IOL Calculation Formulas: SRK/T vs Haigis vs Barrett — A Practical Comparison
Choosing the right IOL formula for your cataract patients isn't just academic — the formula choice affects visual outcomes. This guide explains when each formula performs best and how to document IOL biometry efficiently.
Why Formula Choice Matters for Refractive Outcomes
In the era of premium IOLs — toric, extended depth of focus, multifocal — patients arrive with high refractive expectations after cataract surgery. A mean absolute error (MAE) of 1.0 D was acceptable in 1990. Today's patients expect to be within 0.5 D of target in most cases.
Formula selection is one of the most controllable variables in achieving this. The right formula for the eye in front of you can reduce prediction error by 30–50% compared to a suboptimal choice.
A Brief History of IOL Calculation
1st generation formulas (theoretical): Fyodorov, Colenbrander — based on thin lens optics. Poor accuracy for outlier eyes.
2nd generation (regression-based): SRK (Sanders-Retzlaff-Kraff), first published 1980. Simple but systematically inaccurate for short and long eyes.
3rd generation: SRK/T, Holladay 1, Hoffer Q — improved regression formulas with anatomical modifications. Still in widespread use.
4th generation: Haigis (uses three constants), Holladay 2 — incorporate more biometric variables.
New generation: Barrett Universal II, Olsen, Kane, PEARL-DGS — use ray tracing or machine learning. Consistently outperform 3rd generation in large validation studies.
SRK/T: When It Still Works
SRK/T (Theoretical) remains the most widely installed formula globally and is adequate for:
- Average eyes: Axial length 22.0–24.5 mm, K-readings 42–46 D
- When only partial biometry is available (e.g. A-scan without IOLMaster data)
- Quick sanity check when comparing outputs from multiple formulas
Known weaknesses:
- Long eyes (>26 mm): systematically myopic prediction (over-powers the IOL) — use Kane or Barrett instead
- Short eyes (<21 mm): hyperopic shift — Hoffer Q or Holladay 2 typically outperform SRK/T
- Post-keratorefractive eyes: do not use SRK/T at all
Haigis: The Three-Constant Advantage
The Haigis formula uses three lens constants (a0, a1, a2) rather than a single A-constant, allowing for optimisation of the effective lens position (ELP) across a range of anterior chamber depths (ACD).
Best suited for:
- Eyes where ACD is an outlier (very shallow or very deep anterior chamber)
- Clinics with a well-established dataset for Haigis constant optimisation
- Sulcus-fixated IOLs (formula adapts to changed ELP)
Requirement: Accurate ACD measurement. Haigis performs poorly when ACD data is from A-scan (contact method) rather than non-contact optical biometry.
Post-refractive surgery: The Haigis-L formula was specifically developed for LASIK/PRK eyes and accounts for the underestimation of corneal power by standard keratometry in these patients.
Barrett Universal II: The Current Gold Standard
Multiple head-to-head studies (including the large ESCRS registry study) have placed Barrett Universal II (BU II) at or near the top of formula rankings across the full range of axial lengths.
Strengths:
- Performs consistently well across short, average, and long eyes
- Handles partial coherence interferometry and swept-source OCT biometry equally well
- The Barrett Toric Calculator is considered the most reliable toric IOL planning tool available
- Freely available at www.apacrs.org (web-based calculator)
Inputs required:
- Axial length, K1, K2, K axis, ACD, LT (lens thickness), WTW (white-to-white)
- All available from modern biometers (IOLMaster 700, Lenstar 900)
When BU II may not be optimal:
- Post-keratorefractive corneas — use Barrett True-K instead (a specific version of Barrett for post-LASIK/PRK)
- Very high astigmatism with irregular corneas — consider total keratometry (TK) data if available
Kane Formula: The Machine Learning Challenger
The Kane formula uses a combination of theoretical optics and machine learning, trained on over 30,000 cases. In multiple independent validations (including the Kane original publication and subsequent studies), it performs at least as well as Barrett Universal II and often outperforms it in extreme axial lengths.
Particularly strong for:
- Long eyes (>26 mm): Kane shows lower mean absolute error than most formulas in published studies
- Short eyes (<21 mm): comparable to Hoffer Q and Holladay 2
Input: Available as an app and web calculator. Requires sex (affects prediction) in addition to standard biometric inputs.
Practical Decision Framework
| Eye type | Recommended formula(s) |
|---|---|
| Normal axial length (22–24.5 mm) | Barrett Universal II, Kane, SRK/T |
| Long eye (>24.5 mm, especially >26 mm) | Barrett Universal II, Kane |
| Short eye (<22 mm) | Hoffer Q, Holladay 2, Barrett Universal II |
| Deep anterior chamber (>3.5 mm) | Haigis (optimised constants), Barrett UII |
| Shallow anterior chamber (<2.5 mm) | Hoffer Q, Holladay 2 |
| Post-LASIK/PRK | Barrett True-K, Haigis-L, PEARL-DGS |
| Post-RK | Masket method, Barrett True-K (with history) |
| Toric IOL planning | Barrett Toric Calculator |
Practical workflow recommendation: Run at least three formulas for each case and compare outputs. For most eyes, Barrett UII should be one of the three. If all three agree within 0.5 D, proceed with confidence. If there is significant disagreement, investigate the biometry data for measurement error before proceeding.
Documenting IOL Biometry in Clinit
Clinit's optical module includes a dedicated IOL biometry calculator that:
- Accepts input data from IOLMaster 700 (CSV export) or manual entry
- Calculates target refraction across SRK/T, Haigis, Barrett Universal II, and Hoffer Q simultaneously
- Displays formula outputs side by side for quick comparison
- Records the surgeon's selected formula, lens model, targeted refraction, and actual postoperative outcome
- Generates a biometry report for the surgical record and insurance documentation
- Tracks prediction error over time by surgeon and formula for internal audit and A-constant optimisation
Consistent postoperative outcome recording allows your clinic to optimise A-constants and lens constants progressively, improving prediction accuracy with every case you document.
