Evaluation of the DiversiLab typing method in a multicenter study assessing horizontal spread of highly resistant gram-negative rods

Abstract

The worldwide prevalence of highly resistant Gram-negative rods (HR-GNR) is increasing rapidly. Reliable typing methods are needed to detect and control outbreaks and to monitor the effectiveness of infection control programs in endemic situations. In this study, we investigated the performance of the DiversiLab typing method in comparison with the amplified fragment length polymorphism (AFLP) typing method. Six hundred fifty-three HR-GNR isolates, which were obtained during a 6-month prospective survey in 18 Dutch hospitals, were typed by AFLP and DiversiLab. Subsequently, the sensitivity and specificity of DiversiLab were calculated, using AFLP as the reference method. In addition, results were compared by means of epidemiological linkage, and Cohen's kappa for agreement was calculated. DiversiLab considered significantly more isolates (275) to belong to a cluster than AFLP (198) (P < 0.001). In direct comparison, the sensitivity was 83.8%, and the specificity was 78.6%. When epidemiological linkage was included in the analysis, DiversiLab considered eight isolates as secondary cases, which were considered unique in AFLP. Only two secondary cases, according to AFLP, were missed by DiversiLab. This results in a kappa for agreement of 0.985. In daily practice, a typing method has to be used in combination with epidemiological information. When this was done, DiversiLab was shown to be a reliable method for the typing of HR-GNR. This, in combination with the ease of use and the speed, makes DiversiLab an appropriate method for screening in routine clinical practice. When a cluster is suspected and the consequences of these findings are substantial, a confirmatory analysis should be performed.