In the present retrospective study, we assessed the performance of the SRK/T formula using ultrasonic biometry in eyes with short and long ALs seperately.
Group 1 (Short ALs)
The AL is the most important factor in IOL calculation. Any measurement error in the AL of a short eye could have a larger effect on final refractive error. Compression of the eye is believed to be part of the cause of AL shortening error [
1‐
3,
17,
18]. A mean shortening of 0.25–0.33 mm has been reported between applanation and immersion axial length measurements, which can translate into an error of IOL power by approximately 1 D [
3,
17‐
19].
There are different studies to evaluate the predictive accuracy of various IOL power calculation formulas in eyes with short AL by using different IOL calculation methods and different results are reported. Sander et al. [
6,
16] and Narvaez et al. [
20] reported that the SRK/T effective and no difference between any of third and fourth generation formulas at errors. Wang et al. [
21] showed that the SRK/T and Hoffer Q were equal. Gavin and Hammond [
17], Aristodemou et al. [
12], Kapadia et al. [
22], Hoffer Q [
5], Szaflik et al. [
23], Day et al. [
24] showed that the Hoffer Q formula more accurate, contrary to Maclaren et al. [
25], Terzi et al. [
26], Moschos et al. [
27] and Roh et al. [
18] reported that the Haigis formula was more accurate than the other formulas.
In our study, the mean PE was 0.017 ± 0.58 D (range from −1.060 to 1.015) and there was a little tendency towards hyperopia. We found a prediction accuracy of 57.89 % for refractive errors of ±0.50 D, a prediction accuracy of 94.74 % for refractive errors of ±1.00D using SRK/T in eyes with short ALs (Table
2). The MAE of our study was 0.48 ± 0.29D (0.02 to 1.060). These results were similar to the previous studies or better than.
The ME was 0.87 D ± 0.829 D with SRK/T formula in 41 eyes with AL<22.00 mm using IOL master(IOLm) in the study conducted by Gavin and Hammond [
17], 0.834 ± 0.262 D in eyes with AL<22.00 mm (
n = 10; relatively small size series) by Hoffer et al. [
5], 0.53 ± 0.25 D in 25 eyes with AL< 22.00 mm using IOLm by Roh et al.[
18], 0.78 ± 0.66 D in 33 eyes with AL<22.00 mm using optical biometry by Wang et al. [
21], 0.91 ± 0.64 D in 163 eyes with AL <22.00 mm using IOLm by Day et al., and 0.41 ± 0.23 D by Moschos et al. [
27]. Contrary to the ME was −1.45 ± 0.14 D in 76 eyes with mean AL =20.79 mm by Maclaren et al. [
25] and −0.96 ± −1.24 D in eyes with axial length < 21,00 D by Kapadia et al. [
22]. Kapadia et al. reported that the postoperative SE within ± 1,0D was 80 % with SRK/T formula using A-Scan biometry.
In this study, we found a negative correlation between AL and K1, K2, IOLp; practically, as AL decreased, K1, K2,and IOLp increased. A negative significiant correlation was found between PEand and achieved SE, but no correlation between PE and the other parameters.
Groups 2 (Long ALs)
The main difficulties in IOL power calculations for long eyes may be partly due to the anatomy of the posterior pole (posterior staphyloma). Posterior staphyloma decrease the accuracy of preoperative biometry. A-Scan biometry has a disadvantage compared with optic biometry and immersion biometry for accurate AL measurement [
28‐
30]. Because of this, using A-Scan biometry with B-Scan mod together is recommended [
29]. In our patients, sometimes A-Scan biometry was combined with B-Scan mod for detecting side of staphyloma.
The SRK/T formula probably the most accurate formula for long eyes and is now widely used. Holladay et al. [
4], Sanders et al. [
16],Hoffer Q [
5], Kapadia et al. [
22], Maclaren et al. [
25], Donoso et al. [
31], Kapamajian and Miller [
32], Aristodemou et al. [
12], El-Nafees et al. [
33] and Chua et al. [
34] were reported that SRK/T formula was more accurate than the other formulas in long eyes. Haigis et al. [
15,
35], Terzi et al. [
26], Bang et al. [
36] and Roessler et al. [
30] reported that the Haigis formula more accurate than the SRK/T formula and the others. Mitra et al. [
37] and Petermeier et al. [
38] reported that the SRK/T, Haigis or; Holliday were equal. Wang et al. [
39] reported that the SRK/T and Haigis formula were comparable.
Sanders et al. [
16] reported that for errors less than 0.5D was 45 %, less than 1.0D the results was 85 %, and greater than 2 D was 2.5 % by the SRK/T formula. In that study, there was no difference between SRK/T formula and the others. In the study conducted by Petermeier et al. [
38], postoperative SE was −1.42 ± 1.33D (−3.94 to +1.0) (positive dioptre IOL group (
n = 30) and postoperative SE was within ±0.5 D in 45.5 % of cases, and within ±1.0 D in 77.3 % of cases. Zaldivar et al. [
29] reported that 92%of eyes were within ± 1.0D when using SRK/T formula in cases of plus power IOLs, and 54 % with the SRK/T in the cases of minus power IOLs. Maclaren et al. [
25] did a retrospective analysis in 75 eyes having cataract surgery with zero- or negative-powered IOLs using SRK/T formula and A-scan, B-scan, and optic biometry. They also reported that forty-one percent of 75 patients analyzed were within ±1.00 D of the predicted refraction and 95 % confidence interval, 0.89-1.39 D. Kapadia et al. [
22] reported that the MAE −0.59 ± 0.91D, −0.46 ± 0.24 D, 0.24 ± −0.05 D in eyes with axial length 24–27 mm (
n = 28), 27–29 mm (
n = 27), and >29 mm (
n = 25) respectively, using SRK/T formula and A-Scan biometry. The postoperative SE was within ± 1 D in 67.85 % of cases when using SRK/T formula (Haigis equall; 68 %) in their study.
Ghanem and El-Sayed [
28] reported that the postoperative SE was ± 1.0 D of assumed refraction in 75 %, the refractive outcome was within ±1.0D in 45 %, and there was a tendency toward hyperopia with SRK/T formula (
n = 127, AL ≥ 26 mm). In the study conducted by Holladay et al. [
4,
7], the ME was −0.194 D, and the MAE was 0.345 ± 0.401 D in eyes with AL 24.5–26.0 mm, the ME was 0.041, the MAE was 0.442 ± 0.56 D in eyes with AL greater than 26.0, and the MAE was 0.38 ± 0.47 D in all long eye. In a study consisting of more than 300 long eyes, Aristodemou et al. [
12] reported that the SRK/T had the lowest MAE, with statistically significant differences for ALs of 27.00 mm or longer. Mitra et al. [
37] found the ME was +0.92 D with SRK/T formula using applanation ultrasonography in Indian myopic population with long axial lengths (24.75–32.35 mm). Wang et al. [
39] reported that the MAE was 0.45 ± 0.10 D with the SRK formula in eyes with AL more than 26 mm (
n = 75). Narvaez et al. [
20] reported the MAE was 0.49 ± 0.39 (0.00 ± 2.26 D; 24.5–26.0 mm), the MAE was 0.55 ± 0.64 D (range 0.04 ± 3.48 D; greater than 26 mm) in totally 181 eyes. El Nafees et al. [
33] reported that the ME was +0.04 D (25–27 mm), +0.15D (27–29 mm), +0.33D (29–31.4 mm) with SRK/T and the MAE was less than 1.0D in 81.3 % eyes (
n = 53 eyes). Chua et al. [
34] reported that the ME was 0.18 D for eyes using SRK/T with ALs greater than 25 mm, Kapamajian and Miller [
32] reported the mean PE was +1.16D and Roessler et al. [
30] reported the MAE was 1.01 ± 0.61D with the SRK/T formula using optical biometry in long eyes.
In our study, the mean AL was 25.77 D (range from 24.60 to 32.90) and the mean IOLp was 15.80 (range from −5.00 to 20.50D). Only one patient had a negative IOL power. The postoperative SE was 0.42 ± 0.96 D (range; −2.62 to 2.75 D), within ± 1 D in 70.97 % of cases and within ± 0.5 D in 50 % of cases. The mean PE of in long ALs was −0.004 ± 0.93 D (range from −1.83 D to 3.55D) and there was a little tendency towards myopia. The MAE was 0.68 ± 0.62D (range 0.005 to 3.55D). We showed a prediction accuracy of 50 % for refractive errors of ±0.50 D, a prediction accuracy of 70.97 % for refractive errors of ±1.00D using SRK/T formula in eyes with long ALs (Table
2). We found a positive significant relationship between attempted preoperative SE and achieved postoperative SE (
r = 0.255;
p = 0.045). These results showed that refractive outcomes similar to the preoperative target refractive prediction ± 1D were reached. Only a few refractive surprises may be due to the AL errors in ultrasonic biometry or use of inappropriate formula.
Additionaly, the other results of our study can be summarized as follows:
1.
We found no significant relation-ship between PE and AL,K1,K2, IOLp, IOL types, age in both groups.
2.
We found a significant negative correlation for short ALs, contrary to a significant positive correlation for long ALs, between PE and achieved SE.
3.
We found no significant relation-ship was found between MAE and AL, K1, K2, age, attempted SE, achieved SE in both groups.
4.
We found a significant negative correlation between AL and K1,K2, IOLp in short ALs, a significant positive correlation between AL and K1,K2, contrary to negative AL and IOLp in long ALs.
This study has some weakness. Firstly, it is a retrospective analysis. Secondly, the relatively sample size (38 eyes) and a narrow range (20.50–21.99 D) of AL for the short eyes. Thirdly, only one formula(SRK/T) was used for IOL calculation and not compared with other formulas. Finally, different IOL types and IOL constants were used. On the other hand, our study also has some strength. Firstly, all surgeries and procedures were performed by a single surgeon with the same technique and devices. Secondly, relatively large sample size for long eyes (n = 62). Finally, using only one formula, the SRK/T formula, is an advantage of this study so as to determine the performance of a single formula.