The cards were close to serum results
Two dried-blood-spot methods produced results close to conventional serum testing for antibodies against Trypanosoma cruzi in a prospective agreement study. Against serum, Whatman filter cards had 95.55% sensitivity and 100% specificity, while Cobas Plasma Separation Cards, or PSCs, had 94.86% sensitivity and 100% specificity. Cohen's kappa, a statistic used to measure agreement between classifications, was 0.910 for Whatman and 0.897 for PSC, a level the analysis described as almost perfect in this sample.
Sensitivity here means the share of serum-positive samples that a card also identified, while specificity is the share of serum-negative samples it left negative. The reported 95% confidence interval for Whatman's sensitivity ranged from 92.32% to 97.51%; PSC's ranged from 91.49% to 96.99%. For both devices, the reported specificity interval ran from 94.9% to 99.9%.
The gap appeared near the cutoff
Researchers analyzed surplus paired serum and whole-blood samples from adult patients at the International Health Unit of Hospital Universitari Vall Hebron whose anti-T. cruzi serology had been requested. Whole blood was processed into dried blood spots on either Whatman filter cards or PSC cards and tested with the Elecsys Chagas assay, with paired serum serving as the reference comparator. Of 387 DBS pairs initially collected, five were withdrawn because of operator errors, leaving 382 valid paired results.
For the initial comparison, a COI, or cut-off index, of 1.0 or more counted as reactive and a value below 1.0 as non-reactive. The cards produced lower numerical signals than serum. Median COI was 306.6 in serum, compared with 96.5 on Whatman and 63.3 on PSC; both DBS medians were lower than serum, with p-values below 0.05.
The difference was concentrated among samples with low serum COI values. Of 292 samples that serum classified as positive, 13 Whatman results, or 4.45%, and 15 PSC results, or 5.14%, were incorrectly classified as negative. In those discordant cases, the median serum COI was 4.67, while the median was 0.41 on Whatman and 0.34 on PSC.
Changing the cutoff changed the balance
Researchers then tested lower, device-specific thresholds: at least 0.21 for Whatman and at least 0.23 for PSC, rather than the established serum threshold of at least 1. With those settings, sensitivity was 100% for Whatman and 99.7% for PSC. Specificity was 97.8% for both, showing the tradeoff between missing fewer positives and allowing more false-positive classifications.
Despite the difference in signal size, the two cards almost always made the same yes-or-no call. Among 382 matched samples, both devices classified 277 as positive and 103 as negative; only two samples, or 0.5%, were discordant. Both were positive on Whatman and negative on PSC, producing a device-to-device kappa of 0.987.
The numerical results showed a more consistent difference between the devices. Whatman COI values were higher than PSC values, with a paired difference that had a p-value below 0.001, but the values also tracked one another closely, with a correlation coefficient of 0.974 and a correlation p-value below 0.0001. A Bland-Altman analysis, which estimates how far paired measurements differ, found a bias of 0.09 on a base-10 COI scale and 95% agreement limits between -0.36 and 0.53. Some 379 of 382 cases, or 99.2%, fell within those limits.
A useful result with clear limits
The authors interpret optimized DBS testing with either device as a reliable and viable alternative to invasive serum sampling, particularly for screening where access to health facilities is limited.
The conclusion has clear boundaries. This was a single-center prospective agreement study using adult surplus samples under controlled laboratory conditions. The authors say the findings should be understood as evidence of agreement between DBS devices, not as a measure of clinical diagnostic accuracy. The controlled laboratory setting differs from field settings, so the study does not establish how the cards would perform outside those conditions.
The study was supported by Roche Diagnostics International Ltd and the Instituto de Salud Carlos III. The funders were reported to have had no role in the design, data collection, analysis, publication decision or manuscript preparation, and the authors declared no competing interests.
Paper data and sources
Original title: Comparison of Cobas Plasma Separation Cards and Whatman filter paper cards versus conventional serum samples for anti-Trypansoma cruzi antibody detection.
Authors: Aroa Silgado, Alejandro Mediavilla, Patricia Martínez-Vallejo et al.
Journal/Repository: PLoS neglected tropical diseases
Status: Peer-reviewed
First online: 2026-08-21
DOI: 10.1371/journal.pntd.0014554
Original paper · Full text