**Purpose.**:
To examine the interrelationships among the Ocular Response Analyzer (ORA; Reichert Ophthalmic Instruments, Buffalo, NY), Goldmann applanation tonometer (GAT), and corneal geometry measurements in a young, healthy sample.

**Methods.**:
Central corneal radius, ORA, GAT, and central corneal thickness (CCT) measurements were taken in 99 subjects (age, 21 ± 2 years) who were free of ocular and systemic disease.

**Results.**:
The mean ± SD corneal hysteresis (CH) and corneal resistance factor (CRF) were 10.4 ± 1.2 and 10.1 ± 1.5 mm Hg, respectively. The Bland-Altman 95% limits of agreement of ORA Goldmann-correlated IOP (ORAg) and ORA corneal-compensated (ORAcc) IOP with reference to GAT were −4.5 to +6.0 and −4.1 to +6.8 mm Hg, respectively. The full equations used by the ORA to calculate ORAcc and CRF were reconstructed. The statistically significant effect of CCT on GAT became redundant if CRF was included in a multivariate regression analysis. Both CH and CRF were associated with CCT (*R* ^{2} = 0.252 and 0.290, respectively).

**Conclusions.**:
Sample CH and CRF were consistent with those reported in the literature. ORAg and ORAcc agreed poorly with GAT. CRF appears to be at least a partial description of corneal rigidity. The ocular determinants of CH are unclear.

^{ 1 }However, its accuracy is affected by the interindividual variation in corneal geometric properties, such as thickness

^{ 2 –5 }and curvature.

^{ 6,7 }Corneal biomechanical properties such as Young's modulus are also believed to influence GAT readings,

^{ 8 –12 }but the traditional methods used to determine these properties (strip extensiometry and membrane inflation) cannot be performed in vivo.

^{ 13,14 }An air pulse of increasing force lasting approximately 20 ms is directed onto the eye, causing progressive corneal deformation through a first, inward applanation state (P1) to indentation. A second, outward applanation (P2) is achieved as the cornea returns to its original shape. Infrared light is used to determine the point of corneal flattening by using the cornea as a mirror, as previously described.

^{ 15 }The air pulse force is noted at P1 and P2 for use in calculating four parameters, as discussed below.

^{ 13,14 }The former is intended to quantify the viscoelastic mechanical damping ability of the cornea, whereas the latter is thought to describe its overall viscoelastic resistance.

^{ 14,16 }Neither of these variables can be considered corneal properties, because they are responses that are specific to the ORA measurement process. In contrast, true properties such as thickness and Young's modulus are ideally invariant to the measurement technique.

^{ 13,16 }However, the definitions and validity of CH, CRF, and ORAcc have not been convincingly demonstrated. The goals of this study were to collect normative ORA data in young normal subjects and to evaluate the descriptions of the ORA parameters based on their intercorrelations and associations with other clinically measured variables.

^{ 17 –19 }

^{ 20 }where each ORAg value was to be within 2.4 millimeters of mercury (mm Hg) of their mean. This latter figure was based on an unpublished pilot study performed by the authors. However, the mean of the best three readings was selected for analysis, to avoid inclusion of borderline waveforms, as the evaluation of measurement quality is subjective. Each of the three readings used was to have its ORAg within 2.3 mm Hg of the mean (also based on pilot study data). Additional readings were taken to meet these requirements, if necessary. The cornea was then anesthetized and the tear film stained with 1 drop of 0.5% proxymetacaine hydrochloride (Alcaine; Alcon Laboratories, Frenchs Forest, NSW, Australia) and 2% sodium fluorescein (Minims; Bausch & Lomb, North Ryde, NSW, Australia) respectively, before the IOP was measured (Goldmann AT 900; Haag-Streit, Bern, Switzerland). The average of three readings, within ±2 mm Hg of their mean, was used for analysis. Last, the CCT was measured by ultrasonic pachymeter (Pocket II Precision Pachymeter; Quantel Medical, Clermont-Ferrand, France). Three consecutive readings within ±5 μm of their average were taken, and the mean analyzed.

^{ 21 }An assistant recorded GAT results and turned the measurement drum away from the recorded value after each attempt.

^{ 22 }was used to calculate the required sample size, which was 82.

^{ 23 }

^{ 24 }Corneal geometry has a more established role in tonometry research, and thus was entered as the first block for these regressions. Where CH and CRF were potential predictors, they were assessed in separate models. Because all ORA parameters are calculated linearly from two basic applanations (equations 1 –4), their simultaneous inclusion would cause misestimation of their coefficients and their multivariate coefficient of determination (

*R*

^{2}) to erroneously reflect the explanatory effect of P1 and P2.

Parameter (Units) | Mean (SD) |
---|---|

Intraocular pressure | |

GAT, mm Hg | 13.9 (2.9) |

ORAg, mm Hg | 14.7 (3.2) |

ORAcc, mm Hg | 15.3 (2.9) |

Corneal geometry | |

CCR, mm | 7.80 (0.24) |

CCT, μm | 546.0 (29.9) |

Corneal biomechanics | |

CH, mm Hg | 10.4 (1.2) |

CRF, mm Hg | 10.1 (1.5) |

Basic ORA applanations | |

P1, mm Hg | 19.9 (3.3) |

P2, mm Hg | 9.5 (3.1) |

*r*, both

*P*> 0.05).

*r*= 0.931,

*P*< 0.001). With the current data, each ORA parameter could be virtually entirely predicted by P1 and P2, by using linear regression (

*R*

^{2}≥ 0.999). These equations were algebraically rearranged to the form presented by Reichert, and are shown in equations 1 to 4:

Predictor | GAT | ORAg | ORAcc |
---|---|---|---|

CCR, mm | −2.39 [1.19] | 0.61 [1.27] | 1.30 [1.20] |

CCT, 10 μm | 0.19 [0.10] | 0.33 [0.10] | 0.07 [0.10] |

Constant | 22.2 [10.0] | −8.1 [10.8] | 1.6 [10.2] |

R ^{2} | 0.068 | 0.104 | NA |

*P*= 0.868 and −0.05 [0.30] mm Hg ORAg/mm Hg CH,

*P*= 0.879). For ORAcc, CH was a significant predictor and also caused CCT to gain significance (−1.11 [0.26] mm Hg ORAcc/mm Hg CH,

*P*< 0.001, and 0.29 [0.10] mm Hg ORAcc/10 μm CCT,

*P*= 0.006, respectively). The final model

*R*

^{2}was 0.185.

*P*< 0.001). Where applicable, its inclusion also caused CCR and CCT to lose their significance in each model (all

*P*> 0.05). The respective

*R*

^{2}values were 0.266, 0.552, and 0.152.

^{ 25 –30 }These citations also report the SD for CH and CRF to lie between 1.3 and 2.0, and 1.5 and 2.0 mm Hg, respectively.

^{ 25 –30 }Only the seminal paper was an exception, which reported a mean CH of 9.6 mm Hg.

^{ 14 }Slight modifications to the ORA before its public release may account for this discrepancy.

^{ 31 –35 }although some studies find no mean difference between GAT and both ORA IOP measurements.

^{ 6,29,36 }

^{ 6,29,32 }and glaucoma subjects

^{ 33,35,36 }also found the 95% limits of agreement between the ORA IOP estimates and GAT to span a range of approximately 10 mm Hg or greater. A much lower 95% range of 6.6 mm Hg was observed in healthy subjects by Oncel et al.

^{ 34 }; however, the reasons for these narrower limits are not evident.

^{ 14 }The equations determining ORAcc and CRF are also linear; however, calibration coefficients that have not been reported are used in their calculation.

^{ 13 }These values were determined and presented in equations 3 and 4.

*R*

^{2}for these equations indicates that each determinant of the calculated ORA parameters, such as IOP, corneal geometry, corneal biomechanical behavior, and possibly other unknown factors must be represented through P1, P2, and constants alone. Luce

^{ 14 }claimed that two applanations are necessary to separate the IOP and corneal effects. However, this claim is questionable, given that the information contained within P1 and P2 is extremely similar (

*R*

^{2}= 0.867,

*P*< 0.001). In fact, the suitability of the NCT process for determining corneal biomechanics and the true IOP has not been demonstrated with reference to traditional biomechanical testing and manometry. Studies inducing pharmacologic changes in IOP

^{ 37 }and using finite element analysis

^{ 38 }suggest that the calculation of ORAcc, CH, and CRF could at least be improved.

^{ 13 }

^{ 2,39 }The accepted interpretation of the positive association between applanation tonometry and CCT measurements is that an increased corneal thickness provides greater resistance to applanation tonometry, causing overestimation and vice versa.

^{ 2 –5 }The current results indicate that a 0.19 [0.10] mm Hg GAT overestimation would be expected if CCT was thicker than average by 10 μm, which is toward the lower end of the range of 0.13 to 0.34 mm Hg/10 μm CCT reported in recent work.

^{ 2,40 –43 }

^{ 44,45 }The cornea may behave as a comparatively more rigid structure during the NCT process because of its viscoelastic properties; the resistance of a viscoelastic material to deformation is greater if the stress is applied at a faster rate.

^{ 44 }The applanation process is performed over the order of milliseconds during NCT/ORA,

^{ 14,15 }in comparison to seconds for GAT. Another possible explanation is that overall corneal resistance to NCT is greater due to its larger nominal applanation area.

^{ 15,46 }In contrast, ORAcc and CCT were unrelated, as previously observed.

^{ 6,29,32 }This finding is consistent with the claim that ORAcc is an IOP estimate with reduced corneal dependence.

^{ 13,16 }

^{ 32,47,48 }and ORAg

^{ 28,32,49 }agrees with work done on normal eyes. These findings suggest that under physiological conditions, CH is independent to the IOP and does not describe a cornea-related GAT error. To our knowledge, the effects of corneal hysteretic behavior or viscoelasticity on GAT accuracy have not been studied. It is thus difficult to infer whether this null result supports or refutes the validity of CH.

^{ 6,32,47 }and ORAg

^{ 28,32,49 }also appear in the literature. These co-variations support the description of CRF as a response related to corneal resistance, since its inclusion in the model caused CCT to lose its statistical significance when GAT or ORAg was the outcome variable. The effect of CCT causing tonometric overestimation is therefore encompassed by CRF.

*R*

^{2}= 0.266) was much higher than the combined influence of CCR and CCT (

*R*

^{2}= 0.068). Two potentially inclusive interpretations are proposed. First, because corneal biomechanics affect the accuracy of GAT,

^{ 8 –12 }CRF may have a stronger relationship with GAT because it is an overall composite response related to both corneal biomechanics and geometry. Second, the implication that at least 26.6% of the variation in GAT occurs because of corneal errors appears fallaciously high. Since it is highly improbable that the equation derived to calculate CRF is entirely specific in separating the corneal and IOP contributions to P1 and P2, CRF is likely to be IOP dependent and hence only partially valid. This theory is supported by the much stronger relationship between ORAg and CRF. It is inconceivable that corneal interference could be responsible for over half of the variation in NCT measurements (

*R*

^{2}= 0.552). A portion of this shared variance probably occurs due to the IOP affecting both parameters.

^{ 28 –30,50 }The findings in Table 3 are thus consistent with those in previous work. However, other factors such as corneal biomechanics and possibly IOP may contribute to CH and/or CRF given that most of their variation was not explained by CCT (

*R*

^{2}= 0.252 and 0.290, respectively).

^{ 2 –5 }CRF was thus developed to correlate strongly with CCT,

^{ 13 }as also observed in the present study. However, it is still unclear how strongly CRF is related to corneal rigidity.

^{ 20 }Quantitative analysis of the waveform may also be of interest and is now possible with version 2.04 of the ORA software. The waveform morphology can potentially describe the corneal and IOP response to the air pulse more completely

^{ 14 }; however, the relationship of the varying aspects of the ORA signal to their underlying ocular determinants is unknown. Conversely, it is more ideal to quantify ORA results in terms of their waveform morphology rather than the calculated ORA parameters, because the latter makes unvalidated assumptions on the underlying quantity and quantities being measured.

*Ophthalmologica*1957;134:221–242. In: Ritch R Caronia RM eds. Classic Papers in Glaucoma. The Hague, The Netherlands: Kugler Publications; 2000;155–162.