To assess the accuracy of IOL power calculation formulae in children less than 2 years of age.
Retrospective, comparative study, comprising of 128 eyes of 84 children.
We analyzed records of children less than 2 years with congenital cataract who underwent primary IOL implantation. Data were analyzed for prediction error using the 4 commonly used IOL power calculation formulae. We calculated the absolute prediction error with each of the formulae and the formula that gave least variability was determined. The formula that gave the best prediction error was determined.
Mean age at surgery was 11.7 ± 6.2 months. Absolute prediction error was found to be 2.27 ± 1.69 diopters (D) with SRK II, 3.23 ± 2.24 D with SRK T, 3.62 ± 2.42 D with Holladay, and 4.61 ± 3.12 D with Hoffer Q. The number of eyes with absolute prediction error within 0.5 D was 27 (21.1%) with SRK II, 8 (6.3%) with SRK T, 12 (9.4%) with Holladay, and 5 (3.9%) with Hoffer Q. Comparison between different formulae showed that the absolute prediction error with SRK II formula was significantly better than with other formulae ( P < .001). Prediction error with SRK II formula was not affected by any factor such as age ( P = .31), keratometry ( P = .32), and axial length ( P = .27) of the patient. Axial length influenced the absolute prediction error with Holladay ( P = .05) and Hoffer Q formulae ( P = .002). Mean keratometry influenced prediction error ( P = .03) with SRK T formula.
Although absolute prediction error tends to remain high with all present IOL power calculation formulae, SRK II was the most predictable formula in our series.
With the advances in microsurgical devices, instruments, high-molecular-weight viscoelastic agents, and intraocular lens (IOL) designs, the acceptance of primary intraocular lens implantation in children less than 2 years of age has gradually increased. The unique challenge faced by a pediatric cataract surgeon is that of increased uncertainty regarding IOL power calculation.
In children less than 2 years of age, eyes have shorter axial length (AL), steeper cornea with higher keratometry values (K), and shallower anterior chamber depth (ACD). These eyes undergo rapid elongation in axial length and increase in the dimension of cornea, depth of anterior chamber, and capsular bag. The cornea becomes less steep and there is a decline in the power of the crystalline lens and cornea during this period. This tends to result in a myopic shift postoperatively, if an emmetropic power IOL is implanted. Although undercorrection of the IOL power is routinely carried out, the accuracy of the final refractive outcome depends on the precision of the IOL power calculation. In the case of younger children, prediction error in IOL power is affected by the errors in the biometry under general anesthesia, position of the implanted IOL, and age-related undercorrection planned by the surgeon. However, axial length and keratometry measurements done under general anesthesia tend to be less precise than biometry in the supine position because of improper centration and fixation. Hence, selection of appropriately powered IOL becomes a complex issue.
In addition, the current formulae used for calculating IOL power may not be accurate in children with smaller eyes. A few studies have been conducted to resolve this challenge, but none have come up with a definitive conclusion and all have their own drawbacks. Furthermore, Moore and associates, in their study in 203 pediatric eyes, showed that prediction error tends to be greater in children less than 2 years of age. Since children of this age group are affected the most by the myopic shift, it is essential to derive the best IOL power calculation formula for them.
The objective of this study was to assess the predictability of desired postoperative refractive outcomes with the commonly used IOL power calculation formulae (Sanders-Retzlaff-Kraff II [SRK II], Sanders-Retzlaff-Kraff T [SRK T], Holladay I, and Hoffer Q formulae) in children less than 2 years of age who had undergone primary IOL implantation.
Materials and Methods
We retrospectively analyzed the records of all children less than 2 years of age who had undergone cataract surgery with primary IOL implantation at our institute between January 1, 2006 and December 2007. We included the patients with a minimum follow-up of 4 weeks. We excluded cases with history of ocular trauma, sulcus fixated IOLs, and evidence of any ocular or systemic abnormality.
Preoperatively, a detailed history and complete ocular examination was done for all children. Intraoperatively, a detailed examination under anesthesia (EUA) was performed. Intraocular pressure was measured with a Perkins MK2 hand held applanation tonometer (Hagg Streit UK Ltd, Harlow, Essex, England). Keratometry and axial length were obtained under general anesthesia. Keratometry was performed using a Nidek KM 500 handheld autokeratometer (Nidek Inc, Fremont, California, USA). A minimum of 3 readings were taken and an average of the readings was chosen for IOL power calculation. Biometry was done using standard applanation technique (OcuScan; Alcon Laboratories, Fort Worth, Texas, USA). OcuScan is equipped to provide IOL power using various IOL formulae, namely SRK II, SRK T, Holladay I, and Hoffer Q. Ten readings were taken with a sharp retinal spike. Calculations were made using standard deviation (SD) less than 0.1 mm and the average reading. The SRK II formula was used to decide the IOL power. Even in cases of unilateral cataract, biometry for both eyes was performed to rule out any discrepancy in the IOL power. Where discrepancy was noted, it was cross-checked with another surgeon before choosing a desired IOL power. IOL power was calculated according to Enyedi’s correction. All patients had records of keratometry, axial length, and implanted IOL power. For all these patients, appropriate IOL power calculation was obtained according to the various formulae. The operated eyes were implanted with either rigid polymethyl methacrylate (PMMA) nonfoldable lenses or acrylic hydrophobic foldable lenses.
After injecting high-molecular-weight viscoelastic in the anterior chamber, a continuous curvilinear capsulorhexis was initiated with a cystitome and completed with pediatric capsulorhexis forceps (23 G, Model no. IG- 3984, Pediatric Capsulorhexis forceps; Indo-German, Mumbai, India). The anterior capsule was stained with trypan blue 0.5% in cases with total cataract and in other cases when needed. Lens aspiration was performed through 2 corneal incisions using automated bimanual irrigation and aspiration. An intraocular lens was implanted through a scleral tunnel or clear corneal incision according to the surgeon’s preference. Hydrophobic acrylic foldable lenses (Acrysof SA60 AT, MA60AC, or SN60WF; Alcon Laboratories) with an optic diameter of 6 mm and overall diameter of 13 mm or rigid PMMA (Ocular Vision SF 102; Eye Care, Vadodara, India) IOL with an optic diameter of 5.5 mm and overall diameter of 12 mm were implanted. PMMA IOLs were implanted through scleral tunnel incisions. In accordance with the surgeon’s preference, primary posterior capsulotomy (PPC) was done before or after implanting the IOL. In cases where the primary capsulotomy was done after the implantation of the IOL, a vitrectomy probe was passed under the IOL edge and PPC was done using the vitrector in direct contact with the posterior capsule with a cut rate of 600 cuts/minute and low vacuum of 100 to 150 mm Hg. This was followed by limited anterior vitrectomy. All the IOLs were implanted in the capsular bag. The incision was sutured with a 10–0 nonabsorbable monofilament nylon suture (Aurolab, Tamil Nadu, India). The size of the anterior capsulorhexis was aimed at 5 to 5.5 mm and that of posterior capsulotomy at 4 to 4.5 mm. The interval between surgeries in bilateral cases ranged from 1 to 4 weeks.
All children were examined on the first postoperative day. They were prescribed topical tobramycin 0.3% eye drops 4 times a day for a week. Then for the next 2 to 3 weeks, atropine sulfate 1% eye ointment twice a day was prescribed for children less than 1 year of age, and homatropine bromide 2% eye drops 2 times a day was prescribed for children more than 1 year of age. Prednisolone acetate 1% eye drops were prescribed 8 to 10 times a day initially and were gradually tapered off over 6 weeks. EUA was done 1 to 4 weeks postoperatively. Complete eye examination including measurement of intraocular pressure and retinoscopy was done. The sutures were removed at 4 weeks postoperatively. The retinoscopy readings after suture removal were taken into consideration for the purpose of analysis. A trained optometrist who was not aware of the amount of undercorrection did the retinoscopy.
The data included the patient’s age at surgery, keratometry, axial length, power of the implanted IOL, type of IOL, IOL position, and refractive status. Postoperative refraction was determined using retinoscopy at the time of EUA at 4 weeks postoperatively for all patients. Refractive error was converted into spherical equivalent (spherical equivalent = sphere + cylinder/2) in diopters (D). Using axial length, keratometry, and the manufacturer’s “A” constant obtained at the time of surgery, expected refraction was calculated with each of the 4 formulae.
Prediction error and absolute prediction error were calculated for each formula as follows:
Prediction error = Target refraction − Actual refraction
A b s o l u t e error = | T arg e t refraction − Actual refraction |
Descriptive statistics were used to represent the distribution of the prediction errors with each of the formulae. Kruskal-Wallis test was used to evaluate the differences in the prediction errors among the formulae. As both eyes of a few children had undergone surgery, the cluster of data for each child was considered as the primary sampling unit for analysis. Multivariate regression models were also built to evaluate the effect of age, axial length, and keratometry on the absolute prediction errors, with each of the formulae. Statistical analyses were performed using commercial software ([Stata version 11; StataCorp, College Station, Texas, USA] and [SPSS version 18; SPSS, Inc, Chicago, Illinois, USA]). A P value less than .05 was considered statistically significant.