A Bayesian network meta-analysis on comparisons of intraocular lens power calculation methods for paediatric cataract eyes

Author: Hong Y, Sun Y, Xiao B, Ainiwaer M, Ji Y

Geographical coverage: Asia, North America, the Middle East and Africa,

Sector: Intraocular lens

Sub-sector: Accuracy

Equity focus: Not reported

Study population: Paediatric patients with cataracts

Review type: Effectiveness review

Quantitative synthesis method: Meta-analysis

Qualitative synthesis method: Not applicable

Background

Cataract surgery is generally safe and effective in children, yet accurate calculation of intra-ocular lens (IOL) power remains problematic. Most IOL power formulae and biometry devices are designed for adults, whereas paediatric eyes exhibit higher keratometry, shorter axial length and a shallower anterior chamber, pre-disposing them to larger postoperative refractive errors. Numerous studies have compared adult-based formulae in children, but inconsistent findings hinder clear guidance. A systematic comparison is therefore required to inform surgical planning and optimise visual outcomes in paediatric patients.

Objective

To compare and rank the accuracy of available formulae for calculating IOL power in paediatric eyes.

Main findings

Thirteen studies encompassing 1,781 paediatric eyes were included. Four studies were conducted in China, two in India, two in the United States and one each in Saudi Arabia, Canada, Nepal, Egypt and Korea. Twelve studies were retrospective; one was prospective. Patient ages ranged from 1.2 to 216 months, with mean axial lengths of 19.2–22.3 mm. Formulae assessed were: Barrett Universal II (Barrett UII), Haigis, Holladay 1, Holladay 2, Hoffer Q, Olsen, SRK/T and SRK II.

• Pairwise meta-analysis
  • Prediction error (PE) ±0.50 dioptres (D): SRK/T out-performed SRK II (risk ratio [RR] 1.15; 95 % CI 1.03–1.30).
  • PE ±1.00 D: SRK/T (RR 1.10; 95 % CI 1.02–1.18) and Holladay 1 (RR 1.15; 95 % CI 1.01–1.30) were superior to SRK II.
  • No other significant differences were found.
• Bayesian network meta-analysis
  • Integrating direct and indirect evidence revealed no statistically significant differences between formulae for either PE threshold.
  • Ranking by SUCRA
    • ±0.50 D: Barrett UII (80.3 %), Haigis (67.2 %), Holladay 1 (64.5 %), SRK/T (63.4 %).
    • ±1.00 D: Holladay 1 (73.6 %), SRK/T (68.6 %), Hoffer Q (53.4 %).
      These results support Barrett UII, SRK/T and Holladay 1 as the three most accurate formulae overall.
• Age-stratified analysis
  • Children ≤2 years: SRK II ranked highest (SUCRA 89.9 % for ±0.50 D; 89.0 % for ±1.00 D), followed by SRK/T and Holladay 1. Barrett UII performed poorly, likely owing to limited data in this group.
  • Children >2 years: Barrett UII ranked first (SUCRA 92.1 % for ±0.50 D; 88.4 % for ±1.00 D), with SRK/T, Haigis and Holladay 1 also performing well.

Methodology

Searches of PubMed, Web of Science, the Cochrane Library and EMBASE up to December 2021 identified studies involving paediatric cataract surgery with in-the-bag IOL implantation that compared at least two formulae and reported the proportion of eyes with PE within ±0.50 D and/or ±1.00 D. No restrictions were placed on language or publication year. Reference lists were hand-searched. Two reviewers independently screened, extracted data and assessed risk of bias using QUADAS-2; discrepancies were resolved by consensus or a third reviewer. Fixed-effects models were applied when heterogeneity was low (I² < 50 %); otherwise random-effects models were used. A Bayesian network meta-analysis integrated direct and indirect comparisons, and formulae were ranked by SUCRA.

Applicability / external validity

Variability in biometric techniques, patient ages and IOL models, together with sparse data for some formulae (particularly Barrett UII in very young children), may limit generalisability. Inclusion of both eyes from individual patients in several studies could introduce correlation bias. Future research should incorporate age-specific data and compare emerging formulae.

Geographic focus

No geographical restrictions were applied; studies originated from Asia, North America, the Middle East and Africa, reflecting broad global practice.

Summary of quality assessment

Overall confidence in the review’s conclusions is high. Searches were comprehensive and unrestricted by language; eligibility criteria were explicit; dual processes for screening, extraction and quality appraisal were employed; study characteristics were well described; and heterogeneity was explored. The review did not, however, supply a list of excluded studies.