Diagnostic Agreement Between Invasive and Non-invasive Assessments for Preoperative Dry Eye Disease Screening in Patients with Cataracts

Abstract

Introduction

The aim was to assess the incidence of dry eye disease (DED) and determine the cut-off value for non-invasive keratography tear break-up time (NIKBUT) in patients of European descent with cataracts and without significant comorbidities and to test the diagnostic agreement between invasive and non-invasive protocols.

Methods

DED symptoms and signs were evaluated preoperatively, including the SPEED II questionnaire, fluorescein break-up time (FBUT), ocular surface staining (OSS), and NIKBUT first (NIKBUT-f) and average (NIKBUT-avg). DED diagnosis was based on a SPEED score > 5 and FBUT ≤ 5 s or OSS > 1. The receiver-operating characteristic curves were determined for NIKBUT's detection of DED. The established NIKBUT-f cut-off was used instead of FBUT for alternative DED diagnostic protocols.

Results

Out of 715 patients, 217 were eligible, with a mean age of 70.9 ± 7.9 years. Thirty per cent of patients manifested both signs and symptoms of DED; 21% had symptoms only and 22% signs only. Forty-five per cent of asymptomatic patients presented abnormal FBUT and/or OSS. FBUT was shorter than NIKBUT-f (p < 0.001, Wilcoxon) with a mean difference of 2.5 s and narrower limits of agreement for shorter BUT. The optimal diagnostic cut-off for NIKBUT-f was 7.3 s, with 69.7% sensitivity and 62.3% specificity. Replacing FBUT with the optimised NIKBUT-f threshold resulted in 2% false negatives and the highest (91%) agreement compared with other studied protocols. The fully non-invasive protocol based on SPEED and NIKBUT alone resulted in 9% of patients not being assigned to DED diagnosis.

Conclusions

To maximise the agreement with eye care professional diagnosis and to minimise the number of false negatives, it is recommended to decrease the standard NIKBUT-f cut-off value of 10 s for patients with cataracts. A fully non-invasive approach will result in fewer patients being referred for preoperative eye surface treatment. Ocular staining should not be omitted from the DED diagnosis.

Key Summary Points

Why carry out this study?

Routine ocular surface screening before cataract surgery is essential to reduce the risk of postoperative dry eye disease.

There is no uniform diagnostic protocol for this group of patients, i.e., older individuals with cataracts.

Screening with non-invasive measurements can replace slit-lamp eye care professional (ECP) evaluation, but the cut-off values suggested by the Dry Eye Workshop Reports are not appropriate for elderly patients with cataracts.

What was learned from the study?

For patients with cataracts, it is recommended to lower the cut-off value for the noninvasive tear film break-up time to 7 s to maximise diagnostic accuracy with the professional assessment by ECP and minimise the number of false-negative results.

A fully non-invasive approach results in fewer patients being referred for preoperative eye surface treatment than examination including fluorescein.

Introduction

Undiagnosed dry eye disease (DED) prior to cataract surgery may lead to inaccurate keratometric measurements, errors in intraocular lens power calculation and an increased risk of exacerbating DED signs and symptoms after cataract surgery. Preoperative diagnosis of DED is essential to implement appropriate treatment strategies to improve both postoperative refractive precision and patient-reported outcomes [1]. Optimally, DED screening should be routinely performed as a first preoperative assessment in patients referred to cataract surgery.

The clinical criteria for diagnosing DED in the elderly, cataract population are not sufficiently well defined in terms of both clear cut-off values and a common diagnostic protocol [2, 3]. The existing cut-off values for various tests were established for different age groups, usually young adults or middle-aged individuals with varied ethnicities [4, 5], and the proposed diagnostic protocols are sometimes difficult to perform in a basic clinic. The American Society of Cataract and Refractive Surgeons (ASCRS) protocol for preoperative examination of DED requires the use of specialized equipment that is not commonly available in clinics, such as MMP-9 enzyme tests and osmometers [1]. To ensure broad adoption, clinical tests must be easily accessible, optimally cost-effective and supported by validated threshold values to distinguish between normal and diseased subjects in specific populations, e.g. older patients with cataracts [6, 7].

The Dry Eye Workshop (DEWS) Report provided by the Tear Film & Ocular Surface Society (TFOS) has developed a suggested guideline for assessing the ocular surface based on the scientific literature (TFOS DEWS II and III). The most common diagnostic method is fluorescein tear film break-up time (FBUT), with a 10-s cut-off value [8]. Newer studies and the most recent TFOS DEWS III Report suggest a lower diagnostic threshold of 5 s to differentiate between normal and patients with DED [4, 9, 10]. However, due to FBUT's invasive nature and the difficulty in controlling the amount of dye applied, which may affect the results [9, 11], DEWS recommends using non-invasive methods to measure tear film break-up time. Several cut-off values for non-invasive tear break-up time (NIBUT) have been established to date. TFOS DEWS II and III proposed a 10-s threshold for NIBUT as the first detected tear film break-up [4], but many studies suggest lower values, ranging from 4.8 to 9 s. [4, 5, 12]. However, the mean age of subjects studied is usually much younger (range 24–39 years) than for the average patients with cataracts. To the best of our knowledge, no studies have been conducted to establish an appropriate NIBUT threshold for differentiating DED from normal eyes in elderly individuals with cataracts. Ageing affects tear production, meibomian gland function and goblet cell density independently from other known risk factors [13, 14]. Therefore, cut-off values may need some adjustment to account for the inferior ocular surface in older patients.

The lack of clear diagnostic criteria leads to differences in the reported prevalence of preoperative DED among patients with cataracts, which ranges from 50 to 80% [15–18]. The discrepancy in results may also depend on the population studied and coexisting consistent risk factors for DED. Finally, increasing access to technologies that enable NIBUT measurements, often as an additional module in different keratographs or slit lamps, raises the question of appropriate diagnostic thresholds. Device software differs in how it objectively analyses tear-film breakup, which affects the recorded values [19, 20]. Many of these devices use their own NIBUT parameter nomenclature; for example, the Keratograph (Oculus, Germany) introduced non-invasive keratograph break-up time, NIKBUT.

A universal, simple, validated screening protocol for DED in patients with cataracts is needed, incorporating noninvasive tear film assessments with reference values adopted for this group. The challenges include discrepancies in reported DED prevalence, coexisting systemic and ocular diseases, diagnostic test thresholds, and the choice and number of dry eye tests required. To address this issue, the aim of the present study was to assess the prevalence of dry eye in elderly white patients with cataracts who had no systemic or ocular comorbidities to determine the appropriate cut-off value for NIKBUT in this group.

Methods

Study Design and Participants

The study was approved by the Ethical Committee of Wroclaw Medical University (no. 734/2022) and followed the tenets of the Declaration of Helsinki and good clinical practice. Written informed consent was obtained from all participants prior to being enrolled. This study was conducted as part of a broader prospective controlled clinical study (registration number at Polish National Center of Science 2021/42/E/ST7/00345), with data collected in one clinical ophthalmology site in Poland, and included consecutive patients who were at least 50 years of age, scheduled to undergo cataract surgery from January 2022 to January 2025.

Ocular exclusion criteria were any intra- or extraocular surgical procedure in the previous 6 months, any corneal laser vision correction surgery in either eye, eyelid malposition, significant blepharitis, ocular rosacea, allergic conjunctivitis, infectious conjunctivitis or significant corneal abnormalities (e.g. epithelial basement membrane disease, EBMD), pterygium, use of contact lenses, topical antiglaucoma medications, topical antibiotics, topical nonsteroidal anti-inflammatory drugs, and topical steroid or cyclosporin in either eye. Patients who already had a dry eye disease diagnosis and used eye lubricants on a regular basis were also excluded. General exclusion criteria were Sjogren syndrome and other autoimmune diseases or systemic medications (such as antihistamines, antidepressants, anxiolytics and isotretinoin) known to impair the ocular surface [8]. Patients with well-controlled type II diabetes without diabetic retinopathy or maculopathy, well-controlled and stable hypothyroidism and mild hypertension were included in the analysis [21].

Preoperatively, all eligible subjects underwent evaluation of DED symptoms and signs. Symptoms were assessed using the Preoperative OSD SPEED II^© questionnaire (Ocular Surface Disease Standard Patient Evaluation of Eye Dryness) adopted by the American Society of Cataract and Refractive Surgery (ASCRS) Cornea Clinical Committee [1], which included additional questions to help screen for other types of ocular surface disorders, blepharitis, visual significance and potential infection risk. These extra questions do not have scoring, but the number of checked red boxes (known as red flags) are recorded for reference (maximum 18). The higher the number of red flags, the greater is the suspicion of visually significant DED, requiring preoperative treatment [1]. Permission for using the ASCRS SPEED II^© OSD questionnaire was given by its developer, Christopher E. Starr. Hereafter, the questionnaire is referred to as SPEED II.

The DED signs were tested in both eyes with non-invasive and invasive tests, starting from the least invasive, according to TFOS DEWS II recommendations [22]. First, the tear meniscus height (TMH) and the first and average non-invasive keratograph break-up times (NIKBUT-f, NIKBUT-avg) were examined with the Keratograph 5M (K5M, Oculus, version v3.10.0.0). Central TMH was measured with digital calipers using the K5M software. NIKBUT-f was assessed as the time between the double blink and the appearance of the first distortion in the reflected Placido rings. NIKBUT-avg was the average time of all tear film break-ups occurring in the measured period of up to 25 s (time limit set by the device’s software). Two K5M measurements were taken for each eye, separated by 2 min, and the average of each score was further analysed. Ocular hyperaemia was assessed with the R-scan for bulbar redness. An image of the entire eye was taken with a high depth of field, and the results of the automated grading of the image for the JENVIS 0–5 grading scale were recorded (User Guide Keratograph 5M) [23]. Then, slit-lamp assessments were conducted in the following order: conjunctival hyperaemia evaluation, lid-parallel conjunctival fold (LIPCOF) grading, fluorescein break-up time, and corneal and conjunctival fluorescein staining. Ocular hyperaemia was graded from 0 to 3 (0 = no hyperaemia, 1 = mild, 2 = intense, 3 = very intense) [24]. LIPCOF was assessed with a slit lamp using the grading scale proposed by Pult et al. [25]. To measure FBUT, a fluorescein strip (Madhu Instruments Pvt. Ltd) was moistened with a drop of saline solution and applied to the temporal canthal lid margin area. The slit-lamp was set at 10 × magnification with cobalt blue light and a Wratten 12 yellow filter to optimise observation of the tear film. Using a stopwatch, the time was recorded (in seconds) from the last complete blink to the first appearance of a dark spot. The final FBUT score was calculated as the average of three consecutive measurements. Fluoresceine corneal and conjunctival staining (Ocular Staining Score, OSS) was based on the Oxford schema. Staining ranges from 0 to 15 for the total exposed inter-palpebral conjunctiva and cornea as a sum of 0 to 5 grading for the cornea, temporal conjunctiva and nasal conjunctiva [26]. Grade 1 staining means < 10 punctate spots; therefore, OSS > 1 corresponds to ocular staining criteria for DED proposed by TFOS DEWS III [10, 26].

The diagnosis of DED was made when the SPEED II score was > 5 points [10, 27, 28] and at least one of the following criteria were met: FBUT ≤ 5 s [9] or OSS > 1 [10, 26]. The percentage distribution analysis of other positive signs of DED in examined subjects was based on the following cut-off values: TMH < 0.2 mm; LIPCOF ≥ 1; ocular hyperaemia (slit lamp and K5M) > 1.

In addition, an alternative approach to DED diagnosis was used, replacing FBUT with the NIKBUT-f with the cut-off value determined here (optimised NIKBUT-f). The following diagnostic protocols were tested:

1. Protocol 1: SPEED, FBUT and/or OSS, as the gold standard;

2. Protocol 2: SPEED, NIKBUT-f < 10 s and/or OSS, as suggested by TFOS DEWS III Report;

3. Protocol 3: SPEED, NIKBUT-f ≤ optimised NIKBUT-f (s) and/or OSS;

4. Protocol 4: SPEED, NIKBUT-avg < 10 s and/or OSS, according to the reported clinical screening protocols for patients with cataracts [17, 18];

5. Protocol 5: SPEED, NIKBUT-avg < 7 s and/or OSS, as the protocol based on K5M evaluation recommended by the manufacturer;

6. Protocol 6: non-invasive: SPEED and NIKBUT-f ≤ optimised NIKBUT-f, as the entirely non-invasive simplified protocol.

Statistical Analysis

As this was an observational study, it was not powered to determine statistical significance. Analyses were completed using Statistica v.13.3 (StatSoft) and the Stat Plus PRO v7.8.4 for Microsoft Excel (AnalystSoft Inc.). To calculate the mode, the kernel density estimation method was used with a Gaussian kernel and the default bandwidth in Matlab R2024b (MathWorks Inc.). Only one eye (right) of each subject was used for the statistical analysis [29]. When the right eye was not eligible, left eye data were used. The Shapiro-Wilk test was used to test the normality of data distribution. Wilcoxon matched pairs test was used to test within-patient differences among FBUT, NIKBUT-f and NIKBUT-avg values, and Bland-Altman analysis was performed to evaluate the agreement between FBUT and NIKBUT. Spearman’s rank correlation coefficient (ρ) was calculated to assess the relationship between measured indices. The sample size required to determine whether the correlation coefficient differs from zero and is statistically valid was calculated according to Hulley et al. [30]. Level of significance was set to 0.05 for all tests. The receiver-operating characteristic curve (ROC) and associated area under the curve (AUC) were determined for NIKBUT-f and NIKBUT-avg. Youden optimal diagnostic cut-off sensitivity and specificity were calculated. The comparisons between normal and patients with DED were performed using Mann-Whitney U test with continuity correction. A Venn diagram was applied to illustrate all possible logical relations between indices used to diagnose DED.

Results

Patient Demographics and DED Index Characteristics

Out of 715 patients, 217 were eligible for the study (37 left eyes, 180 right eyes, 164 females). The mean age of patients was 70.9 ± 7.9 (range: 50–93) years. All were of European descent. One hundred seventeen patients (54%) had systemic hypertension, 48 patients (22%) had diabetes type II without diabetic maculopathy or retinopathy, and 26 patients (12%) had hypothyroidism; all these chronic conditions were well controlled with medical treatment. DED index descriptive statistics for the entire cohort are presented in Table 1.

Table 1.

Median, interquartile range (IQR), 95% confidence intervals (CI) and range of DED indices in the examined population (n = 217)

	Median	IQR	95% CI	Min	Max
SPEED score	6.00	7.00	5.85–7.24	0.00	20.00
Red flags number	4.00	4.00	3.72–4.36	0.00	11.00
FBUT	6.00	6.33	6.46–7.60	1.00	22.33
NIKBUT-F	7.55	8.98	8.73–10.40	1.00	25.00
NIKBUT-AVG	12.81	9.69	12.44–14.04	1.00	25.00
Ocular staining score	0.00	2.00	0.92–1.36	0.00	9.00
TMH	0.33	0.18	0.34–0.38	0.12	0.97
Hyperaemia K5M	1.50	0.60	1.42–1.55	0.50	2.50
Hyperaemia slit lamp	0.00	1.00	0.47–0.65	0.00	3.00
LIPCOF	0.00	1.00	0.43–0.62	0.00	3.00

SPEED Standard Patient Evaluation of Eye Dryness, FBUT fluorescein tear break-up time, NIKBUT-f non-invasive keratography tear break-up time first, NIKBUT-avg non-invasive keratography tear break-up time average, TMH tear meniscus height, K5M Keratograph 5M, LIPCOF lid-parallel conjunctival folds

The distribution of SPEED scores, FBUT, NIKBUT-f and NIKBUT-avg values are shown in Fig. 1. The SPEED score mode was 4.3. The break-up time (BUT) results followed different patterns. Modes of FBUT, NIKBUT-f and NIKBUT-avg were 3.83 s, 5.23 s and 11.01 s, respectively. The Wilcoxon test revealed significant differences between FBUT and NIKBUT-f and between FBUT and NIKBUT-avg (p < 0.001).

Fig. 1.

Histograms of Standard Patient Evaluation of Eye Dryness (SPEED) score, fluorescein tear break-up time (FBUT), non-invasive keratography tear break-up time first and average (NIKBUT-f and NIKBUT-avg)

Table 2 summarizes the number of patients who exceeded the predefined cut-off value for each DED index assessed separately.

Table 2.

The percentage of eyes presenting values above and below specific cut-offs for DED indices in the examined cohort

Testing method	Number of eyes (n = 217)	Percentage of eyes
SPEED score
> 5	111	51
≤ 5	106	49
TMH (mm)
≤ 0.2	28	13
> 0.2	189	87
FBUT (s)
≤ 5	94	43
> 5	123	57
NIKBUT-f (s)
< 10	139	64
≥ 10	78	36
NIKBUT-avg (s)
< 10	71	33
≥ 10	146	67
Ocular staining score (Oxford)
> 1.0	62	29
≤ 1.0	155	71
LIPCOF
> 1	23	11
≤ 1	191	88
Missing	4	1
Ocular hyperaemia SL (slit lamp)
≥ 1	102	47
< 1	111	51
Missing	4	2
Ocular hyperaemia K5M
> 1	145	67
≤ 1	24	11
Missing	48	22

The percentage of patients exceeding the DED cut-off threshold is marked in bold

SPEED Standard Patient Evaluation of Eye Dryness, TMH tear meniscus height, FBUT fluorescein tear break-up time NIKBUT-f non-invasive keratography tear break-up time first, NIKBUT-avg non-invasive keratography tear break-up time average, LIPCOF lid-parallel conjunctival folds, K5M Keratograph 5M

There was a positive correlation between FBUT and NIKBUT-f (ρ = 0.47, p < 0.001, Spearman) and between FBUT and NIKBUT-avg (ρ = 0.45, p < 0.001). The correlation between SPEED score and ocular signs was only significant for FBUT (ρ = 0.189, p = 0.005). Other correlations that were significant for the measured patients are summarized in Table 3.

Table 3.

Statistically significant correlations between measured DED indices, considering the sample size required to determine whether the correlation coefficient differs from zero

Pair of variables	ρ; p	n
Age—hyperaemia K5M	0.185; 0.006	216
Hyperaemia SL—hyperaemia K5M	0.339; < 0.001	166
Hyperaemia SL—LIPCOF	0.351; < 0.001	213
Hyperaemia SL—FBUT	− 0.334; < 0.001	213
Hyperaemia SL—OSS	0.278; < 0.001	213
LIPCOF—FBUT	− 0.232; < 0.001	214
LIPCOF—OSS	0.207; 0.002	214
SPEED score—red flags number	0.436; < 0.001	212

K5M Keratograph 5M, SL slit lamp, hyperaemia K5M assessed with K5M, hyperaemia SL = assessed with slit lamp, OSS ocular staining score, LIPCOF lid-parallel conjunctival folds, FBUT fluoresceine break-up time, SPEED Standard Patient Evaluation of Eye Dryness

DED Diagnosis

Full data of SPEED II, FBUT and OSS were available for 217 subjects. The DED diagnosis criteria were fulfilled for 66 (30%) patients (55 females). The discordance between DED signs and symptoms was observed (Table 4). One hundred eleven patients (51%) reported DED symptoms, whereas one hundred fourteen patients (53%) manifested FBUT ≤ 5 and/or OSS > 1 preoperatively, with 48 of these patients (40%) being asymptomatic. Forty-eight patients (22%) presented DED signs without symptoms, and a similar number (45, 21%) of patients reported symptoms only, without signs of DED. There was no significant difference in the number of red flags in SPEED II between symptomatic patients with signs of DED and symptomatic patients without DED signs (p = 0.51, Mann-Whitney U). The incidence of abnormal FBUT and/or OSS among asymptomatic patients (n = 106) was 45%. Forty per cent of symptomatic patients had no clinical signs of DED. Table 4 summarises the DED diagnoses.

Table 4.

DED diagnosis based on SPEED II questionnaire score and FBUT and/or OSS

n = 217	FBUT > 5 s and/or OSS ≤ 1	FBUT ≤ 5 s and/or OSS > 1
SPEED ≤ 5	58 (27%) (no-DED)	48 (22%) (signs only)
SPEED > 5	45 (21%) (symptoms only)	66 (30%) (DED)

SPEED Standard Patient Evaluation of Eye Dryness, OSS ocular staining score, FBUT fluoresceine break-up time, DED dry eye disease

Patients with DED diagnosis presented shorter NIKBUT-f and NIKBUT-avg, higher OSS and ocular hyperaemia scores (slit lamp only) and higher scoring for additional ASCRS modified SPEED II extra questions (red flags) than patients who did not meet the criteria for DED diagnosis (Table 5). LIPCOF grading as well K5M ocular hyperaemia and TMH showed no statistically significant differences between patients with and without DED. More women and higher incidence of hypothyroidism were found in the DED group. No age differences were noted between patients with and without DED as well as the group with no signs and symptoms and the rest of the patients.

Table 5.

Characteristics of patients with cataracts, with and without dry eye disease

	No DED n = 151	DED n = 66	p*
Demographics
Age	72 (9.0)	71 (7.5)	0.902
Male	42 (28%)	11 (17%)
Female	109 (72%)	55 (83%)
Hypertension	81 (54%)	36 (54%)
Diabetes	33 (22%)	15 (23%)
Thyroid disorder	13 (9%)	13 (20%)
DED tests
SPEED score	4.0 (4.0)	10.0 (6.5)	< 0.001
Red flags number	4 (3)	5 (3)	< 0.001
FBUT	8.0 (5.3)	3.3 (1.3)	< 0.001
NIKBUT-f (s)	9.0 (9.4)	5.0 (5.0)	< 0.001
NIKBUT-avg (s)	13.7 (9.4)	10.8 (7.8)	< 0.001
TMH (mm)	0.33 (0.17)	0.34 (0.22)	0.566
Ocular hyperaemia K5M	1.4 (0.7)	1.5 (0.5)	0.420
Ocular hyperaemia slit lamp	0 (1)	1 (1)	0.002
LIPCOF	0 (1)	0 (1)	0.468
Ocular staining score	0 (1)	2 (2)	< 0.001

Numerical variables are presented as median and interquartile range; categorical data are presented as numbers and percentages

DED dry eye disease, SPEED Standard Patient Evaluation of Eye Dryness, FBUT fluoresceine break-up time, NIKBUT-f non-invasive keratography tear break-up time first, NIKBUT-avg non-invasive keratography tear break-up time average, TMH tear meniscus height, LIPCOF lid-parallel conjunctival folds

Figure 2 shows the agreement between NIKBUT-f and FBUT in the no-DED group and the DED group using Bland-Altman plots, A and B, respectively. In both cases, the mean difference was about 2.5 ± 1.0 s, indicating shorter FBUT. However, the limits of agreement (LoA) were almost twice as wide in the no-DED group than in the DED group: 24.47 s and 15.76 s, respectively. The LoA as well as the difference between NIKBUT-f and FBUT was visibly smaller for shorter BUT. For NIKBUT-avg, the mean difference (NIKBUT-avg – FBUT) was 5.9 ± 1.0 s and 7.0 ± 1.2 s for patients without DED and patients with DED, respectively, and LoAs were 22.56 s and 20.10 s, respectively.

Fig. 2.

Bland-Altman plots showing the agreement between objective first noninvasive break-up time (NIKBUT-f) and fluorescein break-up time (FBUT) for patients with DED (a) and no-DED patients (b) and the agreement between objective average noninvasive break-up time (NIKBUT-avg) and fluorescein break-up time (FBUT) for DED patients (c) and no-DED patients (d). The solid lines represent the mean difference and the dashed lines represent the upper and lower limits of agreement between the DED indices. The error bars show the confidence intervals (CI)

For the entire cohort, the average FBUT was 2.54 s shorter than for NIKBUT-f (95% CI: 1.78 to 3.29 s) and 6.21 s shorter than for NIKBUT-avg (95% CI: 5.46 to 6.96 s) based on Bland-Altman plots (not shown in the figure). LoAs were 22.14 s and 21.95 s for NIKBUT-f and NIKBUT-avg, respectively.

The efficacy of NIKBUT-f and NIKBUT-avg for DED diagnosis was examined using ROC curve analysis (Fig. 3 AB). The area under the curve (AUC) for NIKBUT-f was 0.706 (95% CI 0.633–0.778, p < 0.001) and the proposed cut-off value was 7.3 s with 0.697 sensitivity and 0.623 specificity with a Youden index value of 0.319. For NIKBUT-avg, the AUC was 0.644 (95% CI 0.577–0.730, p = 0.001) and the proposed cut-off was 15.66 s with 0.833 sensitivity and 0.430 specificity with a Youden index value of 0.264.

Fig. 3.

Receiver-operating characteristic (ROC) curves of discrimination ability in the detection of dry eye disease using non-invasive keratography break-up time first (NIKBUT-f) (a) and non-invasive keratography break-up time average (NIKBUT-avg) (b)

Non-invasive Versus Invasive DED Diagnosis

The alternative approach to DED diagnosis was used, replacing FBUT with the NIKBUT-f test, with a determined cut-off value of 7.3 s. Table 6 shows the number of patients categorised into DED positive or DED negative based on clinical signs and symptoms using the above-mentioned cut-off values and different diagnostic protocols. The highest agreement (91%) with an eye care professional (ECP) diagnosis was achieved by protocol 3, closely followed by protocols 4 and 5. Although the number of DED-positive patients was similar for protocol 1 and 4, not the same patients were diagnosed with DED using these two approaches.

Table 6.

Number and percentage of patients with positive DED tests results for invasive and non-invasive diagnostic protocols, with false positively and false negatively diagnosed patients with agreement to protocol 1

	Protocol	DED positive n (%)	DED negative n (%)	False positive n (%)	False negative n (%)	Agreement %
1	FBUT and/or OSS + SPEED	66 (30)	151 (70)	NA	NA	NA
2	NIKBUT-f < 10 s and/or OSS + SPEED	82 (37)	135 (62)	20 (9)	4 (2)	89
3	NIKBUT-f < 7.3 s and/or OSS + SPEED	76 (35)	141 (65)	15 (7)	5 (2)	91
4	NIKBUT-avg < 10 s and/or OSS + SPEED	66 (30)	151 (70)	11 (5)	11 (5)	90
5	NIKBUT-avg < 7 s and/or OSS + SPEED	57 (26)	160 (74)	6 (3)	15 (7)	90
6	NIKBUT-f < 7.3 s + SPEED	61 (28)	156 (72)	15 (7)	20 (9)	84

DED dry eye disease, FBUT fluoresceine break-up time, OSS ocular staining score, NIKBUT-f non-invasive keratography tear break-up time first, NIKBUT-avg non-invasive keratography tear break-up time average, SPEED Standard Patient Evaluation of Eye Dryness, NA not applicable

The Venn diagram for FBUT, NIKBUT-f, OSS and SPEED shows varying patient counts for different diagnostic protocols and summarises the coexisting pathological values considering the optimised NIKBUT-f < 7.3 s (Fig. 4). Only 22 patients met all diagnostic criteria applied cumulatively (n = 22). Despite considering three indices to assess DED signs, 30 patients still reported only positive SPEED.

Fig. 4.

Venn diagram for invasive and non-invasive diagnostic protocols. Cut-off value of non-invasive keratography tear break-up time first (NIKBUT-f) was determined from the receiver-operating characteristic curve

Discussion

In this cohort of generally healthy white patients with cataracts who had not previously been diagnosed with ocular surface disease, nearly one-third manifested both signs and symptoms of DED preoperatively, as assessed by FBUT, fluorescein ocular staining and the SPEED II questionnaire. Almost half of asymptomatic patients presented abnormal FBUT and/or OSS; conversely, > 40% of symptomatic patients did not have clinical signs of DED. Replacing FBUT with optimised NIKBUT-f resulted in only 2% false-negative results while maintaining a relatively low false-positive rate. However, the fully non-invasive screening model failed to identify approximately 9% of patients, highlighting the importance of ocular staining evaluation.

FBUT remains the most accessible method for DED evaluation in every ophthalmology setting. Considering only this single criterion, 43% of patients were below the 5-s threshold in the cohort examined here. Studies on groups of patients referred for cataract surgery with an accepted FBUT cut-off value of 5 s reported higher percentages of patients with preoperative unstable tear film: 57% in a Japanese population [31] and 62.9% in patients of European descent [15], which may be because of possibly less “healthy” populations being studied. Notably, the cohort studied here was rigorously selected to possibly avoid the significant, consistent risk factors of DED described in the TFOS DEWS II and III Digest Epidemiology Section [10]. Nevertheless, hypertension, type II diabetes and hypothyroidism are frequent in the older population [32–34] and selecting patients without these common diseases not only would hardly be feasible but also would create an unrealistic population. While the study included individuals with only mild, pharmacologically controlled hypertension, diabetes and hypothyroidism, 50% of patients with hypothyroidism were diagnosed with DED and 31% with hypertension or type II diabetes. In a recent multicentre study with much broader inclusion criteria (a less healthy population), the prevalence of abnormal FBUT (< 10 s) ranged from 42.9 to 62.3%, demonstrating that the incidence of DED signs depends on the population studied and protocols used [35].

The symptoms of DED were examined with the SPEED II questionnaire, which is more appropriate than OSDI for patients with compromised visual acuity due to cataracts, as it helps to avoid biased responses to questions about eye problems that limit visual tasks such as driving or watching television [36]. In the studied cohort, 40% of symptomatic patients had no signs of DED; conversely, 45% of asymptomatic patients presented abnormal FBUT and/or OSS. Moreover, 22% of all patients exhibited DED signs only, and a further 21% of all patients exhibited DED symptoms only. The discordance between patient-reported symptoms and signs evaluated by ECP is well described [10]. As we excluded patients with significant blepharitis, allergy, corneal EBMD and any comorbidities that could be responsible for neuropathic pain or neurotrophic conditions, the overrating of symptoms may have been caused by the general eye strain and decreased well-being in preoperative patients with cataracts, with aggravation of eye irritation. The study demonstrated a correlation between the SPEED II score and the number of red flags, and the statistically higher number of flags in the DED group, indicating a higher incidence of episodes of mild blepharitis, typical for ageing eyes (significant blepharitis was excluded), fluctuating vision and the voluntary, irregular use of eye lubricants. For clinicians, red flags identified in the SPEED II questionnaire indicate the need for preoperative eyelid hygiene and the preventive use of artificial tears following cataract surgery. Therefore, using the ASCRS-modified SPEED II with additional questions appears more appropriate for preoperative DED screening than the OSDI.

Various diagnostic protocols used for DED and several cut-off values were proposed for DED indices. As noninvasive methods for assessing tear-film break-up time become more widely available, clinicians performing preoperative DED screening in patients with cataracts need clear recommendations regarding diagnostic thresholds. Notably, indices similar to NIKBUT-f are recorded by all devices dedicated to noninvasive tear film assessment. The TFOS DEWS II Report recommends a 10-s threshold for NIKBUT-f only, while published studies refer to NIKBUT-f, NIKBUT-avg or both [3, 5, 17, 18]. Therefore, we compared the “gold standard” ECP protocol using the FBUT, OSS and SPEED II with alternative protocols based on NIKBUT-f and NIKBUT-avg. The cut-off values for protocols 2, 4 and 5 were derived from TFOS DEWS II/III, studies reporting DED prevalence in patients with cataracts and K5M User Guide recommendations, respectively [8, 10, 17, 18, 23]. The DEWS II/III recommendation (protocol 2) may overestimate the prevalence of DED in this older cohort, prompt unnecessary diagnostic workups and adversely influence counselling about premium intraocular lenses, given that DED is a contraindication to multifocal IOLs.

Although both false-negative and false-positive DED diagnoses are undesirable, clinically, unrecognized DED cases are of greater importance for preoperative screening. Interestingly, protocols employing NIKBUT-avg (protocol 4 and 5) yielded two to three times more false-negative patients than protocols based on NIKBUT-f (protocol 2 and 3). Specifically, protocol 5, based on the K5M threshold used in the JENVIS Pro Dry Eye Report, failed to detect the highest number of DED cases in this cohort.

Busy cataract clinics would benefit from a simple preoperative DED screening protocol performed by a technician. Unfortunately, the simplified, quick, fully non-invasive diagnostic approach (protocol 6) demonstrated the poorest agreement with ECP diagnosis and the greatest false-negative rate. This supports the validity of including ocular staining in preoperative DED screening, which is consistent with the observations of Wolffsohn et al. in a much larger group of patients [36]. Notably, patients with a negative result on the simplified DED protocol due to high NIKBUT-f showed clinically significant OSS, with only three exceptions.

Studies reporting correlations between FBUT and NIKBUT present varying results [3, 4]. This study reported similar mean differences between FBUT and K5M as a study by Wang et al. (2 s and 6.9 s for NIKBUT-f and NIKBUT-avg, respectively), although they studied a young cohort [4]. These results are consistent with the well-known destabilizing effect of fluorescein on the tear film [22]. Narrower limits of agreement between NIKBUT-f and FBUT were observed for DED than for patients without DED, suggesting the possibility of establishing a rational threshold for clinical application. A similar observation of better agreement for less stable tear film was previously described by Szczesna-Iskander et al. in a younger group, which showed a quasi-monotonical increase in the width of the limits of agreement with the mean BUT [20]. They also showed that the width of the limits of agreement of the average BUT below 10 s was equal to 14 s and above 10 s was 27.5 s; this is a similar result to that obtained in this study for the DED group (15.8 s) and no-DED (24.5 s) group, respectively. Considering the mean difference between FBUT and NIKBUT-f (2.5 s), the NIKBUT-f cut-off value of 10 s may need optimisation, in line with the results of the ROC performed in this study.

The threshold value for NIKBUT-avg was also established, as this index is used in the DED diagnostic protocol in the K5M software (JENVIS Pro Dry Eye Report). The K5M manual classifies tear film stability into three levels, where NIKBUT-avg ≥ 7 s to < 14 s is considered critical and NIKBUT-avg < 7 s indicates unstable tear film. The reported clinical screening protocols for patients with cataracts consider a NIKBUT-avg < 10 s to be the diagnostic threshold for DED [17, 18]. With this value, a percentage of DED-positive subjects in a population of Norwegian patients with cataracts, similar in age to the cohort studied here, was 29.8%, which is also similar to the results for protocol 4 in our group [17]. However, using diagnostic protocol 4, 5% of patients were falsely classified compared to the ECP diagnosis. Setting a threshold at 7 s would lead to even more undiagnosed patients (Table 6). Therefore, protocols using NIKBUT-avg as a DED diagnostic criterion must be used with caution. Considering the wide limits of agreement between FBUT and NIKBUT-avg and the generally averaging nature of this indicator, setting a single cut-off value may not be feasible. The NIIKBUT-avg threshold established here is twice as high as the K5M definite cut-off and needs further clinical evaluation, preferably for a less healthy population.

The Venn diagram showed that the overlap of specific signs and symptoms yields varying patient counts. Only 10% of all subjects were concurrently positive for FBUT, NIKBUT-f, SPEED and OSS. Therefore, using multiple tests for preoperative screening may not be helpful. Furthermore, LIPCOF grading and K5M ocular hyperaemia showed no significant differences between patients with and without DED in this elderly population. The highest number of patients with a single positive DED indicator was observed for the SPEED questionnaire alone (n = 30), once again highlighting the overreporting of symptoms in the ageing population. Notably, the only significant correlation between the SPEED score and ocular signs in this cohort was with FBUT.

The limitations of this study include the lack of osmolarity and MMP-9 enzyme testing proposed by ASCRS as part of preoperative screening for ocular surface disease, since these tests are not reimbursed in most European Union countries. According to TFOS DEWS III, inflammation is more frequent in severe cases of DED and MMP-9 is less sensitive in the more common mild-to-moderate cases of DED [10]. For conjunctival staining, lissamine green was not used because this vital dye has no valid registration for human use in Poland. Although the TFOS DEWS II and III Reports recommend lid wiper epitheliopathy (LWE) assessment in DED diagnosis, it was not assessed here to maintain diagnostic simplicity. A recent study by Wolffsohn et al. found that the incidence of LWE is similar in patients with OSDI scores above and below the dry eye cut-off; therefore, including this parameter is unlikely to affect the diagnosis [36]. Gender imbalance may also be considered as a limitation of this study, with 75% of subjects being women. This could have influenced the DED prevalence in our group. The data presented show that the ratio of patients with DED to those without DED was 1:2 among females and 1:4 among men, which is consistent with female gender being a well-known risk factor for DED [21]. However, the aim was to examine the prevalence of DED in patients presenting for cataract surgery during a specific time period and not to attempt to balance gender.

The study's strengths include the use of the SPEED II questionnaire, which avoids bias due to visual impairment in patients with cataracts. Palkovits et al. demonstrated that the impact of cataracts on visual function assessed by the OSDI is significant [37]. Therefore, SPEED II may provide a more reliable assessment of DED-related symptoms before cataract surgery than the commonly used OSDI. Furthermore, the rigorously selected population of generally healthy individuals with cataracts belonging to the same ethnic group, being of European descent, with a mean age of 70 years is also an advantage. Older populations are infrequently included in DED research; therefore, developing validated cut-off values and simple diagnostic protocols suitable for the busy clinical environment is essential to optimise preoperative management and enhance postoperative outcomes.

The results recommend lowering the current NIBUT threshold to 7 s. FBUT can be replaced by an adjusted NIBUT, but ocular-surface staining cannot be omitted. A simple diagnostic protocol consisting of the SPEED questionnaire, NIKBUT-f and fluorescein staining can be applied in clinical practice for quick and robust preoperative DED screening in patients with cataracts. For this task, a well-trained technician or an optometrist could support the ECP.

Conclusions

Approximately one-third of generally healthy white patients with cataracts showed both signs and symptoms of DED prior to surgery, and similar proportion of patients showed neither signs nor symptoms of DED. Screening based on a fully non-invasive shortened protocol (SPEED II + NIKBUT-f) would refer fewer patients for preoperative eye surface treatment than examination with fluorescein dye. This suggests that assessment of ocular surface staining in DED screening is important to avoid underdiagnosis of DED prior to cataract surgery. Furthermore, diagnostic protocols with NIKBUT-avg should be avoided because of the risk of missed DED cases. Finaly, a lower NIKBUT-f cut-off value than that proposed by TFOS DEWS III is likely, as the study focused on an elderly population. Replacing FBUT with an optimised NIKBUT-f threshold resulted in the highest agreement between these two protocols, suggesting that non-invasive methods may be suitable for streamlining preoperative screening.

Author Contribution

Conceptualization: Maria Muzyka-Wozniak, Dorota H. Szczesna-Iskander; Data acquisition: Izabela Kaminska, Patrycja Piwowarczyk, Maria Muzyka-Wozniak; Formal analysis: Maria Muzyka-Wozniak, Dorota H. Szczesna-Iskander, Patrycja Piwowarczyk; Writing—original draft preparation: Maria Muzyka-Woźniak; Writing—review and editing: Maria Muzyka-Wozniak, Dorota H. Szczesna-Iskander.

Funding

This study was supported by The Polish National Science Centre (SONATA BIS Grant 2021/42/E/ST7/00345). The journal's Rapid Service Fee will be financed from funds of the Wrocław University of Science and Technology (authors) obtained as part of a scientific project from the National Science Centre.

Data Availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflict of Interest

Maria Muzyka-Wozniak, Izabela Kaminska, Patrycja Piwowarczyk and Dorota H. Szczesna-Iskander declare that they have no competing interests.

Ethical Approval

The study was approved by the Bioethics Committee at the Medical University of Wroclaw, Poland (no. KB—734/2022) and followed the tenets of the Declaration of Helsinki of 1964 and its later amendments. Written informed consent was obtained from all participants prior to being enrolled.