Pediatric intraocular lens (IOL) implantation has become routine at many medical centers. Studies show a long-term myopic shift in children with pseudophakia, particularly in those younger than 2 years but also in older children in the first decade.1–6 Using these follow-up data, surgeons have tried to determine which IOL power to use in a child to achieve the desired long-term postoperative refraction. However, these studies ultimately rely on the surgeon's ability to select an IOL that achieves the initial postoperative desired refraction. If the target is not achieved, the outcome may be poor and the long-term data not applicable. Studies show that IOL power calculation in children has an error range of 2.5 to 5.0 diopters (D).1,7,8 Other studies suggest that newer theoretical formulas, such as the Holladay 1,9 Hoffer Q,10 and SRK/T,11 predict postoperative refraction more accurately than older regression formulas such as the SRK and SRK II.9,10,12
We compared 5 IOL calculation formulas to determine which better predicts the postoperative refractive outcome in pseudophakic children.
Patients and Methods
The charts of 158 consecutive patients 2 to 17 years old operated on by 1 of 2 staff surgeons (D.S.R., A.V.L.) were retrospectively reviewed. The cohort comprised 206 cataract extractions with primary or secondary posterior chamber IOL implantation performed between May 1992 and April 2000 at the Hospital for Sick Children. Posterior capsulotomy and anterior vitrectomy were routinely performed in every patient. Optic capture in the capsular bag was performed in 37 cases. In 9 cases, the optic was placed in the sulcus. This was taken into account in the IOL calculations.
Data were collected from 2 subsets of patients: those with available data 2 to 3 months after surgery and those with available data 2 to 6 months after surgery. The measured outcome was the actual refraction at 2 to 3 months and 2 to 6 months postoperatively versus the desired refraction.
Excluded were patients with corneal scarring involving the visual axis, intraocular congenital anomalies precluding overrefraction (eg, persistent hyperplastic primary vitreous), and penetrating keratoplasty (combined with cataract extraction or before 2 to 3 months and 2 to 6 months postoperatively) and patients without preoperative axial length measurements, preoperative keratometry measurements, or a follow-up refraction 2 to 3 months and 2 to 6 months postoperatively. All types of cataracts were included in patients who were in the study.
Parameters taken from the surgeon's office chart or the clinic hospital medical record included sex, date of birth, surgery date, IOL A-constant provided by the manufacturer, IOL power, preoperative axial length, corneal curvature, IOL position, postoperative time at refraction, and postoperative refraction.
Axial length measurements were done by a skilled technician or by the surgeon in his office, in some cases with general anesthesia, using standard contact A-scan biometry techniques. The technician used a B-scan-S standardized A-scan (Biovision, B.V. International) at a speed of 1641 m/second for the cornea, 1532 m/second for the anterior chamber, 1641 m/second for the lens, and 1532 m/second for the vitreous. In the clinic, 1 surgeon (D.S.R.) used an I3SYSTEM-ABD ophthalmic ultrasound method (Innovative Imaging Inc.) with 4-gate measurements and combined velocities of 1532 m/second and 1641 m/second in phakic eyes and the other surgeon (A.V.L.), an ultrasound biometer (model 810, Zeiss Humphrey Instruments) with an average speed of 1555 m/second. Both surgeons used the latter machine in the operating theater. Repeated measurements were taken until 3 to 5 measurements or 10 measurements were obtained with sharp retinal spikes. When 5 or fewer measurements were obtained, the mean was used. With the 10-measurement technique, the high and low values were discarded before averaging. This method reduces possible inaccuracies of contact A-scan compared to the immersion technique due to corneal compression.13
For older cooperative children, the technician or surgeon measured the corneal curvature with the patient awake using a table-mounted keratometer (Bausch & Lomb, Inc.). Corneal measurements in younger children were taken using general anesthesia and a manual floor-mounted keratometer (Bausch & Lomb) or automated keratometer (Retinomax K-plus, Nikon Inc.). No subgroup analysis was possible because of the small number of cases.
All patients received a single-piece poly(methyl methacrylate) IOL (6693B, SP13UB, or SP24UB, Chiron Vision). The IOL was placed through a 2.0 to 3.0 mm long and 6.0 to 7.0 mm wide scleral tunnel incision that was closed with 10-0 polyglactin or nylon sutures. No sutures were cut postoperatively as cylinder was not a significant problem in any patient. Postoperative refraction was obtained by retinoscopy by the surgeon or by the community ophthalmologist or optometrist to whom the patient had been discharged. For data analysis, postoperative refraction was converted to a spherical equivalent.
Five IOL power formulas, 2 regression (SRK and SRK II) and 3 theoretical (Holladay 1, Hoffer Q, and SRK/T) were used to predict refractive outcome by the implanted IOL power. The manufacturer's A-constant was used with the SRK, SRK II, and SRK/T formulas. This was not done with the other 2 formulas as both use a personalized anterior chamber depth (ACD) factor. The Holladay formula calculates a personalized ACD factor that is the sum of the corneal height, the thickness of the cornea, and the distance from the iris plane to the IOL's principal plane. This last value is termed the surgeon factor (SF). Since the SF cannot be known before surgery, Holladay et al.9 calculated it from a series of postoperative eyes with 1 IOL style and constructed empirical conversion formulas to correspond between the A-constant and the SF and ACD. The Hoffer Q formula predicts ACD based on a personalized ACD for any lens style as well as the axial length and keratometry reading of the individual eye without using Fyodorov and coauthors' method,14 in which the axial length and K-reading are used to predict the corneal height formula (used by the Holladay and SRK/T but not by the Hoffer Q). The SRK and SRK II formulas were calculated using Excel 2000 spreadsheets (Microsoft Corp.) based on the published SRK equation12 and SRK II modifications.10 The Holladay 1, Hoffer Q, and SRK/T formulas were calculated using the Holladay IOL Consultant program (version 2.20, Holladay Consultant, Inc.). This program was used to calculate the SF for the Holladay formula and the ACD for the Hoffer Q formula.9 The IOL position in the sulcus or in the bag was a parameter in Holladay IOL Consultant calculations.
The target refraction (2- to 3-month follow-up/2- to 6-month follow-up) in each case was selected preoperatively by the surgeon. Target refractions ranged from −2.59 to 1.74 D/2.69 to 6.85 D (mean −0.01 ± 0.73 D/0.35 ± 1.60 D; median −0.11 D/0.01 D), seeking a short-term postoperative refraction that would best lead to long-term suitability based on the literature, refraction in the other eye, personal experience of the surgeon, outcome in the other eye, and other factors. The postoperative refraction was measured at a vertex distance of 15.0 mm (usual spectacle position). The prediction error was calculated by subtracting the actual postoperative refraction at 2 to 3 months and 2 to 6 months from the target refraction with each of the 5 formulas. The McClatchey logarithmic formula, which predicts the age-related myopic shift in the pseudophakic eye,2,3,15 was taken into account in determining the target refraction.
The intraclass correlation coefficient (ICC) was used to evaluate the level of chance corrected agreement between the target and actual postoperative refractions. An ICC of 1 depicts full agreement. No agreement between the target and actual postoperative refractions produces a value of 0 (Table 1).
The Ethics Board of the Hospital for Sick Children Research approved the study. All patients or their caretakers signed informed consent in person or by mail, which allowed the investigators to approach the patient's community physicians for follow-up data. Data from these physicians were obtained by mail.
Results are presented as follows: 2- to 3-month follow-up/2- to 6-month follow-up.
The study evaluated 31/49 patients (19/34 boys) who had 34/59 IOL implantations and for whom all necessary data were available. Data for 127/109 patients (172/147 IOL implantations) were lacking because preoperative data for 36 patients (42 IOL implantations) were missing and follow-up refractions for 91/102 patients in the desired postoperative time window (130/140 IOL implantations) could not be found. Thirty patients (36 IOL implantations) lacked both preoperative and postoperative data. In 1 patient with traumatic corneal laceration, the fellow eye was measured and the results were used to choose the IOL power for the operated eye. The age of the study group ranged from 18 months to 17 years 8 months/22 months to 18 years (median 6 years 5 months/7 years 5 months). A total of 29/50 IOLs were bag fixated, and 5/9 IOLs were placed in the sulcus. Separate statistical analysis was not possible due to the disparity in sample sizes. The postoperative refraction was performed a mean of 2 months 6 days/2.5 months (range 2 months 1 day to 2 months 29 days/2 months 1 day to 5 months 15 days).
Thirty-eight keratometry and biometry ultrasounds were done by the surgeon (A.V.L., D.S.R.) using general anesthesia, 12 measurements were done preoperatively by the technician with the patient awake, and 5 were done by the surgeon (D.S.R.) in the office with the patient awake. In 4 cases, it could not be determined where the measurements had taken place. Subanalysis of the different instruments used was not possible because of the small number of patients who had biometry while awake. The median axial length was 23.06 mm/22.00 mm (range 19.79 to 25.41 mm/19.24 to 26.69 mm). Plotting the age as an independent variable against the axial length produced a logarithmic curve (Figure 1). The mean corneal power was 43.54 ± 1.72 D/43.99 ± 1.96 D (range 39.74 to 46.68 D/38.50 to 49.25 D).
On average, each theoretic formula missed the target refraction by approximately 1.1 D/1.3 D (Table 2). The regression formulas (SRK and SRK II) performed slightly worse. The range of error was large. In some cases, the target refraction was missed by 3.5 to 5.5 D/6.0 to 9.0 D (Table 2). A prediction error greater than 2.0 D was observed in 9% to 18%/23% to 39% of operated eyes. The ICC indicated poor to fair agreement between the target refraction and actual postoperative refraction.
We compared 5 commonly used IOL calculation formulas in terms of their ability to predict refractive outcomes in pediatric IOL implantation. The measured outcome was the actual refraction 2 to 3 months and 2 to 6 months postoperatively versus the target refraction. We chose this time interval as the eye has healed and long-term effects have not yet taken place.
There are 4 published studies of the refractive outcome after pediatric cataract extraction and IOL implantation (Table 3A), one of which is by Arffa and coauthors.7 They used the SRK formula, not more advanced formulas such as the SRK II or SRK/T. We thought it would be of interest to compare the predictive error in our study and that in the study by Arffa and coauthors to observe the trends of predictive errors in 2 studies using the same formula. Furthermore, we wanted to determine whether the SRK II formula provides better predictions than the SRK not only in adults, for whom it was designed, but also in children and toddlers having lensectomy or cataract extraction.
Using the ICC method, we found all 5 formulas to be at best only fair to good at successfully achieving the target refraction. Although the difference was probably not clinically significant, the theoretical formulas performed slightly better than the regression formulas. The mean absolute prediction error in our study was similar to that in other series of pediatric eyes.1,7,16,17 The range was wider in our group, although it was hard to compare our data with those in other studies that used different target time intervals and generalizations in the use of certain parameters for IOL power calculation (Table 3B). We believe the range in the prediction error was too high to be clinically acceptable.
Why do these formulas appear to be less accurate for pediatric eyes despite long records of success in adults? Several authors have devised a logarithmic formula to predict the myopic shift in pseudophakic eyes from infancy to adulthood.2,3,15 Six months after surgery, little shift should occur and the refraction should remain close to the surgical target (range −0.33 to −0.01 D, mean −0.07 ± 0.02 D, median −0.05 ± 0.02 D). Instead, we observed a much larger range of refractive errors at 2 to 3 months (range −2.59 to 1.74 D, mean −0.01 ± 0.73 D, median −0.11 D) and even more at 6 months (range 2.69 to 6.85 D, mean 0.35 ± 1.60 D, median 0.01 D). The change between 2 months and 6 months was toward greater hyperopic errors. This suggests that the magnitude and direction of the errors we observed were caused by the expected myopic shift described by McClatchey and Parks.2 Between 2 months and 6 months postoperatively, factors such as continued molding of the soft pediatric sclera and progressive posterior capsule fibrosis may play a role.18 In fact, as we relied in part on the data of McClatchey and Parks to set our targets, any shift over time should have been in keeping with their predictions. Although their data can be helpful in predicting the correct surgical target refraction, they are only useful if the target is achieved.
We were concerned that inaccuracy might occur because pediatric eyes are shorter. However, our cohort, with a youngest age of 2 years at surgery, had axial lengths in the lower range of normal adult eyes. Some children even had longer eyes than the average in adults. Our data were similar to previously reported normative data.19 Larsen19 measured the sagittal growth of the eye from birth to puberty in 1852 eyes using ultrasound. Plotting age as an independent variable against the axial length in our 59 patients produced a logarithmic graph that demonstrated similar growth to the growth curve for the length of the optic axis in Larsen's study. As a result, we do not believe that a shorter axial length was a significant factor in IOL formula inaccuracy in our population. Unfortunately, our small sample size yielded axial length subgroup analyses with an unacceptably low power. There may be more accurate ways to perform axial length measurements such as partial coherence interferometry.20,21 The speed and the noninvasive nature of these noncontact methods might make them preferable in the pediatric population in the future.
Do other factors influence the accuracy of the newer theoretical formulas? The ACD is the sum of the corneal thickness, corneal height, and another value. Corneal height is calculated using axial length and keratometry readings. In the Holladay and SRK/T formulas, the distance from the iris plane to the IOL principal plane is called the SF, while the Hoffer Q formula uses only the ACD.9–11 The mean and range of the corneal curvature in our patients (43.99 ± 1.96 D and 38.50 to 49.25 D, respectively) were also close to the normal values in adults reported in the literature (mean 43.81 ± 1.60 D22; mean 43.83 ± 1.56 D, range 39.38 to 43.37 D10). This is not surprising as the mean corneal curvature in babies rapidly reaches that of adults (mean 48.4 ± 1.7 D in neonates, mean 45.9 ± 2.3 D at 1 month, mean 42.9 ± 1.3 D at 36 months).23 The mean age of our patients (7 ± 5 years) was well into the age at which corneal curvature reaches that in an adult eye. Therefore, we do not believe that corneal curvatures played a role in the inaccuracy of the IOL formulas in our study.
Corneal thickness in children 2 years of age is also equivalent to adult readings of 0.52 mm.24 As our youngest patient was 1 year 11 months old, we do not think corneal thickness in children played a role in the inaccuracy of the IOL formulas in our study.
The distance from the iris plane to the IOL's principal plane (SF) was calculated from a series of postoperative adult eyes. Pseudoaccommodation amplitude in pediatric pseudophakic patients can cause a mean anterior IOL movement of 0.42 mm at near,25 perhaps because of the increased scleral elasticity in children. Residual capsule fibrosis is more aggressive in children than in adults and can occur in 51% to 96% of cases if no posterior capsulotomy is done at the time of surgery. Both factors could affect the IOL's position in the eye, perhaps making it less predictable than in adults. This would have an adverse affect on all the formulas.
Another source of error might be difficulty obtaining axial length and corneal curvature measurements in uncooperative children, although all our measurements were done by an experienced technician (21%) or surgeon (5%) on compliant awake children or by the surgeons with the child under general anesthesia (67%). If the technician was concerned about accuracy, he indicated it on the report and the surgeon would remeasure with the patient under anesthesia. The surgeons used measurements taken under anesthesia when there was a concern about the accuracy or unfeasibility of data taken with the patient awake. Most of our patients were measured under general anesthesia, reducing the impact of a lack of cooperation on measurement accuracy. As most measurements were done under optimal conditions in the operating theater by the experienced surgeons (no trainees performed the biometry), we believe our averaging of best measurements was accurate.
Postoperative refraction in children can be difficult. Many ophthalmologists performed our postoperative refractions as some children were under community care 2 to 6 months postoperatively.
Biometry can be done by contact or immersion techniques.26 We used only the contact technique to measure axial length. Corneal compression is the most common cause of a shortened axial length measurement.27 For example, a 1.0 mm error in axial length can result in a 2.5 to 3.0 D error in postoperative refraction.28 In infants and children, the cornea and sclera are more elastic and distensible than in adults.29 Elastic fibers in the sclera can be stretched beyond their limit of elasticity when not supported by a fully developed collagen structure.30 The entire globe can therefore be distended until approximately 3 years after birth.30 Corneal compression during axial length biometry might, therefore, be more common and greater than in adults. In a more elastic pediatric eye, the propensity and magnitude of this measurement error may increase. Using this method in 20 eyes that were shorter than 22.0 mm might have decreased corneal compression.
The axial length of the globe is indirectly recorded by multiplying the assumed velocity and the time it takes the sound wave to reflect back to the ultrasound machine from various structures in the eye, such as the anterior capsule of the lens, posterior capsule of the lens, and retina.31,32 It relies on the assumed speed of sound within the aqueous, lens, and vitreous.33 Hoffer34 measured each component of the axial length individually, with the A-scan ultrasound reset for the particular speed of ultrasound for each ocular tissue in the path of the ultrasound beam. When compared with a single measurement using an “average” tissue velocity for the entire eye, however, the more complex method was less accurate. Some of our biometry units use an average sound wave velocity to determine the axial length. This assumption may not be accurate in the pediatric eye, in which a relatively greater intraocular volume is occupied by the crystalline lens19 and in which tissue viscosities may be different than in adults.35 Inaccurate axial length measurements contribute to prediction error.
One weakness of our study was the small number of children. Even so, we do not believe there was selection bias because patients were excluded for missing data and no other reason. Subgroup analysis would be interesting, but we did not have sufficient statistical power to draw conclusions about the subgroups.
In summary, despite efforts to assist surgeons in selecting the appropriate targets for postoperative refraction in pediatric IOL implantation to achieve the desired long-term refractive outcome, our ability to achieve that target may be unsatisfactory. Factors that may contribute to this inaccuracy include axial length errors, especially contact ultrasound; K-reading errors; pseudoaccommodation; capsule fibrosis; measurement errors in children; corneoscleral elasticity; and differences in media viscosity compared to the normal adult values. In children younger than 2 years, axial length, corneal curvature, and corneal thickness differences from those in adults may make the inaccuracy even greater. If the target is not achieved, the effects of long-term expected refractive change with the child's growth might be amplified to create even greater adverse outcomes. Further prospective research is needed to improve our ability to achieve postoperative refractive targets.
We intend to gather data to better analyze which newer formulas will improve our ability to achieve more accurate prediction of postoperative refractive errors. Until then, it may be important that the process of informed consent for IOL implantation include discussion of the likelihood of undesirable refractive outcomes, in which case glasses or contact lenses may be required to achieve the desired outcome. As we move from the excellent refractive control of aphakic contact lenses toward the continued expansion of IOL implantation in pediatric eyes, this becomes increasingly important.
1. Andreo LK, Wilson ME, Saunders RA. Predictive value of regression and theoretical IOL formulas in pediatric intraocular lens implantation. J Pediatr Ophthalmol Strabismus 1997; 34:240-243
2. McClatchey SK, Parks MM. Theoretic refractive changes after lens implantation in childhood. Ophthalmology 1997; 104:1744-1751
3. McClatchey SK, Dahan E, Maselli E, et al. A comparison of the rate of refractive growth in pediatric aphakic and pseudophakic eyes. Ophthalmology 2000; 107:118-122
4. Spierer A, Desatnik H, Blumenthal M. Refractive status in children after long-term follow up of cataract surgery with intraocular lens implantation. J Pediatr Ophthalmol Strabismus 1999; 36:25-29
5. Dahan E, Drusedau MUH. Choice of lens and dioptric power in pediatric pseudophakia. J Cataract Refract Surg 1997; 23:618-623
6. Enyedi LB, Peterseim MW, Freedman SF, Buckley EG. Refractive changes after pediatric intraocular lens implantation. Am J Ophthalmol 1998; 126:772-781
7. Arffa RC, Donzis PB, Morgan KS, Zhou YJ. Prediction of aphakic refractive error in children. Ophthalmic Surg 1987; 18:581-584
8. Lesueur L, Chapotot E, Arne JL, et al. La prédictabilité de l'amblyopie chez l'enfant amétrope children. Apropos de 96 cas. J Fr Ophtalmol 1998; 21:415-424
9. Holladay JT, Prager TC, Chandler TY, et al. A three-part system for refining intraocular lens power calculations. J Cataract Refract Surg 1988; 14:17-24
10. Hoffer KJ. The Hoffer Q formula: a comparison of theoretic and regression formulas. J Cataract Refract Surg 1993; 19:700-712; errata, 1994; 20:677
11. Retzlaff JA, Sanders DR, Kraff MC. Development of the SRK/T intraocular lens implant power calculation formula. J Cataract Refract Surg 1990; 16:333-340; correction, 528
12. Sanders DR, Retzlaff JA, Kraff MC, et al. Comparison of the SRK/T formula and other theoretical and regression formulas. J Cataract Refract Surg 1990; 16:341-346
13. Shammas HJ. A comparison of immersion and contact techniques for axial length measurement. Am Intra-Ocular Implant Soc J 1984; 10:444-447
14. Fedorov SN, Kolinko AI, Kolinko I. Estimation of optical power of the intraocular lens [Russian]. Vestn Oftalmol 1967; 80:27-31
15. McClatchey SK. Intraocular lens calculator for childhood cataract. J Cataract Refract Surg 1998; 24:1125-1129
16. Lesueur L, Arne J-L, Chapotot E. La prédictabilité du calcul de la puissance de l'implant dans le traitement de la cataracte chez l'enfant. J Fr Ophtalmol 1998; 22:209-212
17. Tromans C, Haigh PM, Biswas S, Lloyd IC. Accuracy of intraocular lens power calculation in paediatric cataract surgery. Br J Ophthalmol 2001; 85:939-941
18. Hosal BM, Biglan AW. Risk factors for secondary membrane formation after removal of pediatric cataract. J Cataract Refract Surg 2002; 28:302-309
19. Larsen JS. The sagittal growth of the eye. IV. Ultrasonic measurement of the axial length of the eye from birth to puberty. Acta Ophthalmol 1971; 49:873-886
20. Meyer F, Renard JP, Roux L, et al. Intérêt d'un nouveau biomètre non-contact pour le calcul de la puissance des lentilles intra-oculaires cristalliniennes. J Fr Ophtalmol 2001; 24:1060-1066
21. Verhulst E, Vrijghem JC. Accuracy of intraocular lens power calculations using the Zeiss IOL master. A prospective study. Bull Soc Belge Ophtalmol 2001; 281:61-65
22. Hoffer KJ. Biometry of 7,500 cataractous eyes. Am J Ophthalmol 1980; 90:360-368; correction, 890
23. Weale RA. A Biography of the Eye; Development, Growth, Age. London, HK Lewis, 1982
24. Ehlers N, Sørensen T, Bramsen T, Poulsen EH. Central corneal thickness in newborns and children. Acta Ophthalmol 1976; 54:285-290
25. Lesiewska-Junk H, Kaluzny J. Intraocular lens movement and accommodation in eyes of young patients. J Cataract Refract Surg 2000; 26:562-565
26. Byrne SF, Green RL. Ultrasound of the Eye and Orbit. St Louis, MO, Mosby-Year Book, 1992
27. Binkhorst RD. The accuracy of ultrasonic measurement of the axial length of the eye. Ophthalmic Surg 1981; 12:363-365
28. Hoffer KJ. Preoperative cataract evaluation. In: Caldwell DR, ed, Cataract; Transactions of the New Orleans Academy of Ophthalmology. New York, NY, Raven Press, 1988; 24
29. Waring GO III, Rodrigues MM. Congenital and neonatal corneal abnormalities. In: Tasman W, Jaeger EA, eds, Duane's Ophthalmology on CD-ROM. Philadelphia, PA, Lippincott Williams & Wilkins, 1999
30. Watson P. Diseases of the sclera and episclera. In: Tasman W, Jaeger EA, eds, Duane's Ophthalmology on CD-ROM. Philadelphia, PA, Lippincott Williams & Wilkins, 1999
31. Lowe RF. Time-amplitude ultrasonography for ocular biometry. Am J Ophthalmol 1968; 66:913-918
32. Lizzi FL, Burt WJ, Coleman DJ. Effects of ocular structures on propagation of ultrasound in the eye. Arch Ophthalmol 1970; 84:635-640
33. Ludwig GD. The velocity of sound through tissues and the acoustic impedance of tissue. J Acoust Soc Am 1950; 22:862-866
34. Hoffer KJ. Preoperative cataract evaluation: intraocular lens power calculation. Int Ophthalmol Clin 1982; 22(2):37-75
35. Berman ER, Michaelson IC. The chemical composition of the human vitreous body as related to age and myopia. Exp Eye Res 1964; 3:9-15