Iron deficiency (ID) is the most common nutritional deficiency. ID diagnosis requires ferritin measurement because clinical findings are poor and nonspecific. We studied the diagnostic value of blue sclera, which was scarcely reported as a specific and sensitive sign of ID. We enrolled 74 patients suspected of having ID. Pictures of their eyes were taken using a smartphone under similar daylight conditions. Three independent physicians graded the scleral color, and a computer analysis yielded the blue percentile of the sclera image. Final analysis included 67 patients (mean age 59.9 ± 20.1 years). Fifty-one had ID. Subjective blue scleral color was associated with ID for physician 1 (64.5% vs. 86.1%, p = 0.03). Sensitivity was 60.8% (CI95: 46.1%; 74.2%), specificity 68.8% (CI95: 41.3%; 89%), and positive predictive value 86.1% (CI95: 70.5%; 95.3%). A marginal difference was observed for other physicians (p = 0.05). Computer analysis showed higher blue in the ID group (p = 0.04). The area under the receiver operating characteristic (ROC) curve was 0.7 (0.54; 0.85). Sensitivity was 78.4% (CI95: 63.7%; 88.7%), specificity was 50% (CI95: 24.7%; 75.3%). Assessment of blue sclera was not influenced by iris color, sex, or anemia. We showed that blue sclera has good positive predictive value for ID diagnosis, and computer analysis was correlated to clinical assessment. Improvement of this innovative, non-invasive method could provide an easy handling and inexpensive diagnosis tool for ID.
Keywords: ROC curve; anemia; diagnostic imaging; iron metabolism disorders; sclera; smartphone.