Introduction

Tonometry is a common and essential diagnostic in veterinary ophthalmic examinations. In conjunction with clinical signs, intraocular pressure (IOP) measurement can support the diagnosis and management of many ocular diseases, most notably glaucoma and uveitis.¹ Before interpreting tonometric readings, it is important to understand how IOP is controlled and what intrinsic and extrinsic factors can change it. Extraocular muscle tone, scleral rigidity, and aqueous humor production and drainage determine the IOP of each globe.² A variety of physiologic factors such as changes in CNS output, blood pressure, venous drainage, and partial pressure of oxygen and carbon dioxide in arterial blood can affect aqueous humor production and drainage.³ Intraocular pressure can also be influenced by age, breed, diurnal variation, and season.^1,4 In addition, IOP can be impacted by extrinsic factors including sedatives, anesthetic agents, topical ophthalmic medications, measurement technique, restraint, and body position.^5–12 Pathological ocular surface changes in corneal thickness and rigidity also have the ability to alter tonometric readings, the degree of which depends on the magnitude of corneal change and the tonometer used.¹

Modern applanation tonometers, such as the Tono-Pen XL, are based on the Mackay-Marg principle which uses both indentation and applanation to calculate IOP.^12,13 The Tono-Pen XL contains a small plunger attached to a micro-strain gauge that converts physical movement of the plunger into electrical signals that are interpreted by the digital microprocessor.¹² Resistance increases as the plunger contacts the cornea until applanation is reached.¹³ A slight decrease in force against the plunger occurs at the point of applanation followed by a further increase.¹³ This dip in the resistance curve is the calculated IOP measurement.¹³ The Tono-Pen XL averages 4 measurements to generate a reading, whereas the newer Tono-Pen AVIA averages 10 measurements.¹² The processor calculates a coefficient of variance (CV) to help interpret the reliability of the average.

Modern applanation tonometers are inherently relatively accurate in estimating IOP, and have low variability. Given these properties and that the majority of extrinsic factors serve to increase IOP, making it easy and common to obtain a falsely elevated IOP reading but rarely a falsely lowered one, in clinical practice, the first (reasonable) or lowest IOP reading with a CV < 5% is often designated as the globe’s value. However, in some practices and commonly in research studies where IOP is measured, the mean of triplicate low-error readings is used as the value.^14–21 This is done to increase both accuracy and precision but requires additional time and effort. To the authors’ knowledge, there is no published veterinary data that quantify the improvement in accuracy and precision resulting from triplicate measurements compared with a single measurement. The additional time and effort required to obtain triplicate readings is only justified if the increase in accuracy and precision over a single reading is clinically meaningful.

The purpose of this study, involving data from 4 previously conducted canine studies,^8–11 was to quantify and assess for clinical significance both the assumed improvement in accuracy of the lowest or averaged triplicate value over the first IOP measurement in clinical practice, and the assumed improvement in both accuracy and precision of averaged values over single (first) IOP measurements when statistically analyzed in research studies. Lowest or averaged triplicate IOP measurements were assumed to be most accurate. Improvement in accuracy and clinical impact of using the lowest or averaged IOP value over the first measurement in clinical practice was quantified by calculating interpercentile ranges of differences of an individual dog determined by using the lowest, averaged or first value. Bias in clinical practice was estimated with the mean or median of these differences. Improvement in accuracy in research studies of using averaged values over a single IOP measurement was quantified by calculating prediction intervals of differences between-treatment means calculated from averaged or first IOP values. Bias in research studies was estimated with the mean of these differences. The full range of probable differences were assessed for potential clinical significance both with respect to an individual dog and research studies. Improvement in precision in research studies due to taking 3 versus 1 IOP measurement was quantified by calculating reduction in SDs and SEs of treatment means. In addition to mean differences and variability, results of inferential statistical analyses (P values) of the 4 research studies were compared to assess increase in statistical power and decrease in Type I error due to triplicate measurements. The hypothesis was that accuracy and precision of the first IOP readings would not differ in a clinically meaningful manner from lowest or averaged triplicate IOP values.

Materials and Methods

Data from 4 previous studies^8–11 evaluating the effect of various anesthetic drug protocols on IOP were analyzed retrospectively. All of these studies were conducted at the same institution (University of Georgia College of Veterinary Medicine) using the same tonometer (Tono-Pen XL; Medtronic Solan). The IOP readings were collected in triplicate at all time points by the same trained individual (EH) masked to the treatment groups.

All studies involved random-source dogs (from municipal pounds) being used for veterinary student surgical exercises. A total of 116 dogs were included across the 4 studies. In 1 (crossover design) study,⁹ some dogs received 3 separate treatment protocols, yielding a total of 135 baseline comparisons. Protocols for all studies were approved by the University of Georgia Animal Care and Use Committee, and husbandry was provided according to established institutional guidelines. Age was not recorded because a definitive age could not be established for most dogs, but all were determined to be adults based on physical examination.

Every dog in each study received a complete ophthalmic examination by an experienced individual masked to the treatment groups. The ophthalmic exams consisted of a quantitative tear test (Schirmer tear test strips; Schering-Plough Animal Health Corp), fluorescein staining (Fluor-I-Strip-A.T.; Bausch & Lomb Pharmaceuticals Inc), applanation tonometry, biomicroscopy, and indirect ophthalmoscopy with pupillary dilation. Any dogs deemed unhealthy on physical examination, or with abnormal packed cell volume (reference range: 35% to 57%), plasma total protein value (reference range: 5.2 to 7.3 g/dL), or ophthalmic examination were excluded from the studies.

Before the first IOP measurement at each time point, 1 drop of 0.5% proparacaine (Bausch & Lomb Pharmaceuticals Inc) was applied topically to each eye. All IOP measurements were performed with the dog standing, sitting, or in sternal recumbency with the head raised, and care taken not to occlude the jugular veins or place pressure on the globe while retracting the eyelids. The same tonometer was factory-calibrated before each study and manually calibrated each day before data collection. Only IOP measurements with < 5% error were recorded, and 3 readings were recorded for each eye at each time point. All dogs in each study were medicated and measured at the same time of day (afternoons) to account for diurnal variation of IOP.

Study A⁸ evaluated the effects of ketamine, diazepam, and the combination of ketamine and diazepam in 50 clinically normal dogs in which premedication was not administered. Dogs were randomly allocated to 1 of 5 groups. Dogs received ketamine alone (5 mg/kg or 10 mg/kg IV), ketamine (10 mg/kg IV) with diazepam (0.5 mg/kg IV), diazepam alone (0.5 mg/kg IV), or saline (0.9% NaCl) solution (0.1 mL/kg IV). IOPs were measured immediately before and after injection and at 5, 10, 15, and 20 minutes after injection.

Study B⁹ evaluated the effects of graded doses of propofol in 11 clinically normal dogs. There were 3 treatment groups: propofol (8 mg/kg IV) until loss of jaw tone, propofol until loss of jaw tone plus 20%, and propofol until loss of jaw tone plus 50%. Atracurium (0.1 mg/kg IV) was administered in all treatments immediately after the propofol. All dogs received the 3 treatments in a randomized order, with at least a 1-week interval between treatments (crossover design). Direct arterial blood pressure and IOP were measured at baseline, after 5 minutes of pre-oxygenation (before induction), before intubation, and after intubation.

Study C¹⁰ evaluated the effects of lidocaine or diazepam administered intravenously (IV) before induction of anesthesia with propofol (8 mg/kg IV until loss of jaw tone) followed by atracurium (0.3 mg/kg IV) and orotracheal intubation in 33 clinically normal dogs. The effect of lidocaine applied topically to the larynx after induction with propofol was also evaluated. Dogs were assigned into the following treatment groups; lidocaine (2 mg/kg IV) with saline (0.9% NaCl) solution (0.1 mL/kg) topically applied to the larynx; saline solution (0.1 mL/kg IV) with lidocaine (2 mg/kg) topically applied to the larynx; diazepam (0.25 mg/kg IV), with saline solution (0.1 mL/kg) topically applied to the larynx; or saline solution (0.1 mL/kg IV) with saline solution (0.1 mL/kg) topically applied to the larynx. IOPs were measured in each eye before premedication, before induction, before intubation, and after intubation.

Study D¹¹ evaluated the effects of propofol and thiopental induction in 22 clinically normal dogs. Dogs were randomly assigned to receive propofol (8 mg/kg IV) or thiopental (18 mg/kg IV) until loss of jaw tone. Direct arterial blood pressure, arterial blood gasses, and IOP were measured at baseline, after pre-oxygenation but before induction, before endotracheal intubation, and after intubation.

Statistical analysis

All analyses were performed using SAS version 9.4 (SAS Institute Inc). A significance threshold of 0.05 was used.

Differences between averaged triplicate and first IOP values, averaged and lowest IOP values, and first and lowest IOP measurements were analyzed. Histograms and Q-Q-plots of differences were examined to evaluate normality. Descriptive statistics of these differences were performed for baseline only and all triplicate IOPs across the 4 studies. A Bland-Altman plot was also constructed for differences between averaged and first IOP.

To account for potential within dog correlation of differences, 1-way random effects models were used to estimate conditional means (model-based estimates) for averaged triplicate values, first IOP measurements, and differences between averaged and first values. The SDs and intra-class correlation coefficients (ICCs) of these differences were also computed. A single random intercept for each dog was used. The model intercept estimated the conditional mean, and the square root of the total model variance (dog variance + residual variance) estimated the SD. The ICC was estimated by dividing the residual variance by total model variance.

Mean (± SD of means) and SD (± SD of SD) of IOPs were calculated for each time point within each treatment group, for right and left eyes separately, for each study and for all studies together, based on alternatively averaged triplicate and first IOPs. Mean differences in means and average percent reduction in SDs both with associated 95% prediction interval (PI) were also calculated for each study separately and all studies together. Predicted SEs and 95% and 99% confidence intervals (CIs) of the mean difference were calculated for group sample sizes of 5 to 100 eyes.

Separate linear mixed-models (LMMs) were used to test for differences between treatments at each time point for each study (A through D) alternatively using first IOP measurements or, averaged triplicate IOP values. Each LMM included fixed factors of treatment, time point and a treatment by time point interaction and random intercepts for each dog and for each eye within each dog. Simple effects were tested to compare treatments at each time point and multiple comparisons were adjusted for using the Tukey test. Conditional means and their SEs were estimated for each time point, treatment and study. P values for each between-treatment comparison of treatments at each time point for each study (107 between-treatment comparisons) were recorded for each alternative analysis.

An LMM using all individual triplicate values was constructed to estimate ICCs for each study and for all studies together. An additional random intercept for each triplicate was added to the LMMs described above to have the residual error represent within triplicate error. The residual error SD was recorded as an estimate of intra-triplicate error (ie, measurement error).

Another LMM was used to test for a decrease or increase from first to third IOP readings separately for averaged triplicate IOP category (< 10, ≥ 10 to 12, ≥ 12 to 14, ≥ 14 to 16, ≥ 16 to 18, ≥ 18 to 20, ≥ 20 mm Hg). This LMM included an additional covariate for triplicate (1st, 2nd, or 3rd).

Satterthwaite degrees of freedom method and restricted maximum likelihood estimation were used in all random and linear mixed effects models. Histograms and Q-Q-plots of model residuals were examined and confirmed the assumption of normality.

Results

The mean ± SD of all 270 baseline single or averaged triplicate IOP measurements from 116 dogs was 14.2 ± 3.3 mm Hg for first readings compared to 14.0 ± 3.1 mm Hg for averaged values, and 12.9 ± 2.9 mm Hg for lowest readings. These means were very close to conditional means ± SD estimates from a linear mixed model that accounted for repeated baseline measurements within dogs: 14.3 ± 3.4 mm Hg for first readings compared to 14.2 ± 3.1 mm Hg for averaged values, and 13.1 ± 3.0 mm Hg for lowest readings. The increased precision as quantified by reduction in SD was 6% when using averaged compared to first baseline values.

Table 1 lists descriptive statistics for the differences between averaged triplicate values and first readings for just baseline and for all data (n = 1432), which were approximately normally distributed. The conditional mean (95% CI) difference between averaged values and first readings over all IOP measurements was –0.08 mm Hg (–0.17 to –0.001; LMM, P = 0.047) indicating a statistically significant yet clinically negligible bias of repeated measurements. The SD of these differences was 1.4 mm Hg; the 2.5th to 97.5th interpercentile range was –3 to 2.7 mm Hg and the 1st to 99th interpercentile range was –4 to 3.7 mm Hg. The ICCs of averaged to first differences for dog were 0% for all baseline data and 2.6% for all data. This indicated that there is negligible within-dog correlation of the differences, and repeated IOP measurement differences from the same dogs (ei, both eyes) can be treated as independent for both descriptive and inferential statistics (LMMs and descriptive model mean differences and SDs were the same to 1 decimal point). A Bland-Altman plot aided visualization of these mean differences with 95% limits of agreement (Figure 1). The Bland-Altman plot showed the tendency for the averaged triplicate value to be higher than the first reading for lower IOP values and lower than the first reading for higher IOP values. It also showed that the variability of the differences appears to increase with increasing IOP values.

View Full Size

Bland-Altman plot of averaged triplicate intraocular pressure (IOP) values versus first IOP readings (n = 1432) in healthy dogs. The middle dashed line represents the mean difference and the lower and upper dashed lines represent the lower and upper 95% limits of agreement.

Citation: American Journal of Veterinary Research 83, 4; 10.2460/ajvr.21.08.0114

• Download Figure
• Download figure as PowerPoint slide

Descriptive statistics for the differences between averaged triplicate IOP values and lowest IOP readings, which were left-skewed and not normally distributed, were tabulated (Table 1). The median (IQR) difference between averaged and lowest readings over all IOP measurements was 1 mm Hg (0.7 to 1.7) indicating an expected negative bias of taking the lowest rather than the average of triplicate measurements. The 2.5th to 97.5th interpercentile range for these differences was 0 to 3.7 mm Hg and the 1st to 99th interpercentile range was 0 to 4.3 mm Hg. The ICCs of average minus lowest differences for dog were 8% for all baseline data and 9% for all data which meant that within-dog correlation was low.

Descriptive statistics were also summarized for the differences between first and lowest IOP readings, which were also left-skewed and not normally distributed (Table 1). The median (IQR) difference between first and lowest readings over all IOP measurements was 1 mm Hg (0 to 2 mm Hg) indicating an expected negative bias of taking the lowest compared to the first of triplicate measurements. The 2.5th to 97.5th interpercentile range was 0 to 5 mm Hg and the 1st to 99th interpercentile range was 0 to 8 mm Hg. The ICCs of first minus lowest differences for dog were 0% for all baseline data and 6% for all data, which meant that within-dog correlation was low.

Measured IOP significantly increased from the first to the third reading by a mean of 0.3 mm Hg for readings with triplicate averages < 10mm Hg (P = 0.015); significantly decreased by a mean of 0.2 mm Hg for readings with triplicate averages 14 to 20 mm Hg (P = 0.025); and decreased by a mean of 0.4 mm Hg for readings with a triplicate average > 20 mm Hg (P = 0.063).

Means of study group means calculated from averaged triplicate IOP values (n = 160) for each study, treatment, time point, and eye were 0.05 mm Hg less than those based on first IOP readings, with a 95% PI ranging from the first being 1 mm Hg higher to 0.9 mm Hg lower (Supplementary Table S1). These PIs have 95% probability of including a future value of the difference in treatment means and can be used to interpret clinical significance of the range of probable differences. Only 4 differences had absolute values ≥ 1 mm Hg. The mean of study group SDs calculated from averaged IOP values was 3.3 mm Hg, which was 4% less than (95% PI; 27% increase to 35% decrease) the value based on first IOP readings (3.5 mm Hg). The impact of triplicate measurements on group means and accuracy decreases with increased sample sizes. To aid in planning future studies, CIs were calculated on the difference in group means based on first versus averaged triplicate for various sample sizes (Supplementary Table S2). For example, in a study with 10 eyes/group, the 95% CI of the mean difference in group means would be -0.9 to 0.8 mm Hg versus in a study with 60 eyes per group (95% CI; –0.4 to 0.3 mm Hg).

The ICC of all data was 0.82 (Supplementary Table S3), indicating excellent within-triplicate correlation, with 82% of the total variation in IOP values between triplicates (due to dogs, eyes within dogs and time points) and 18% within triplicates. The larger the ICC the lower the increase in precision due to triplicate measurements, and conversely the smaller the ICC the more the increase in precision. The SEs from conditional means in LMMs were compared to assess the impact of analyzing first versus averaged on precision and statistical power, which would increase with SE reduction. A 3% to 9% (0.03 to 0.09 mm Hg) reduction in SEs was observed when estimated from the averaged values versus first readings. Overall, for all studies, the mean SE decreased from 1.06 to 1.00 mm Hg (6%), which was not a clinically meaningful decrease in SE.

For all between-treatment comparisons for all studies and time points (n = 107), the P values from analysis using averaged triplicate values were lower than those using the first IOP reading 46% (49/107) of the time. Which P value was lower was mostly driven by the change in mean differences rather than decrease in variability.

For all between-treatment comparisons for all studies and time points, 95% (102/107) of statistically significant interpretations agreed. For 2% (2/107) of the between-treatment comparisons, the analysis of averaged triplicate IOP values was interpreted as statistically significant (P < 0.05) while the analysis of first IOP measurements was not (Supplementary Table S4). For these 2 comparisons the estimated treatment difference was higher and the SE was lower when using the averaged IOP value. For 3% (3/107) of the between-treatment comparisons, the analysis of first IOP measurements was interpreted as statistically significant (P < 0.05) while the analysis of averaged IOP values was not. For these 3 comparisons the estimated treatment difference was higher when using the first IOP measurement, which caused there to be a significant difference even with the minimally increased SE. Of 11 comparisons found significant with averaged data, 2 (18%) were not found significant with first measurements. Of 96 comparisons not found significant with averaged data, 3 (3%) were found significant with first measurements.

Differences in estimated treatment differences (bias) for each between-treatment comparison were on average only 0.03 mm Hg, indicating no bias in estimated treatment differences in using first compared to averaged triplicate IOP values. For 46% (49/107) of the comparisons the estimated treatment difference was greater using averaged IOP values compared to 54% (58/107) which were greater using first IOP measurements.

Discussion

In the present study, using the first IOP reading did not result in any clinically significant bias nor likely any clinically significant reduction in accuracy compared to using the lowest or averaged triplicate IOP value. The mean difference (bias) between first and averaged value was negligible (–0.08 mm Hg). The first IOP measurement was biased to be higher than the lowest by definition with a median difference of 1 mm Hg. Likewise, the averaged triplicate IOP value was biased to be higher than the lowest with a median difference of 1 mm Hg.

Probable improvement in accuracy from using averaged triplicate over first IOP measurement in clinical practice was quantified with the 2.5th to 97.5th interpercentile range of the differences, which was –3 to 2.7 mm Hg (slightly wider but practically the same as the 95% limits of agreement, which were –2.8 to 2.6 mm Hg (Figure 1; minimum value, –7; maximum value, 5.7; with widest range noted at higher IOPs). For example, if a dog’s first IOP reading is 15 mm Hg, its averaged IOP value would most likely be between 12 and 17.7 mm Hg.

For the difference between the averaged triplicate and the lowest IOP measurement, the 2.5th to 97.5th interpercentile range was 0 to 3.7 mm Hg (minimum value, 0; max value, 9.33; with the widest range noted at higher IOPs). For example, if a dog’s lowest of 3 IOP readings is 15 mm Hg, its averaged IOP value would most likely be between 15 and 18.7 mm Hg. Fortunately for the comparison of these multireading approaches (averaged versus lowest IOP), the difference of values will be apparent at the time of measurement, and if of clinical concern, could prompt additional confirmatory readings to improve accuracy.

Probable improvement in accuracy from using the lowest over the first IOP measurement in clinical practice was quantified with the 2.5th to 97.5th interpercentile range of the differences, which was 0 to 5 mm Hg (minimum value 0; maximum value, 11). For example, if a dog’s lowest of 3 IOP readings is 15 mm Hg, its first IOP reading would most likely be between 15 and 20 mm Hg. Of course, unless multiple readings are taken, it is impossible to know if the reading obtained is the lowest. If multiple readings are taken, using the lowest is reasonable, especially given that the lowest IOP reading is likely the most accurate. This is because it is difficult to falsely lower IOP measurement values but easy for intrinsic and extrinsic factors to falsely elevate them.^1,6,22

Based on these interpercentile ranges, probable differences for an individual dog were all < 5 mm Hg and would be unlikely to impact the diagnosis and management of the vast majority of clinical patients. This said, for a dog being monitored for glaucoma or uveitis, where an even smaller difference could impact how the dog is managed, it could be more important to obtain multiple readings.

In research studies, using the first IOP measurement did not result in any clinically significant bias nor likely a clinically significant difference in accuracy or precision compared to using the averaged triplicate value. The mean difference (bias) between treatment means using first and averaged values was negligible (–0.05 mm Hg). Improvement in accuracy with averaged versus first IOP values, quantified with 95% PIs of the differences in treatment group means (with 5 to 11 eyes each), was small (ranging from first being 1 mm Hg higher to 0.9 mm Hg lower than averaged). Improvement in precision with averaged versus first IOP values was also minimal, with treatment group SDs on average decreased by only 4% (0.2 mm Hg) and conditional mean SEs on average decreased by only 6% (0.06 mm Hg).

As expected, only collecting 1 reading instead of 3 results in some loss of statistical power due to this minimal increase in precision, though less than half (46%) of P values were smaller from the analyses of averaged compared to first IOP values. This is because the change in estimated treatment differences (mean of absolute difference in estimated treatment differences was 0.35 mm Hg) was of a higher magnitude then the mean decrease in SE.

The IOP increased 0.3 mm Hg from the first to the third reading in dogs with a low IOP (< 10 mm Hg) and decreased 0.2 mm Hg in dogs with a normal IOP (14 to 20 mm Hg). The decrease with repeated measurements in dogs with a normal IOP was due to forcing aqueous humor out the iridocorneal angle (aqueous massage effect). This effect is especially prominent in glaucomatous eyes, so using the first measurement may avoid this decrease and underestimation that occurs when using averaged or lowest values.^23,24 The increase in dogs with a low IOP may reflect a regression to the mean, whereby the first IOP reading may have been artificially low. Alternatively, repeated pressure on a globe with low IOP may affect aqueous outflow differently. When a first IOP reading < 10 mm Hg is obtained, taking an average of 3 measurements is likely to produce a slightly higher value, which may or may not be advantageous in clinical decision-making.

Repeated applanation tonometry is more laborious for investigators, especially when a large number of subjects are evaluated over multiple time points. Across the 4 studies included, at least 2,864 IOP measurements could have been avoided if only the first low-error reading was obtained. A significantly increased incidence of iatrogenic adverse effects caused by repeated corneal contact by the tonometer tip, such as superficial corneal irritation or even ulceration, seems unlikely but should be considered.^21,25 For study design, the benefit of increased power versus greater time, potential for adverse effects, and risk of statistical type I error due to possible drift of multiple IOP measurements should be weighed. Power analyses should be used to quantify increased power to aid in this cost-benefit decision. Intratriplicate SDs (measurement errors) and ICC values (for normal dogs) provided in Supplementary Table S3 can be used as inputs for power analyses.

Measuring IOP in triplicate is standard practice in veterinary and human ophthalmic research.^25–29 In clinical practice, however, the first low error reading is often the assigned IOP value for the globe. The supposed superiority of averaged triplicate values versus the first low error reading for research has not been previously evaluated in the veterinary literature. One justification for the practice of obtaining triplicates is the thought that single readings have greater variability than averaged multiple readings and that the resulting increase in accuracy may be clinically significant. In people, Dielemans et al³⁰ documented that intraobserver and interobserver variability with the Goldman applanation tonometer were decreased by 9% and 11%, respectively, using a median of triplicate compared to a single measurement. Kass et al³¹ recommended measuring the IOP in a research setting at least twice to reduce variability and a third time if the difference was > 2 mm Hg. It is commonplace in human pharmaceutical development to average 3 consecutive IOP measurements to presumably increase the reliability of the value.²⁵ However, in review of the results of 1 human ophthalmic pharmaceutical clinical trial by Stewart et al,²⁵ the averaged triplicate IOP value did not differ significantly from the first measurement. Repeated measurements where the previous values are known also introduces the possibility of bias by the tonometer operator.

The present study was limited to ophthalmically normal dogs undergoing applanation tonometry, specifically with a Tono-pen XL by a single operator, so the results of averaged triplicate, lowest, and first IOP approaches for dogs with ophthalmic abnormalities, other species, other tonometer types, and other or multiple operators may not be applicable and should be evaluated by further multi-study reviews.

The value displayed by modern handheld applanation tonometers is an average of multiple measurements (4 to 10 depending on the model) and the CV is also calculated by the microprocessor and shown to help the operator determine when additional measurement(s) may be warranted. The Tono-Pen Avia averages 10 measurements, compared with the Tono-Pen’s 4, which may further reduce the variability associated with single IOP readings and the theoretical need for averaged triplicate values. The results of the present study suggested these tonometer features may be sufficient to control for clinically significant variability associated with single (first) IOP readings and reduce the need for triplicate measurements in veterinary ophthalmic clinical practice and, with sacrifice of some statistical power, research.

Supplementary Materials

Supplementary materials are posted online at the journal website: avmajournals.avma.org