A Statistical Evaluation of the Chemical Composition of Heroin Samples with the View to Discriminating Between Their Possible Sources

Abstract

The concentrations, in parts-per-billion, of 39 trace elements were measured for each of 139 heroin samples that were known to have come from either Chinese or non-Chinese sources. It was found that after dichotomizing the distribution of the concentration of each element, a statistical model based on a linear combination of only eight elements, each significant at the 20% level of significance, explained as much as 78% of the variation in the probability of a heroin sample being of a Chinese source. This model had a high ability to discriminate between heroin samples from Chinese and non-Chinese sources, with a positive predictive value of 95% and a false positive rate of 15%.

This is a preview of subscription content, access via your institution.

References

  1. 1.

    Shine JP, Ika RV, Ford TE. Multivariate statistical examination of spatial and temporal patterns of heavy contamination in New Bedford harbor marine sediments. Environ Sci Technol. 1995;29:1781–1788.

    CAS  Article  Google Scholar 

  2. 2.

    Massart L, Kaufman L. Hierarchical nonheirarchical clustering strategy and application to classification of iron meteorites according to their trace element patterns. Analytic Chem. 1982;54:911–917.

    CAS  Article  Google Scholar 

  3. 3.

    Neumann H. Comparison of heroin by capillary gas chromatography in Germany. Forensic Sci Intl. 1994;69:7–16.

    CAS  Article  Google Scholar 

  4. 4.

    Perillo BA, Klein RF, Franzosa ES. Recent advances by the U.S. Drug Enforcement Administration in drug signature and comparative analysis. Forensic Sci Intl. 1994;69:1–6.

    CAS  Article  Google Scholar 

  5. 5.

    Armitage P, Berry G. Statistical Methods in Medical Research. Oxford: Blackwell Scientific Publications; 1994.

    Google Scholar 

  6. 6.

    Mardia KV, Kent JT, Bibby JM. Multivariate Analysis. London: Academic Press; 1979.

    Google Scholar 

  7. 7.

    Collet D. Modelling Binary data. New York: Chapman and Hall; 1994.

    Google Scholar 

  8. 8.

    STATA Reference Manual Release 4, Stata Press, College Station, Texas.

  9. 9.

    MATLAB Version 4 User’s Guide. The Maths Works Inc. New Jersey: Prentice Hall; 1995.

    Google Scholar 

Download references

Author information

Affiliations

Authors

Corresponding author

Correspondence to A. Ekangaki PhD.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Ekangaki, A., Skopec, J., Skopec, Z. et al. A Statistical Evaluation of the Chemical Composition of Heroin Samples with the View to Discriminating Between Their Possible Sources. Ther Innov Regul Sci 32, 229–241 (1998). https://doi.org/10.1177/009286159803200131

Download citation

Key Words

  • Positive predictive value
  • Sensitivity
  • False positive rate
  • Logistic regression