Reporting of Statistical Results in Prosthodontic and Implantology Journals:: P Values or Confidence Intervals?

• Autores: Dimitrios Kloukos, Padhraig S. Fleming, Spyridon N. Papageorgiou, Haralampos Petridis, Nikolaos Pandis
• Localización: The International Journal of Prosthodontics, ISSN-e 1942-4426, Vol. 27, Nº. 5, 2014, págs. 427-432
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • Purpose: Confidence intervals (CIs) are integral to the interpretation of the precision and clinical relevance of research findings. The aim of this study was to ascertain the frequency of reporting of CIs in leading prosthodontic and dental implantology journals and to explore possible factors associated with improved reporting. Materials and Methods: Thirty issues of nine journals in prosthodontics and implant dentistry were accessed, covering the years 2005 to 2012: The Journal of Prosthetic Dentistry, Journal of Oral Rehabilitation, The International Journal of Prosthodontics, The International Journal of Periodontics & Restorative Dentistry, Clinical Oral Implants Research, Clinical Implant Dentistry and Related Research, The International Journal of Oral & Maxillofacial Implants, Implant Dentistry, and Journal of Dentistry. Articles were screened and the reporting of CIs and P values recorded. Other information including study design, region of authorship, involvement of methodologists, and ethical approval was also obtained. Univariable and multivariable logistic regression was used to identify characteristics associated with reporting of CIs. Results: Interrater agreement for the data extraction performed was excellent (kappa = 0.88; 95% CI: 0.87 to 0.89). CI reporting was limited, with mean reporting across journals of 14%. CI reporting was associated with journal type, study design, and involvement of a methodologist or statistician. Conclusions: Reporting of CI in implant dentistry and prosthodontic journals requires improvement. Improved reporting will aid appraisal of the clinical relevance of research findings by providing a range of values within which the effect size lies, thus giving the end user the opportunity to interpret the results in relation to clinical practice.