IOL Power Calculation Formulas in Post-LASIK Eyes Using Two Biometry Systems in Vietnam

Abstract

Background:

Cataract surgery aims to restore clear vision by replacing the clouded crystalline lens with an intraocular lens (IOL). Accurate IOL power calculation is crucial to achieve desirable postoperative refractive outcomes.

Objective:

The aim of this study was to compare the predictive accuracy of multiple intraocular lens (IOL) power calculation formulas in eyes with prior LASIK surgery using two biometric devices, LenStar 900 and IOLMaster 500, combined with immersion ultrasound (US).

Methods:

This retrospective observational study included 37 eyes of 29 patients who underwent previous LASIK and subsequent cataract surgery. Biometric measurements included axial length, keratometry, anterior chamber depth, and lens thickness, which were obtained using LenStar 900 (19 eyes) or IOLMaster 500 combined with US (18 eyes). IOL power was calculated using the Shammas PL, Haigis-L, Barrett TK no history, Shammas Cooke, and EVO 2.0 (without PK1/PK2). The mean absolute error (MAE) at 3 months postoperatively was assessed for each formula within each device group.

Results:

For LenStar 900, the MAE ranged from 0.389 ± 0.329 D (Barrett TK no history) to 0.574 ± 0.689 D (Shammas Cooke), with no significant differences among the formulas. For IOLMaster 500 + US, the MAE ranged from 0.423 ± 0.210 D (Barrett TK no history) to 0.601 ± 0.510 D (Shammas Cooke). Comparisons between devices revealed significantly lower MAE with LenStar 900 for Shammas PL, Shammas Cooke, and EVO 2.0, whereas differences for Haigis-L and Barrett TK showed no significant differences.

Conclusion:

IOL power prediction in post-LASIK eyes varied according to the biometry device. Barrett TK had no history, and EVO performed well with LenStar 900, while Haigis-L showed consistent accuracy across devices, suggesting that it may be the most reliable formula when using non-synchronous biometry systems.

1. BACKGROUND

Cataract surgery aims to restore clear vision by replacing the clouded crystalline lens with an intraocular lens (IOL). Accurate IOL power calculation is crucial to achieve desirable postoperative refractive outcomes. In eyes that have previously undergone myopic LASIK, altered corneal curvature and thickness pose challenges for standard IOL formulas, often leading to refractive surprises if conventional methods are applied (1).

In Vietnam, the number of post-LASIK eyes that underwent cataract extraction with IOL implantation remains limited. Moreover, data collection spans many years, during which the availability and type of biometric devices vary owing to equipment malfunction, replacement, or institutional constraints. As a result, patients in a single cohort may have undergone ocular measurements using different devices, creating heterogeneity in the recorded biometric data. This practical situation raises an important question: Does the choice of the biometric system affect the predictive accuracy of IOL power formulas in post-LASIK eyes (2, 3).

Several formulas have been specifically developed or modified for eyes after refractive surgery, including Shammas PL, Haigis-L, Barrett True-K no history, Shammas Cooke, and EVO 2.0, without PK1 and PK2. These formulas incorporate corneal measurements and axial length in ways that attempt to compensate for previous LASIK-induced changes; however, studies comparing their performance under different measurement devices are scarce (4).

The two widely used biometric systems are the LenStar 900 and IOLMaster 500 combined with immersion ultrasound (US). LenStar 900 performs optical low-coherence reflectometry to measure axial length, corneal power, anterior chamber depth, and lens thickness in a single session. In contrast, the IOLMaster 500, often supplemented with immersion ultrasound, allows accurate axial length and corneal power measurements, with ultrasound providing additional anterior chamber depth and lens thickness data (5, 6).

2. OBJECTIVE

This study aimed to evaluate and compare the accuracy of five IOL power calculation formulas in post-myopic LASIK eyes by using these two different biometry devices. The mean absolute error (MAE) at three months postoperatively was chosen as the primary outcome, allowing us to assess which formula-device combinations provide the most reliable predictions for this challenging patient population.

3. MATERIAL AND METHODS

Study Design and Participants

This retrospective observational comparative study was conducted at the Vietnam National Institute of Ophthalmology and High-Tech Eye Center of Dong Do Hospital, Hanoi, Vietnam, from March 2016 to April 2025. The study adhered to the tenets of the Declaration of Helsinki and was approved by the Institutional Review Board of Hanoi Medical University. Informed consent was obtained from all participants. Our Institutional Review Board approved this (Ref: 886/GCNHDDDNCYSH-DHYHN, dated April 06, 2023).

We included patients with post-myopic LASIK who were scheduled for cataract surgery with intraocular lens implantation.

The inclusion criteria were as follows.

• Age ≥ 18 years,

• History of myopic LASIK in at least one eye,

• Availability of preoperative biometry and keratometry data,

• Follow-up at 1 and 3 months postoperatively,

Exclusion criteria included:

• History of other corneal refractive surgery (e.g., PRK, RK)

• Ocular comorbidities affecting visual outcomes (e.g., keratoconus, glaucoma, uveitis),

• Intraoperative or postoperative complications affecting refraction.

Biometric Measurement

Two groups were defined based on the measurement device used:

a) LenStar 900 (Haag-Streit AG, Switzerland) – 19 eyes: axial length (AL), keratometry (K1, K2), anterior chamber depth (ACD), and lens thickness (LT) were measured simultaneously using optical low-coherence reflectometry (OLCR).

b) IOLMaster 500 (Carl Zeiss Meditec, Germany) combined with immersion ultrasound – 18 eyes: AL, K1, and K2 were measured using IOLMaster 500, and ACD and LT were measured using immersion ultrasound when OLCR was unavailable. WTW was measured manually using a slit-lamp mounted caliper (or surgical caliper). Central corneal thickness (CCT) was measured by ultrasound pachymetry (A-scan based).

The intraocular lens (IOL) power for each case was calculated using five different formulas that are widely applied post-LASIK cataract surgery:

• Shammas PL formula,

The Shammas PL method was applied using the ASCRS IOL Calculator for Eyes with Prior Myopic LASIK/PRK (American Society of Cataract and Refractive Surgery, available at: https://iolcalc.ascrs.org). This approach estimates corneal power without requiring historical data, and adjusts the effective lens position (ELP) accordingly.

• Haigis-Lformula,

The Haigis-L formula, also provided in the ASCRS online calculator, was used. It is specifically designed for eyes that have undergone myopic LASIK/PRK and incorporates modified corneal power estimation into the Haigis formula to improve accuracy.

• Barrett True-K (no history) formula,

The Barrett True-K no-history formula, accessed through the ASCRS calculator, was used for cases where pre-LASIK keratometry was unavailable. This method derives the true corneal power based on current biometric data and provides reliable predictions without prior surgical data.

• Shammas-Cooke formula,

The Shammas-Cooke formula was applied using a dedicated online calculator (http://www.iolcalc.com or the Shammas-Cooke Calculator website). This formula is designed for post-refractive eyes and combines the elements of the Shammas approach with adjustments proposed by Cooke to refine the accuracy of corneal power estimation.

• EVO 2.0 formula without PK1 and PK2.

The EVO 2.0 formula was used via the EVO IOL online calculator (Emmetropia Verifying Optical Software). For this study, the calculation was performed without PK1 and PK2 inputs because these data were unavailable. EVO 2.0 applies ray-tracing–based modeling to derive corneal power and optimize refractive prediction in post-refractive eyes.

All calculations were performed independently for each eye. For consistency, the same set of biometric inputs (axial length, keratometry, anterior chamber depth, and lens thickness) was used across all formulas, with data obtained from the LenStar 900 or IOLMaster 500 combined with immersion ultrasound, as appropriate.

Predicted refractive outcomes were calculated using each formula for the selected IOL, and the mean absolute error (MAE) was determined as the absolute difference between the predicted and actual postoperative spherical equivalents at 3 months.

Surgical Procedure

All cataract surgeries were performed using standard phacoemulsification with posterior-chamber IOL implantation. Multiple experienced surgeons participated in the procedures, following a standardized surgical protocol to ensure consistency in the technique. The choice of IOL power was based on the preoperative biometry results of each measurement group. Refraction was measured 1 and 3 months postoperatively.

Outcome Measures

The primary outcome was MAE of each IOL formula at 3 months postoperatively. The secondary outcomes included the comparison of the MAE between the two device groups (LenStar 900 vs. IOLMaster 500 + immersion ultrasound).

Statistical Analysis

All statistical analyses were performed using SPSS version 30 software (IBM Corp., Armonk, NY, USA). Continuous variables are presented as mean ± standard deviation (SD) or median with interquartile range, depending on the normality of distribution (Shapiro-Wilk test).

• Within-group comparisons of MAE among formulas were performed using Friedman’s two-way analysis of variance by ranks with post-hoc Bonferroni correction.

• Between-group comparisons of MAE for each formula were performed using the Mann-Whitney U test because of the non-normal distribution.

• A p-value < 0.05 was considered statistically significant.

Tables and figures were constructed to present descriptive statistics, mean ranks, and p-values for both intra- and intergroup comparisons.

4. RESULTS

Demographic and Preoperative Characteristics

A total of 37 post-myopic LASIK eyes from 29 patients were included in this study. Nineteen eyes were measured using LenStar 900 (Group 1) and 18 eyes were measured using IOLMaster 500 combined with immersion ultrasound (Group 2). The mean age was 44.0 ± 9.28 years in Group 1 and 47.2 ± 9.39 years in Group 2 (p = 0.32). The mean interval from prior LASIK to cataract surgery was significantly longer in Group 2 (16.9 ± 3.7 years) than in Group 1 (12.7 ± 3.9 years, p < 0.01). logMAR (UCVA) in Group 0.40 ± 3.00 logMAR (mean 1.48 ± 0.74) in Group 1 and 0.50 to 3.00 logMAR (mean 1.62 ± 0.85) in Group 2 (p = 0.53). Corneal keratometry (K1, K2, and mean K) was slightly steeper in Group 2, with borderline significance for K1 (38.06 ± 2.75 D vs 36.99 ± 2.88 D, p = 0.08) and mean K (38.47 ± 2.75 D vs 37.70 ± 2.83 D, p = 0.07). Corneal astigmatism was higher in Group 1 (-1.26 ± 0.77 D vs -0.84 ± 0.29 D, p = 0.04).

Axial length (AL) was comparable between groups (28.57 ± 2.93 mm vs 29.39 ± 1.72 mm, p = 0.12). Central corneal thickness (CCT) and lens thickness (LenT) were greater in Group 2 (476.9 ± 30.3 μm vs 443.8 ± 42.2 μm, p < 0.01; 4.49 ± 0.43 mm vs 3.85 ± 0.61 mm, p < 0.01, respectively). The anterior chamber depth (ACD) and horizontal corneal diameter (WTW) were similar between groups. The mean implanted IOL power was higher in Group 1 (18.03 ± 5.64 D) than in Group 2 (14.03 ± 3.70 D, p < 0.01) (Table 1).

Table 1. Demographics and Preoperative Characteristics of Study Eyes Abbreviations: UCVA = uncorrected visual acuity; BCVA = best-corrected visual acuity; K1/K2 = keratometry values; Mean K = mean keratometry; CCT = central corneal thickness; ACD = anterior chamber depth; WTW = white-to-white corneal diameter; IOL = intraocular lens; US = ultrasound biometry.

Variable	LenStar 9000 (n=19)	IOLMaster 500 + US (n=18)	p-value
Age (years)	44.0 ± 9.28	47.2 ± 9.39	0.32
Time since LASIK (years)	12.7 ± 3.86	16.9 ± 3.70	<0.01
Pre-op UCVA (logMAR)	1.48 ± 0.74	1.62 ± 0.85	0.53
K1 (D)	36.99 ± 2.88	38.06 ± 2.75	0.08
K2 (D)	38.31 ± 2.75	38.88 ± 2.76	0.35
Mean K (D)	37.70 ± 2.83	38.47 ± 2.75	0.07
Corneal astigmatism (D)	-1.26 ± 0.77	-0.84 ± 0.29	0.04
Axial length (mm)	28.57 ± 2.93	29.39 ± 1.72	0.12
CCT (μm)	443.8 ± 42.18	476.9 ± 30.3	<0.01
ACD (mm)	3.44 ± 0.45	3.45 ± 0.29	0.92
Lens thickness (mm)	3.85 ± 0.61	4.49 ± 0.43	<0.01
WTW (mm)	11.81 ± 0.33	11.75 ± 0.37	0.58
Implanted IOL power (D)	18.03 ± 5.64	14.03 ± 3.70	<0.01
UCVA 3 mo (logMAR)	0.36 ± 0.23	0.33 ± 0.17	0.56
BCVA 3 mo (logMAR)	0.20 ± 0.15	0.21 ± 0.18	0.74

Postoperative Visual Acuity

At 3 months, UCVA improved to 0.36 ± 0.23 logMAR in Group 1 and 0.33 ± 0.17 logMAR in Group 2 (p = 0.56). Best-corrected visual acuity (BCVA) was also comparable (0.20 ± 0.15 logMAR vs 0.21 ± 0.18 logMAR, p = 0.74), indicating similar functional outcomes regardless of the biometry device used.

Mean Absolute Error (MAE) Within Groups

In the LenStar 900 group, the mean MAE ranged from 0.389 ± 0.329 D (Barrett TK no history) to 0.574 ± 0.689 D (Shammas Cooke). Friedman’s test revealed no significant differences between the five formulas (χ² = 6.197, df = 4, p = 0.185). Post hoc pairwise comparisons with Bonferroni adjustment showed no statistically significant differences between the formulas (all p_adj > 0.05) (Table 2).

Table 2. Descriptive statistics and Friedman test for mean absolute error (MAE) at 3 months – LenStar 900 (n = 19). Abbreviations: MAE = mean absolute error; SD = standard deviation; D = diopter; PK = posterior corneal power. Statistical test: Friedman test with post-hoc pairwise comparisons adjusted by Bonferroni correction.

Formula	Mean ± SD (D)	Range (D)	Mean Rank	Post-hoc Adj p-value
Shammas PL	0.487 ± 0.420	0.08–2.00	3.26	1.000
Haigis-L	0.546 ± 0.420	0.03–1.67	3.34	1.000
Barrett TK no history	0.389 ± 0.329	0.04–1.21	2.45	1.000
Shammas Cooke	0.574 ± 0.689	0.00–3.05	3.37	1.000
EVO 2.0(Without PK1/PK2)	0.422 ± 0.451	0.03–2.04	2.58	1.000

In the IOLMaster+US group, the mean MAE ranged from 0.451 ± 0.278 D (Haigis-L) to 0.851 ± 0.335 D (Shammas PL). Friedman’s test demonstrated a significant difference between formulas (χ² = 23.922, df = 4, p < 0.001). Post-hoc comparisons indicated that Haigis-L had a significantly lower MAE than Shammas PL (p_adj < 0.001), Shammas Cooke (p_adj = 0.037), and Barrett TK (p_adj = 0.007). Shammas PL had a significantly higher MAE than Haigis-L (p_adj < 0.001) and EVO (p_adj = 0.019) (Table 3).

Table 3. Descriptive statistics, Friedman, and post-hoc results for mean absolute error (MAE) at 3 months – IOLMaster 500 + US (n = 18) Abbreviations: MAE = mean absolute error; SD = standard deviation; D = diopter; PK = posterior corneal power; ns = not significant. Statistical test: Friedman test was used to compare multiple formulas, followed by post-hoc pairwise comparisons with Bonferroni correction.

Formula	Mean ± SD (D)	Range (D)	Mean Rank	Post-hoc Adj p-value
Shammas PL	0.851 ± 0.335	0.09–1.36	4.00	–
Haigis-L	0.451 ± 0.278	0.05–1.12	1.78	<0.001 (vs Shammas PL), 0.037 (vs Shammas Cooke), 0.007 (vs Barett TK no history)
Barrett TK no history	0.759 ± 0.307	0.02–1.34	3.56	0.007 (vs Haigis- L)
Shammas Cooke	0.804 ± 0.328	0.19–1.40	3.31	0.037 (vs Haigis- L)
EVO 2.0 (without PK1/PK2)	0.649 ± 0.231	0.28–1.09	2.36	0.019 (vs Shammas PL), ns (vs others)

Between-Group Comparisons

Abbreviations: MAE = mean absolute error; D = diopter. Statistical test: Mann–Whitney U test was used for between-group comparisons (LenStar 900 vs IOLMaster 500 + US). Error bars represent ± standard deviation (SD).

Comparison of MAE between LenStar 900 and IOLMaster+US revealed that Shammas PL, Barrett TK no history, and EVO 2.0 had significantly lower errors in the LenStar group (p < 0.01 all). Shammas Cooke showed a trend toward lower MAE in LenStar (p = 0.007), whereas the Haigis-L performance was comparable between devices (p = 0.715) (Figure 1). These results suggest that LenStar 900 may provide slightly better predictive accuracy for certain formulas, whereas Haigis-L accuracy is similar between devices.

Figure 1. Between-group comparison of mean absolute error (MAE) at 3 months – LenStar 900 vs IOLMaster 500 + US (Mann–Whitney U test).

5. DISCUSSION

Accurate intraocular lens (IOL) power calculation in post-LASIK eyes remains challenging worldwide because of the altered anterior corneal curvature and variable corneal indices (7, 8). In our study, we evaluated five IOL formulas, Shammas PL, Haigis-L, Barrett TK, Shammas Cooke, and EVO 2.0, without PK1 and PK2, using two different biometric approaches: LenStar 900 and IOLMaster 500 combined with immersion ultrasound (US).

Formula Performance within Each Device

For LenStar 900, the mean absolute error (MAE) ranged from 0.389 ± 0.329 D (Barrett TK no history) to 0.574 ± 0.689 D (Shammas Cooke), with no significant differences among the formulas (Friedman χ² = 6.197, df = 4, p = 0.185). These findings suggest that all formulas performed comparably in eyes measured with the fully integrated LenStar 900. This aligns with prior studies showing that multi-parameter formulas, such as Barrett TK and Shammas PL, yield similar refractive predictability when complete biometric data are available (9). The lowest mean absolute errors (MAE) were found for Barrett TK (0.389 ± 0.329 D) and EVO (0.422 ± 0.451 D), consistent with prior studies reporting the high accuracy of these formulas in post-LASIK eyes (7, 10). Melles et al. reported MAE values of 0.35–0.45 D for Barrett TK (11). In a recent multicenter Japanese study (2025), EVO 2.0, most likely applied in its K-only version, demonstrated a mean absolute error (MAE) of 0.48 D. This was slightly better than that of Haigis-L (0.51 D) and comparable to those of Shammas (0.46 D) and Pearl-DGS (0.46 D). However, EVO 2.0 performed worse than Barrett True-K no-history (0.43 D) and the ASCRS average method (0.42 D). These results indicate that EVO 2.0, without posterior keratometry, provides a reasonable alternative in post-LASIK eyes, surpassing some legacy no-history formulas, but it may not consistently reach the predictive accuracy achieved by the Barrett True-K no-history or ASCRS average (12). Haigis also highlighted the importance of formulas incorporating independent anterior chamber depth (ACD) and lens thickness measurements in post-refractive eyes(13).

IOLMaster 500 + US group: The mean MAE ranged from 0.451 ± 0.278 D (Haigis-L) to 0.851 ± 0.335 D (Shammas PL), with a significant difference between the formulas (Friedman test, p < 0.001). The Post-hoc analysis indicated that Haigis-L outperformed Shammas PL, Shammas Cooke, and Barrett TK. This finding supports previous evidence that Haigis-L, when combined with accurate AL, ACD, and LT measurements, even from non-synchronous biometry maintains good predictability (9). The higher MAE for Shammas PL in this group may reflect the sensitivity of the formula to incomplete or less precise biometry inputs, consistent with previous reports (1). EVO 2.0, although lacking PK1 and PK2, showed intermediate MAE, suggesting its adaptability, but highlighting that missing central corneal data may reduce predictive accuracy in post-LASIK eyes, which is consistent with international reports emphasizing Haigis-L as reliable for post-LASIK eyes, particularly when ACD and lens thickness are independently measured (14, 15).

Comparison Between Biometric Approaches

Our study demonstrated that the predictive accuracy of IOL power calculation formulas in post-LASIK eyes varies according to the biometric device used. Across formulas, LenStar 900 generally showed a lower MAE than IOLMaster 500 combined with immersion ultrasound for Shammas PL, Barrett TK no history, Shammas Cooke, and EVO 2.0 (without PK1/PK2), whereas Haigis-L performance remained comparable between devices (p = 0.715). This finding is consistent with previous reports indicating that optical low-coherence reflectometry (OLCR) devices, which simultaneously measure axial length, keratometry, anterior chamber depth, and lens thickness, may improve predictive accuracy, particularly for formulas sensitive to incomplete biometric inputs.

The Shammas PL, Barrett TK no history, and EVO 2.0 formulas demonstrated significantly lower MAE when using LenStar 900 (p < 0.01), highlighting the impact of synchronous multi-parameter measurements on formula performance. In contrast, Haigis-L maintained similar accuracy across both devices, confirming its robustness and reliability even when biometry measurements were obtained using non-synchronous systems. Our results align with the findings of Wang et al. (2015) (9), who emphasized that post-refractive surgery IOL calculation formulas are highly dependent on the precise measurement of anterior segment parameters.

Interestingly, Shammas Cooke showed a trend toward better performance with LenStar 900 (p = 0.007), suggesting that while modern formulas improve predictability, some formulas remain sensitive to measurement variability. These findings reinforce the clinical importance of selecting appropriate biometry methods and formulas to minimize refractive surprises in post-LASIK eyes. Overall, the study supports the preferential use of OLCR-based devices, such as LenStar 900, for post-refractive surgery eyes, while confirming the relative reliability of Haigis-L across devices (16, 17).

Practical Implications in the Vietnamese Context

In Vietnam, many ophthalmic centers, especially in provincial or rural settings, lack integrated OLCR devices. The IOLMaster 500 is widely available; however, some centers still rely on ultrasound immersion for ACD and lens thickness measurements. Our data suggest that the combination of IOLMaster 500 + US is an acceptable alternative when OLCR is unavailable, provided the Haigis-L formula is used. The MAE values obtained using this approach were within clinically acceptable limits (<0.5 D), supporting its use in resource-limited settings. This aligns with practical recommendations, suggesting that multiparameter formulas can compensate for incomplete measurements when biometric devices are not fully integrated(18).

Clinical Relevance

Our results indicate that for post-LASIK eyes:

• Barrett TK no history and EVO (without PK1/PK2) formulas are the most reliable when advanced OLCR biometry is available.

• Haigis-L is the preferred formula when using IOLMaster 500 + US, producing an MAE comparable to that of the OLCR-based measurements.

• Centers without integrated OLCR devices can achieve acceptable refractive outcomes by using a combination of optical and ultrasound biometry with appropriate formulas.

Limitations of the study

The sample size was relatively small (n = 37), and the follow-up period was limited to 3 months postoperatively. Only post-LASIK eyes were included, and extrapolation to post-PRK or SMILE eyes should be performed with caution. Newer formulas, such as Kane or EVO 2.0 (with PK1/PK2), were not assessed.

Future Directions

Future multicenter studies in Vietnam should evaluate larger cohorts, include newer formulas, and consider device availability in regional centers. Optimizing formula selection based on device capabilities will maximize the postoperative refractive outcomes.

6. CONCLUSION

The predictive accuracy of IOL power calculation formulas in post-LASIK eyes varied with the biometry device used. Barrett TK no history and EVO (without PK1/PK2) performed favorably with LenStar 900, whereas Haigis-L and EVO (without PK1/PK2) yielded lower prediction errors in the IOLMaster 500 + ultrasound group. Haigis-L showed consistent accuracy across devices, indicating that it may be the most reliable formula, especially when using nonsynchronous biometry systems.

Data Access Statement:

All relevant data supporting the findings of this study have been included in the manuscript.

Author’s Contribution:

Each author was included in all phases of preparation of this article. Final proofreading was made by the first author before printing.

Conflicts of interest:

There are no conflicts of interest.

Funding Statement:

This study received no specific grants from any funding agency in the public, commercial, or not-for-profit sectors.