Key words IOL power calculation - SMILE - myopia - cataract - refractive surgery
Schlüsselwörter Katarakt - refraktive Chirurgie - IOL-Stärkenberechnung
Introduction
With the increasing number of patients with ametropia, more people choose corneal
refractive surgery. Small incision lenticule extraction (SMILE) has been one of the most
popular refractive correction surgeries all over the world. It is expected to offer better
biomechanical stability than procedures that involve flap creation, such as laser in
situ keratomileusis (LASIK) or photorefractive keratotomy (PRK). Patients who have had
SMILE surgery still cannot avoid future cataract surgery with aging. However, studies have
failed to compare the accuracy of intraocular lens (IOL) power calculation formulae in eyes
after SMILE.
There are two main sources of error in calculating IOL power after refractive surgery,
including corneal power measurement error [1 ], [2 ] and effective lens position (ELP) error [3 ]. The error of corneal power is in itself a two-sided issue. First, corneal
topographers do not consider the change of the back-to-front corneal radius ratio (B/F
ratio) that occurs after excimer laser ablation of the anterior cornea. Thus, an
inappropriate refractive index is used for calculating corneal power. Second, most manual
topographers do not authentically measure the curvature of the central cornea, which is the
flattest area after myopic ablation. The second main error is the estimation error of ELP.
Although this is challenging in virgin eyes as well, it imposes additional challenges after
refractive surgery, especially in formulae that use corneal power to estimate ELP [4 ]. All these errors lead to the underestimation of IOL power in
myopic ablation and the opposite in hyperopic ablation.
Over the past few decades, several formulae have been described to address the accuracy of
the IOL power calculation in eyes after myopic LASIK/PRK surgery. These formulae are divided
into two categories according to whether or not to use the historical data before refractive
surgery [3 ]. The formulae with historical data would require
knowledge of preoperative data and stable postoperative refractions [3 ]. For example, surgically induced refractive change at the corneal plane is needed
in Barrett True-K, Masket, or Modified Masket method [5 ], [6 ]. In addition, pre-refractive surgery keratometry is required in the
Double-K or clinical history method (CHM) [7 ] – [9 ]. CHM was considered the gold standard to calculate IOL power for
eyes with previous LASIK/PRK [8 ]. However, as the historical data
are not available or not credible, the formulae with historical data were proven to be not
as accurate as thought. Since then, several formulae that do not rely on historical data
have been proposed, including Barrett True-K no history, Hill Potvin Shammas PM, Haigis-L,
Shammas-PL, and others [5 ], [10 ] – [14 ]. However, most of them are derived based on
empirical regression analysis, resulting in the accuracy of IOL power calculations for
patients after excimer surgery being lower than that of the virgin eyes [15 ], [16 ], [17 ].
Theoretically, the optical raytracing method based on the Snell law is a better solution in
calculating IOL power after refractive surgery [18 ]. The
raytracing method relies on the true corneal curvature to calculate the corneal power, which
is not subject to the corneal power error [19 ]. Moreover, unlike
in the case of third-generation theoretical formulae, the raytracing method does not use the
corneal power to evaluate ELP [20 ]. Thus, the ELP error could
also be ignored. Previous studies indicated that the raytracing method had an encouraging
outcome in calculating IOL power for eyes after myopic LASIK/PRK surgery [21 ], [22 ], [23 ]. The purpose of this study was to evaluate the accuracy of the raytracing method
that was performed by a rotating Scheimpflug camera combined with a Placido disc corneal
topographer in calculating IOL power for myopic eyes after SMILE surgery.
Patients and Methods
Ethical Approval
This study was performed in accordance with the ethical standards stated in the 2013
Declaration of Helsinki and was approved by the Tianjin Eye Hospital ethics committees.
Informed written consent was obtained from all participants.
Study Participants
This was a retrospective study that comprised patients who had SMILE refractive surgery
for treatment of myopic and/or myopic astigmatism between May 1, 2020, and December 31,
2020, at Tianjin Eye Hospital. Inclusion criteria were (1) age 18 years or older; (2) no
complications during or after SMILE surgery; (3) manifest refraction was performed at
least 6 months postoperatively; and (4) best-corrected distance visual acuity (BCVA) of
20/25 or better at 5 m distance. Exclusion criteria were (1) the presence of active ocular
disease; (2) previous ocular trauma or ocular surgery; and (3) systemic diseases such as
diabetes or connective tissue disorders.
Surgery and Measurement
The same surgeon performed all SMILE procedures using a femtosecond laser platform
(VisuMax, Carl Zeiss, AG, Jena, Germany) with a 500-kHz repetition rate. The parameters
used included an optical zone of 6.5 to 7.0 mm, a cap diameter of 7.5 to 8.0 mm, a
predetermined cap thickness of 120 µm, and an energy of 125 to 160 nJ. The side cut was
placed at the 12-oʼclock position of the cornea with an angle of 90 degrees and a
circumferential width of 2.0 to 3.0 mm. Following removal of the lenticule, the incision
was flushed with balanced salt solution (Alcon Laboratories, Inc., TX, USA). The optical
measurement was performed by Scheimpflug tomography (Oculus, Pentacam AXL, Wetzlar,
Germany) before and 6 months after SMILE surgery. Axial length was measured by a
noncontact biometer (Lenstar LS-900; Haag-Streit, AG, Koeniz, Switzerland).
IOL Power Calculation and Optimization
The preoperative IOL power calculation was calculated by the raytracing method directly
from Scheimpflug tomography (Pentacam AXL, Wetzlar, Germany), which utilized a C constant
to predict the postoperative IOL position. The raytracing calculations were performed
using Pentacam AXL software (version 1.22r05) over a 4.0-mm optical zone directly by the
built-in calculation of “Olsen Raytracing”. The refractive prediction errors (PEs) and IOL
powers after SMILE surgery of Double-K SRK/T, Hill-Potvin Shammas, Barrett True-K no
history, and raytracing were also performed directly by using Pentacam. The following
formulae were back-calculated in the American Society of Cataract and Refractive Surgery
(ASCRS) calculator: Masket method, Modified Masket method, Barrett True-K, Haigis-L, and
Shammas-PL formulae. Lens factors (constants) of each formula were optimized by zeroing
out the mean arithmetic PE. Haigis constants (a0, a1, a2) were obtained from the User
Group for Laser Interference Biometry website.
Outcomes Measures
IOL power calculations were performed using the raytracing method before the SMILE
procedure, and the IOL power corresponding to the minimum myopic refractive error was
recorded. Six months after the SMILE surgery, the same IOL power was selected and the
IOL-induced refractive error at the corneal plane was calculated using the raytracing
method, formulae with historical data (Barrett True-K, Double-K SRK/T, Masket, Modified
Masket), and formulae without historical data (Barrett True-K no history, Haigis-L, Hill
Potvin Shammas PM, Shammas-PL). The difference between the IOL-induced refractive error at
the corneal plane before and after the surgery was defined as IOL-Dif. In addition, the
alteration of the spherical equivalent before and after the SMILE procedure was calculated
in manifest refraction at the corneal plane (SMILE-Dif). The refractive PE was defined as
the difference between the SMILE-Dif and IOL-Dif with different methods and formulae for
the same IOL power and model (SN60WF, Alcon Laboratories, TX, USA) [24 ].
For example, if the patientʼs refraction was − 6.00 D (spherical equivalent) before SMILE
surgery, the IOL power corresponding to the − 0.21 D was + 15 D, then − 0.21 D was
recorded as the minimum myopic refractive error. Six months after the SMILE surgery, the
same + 15 D IOL power was selected and the IOL-induced refractive error at the corneal
plane was + 6.76 D. Then, the IOL-Dif can be calculated as + 6.76 D-(− 0.21 D), which was
+ 6.97 D. In the meanwhile, the patientʼs refraction was also required post-operation,
which in this case was + 0.50 DS, and that made the SMILE-Dif 6.5 D. Taken together, the
PE equaled 0.47 D.
The primary outcome was the mean arithmetic prediction error (ME). In order to eliminate
the bias of the lens factor, the MEs for each formula was made to equal zero by changing
the constant individually for each formula [25 ]. The median
absolute prediction error (MedAE) and mean absolute prediction error (MAE) were defined as
the median and mean arithmetic value that turn all negative PEs into positive values,
respectively. In addition, the percentages of eyes within PEs of ± 0.25 D, ± 0.50 D,
± 0.75 D, and ± 1.00 D were also calculated for different formulae. Subgroup analysis was
performed according to different anterior chamber depths (cutoff value 3.5 mm), axial
lengths (cutoff value 26 mm), B/F ratio (cutoff value 73%), keratometry (cutoff value
38 mm), lens thickness (cutoff value 3.5 mm), and preoperative refraction (cutoff value
− 6 D).
Statistical Analysis
The Kolmogorov-Smirnov test was used to check the data distribution for normality. Each
group of statistics with a normal distribution was shown with a mean (standard deviation),
while those with non-normal distribution were indicated with a median (lower quartiles,
upper quartiles). For the analysis of MedAE and MAE differences, Friedman signed-rank or
Studentʼs t-test was used. Bonferroni correction was applied for multiple tests. The
percentages of the targeting refraction within ± 0.25 D, ± 0.50 D, ± 0.75 D, and ± 1.00 D
were compared using Cochranʼs Q test. Statistical analysis was performed using the
Statistical Product and Service Solutions (SPSS, version 24 for Mac). A probability level
of 0.05 was considered statistically significant.
Results
Baseline Data
A total of 70 eyes of 70 patients were eventually included in this study. The mean age
was 21 ± 3 (18 ~ 32) years and 36 patients (51.4%) were female. All 70 eyes were from the
same side of the patient and had available manifest refraction data before and after
6-month SMILE. The mean preoperative spherical equivalent was − 5.67 ± 1.49 D
(− 9.50~-2.125 D). The mean IOL power before refractive surgery by the raytracing method
was 14.67 ± 2.20 D (11.0 ~ 19.5 D) with a mean predicted spherical equivalent of
− 0.18 ± 0.09 D (− 0.33 ~ 0.02 D) (all detailed data of IOL-Dif and SMILE-Dif data are
listed in Supplementary Table 1). The baseline characteristics of this study cohort are
summarized in [Table 1 ].
Table 1 Baseline characteristics.
Parameter
Values
SE: spherical equivalent; D: diopter; K: keratometry; B/F ratio: back-to-front
corneal radius ratio
Age (years)
21.00 ± 3 (18 – 32)
Sex (male : female)
34 : 36
Preoperative SE (D)
− 5.67 ± 1.49 (− 9.50 ~ − 2.125)
Postoperative SE (D)
− 0.23 ± 0.08 (− 0.75 ~ 0.25)
Axial length (mm)
26.16 ± 0.88 (24.58 ~ 28.00)
Anterior chamber depth (mm)
3.80 ± 0.23 (3.31 ~ 4.25)
Preoperative K (D)
42.95 ± 1.25 (40.25 ~ 46.60)
Postoperative K (D)
38.56 ± 1.62 (34.55 ~ 42.05)
Lens thickness (mm)
3.50 ± 0.15 (3.10 ~ 3.84)
White-to-White (mm)
11.87 ± 1.43 (11.00 ~ 12.80)
B/F ratio (%)
73.59 ± 2.29 (69.10 ~ 79.80)
Outcomes of Different Formulae
[Table 2 ] shows the outcomes of the including nine formulae.
The raytracing method produced the lowest MAE (0.26 ± 0.24 D) and MedAE (0.16 D) in
refractive prediction, which was lower than Double-K SRK/T (p = 0.004 and p < 0.001),
Shammas-PL (both p < 0.001), and Haigis-L (both p < 0.001) formulae. Raytracing also
had a lower MAE than Barrett True-K no history formula (p = 0.048). Additionally, the
Modified Masket method had a lower MAE than Shammas-PL (p = 0.011) and Haigis-L formulae
(p = 0.039). The Masket method was better than the Shammas-PL formula in terms of MAE
(p = 0.030) and MedAE (p = 0.048). Refractive PEs of all formulae are demonstrated in
[Fig. 1 ].
Table 2 Refractive outcomes of different formulae
(number = 70).
Formulae
Absolute prediction error (D)
Proportion of PE% (number)
Mean ± SD
Median (Q1, Q3)
± 0.25 D
± 0.50 D
± 0.75 D
± 1.00 D
*Significant difference compared with the Olsen Raytracing method after
Bonferroni correction.
With history data
Masket
0.36 ± 0.31
0.26 (0.14, 0.48)
50.0 (35)
77.1 (54)
88.6 (62)
95.7 (67)
Modified Masket
0.35 ± 0.26
0.31 (0.15, 0.50)
41.4 (29)
75.7 (53)
91.4 (64)
98.6 (69)
Barrett True-K
0.40 ± 0.34*
0.31 (0.12, 0.62)
47.1 (33)
67.1 (47)
81.4 (57)*
95.7 (67)
Double-K SRK/T
0.48 ± 0.39*
0.39 (0.18, 0.68)*
30.0 (21)*
62.9 (44)
77.1 (54)*
92.9 (65)
Without history data
Olsen Raytracing
0.26 ± 0.24
0.16 (0.07, 0.48)
64.3 (45)
81.4 (57)
95.7 (67)
100.0 (70)
Hill Potvin Shammas PM
0.41 ± 0.31
0.32 (0.16, 0.55)
37.1 (26)*
65.7 (46)
87.1 (61)
92.9 (65)
Barrett True-K no history
0.43 ± 0.35
0.37 (0.16, 0.61)
41.4 (29)
67.1 (47)
87.1 (61)
91.4 (64)
Haigis-L
0.52 ± 0.43*
0.43 (0.16, 0.74)*
31.4 (22)*
60.0 (42)
75.7 (53)*
82.9 (58)*
Shammas-PL
0.54 ± 0.41*
0.45 (0.21, 0.75)*
30.0 (21)*
54.3 (38)*
75.7 (53)*
87.1 (61)*
Fig. 1 Box diagram of the refractive prediction error of all formulae (sorted
by the median absolute prediction error in ascending order).
Moreover, Cochranʼs Q test showed that all nine formulae had significant statistical
differences in percentages of eyes within a PE of ± 0.25 D (p < 0.001), ± 0.50 D
(p = 0.002), ± 0.75 D (p < 0.001), and ± 1.00 D (p < 0.001). The raytracing method
showed the highest percentages of eyes within a PE of ± 0.25 D (64.3%), ± 0.50 D (81.4%),
± 0.75 D (95.7%), and ± 1.00 D (100.0%). [Table 1 ] also showed
the significant difference between the raytracing method and the others after Bonferroni
correction. In pairwise comparison, Haigis-L had a significantly smaller percentage of
eyes within a PE of ± 1.00 D than that of the Modified Masket method (p = 0.001), Masket
method (p = 0.014), Barrett True-K no history (p = 0.014), and Shammas-PL (p = 0.014). The
Masket method achieved a higher percentage of eyes within a PE of ± 0.50 D than Shammas-PL
(p = 0.04). Stacked histogram showed the percentage of eyes within a given diopter range
of predictive refraction outcome ([Fig. 2 ]).
Fig. 2 Stacked histogram comparing the percentage of cases within a given
diopter range of refractive prediction error (sorted by the percentage of eyes within
a prediction error of ± 0.250 D in descending order).
Subgroup Analysis
As the raytracing method considered multiple factors to calculate IOL power, including
anterior chamber depth, lens thickness, and B/F ratio, we performed a subgroup analysis of
these parameters. In addition, a subgroup analysis was also performed according to axial
length, keratometry, and preoperative SE. [Table 3 ] shows the
refractive outcomes from the subgroup analysis. There were no statistically significant
differences in terms of the MedAE or MAE in the different subgroups (all p < 0.05).
Moreover, no difference was found in the percentage of eyes within a PE of ± 0.25 D,
± 0.50 D, ± 0.75 D, or ± 1.00 D as well (all p < 0.05). These results showed that
raytracing had excellent consistency in calculating IOL power after myopic SMILE.
Table 3 Refractive outcomes of Olsen Raytracing formula in different
groups.
Different groups
Absolute prediction error (D)
Proportion of PE% (number)
Mean ± SD
Median (Q1, Q3)
± 0.25 D
± 0.50 D
± 0.75 D
± 1.00 D
ACD: anterior chamber depth; AL: axial length; B/F ratio: back-to-front corneal
radius ratio; D: diopter, K: keratometry; LT: lens thickness; Pre-SE:
preoperative spherical equivalent
Pre-SE > − 6 (42)
0.24 ± 0.23
0.16 (0.05, 0.33)
69.0 (29)
83.3 (35)
95.2 (40)
100.0 (42)
Pre-SE ≤ − 6 (28)
0.30 ± 0.26
0.18 (0.09, 0.50)
57.1 (16)
78.6 (22)
96.4 (27)
100.0 (28)
K < 38 mm (24)
0.26 ± 0.24
0.17 (0.07, 0.49)
66.7 (16)
83.3 (20)
95.8 (23)
100.0 (24)
K ≥ 38 mm (46)
0.26 ± 0.24
0.16 (0.06, 0.44)
63.0 (29)
80.4 (37)
95.7 (44)
100.0 (46)
AL < 26 mm (32)
0.27 ± 0.22
0.21 (0.10, 0.43)
62.5 (20)
81.3 (26)
96.9 (31)
100.0 (32)
AL ≥ 26 mm (38)
0.25 ± 0.26
0.16 (0.03, 0.49)
65.8 (25)
81.6 (31)
94.7 (36)
100.0 (38)
ACD < 3.5 mm (31)
0.27 ± 0.24
0.20 (0.09, 0.49)
58.1 (18)
80.6 (25)
96.8 (30)
100.0 (31)
ACD ≥ 3.5 mm (39)
0.25 ± 0.25
0.16 (0.04, 0.48)
69.2 (27)
82.1 (32)
94.9 (37)
100.0 (39)
LT < 3.5 mm (35)
0.26 ± 0.24
0.20 (0.06, 0.48)
62.9 (22)
82.9 (29)
97.1 (34)
100.0 (35)
LT ≥ 3.5 mm (35)
0.26 ± 0.25
0.16 (0.07, 0.49)
65.7 (23)
80.0 (28)
94.3 (33)
100.0 (35)
B/F ratio < 73% (28)
0.26 ± 0.23
0.17 (0.07,0.49)
60.7 (17)
85.7 (24)
96.4 (27)
100.0 (28)
B/F ratio ≥ 73% (42)
0.26 ± 0.25
0.16 (0.04,0.44)
66.7 (28)
78.6 (33)
95.2 (40)
100.0 (42)
Discussion
Since the inception of SMILE almost 10 years ago, the procedure has been rapidly growing in
popularity [26 ]. IOL power calculations in eyes after SMILE will
inevitably become a challenging task for most ophthalmologists. Due to the limited number of
patients who previously had myopic SMILE surgery and then had cataract surgery, it is
difficult to compare the accuracy of different IOL power calculation formulae by directly
calculating refractive PE. In this study, a theoretical model was adopted, that refractive
PE could be indirectly obtained by calculating the difference between the residual spherical
equivalent predicted by the standard method and the residual spherical equivalent predicted
by the targeted formula for the same IOL power and model [17 ], [24 ]. Olsen Raytracing was used as the standard and
benchmark formula to calculate IOL power before refractive surgery, since it had the best
outcomes in terms of accuracy for long eyes compared with Barrett Universal II, Haigis, or
third-generation formulae [15 ]. Furthermore, the raytracing
method has been proven to be highly effective in estimating the changes in corneal power and
calculating IOL power after LASIK/PRK [18 ], [27 ]. We evaluated the accuracy of IOL power calculations after SMILE
using raytracing and compared the outcomes using formulae with historical data (Barrett
True-K, Double-K SRK/T, Masket, Modified Masket) and without historical data (Barrett True-K
no history, Haigis-L, Hill Potvin Shammas PM, Shammas-PL). Our outcomes showed that
raytracing is the most accurate method in predicting and achieving the target refraction in
calculating IOL power for myopic eyes after SMILE, with good consistency in IOL power
calculations of eyes with different axial lengths, anterior chamber depth, keratometry, lens
thickness, B/F ratio, and preoperative refraction.
As mentioned above, formulae used to calculate IOL power after corneal refractive surgery
can be divided into two categories, including with historical data and without. In this
study, Barrett True-K, Double-K SRK/T, Masket, and Modified Masket require knowledge of
preoperative refractive data. One previous study indicated that Barrett True-K had no
significant differences with Masket and Modified Masket in calculating IOL power after
excimer laser ablation [16 ], which was consistent with our
results in eyes after SMILE. Although our study showed that raytracing had the lowest mean
arithmetic PE among the four formulae with historical data, a significant difference was
only found between raytracing and Double-K SRK/T. In addition, this study also enrolled
formulae that did not require pre-refractive history data, including Barrett True-K no
history. In the early years, Haigis-L and Shammas-PL formulae had good predictability of IOL
power calculations when refractive historical data was unknown. With the introduction of
Barrett True-K no history, its accuracy has been widely recognized in calculating IOL power
in eyes after refractive surgery [28 ], [29 ]. However, a recent meta-analysis found that Barrett True-K no history had no
significant difference from Haigis-L and Shammas-PL [30 ]. The
same result was found in our study.
Unlike the third- and fourth-generation formulas, the raytracing method directly addresses
the curvature of the corneal center, which truly reflects the change in total corneal power
after refractive surgery [31 ]. The true net power (TNP; apex
zone 4 mm) is calculated from the measurement of both corneal surfaces with the real ratio
between the anterior and posterior corneal radius, which is much lower than the anterior
simulated keratometry [12 ]. Our results showed that raytracing
had a lower MedAE than Shammas-PL and Haigis-L, but not Barrett True-K no history and Hill
Potvin Shammas PM. Similar results were found in eyes after myopic LASIK/PRK [32 ], [33 ]. Yet raytracing had a higher
percentage of eyes within a PE of ± 0.25 D than Hill Potvin Shammas PM and within a PE of
± 0.75 D than Barrett True-K no history. These results demonstrated that it had a better
outcome than all mentioned formulae without pre-refractive historical data in calculating
IOL power after SMILE.
The raytracing method uses a special C constant to estimate ELP [21 ]. C constant is a constant based on preoperative anterior chamber depth, lens
thickness, and IOL constant, and is no longer dependent on axial length or corneal
curvature. Therefore, ELP errors could be prevented [34 ].
Moreover, the raytracing method uses measurements of both the anterior and posterior corneal
radii rather than fictitiously assuming a constant ratio of anterior to posterior corneal
curvature to determine total corneal power. Thus, it could also avoid the corneal power
error in calculating IOL power after refractive surgery [34 ].
For eyes with an abnormal corneal curvature, raytracing displays its unique advantages
compared with SRK/T and Haigis formulae [35 ]. Meanwhile, as the
cornea is aspheric and pupil diameter is not fixed, it is not accurate to only consider the
paraxial optical path, which would cause a large aberration. The raytracing method considers
several factors such as corneal irregularity, pupil diameter, and IOL thickness to minimize
the aberration, which is also one of the advantages between the raytracing method and
traditional formulae in calculating IOL power [36 ], [37 ]. Several studies have been devoted to comparing the accuracy of
the raytracing method with traditional formulae for calculating IOL power after refractive
surgery and demonstrated its superiority [38 ], [39 ].
These theoretical methodological advantages of the raytracing method have been previously
proven in patients with a history of previous myopic laser vision correction who underwent
cataract surgery and IOL implantation. For instance, in Saviniʼs report [18 ], the percentage of eyes within a PE of ± 0.50 D (± 1.00 D)
obtained by raytracing was 71.4% (85.7%), with the MedAE being + 0.25 D. Gjerdrum et al.
[40 ] yielded even better results. The Anterion-OKULIX
calculations showed a higher percentage of eyes with PEs within ± 0.25, ± 0.5, and ± 0.75
(60%, 88%, and 100%, respectively). These results seem comparable to our findings in
post-SMILE eyes. Lazaridis et al. [24 ] reported 81.9% of eyes
within a PE of ± 0.50 D in a study of 204 eyes undergoing SMILE. Similar results were found
in our study, in which the percentage of eyes within a PE of ± 0.50 D was 81.4%.
Our present data are also endorsed by the Lischke et al. study, which analyzed the first
cohort of post-SMILE eyes undergoing cataract surgery and IOL implantation. Although the
study is limited by its relatively small sample size, its result is highly coherent with our
findings, in which raytracing showed the smallest mean absolute error (0.40 D) and yielded
the largest percentage of eyes within ± 0.50/± 1.00 D (82/91%) compared to other empirically
optimized formulae available in the ASCRS post-keratorefractive surgery IOL power online
calculator [41 ].
This study has several limitations. One limitation is the retrospective study design with a
limited sample. Further prospective analysis should include more patients. Second, as SMILE
has only been employed by clinicians since 2011, sufficient empirical data on IOL power
prediction accuracy do not exist. Therefore, in this study, a theoretical model was used,
which involved the virtual implantation of the same IOL before and after SMILE instead of
actual implanting IOL in post-SMILE patients. It would be more ideal to gather and analyze a
cohort of post-SMILE patients who have undergone cataract extraction with IOL implantation.
Third, only formulae with a broad application were included in this study. The others, such
as Atlas-, Galilei- or OCT-based corneal measurements, were not enrolled because of their
limited application [10 ], [11 ].
Finally, since Olsen Raytracing was chosen as the standard method to calculate IOL power
before refractive surgery, it may create potential bias that the postop values favored
itself when comparing with other different approaches. However, the meaning of “bias” is
complicated and vague. Zhu et al. conducted research comparing the stability of different
formulae after SMILE surgery [42 ]. In their study, a concept of
equivalent IOL power (EILD) was introduced, and a comparison of the same formula before and
after SMILE surgery was made. Theoretically, if the formula is more stable, it could create
more “bias” when chosen as the benchmark formula to calculate IOL power before refractive
surgery, since it favors its own result. However, on the other hand, when choosing a
standard formula before SMILE, the most stable formula should be the first choice. For
instance, in the above paper, the Barrett True-K formula was found to be more stable than
SRK/T, Holladay 1, and Haigis and was recommended to calculate IOL power. Therefore, the
“bias” may occur but can hardly be eliminated.
Conclusion
Previous studies indicated that the raytracing method provided great accuracy for IOL power
calculations after myopic LASIK/PRK surgery. But its accuracy has not been described for
eyes that underwent SMILE surgery, especially when raytracing was chosen as the standard
method to calculate IOL power before SMILE. In this study, we have demonstrated that
raytracing is the most accurate method in predicting and achieving the target refraction in
calculating IOL power for myopic eyes after SMILE. It has a good consistency in IOL power
calculations of eyes with different axial lengths, anterior chamber depths, keratometry,
lens thickness, B/F ratio, and preoperative refraction.