Compositional data analysis

Our plain-English summary of Compositional data analysis, by L. Barabesi, P. Sartorelli (Occupational and Environmental Medicine, 2020).

The study

• Authors: L. Barabesi, P. Sartorelli
• Journal: Occupational and Environmental Medicine
• Published: 2020;77(5):352
• Type of article: Letter, Comment
• Cited by: 3 articles, as counted by NIH iCite
• Read it: PubMed · Publisher

In plain English

This is a letter to the editor, not a research study. It reports no new patients or data. The authors, from the University of Siena in Italy, criticised the statistics in an earlier study by Keskitalo et al. That study looked at patients with asbestosis. It examined how the mix of cell types in fluid washed from the lungs (bronchoalveolar lavage fluid) related to smoking, to the number of asbestos bodies, and to survival. The letter argued that these cell counts were percentages adding up to 100, a type of data called compositional data. The authors wrote that standard statistical methods give biased results with such data, and that Keskitalo et al did not use the methods designed for it. They raised two further concerns. Eosinophil counts were likely to include zero values. The survival analysis should also have allowed for this type of data, for example with a Cox regression method described by McGregor et al. The authors concluded that Keskitalo et al's conclusions were not supported by a sound statistical analysis.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Letter to the editor commenting on the statistical methods of a previously published study; no original data
• Who was studied: No population was studied. The letter discusses a study by Keskitalo et al of patients with asbestosis.
• Exposure: Not studied in this letter. It discusses the counts of asbestos bodies and bronchoalveolar lavage fluid cell differential counts used in the Keskitalo et al study.
• Main results:
  • The authors pointed out that the cell differential counts in the Keskitalo et al study were percentages summing to 100, which makes them compositional data
  • Quoting a statistics textbook, the authors stated that applying standard statistical methods to compositional data leads to biased results
  • The authors stated that Keskitalo et al did not use techniques designed for compositional data
  • The authors noted that eosinophil counts were likely to include zero values because measurements often fall below the detection limit, and that zero values may cause problems for inference
  • The authors stated that the survival analysis should account for the nature of the data, and pointed to a Cox regression formulation by McGregor et al for compositional covariates
• The authors' conclusion: The authors concluded that the conclusions of Keskitalo et al were not supported by a sound statistical analysis

Topics

The National Library of Medicine indexes this study under subjects our site covers: Asbestosis, Lung cancer.

Related studies

• 'Compositional data analysis' by Barabesi and Sartorelli: Authors' response — R. Kaarteenaho, E. Keskitalo (2020)
• Mesothelioma trends in the United States: an update based on Surveillance, Epidemiology, and End Results Program data for 1973 through 2003 — B. Price, A. Ware (2004)
• Ordinal regression models for epidemiologic data — B.G. Armstrong, M. Sloan (1989)
• Time trend of mesothelioma incidence in the United States and projection of future cases: an update based on SEER data for 1973 through 2005 — B. Price, A. Ware (2009)
• Integrative analysis of microRNA, mRNA and aCGH data reveals asbestos- and histology-related changes in lung cancer — P. Nymark, M. Guled, et al. (2011)
• DREM 2.0: Improved reconstruction of dynamic regulatory networks from time-series expression data — M.H. Schulz, W.E. Devanny, et al. (2012)

Browse more research in Occupational and Environmental Medicine, or return to medical research.