Bias due to conjunctiva hue and the clinical assessment of anemia.

The influence of conjunctiva hue on the clinical evaluation of anemia was tested by three educated non-clinicians, trained for such a purpose, in 219 healthy ambulatory subjects. Over 27% of conjunctivae were classified as pink and 72.8% as reddish; the three observers agreed on 8.2% of the pink and on 49.8% of the reddish. Anemia (hemoglobin below 10.0 g/dl) was detected in 5.4% of the subjects with pink hue, and in 3.4% of subjects with reddish hue. Between 39.7 and 62.2% of the subjects with pink conjunctiva were misclassified as anemic by at least one worker, and between 20.0 and 50.0% of those with reddish conjunctiva were misclassified as not anemic. Overall agreement on conjunctiva hue for the 219 worker pairs ranged from 69.0 to 75.0%; the three kappa coefficients were between 0.27 and 0.34. The importance of conjunctiva hue variation will depend on the objective of the assessment and on the availability of laboratory tests. Bias due to conjunctiva hue should be stressed when clinicians and field personnel are under training.