Statistical Modeling for Quality Assurance of Human Papillomavirus DNA Batch Testing

J Low Genit Tract Dis. 2018 Jul;22(3):219-224. doi: 10.1097/LGT.0000000000000391.

Abstract

Objectives: Our objective was to simulate the distribution of human papillomavirus (HPV) DNA test results from a 96-well microplate assay to identify results that may be consistent with well-to-well contamination, enabling programs to apply specific quality assurance parameters.

Materials and methods: For this modeling study, we designed an algorithm that generated the analysis population of 900,000 to simulate the results of 10,000 microplate assays, assuming discrete HPV prevalences of 12%, 13%, 14%, 15%, and 16%. Using binomial draws, the algorithm created a vector of results for each prevalence and reassembled them into 96-well matrices for results distribution analysis of the number of positive cells and number and size of cell clusters (≥2 positive cells horizontally or vertically adjacent) per matrix.

Results: For simulation conditions of 12% and 16% HPV prevalence, 95% of the matrices displayed the following characteristics: 5 to 17 and 8 to 22 total positive cells, 0 to 4 and 0 to 5 positive cell clusters, and largest cluster sizes of up to 5 and up to 6 positive cells, respectively.

Conclusions: Our results suggest that screening programs in regions with an oncogenic HPV prevalence of 12% to 16% can expect 5 to 22 positive results per microplate in approximately 95% of assays and 0 to 5 positive results clusters with no cluster larger than 6 positive results. Results consistently outside of these ranges deviate from what is statistically expected and could be the result of well-to-well contamination. Our results provide guidance that laboratories can use to identify microplates suspicious for well-to-well contamination, enabling improved quality assurance.

MeSH terms

  • Adult
  • Algorithms
  • DNA, Viral / genetics
  • DNA, Viral / isolation & purification*
  • Diagnostic Errors
  • Early Detection of Cancer / methods*
  • Female
  • Humans
  • Middle Aged
  • Models, Statistical
  • Molecular Diagnostic Techniques / methods*
  • Papillomaviridae / genetics
  • Papillomaviridae / isolation & purification*
  • Papillomavirus Infections / diagnosis*
  • Papillomavirus Infections / epidemiology
  • Prevalence
  • Quality Assurance, Health Care*
  • Specimen Handling / methods*

Substances

  • DNA, Viral