Limitations in the Statistical Analysis of Normalised Cigarette Smoke Analyte Yield per Milligram of Nicotine Yield

Yields of selected mainstream smoke analytes expressed per milligram of nicotine yield (nicotine ratio) and ceilings on these ratios have been proposed by WHO as part of future cigarette product regulation. This paper describes the different approaches required for precision assessment, depending on whether yields or nicotine ratios are being studied. The widely used approach of assessment of yield precision is to perform a collaborative study using a standardised method. However, for assessment of ratio precision the measurement of smoke analyte and smoke nicotine yields are often not carried out on the same set of cigarettes (unpaired) due to analytical constraints and therefore the statistical approach described in ISO 5725 is inappropriate due to the various replicate combinations. In this paper, the precision of ratios was computed with unpaired measurements for NNN and nicotine yield data for the CM6 monitor test piece and the Kentucky Reference 1R5F cigarette carried out during a collaborative study in 2011 (1). A sampling technique, based on the draw of the most representative ratios, has been used to evaluate the range of both estimated repeatability and reproducibility under the ISO smoking regime that might be expected when comparing data between different laboratories. This statistical evaluation highlighted that a robust estimate of repeatability and reproducibility could not be determined for ratios obtained with unpaired measurements, using the method defined by ISO5725-2