Assessment of multivariate logistic regression analysis in articles published in Turkish cardiology journals

OBJECTIVES
We aimed to assess the use and reporting-quality of multivariate logistic regression analysis (MVLRA) in articles published in two Turkish cardiology journals.

STUDY DESIGN
We reviewed all original articles published in two Turkish cardiology journals (the Anatolian Journal of Cardiology and Archives of the Turkish Society of Cardiology) between January 2010 and August 2011. The articles that used MVLRA were analyzed comprehensively based on 10 predefined criteria.

RESULTS
A total of 212 articles were reviewed, of which MVLRA was used in 33 (15.6%). Twenty-nine articles (%87.9) properly included the main components of the MVRLA, namely, odds ratios, 95% confidence intervals, and p values. However, none of the articles reported MVRLA-related data such as the modeling type, validation, goodness-fit, multicollinearity and interaction tests. There were severe reporting flaws and faults as to the ratio of the total number of events or sample size to the number of independent variables included into the MVLRA model, the use of fitness procedures, and how the independent variables were selected.

CONCLUSION
Our results indicate that MVLRA has become a standard statistical method in the Turkish cardiology literature. However, overall reporting of MVLRA data still has seriously inadequate and inaccurate aspects.