Comparison between two commercially available serological tests and polymerase chain reaction in the diagnosis of Cryptosporidium in animals and diarrhoeic children

For the detection of Cryptosporidium species in 804 animals and 165 diarrhoeic children (<10 years) in Egypt, two copro-antigen tests, the RIDASCREEN® Cryptosporidium test [enzyme immunoassay (EIA)] and the RIDA®QUICK Cryptosporidium/Giardia Combi [immuno-chromatographic test (ICT)] as well as polymerase chain reaction (PCR) were used. Prevalence of Cryptosporidium was 15.0, 19.5 and 32.3 % in animals and 2.4, 6.7 and 49.1 % in children using EIA, ICT and PCR, respectively. Using PCR as reference method, animal samples sensitivity (Se) of the EIA was 46.5 % when questionable samples were considered positive, whereas specificity (Sp) was 100 %. Se of the ICT was 60.4 % while Sp was 100 %. Positive predictive values (PPVs) for both EIA and ICT test were 100 %, and negative predictive values (NPVs) for EIA were 79.7 and 84.1 % for ICT. For the children samples, the Se of EIA was 5 %, Sp was 100 %, PPV was 100 % and NPV was 52.2 %, while the Se of ICT was 13.6 %, Sp was 100 %, PPV was 100 % and NPV was 54.6 %. The Kappa score of agreement between PCR and ICT was 67.4 %, 54.1 % between PCR and EIA and 84.4 % between ICT and EIA. Until the second serial dilution of the EIA and ICT test, 9 × 103 oocysts/μl of Cryptosporidia was detected, whereas in PCR, they were detected until the sixth serial dilution. Copro-antigen tests were easy to perform and less time-consuming but less sensitive compared to PCR. They obviously are best applicable for screening and epidemiological studies of large numbers of subjects, for batch specimen processing and in isolated or rural areas where reliable tests like PCR are unfeasible. When in children, a single stool sample is used for the diagnosis of clinical cases; better results can be obtained when non-standardized PCR due low specificity is coupled with copro-antigen tests.