A demonstration of lack of variability among six tuberculin skin test readers.

The variability of tuberculin skin test readings among six trained and experienced readers was evaluated using a modified sliding caliper method. Each of 537 tests were read independently by two readers. There were 23 disagreements between paired readers resulting in an overall interobserver reliability of 95.7 per cent. In 82 per cent of the paired readings the results were different by 2 mm or less. The observer lack of variability was likely due to the training and experience of the readers.