Study of reliability and concordance between two VDRL tests for syphilis investigation in patients with HIV followed-up at a university hospital in Rio de Janeiro in 2017–2019

Abstract

Introduction: The laboratory diagnosis of syphilis is given by a positive treponemal test and a non-treponemal test, with VDRL (Veneral Disease Research Laboratory) being the “gold standard”. Objective: To compare two tests commercially validated for biological fluids and analyzed by different operators, in order to assess their performance in detecting high (≥:8) and low (≤1:2) titrations, as well as to determine the agreement between results in paired serum samples from patients with syphilis and living with HIV. Methods: Cross-sectional study, approved by the Research Ethics Committee of the teaching hospital Gaffrée e Guinle (HUGG), under CAAE 66558117.0.0000.5258. The study population was composed by patients diagnosed with syphilis and confirmed by the positivity of one or more treponemal tests. All samples were analyzed simultaneously by two different operators, each using a kit: VDRL WAMA Diagnóstica®, São Carlos, SP, Brazil; VDRL Brás, Laborclin®, Pinhais, PR, Brazil. The SPSS statistical program was used. Results: 110 serum samples from patients diagnosed with syphilis treated at HUGG were analyzed. The frequency of high VDRL titrations among patients, following the VDRL criterion ≥1:8, was practically the same in both tests, with 68% in VDRL Laborclin and 69% in VDRL WAMA (p = 0.87) and VDRL ≤1:2, 80% for WAMA and 83% for Laborclin (p = 0.72). The results of VDRL were tabulated in pairs; then the Cohen’s Kappa coefficient of agreement was calculated (Κ) 0.32 (95%CI 0.21–0.41; p<0.00001), as well as the weighted Kappa (Kw) and the intraclass correlation coefficient (ICC) 0.89 (95%CI 0.84–0.92; p<0.00001). The Bland-Altman diagram was also used. We found poor agreement between the VDRL tests when results were nominally concordant, that is, with the same titles in both tests. However, if partial agreement is considered, the interpretation of the magnitude of agreement estimators was almost complete (≥0.80). Conclusion: Reliability and agreement were high between the VDRL tests of both manufacturers when considering the close titrations (up to two dilutions). Further reliability and agreement studies are essential between the non-treponemal tests available and used in Brazil.