Visual Psychophysics and Physiological Optics

Comparing Methods to Estimate the Human Lens Power

Jos J. Rozema,1,2 David A. Atchison,3 and Marie-José Tassignon1,2

PURPOSE. To compare the accuracy of different methods of calculat- lar refraction, to estimate the power of an equivalent lens at a
ing human lens power when lens thickness is not available. location near that of the lens. Since these biometric parameters
METHODS. Lens power was calculated by four methods. Three meth- are easily determined, such methods can provide a quick esti-
ods were used with previously published biometry and refraction mate of the equivalent lens power.
data of 184 emmetropic and myopic eyes of 184 subjects (age range, The most well known of these methods was proposed by
18–63 years; spherical equivalent range, ⫺12.38 to ⫹0.75 D). These Bennett,6 who used a thick-lens description that makes as-
three methods consist of the Bennett method, which uses lens sumptions about the shape and refractive index distribution of
thickness, a modification of the Stenström method and the Bennett- the lens based on the Gullstrand-Emsley schematic eye.7 From
Rabbetts method, both of which do not require knowledge of lens this, he could calculate the equivalent lens power in a way that
thickness. These methods include c constants, which represent dis- has been shown to be accurate in comparison with phakom-
tances from lens surfaces to principal planes. Lens powers calculated etry.8 However, his method requires knowledge of the lens
with these methods were compared with those calculated using thickness, which is sometimes not available.
phakometry data available for a subgroup of 66 emmetropic eyes (66 Other methods do not require this knowledge of the lens
subjects). thickness, such as the approaches proposed by Stenström9,10
and by Bennett and Rabbetts.11 These approaches might be
RESULTS. Lens powers obtained from the Bennett method corre- useful in a clinical practice using biometry devices that do not
sponded well with those obtained by phakometry for emme- provide lens thickness (e.g., IOL Master; Carl Zeiss Meditec,
tropic eyes, although individual differences up to 3.5 D occurred. Dublin, CA), or in analysis of historical biometry data.
Lens powers obtained from the modified-Stenström and Bennett- The purposes of this study are (1) to verify the agreement
Rabbetts methods deviated significantly from those obtained with that Dunne et al.8 found between the Bennett method and
either the Bennett method or phakometry. Customizing the c phakometry; to (2) compare lens powers obtained with the
constants improved this agreement, but applying these constants Bennett method, our modification of the Stenström method,
to the entire group gave mean lens power differences of and the Bennett-Rabbetts method for previously published data
0.71 ⫾ 0.56 D compared with the Bennett method. By further of emmetropic and myopic eyes, and (3) to provide custom-
optimizing the c constants, the agreement with the Bennett ized constants to optimize the performance of these three
method was within ⫾1 D for 95% of the eyes. methods. The results allow improvement of our statistical eye
CONCLUSIONS. With appropriate constants, the modified Sten- model12 by including a more reliable method to estimate lens
ström and Bennett-Rabbetts methods provide a good approxi- power when lens thickness is not available.
mation of the Bennett lens power in emmetropic and myopic
eyes. (Invest Ophthalmol Vis Sci. 2011;52:7937–7942) DOI:
10.1167/iovs.11-7899 METHODS
Subjects
O cular refraction is determined by axial length, anterior
chamber depth, corneal power ,and lens power. Al-
though axial length and keratometry measurements have be-
To estimate the accuracy of the lens power calculations with respect
to phakometry, we need the biometry and phakometry data of a
population of normal subjects. For this purpose, we used previously
come routine clinically, determining lens power is problem-
published data by Atchison et al.13 for a group of 66 eyes of 66
atic, as the lens radii of curvature and refractive index
emmetropic subjects (32 men, 34 women; 62 Caucasian, 4 non-Cau-
distribution are usually not available. Although techniques
casian). The average spherical equivalent refraction of this group was
have been proposed in the literature to estimate the radii in
⫹0.01 ⫾ 0.38 D (range, – 0.88 to ⫹0.75), and the mean age was
vivo,1–5 they are currently too complicated to be used in
42.4 ⫾ 14.4 years (range 19 – 69).
large-scale studies or clinical practice.
To compare the results of the three power calculation methods for
Because of this impracticality, various methods have been
a wider range of refractions, the first dataset was supplemented by a
proposed that use ocular biometry, such as keratometry, ocular
second set from the same research group.14 This dataset contained 118
axial length, anterior chamber depth, lens thickness, and ocu-
eyes of 118 emmetropic and myopic subjects (43 men, 75 women; 74
Caucasian, 44 non-Caucasian) with a mean spherical equivalent refrac-
tion of ⫺2.69 ⫾ 2.79 D (range, ⫺12.3 to ⫹0.75 D) and an average
From the 1Department of Ophthalmology, Antwerp University subject age of 25.4 ⫾ 5.1 years (range, 18 –36 years). No phakometry
Hospital, Edegem, Belgium; the 2Department of Medicine, Antwerp data were available for this second dataset.
University, Wilrijk, Belgium; and the 3School of Optometry and Insti- Inclusion criteria were stringent, to ensure that only healthy eyes were
tute of Health and Biomedical Innovation, Queensland University of included. These entailed, among others, corrected visual acuity better than
Technology, Brisbane, Queensland, Australia. 6/6 on an ETDRS chart, an intraocular pressure below 21 mm Hg, and a
Submitted for publication May 18, 2011; revised July 29, 2011; Pelli-Robson contrast sensitivity of 1.65 or better for subjects of 40 years of age
accepted August 18, 2011.
and younger and a contrast sensitivity of 1.50 or better for subjects older than
Disclosure: J.J. Rozema, None; D.A. Atchison, None; M.-J.
Tassignon, None 40 years of age. In the myopic dataset, eyes with astigmatism larger than 0.5
Corresponding author: Jos J. Rozema, Department of Ophthalmol- D were also excluded.
ogy, Antwerp University Hospital, Wilrijkstraat 10, 2650 Edegem, Bel- The subjects’ eyes were not dilated or cyclopleged before testing,
gium; [email protected]. which might have caused some degree of accommodation in some of

Investigative Ophthalmology & Visual Science, October 2011, Vol. 52, No. 11
Copyright 2011 The Association for Research in Vision and Ophthalmology, Inc. 7937

Downloaded from iovs.arvojournals.org on 03/27/2024

 7938 Rozema et al. IOVS, October 2011, Vol. 52, No. 11

the younger subjects, resulting in slightly more hyperopic refraction, ⫺ 1000n共Scv ⫹ K兲 1000n
increased lens thickness, and decreased anterior chamber depth. P L, Bennett ⫽ ⫹ (1)
1000n ⫺ 共ACD ⫹ c1 T兲共Scv ⫹ K兲 c2 T ⫹ V
The data collection complied with the Declaration of Helsinki and
received ethical committee approval from the QUT University Human
with n ⫽ 4/3 the aqueous and vitreous index, c1T ⫽ 1000n(n ⫺ nL)T/
Research Ethics Committee and the Prince Charles Hospital Human
(nLPLrLa) the distance between the anterior lens surface and first lenticular
Research Ethics Committee. All subjects gave written informed con-
principal plane, and c2T ⫽ 1000n(n ⫺ nL)T/(nLPLrLa) the distance between
sent before participation.
the posterior lens surface and second lenticular principal plane. The latter is
negative because the principal plane is in front of the back surface. Bennett
Biometry
estimated the c1 and c2 constants using the Gullstrand-Emsley eye model, for
Subjects’ refractions were determined monocularly using Jackson crossed which the lens refractive index nL ⫽ 1.416.
cylinders in a phoropter. Keratometry was performed with a corneal topog-
rapher (E300; Medmont, Vermont, VIC, Australia), and axial length, anterior Modified-Stenström Method: Lens Power
chamber depth, lens thickness, and vitreous depth were measured by A-scan Calculation if Lens Thickness Is Not Known
ultrasonography (Quantel Medical AXIS-II). For the emmetropic group, the
radii of curvature of the anterior and posterior lens surfaces, as well as the lens If lens thickness T is not available, one can estimate the lens power PL
equivalent refractive index, were obtained by analyzing Purkinje images, using Stenström’s method,9,10 which provides the lens power refer-
refraction, and biometry, by using a setup and calculations13 similar to that enced to its anterior vertex rather than to the principal planes. We
described by Rosales and Marcos.3 Note that phakometry data were not modified the method by including the parameter cSten, which is the
available for the second dataset. estimated distance between the anterior lens surface and the first
lenticular principal plane. The modified-Stenström method is given by:
Bennett’s Method: Lens Power Calculation Using
Known Lens Thickness 1000n共Peye ⫺ K兲
P L, Sten ⫽ (2)
Bennett’s method6 calculates lens power PL when lens thickness T is 1000n ⫺ K共ACD ⫹ cSten兲
available by keeping the distances from the surfaces to the principal
planes of the lens in the same proportion as in the lens of the Gullstrand- using the parameters in Table 1, with n ⫽ 1.336. This equation
Emsley eye model.7 Using the parameters defined in Table 1, contains the equivalent power of the eye Peye. Based on Stenström’s
the steps in his method can be combined as the single equation: derivation, we calculated this as:

1
P eye ⫽ 关1000n ⫺ 共ACD ⫹ cSten兲K ⫺ SPP共L ⫺ ACD ⫺ cSten兲
2共L ⫺ ACD ⫺ cSten兲

⫹ 冑共1000n ⫺ 共ACD ⫹ cSten兲K ⫺ SPP共L ⫺ ACD ⫺ cSten兲兲2 ⫺ 4共ACD ⫹ cSten兲K SPP共L ⫺ ACD ⫺ cSten兲兴 (3)

Here the ocular refraction at the first principal plane of the eye SPP is A simplification of equation 3 was proposed by van Alphen,15
used. Lens power PL can be found by substituting the value for Peye who used the approximation Peye ⫽ 1392/L ⫺ SPP. However, since
derived from equation 3 into the right side of equation 2. this simplification deviates considerably from values obtained from

TABLE 1. Overview of the Parameters Used

Parameter Unit Calculation Description

S D Spherical refraction at spectacle back vertex plane

SCV D S/(1 ⫺ 0.014 S) Spherical refraction at corneal vertex
SPP D S/(1 ⫺ 0.0155 S) Spherical refraction at first principal plane of the eye
K D Corneal power
ACD mm Anterior chamber depth (corneal epithelium to anterior lens)
T mm Lens thickness
L mm Axial length
V mm L-ACD-T Vitreous depth
PL D Lens power
n — Refractive index of aqueous and vitreous humors
nL — Refractive index of crystalline lens
PL,Bennett D Equation 1 Lens power using Bennett method
rLa mm Anterior radius of curvature of lens
rLp mm Posterior radius of curvature of lens
PLa D (nL ⫺ n)/rLa Power of anterior lens surface
PLp D (n ⫺ nL)/rLp Power of posterior lens surface
c1T mm 1000 n(n ⫺ nL)T/(nLPLrLp) Distance between anterior lens surface and first principal plane of lens
c2 T mm 1000 n(n ⫺ nL)T/(nLPLrLa) Distance between posterior lens surface and second principal plane of lens
PL,Sten D Equation 2 Lens power using modified-Stenström method
Peye D Equation 3 Equivalent power of combination of eye and a thin correcting lens placed
at the cornea
cSten mm Equation 2 ⫹ 3 solved for cSten Distance between anterior lens surface and first principal plane of lens
PL,BR D Equation 4 Lens power using Bennett-Rabbetts method
cBR mm Equation 4 solved for cBR Distance between thin lens position and anterior lens surface

Downloaded from iovs.arvojournals.org on 03/27/2024

 IOVS, October 2011, Vol. 52, No. 11 Estimating Lens Power 7939

equation 3 for cSten ⬎ 0 mm, we did not include it in our analy- used instead, to estimate values numerically. Means and standard de-
sis. viations of these optimal c constants were called the customized c
constants and are given in Table 2.
Bennett-Rabbetts Method: Lens Power Calculation
if Lens Thickness Is Not Known Statistics
Another approach to calculating PL without knowing T is to modify an The significance level was P ⬍ 0.05 (analyses by Excel; Microsoft; SPSS
equation proposed by Bennett and Rabbetts11 for the purpose of ver. 12; Chicago, IL).
calculating the spherical refraction of an eye when its biometry is
known. They replaced the lens with an equivalent thin lens located at
the midpoint between the lenticular principal planes, using the Ben- RESULTS
nett-Rabbetts eye model.11 If the ocular refraction at the corneal vertex
Agreement between Calculated and Phakometry
SCV is known, their equation can be rewritten to give PL:
Derived Lens Powers for Emmetropes
L共Scv ⫹ K兲 ⫺ 1000n The mean lens power determined with phakometry was PL ⫽

冉 冊
P L, BR ⫽ (4) 22.87 ⫾ 2.42 D, which may be considered the target value that
ACD ⫹ cBR
共L ⫺ ACD ⫺ cBR兲 共Scv ⫹ K兲 ⫺ 1 the calculation methods must approximate (Table 2). Using
1000n
both the Gullstrand-Emsley and Bennett-Rabbetts eye models,
with n ⫽ 1.336 and cBR the distance between the anterior lens surface
the lens powers with the Bennett method were not signifi-
and the thin lens position. This parameter can be found by solving
cantly different from the phakometry powers. Using the cus-
equation 4 for cBR when PL is known.
tomized c constants did not improve the agreement. A Bland-
Altman plot shows that the differences between Bennett and
Phakometry phakometry lens power remained between ⫾3 D (Fig. 1a) and
for 45% and 50% of the eyes were less than ⫾1 D (Table 2).
Using the lens surface radii of curvature and lens refractive index These differences did not correlate with subject age (Pear-
determined using phakometry, along with the lens thickness, the son ⬍ 0.01, P ⬎ 0.05), which excludes accommodation as a
lens equivalent power was calculated using the thick lens formula16: possible source of these differences.
Using the Gullstrand-Emsley and Bennett-Rabbetts eye mod-
T els, the modified-Stenström and Bennett-Rabbetts methods
P L ⫽ PLa ⫹ PLp ⫺ P P (5)
1000nL La Lp gave lens powers that were approximately 1.5 D lower and
were significantly different from phakometry lens powers
with PLa and PLp as defined in Table 1. (paired t-tests, P ⬍ 0.01). By customizing the c constants, the
differences with phakometry reduced remarkably to nonsignifi-
Comparing Lens Powers with the cance (paired t tests, P ⬎ 0.05), and for ⬃40% of the eyes the
Different Methods differences were less than ⫾1 D (Table 2).
The Pearson correlation coefficients between the calculated
To compare lens powers obtained with the methods detailed above,
and phakometry lens powers were high (Table 2) and indepen-
we determined the c constants c1, c2, cSten, and cBR for both Gullstrand-
dent of the eye model used. The correlation coefficients were
Emsley and Bennett-Rabbetts eye models. As both eye models will
higher for the Bennett method than for the modified-Stenström
differ from actual ocular biometry, we determined the optimal c con-
and the Bennett-Rabbetts methods.
stants also for each eye individually. For the Bennett method, these
constants were easily determined by filling in the available phakometry
Comparison of the Three Lens Power Calculation
of the emmetropic dataset into the formulas for c1 and c2 in Table 1,
using n ⫽ 1.336. The optimal c constants of the modified-Stenström
Methods for the Whole Population
and Bennett-Rabbetts methods were found by using the phakometry Figure 2a shows lens powers calculated using the Bennett
lens powers of the emmetropic dataset for PL and solving equations 2 method with customized c constants calculated for the combi-
and 3 and equation 4 for the c constants, also using n ⫽ 1.336. The nation of the two datasets (184 eyes) as a function of axial
analytical solution for cSten in the modified-Stenström method was length L. The lens power has a negative correlation with axial
mathematically complicated and could not be used in Excel (Microsoft, length for L ⬍ 24 mm (r ⫽ – 0.624; P ⬍ 0.001), with a slope
Redmond, WA); Mathematica (Wolfram Research, Champaign, IL) was that matches that of the measured lens power data. Above L ⫽

TABLE 2. Comparison of the Measured and Calculated Lens Powers Using the Biometry and Phakometry of the Emmetropic Data

Pearson Correlation
Within ⴞ1 D Coefficients with
Method Symbol Eye Model c Constants Average from PL (%) Phakometry

Phakometry PL 22.87 ⫾ 2.42 D

Bennett PL,Bennett Gullstrand-Emsley c1 ⫽ 0.596; c2 ⫽ ⫺0.358 22.50 ⫾ 2.02 D 45.5 0.778 (P ⬍ 0.001)
Bennett-Rabbetts c1 ⫽ 0.599; c2 ⫽ ⫺0.353 22.74 ⫾ 2.03 D 50.0 0.779 (P ⬍ 0.001)
Customized c1 ⫽ 0.571 ⫾ 0.028 22.54 ⫾ 2.00 D 45.5 0.778 (P ⬍ 0.001)
c2 ⫽ ⫺0.378 ⫾ 0.029
Modified Stenström PL,Sten Gullstrand-Emsley cSten ⫽ 2.145 mm 21.04 ⫾ 1.94 D 19.7 0.720 (P ⬍ 0.001)
Bennett-Rabbetts cSten ⫽ 2.221 mm 21.36 ⫾ 1.97 D 27.3 0.720 (P ⬍ 0.001)
Customized cSten ⫽ 2.875 ⫾ 0.763 mm 22.78 ⫾ 2.12 D 42.4 0.721 (P ⬍ 0.001)
Bennett-Rabbetts PL,BR Gullstrand-Emsley cBR ⫽ 2.230 mm 21.21 ⫾ 1.96 D 24.2 0.720 (P ⬍ 0.001)
Bennett-Rabbetts cBR ⫽ 2.306 mm 21.54 ⫾ 1.99 D 36.4 0.720 (P ⬍ 0.001)
Customized cBR ⫽ 2.891 ⫾ 0.778 mm 22.81 ⫾ 2.13 D 40.9 0.721 (P ⬍ 0.001)

n ⫽ 66 eyes.

Downloaded from iovs.arvojournals.org on 03/27/2024

 7940 Rozema et al. IOVS, October 2011, Vol. 52, No. 11

FIGURE 1. Bland-Altman plots showing the differences between the phakometry lens powers and the lens powers calculated using (a) the Bennett
method and (b) the Bennett-Rabbetts method. Powers for the calculation methods are shown for both eye model c constants and customized c
constants. As the modified-Stenström and the Bennett-Rabbetts methods gave similar lens powers, results are shown only for the latter.

24 mm, approximately corresponding with the onset of myo- To improve the matches of the modified-Stenström and
pia, the lens power plateaus to become constant (r ⫽ – 0.036; Bennett-Rabbetts methods with the Bennett method, a sec-
P ⬎ 0.05). Because phakometry was not available for the ond c constant (customized 2) was determined for the mod-
second dataset, this plateauing could not be confirmed exper- ified-Stenström and Bennett-Rabbetts methods that mini-
imentally. However, a similar trend was found in the raw data mized the mean lens power difference with the Bennett
published by Sorsby et al.17 Thus, in the absence of phakom- method over the entire population. When these customized
etry data for the entire dataset, the Bennett power with cus- 2 constants were used, the lens power differences with the
tomized c constants was used as a benchmark. This choice is Bennett method were no longer statistically significant (P ⬎
based on the observation of Dunne et al.8 that the Bennett 0.05) and were within ⫾1 D for about 95% of eyes (Table 3).
power corresponds well with phakometry in myopic refrac- For both methods, the power differences with the Bennett
tions up to –9.37 D, including the long eyes for which the method were correlated significantly with axial length L
plateauing is shown in Figure 2a. (r ⫽ 0.390, P ⬍ 0.001 and r ⫽ 0.329, P ⬍ 0.001 for the
The mean powers with the modified-Stenström and Ben- modified-Stenström and the Bennett-Rabbetts methods, re-
nett-Rabbetts methods, using the Gullstrand-Emsley and Ben- spectively; Fig. 2b).
nett-Rabbetts eye models, were 0.5 to 1.0 D less than the
mean powers obtained with the Bennett method and its
customized c constants (Table 2). These differences were DISCUSSION
statistically significant (paired t-test, P ⬍ 0.001). Using the
customized c constants, the modified-Stenström and Ben- Using the customized c constants all three lens power calcula-
nett-Rabbetts methods each yielded lens powers that were tion methods are in reasonable agreement with the phakom-
0.71 ⫾ 0.56 D greater than those with the Bennett method etry lens power for emmetropic eyes. This answers the first
(Table 3), and this difference was also statistically significant purpose of this article, which was to confirm the agreement
(P ⬍ 0.001). that Dunne et al.8 found between the Bennett method and

FIGURE 2. (a) Phakometry lens power and powers calculated using the Bennett method with customized c constants, plotted as a function of axial length
L. (b) Difference between the lens power values calculated using the modified-Stenström or Bennett-Rabbetts methods with the customized 2
constants and the Bennett method with customized c constants, plotted as a function of axial length L. The calculated powers use data from both
datasets (184 eyes), whereas the phakometry powers contain data only from the first dataset (66 eyes).

Downloaded from iovs.arvojournals.org on 03/27/2024

 IOVS, October 2011, Vol. 52, No. 11 Estimating Lens Power 7941

TABLE 3. Comparison of the Measured and Calculated Lens Powers Using the Biometry of Both the Emmetropic and Myopic Datasets

Within ⴞ1 D Pearson Correlation

from Coefficient with
Method Symbol Eye Model c Constants Average PL,Bennett (%) Phakometry

Bennett PL,Bennett Customized c1 ⫽ 0.571 ⫾ 0.028 22.31 ⫾ 1.72 D

c2 ⫽ ⫺0.378 ⫾ 0.029
Modified Stenström PL,Sten Gullstrand-Emsley cSten ⫽ 2.145 mm 21.30 ⫾ 1.61 D 61.4 0.942 (P ⬍ 0.001)
Bennett-Rabbetts cSten ⫽ 2.221 mm 21.62 ⫾ 1.63 D 71.7 0.943 (P ⬍ 0.001)
Customized* cSten ⫽ 2.875 ⫾ 0.763 mm 23.01 ⫾ 1.76 D 64.7 0.947 (P ⬍ 0.001)
Customized 2 cSten ⫽ 2.550 mm 22.30 ⫾ 1.69 D 95.1 0.945 (P ⬍ 0.001)
Bennett-Rabbetts PL,BR Gullstrand-Emsley cBR ⫽ 2.230 mm 21.45 ⫾ 1.62 D 67.9 0.946 (P ⬍ 0.001)
Bennett-Rabbetts cBR ⫽ 2.306 mm 21.77 ⫾ 1.64 D 78.8 0.947 (P ⬍ 0.001)
Customized* cBR ⫽ 2.891 ⫾ 0.778 mm 23.02 ⫾ 1.76 D 66.8 0.950 (P ⬍ 0.001)
Customized 2 cBR ⫽ 2.564 mm 22.31 ⫾ 1.69 D 95.1 0.948 (P ⬍ 0.001)

n ⫽ 184 eyes.
* Customized c constants of Table 2.

phakometry. However, for individual eyes, differences be- lens power differences to within ⫾1 D for ⬃95% of the eyes.
tween calculated and phakometric power of up to 3.5 D If lens thickness is not available, both methods with the cus-
occurred (Fig. 1, Table 2), which is considerably larger than tomized 2 constants may be considered to be good approxi-
the differences of up to 0.77 D that Dunne et al. reported for mations of the Bennett method.
the Bennett method. These differences could result from bio- Although the three calculation methods now match well
metric errors and Bennett’s assumption that the lens shapes of with one another for a wide range of refractions, there are still
the eye models are representative for all eyes (the ratio PLp/PLa theoretical issues to consider. The first is that the modified-
of posterior to anterior lens powers was 1.52 ⫾ 0.19 for pha- Stenström and Bennett methods produce the same results
kometry, but 1.67 and 1.70 for the Gullstrand-Emsley and when both lenticular principal planes coincide (i.e., c1T ⫽ cSten
Bennett-Rabbetts eye models, respectively). Using the argu- and c2T ⫽ T ⫺ cSten). This can be confirmed mathematically by
ment of Bennett6 and Dunne et al.8 that lens power provided comparing equation 1 with equations 2 and 3 for the special
by the Bennett method is likely to be more accurate than
case when Spp ⫽ SCV ⫽ 0. The more general case, when Spp
phacometry, because of the inherent difficulties in performing
and SCV are different from 0, could only be confirmed numer-
the latter accurately, we considered that the Bennett method
derived power as a reasonable approximation of the real equiv- ically because of the mathematically complicated equation 3.
alent lens power and used it as a reference to compare the Although this seems to point at some common origin of both
modified-Stenström and Bennett-Rabbetts methods. formulas, the meaning of this observation remains unclear.
For the Bennett method, the choice of eye model did not A second relationship was found between the modified-
influence the calculated lens power significantly, which may Stenström and Bennett-Rabbetts methods, which, despite be-
be a consequence of the fact that the method is based on ray ing mathematically very different, produced very similar lens
tracing of a thick lens model rather than a thin lens approxi- powers. Again, a possible relationship between both methods
mation. It can be used accurately for emmetropic eyes with could not be investigated further due to the complexity of
either the c1 and c2 constants of the two eye models or the equation 3.
customized constants derived in this work. As lens power depends on lens refractive index, one could
The second purpose of the paper was to compare lens expect a correlation between the c constants and lens refrac-
powers obtained with the Bennett method, the modified-Sten- tive index values nL determined from phakometry. For this
ström method, and the Bennett-Rabbetts method for emme- reason, the results of the lens power calculations were given
tropic and myopic eyes. The modified-Stenström and Bennett- for each eye model separately. However, a significant correla-
Rabbetts methods gave lens powers that were significantly tion with nL was seen only for c1 of the Bennett method; the
lower than those given by phakometry (mean, 1.6 D) and the other c constants were either constant or randomly distributed.
Bennett method (mean, 1.3D) for emmetropic eyes and for the Finally, we would like to point out that one could also use
Bennett method in combined emmetropic and myopic eyes IOL calculation formulas, such as the Hoffer Q18 or the SRK/T
(mean, 0.8 D). formula19,20 to calculate the lens power, provided appropriate
The third purpose of the paper was to provide customized values for the IOL constants are used. Here, one has to deal
constants to optimize the performances of the Bennett, modi- with the added difficulty of estimating the final postoperative
fied-Stenström, and Bennett-Rabbetts methods. For the Bennett
position of the lens,21,22 which may explain the large variety in
method using customized c constants made little difference in
IOL calculation formulas in the literature.
the results, but in emmetropic eyes the customized c of 2.875
and 2.891 mm for modified-Stenström and Bennett Rabbetts In conclusion, if lens thickness is known the equivalent lens
methods, respectively, gave nonsignificant lens power differ- power is best calculated using the Bennett method with either
ences with phakometry and produced more accurate results the published or the customized c constants. The modified-
than the constants of the eye models. When comparing lens Stenström and Bennett-Rabbetts methods, with appropriate c
powers for combined emmetropic and myopic eyes, the cus- constants, provide reasonable approximations of equivalent
tomized c constants for emmetropic eyes produced systematic lens power when lens thickness is not known. These methods
lens power differences between the Bennett method and the allow the application of the concept of our statistical eye
modified-Stenström and Bennett-Rabbetts methods. This was model12 to datasets without lens thickness or can be included
improved by new customized 2 constants for the latter two in the software of a biometry device alongside IOL calculation
methods (c ⫽ 2.550 and 2.564 mm for modified-Stenström and formulas, thus providing physicians with access to the impor-
Bennett Rabbetts methods, respectively), which brought the tant parameter of lens power.

Downloaded from iovs.arvojournals.org on 03/27/2024

 7942 Rozema et al. IOVS, October 2011, Vol. 52, No. 11

Acknowledgments 11. Rabbetts RB. Bennett and Rabbetts’ Clinical Visual Optics. 4th
ed. London: Butterworth-Heinemann: 2007:223–227,412–
The authors thank Kristien Wouters for statistical advice and Rafael 413,425– 426.
Irribaren and Wolfgang Haigis for discussions on the topic of this work. 12. Rozema JJ, Atchison DA, Tassignon MJ. Statistical eye model for
normal eyes. Invest Ophthalmol Vis Sci. 2011;52:4525– 4533.
References 13. Atchison DA, Markwell EL, Kasthurirangan S, Pope JM, Smith G,
Swann PG. Age-related changes in optical and biometric charac-
1. Dunne MCM. Scheme for the calculation of ocular components in
teristics of emmetropic eyes. J Vis. 2008;8:29.
a 4-surfaced eye without the need for measurement of the anterior
crystalline lens surface Purkinje images. Ophthalmic Physiol Opt. 14. Atchison DA. Optical models for human myopic eyes. Vision Res.
1992;12:370 –375. 2006;46:2236 –2250.
2. Garner LF. Calculation of the radii of curvature of the crystalline 15. Van Alphen, GWHM. On emmetropia and ametropia. Ophthalmo-
lens surfaces. Ophthalmic Physiol Opt. 1997;17:75– 80. logica. 1961;(suppl):1–92.
3. Rosales P, Wendt M, Marcos S, Glasser A. Changes in crystalline 16. Pedrotti LS, Pedrotti FL. Optics and Vision. Upper Saddle River, NJ:
lens radii of curvature and lens tilt and decentration during dy- Prentice Hall; 1998:77.
namic accommodation in rhesus monkeys. J Vis. 2008;8:1–12. 17. Sorsby A, Benjamin B, Davey JB, Sheridan M, Tanner JM. Emmetro-
4. Dubbelman M, Van der Heijde GL. The shape of the aging human pia and its aberrations; a study in the correlation of the optical
lens: curvature, equivalent refractive index and the lens paradox. components of the eye. Spec Rep Ser Med Res Counc (GB).
Vision Res. 2001;41:1867–1877. 1957;11(293):1– 69.
5. Rosales P, Dubbelman M, Marcos S, van der Heijde R. Crystalline 18. Hoffer KJ. The Hoffer Q formula: a comparison of theoretic and
lens radii of curvature from Purkinje and Scheimpflug imaging. J regression formulas. J Cataract Refract Surg. 1993;19:700 –712.
Vis. 2006;6:1057–1067. (Zuberbuhler B, Morrell AJ. Errata in printed Hoffer Q formula. J
6. Bennett AG. A method for determining the equivalent lens powers Cataract Refract Surg. 2007;33:2; author reply: 2–3).
of the eye and its crystalline lens without resorting to phakometry.
19. Retzlaff JA, Sanders DR, Kraff MC. Development of the SRK/T
Ophthalmic Physiol Opt. 1988;8:53–59.
intraocular lens implant power calculation formula. J Cataract
7. Emsley HH. Visual Optics. Vol. 1. 5th ed. London: Butterworth;
Refract Surg. 1990;16:333–340.
1952:40 – 42,344,360 –361.
20. Sheard RM, Smith GT, Cooke DL. Improving the prediction accu-
8. Dunne MCM, Barnes DA, Royston JM. An evaluation of Bennett’s
method for the determining the equivalent lens powers of the eye racy of the SRK/T formula: the T2 formula. J Cataract Refract
and its crystalline lens without resorting to phakometry. Ophthal- Surg. 2010;36:1829 –1834.
mic Physiol Opt. 1989;9:69 –71. 21. Gobin L, Rozema JJ, Tassignon M-J. Review of IOL power
9. Stenström S. Untersuchungen über die Variation und Kovaria- calculation: a theoretical analysis of proposed formulas. In: Garg A,
tion der optischen Elemente des menschlichen Auges. PhD thesis. Lin JT, Lathkany R, Bovet J, Halgis W, eds. Mastering the Tech-
Uppsala, Sweden: University of Uppsala, 1946. niques of IOL Power Calculations. 2nd ed. New Delhi, India;
10. Stenstrom S. Investigation of the variation and the correlation of Jaypee 2009:Chapter 54.
the optical elements of human eyes. Part II, Translated by Woolf 22. Olsen T. Calculation of intraocular lens power: a review. Acta
D. Am J Optom Arch Am Acad Optom. 1948;25:286 –299. Ophthalmol. Scand. 2007;85:472– 485.

Downloaded from iovs.arvojournals.org on 03/27/2024