Olsen Thomas
University Eye Clinic, Aarhus Kommunehospital, Aarhus, Denmark.
J Cataract Refract Surg. 2006 Mar;32(3):419-24. doi: 10.1016/j.jcrs.2005.12.139.
To investigate methods to predict the effective postoperative anterior chamber depth (ACD) based on a large patient sample.
University Eye Clinic, Aarhus Kommunehospital, Aarhus, Denmark.
Based on 6698 consecutive cataract operations with recorded postoperative refractive results, the postoperative effective ACD was calculated in each case and studied by multiple linear regression for covariance with a number of preoperatively defined variables including the axial length by ultrasonography, preoperative ACD, lens thickness, corneal radius by keratometry, subjective refraction, patient age, and corneal white-to-white diameter, the latter of which was available in a subgroup of 900 cases.
The postoperative effective ACD was significantly correlated with 6 preoperative variables (in decreasing order): axial length, preoperative ACD, keratometry reading, lens thickness, refraction, and patient age (R = 0.49, P < .000001). Age showed the weakest correlation (P = .02) and could be omitted with no significant decrease in the total correlation coefficient. Using the 5 most significant variables, the ACD could be predicted according to a regression formula with an accuracy of 82.1% of the predictions within 0.5 mm. When this ACD algorithm was used in retrospect in the intraocular lens (IOL) power calculation, the refractive prediction error decreased by 10% from the error associated with a previously published 4-variable algorithm and decreased 28% from the error using no individual ACD method other than the average ACD (P < .00001).
The postoperative ACD was significantly correlated with and hence predictable by a 5-variable regression method incorporating the preoperative axial length, ACD, keratometry reading, lens thickness, and refraction as the most significant variables. The statistical relationship can be used to create a new ACD prediction algorithm to incorporate in a modern "thick lens" IOL power calculation formula with significant improvement in the accuracy of the refractive predictions as a result.
基于大量患者样本研究预测术后有效前房深度（ACD）的方法。
丹麦奥胡斯市奥胡斯市立医院大学眼科诊所。
基于6698例连续白内障手术且记录了术后屈光结果，计算每例患者的术后有效ACD，并通过多元线性回归研究其与多个术前定义变量的协方差，这些变量包括超声测量的眼轴长度、术前ACD、晶状体厚度、角膜曲率计测量的角膜半径、主观验光、患者年龄以及角膜白对白直径（后者在900例亚组中可得）。
术后有效ACD与6个术前变量显著相关（按相关性递减顺序）：眼轴长度、术前ACD、角膜曲率计读数、晶状体厚度、验光和患者年龄（R = 0.49，P <.000001）。年龄显示出最弱的相关性（P =.02），省略该变量后总相关系数无显著下降。使用5个最显著变量，可根据回归公式预测ACD，预测准确率为82.1%，预测值在0.5mm范围内。当此ACD算法回顾性用于人工晶状体（IOL）屈光度计算时，屈光预测误差比与先前公布的四变量算法相关的误差降低了10%，比使用除平均ACD外无个体ACD方法的误差降低了28%（P <.00001）。
术后ACD与包含术前眼轴长度、ACD、角膜曲率计读数、晶状体厚度和验光作为最显著变量的五变量回归方法显著相关，因此可通过该方法预测。这种统计关系可用于创建一种新的ACD预测算法，以纳入现代“厚晶状体”IOL屈光度计算公式，从而显著提高屈光预测的准确性。
