Comparing TSPOT assay results between an Elispot reader and manual counts

Background: The interferon gamma release assay, TSPOT.TB (TSPOT) can be read by several methodologies, including an Elispot reader or manually by technician. We compared the results from these two counting methods. Methods: Automated and manual TSPOT results among 2481 United States health care workers were compared. Cohen's kappa coefficient was used to determine the inter-rater agreement. Univariate and multiple logistic regression were used to investigate selected variable contributions. Results: No prognostic factors were associated with agreement of TSPOT results between counting methods. Agreement between TSPOT results were 92.3%, 89.5%, 93.0%, and 93.1% at baseline, and at follow-up at 6, 12, and 18 months, respectively. The inter-rater agreement for all test results was good (kappa = 0.71). There was a significant difference between individual technicians kappa coefficients (p < 0.001), but no significant increase in agreement over time for technicians (p = 0.394). Conclusion: Commercial Elispot readers and manual counts have good agreement of TSPOT results in a low TB burden setting. Levels of agreement differed between individual technicians and automated reader from moderate to very good, indicating borderline results may be misinterpreted due to inter-rater variability. With no latent tuberculosis infection (LTBI) gold standard, it cannot be determined if one TSPOT reading method is better than another.