These tools will no longer be maintained as of December 31, 2024. Archived website can be found here. PubMed4Hh GitHub repository can be found here. Contact NLM Customer Service if you have questions.


Pubmed for Handhelds

PUBMED FOR HANDHELDS

Search MEDLINE/PubMed

  • Title: Statistical analysis of joint short-term and long-term survival in resuscitation research.
    Author: Babbs CF.
    Journal: Resuscitation; 2007 Nov; 75(2):323-31. PubMed ID: 17583410.
    Abstract:
    OBJECTIVE: To develop statistical tools that use combined initial survival data and post-resuscitation survival data to test the null hypothesis that true, population-wide outcomes following experimental CPR interventions are not different from control. METHOD: A new test statistic, d(2), for evaluating Type 1 error is derived from a bivariate, two-dimensional analysis of categorical initial resuscitation and post-resuscitation survival data, which are statistically independent because they are obtained during non-overlapping periods of time. The d(2) test statistic, which is distributed as a chi-squared distribution, is derived from first principles and validated using Monte Carlo methods of computer simulation for thousands of clinical trials. RESULTS: Under the null hypothesis, the normalized difference in the proportions of patients surviving the initial resuscitation period and the normalized difference in the proportions of such short-term survivors that also survive the post-resuscitation period are jointly distributed in a two-dimensional space as a bivariate standard normal distribution, against which observed intervention and control outcomes can be compared in a test of statistical significance. Typically this two-dimensional approach has greater statistical power to detect true differences, compared to conventional one-dimensional tests. Smaller group sizes (Ns) are usually required to reach statistical significance when both initial survival and post-resuscitation survival are considered together. Such two-dimensional analysis is easily extended to meta-analysis of multiple trials. CONCLUSIONS: A straightforward, easy-to-use bivariate test for Type I errors in statistical inference can be done for resuscitation studies reporting both short-term and long-term survival data. Acceptance of such two-dimensional tests of the null hypothesis, as proposed by Hallstrom, can save time, money, effort, and disappointment in the difficult and sometimes frustrating field of resuscitation research.
    [Abstract] [Full Text] [Related] [New Search]