Near-body drugs of abuse testing in a post mortem medico-legal population in South Africa: a pilot study

A research report submitted to the Faculty of Health Sciences, University of the Witwatersrand, in partial fulfilment of the requirements for the degree of Master of Medicine Johannesburg 19 September 2016Background: Forensic toxicological analysis in SA is fraught with delay, partly due to lack of equipment as well as lack of experienced staff, but also due to the overwhelming number of backlogged Forensic Chemistry Laboratory (FCL) cases, as well as new cases being submitted to the FCL for generalised untargeted toxicology screening using gold-standard chromatographic techniques. These delays continue to result in significant adverse legal, social and administrative consequences. Reliable rapid near-body screening toxicological tests would reduce the load on the Forensic Chemistry Laboratories and improve efficiency of the South African criminal justice system. Methods: This was a prospective, observational and transverse study examining a sample population from the Johannesburg Forensic Pathology Service (FPS) Medico-Legal Mortuary in South Africa. Whole blood samples were taken from femoral vessels of decedents. Blood samples were processed and qualitatively analysed using near-body tests (Randox Drugs of Abuse (DOA) I and Narcotics Detector(ND) immunoassays) and liquid chromatography coupled with mass spectrometry (LC-MS). Amphetamines, tetrahydrocannabinol (THC) and opiates were the drugs of abuse examined in this study. Results: In total 102 whole blood samples were collected and processed from 102 decedents between March – May 2014 and subsequently analysed. Randox and ND performed well when compared to LC-MS. The poorest performance was in the diagnosis of THC, where the specificity of Randox was low (84%) and the sensitivity of ND was very low (28.6%). Although the confidence intervals were wide positive predictive values for THC were also poor using both Randox (32%, 95% CI 0.15-0.55) and ND(40%, 95% CI 0.07-0.83). However, negative predictive values for THC were good with narrow confidence intervals using both Randox (100%, 95% CI 0.94-1) and ND(94%, 95% CI 0.88-0.98) . Negative predictive values for opiates (ND - 98% and Randox - 94.8%) and amphetamines (ND - 100% and Randox - 100%) using both ND and Randox were also good with narrow confidence intervals. The results for amphetamines should be treated with caution however, as the overall positivity rate was very low. Positive predictive values for opiates and amphetamines appeared good, but due to the low positive test prevalence in this population, the confidence intervals were wide and these values are possibly not accurate. There was significant correlation shown between LC-MS, Randox and ND using Cohen’s kappa (fair to substantial correlation) and the McNemar two-tailed test (p