Network meta-analysis of no-history methods to calculate intraocular lens power in eyes with previous myopic laser refractive surgery

Authors: Wen D, Yu J, Zeng Z, McAlinden C, Hu L, Feng K, et al.

Geographical coverage: Europe, Asia, and North America

Sector: Biomedical

Sub-sector: Treatment

Equity focus: None

Study population: People who had received surgery for myopia and had subsequently undertaken IOL power calculations.   

Review type: Effectiveness review

Quantitative synthesis method: Meta-analysis using no history methods.

Qualitative synthesis method: Not applicable

Background:

LASIK, PRK, and LASEK are frequently used surgical techniques to rectify myopia. It’s widely acknowledged that past corneal laser refractive surgeries can pose challenges in obtaining precise biometry during cataract surgery. Over the last two decades, numerous strategies have been devised to enhance the precision of intraocular lens (IOL) power estimation in such cases. Given the challenges in accessing past data, methods that don’t rely on historical data are considered optimal for calculating IOL power. Research has also indicated that these no-history methods can yield superior results compared to those that use historical data.

Objectives: To systematically compare and rank the predictability of no-history intraocular lens (IOL) power calculation methods after myopic laser refractive surgery.

Main findings:

Taking into account the three outcome measures – highest percentages of eyes with a prediction error (PE) within ±0.50 and ±1.00 D, lowest mean absolute error (MAE), and lowest median absolute error (MedAE), researchers identified the top three no-history formulas for intraocular lens (IOL) power estimation in eyes that had undergone myopic corneal laser refractive surgery. These were: ORA, BESSt, and Triple-S (D-K SRK/T).

A total of 19 studies were included in the network meta-analysis. Included trials were published from 2009 to 2019. Almost all trials involved three or more formulas, with the exception of three atudies which had only two formulas. Of the included 19 trials, four recruited participants from Europe, six from Asia, and nine from North America. The majority of included studies (n=18) were regarded as high quality.

The studies analysed 1,098 eyes using 19 different formulas. The network meta-analysis revealed that certain formulas (Okulix, ORA, BESSt, Triple-S (D-K SRK/T), and Fourier-Domain OCT-Based) were more accurate in predicting the percentage of eyes with a prediction error (PE) within ±0.50 D compared to others. The top four formulas, according to the surface under the cumulative ranking curve (SUCRA) values, were Okulix, ORA, BESSt, and Triple-S (D-K SRK/T). The ORA formula showed fewer errors than the Shammas-PL formula. The top four formulas based on SUCRA values were Triple-S, BESSt, ORA, and Fourier-Domain OCT-Based. The formulas with the lowest median absolute error (MedAE) were SToP (SRK/T), ORA, Fourier-Domain OCT-Based, and BESSt.Node-splitting analysis in terms of PE within ±0.50 or ± 1.00 D and MAE showed significant consistency (p >.05). The design-by-treatment interactions model was used and found that the global inconsistency existed in the MAE (p = .0046). The funnel plots showed that the included studies lie symmetrically around the 0 line (vertical line) with respect to the PE within ±0.50 or ±1.00 D and MAE.

On the basis of evidence from this analysis, authors propose the following evidence-based guidelines for policy: ORA, BESSt and Triple-S (D-K SRK/T) formulas provided the highest percentages of eyes with a PE within ±0.50 and ±1.00 D, as well as the lowest MAE and MedAE. Authors also suggest that future research should address other formulas to confirm accuracy of methods.

Methodology:

The review included case series studies involving patients who had cataract surgery following corneal myopia laser refractive surgery. No-history formulas were used to calculate the intraocular lens (IOL) power. The most accurate combination of methods that modify keratometric readings was selected when combined with different IOL power formulas. Studies were compared based on two or more formulas and reported at least one of the specified outcome measurements. Exclusions were made for studies that contained only one no-history method, included patients who had undergone hyperopic corneal laser refractive surgery or radial keratotomy, used Single-K formulas when Double-K should have been applied, were classified as no-history methods but used some historical data, or analysed methods or formulas deemed inappropriate. There were no language restrictions. Authors conducted a systematic literature review using PubMed, EMBASE, the Cochrane Library, and the U.S. trial registry for trials published up to August 2019. Additionally, authors manually examined the reference lists of clinical trials, related meta-analyses, and systematic reviews to identify relevant studies.

Study screening was performed by two independent investigators. They retrieved the full-text articles that appeared relevant after reviewing the titles and abstracts. They independently assessed full-text articles for final eligibility. Any discrepancy was resolved by focused discussion or consultation with an additional investigator. Additionally, two investigators independently extracted information into an electronic database. For data that was missing or could not be directly obtained, the authors of trial reports were contacted or GetData GraphDigitizer was used to read data from figures.

To evaluate the study quality, the Quality Appraisal Tool for case series studies using a modified Delphi technique developed by the Institute of Health Economics was used.

Traditional pairwise meta-analyses were performed using random-effects models. Odds ratios with 95% CIs were calculated for binary outcomes, while weighted mean differences with 95% CIs were calculated for continuous outcomes. MedAE was analysed descriptively. Positive outcomes had odds ratios greater than 1, indicating beneficial effects of the first formula over the second. Negative outcomes had weighted mean differences less than 0, indicating the same. Statistical heterogeneity was investigated using the I² statistic. Network meta-analysis was performed to incorporate indirect comparisons, with relative rankings estimated using ranking probabilities and surface under the cumulative ranking curve. Inconsistency between direct and indirect evidence was assessed using a “node-splitting” approach. Publication bias was evaluated using funnel plots, with subgroup analyses conducted for methods evaluated in at least three different studies and 100 eyes.

Applicability/external validity: Authors note that this network meta-analysis has several limitations, which affect the applicability and external validity of review findings. Furthermore, included trials were conducted in Europe, Asia and North America. Therefore, it is unclear whether the conclusions of study apply to other populations.

Geographic focus: Included studies were conducted in Europe, Asia and North America, with a small number conducted in low- and middle-income settings.

Summary of quality assessment:

The approaches used to identify, select and critically appraise studies were generally rigorous, with at least two authors completing all key tasks. However, the search was somewhat limited in that it did not seek to incorporate unpublished material; it is also unclear what the start dates for the searches were. The approach to the analysis of the data was very robust; however, as the authors acknowledge themselves, the issue of possible correlation between data obtained from ‘pairs’ of eyes was not dealt with. For these reasons, we have medium confidence in the findings of this review.