Discrepancies between Intraocular LensImplant Power Prediction Formulasin Pediatric Patients

Maya Eibschitz-Tsimhoni, MD,1 Omer Tsimhoni, PhD,2 Steven M. Archer, MD,1 Monte A. Del Monte, MD1

Purpose: The SRK II, SRK/T, Holladay I, and Hoffer Q intraocular lens power prediction formulas have beenclaimed to be interchangeable in their predicted postoperative refractive outcome among pediatric patients. Inthis study, we evaluated this clinical perception.

Design: Mathematical analysis.Methods: Analytical prediction of implant power using keratometry values up to 55 diopters and axial length

values as short as 16 mm was performed for 2 different refractive goals using the optimized intraocular lensconstants for the SRK II, SRK/T, Holladay I, Hoffer Q, and Haigis formulas. Comparison graphs for the predictedimplant power of each formula were constructed and differences between predicted results of the formulas wereplotted.

Main Outcome Measure: Predicted implant power.Results: Significant differences in intraocular lens power prediction were found among the Hoffer Q,

Holladay I, and SRK II formulas in the pediatric range of axial length and keratometry values. The Holladay I andHaigis formulas were found to be similar in their intraocular lens power prediction. The SRK/T was comparablewith the Holladay I and Haigis formulas, but still differed in the high keratometry values.

Conclusions: This analysis demonstrates differences in the intraocular lens power prediction among com-monly used formulas for axial length and keratometry values in the pediatric range. It is unclear under whatcircumstances each of these formulas may be preferred in the pediatric population. Ophthalmology 2007;114:

383–386 © 2007 by the American Academy of Ophthalmology.

Several formulas can be used to predict the intraocular lens(IOL) power needed to achieve a desired refractive goal aftercataract surgery.1 Because all formulas were derived modeledon data from adult eyes, it is as yet unclear whether they can beapplied in children with the same degree of accuracy. Previousreports did not find a significant difference between formulasin their prediction of postoperative refraction, including SRKII, Hoffer Q, Holladay I, and SRK/T, when used in children.2,3

Since these reports were published, the acceptable age for IOLimplantation has become progressively younger.4 However,no standard guidelines for the IOL power predictionformulas have been established. In this study, we reportour analysis of the different IOL prediction formulascommonly used in children.

Originally received: December 2, 2005.Accepted: June 28, 2006. Manuscript no. 2005-1174.1 Department of Ophthalmology and Visual Sciences, Kellogg Eye Center,University of Michigan, Ann Arbor, Michigan.2 Department of Industrial and Operations Engineering, University ofMichigan, Ann Arbor, Michigan.

Supported by the Wayne and Joan Webber Foundation, Clinton, Michigan.

The authors have no conflicts of interest regarding the article.

Correspondence to Maya Eibschitz-Tsimhoni, MD, Department of Oph-thalmology and Visual Sciences, Kellogg Eye Center, University of Mich-

igan, 1000 Wall Street, Ann Arbor, MI 48105. E-mail: mayae@umich.edu.

© 2007 by the American Academy of OphthalmologyPublished by Elsevier Inc.

Materials and Methods

An analytical prediction of IOL power over a wide range ofkeratometry (K) and axial length (AL) values was performed usingthe SRK II, SRK/T, Holladay I, Hoffer Q, and Haigis equations.5–9

A spreadsheet was generated with all the possible combinations ofK values from 40 to 55 diopters (D) and AL values from 16 to 28mm. For each combination (e.g., K � 45 D, AL � 19 mm), an IOLprediction was made with each equation for desired refractiveoutcomes of plano and �6.00 D. The equations were implementedas functions in a spreadsheet program (Microsoft Excel VBA;Microsoft, Redmond, WA) with AL and K as input parameters andpredicted IOL power as output. The optimized IOL constants forthe SA60AT lens (Alcon Laboratories, Fort Worth, TX) with anominal A constant of 118.4 were used.10 The results of thesecalculations were confirmed to match the analytical definition ofthe equations5–9 and were validated against results from the IOL-Master (Carl-Zeiss Meditec, Dublin, CA) over a wide range ofvalues. Pairwise comparisons were made among the 5 equationsfor a desired refractive outcome of plano and �6.00 D. Contourplots of the comparisons then were made for a refractive outcomeof plano.

Results

The contour plots illustrate differences between the predicted IOLpowers for a postoperative target refraction of plano using the SRKII, SRK/T, Holladay I, Hoffer Q, and Haigis equations (Fig 1). The

contour plots show how the prediction formulas diverge as a

383ISSN 0161-6420/07/$–see front matterdoi:10.1016/j.ophtha.2006.06.063

384

Eibschitz-Tsimhoni et al � Discrepancies of IOL Power Formulas in Pediatric Patients

function of the AL and K values. Significant differences in IOLpower prediction are found among the Hoffer Q, Holladay I, andSRK II formulas in the pediatric range of axial length and kera-tometry values (Fig 1A–D). The Holladay and Haigis formulaswere found to be similar in their IOL prediction (Fig 1E). TheSRK/T was comparable with the Holladay and Haigis formulas,but still differed in the high K values (Fig 1F). The differencesbetween the formulas were found to be similar when a targetrefraction of �6.00 D was chosen.

Discussion

The placement of an IOL in children undergoing cataractsurgery is gaining wider acceptance.4,11–13 In addition, withimproved surgical equipment and techniques, the accept-able age for IOL implantation is becoming progressivelyyounger.4 With the trend toward implanting IOLs in infantswith shorter AL values and higher K values, there is a needto understand the accuracy and the differences betweenprediction formulas at lower extremes of ALs and at thehigher extremes of Ks. As shown by Gordon and Donzis,14

the average AL is 16.8�0.6 mm in normal newborns,19.2�0.7 mm in the 0- to 1-year age group, 20.2�0.3 mmin the 1- to 2-year age group, and 21.4�0.1 mm in the 2- to3-year age group. The average K value is 51.2�1.1 D innormal newborns, 45.2�1.3 D in the 0- to 1-year age group,44.9�0.9 D in the 1- to 2-year age group, and 44.1�0.3 Din the 2- to 3-year age group. Flitcroft et al15 reported thatthe mean AL in eyes with congenital cataracts was18.57 mm, with a range of 16.44 to 20.00 mm at a mean ageat surgery of 24 weeks (range, 4–94 weeks). The meanK value before surgery for the congenital cataract group inthis study was 47.78 D, with a range of 43.55 to 57 D. Thesefindings are shown in Figure 2.

Previous reports suggest that there is no important dif-ference among the SRK II, Hoffer Q, Holladay I, andSRK/T equations when used in children. However, thesereports most likely were not adequately powered to findmeaningful differences in the short axial length group, ifthey exist. Andreo et al3 stated that there was little differ-ence between SRK II, SRK/T, Holladay I, and Hoffer Qformulas in short, medium, and long eyes in providingadequate predicted refraction. The mean errors noted in thatstudy were between 1.23 to 1.33 D in long eyes, 0.98 to 1.03D in medium eyes, and 1.41 to 1.8 D in short eyes. But onlythe mean of a small number of patients (n � 17) with ALsless than 22.0 mm (range, 18.6–22.0 mm) was evaluated inthe group with short eyes, perhaps obscuring differences atshorter ALs.

Mezer et al2 found that prediction formulas, includingHoffer Q, Holladay I, SRK/T, SRK, and SRK II equations,provide similar outcomes in patients between 2 and 17 years

4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™Figure 1. Intraocular lens power prediction comparison for an emmetropHaigis equations. Equal difference points are connected by lines and labela nominal A constant of 118.4 were used: a0 � 0.357, a1 � 0.261, a2 �A � 118.7 for SRK/T, A � 119.0 for SRK II.10 The difference between

marked on the plot.14,15 D � diopters.

of age. All formulas were found to be similarly unsatisfac-tory (as defined by Mezer et al) in achieving the targetrefraction. The average differences between predicted andactual postoperative refraction using mean errors rangedfrom 1.06�0.79 D to 1.79�1.47 D. However, only themean error for all patients was reported. Patients were notsubgrouped, and differences as a function of ALs andK values were not defined.

Neely et al16 also found that the SRK II, SRK/T, andHolladay I formulas had no significant difference in lenspower predictability in pediatric patients and that the Hoffer Qformula showed a positive bias of 1.2 D. The authors alsoshowed that the SRK II, SRK/T, Holladay I, and Hoffer Qformulas were increasingly less accurate at predicting theneeded IOL implant power at smaller axial lengths.

Only a small number of adult patients with short ALshave been studied and reported in the literature. In a studyby Hoffer7 of 500 eyes, only 36 eyes had ALs less than22 mm, and the average AL was relatively large (by pedi-atric standards), 21.43�0.69 mm. Apart from the smallnumbers reported, the different characteristics of these short

™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™ractive goal (Rs � 0) for the SRK II, SRK/T, Holladay I, Hoffer Q, andhe optimized intraocular lens constants for the Alcon SA60AT lens with53, personalized anterior chamber depth � 5.37, surgeon factor � 1.60,ulas’ predictions at a few typical axial length and keratometry values are

Figure 2. Typical pediatric axial lengths and keratometry values. Meanand 1 standard deviation shown except for congenital cataract eyes inwhich the mean and range of 35 eyes are shown (derived from datareported in Gordon and Donzis14 and Flitcroft et al15). D � diopters.

™™™ic refed. T

0.1form

385

Ophthalmology Volume 114, Number 2, February 2007

adult eyes may make it difficult to extrapolate conclusionsto the pediatric population.

Our mathematical comparison of the predicted IOLpower among the SRK II, SRK/T, Holladay I, Hoffer Q, andHaigis formulas over a range of ALs from 16 to 28 mm andover a range of K values from 40 to 55 D for differentrefractive goals (plano and �6.00 D) demonstrates that onlya narrow range, more typical of older children and adulteyes, is insensitive to the formula chosen. Addressing thisquestion analytically gives the advantage of studying everyspecific AL and K value combination without having hadenough patients with this specific value combination. Thiswould be especially difficult in infants and newborns. Inaddition, clinical studies introduce other factors such as ALand K value measurement errors, surgeon factors, and re-fractive changes of the eye with time that may cause prob-lems with interpreting the results. Currently, the pediatricophthalmologist must rely on these IOL power predictionformulas drawn largely from studies on adult eyes. Whichformula to use in infants and children and when to use itremain matters of conjecture.

References

1. Eibschitz-Tsimhoni M, Archer SM, Del Monte MA. Intraoc-ular 1ens power calculation in children. Surv Ophthalmol. Inpress.

2. Mezer E, Rootman DS, Abdolell M, Levin AV. Early post-operative refractive outcomes of pediatric intraocular lensimplantation. J Cataract Refract Surg 2004;30:603–10.

3. Andreo LK, Wilson ME, Saunders RA. Predictive value ofregression and theoretical IOL formulas in pediatric intraoc-ular lens implantation. J Pediatr Ophthalmol Strabismus 1997;34:240–3.

4. Lambert SR, Lynn M, Drews-Botsch C, et al. Intraocular lens

implantation during infancy: perceptions of parents and the

386

American Association for Pediatric Ophthalmology andStrabismus members. J AAPOS 2003;7:400–5.

5. Sanders DR, Retzlaff J, Kraff MC. Comparison of the SRK IIformula and the other second generation formulas. J CataractRefract Surg 1988;14:136–41.

6. Holladay JT, Prager TC, Chandler TY, et al. A three-partsystem for refining intraocular lens power calculations. J Cat-aract Refract Surg 1988;14:17–24.

7. Hoffer KJ. The Hoffer Q formula: a comparison of theoreticaland regression formulas. J Cataract Refract Surg 1993;19:700–12.

8. Retzlaff J, Sanders DR, Kraff MC. Development of the SRK/Tintraocular lens implant power calculation formula. J CataractRefract Surg 1990;16:333–40.

9. University-Augenklinik Würzburg. Department of Echogra-phy. IOL calculation according to HAIGIS. Available at:http://www.augenklinik.uni-wuerzburg.de/uslab/ioltxt/haie.htm. Accessed June 4, 2006.

10. University-Augenklinik Würzburg. ULIB: user group for laserinterference biometry. Optimized IOL constants for the ZEISSIOL Master calculated from patient data on file (as of May 23,2006).Availableat:http://www.augenklinik.uni-wuerzburg.de/ulib/c1.htm. Accessed June 4, 2006.

11. Wilson ME. Intraocular lens implantation: has it become thestandard of care for children? Ophthalmology 1996;103:1719–20.

12. Wilson ME, Bluestein EC, Wang XH. Current trends in theuse of intraocular lenses in children. J Cataract Refract Surg1994;20:579–83.

13. Superstein R, Archer SM, Del Monte MA. Minimal myopicshift in pseudophakia versus aphakic pediatric cataractpatients. J AAPOS 2002;6:271– 6.

14. Gordon RA, Donzis PB. Refractive development of the humaneye. Arch Ophthalmol 1985;103:785–9.

15. Flitcroft DI, Knight-Nanan D, Bowell R, et al. Intraocularlenses in children: changes in axial length, corneal curvature,and refraction. Br J Ophthalmol 1999;83:265–9.

16. Neely DE, Plager DA, Borger BA, Golub RL. Accuracy ofintraocular lens calculations in infants and children undergo-

ing cataract surgery. J AAPOS 2005;9:160–5.