Sperduto R D, Hiller R, Podgor M J, Palmberg P, Ferris F L, Wentworth D
Am J Epidemiol. 1986 Dec;124(6):994-1003. doi: 10.1093/oxfordjournals.aje.a114489.
Technologic advances in ophthalmic equipment offer the possibility of replacing direct clinical examinations with Reading Center evaluations of data recorded in epidemiologic studies. Clinical and Reading Center examiners made independent ophthalmic diagnoses of 133 right and 132 left eyes of 138 adults in the Visual Acuity Impairment Survey Pilot Study, carried out in three US cities, Boston, Detroit, and Minneapolis, in August 1981-December 1982. The Reading Center diagnosed eye conditions using only photographic and visual field data collected at the time of the clinical examination. In the comparisons of clinical and Reading Center evaluations reported here, only eyes judged by the examiners to have pathology severe enough to reduce visual acuity to 6/9 or worse were classified as having pathology. (No visual acuity criterion was required for the diagnosis of glaucoma or diabetic retinopathy.) There was agreement in diagnostic assessments between clinical and Reading Center examiners in about 80% of eyes. The kappa statistic, which adjusts for chance agreement, was in the fair to good range: 0.60 for 133 right eyes and 0.62 for 132 left eyes. When the Reading Center examiners were provided with additional information on medical history, refractive error and best corrected visual acuity, the agreement between clinical and Reading Center assessments among the subset of eyes with 6/9 or worse vision again was in the fair to good range, with kappas of 0.61 for 45 right eyes and 0.68 for 48 left eyes. Inter-observer agreement between Reading Center examiners in diagnosing pathology was in the good to excellent range. Use of Reading Centers in future epidemiologic studies should be considered, but elimination of the clinical examinations is not recommended until modifications in the protocol described here have been made and shown to improve levels of agreement between clinical and Reading Center examiners.
眼科设备的技术进步使得在流行病学研究中用阅读中心对记录数据的评估来取代直接临床检查成为可能。在1981年8月至1982年12月于美国波士顿、底特律和明尼阿波利斯三个城市开展的视力损害调查试点研究中，临床检查人员和阅读中心检查人员对138名成年人的133只右眼和132只左眼进行了独立的眼科诊断。阅读中心仅使用临床检查时收集的照片和视野数据来诊断眼部疾病。在本文报告的临床评估与阅读中心评估的比较中，只有检查人员判断病理严重到足以使视力降至6/9或更差的眼睛才被归类为患有病理疾病。（青光眼或糖尿病视网膜病变的诊断无需视力标准。）临床检查人员和阅读中心检查人员在约80%的眼睛诊断评估上达成一致。考虑到偶然一致性进行调整的kappa统计量处于中等至良好范围：133只右眼为0.60，132只左眼为0.62。当为阅读中心检查人员提供有关病史、屈光不正和最佳矫正视力的额外信息时，在视力为6/9或更差的眼睛子集中，临床评估与阅读中心评估之间的一致性再次处于中等至良好范围，45只右眼的kappa值为0.61，48只左眼的kappa值为0.68。阅读中心检查人员在诊断病理方面的观察者间一致性处于良好至优秀范围。未来的流行病学研究应考虑使用阅读中心，但在此处所述方案进行修改并证明可提高临床检查人员与阅读中心检查人员之间的一致性水平之前，不建议取消临床检查。
