目的：探究OCULUS Keratograph 5M(简称K5M)眼表综合分析仪用于干眼泪膜分度诊断的诊断价值。

方法：收集门诊通过常规干眼检查手段诊断为干眼的患者50例100眼与同时期可疑干眼的非干眼患者20例40眼，通过K5M眼表分析仪采集以上患者的即时眼表参数，包括第1次泪膜破裂时间(NIBUTf)、平均泪膜破裂时间(NIBUTav)、泪河高度(TMH)，进行K5M眼表分析仪与常规干眼检查对干眼泪膜相关指标分度的诊断效能试验。

结果：K5M眼表分析仪对干眼组与非干眼组NIBUTf、NIBUTav、TMH均有差异(P<0.05)。K5M眼表分析仪的NIBUTf与常规检查诊断BUT的分度诊断符合率为78.6%，ROC曲线下面积为0.795，标准误为0.042，P<0.01，95%的置信区间为(0.712，0.878)； K5M眼表分析仪的NIBUTav与常规检查诊断BUT的符合率为73.6%，ROC曲线下面积为0.853，标准误为0.033，P<0.01，95%的置信区间为(0.788，0.917)； K5M眼表综合分析仪的TMH与常规检查诊断SⅠt的符合率为87.9%，ROC曲线下面积为0.795，标准误为0.044，P<0.01，95%的置信区间为(0.709，0.880)。通过NIBUTf结合TMH对干眼的探究性诊断的灵敏度为98%，特异度为40%，正确指数为38%，阳性似然比1.63，阴性似然比0.05，比值比OR为32.6。通过NIBUTav结合TMH对干眼的探究性诊断的灵敏度为86%，特异度为75%，正确指数为61%，阳性似然比3.44，阴性似然比0.187，比值比OR为18.39。

结论：K5M眼表分析仪能够为干眼的单一泪膜指标分度诊断提供准确可靠的诊断价值； TMH结合NIBUTf对干眼的诊断灵敏度要优于结合NIBUTav，而特异度要差于结合NIBUTav。

METHODS: Totally 100 eyes diagnosed with dry eye by conventional means and 40 non-dry eye eyes with suspected dry eye at the same time were enrolled through OPC. The immediate ocular surface parameters of the above cases were collected by the K5M ocular surface analyzer, including the first tear film rupture time(NIBUTf), the average tear film rupture time(NIBUTav), and the tear river height(TMH)which were utilized to evaluated the diagnostic efficacy of K5M compared to routine ocular surface examination on the index of dry eye tear film.

RESULTS: The comparison of NIBUTf, NIBUTav and TMH between dry eye group and non-dry eye group by K5M eye surface comprehensive analyzer showed statistically significant differences(all P<0.05). The coincidence rate between the NIBUTf of K5M ocular surface analyzer and the BUT of routine diagnosis was 78.6%, and the area under the ROC curve as well as the standard error were 0.042 and 0.795(P<0.01)with 95% confidence interval(0.712, 0.878); The coincidence rate between NIBUTav of K5M and BUT of routine examination diagnosis was 73.6%, and the area under the ROC curve as well as the standard error were 0.853 and 0.033(P<0.01), with 95% confidence interval(0.788, 0.917).The coincidence rate between TMH of K5M and the SⅠt of conventional diagnosis was 87.9%, and the area under the ROC curve as well as the standard error were 0.795 and 0.044(P<0.01), with the 95% confidence interval(0.709, 0.880). The sensitivity, specificity, correctness index, positive likelihood ratio, negative likelihood ratio, and odds ratio of the exploratory diagnosis of dry eye by NIBUTf combined with TMH were 98, 40, 38, 1.63, 0.05, and 32.6%, respectively. The sensitivity, specificity, correctness index, positive likelihood ratio, negative likelihood ratio, and odds ratio of the exploratory diagnosis of dry eye by NIBUTav combined with TMH were 86, 75, 61, 3.44, 0.187, 18.39%, respectively.

CONCLUSION: K5M ocular surface analyzer can provide accurate and reliable diagnostic value for single tear film index grading diagnosis of dry eye; compared with TMH combined with NIBUTav, TMH combined with NIBUTf was more sensitive but less specific in diagnosing dry eye.