Noncontact Esthesiometer: Large Cohort Study on Agreement with Contact Method and Correlations with Corneal Nerves and Ocular Surface

Abstract

Purpose

To evaluate the repeatability of noncontact Brill esthesiometer (NBE), its agreement with Cochet–Bonnet esthesiometer (CBE), and the relationship between corneal sensitivity, corneal nerve metrics, and ocular surface parameters.

Design

Cross-sectional study.

Subjects

This study included 880 eyes, comprising healthy controls (80 eyes), dry eye disease (DED) (170 eyes), neuropathic corneal pain (NCP) (140 eyes), postrefractive surgery (P-RS) (176 eyes), and diabetes mellitus (DM) (314 eyes).

Methods

Noncontact Brill esthesiometer and CBE measurements were obtained at central and 4 peripheral corneal quadrants. Both measurements were converted to millinewtons (mN) for analysis. In vivo confocal microscopy for 7 corneal nerve parameters and 4 ocular surface parameters were assessed.

Main Outcome Measures

Agreement between NBE and CBE measurements. Correlations between NBE measurements, corneal nerve metrics, and ocular surface characteristics.

Results

Noncontact Brill esthesiometer presented excellent intraobserver and interobserver repeatability with both intraclass correlation coefficients >0.980 (P < 0.001). Noncontact Brill esthesiometer showed significantly higher values in mN than CBE (all P < 0.001). Noncontact Brill esthesiometer and CBE measurements showed good agreement in eyes with better corneal sensitivity (NBE <0.085 mN), whereas the agreement decreased at higher values. Noncontact Brill esthesiometer and CBE measurements significantly correlated across all cohorts (all P < 0.05). Postrefractive surgery group had the highest proportion of eyes within 95% limits of agreement (98.3%) and the strongest correlation between NBE and CBE measurements (r = 0.853, P < 0.001). Noncontact Brill esthesiometer measurements significantly correlated with corneal nerve parameters in the control, DED, P-RS, and DM groups (r = –0.219 to –0.545), while CBE showed significant correlations only in the DED and P-RS groups with comparable magnitude (r = –0.251 to –0.554, all P < 0.05). The strongest significant correlations for both NBE and CBE were in P-RS group (NBE: r = –0.514 to –0.545; CBE: r = –0.554). After adjustment for confounders, NBE was significantly associated with corneal nerve fiber length and width, while CBE showed significant association only with corneal nerve fiber fractal dimension. Noncontact Brill esthesiometer measurements, to a greater extent than CBE, significantly correlated with ocular surface parameters in the control, NCP, and DM groups (r = –0.126 to –0.465, all P < 0.05).

Conclusions

Noncontact Brill esthesiometer and CBE measurements showed good agreement at better corneal sensitivity and are not interchangeable. Noncontact Brill esthesiometer might link better with corneal nerve and ocular surface changes compared with CBE.

Financial Disclosures

The author has no/the authors have no proprietary or commercial interest in any materials discussed in this article.

The cornea is the most densely innervated peripheral tissue, with a central nerve density of approximately 7000 nociceptors per square millimeter.¹ Impairment of corneal innervation initiates a vicious cycle in which diminished trophic support leads to decreased lacrimation and impaired blinking reflexes, ultimately resulting in epithelial cell damage.² Corneal sensory abnormalities occur in various ocular surface conditions. Decrease or absence of corneal sensitivity is a defining feature and diagnostic hallmark of neurotrophic keratitis.² In conditions such as herpetic keratitis, dry eye disease (DED), contact lens wear, and corneal surgery, alterations in corneal sensitivity are typically secondary to the underlying pathology.^3,4 In neuropathic corneal pain (NCP), corneal sensitivity may be heterogenous rather than consistently reduced, reflecting corneal nerve dysfunctions.⁵ Long-term use of topical antiglaucoma medications and systemic diseases such as diabetes mellitus (DM) and vitamin B12 deficiency can also lead to reduced corneal sensitivity.6, 7, 8 Accurate evaluation of corneal sensitivity is crucial for the timely identification and monitoring of corneal nerve damage.

Several types of esthesiometers are currently available for assessing corneal sensitivity.⁹ The Cochet–Bonnet esthesiometer (CBE) (Luneau Ophthalmologie) is a handheld device with a nylon monofilament (0.08 or 0.12 mm diameters) adjustable from 0–6.0 cm to deliver varying intensities of stimuli.¹⁰ Cochet–Bonnet esthesiometer is widely used due to its portability, quantifiable measurements, and straightforward operation and is regarded as the gold standard in clinical practice.¹¹ Previous studies have indicated significant correlations between corneal nerve metrics assessed by in vivo confocal microscopy and CBE measurements in various ocular conditions.12, 13, 14 In eyes with herpes simplex keratitis, corneal nerve fiber density (CNFD) was reported to be positively correlated with CBE measurements in both the central (r = 0.53) and peripheral cornea (r = 0.27, both P < 0.001).¹² In eyes following LASIK, corneal nerve fiber length (CNFL) was positively correlated with CBE (r = 0.31, P < 0.001).¹⁵ Similarly, CNFD was positively correlated with CBE in non-Sjögren DED (r = 0.38, P = 0.041).¹⁶ However, CBE requires direct contact with the corneal surface, posing a risk of corneal damage or infection.¹⁷ In addition, the truncated stimulus intensity range and nonlinear intensity increments may not provide a precise estimate.¹⁸ Considering that CBE measurements increment in millimeters (typically 5 mm) and humidity affects the nylon filament bending, CBE may be unreliable for detecting minor sensitivity changes.¹⁹

The noncontact Brill esthesiometer (NBE) (Brill Engines), approved by the US Food and Drug Administration in 2023, is a portable device that functions as either a slit lamp attachment or a handheld instrument.²⁰ Noncontact Brill esthesiometer delivers ambient air pulses at room temperature directed to the cornea surface and has demonstrated good repeatability in healthy individuals.²⁰ However, factors such as flow-rate-dependent evaporation, cooling, and tear film depletion due to the gas-based stimulus may affect its performance.²¹ Whether the stimulus range of NBE adequately captures minor changes in corneal sensitivity across different ocular surface diseases remains unclear. Moreover, there is still limited understanding regarding the differences and correlations between NBE and CBE measurements. The relationships between NBE measurements and corneal nerve metrics, and clinical ocular surface assessments remain largely unexplored, with existing publications primarily focusing on chronic ocular graft-versus-host disease.²²

This study aimed to assess the corneal sensitivity using the NBE in a large cohort, including healthy individuals and patients at risk of corneal neuropathic changes, such as DED, NCP, postrefractive surgery (P-RS), and DM. The performance of the NBE was evaluated by assessing its repeatability and its agreement with measurements from the CBE to validate its clinical applicability. In addition, correlations between NBE measurements, corneal nerve metrics, and ocular surface characteristics were analyzed across various groups. This provides further insight into the clinical relevance of NBE as a surrogate marker of corneal nerve status and function.

Methods

Study Design and Population

A single-center, cross-sectional study was conducted at the Singapore National Eye Center and Singapore Eye Research Institute. Participants were recruited between March 2024 and March 2025. The inclusion and exclusion criteria for the study population are listed in Table 1. The study was approved by the Institutional Review Board of SingHealth, Singapore (No. 2022/2046), and was conducted in accordance with the tenets of the Declaration of Helsinki. Written informed consent was obtained from all participants.

Table 1.

Inclusion and Exclusion Criteria for the Study Population

Inclusion Criteria

•≥21 years of age •No history of contact lens wear within the past 2 weeks •Healthy cohort: participants with no history of ocular or systemic diseases, ocular trauma, or ocular surgeries, and did not use topical eye drops. •DED cohort: participants with Ocular Surface Disease Index (OSDI) score ≥13 and signs of DED, including tear breakup time (TBUT) <10 s and/or the presence of ocular surface staining (>5 corneal spots, >9 conjunctival spots)23 •NCP cohort: Participants with (1) persistent (>3 months) neuropathic ocular symptoms, including burning, stinging, photophobia, pain, severe dryness; (2) absent or minimal ocular surface fluorescein staining, with National Eye Institute (NEI) score <2; and (3) corneal nerve abnormalities as detected by IVCM, such as decreased corneal nerve length and density, presence of neuromas; no response to conventional dry eye treatments for ≥6 months5 •Postrefractive surgery cohort: participants who have undergone corneal laser refractive surgery (i.e., LASIK, KLEx, or advanced surface ablation) within the past 3 months24,25 •DM cohort: participants diagnosed with type 2 DM based on the following criteria: fasting plasma glucose >7.0 mmol/L, or symptoms of hyperglycemia accompanied by plasma glucose >11.1 mmol/L, or a 2-hour plasma glucose >11.1 mmol/L after a 75-g oral glucose tolerance test. Subjects have a known duration of type 2 DM for >3 months.26

Exclusion criteria

•Eyes with active inflammation or infection •Eyes that have undergone corneal surgery other than corneal laser refractive surgery •Presence of corneal abrasion or corneal epithelial defects •Eyes with prolonged use of topical antiglaucoma medications •Diseases that may affect corneal sensitivity, such as intracranial neoplasm, stroke, or a history of trigeminal nerve surgery •Other systemic diseases that affect the ocular surface, such as systemic autoimmune disorders.

DED = dry eye disease; DM = diabetes mellitus; IVCM = in vivo confocal microscopy; KLEx = keratorefractive lenticule extraction; NCP = neuropathic corneal pain.

Measurements of Corneal Sensitivity

Corneal sensitivity was assessed at a single site during a single visit. The assessments were started with NBE, which was performed twice with a 5-minute interval to evaluate repeatability, followed by CBE measurements after an additional 5-minute interval by 2 experienced ophthalmologists (M.Y and C.L) under controlled temperature (21°C–24°C) and relative humidity (55%–56%).

The CBE features a retractable 0.12-mm diameter nylon filament with a 0–6.0 cm scale. Shorter filament lengths apply greater pressure. Measurements begin with placing the fully extended filament against the cornea and retracting in 0.5-cm increments until a blink or verbal acknowledgment was elicited. The maximum filament length that triggers a response was recorded as the corneal sensitivity.¹⁸ Measurements from the central and 4 peripheral quadrants of the cornea were averaged to represent the overall sensitivity of each eye.

Noncontact Brill esthesiometer assessments were conducted following the manufacturer’s instructions. The NBE was mounted on a slit lamp and the nozzle was positioned 4 ± 1 mm from the patient’s cornea, with this distance accurately maintained by an electronic positioning system that converged 2 light-emitting diodes into a single light-emitting diode spot on the cornea. Noncontact Brill esthesiometer delivered pulsed airflow from level 1–5, corresponding pressures of 1–10 mbar, targeting a 0.4-mm diameter area on the cornea. Stimulation began at the lowest level and gradually increased in intensity. The time between each measurement was 10 seconds. The lowest stimulation level and the associated air pulse pressure perceived by the patient were recorded.²⁷ Measurements were first obtained at the central cornea. For peripheral corneal assessment, the nozzle was repositioned to the superior, inferior, nasal, and temporal quadrants, while the participant maintained forward fixation. The same measurement protocol was applied at each location. The overall sensitivity was assessed by averaging the pressure measurements at the central and the 4 peripheral corneal quadrants, and all comparisons between NBE and CBE measurements were based on the overall corneal sensitivity. Intraobserver repeatability was assessed from 2 consecutive NBE measurements performed by the same operator with 5 minutes apart. Interobserver repeatability was evaluated by a second operator repeating the NBE measurements 1 hour after.

Data Processing

To enable a direct quantitative and statistical comparison on the same scale, the measurements from CBE and NBE were standardized to millinewtons (mN). As per the manufacturer’s instructions, the CBE values in cm were transformed to pressure (g/mm²) and then converted to mN using the nylon filament section area (S = 0.0113 mm²). Similarly, NBE measurements, initially in mbar, were converted to mN. Table 2 summarizes these conversions. The higher pressure values in mN indicate the lower corneal sensitivity.

Table 2.

Conversion of the Measurement Units in Cochet–Bonnet Esthesiometer and Noncontact Brill Esthesiometer

Cochet–Bonnet Esthesiometer				Noncontact Brill Esthesiometer
Length of nylon filament (cm)	Pressures values in g/mm2	Pressures values in mN	Pressures values in mbar	Pressures values in mN
6.0	0.4	0.020	2.0	0.025
5.5	0.5	0.025	3.0	0.038
5.0	0.55	0.027	4.0	0.050
4.5	0.7	0.035	5.0	0.063
4.0	0.8	0.039	6.0	0.075
3.5	1.0	0.049	7.0	0.088
3.0	1.4	0.069	8.0	0.100
2.5	1.8	0.089	9.0	0.113
2.0	2.8	0.138	10.0	0.126
1.5	5.1	0.251
1.0	10.3	0.508
0.5	15.9	0.784

Ocular Surface Subjective and Objective Assessments

Schirmer I test was performed without topical anesthesia for 5 minutes. Corneal fluorescein staining was evaluated by the National Eye Institute (NEI) scale across 5 corneal regions (0–15). Ocular surface fluorescein staining was assessed with the Oxford grading (0–5). Tear breakup time (TBUT) was the lapse between the end of blinking and tear breakup. All examinations were performed by masked and experienced ophthalmologists with published protocols.^26,28 Ocular Surface Disease Index questionnaire consists of 12 questions with a total score of 0–100.²⁹ To minimize the impact on tear film and ocular surface status, clinical assessments were conducted in the following order: Ocular Surface Disease Index, corneal sensitivity, TBUT, corneal and ocular surface staining, and Schirmer test. All the assessments and the corneal sensitivity measurements were conducted on the same day for each participant.

In Vivo Confocal Microscopy Image Acquisition and Analysis

In vivo confocal microscopy were performed using the Heidelberg HRT3 Rostock Cornea Module (Heidelberg Engineering GmbH) after the ocular surface assessments to examine sub-basal corneal nerve plexus by an independent, masked, and experienced ophthalmologist following the published protocol.^30,31 Images were taken of the central cornea, and the nasal, temporal, superior, and inferior quadrants approximately 3 mm from the corneal apex. For each scanned area, 5 best-focused images were selected and analyzed using the ACCMetrics software (University of Manchester).³² The following parameters were obtained: CNFD (number of main nerve fibers/mm²), CNFL (the total length of fibers in mm/mm²), corneal nerve branch density (CNBD) (number of branch points on the main fibers/mm²), corneal total branch density (CTBD) (total number of branch points/mm²), corneal nerve fiber area (CNFA) (total nerve fiber area mm²/mm²), corneal nerve fiber width (CNFW) (mean nerve fiber width in mm/mm²), and corneal nerve fiber fractal dimension (CFracDim) (measurement of the spatial loss of corneal nerves, a high CFracDim value indicates an evenly distributed complex nerve fiber structure).³³

Statistical Analysis

All data were presented as mean ± standard deviation (SD), percentage or median (interquartile range). Bland–Altman analysis and intraclass correlation coefficient (ICC) were used to evaluate the intraobserver and interobserver repeatability. An ICC value of 0.75–0.90 indicates good repeatability, and ICC >0.90 indicates excellent repeatability.³⁴ Coefficient of repeatability was calculated as 1.96 SD of the differences the between repeated measurements. Bland–Altman analysis was used to evaluate the interinstrument agreement. The 95% limits of agreement (LOA) were calculated as the mean differences between the NBE and CBE measurements ± 1.96 SD of the differences. The mean values of the right and left eyes were used, and the correlations between the corneal sensitivity, corneal nerve metrics, and ocular surface assessments were performed using the Pearson correlation coefficient. Multivariate linear mixed models were used to determine the associations between NBE and CBE measurements and corneal nerve parameters to adjust for potential confounders, including age, sex, and study cohorts. The Kolmogorov–Smirnov test was used to analyze the normality of the data. Wilcoxon signed-rank test compared measurements between devices, and Kruskal–Wallis test followed by Dunn post hoc test assessed differences across cohorts. The required sample size was calculated from the preliminary data of 30 eyes on the agreement between the NBE and CBE measurements. The mean difference between NBE and CBE measurements was 0.025 mN, with a mean SD of 0.015 mN, and maximum allowed difference of 0.018 mN. Hence, 76 eyes, with a power of 80% and at a 5% level of significance were sufficient to detect the correlation between NBE and CBE measurements. Although this sample size was sufficient for the primary agreement analysis, a substantially larger cohort of 880 eyes was included in our study to improve the precision of agreement analysis, and allow a reliable subgroup and correlation analysis. Statistical analysis was performed with SPSS version 26 (IBM SPSS Statistics).

Results

Demographic Information

A total of 880 eyes from 440 participants were included, comprising 80 eyes from healthy individuals (mean age: 45.0 ± 20.2 years; 42.5% female), 170 eyes with DED (mean age: 63.8 ± 12.1 years; 77.6% female), 140 eyes with NCP (mean age: 56.7 ± 13.7 years; 85.7% female), 176 P-RS eyes (mean age: 26.8 ± 5.1 years; 50.0% female), and 314 eyes from patients with DM (mean age: 58.2 ± 11.7 years; 35.7% female). The overall study population comprised 248 (56.4%) women, with a mean age of 57.2 ± 14.7 years.

Intraobserver and Interobserver Repeatability of NBE Measurements

Noncontact Brill esthesiometer presented excellent intraobserver and interobserver repeatability, with intraobserver ICC of 0.997 (95% confidence interval [CI]: 0.994 to 0.998, P < 0.001) and interobserver ICC of 0.984 (95% CI: 0.972 to 0.991, P < 0.001). The coefficient of repeatability was 0.0031 mN (95% CI: 0.0026 mN to 0.0038 mN) for intraobserver repeatability and 0.0042 mN (95% CI: 0.0027 mN to 0.0065 mN) for interobserver repeatability. This indicated a narrow range of differences between repeated NBE measurements, demonstrating good repeatability for both intraobserver and interobserver assessments. Bland–Altman analysis indicated good agreement for both repeated measurements and between observers, with mean differences of 0.000 ± 0.002 mN (95% LOA: –0.004 to 0.006 mN) and 0.001 ± 0.003 mN (95% LOA: –0.005 to 0.041 mN), respectively (Fig 1A, B).

Figure 1.

Bland–Altman plots showing the repeatability of NBE and the pattern of change in pressure exerted by the NBE and CBE. (A) Intraobserver repeatability, (B) interobserver repeatability, and (C) comparison of the pattern of change in pressure exerted by the NBE and CBE with progressive increase in stimulus intensity. CBE = Cochet–Bonnet esthesiometer; LOA = limits of agreement; mN = millinewton; NBE = noncontact Brill esthesiometer.

NBE and CBE Measurements in 5 Groups

A comparison of the pressure exerted by NBE and CBE is presented in Figure 1C. Noncontact Brill esthesiometer consistently showed significantly higher measurements than CBE across all cohorts (Table 3, all P < 0.001). In addition, NBE measurements were significantly higher in the DED (0.044 ± 0.021 mN), NCP (0.041 ± 0.019 mN), P-RS (0.041 ± 0.014 mN), and DM cohorts (0.045 ± 0.022 mN) compared with controls (0.034 ± 0.010 mN, all P < 0.05), indicating reduced corneal sensitivity in these groups. Diabetes mellitus group presented significantly higher NBE measurements than NCP group (0.045 ± 0.022 mN vs. 0.041 ± 0.019 mN, P = 0.044), suggesting greater corneal sensitivity loss. However, no significant differences were observed in CBE measurements (Table 3). Additionally, NBE measurements were comparable across the 5 cornea regions (all P > 0.05) (Supplementary Table 1, available at www.ophthalmologyscience.org).

Table 3.

Comparison of the NBE and CBE Measurements

Study Cohorts	NBE (mN)	CBE (mN)	P Value∗
Whole cohort	0.043 ± 0.020	0.023 ± 0.018	<0.001
Control	0.034 ± 0.010	0.021 ± 0.002	<0.001
DED	0.044 ± 0.021	0.023 ± 0.007	<0.001
NCP	0.041 ± 0.019	0.023 ± 0.032	<0.001
P-RS	0.041 ± 0.014	0.024 ± 0.006	<0.001
DM	0.045 ± 0.022	0.022 ± 0.003	<0.001
P value†	<0.001	0.625	-
Control vs. DED	<0.001	>0.999	-
Control vs. NCP	0.025	>0.999	-
Control vs. P-RS	0.048	>0.999	-
Control vs. DM	<0.001	>0.999	-
DED vs. NCP	0.264	>0.999	-
DED vs. P-RS	0.352	>0.999	-
DED vs. DM	>0.999	>0.999	-
NCP vs. P-RS	>0.999	>0.999	-
NCP vs. DM	0.044	>0.999	-
P-RS vs. DM	0.109	>0.999	-

CBE = Cochet–Bonnet esthesiometer; DED = dry eye disease; DM = diabetes mellitus; mN = millinewton; NBE = noncontact Brill esthesiometer; NCP = neuropathic corneal pain; P-RS = postrefractive surgery. Bold values indicate statistically significant results (P < 0.05).

^∗NBE measurements vs. CBE measurements.

^†Bonferroni post hoc test across 5 subgroups.

Agreement between the NBE and CBE Measurements

Bland–Altman analysis of the whole cohort demonstrated an overall high level of agreement between NBE and CBE measurements. However, in cases with higher NBE values (>0.085 mN), representing worse corneal sensitivity, the dispersion between the 2 measurements increased, with a greater proportion of data points falling outside the 95% LOA (Fig 2). The DM group showed the highest mean difference between NBE and CBE (0.024 ± 0.020 mN) and the lowest proportion (94.3%) within the 95% LOA. The control group had the lowest mean difference (0.012 ± 0.007 mN), significantly lower than those of the DED (0.023 ± 0.021 mN) and DM groups (0.024 ± 0.020 mN, both P < 0.05). The P-RS group had the highest proportion of eyes (98.3%) within the 95% LOA (Table 4).

Figure 2.

Bland–Altman plots showing the agreement between the NBE and CBE measurements. (A) The whole cohort, (B) the control group, (C) the DED group, (D) the NCP group, (E) the P-RS, and (F) the DM group. The solid red lines indicate the mean difference, and the dotted red lines indicate the upper and lower 95% LOA. CBE = Cochet–Bonnet esthesiometer; DED = dry eye disease; DM = diabetes mellitus; LOA = limits of agreement; mN = millinewton; NBE = noncontact Brill esthesiometer; NCP = neuropathic corneal pain; P-RS = postrefractive surgery.

Table 4.

Agreement between the CBE and NBE Measurements

Study Cohorts	Mean Difference of NBE and CBE (mN)	95% LOA (mN)	Proportion of Eyes within 95% LOA (%)
Whole cohort	0.020 ± 0.024	–0.027 to 0.067	96.3
Control	0.012 ± 0.007	–0.004 to 0.031	97.5
DED	0.023 ± 0.021	–0.014 to 0.062	95.3
NCP	0.017 ± 0.032	–0.046 to 0.079	95.7
P-RS	0.021 ± 0.012	–0.003 to 0.045	98.3
DM	0.024 ± 0.020	–0.017 to 0.064	94.3
P value (post hoc test)	<0.001	-	-
Control vs. DED	0.039	-	-
Control vs. NCP	>0.999	-	-
Control vs. P-RS	0.782	-	-
Control vs. DM	0.016	-	-
DED vs. NCP	0.079	-	-
DED vs. P-RS	>0.999	-	-
DED vs. DM	>0.999	-	-
NCP vs. P-RS	>0.999	-	-
NCP vs. DM	0.009	-	-
P-RS vs. DM	>0.999	-	-

CBE = Cochet–Bonnet esthesiometer; DED = dry eye disease; DM = diabetes mellitus; LOA = limits of agreement; mN = millinewton; NBE = noncontact Brill esthesiometer; NCP = neuropathic corneal pain; P-RS = postrefractive surgery. Bold values indicate statistically significant results (P < 0.05).

Correlation between NBE and CBE Measurements

A significant positive correlation was observed between NBE and CBE measurements across the whole cohort (r = 0.400, P < 0.001) and all subgroups: control group (r = 0.695, P < 0.001), DED group (r = 0.340, P < 0.001), NCP group (r = 0.150, P = 0.020), P-RS (r = 0.853, P < 0.001), and DM group (r = 0.531, P < 0.001; Fig 3).

Figure 3.

Scatter plots with quadratic fitted line showing the correlation between the NBE and CBE measurements. (A) The whole cohort, (B) the control group, (C) the DED group, (D) the NCP group, (E) the P-RS, and (F) the DM group. CBE = Cochet–Bonnet esthesiometer; DED = dry eye disease; DM = diabetes mellitus; NBE = noncontact Brill esthesiometer; NCP = neuropathic corneal pain; P-RS = postrefractive surgery.

Correlation and Association between Corneal Sensitivity and Corneal Nerve Parameters

In the whole cohort, NBE measurements showed correlations with corneal nerve metrics of comparable strength to those of CBE, but involving a broader range of nerve parameters. Specifically, NBE measurements were significantly negatively correlated with CNFD, CNBD, CNFL, CTBD, CNFA, and CFracDim (r = –0.167 to –0.243), and positively correlated with CNFW (r = 0.209, all P < 0.05). Cochet–Bonnet esthesiometer measurements were significantly negatively correlated with CNBD, CNFL, CTBD, CNFA, and CFracDim (r = –0.151 to –0.203, all P < 0.05). The strongest correlations were observed for CNFD with NBE and for CNFL with CBE (Table 5).

Table 5.

Correlation between Corneal Sensitivity and Corneal Nerve Parameters

Parameters	Statistic	Whole Cohort		Control		DED		NCP		P-RS		DM
		NBE (mN)	CBE (mN)	NBE (mN)	CBE (mN)	NBE (mN)	CBE (mN)	NBE (mN)	CBE (mN)	NBE (mN)	CBE (mN)	NBE (mN)	CBE (mN)
CNFD (/mm2)	r	–0.231	–0.151	–0.445	–0.227	–0.133	–0.131	0.151	0.036	–0.514	–0.554	–0.320	–0.017
	P	<0.001	0.018	0.005	0.164	0.226	0.286	0.318	0.814	0.042	0.026	0.005	0.883
CNBD (/mm2)	r	–0.167	–0.154	–0.215	–0.069	–0.221	–0.283	0.052	–0.03	–0.520	–0.318	–0.179	0.025
	P	0.007	0.015	0.188	0.677	0.042	0.019	0.728	0.843	0.039	0.231	0.119	0.827
CNFL (mm/mm2)	r	–0.243	–0.203	–0.437	–0.223	–0.108	–0.251	0.12	0.013	–0.545	–0.255	–0.298	–0.032
	P	<0.001	0.001	0.005	0.172	0.327	0.039	0.421	0.933	0.029	0.341	0.009	0.785
CTBD (/mm2)	r	–0.170	–0.154	–0.133	–0.035	–0.219	–0.350	–0.007	–0.048	–0.291	–0.134	–0.207	0.013
	P	0.006	0.016	0.421	0.833	0.044	0.003	0.962	0.747	0.274	0.622	0.07	0.91
CNFA (mm2/mm2)	r	–0.174	–0.195	–0.149	–0.051	–0.078	–0.382	–0.023	–0.076	–0.412	–0.152	–0.285	–0.111
	P	0.005	0.002	0.365	0.759	0.476	0.001	0.88	0.611	0.112	0.575	0.012	0.336
CNFW (mm/mm2)	r	0.209	–0.003	0.327	0.169	0.083	–0.026	–0.102	0.07	–0.192	–0.176	0.367	0.078
	P	0.001	0.962	0.042	0.305	0.451	0.832	0.495	0.64	0.476	0.514	0.001	0.498
CFracDim	r	–0.186	–0.156	–0.226	–0.139	–0.048	–0.316	0.105	0.107	–0.539	–0.279	–0.369	–0.065
	P	0.002	0.014	0.166	0.398	0.664	0.009	0.483	0.475	0.031	0.296	0.001	0.577

CBE = Cochet–Bonnet esthesiometer; CFracDim = corneal nerve fiber fractal dimension; CNBD = corneal nerve branch density; CNFA = corneal nerve fiber area; CNFD = corneal nerve fiber density; CNFL = corneal nerve fiber length; CNFW = corneal nerve fiber width; CTBD = corneal total branch density; DED = dry eye disease; DM = diabetes mellitus; NBE = noncontact Brill esthesiometer; NCP = neuropathic corneal pain; P-RS = postrefractive surgery. Bold values indicate statistically significant results (P < 0.05).

In the linear mixed models analysis with the adjustment age, gender, and study groups, NBE measurements was significantly associated with CNFL (β = –0.001, P = 0.048) and associated with CNFW (β = 2.900, P = 0.043), whereas CBE measurements was significantly associated with CFracDim (β = –0.049, P = 0.010; Table 6).

Table 6.

Multivariate Linear Mixed Models for the Association between Corneal Sensitivity and Corneal Nerve Parameters

Parameters	NBE		CBE
	β (95% CI)	P value	β (95% CI)	P value
CNFD	–0.001 (–0.004 to 0.002)	0.447	–0.002 (–0.004 to 0.001)	0.166
CNBD	0.001 (0.000 to 0.003)	0.121	0.000 (–0.001 to 0.000)	0.220
CNFL	–0.001 (–0.002 to 0.001)	0.048	0.001 (–0.001 to 0.002)	0.914
CTBD	0.004 (–0.004 to 0.012)	0.325	–0.005 (–0.011 to 0.002)	0.416
CNFA	–0.119 (–7.487 to 7.248)	0.974	0.133 (–1.810 to 2.075)	0.893
CNFW	2.900 (0.061 to 5.739)	0.043	0.268 (–0.483 to 1.018)	0.481
CFracDim	–0.118 (–0.262 to 0.027)	0.110	–0.049 (–0.086 to –0.012)	0.010

CBE = Cochet–Bonnet esthesiometer; CFracDim = corneal nerve fiber fractal dimension; CI = confidence interval; CNBD = corneal nerve branch density; CNFA = corneal nerve fiber area; CNFD = corneal nerve fiber density; CNFL = corneal nerve fiber length; CNFW = corneal nerve fiber width; CTBD = corneal total branch density; NBE = noncontact Brill esthesiometer. Bold values indicate statistically significant results (P < 0.05).

Subgroup analyses revealed different strength and patterns of correlations across groups. In the control group, NBE measurements significantly negatively correlated with CNFD and CNFL (r = –0.437 and –0.445, both P = 0.005), and significantly positively correlated with CNFW (r = 0.327, P = 0.042), whereas CBE measurements had no significant correlations. In the DED group, NBE measurements significantly negatively correlated with CNBD and CTBD (r = –0.219 and –0.221, both P < 0.05), while CBE measurements showed broader and stronger negative correlations with CNBD, CNFL, CTBD, CNFA, and CFracDim (r = –0.251 to –0.382, all P < 0.05). Postrefractive surgery group exhibited the strongest correlations across all cohorts. Noncontact Brill esthesiometer measurements significantly negatively correlated with CNFD, CNBD, CNFL, and CFracDim (r = –0.514 to –0.545, all P < 0.05), while CBE measurements significantly correlated only with CNFD (r = –0.554, P = 0.026). In the DM group, NBE measurements significantly negatively correlated with CNFD, CNFL, CNFA, and CFracDim (r = –0.285 to –0.369, all P < 0.05), and positively correlated with CNFW (r = 0.367, P = 0.001), while CBE showed no significant correlations. No significant correlations were observed in the NCP group.

Correlation between Corneal Sensitivity and Ocular Surface Assessments

In the whole cohort, NBE measurements significantly negatively correlated with both Oxford and NEI scores (r = –0.122 and –0.206, both P < 0.05), whereas CBE measurements significantly negatively correlated with Oxford scores (r = 0.141, P = 0.001). In the DED group, CBE measurements significantly positively correlated with Oxford and NEI scores (r = 0.506 and 0.583, both P < 0.05). In the NCP group, NBE measurements significantly negatively correlated with Schirmer test values, Oxford scores, and NEI scores (r = –0.144, P = 0.030; r = –0.227, P = 0.015; and r = –0.352, P < 0.001, respectively), whereas CBE measurements significantly negatively correlated with TBUT (r = –0.246, P = 0.006). Noncontact Brill esthesiometer measurements significantly negatively correlated with TBUT (r = –0.465, P = 0.013) in the P-RS group, as well as with Oxford and NEI scores (r = –0.178 and –0.126, both P < 0.05) in the DM group. No significant correlations were observed between NBE or CBE measurements and ocular surface assessments in the control group. Moreover, no significant correlations were found between CBE measurement and ocular surface assessments in P-RS or DM groups (Table 7).

Table 7.

Correlation between Corneal Sensitivity and Ocular Surface Assessments

Parameters	Statistic	Whole Cohort		Control		DED		NCP		P-RS		DM
Ocular surface parameters
Schirmer test (mm)		5.93 ± 6.45		13.85 ± 11.13		2.63 ± 2.22		5.81 ± 8.19		11.09 ± 7.50		8.89 ± 8.65
TBUT (s)		4.03 ± 3.17		5.87 ± 2.07		3.00 ± 1.41		3.68 ± 2.03		4.71 ± 2.14		4.72 ± 2.34
Oxford score		0.00 (0.00-0.00)		0.00 (0.00-0.00)		0.00 (0.00-1.00)		0.00 (0.00-0.00)		0.00 (0.00-0.00)		0.00 (0.00-1.00)
NEI score		0.00 (0.00-0.00)		0.00 (0.00-0.00)		1.00 (0.00-2.00)		0.00 (0.00-0.00)		0.00 (0.00-0.00)		0.50 (0.00-2.00)
OSDI score		13.73 ± 17.89		0.09 ± 0.44		23.82 ± 11.42		21.98 ± 17.25		19.58 ± 9.56		5.63 ± 11.43
Correlation analysis
		NBE (mN)	CBE (mN)	NBE (mN)	CBE (mN)	NBE (mN)	CBE (mN)	NBE (mN)	CBE (mN)	NBE (mN)	CBE (mN)	NBE (mN)	CBE (mN)
Schirmer test (mm)	r	–0.003	0.012	–0.150	–0.098	–0.130	–0.356	–0.144	0.100	0.012	0.044	0.080	–0.112
	P	0.932	0.763	0.321	0.518	0.545	0.147	0.030	0.142	0.950	0.825	0.162	0.051
TBUT (s)	r	0.071	0.006	–0.132	–0.077	0.154	–0.123	0.062	–0.246	–0.465	–0.343	0.005	0.058
	P	0.101	0.885	0.382	0.610	0.409	0.518	0.492	0.006	0.013	0.074	0.928	0.313
Oxford score	r	–0.122	0.141	0.110	–0.068	0.068	0.506	–0.227	0.164	0.338	0.289	–0.178	–0.045
	P	0.005	0.001	0.479	0.661	0.681	0.002	0.015	0.084	0.079	0.136	0.002	0.432
NEI score	r	–0.206	–0.024	–0.018	0.254	0.166	0.583	–0.352	0.113	0.303	0.358	–0.126	–0.055
	P	<0.001	0.582	0.906	0.096	0.312	<0.001	0.001	0.235	0.117	0.062	0.028	0.336
OSDI score	r	0.013	0.100	0.303	–0.706	–0.200	0.146	0.033	–0.079	0.289	0.038	0.148	0.155
	P	0.852	0.155	0.194	0.749	0.136	0.375	0.744	0.445	0.487	0.928	0.168	0.149

CBE = Cochet–Bonnet esthesiometer; DED = dry eye disease; DM = diabetes mellitus; NBE = noncontact Brill esthesiometer; NCP = neuropathic corneal pain; NEI = National Eye Institute; OSDI = Ocular Surface Disease Index; P-RS = postrefractive surgery; TBUT = tear breakup time. Bold values indicate statistically significant results (P < 0.05).

Discussion

Our study demonstrated that NBE provides a reliable assessment of corneal sensitivity with high intraobserver and interobserver repeatability. Noncontact Brill esthesiometer consistently measured higher values than CBE, therefore, consistent use of the same instrument is necessary for longitudinal assessment. A good interinstrument agreement was observed, particularly in eyes with better corneal sensitivity, with the control group exhibiting the lowest mean difference and narrowest 95% LOA. Significant positive correlations were observed between NBE and CBE measurements across all cohorts, with the strongest correlation identified in the P-RS group. Furthermore, NBE measurements were significantly correlated with corneal nerve metrics in the control, P-RS, and DM groups, with correlations of comparable magnitude but involving more nerve parameters than CBE. Similarly, NBE measurements were significantly correlated with ocular surface assessments in the NCP, P-RS, and DM groups, showing correlations with more parameters than CBE. To our knowledge, this is the first large-cohort study covering various ocular surface diseases assessing the agreement between NBE and CBE, and the correlations between NBE measurements, corneal nerve metrics, and ocular surface assessments.

Since the 1990s, several noncontact corneal esthesiometers have been developed. The Belmonte noncontact corneal esthesiometer and its modified version, the CRCERT–Belmonte esthesiometer, are gas-based devices that stimulate the temperature-sensitive, velocity-sensitive, polymodal receptors, and potentially chemically sensitive polymodal receptors due to its hyperosmolar effect.^35,36 However, they are currently not commercially available. Additionally, the CRCERT–Belmonte disperses gas across the entire cornea surface and stimulates more receptors, limiting regional assessment precision.³⁷ The noncontact Swiss liquid jet aesthesiometer for corneal sensitivity delivers liquid jet with adjustable stimulus pressure of 100–1500 mbar.³⁸ However, Swiss liquid jet aesthesiometer for corneal sensitivity is not commercially available and the stimulated area varies with the pressure and speed of jet.³⁸ The poor portability of CRCERT–Belmonte and Swiss liquid jet aesthesiometer for corneal sensitivity also restricts their accessibility, while NBE addresses this limitation.

We demonstrated an excellent intraobserver and interobserver repeatability for NBE, with both ICC values >0.980, and coefficient of repeatability values of 0.0065 and 0.0031, consistent with the findings of Ruiz-Lozano et al,²⁷ who reported the ICC values >0.900 for intraobserver and interobserver repeatability in a healthy cohort.

The NBE and CBE differ in the mode of corneal stimulus generation, the types of corneal receptors activated, and the area of stimulation, which may cause the variations in their measurement thresholds. The NBE delivers air pulses without direct contact, activating the mechanosensory pressure receptors, polymodal nociceptors, less abundant cold receptors, and chemically sensitive polymodal receptors as the air pulse may induce localized hyperosmolarity.^17,37 This broader stimulation may activate more corneal terminal receptors.³⁹ The CBE, in comparison, applies purely mechanical pressure through a more localized point of contact, activating the mechanoreceptors.¹⁸ These differences explain the significantly higher measurements from NBE than those from CBE. Early studies also found significantly higher NBE measurements than CBE in DED and healthy individuals.^21,27 Our study reinforces this finding with a wider range of ocular surface pathologies, indicating that NBE might be more sensitive in detecting minor corneal sensitivity changes than CBE. Moreover, because measurements of NBE and CBE are not interchangeable, using different devices for patient follow-up would not be appropriate.

Our study demonstrated good agreement between NBE and CBE when NBE measurement was below 0.085 mN (6.8 mbar), and the agreement declined at higher NBE values. Across all cohorts, increasing applied pressure led to a growing bias between NBE and CBE, resulting in reduced agreement. This observation aligns with Ruiz-Lozano et al²⁷ reporting good agreement when NBE readings were at level 1 or 2 but decreased at higher levels. A possible explanation is the different pressure change patterns between NBE and CBE. Noncontact Brill esthesiometer exhibits a consistent linear increase in pressure across intensity levels, whereas CBE exerts an approximately linear pressure change only at nylon filament length of 6.0–2.5 cm and shows exponential change at filament length <2.0 cm.^17,20 Therefore, the bias of measurements between devices is more pronounced at reduced corneal sensitivity. Among our study cohorts, the control group exhibited the lowest mean difference and narrowest 95% LOA, reflecting normal corneal sensitivity. In contrast, the DM cohort showed larger mean difference and wider 95% LOA due to the eyes with lower corneal sensitivity. Our findings suggest that the agreement between NBE and CBE depends on the corneal sensitivity levels, with better agreement in better corneal sensitivity and greater discrepancies in reduced corneal sensitivity.

Noncontact Brill esthesiometer measurements significantly positively correlated with CBE measurements, indicating that both methods capture comparable trends across various cohorts but are not directly interchangeable. Additionally, the underlying mechanisms of the corneal sensitivity abnormalities varied among the 4 cohorts of corneal neuropathy, which may contribute to the variations in the correlation between both measurements. The P-RS cohort showed the strongest correlation, possibly due to the younger age, healthier preoperative corneal nerve status, and more homogeneous corneal nerve injury pattern from nerve transection. In contrast, NCP involves heterogeneous and complex nerve abnormalities, causing hypesthesia or hyperesthesia,⁴⁰ which may lead to greater measurement variability and weaker correlation.

In the whole cohort, NBE measurements showed correlations with corneal nerve metrics that were comparable in strength to those of CBE, but involved a broader range of nerve parameters. Linear mixed models analyses further demonstrated that, after adjusting for potential confounders, NBE remained significantly independently associated with CNFL and CNFW, while CBE only showed a significant association with CFracDim. In addition, we identified statistically significant but low to moderate negative correlations between NBE measurements and corneal nerve metrics across control, DED, P-RS, and DM cohorts. These findings suggest that functional sensory measurements and corneal nerve morphology may not change in parallel. Corneal nerve loss can precede measurable sensory impairment. In addition, abnormal regeneration and disrupted integration after corneal nerve injuries may result in nerves that are anatomically intact but functionally deficient.^41,42 However, NBE measurements correlated with more nerve parameters in the control, P-RS, and DM groups than CBE, indicating that NBE might better reflect functional changes and the underlying corneal nerve characteristics in these cohorts. Our findings align with a previous report on patients with ocular graft-versus-host disease, showing that NBE values significantly and negatively correlated with the total, main trunk, and branch densities of corneal nerves.^5,22 The absence of correlations in the NCP cohort may due to its heterogeneous with mixed presentation of reduced or increased corneal sensitivity.^5,40

We found significant correlations between NBE measurements and ocular surface parameters in the NCP, P-RS, and DM groups. In contrast, CBE measurements showed significant correlations in the DED and NCP groups, and with fewer ocular surface parameters. While these findings suggest that NBE may be more closely linked to relevant clinical features, the observed correlation coefficients were generally low to moderate, which may reflect the multifactorial nature of ocular surface conditions and corneal nerve status. Noncontact Brill esthesiometer measurements significantly negatively correlated with TBUT in the P-RS group, possibly reflecting postoperative corneal denervation and altered mucin expression.⁴³ Of note, our results suggested that greater ocular surface epitheliopathy (higher NEI and Oxford scores) correlated negatively with NBE measurement in NCP and DM groups but correlated positively with NBE in controls and with CBE in both DED and control groups. Such discrepancies exist across diseases and even among studies examining the same disease. For instance, De et al⁴⁴ found a negative correlation between corneal staining and noncontact air jet esthesiometer measurements in DED, while Bourcier et al³ reported a positive correlation. One possible explanation is that ocular surface epithelial defect affects the measurements of an air-based esthesiometer, as the air jet can penetrate the leaky corneal epithelium and directly stimulate sensory nerves, inducing hypersensitivity.⁴⁴ Air-based esthesiometer can also detect hyperalgesia around epithelial wounds,⁴⁵ while air-mediated hyperalgesia cannot be elicited by CBE. Another hypothesis is that the signs of corneal epitheliopathy are characterized by a complex interplay between ocular surface and corneal somatosensory dysfunction. Variations in hyposensitivity and hypersensitivity may indicate different stages of the disease. Of note, the case population is likely heterogeneous, with patients differing in their severity of corneal neuropathy across subgroups. This underlying heterogeneity may have contributed to the observed differences between instruments as well as their correlations with other clinical metrics.

Our study has several limitations. The corneal sensitivity was assessed at a single time point with no longitudinal observations for disease progression. Disease severity was not stratified, which will be included in future studies. Future investigations can evaluate the effectiveness of NBE in long-term monitoring and explore the biological factors underlying the variability between NBE and CBE measurements. Additionally, CBE measurements for individual corneal regions were not included. Future studies incorporating region-specific comparisons between NBE and CBE would provide more insight and further validate their agreement.

In conclusion, our study highlights the potential of NBE as a reliable, noninvasive for assessing corneal sensitivity, but with no interchangeable values with CBE. Compared to CBE, NBE might correlate with corneal nerve features and clinical ocular surface characteristics better. Our findings provide a better understanding of the applicability of NBE in clinical practice with a broader spectrum of ocular surface conditions.

Manuscript no. XOPS-D-26-00122.