Purpose: To explore whether slit lamp settings may influence measurements made with the LOCS III grading system. The following were tested using a total of 221 subjects: 1. Test-retest variability (with and without the neutral density filter). 2. Readings with and then without the neutral density filter. 3. Readings with maximum and minimum illumination (filter and rheostat). 4. Examinations with different slit beam width. 5. Light output from 10 slit lamps (Haag-Streit 900BM) was measured using a lightmeter.
Results: The 95% limits of agreement (test-retest examination) for nuclear opacity (NO), nuclear colour (NC), cortical opacity (C), and posterior subcapsular lens opacity (P) were 0.66, 0.60, 0.62 and 0.39, respectively, using standard settings. Corresponding results with the neutral density filter were similar. Examinations performed with and without the neutral density filter showed that the 95% limits of agreement increased by a factor of at least 1.7 compared with test retest data (NO and NC) and 2.2 for (C and P) (p < 0.001 (f test)). Maximum vs. minimum brightness settings increased variability by a factor of at least 2.1 (NO and NC) and 3 (C and P) (p < 0.001 (f test)). Changing beam width measurements produced a significant systematic measurement bias of 0.3 for NO and 0.4 for NC (P < 0.01 (t test)), a wider beam giving a higher score. Individual slit lamps may vary by a factor of four in their light output levels for apparently identical settings. The range of illumination produced by a slit lamp is 46-fold.
Conclusions: For nuclear opacity and nuclear colour measurements, changing settings between examinations increases variability without evidence of systematic bias. However, using a thicker slit beam induced a systematic bias. For cortical and posterior subcapsular lens opacity, varying the illumination had more marked effects on reproducibility without a systematic bias.