Maximum blink interval is associated with tear film breakup time: A new simple, screening test for dry eye disease

The prevalence of dry eye disease (DED) is increasing worldwide, and its diagnosis often needs dedicated reagents and machines. We investigated the usefulness of maximum blink interval (MBI) (the length of time that participants could keep their eyes open) in screening for DED. This cross-sectional study included 292 patients (194 with DED and 98 without DED) recruited between September 2016 and September 2017. We compared the MBI between patients with and without DED; examined correlations between MBI and other clinical features of DED, including subjective symptoms (Dry Eye-Related Quality-of-Life Score), tear film breakup time (TFBUT), cornea fluorescence score (CFS), and Schirmer test I value; and determined the optimal cutoff value of MBI to suspect DED using a receiver operating characteristic (ROC) analysis. The MBI was significantly shortened in DED group compared to the non-DED group (10.0 ± 9.1 vs. 24.3 ± 38.2 seconds, p < 0.001). TFBUT was strongly positively correlated with MBI (r = 0.464), whereas CFS was negatively correlated with MBI (r = −0.273). The area under the ROC curve was 0.677, and the optimal MBI cutoff value was 12.4 seconds, providing a sensitivity of 82.5% and specificity of 51.0% to suspect DED. In conclusion, MBI may be a simple, useful test for screening DED.

number of seconds the eyes can stay open without blinking 17 , can be simple a method to assess for DED. However, there have been few studies to assess the usefulness of MBI in the assessment of DED and to study the relationship between MBI and other clinical features of DED 15,17,21 .
In this study, we aimed to assess the usefulness of MBI in screening for DED by (i) comparing MBI between non-DED and DED patients, (ii) examining correlations between MBI and other test results for DED assessment, and (iii) conducting a receiver operating characteristic (ROC) analysis to identify the optimal MBI cutoff value to suspect DED. In addition, we assessed the blink interval period (BIP), which is the difference between MBI and TFBUT, to explore the mechanism of MBI.

Results
Participants' characteristics. We enrolled 292 participants in this study. Table 1 shows the patients' characteristics. All participants underwent complete examination and were eligible for analysis. The average age was 62.3 ± 14.9 years, and 82.2% of the participants were women. Based on the Asia Dry Eye Society (ADES) diagnostic criteria 19 , 98 participants were diagnosed as non-DED (33.6%), whereas 194 participants were diagnosed to have DED (66.4%). Positive subjective symptom rates, the Dry Eye-Related Quality-of-Life Score (DEQS) 22 , and corneal fluorescein staining (CFS) scores were significantly higher in the DED patients than in the non-DED participants. TFBUT was significantly lower in the DED patients than in the non-DED participants. The Schirmer test value was not significantly different between the groups.

Maximum blink interval (MBI) and blink induction period (BIP) between non-DED and DED.
MBI and BIP were compared between the non-DED and DED groups (Table 1). MBI was significantly shorter in the DED group than in the non-DED group (9.5 ± 6.5 vs. 14.4 ± 8.9 seconds, p < 0.001). BIP was also significantly shorter in the DED group than in non-DED group (8.4 ± 9.0 vs. 11.2 ± 7.7 seconds, p < 0.001).

Correlations between MBI and other clinical assessment findings in DED.
We examined the relationship between MBI and other clinical assessment findings in DED (DEQS, TFBUT, CFS and Schirmer test Ι) using the Pearson correlation test (Table 2). MBI was strongly correlated positively with TFBUT (r = 0.464) and negatively correlated with CFS (r = −0.273).
The cutoff value of MBI for detecting DED. Figure 1 shows the ROC curve illustrating the balance between sensitivity and specificity of different values of MBI in suspecting DED. The area under the ROC curve (AUC) was 0.677 (95%CI: 0.610-0.744). The optimum MBI cutoff value was 12.4 seconds, providing the sensitivity of 82.5% and the specificity of 51.0%.

Discussion
The new diagnostic criteria for DED have been recommended by ADES, and the diagnosis of DED can be made by subjective symptoms and decreased TFBUT (tear film instability) 19 . In this study, we showed the usefulness of MBI measurement as a simple DED screening method. The evaluation of tear film stability is important in the diagnosis of DED 1,19 , MBI measurement can be useful for many people if it can substitute for the measurement of TFBUT. Our study demonstrated that MBI is strongly correlated with TFBUT (r = 0.464), indicating that tear breakup stresses the corneal surface, stimulates the underlying nociceptors, and induces blinking. Although MBI is thought to be associated with subjective symptoms, our correlation analysis showed that there was no significant correlation between MBI and subjective symptoms (DEQS), suggesting that MBI is independent from subjective symptoms and reflects the influence of TFBUT. Since the ADES diagnostic criteria are based on TFBUT and subjective symptoms, we concluded that MBI is suitable for the new ADES diagnostic criteria because MBI was associated with TFBUT in this study (Table 2).
Our study demonstrated that MBI was significantly shortened in DED patients compared to non-DED patients, which has also been observed in previous studies 16,17 . We have also suggested the optimal cutoff value of MBI (12.4 seconds) for detecting DED according to the ADES criteria. As shown in Table 3, using a cutoff of 12.4 seconds of MBI for DED screening, the positive predictive value was 76.9% and the positive likelihood ratio was 1.68, indicating that MBI could accurately detect patients with DED. On the other hand, since ADES criteria excluded keratoconjunctival epithelial disorder and Schirmer test for the DED diagnosis 19 , MBI may not be suitable to evaluate keratoconjunctival epithelial disorder and tear-deficient DED, indicating that MBI should not be used to judge the severity or classification of DED.    A variety of assessments have been utilized in evaluating DED, including measurement of TFBUT, CFS, Rose Bengal staining, Schirmer test and osmolality measurement 2 ; however, most of these assessments need dedicated reagents and machines. IBI, another evaluation item that is similar to MBI 14 , also decreases in patients with DED; however, self-measurement of IBI is difficult, and the routine blink rate has not been shown to be related to local corneal and conjunctival factors 17 . Since MBI is a noninvasive and simple test, it is useful for self-screening in people with undiagnosed DED, and it offers the opportunity for making an appropriate diagnosis and subsequent proper use of eye drops for DED, including the use of over-the-counter eye drops.
BIP represents the time between the timing of dark spots occurrence on the ocular surface to the blinking (Fig. 1). This study showed that the BIP was reduced in the DED group compared to the one in the non-DED group ( Table 1), suggesting that the instability of the tear film layer in DED shortens not only the TFBUT but also the BIP. A previous study reported that the total area of the tear film breakup was increased in DED 23 , via the mechanism of the decreased BIP due to the stimulation of the underlying nociceptors 17 . Therefore, we proposed that the mechanism of reduced both TFBUT and BIP (=MBI) is an increase in the number of blinking in patients with DED. Since it is expected that BIP will decrease in the evaporative-type DED, it may be important to study the BIP depending on the type of DED.
This study has a few limitations. This study may have selection bias because it was conducted in a university hospital in Japan and there are more female participants likely due to the predominance of females affected by DED 24 . We did not exclude patients with systemic diseases and systemic treatments. In addition, we did not measure Rose Bengal stain scores, tear osmolality, meibomian gland dysfunction, and corneal sensations to classify the DED in all participants because we wanted to develop a simple screening test that does not require dedicated reagents and machines.
In summary, the results of this study suggest MBI is useful in screening for DED. Specifically, the MBI of 12.4 seconds or shorter was found to suggest a diagnosis of DED based on the 2016 ADES criteria. MBI may be useful for people with undiagnosed DED, and it offers the opportunity for conducting more specific examinations, such as TFBUT, by ophthalmologists.

Participants and Methods
Study design and participants. This cross-sectional observational study included 292 patients recruited between September 2016 to September 2017 at Juntendo University Hospital, Department of Ophthalmology, Tokyo, Japan. A requirement for a written informed consent was waived due to the retrospective observational nature of the study, and it was carried out using the opt-out method on our hospital website. The clinical study was approved by the Juntendo University Hospital, Independent Ethics Committee (approval numbers: 15-185 and 17-088) and adhered to the tenets of the Declaration of Helsinki. Exclusion criteria. We excluded patients with a history of eye lid disorder, ptosis, mental disease, Parkinson disease, and any other disease that affects blinking.

Dry eye disease diagnosis and classification.
Both eyes in all patients underwent complete ophthalmic evaluation, including measurement of best-corrected visual acuity, intraocular pressure (IOP), and assessment of subjective symptoms using the DEQS questionnaire 22 . TFBUT, CFS for kerato-conjunctival vital staining and Schirmer test Ι for reflex tear production were assessed in both eyes. Since blinking is affected by the condition of both eyes via the corneal reflex 18 , the lower TFBUT and Schirmer test Ι value data were used, whereas higher values of CFS were used in this study. We diagnosed DED and non-DED using the 2016 ADES criteria 19 . The 2016 criteria make a diagnosis of DED with two positive items, specifically positive subjective symptoms and decreased TFBUT (≤5 seconds). We diagnosed DED if either eye had a diagnosis of DED by the 2016 criteria.
Environmental conditions. Temperature and humidity of the examination room were controlled at 26 °C in the summer and 24 °C in the winter and 50% relative humidity, according to the Guideline for Design and Operation of Hospital HVAC Systems established by Healthcare Engineering Association of Japan Standard 25 . CFS score, 0-9 ± SD 2.1 ± 1.9 3.5 ± 2.3 <0.001 Schirmer Ι, mm ± SD 6.4 ± 6.6 5.1 ± 5.6 0.080 BIP, sec ± SD 17.6 ± 6.9 5.5 ± 2.6 <0.001

Subjective symptoms and Dry Eye-Related Quality-of-Life Score (DEQS). Subjective symptoms
were assessed by interviewing participants with DED. The DEQS questionnaire was administered to assess the severity of dry eye-associated symptoms and the multifaceted effects of DED on the patients' daily lives 22 . The score derived from this questionnaire is a subjective measurement of DED symptoms where 0 indicates the best score (no symptoms) and 100 indicates the worst score (maximum symptoms).

Tear film breakup time (TFBUT).
TFBUT was measured using a fluorescein dye according to the standard method 19 . To minimize the effect of the test strip on tear volume and TFBUT, a small quantity of the dye was administered with a wetted fluorescein strip. After the dye was instilled, the subject was instructed to blink three times to ensure adequate mixing of the dye with the tears. The time interval between the last blink and the appearance of the first dark spot on the cornea was measured with a stopwatch. The mean value of three measurements was used. The cutoff value of TFBUT ≤5 seconds was used to diagnose DED 19 .

Kerato-conjunctival vital staining (cornea fluorescence staining [CFS]). CFS was graded according
to the van Bijsterveld grading system 26 , dividing the ocular surface into three zones: nasal bulbar conjunctiva, temporal bulbar conjunctiva, and cornea. Each zone was evaluated on a scale of 0 to 3, with 0 indicating no staining and 3 indicating confluent staining. The maximum possible score is 9. Schirmer test I. The Schirmer test I was performed without topical anesthesia after the completion of all other examinations. Schirmer test strips (Ayumi Pharmaceutical Co., Tokyo, Japan) were placed at the outer one-third of the temporal lower conjunctival fornix for 5 minutes. The strips were then removed, and the length of dampened filter paper (in mm) was recorded.

Statistical analyses.
To compare the characteristics of study participants, the t test was used for continuous variables and χ 2 test was used for categorical variables. Pearson correlation coefficients were estimated among MBI, DEQS, TFBUT, and Schirmer test Ι . A ROC analysis was conducted to examine the diagnostic efficacy of MBI for DED. The ROC curve was plotted by computing the sensitivity and specificity using each symmetric value of the rating variable as a possible cutoff point. A point was plotted on the graph for each of the cutoff points; these plotted points were joined by straight lines to form the ROC curve, and the AUC was estimated using the trapezoidal rule. To determine the optimal cutoff value of MBI for detecting DED, the point where sensitivity and specificity were maximized was identified. Data are presented as means ± standard deviations (SDs) or proportions. Statistical analyses were performed using STATA version 14 (Stata Corp, Texas). A P < 0.05 was considered significant.