OBJECTIVES: We reviewed publications with two main objectives: to describe how survival analyses are reported across medical journal specialties and to evaluate changes in reporting across periods and journal specialties. STUDY DESIGN AND SETTING: Systematic review of clinical research articles published in 1991 and 2007, in 13 high-impact medical journals. RESULTS: The number of articles performing survival analysis published in 1991 (104) and 2007 (240) doubled (17% vs. 33.5%; P = 0.000), although not uniformly across specialties. The percentage of studies using regression models and the number of patients included also increased. The presentation of results improved, although only the reporting of precision of effect estimates reached satisfactory levels (53.1% in 1991 vs. 94.2% in 2007; P = 0.000). Quality of reporting also varied across specialties; for example, cardiology articles were less likely than oncology ones to discuss sample size estimation (odds ratio = 0.12; 95% confidence interval: 0.05, 0.30). We also detected an interaction effect between period and specialty regarding the likelihood of reporting precision of curves and precision of effect estimates. CONCLUSION: The application of survival analysis to medical research data is increasing, whereas improvement in reporting quality is slow. We propose a list of minimum requirements for improved application and description of survival analysis.
OBJECTIVES: We reviewed publications with two main objectives: to describe how survival analyses are reported across medical journal specialties and to evaluate changes in reporting across periods and journal specialties. STUDY DESIGN AND SETTING: Systematic review of clinical research articles published in 1991 and 2007, in 13 high-impact medical journals. RESULTS: The number of articles performing survival analysis published in 1991 (104) and 2007 (240) doubled (17% vs. 33.5%; P = 0.000), although not uniformly across specialties. The percentage of studies using regression models and the number of patients included also increased. The presentation of results improved, although only the reporting of precision of effect estimates reached satisfactory levels (53.1% in 1991 vs. 94.2% in 2007; P = 0.000). Quality of reporting also varied across specialties; for example, cardiology articles were less likely than oncology ones to discuss sample size estimation (odds ratio = 0.12; 95% confidence interval: 0.05, 0.30). We also detected an interaction effect between period and specialty regarding the likelihood of reporting precision of curves and precision of effect estimates. CONCLUSION: The application of survival analysis to medical research data is increasing, whereas improvement in reporting quality is slow. We propose a list of minimum requirements for improved application and description of survival analysis.
Authors: Sarah Lacny; Todd Wilson; Fiona Clement; Derek J Roberts; Peter D Faris; William A Ghali; Deborah A Marshall Journal: Clin Orthop Relat Res Date: 2015-11 Impact factor: 4.176
Authors: Rasmus Oestergaard Nielsen; Michael Lejbach Bertelsen; Daniel Ramskov; Merete Møller; Adam Hulme; Daniel Theisen; Caroline F Finch; Lauren Victoria Fortington; Mohammad Ali Mansournia; Erik Thorlund Parner Journal: Br J Sports Med Date: 2018-11-09 Impact factor: 13.800
Authors: Moses M Ngari; Susanne Schmitz; Christopher Maronga; Lazarus K Mramba; Michel Vaillant Journal: BMC Med Res Methodol Date: 2021-04-27 Impact factor: 4.615
Authors: Natasha E Lane; Colleen J Maxwell; Andrea Gruneir; Susan E Bronskill; Walter P Wodchis Journal: EBioMedicine Date: 2015-11-18 Impact factor: 8.143