocular measurements of osmolarity were characterised using the mean difference and 95% limits of agreement. Independent group t tests were used to compare patient characteristics between those with and without self‑reported dry eye and also between men and women.

Patients included

Attendees at an optometric meeting: n=249 (140 men [49.8+14.1 years]; 109 women [39.7+12.7 years]).

Results

There was no statistically significant difference between the DEQ‑5 scores and average tear osmolarity (correlation coefficient 0.02) and highest osmolarity (correlation coefficient 0.03).

The mean DEQ‑5 score was significantly higher among people who self‑reported dry eye compared with those who did not (11.3 versus 5.4; p<0.0001).

No significant differences were seen between the 'yes' and 'no' self‑reported dry eye groups and average osmolarity (p=0.23) and highest osmolarity (p=0.14).

Conclusions

No statistically significant correlation was found between tear osmolarity and ocular symptoms as reported on the DEQ‑5 or with tear osmolarity and a Gestalt self‑assessment of dry eye.

Abbreviations: DED, dry eye disease; DEQ, dry eye questionnaire (DEQ‑5 is the short form of the DEQ).

Table 2 Overview of the Jacobi et al. (2011) study

Study component

Description

Objectives/hypotheses

To assess the differences in the osmolarity in tear samples of people with moderate to severe DED compared with healthy people as controls.

Study design

Prospective, non‑randomised, single centre study. No recruitment period was provided. No follow‑up period was reported.

Setting

Germany. No further details on the setting were given in the paper.

Inclusion/exclusion criteria

Inclusion criteria (DED):

  • a TBUT <5 seconds

  • a Schirmer's test <5 mm

  • OSDI score >83.

Inclusion criteria (controls):

  • asymptomatic

  • a TBUT >10 seconds

  • a Schirmer's test >15 mm.

Primary outcomes

The difference in the osmolarity in tear samples of people with moderate to severe DED compared with healthy people as controls.

Statistical methods

Nonparametric tests were used to compare the results between both groups (Mann–Whitney U test).

Patients included

n=133 people with moderate to severe DED (58 years, 51–64 years; 86 women, 47 men).

n=95 controls (52 years, 48–61 years; 55 women, 40 men).

Results

People with moderate to severe DED showed a tear film osmolarity of 320 mOsm/litre (301–324 mOsm/litre).

The results of the control group were 301 mOsm/litre (298–304 mOsm/litre).

The results revealed a significantly higher tear film osmolarity in people with moderate to severe DED compared with the control group.

The sensitivity was 87%, and the specificity was 81%.

Conclusions

TearLab appears to be an effective objective diagnostic tool in the diagnosis of DED.

Abbreviations: DED, dry eye disease; mm, millimetres; mOsm, milliosmole (one‑thousandth of an osmole); OSDI, Ocular Surface Disease Index; TBUT, tear film breakup time.

Table 3 Overview of the Messmer et al. (2010) study

Study component

Description

Objectives/hypotheses

To evaluate the ability of TearLab to differentiate between people with and without dry eye.

Study design

Prospective cohort, single centre. No recruitment period was provided. No follow‑up period was reported.

Setting

Germany. No further details on the setting were given in the paper.

Inclusion/exclusion criteria

When ≥3 of 6 dry eye signs were present, the patient was recruited into the DED group.

DED signs for inclusion:

  • an OSDI score >15

  • any staining of the cornea or conjunctiva in the typical interpalpebral area

  • a TBUT <7 seconds

  • a Schirmer's test <7 mm in 5 minutes

  • blepharitis or meibomitis.

Primary outcomes

Measurement of tear film osmolarity with TearLab.

Statistical methods

Pearson and Spearman correlations, Wilcoxon and Mann–Whitney U tests.

Patients included

129 people (DED): with 3–6 dry eye signs or symptoms (55 years, 19–86 years; 62.8% women).

71 people (controls): with up to 2 signs or symptoms of DED (39 years, 16–83 years; 62.0% women).

Results

Tear film osmolarity did not show any correlation with the 6 clinical signs of DED.

Osmolarity testing could not discriminate between patients with DED (308.9±14.0 mOsm/litre) and the control group (307.1±11.3 mOsml/litre).

Osmolarity did not correlate with artificial tear use.

Technical problems with the TearLab, reflex tear secretion, or the difficulty in establishing a DED diagnosis with the recommended tests may account for these results.

Conclusions

Tear film osmolarity could not discriminate between people with DED and healthy people as controls.

Abbreviations: DED, dry eye disease; mOsm, milliosmole (one‑thousandth of an osmole); OSDI, Ocular Surface Disease Index; TBUT, tear film breakup time.

Table 4 Overview of the Lemp et al. (2011) study

Study component

Description

Objectives/hypotheses

To assess the diagnostic performance of tear osmolarity (TearLab) compared with other commonly used tests (Schirmer's test, TBUT, staining and meibomian gland grading).

Study design

A prospective, observational, multicentre study. No recruitment period was provided. No follow‑up period was reported.

Setting

10 sites in the EU and USA.

Inclusion/exclusion criteria

Inclusion:

  • 18–82 years of age

  • general population.

Exclusion:

  • active infection of the eye

  • active ocular allergy

  • lid deformity or abnormal lid movement disorder

  • refractive surgery within 1 year of the study visit

  • pregnancy or lactation

  • abnormal nasolacrimal drainage

  • punctal plug placement within 30 days of testing

  • systemic disease known to affect tear production

  • starting or changing the dose of chronic systemic medication known to affect tear production within 30 days of testing (such as antihistamines, antidepressants, diuretics, corticosteroids or immunomodulators)

  • known hypersensitivity to any of the agents used in testing (sodium fluorescein, lissamine green).

Primary outcomes

Sensitivity, specificity, area under the ROC.

Statistical methods

Optimal cut‑off values for each test were determined assuming equal risk for false positive and false negative results. Gaussian distributions were generated based on the mean and standard deviation of normal and dry eye disease populations.

Patients included

Patients from the general population: n=299 (46.3, 18–82 years; 218 women, 81 men).

Results

Tear osmolarity had a 72.8% sensitivity and 92.0% specificity at a cut‑off value of 312 mOsm/litre.

No other clinical sign showed more than 62% performance in both categories. Corneal staining, conjunctival staining, and meibomian gland grading lacked sensitivity (54.0%, 60.3%, and 61.2% respectively), whereas TBUT and Schirmer's test results lacked specificity (45.3% and 50.7% respectively).

Tear osmolarity also had the highest area under the ROC curve (0.89) followed by conjunctival staining (0.83), TBUT (0.81), meibomian gland grading (0.78), corneal staining (0.77), and Schirmer's test (0.71).

Inter‑eye differences in osmolarity correlated with increasing disease severity (r2=0.32; p=0.0001).

Conclusions

Tear osmolarity was the best single objective metric to both diagnose and classify DED. A cut‑off threshold of more than 308 mOsm/litre was most sensitive in differentiating normal from mild to moderate DED.

Abbreviations: DED, dry eye disease; OSDI, Ocular Surface Disease Index; ROC, receiver operating characteristic; TBUT, tear breakup time.

Table 5 Overview of the Sullivan et al. (2012) study

Study component

Description

Objectives/hypotheses

To compare the variability of tear osmolarity over a 3‑month period with that of other commonly used biomarkers used for diagnosing DED and to determine their reproducibility when measuring response to treatment.

Study design

Prospective, longitudinal, observational case series study with an additional treatment arm. No recruitment period was provided. No follow‑up period was reported.

Setting

2 study centres (Spain, Turkey).

Inclusion/exclusion criteria

Inclusion:

  • age >17 years

  • ocular examination showing DED within the 2 years before the study.

Exclusion:

  • eyelid deformity or movement disorder, active ocular infection or allergy

  • laser in situ keratomileusis or photorefractive keratectomy surgery within 1 year of visit 1 or during the study

  • systemic disease known to affect ocular health

  • systemic or topical medications that may have affected ocular health

  • use of artificial tears within 2 hours before their scheduled study visit

  • known sensitivity to any of the agents used in the testing procedures

  • if any of the following criteria applied within 30 days of each visit:

    • change in chronic ocular medication

    • change in systemic medication known to affect ocular health

  • pregnancy or lactation during the study

  • punctal plug placement or cauterisation within 30 days of visit 1 or during the study

  • use of ocular ciclosporin before visit 1, 2, or 3.

Primary outcomes

Variability of tear osmolarity over a 3‑month period.

Statistical methods

The range and standard deviation of each test were reported to compare the variability of the commonly used signs and symptoms of DED. For direct comparison of disease markers, results were expressed as percentages of the total dynamic range of each test. Once normalised, comparisons between the variability of osmolarity and other objective tests were done using the Mann–Whitney nonparametric rank comparison test, and differences were considered significant at p<0.05, with a 95% confidence interval.

Patients included

52 patients completed the study (n=16 mild or moderate DED; n=36 severe DED; age 47.1±16.1 years).

Results

Tear osmolarity (8.7±6.3%) had significantly less variability over a 3‑month period than corneal staining (12.2±8.8%; range p=0.029; variability p=0.040), conjunctival staining (14.8±8.9%; range p=0.0035; variability p=0.002), and meibomian grading (14.3±8.8%; range p=0.0001; variability p<0.0001) across the entire patient population.

Osmolarity also showed less variation than TBUT (11.7±9.0%; p=0.059), Schirmer's tests (10.7±9.2%; p=0.67), and OSDI (9.3±7.8%; p=0.94), although the differences were not significant.

Variation in osmolarity was less for mild dry eye patients (5.9±3.1%) than severe dry eye patients (10.0±6.9%; p=0.038).

After treatment, average osmolarity and variability were lowered from 341±18 mOsm/litre to 307±8 mOsm/litre (n=10; p<0.0001).

A downward trend in symptoms followed changes in osmolarity, declining from 44±17 mOsm/litre to 38±18 mOsm/litre (p=0.35). None of the other signs changed after treatment.

Conclusions

Over a 3‑month period, tear film osmolarity had the lowest variability among commonly used signs of dry eye disease. Osmolarity dropped before changes in symptoms during therapy.

Abbreviations: DED, dry eye disease; mOsm, milliosmole (one‑thousandth of an osmole); OSDI, Ocular Surface Disease Index; TBUT, tear breakup time.

Table 6 Overview of the Tomlinson et al. (2010) study

Study component

Description

Objectives/hypotheses

To compare TearLab with the Clifton osmometer (Clifton Technical Physics, USA).

Study design

Prospective, single centre, single visit comparative study. No follow‑up period was reported.

Setting

The setting was not described.

Inclusion/exclusion criteria

Inclusion (DED):

  • noninvasive TBUT <10 seconds using the HirCal grid10

  • ≥2 positive symptoms using McMonnies questionnaire

  • Schirmer's test score of ≤8 mm in 5 minutes.

Inclusion (controls):

  • asymptomatic

  • Schirmer's test score of ≥15 mm in 5 minutes

  • TBUT of >10 seconds.

Primary outcomes

To compare the OcuSense TearLab osmometer with the Clifton osmometer to determine the comparability of results between the instruments and the diagnostic efficacy of each test for dry eye.

Statistical methods

All data were tested for normality using a Shapiro–Wilk test. A 2‑sampled t test was applied to the data. Bland–Altman analysis was used to assess the level of agreement between the results with the OcuSense TearLab osmometer and Clifton osmometers.

Patients included

Mild to moderate DED: n=15 (41.7+16.9 years; 9 women, 6 men).

Controls: n=21 (35.0+12.8 years; 12 women, 9 men).

Results

Osmolarity values measured with TearLab were 308+6 and 321+16 mOsm/litre for controls and dry eye, respectively, and those measured with Clifton were 310+7 and 323+14 mOsm/litre for controls and dry eye, respectively; these values between patients and controls were significantly different.

Significant correlation was found between TearLab and Clifton measurements (r=0.904; p=0.006).

Bland–Altman analysis showed agreement between techniques; most points were within the 95% confidence limits, and actual values differed by less than 1%.

A cut‑off value of <316 mOsm/litre, taken from the distribution of osmolarity values, was used to diagnose DED with an effectiveness of 73% sensitivity, 90% specificity, and 85% positive predictive value for the TearLab test and 73% sensitivity, 71% specificity, and 65% positive predictive value for the Clifton device in the study samples.

Conclusions

TearLab is a suitable alternative test for diagnosing DED and has the potential to become the gold standard.

Abbreviations: DED, dry eye disease; mOsm, milliosmole (one‑thousandth of an osmole); TBUT, tear film breakup time.