The accuracy of pooling urine samples for the detection of genitalChlamydia trachomatis infection by ligase chain reaction (LCR) was examined. A model was also developed to determine the number of samples to be pooled for optimal cost savings at various population prevalences. Estimated costs included technician time, laboratory consumables, and assay costs of testing pooled samples and retesting individual specimens from presumptive positive pools. Estimation of population prevalence based on the pooled LCR results was also applied. After individual urine specimens were processed, 568 specimens were pooled by 4 into 142 pools and another 520 specimens were pooled by 10 into 52 pools. For comparison, all 1,088 urine specimens were tested individually. The sample-to-cut-off ratio was lowered from 1.0 to 0.2 for pooled samples, after a pilot study which tested 148 samples pooled by 4 was conducted. The pooling algorithm was 100% (48 of 48) sensitive when samples were pooled by 4 and 98.4% (61 of 62) sensitive when samples were pooled by 10. Although 2.0% (2 of 99) of the negative pools of 4 and 7.1% (1 of 14) of the negative pools of 10 tested presumptive positive, all samples in these presumptive-positive pools were negative when retested individually, making the pooling algorithm 100% specific. In a population with 8% genital C. trachomatis prevalence, pooling by four would reduce costs by 39%. The model demonstrated that with a lower prevalence of 2%, pooling eight samples would reduce costs by 59%. Pooling urine samples for detection of C. trachomatis by LCR is sensitive, specific, and cost saving compared to testing individual samples.
Comparison of the ligase chain reaction with cell culture for the diagnosis of Chlamydia trachomatis infection in women.
