Risk of Bias and Reporting Quality in BP180/BP230 ELISA Diagnostic Studies of Bullous Pemphigoid: A Meta-Epidemiological Study.

BACKGROUND

BP180 and BP230 enzyme-linked immunosorbent assays (ELISA) are widely used to support the diagnosis of bullous pemphigoid (BP), yet published sensitivity and specificity vary. We assessed whether methodological limitations and incomplete reporting may contribute to optimistic "headline" accuracy estimates.

METHODS

We performed a meta-epidemiological methodological study of 18 published diagnostic accuracy studies (2004-2025) contributing 42 ELISA index tests (23 BP180, 19 BP230). Risk of bias was assessed with QUADAS-2 and reporting quality with a 16-item STARD-based checklist. We summarized sensitivity and specificity and compared estimates by patient-selection risk (low vs. high/unclear) and cut-off type (manufacturer/prespecified vs. post hoc). No new pooled meta-analysis of diagnostic accuracy was performed.

RESULTS

Patient selection was at high risk of bias in 15/18 studies (83%), and no study was judged low risk in the QUADAS-2 index test domain. Key safeguards were rarely reported: blinding 6/42 tests (14%), participant flow diagram 0/42 (0%), and indeterminate/missing results 14/42 (33%). Median sensitivity/specificity were 0.82/0.94 (BP180) and 0.58/0.98 (BP230). BP180 sensitivity tended to be lower in low-risk versus high/unclear-risk patient-selection studies (0.69 vs. 0.84; p = 0.09; n = 3 low-risk studies).

CONCLUSIONS

Published ELISA accuracy-especially sensitivity-may be overestimated; in practice, a negative ELISA should not be used to rule out BP when clinical suspicion is high.