End‐of‐day assessment of asymptomatic versus highly symptomatic soft contact lens wearers

To investigate differences in key clinical parameters between asymptomatic and highly symptomatic soft contact lens (CL) wearers after 14 h of wear.

populations studied. 5[9][10][11] Discomfort towards the end of the daily lens-wearing period is particularly common, and this 'end-of-day discomfort' has been well-documented in lens wearers, especially for soft lenses [12][13][14][15] which currently account for approximately 86% of worldwide fits. 16Comfort has been found to show a continual and significant decline over the lens-wearing day in symptomatic subjects, 14,15 and Papas et al. 17 showed that removing or replacing a lens (hydrogel or silicone hydrogel) half-way through the day has no significant effect on the reported final end-of-day comfort.This suggests that the underlying cause of discomfort is related to changes occurring in the ocular tissues (as a result of lens wear) rather than changes to the CL itself.The work of Woods et al. 15 also demonstrated that lens material had no effect on end-of-day comfort scores.Building on the work of Papas et al., 17 further work has confirmed that removing or replacing a lens had no significant effect on end-of-day comfort in a group of habitual symptomatic soft lens wearers over a 12-h period. 14The addition of a 'scleral swish' intervention, where the lens was moved from the cornea onto the bulbar conjunctiva and back again following a few blinks (in an attempt to replenish the post-lens tear film), also had no effect on final lens comfort.Other work has established that it is the length of time that a lens is worn rather than the time of day that CL wear started, which affects end-of-day discomfort. 12espite typical daily lens use of more than 12 h, 5,13,18 there are relatively few reports in the literature which investigated clinical performance after more than 8 h; presumably because of the logistical difficulties of organising clinical examinations at antisocial times for both investigators and participants alike.Little is known about the lens surface, ocular surface or tear film at the point just prior to lens removal following a full day of lens wear.It is possible that more marked responses are seen at this timepoint, which in turn may provide further information on the aetiology of CL discomfort, which is likely to be complex and multifactorial.Many studies are confounded by including different lens types, which should ideally be avoided.Investigations carried out at the true 'end-of-day' may allow for shorter clinical studies with fewer subjects due to the potential for more marked clinical responses.
In order to address this gap in our knowledge, this work set out to investigate whether there were differences in five key clinical measures (blinking characteristics, tear meniscus height (TMH), non-invasive tear break-up time (NIBUT), tear film osmolarity and eyelid margin staining) between asymptomatic and highly symptomatic soft lens wearers over the course of a single, typical lens-wearing day (~14 h).These metrics were chosen based on their likelihood of being related to CL discomfort. 19One of the most utilised instruments for identifying asymptomatic and symptomatic lens wearers is the Contact Lens Dry Eye Questionnaire-8 (CLDEQ-8). 20Its authors have recommended a cut-off score of 12 to differentiate between these two groups 21 ; however, larger differences in clinical responses to CL wear may be obtained by separating the symptomatology of these groups further, that is, by recruiting towards the lower and upper ends of the CLDEQ-8 scoring system.Such an approach effectively eliminates the 'middle ground'.
The work was carried out in two phases: Phase 1 aimed to identify subjects who were asymptomatic or highly symptomatic when newly fitted with the same daily disposable hydrogel CL.Phase 2 aimed to investigate differences in the five key clinical measures between the two groups over the course of a full wearing day (~14 h) and to investigate whether there was a relationship between subjective scores of comfort and these key clinical measures.

Study contact lenses
Nelfilcon A CLs (Dailies AquaComfort Plus, Alcon Inc., myalc on.com) were worn by all subjects in both Phase 1 and Phase 2 of the study.The lenses had a back optic zone radius of 8.7 mm and a total diameter of 14.0 mm.This lens was chosen as being a representative example of a modern, commonly prescribed daily disposable CL.Selecting a daily disposable lens eliminated any potential solutionrelated discomfort interactions.Lenses were worn bilaterally and subjects were asked not to use any ocular comfort drops during the study.

Study design
This was a pilot study and all hypotheses were exploratory in nature.No previous data were available for an a priori

Key points
• Five key clinical parameters were compared in asymptomatic versus highly symptomatic contact lens wearers who wore the same lens type over 14 h (reflecting real-world, typical daily lens use).• Upper eyelid margin staining ('lid wiper epitheliopathy') was greater and lower eyelid tear meniscus height was decreased in symptomatic subjects.• Better contact lens comfort scores were associated with lower levels of eyelid margin staining and reduced blink rate.
power analysis calculation over a 14-h wearing period.It was considered that approximately 20-25 subjects for each study group completing Phase 2 would constitute a valid and meaningful investigation.
The work was conducted in two phases which is summarised in Figure 1.Phase 1 was a single-arm, nonrandomised, open-label, bilateral dispensing clinical investigation which was designed to identify asymptomatic and highly symptomatic subjects after being fitted with nelfilcon A CLs for a 2-week period.Phase 2 was a nonrandomised, open-label, bilateral dispensing clinical investigation where eligible asymptomatic and symptomatic subjects from Phase 1 wore nelfilcon A CLs for approximately 14 h on a single study day.Both subjects and investigators were masked as to whether subjects were in the asymptomatic or symptomatic group in both phases of the study.Ethical approval was granted from the University Research Ethics Committee of The University of Manchester prior to subject recruitment.The study conformed to the tenets of the Declaration of Helsinki, and all subjects provided written, informed consent prior to enrolment.
In Phase 1, subjects attended two visits, 2 weeks apart.All subjects were required to be habitual wearers of any daily disposable CL, and further inclusion and exclusion criteria are outlined in Table 1.At Visit 1, various baseline investigations were carried out including subjective refraction, logMAR high contrast visual acuity and slit lamp biomicroscopy.Eligible subjects were fitted with the study CL in both eyes and after 5 minutes, comfort was recorded using visual analogue scales (VAS).For each eye separately, subjects placed an 'X' on a paper-based vertical scale with two extreme anchors ('worst comfort imaginable' and 'best comfort imaginable').The resulting comfort score was measured using a custom measuring scale which had been calibrated for the length between the two anchors on the paper copy.Subjects were discharged from the clinic and asked to return 2 weeks later, after wearing the study lenses for at least 12 hours per day.At Visit 2, subjects completed a CLDEQ-8 questionnaire regarding their experience with the study CLs. 20,21Comfort VAS scores were also recorded for 'after lens application' and 'before lens removal' over the 2 week period.Following lens fit assessments, logMAR high contrast visual acuity and slit lamp biomicroscopy, subjects were exited from Phase 1.In order to separate the symptomatology of the two groups as much as possible, subjects were considered 'asymptomatic' and 'symptomatic' if their CLDEQ-8 score was ≤10 or ≥20, respectively.
Phase 1 aimed to recruit approximately 20-25 subjects from each of the two groups into Phase 2. Those with a CLDEQ-8 score of ≤7 as well as those with a score of ≥20 were automatically invited to take part in Phase 2. For asymptomatic subjects with a score of 8-10, the point at which they were invited to take part in Phase 2 was considered on an ongoing basis.Those with a score of 11-19 were discharged from the study and did not take any further part.
In Phase 2, subjects attended two visits over a single day, where they wore a pair of nelfilcon A CLs for 14 ± 2 h.At Visit 1, subjects attended in the morning without lenses in situ and completed a new informed consent form.Baseline measures were recorded in the following order (least to most invasive with some tests performed in one eye only): high-and low-contrast logMAR visual acuity, blink rate and completeness, TMH, NIBUT, tear film osmolarity, slit lamp biomicroscopy and assessment of eyelid margin staining.All parameters (except for eyelid margin staining) were assessed at baseline (i.e., prior to lens application), after lens fitting and after 14 h of lens wear.Eyelid margin staining was assessed at baseline and after 14 h.More details of the methodology for each of the five key clinical measures are outlined below.
A pair of nelfilcon A CLs was applied, and after 5 min comfort was recorded in each eye using the VAS outlined in Phase 1.After high-and low-contrast logMAR visual acuity and lens fit/surface assessment, the following were carried out with lenses in situ: blink rate and completeness, TMH, NIBUT and tear film osmolarity.Lenses were then removed.Following the application of a new lens pair, the subject was instructed to present for Visit 2, 14 ± 2 h later.
At Visit 2, the following were recorded with lenses in situ: VAS comfort for each eye, high-and low-contrast logMAR visual acuity, lens fit/surface assessment, blink rate and completeness, TMH, NIBUT and tear film osmolarity.Following lens removal, slit lamp biomicroscopy and assessment of eyelid margin staining were carried out before exiting subjects from the study.

Blinking characteristics
Spontaneous blinking was recorded using a custom noninvasive imaging system which has been described in detail previously. 22In brief, a high-speed infrared camera (capturing video at 500 frames per second) was used to record both eyes while a subject watched a 3-min wildlife video clip.The choice of infrared camera prevented the reduction in palpebral aperture present when a white light was used to illuminate the face, 22 and the video content was selected to require low levels of concentration as the type of task can impact blink rate. 23An iPad Teleprompter system (autoc ue.com) was used such that the subject viewed a reflected image of the iPad screen which was positioned approximately 80 cm from the ocular plane.The subject's face was illuminated with an 850 nm infrared lightemitting diode (LED) source (LIU850A, thorl abs.com), and the recording took place in a room with ambient lighting of approximately 300 lux (following a short adaptation period).Both camera and optics were not visible to the subjects during image capture and the purpose of the task was not explicitly divulged in order to avoid unnatural blinking patterns.
Custom semi-automated image analysis (MATLAB, The MathWorks Inc., uk. mathw orks.com) was used to assess blink dynamic metrics from the right eye. 22The first minute of recording was not analysed in order to capture blinks which would represent as natural a pattern as possible.Blink rate was calculated over the remaining 2 min of the recording and the percentage of complete blinks calculated for the first 10 blinks.A complete blink was defined as one where the percentage closure of the eyelids was twothirds or more of the full palpebral aperture height. 24,25

Tear meniscus height
Tear meniscus height was measured non-invasively using a spectral domain optical coherence tomography (SD-OCT) system.The 3D-OCT 2000 (Topcon, topconhealthcare.eu)together with the forehead attachment for anterior segment imaging was used to capture a cross-sectional image of the central tear meniscus using a 3.0-mm anterior line segment scan (scan resolution 1024).The scan was aligned with the centre of the pupil such that it intersected the cornea, tear meniscus and the eyelid.Three good quality images were captured immediately after a blink for the upper and lower tear menisci in both eyes.
Exported images were analysed using custom semiautomated image analysis (MATLAB, The MathWorks Inc., uk. mathw orks.com), where a manual vertical line tool was used to mark the upper and lower extent of the tear meniscus.Each pixel was calibrated to be equivalent to approximately 0.0029 mm, and the height in pixels was automatically converted into mm to give the final measurement.

Non-invasive tear break-up time
NIBUT was measured using a Keeler Tearscope-Plus™ (keeler.co.uk) which consists of an observation tube surrounded by a cold cathode light source allowing the tear film to be viewed in specular reflection. 26The instrument was used on its brightest setting, without a grid, in conjunction with a slit lamp observation system set at 12× magnification.Subjects were asked to blink normally three times and on the third blink to keep their eyes open (without widening the eyes to 'stare').The time taken from the last eye opening to the formation of the first break in the tear film was measured using the integrated timer.This was repeated three times on the right eye.If the subject blinked before the observation of a break, the time to blink was recorded as a valid NIBUT measurement.

Tear film osmolarity
Tear film osmolarity was measured using the TearLab™ Osmolarity System (Trukera Medical, truke ra.com).This is an impedance-based osmometer designed to be used in a clinical setting which samples approximately 50 nL of tear fluid when used.Two consecutive measurements 27,28 were taken from the outer lower tear meniscus of the right eye using single-use test cards.Care was taken to avoid inducing reflex tearing.Daily calibration was performed in accordance with the manufacturer guidelines.

Eyelid margin staining
Lissamine green was applied to the left eye only and eyelid margin staining was graded subjectively.A moistened lissamine green strip (GreenGlo, HUB Pharmaceuticals, hubrx.com) was applied to the inferior bulbar conjunctiva and repeated with a new strip, 5 min later.Upper eyelids were everted and eyelid margin staining was graded using grading scales developed by Korb et al., 29,30 where 'horizontal length' and 'sagittal width' were graded separately from 0 to 3. The average of these two grades was calculated to produce an 'average severity' score for each eyelid margin separately, 30 as follows: no eyelid margin staining (analogous to 'lid wiper epitheliopathy'((LWE); Grade 1 LWE (mild; 0.25-1.0);Grade 2 LWE (moderate; 1.25-2.0)and Grade 3 LWE (severe; 2.25-3.0).The same grading was carried out for the lower eyelid margin.Comparison between the asymptomatic and symptomatic groups at baseline (where relevant), fitting and at 14 h, as well as between timepoints was carried out using 95% confidence intervals (CIs) constructed from least square mean (LSM) differences from the linear mixed model.It was considered that there was a significant difference between groups or timepoints if the 95% CI did not include zero (or 1, if a ratio).Given the exploratory nature of the hypotheses, no adjustment for multiple comparisons was made to control for Type I errors.This approach would minimise the risk of making Type II errors, which in turn, would avoid missing potential important findings that warranted further exploration.

Statistical analysis
Pearson's correlation analyses were conducted between the clinical parameters and VAS comfort scores of the corresponding eye at the 14h follow-up.For TMH and blink metrics, the average VAS score (of the two eyes) was used in the analysis.Tests for the relationship between VAS comfort scores and the parameter were conducted using two-sided 95% CI constructed from the corresponding correlation coefficient.It was considered that there was a significant correlation if the 95% CI did not include zero.
Post-hoc power analysis calculations were carried out for any parameter where no statistically significant difference was found between groups after 14 h.Calculations were made for two independent sample means (twosided at the 0.05 significance level) using the study group sample sizes and the standard deviation of the parameter mean for each group.As a count variable, blink rate was analysed with a generalised linear mixed model using Poisson distribution.For this reason, the power analysis calculation for this parameter was made for two independent sample means with Poisson distributions (two-sided at the 0.05 significance level) using the study group sample sizes and the mean blink rate of the asymptomatic group (as the reference group).Power calculations were solved for what difference would have been detectable at the 80% level.

R ESULTS
In total, 161 subjects were enrolled into Phase 1 with 147 subjects completing this phase.This resulted in the identification of 74 asymptomatic and 21 symptomatic subjects according to the CLDEQ-8 criteria set.It was possible to select only asymptomatic subjects with a CLDEQ-8 score ≤7 to proceed onto Phase 2. The distribution of CLDEQ-8 scores obtained after 2 weeks of nelfilcon A CL wear is shown in Figure 2. Twenty-five asymptomatic and 17 symptomatic subjects proceeded onto Phase 2, and all 42 subjects completed the second phase of the study.Table 2 details the demographics of subjects enrolled into Phase 2. No serious adverse events were reported during either phase.

Blinking characteristics
Timepoint was the only statistically significant factor in the analysis of blink rate (F 2,115 = 38.7,p < 0.0001).Post-hoc testing showed a statistically significant difference for baseline versus fitting (baseline/fitting rate ratio: 0.79, 95% CI: 0.72, 0.87), baseline versus after 14 h (baseline/after 14 h rate ratio: 0.67, 95% CI: 0.61, 0.73) and fitting versus after 14 h (fitting/after 14 h rate ratio: 0.84, 95% CI: 0.78, 0.92) for all subjects with no statistically significant differences between the two groups.No factors were significant in the analysis of the proportion of complete blinks.
After 14 h, the blink rate in the symptomatic group was 1.20 times that of the asymptomatic group (symptomatic/ asymptomatic rate ratio 1.20, 95% CI: 0.91, 1.57) and the difference in the proportion of complete blinks between the symptomatic and asymptomatic groups was −7.27 (95% CI: −29.31, 14.76).Post-hoc power analysis showed that at 80% power, this study would have been able to detect a blink rate ratio of 1.17 and a 34% difference in the proportion of complete blinks between groups after 14 h.
After 14 h, the difference in NIBUT between symptomatic and asymptomatic groups was −1.70 (95% CI: −7.01, 3.62).Post-hoc power analysis showed that at 80% power, this study would have been able to detect a difference in NIBUT of 4 s between groups after 14 h.

Eyelid margin staining
The analysis of upper eyelid margin staining showed a statistically significant effect of the group*timepoint interaction (F 1,40 = 5.7, p = 0.02).Upper eyelid margin staining was greater after 14 h in symptomatic than in asymptomatic subjects (LSM difference: 0.53, 95% CI: 0.01, 1.05) and this was greater after 14 h than baseline in the symptomatic group only (LSM difference: 0.61, 95% CI: 0.16, 1.07).
There was a statistically significant effect of timepoint on lower eyelid margin staining (F 1,40 = 7.0, p = 0.01).Increased lower eyelid margin staining grade was observed after 14 h compared with that at baseline for all subjects (LSM difference: 0.36, 95% CI: 0.09, 0.64).
After 14 h, the difference in lower eyelid margin staining between symptomatic and asymptomatic groups was 0.15 (95% CI: −0.38, 0.68).Post-hoc power analysis showed that at 80% power, this study would have been able to detect a difference in lower eyelid margin staining of 0.66 between groups after 14 h.

Relationship between comfort and clinical measures
After 14 h, there was no significant relationship between VAS comfort and the clinical measures except for upper eyelid margin staining (r = −0.40,95% CI: −0.63, −0.11) and blink rate (r = −0.31,95% CI: −0.57, −0.003).Higher levels of upper lid margin staining and increased blink rate were associated with reduced VAS comfort (Table 3).

DISCUSSION
Phase 1 successfully identified a group of asymptomatic and highly symptomatic subjects fitted with nelfilcon A daily disposable CLs using the CLDEQ-8 questionnaire.A large number of subjects (161) were needed in this part of the study to find sufficient symptomatic subjects who had a CLDEQ-8 score of ≥20.This demonstrates that in studies of this nature (i.e., when subjects are put into a new daily disposable lens), finding asymptomatic subjects is relatively straightforward (74 out of 161 giving a 46% 'success rate'), but significant effort/expense is required to find highly symptomatic subjects (21 out of 161, i.e., a 13% 'success rate').A simpler approach could have been to recruit subjects who were asymptomatic or highly symptomatic in their habitual CLs and to continue using their habitual lenses in the study; however, the advantage of having all subjects wear the same lens type throughout the study meant that confounding factors that have impacted research in this field, such as differences in lens material and design, were avoided.Alternatively, the cut-off score for the symptomatic group could have been reduced from 20 to a lower value on the CLDEQ-8 questionnaire as advocated by Chalmers et al., 21 but avoiding overlap in subject symptomatology was considered an important aspect of the current study design.

Blinking characteristics
Blinking plays a critical role in the spreading of the tear film over the ocular surface from the menisci, and the characteristics of the blink may be influenced by factors such as the amount of lubrication provided by the tear film as well as disruption of the tear film sensed by the corneal nerve fibres.Blink rate is known to increase with CL wear compared with no-lens wear, 23,24 and this was shown in the current work by an increase in the blink rate from pre-lens baseline (LSM 19 blinks/minute) to post-lens fitting (LSM 24 blinks/ minute) and again from fitting to 14 h (LSM 28 blinks/minute).This also suggests that the blink rate increases with greater discomfort.The blink rate in this study (circa 23-30 blinks/ min) was similar to that found in previous work investigating blink rate in soft lens wearers, [31][32][33][34] and any differences between studies are likely to be due to differing methodology.
Despite the variations in symptomatology, no significant difference could be demonstrated between the two groups at any timepoint.After 14 h, the blink rate in the symptomatic group was 1.20 times that of the asymptomatic group.This study was powered sufficiently to detect a blink rate ratio of 1.17.However, despite the commonly held belief that post-hoc power estimates may be appropriate when a statistically non-significant result is observed, the usefulness of this approach has been challenged, with some researchers arguing that post-hoc power calculation is problematic and may not provide meaningful information. 35,36It is probably more important to focus on the observed difference and its confidence interval; this information is presented in Table S1, to assist in the interpretation of the study findings.There is currently no information in the literature on differences in blink rate between these groups to give an indication of what a meaningful clinical difference would be.The literature shows that in dry eye disease, the difference between individuals without and with symptoms ranges from 13 to 29 blinks per minute, [37][38][39] but these values are not necessarily applicable to the lenswearing situation.The blink rate is thought to be associated with corneal sensitivity and disruption of the tear film in the non-lens wearing eye, 31 and some studies have found higher blink rates in lens wearers with greater corneal sensitivity. 32These authors also found increased blink rates in lens wearers with increased dryness at the end of the day.Kojima et al. 33 observed that increased blink rate was related to deterioration in comfort and to increased tear osmolarity, evaporation and tear film instability.In the current work, we observed a statistically significant negative correlation between VAS comfort scores and blink rate after 14 h (i.e., lower comfort equals higher blink rate) which is in agreement with the correlation analysis carried out by Martin-Montañez et al. 32 This result is potentially important because blinking characteristics may provide information about the tear film indirectly.
There was no change in the percentage of complete blinks over a lens-wearing day and no difference between the two groups at fitting or after 14 h.We found a difference of 7% between groups after 14 h and the current work (fewer complete blinks in the symptomatic subjects) was powered sufficiently to detect a difference of 34%.Again, there is no information in the literature on what a meaningful clinical difference would be for these subject groups.Certainly, given these findings it seems likely that this parameter is not a useful metric for distinguishing symptomatology.Previous work has shown that the proportion of incomplete blinks (and therefore also complete blinks) is similar in non-lens and lens wearers and is correlated with the degree of corneal staining. 40Incomplete blinks are thought to increase the effective inter blink period and can have a detrimental effect on the CL surface (reduced wettability, increased tear break-up 41 ) and also potential negative effects on the wiping, eyelid margin epithelial surface.It may be that the differences between our data and those published previously are due to methodology, for example, the type of task used or indeed the precise definition of the term.Further work investigating more intricate aspects of the blink (such as opening/closing speed or duration of various phases of the blink) may elucidate differences between groups with different symptomatology.

Tear meniscus height
Sufficient tear volume is essential in order to maintain an adequate pre-and post-lens tear film, and most of this volume is held within the tear menisci and spread across the ocular/ lens surface during blinking.The TMH gives a direct indication of the tear meniscus volume.Reduced tear volume is likely to lead to inadequate lubrication between the eyelid margin conjunctival epithelium and the lens surface, which may ultimately result in both visual and comfort problems.In the current study, the TMH of the upper and lower eyelids showed a decrease during the course of the lens-wearing day, which has been reported previously using OCT over a 10 h lens wearing period. 42Glasson et al. 43 showed that in the absence of CLs, intolerant lens wearers have reduced TMH compared with a tolerant group.Although the highly symptomatic lens wearers in the current investigation were not 'intolerant', no differences in TMH were evident prior to lens wear in the two groups investigated.
The lower TMH did not differ between the baseline and fitting measurements, and the upper TMH showed reduced values post-fitting compared with baseline, which suggested that the tear film recovers quickly following lens application.Previous work using OCT has shown that tear meniscus volume increases immediately after lens application, 44 which would be expected as a result of reflex tearing and/or additional liquid being applied to the eye during the process, but recovered to baseline levels after 30 min.The timeframe in the current work is similar to this, that is, TMH images were taken 20-30 min after the initial lens application.Yokoi et al. 45 also noted an immediate increase in tear volume following lens application which quickly settled after 5 min and remained similar over a 6-h period.Chen et al. 44 observed reduced lower eyelid tear meniscus volume in symptomatic lens wearers than both asymptomatic lens wearers and asymptomatic non-wearers before lens application and after 30 min.In the current work, no differences between symptomatic and asymptomatic subjects were seen at baseline, but there were significant differences after 14 h.This difference (LSM 0.04 mm) is likely to be clinically meaningful given measurements ranging from 0.1 to 0.3 mm in this work and those of previous studies with CLs using OCT. 46,47Siddireddy et al. 48demonstrated significantly reduced lower eyelid TMH in symptomatic compared with asymptomatic subjects in a single-visit, cross-sectional study using images captured with an Oculus Keratograph 5 M (oculus.de/us/products/keratograph-5m/), which is in agreement with the current work.This is in contrast to the findings of Glasson et al. 49 who observed an increase in TMH in tolerant wearers before and after 6 h of lens wear and no change in intolerants over the same period of time.Chen et al. 42 also found that comfort ratings were positively correlated with tear meniscus volume in self-reported symptomatic and asymptomatic CL wearers, but a significant relationship was not found in the present work.

Non-invasive tear break-up time
A decrease in NIBUT was observed in both groups after fitting and 14 h compared with baseline, but no differences between the two groups were found at any timepoint.This decrease in NIBUT over the course of a day is in line with previous reports on CL wearers, 43 and has also been shown to occur in non-lens wearers. 50However, the reliability of this parameter as an indicator of ocular surface symptoms is less certain.Our data showed a trend towards the asymptomatic group undergoing a greater change over the day than the symptomatic group, but because the timepoint*group interaction in the main statistical model was not significant, it is not possible to conclude this formally.Some studies have been able to show a difference in NIBUT between those with and without symptoms in lens wear, 43,51,52 while others have not. 49,53The reduced tear stability over a CL surface is thought to be due to how the lens impacts the normal structure of the pre-corneal tear film.When a lens is on the eye, the membrane bound mucins of the glycocalyx are mostly hidden from that portion of the tear film formed over the front surface of the lens, resulting in a more weakly anchored tear film as well as altered composition of the lipid layer. 54lthough there was no statistically significant difference between the NIBUT of asymptomatic and symptomatic subjects prior to lens application (at baseline), there was a trend for a reduced NIBUT in the symptomatic compared to asymptomatic subjects (LSM: 15.3 s vs. 19.9s, respectively).However, in their work Glasson et al. 49 found a significant difference in NIBUT (using a similar Tearscope set-up as in the current work) between tolerant and intolerant lens wearers prior to lens wear and advocated the predictive power of this parameter in the likely success of lens wear.We found a difference in NIBUT of 1.3 s between the groups after 14 h (lower in the symptomatic group) and our work was powered sufficiently to detect a difference of 4 s after 14 h.Currently, the literature is inconclusive on what a meaningful clinical difference would be for this parameter.Guillon et al. 52 also found a difference of 1.3 s in the pre-lens NIBUT using a Tearscope when a subset of subjects with the lowest and the highest Ocular Surface Disease Index (OSDI) scores were compared.Although not statistically significant, Kim et al. 53 found a difference in pre-lens NIBUT of 1.2 s after 6 h of lens wear in those with and without symptoms.Both are remarkably similar to the current work.In the Tear Film and Ocular Society (TFOS) DEWS II report, Wolffsohn and co-workers 55 suggested using a clinically meaningful difference in pre-corneal NIBUT of 5 s (for normal vs. dry eye) when powering clinical studies.It is likely that a difference closer to about 1 or 2 s is more appropriate in CL discomfort.
NIBUT was not correlated with comfort scores, although reduced tear break-up time has been associated with increased symptoms of discomfort in previous studies. 51,56,57uillon et al. 52 suggested that NIBUT per se may be too crude a measurement to differentiate wearers with and without discomfort and indicated that the use of other prelens tear film kinetic metrics such as the amount of surface coverage during the interblink period and surface exposure at the time of a blink may be more useful.

Tear film osmolarity
Osmolarity in the symptomatic subjects was lower than that in the asymptomatic subjects at fitting but similar at baseline and after 14 h.Perhaps as a consequence of this hypo-osmolarity, there was a significant difference (i.e., an increase in osmolarity) between fitting and 14 h in the symptomatic group only, that is, symptomatic subjects underwent a bigger change over the day than the asymptomatic wearers.The finding of hypo-osmolarity in the symptomatic group after fitting is somewhat counterintuitive given that hyper-osmolarity has previously been associated with CL discomfort as well as dry eye. 55,58owever, this could have occurred as a result of increased tearing in the symptomatic group or perhaps an inherent inability of these wearers to regulate their tear film homeostasis after application of a CL.Another possibility is that change in osmolarity could drive the symptoms, and future studies involving CL discomfort, particularly end-of-day discomfort should include this metric.We found a difference in osmolarity close to zero between the two groups after 14 h and this work was powered sufficiently to detect a difference of 16 mOSm/L at this timepoint.A meaningful clinical difference for this parameter probably lies somewhere close to this value.Some investigations have previously reported differences of around 10 mOSm/L in lens wearers with and without symptoms, 43,54 whereas differences closer to about 30 mOSm/L may be more appropriate to differentiate normal and dry eye subjects in the absence of lens wear. 28,59Stahl et al. 60 demonstrated a relationship between comfort and CL osmolarity but not between comfort and tear osmolarity, which supports our finding of a lack of correlation between comfort VAS scores and tear film osmolarity.

Eyelid margin staining
Lower eyelid margin staining was similar at baseline and after 14 h for both groups.In general, staining of the lower eyelid increased over the day for both groups.For the upper eyelid, there was significantly more staining after 14 h in the symptomatic than in the asymptomatic wearers, and grades were significantly greater after 14 h compared with baseline for the symptomatic wearers only.Differences of 0.53 and 0.61 Korb grading units for each of these changes, respectively, lie short of the recommended 1.0 grade for a clinically significant change, in work which compared this scale to a new photographic LWE scale. 61In general, however, these results lend support to Korb's original hypothesis 62 that discomfort in CL wearers is caused by reduced fluid film lubrication leading to microtrauma at the eyelid margin, as evidenced with vital dyes.However, to date, there remains no direct evidence that (a) increased friction occurs in symptomatic lens wearers despite findings in vitro demonstrating that different CL surfaces can induce differing amounts of damage to mucinated epithelial cells, 63 (b) lissamine green is actually stainingdamaged cells or cells devoid of mucin coverage in the lid wiper area 64 and (c) trauma occurs at the eyelid wiper. 65,66][69] To lend further support to Korb's hypothesis, the current work demonstrated a significant correlation between VAS comfort scores and upper eyelid margin staining.Higher staining grades were associated with reduced levels of comfort.There are a plethora of publications in the literature which are equivocal, with some finding no association between eyelid margin staining and comfort [70][71][72] and others finding an association 73 or a difference between symptomatic and asymptomatic populations. 74,75his should not be taken to mean that the literature is balanced; rather, those studies showing a difference should generally be taken more seriously than those which did not as, for example, studies which fail to show a difference or an association may be doing so as a result of a lack of statistical power or problems with methodology (notwithstanding the fact that the precise methodologies reported should be scrutinised).For example, many studies, like the current investigation, have used CLDEQ-8 scores as a means to classify subjects into symptomatic and asymptomatic groups.However, no published studies have separated the two groups by eliminating the 'middle ground' CLDEQ-8 scores (i.e., values between 8 and 19) which arguably lends more weight to the current work as the two populations are more likely to be different.No studies to our knowledge have reported investigating eyelid margin staining after 14 h in asymptomatic/ symptomatic wearers, which is when the true end of the CL-wearing day is likely to be.The results indicate that eyelid margin staining can differentiate these groups if the methodology is robust.

Limitations
In this work, only one lens brand was evaluated and it is not possible to extrapolate the findings to other lens designs or materials, including silicone hydrogel materials.Only 1 day of lens wear was evaluated due to the complexity of the study design, which may not be fully representative of the lens-wearing experience.More complex methodologies of assessment may be needed to elucidate fully the underlying mechanisms of CL discomfort including employing larger sample sizes to validate the results from this pilot investigation.

CONCLUSIONS
This work has demonstrated the importance of evaluating the ocular surface and tear film at the time when CL discomfort is occurring.By utilising this approach, this study has shown that of the parameters investigated, the most effective in differentiating asymptomatic from symptomatic wearers after 14 h of lens wear were upper eyelid margin staining and lower TMH.Symptomatic subjects underwent a greater change in upper eyelid margin staining, tear film osmolarity and TMH than asymptomatic subjects over the course of a 14 h wearing day.The parameter with the strongest relationship to comfort was upper eyelid margin staining, with higher scores being associated with lower levels of lissamine green staining.VAS comfort scores were not associated with any of the other clinical measures investigated except for blink rate, where increased blink rate was associated with reduced comfort.

AC K N O W L E D G E M E N T S
We thank our clinical, logistical and administrative colleagues at Eurolens Research for their input into the acquisition of data for this study.We also thank Ross Franklin at Johnson & Johnson Vision Care, Inc. for his support and clinical insight.

CO N F L I C T O F I N T E R E S T S TAT E M E N T
The authors have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Carol Lakkis is a former employee of Johnson & Johnson Vision Care, Inc.

F I G U R E 3
Phase 2: Visual Analogue Scale (VAS) subjective comfort.Least square mean ± 95% confidence interval (Asympt, asymptomatic and Sympt, symptomatic).F I G U R E 4 Phase 2: key clinical parameters.Least square mean ± 95% confidence interval.Asympt, asymptomatic; LWE, lid wiper epitheliopathy; NIBUT, non-invasive tear break-up time; Sympt, symptomatic; TMH, tear meniscus height.T A B L E 3 Correlation coefficient estimates and their 95% confidence intervals between VAS comfort scores and each clinical parameter.

FU
N D I N G I N FO R M AT I O N This work was funded by Johnson & Johnson Vision Care, Inc.
Phase 1: frequency distribution of Contact Lens Dry Eye Questionnaire-8 (CLDEQ-8) scores after 2 weeks of nelfilcon A lens wear.Blue bars indicate scores eligible to proceed onto Phase 2. Subject demographics (Phase 2).