Statistical testing of baseline differences in sports medicine RCTs

A systematic evaluation

Ross L Peterson, Matthew Tran, Jonathan Koffel, Steven D Stovitz

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Background/Aim: The CONSORT (Consolidated Standards of Reporting Trials) statement discourages reporting statistical tests of baseline differences between groups in randomised controlled trials (RCTs). However, this practice is still common in many medical fields. Our aim was to determine the prevalence of this practice in leading sports medicine journals.

Methods: We conducted a comprehensive search in Medline through PubMed to identify RCTs published in the years 2005 and 2015 from 10 high-impact sports medicine journals. Two reviewers independently confirmed the trial design and reached consensus on which articles contained statistical tests of baseline differences.

Results: Our search strategy identified a total of 324 RCTs, with 85 from the year 2005 and 239 from the year 2015. Overall, 64.8% of studies (95% CI (59.6, 70.0)) reported statistical tests of baseline differences; broken down by year, this percentage was 67.1% in 2005 (95% CI (57.1, 77.1)) and 64.0% in 2015 (95% CI (57.9, 70.1)).

Conclusions: Although discouraged by the CONSORT statement, statistical testing of baseline differences remains highly prevalent in sports medicine RCTs. Statistical testing of baseline differences can mislead authors; for example, by failing to identify meaningful baseline differences in small studies. Journals that ask authors to follow the CONSORT statement guidelines should recognise that many manuscripts are ignoring the recommendation against statistical testing of baseline differences.
Original languageEnglish (US)
Article numbere000228
Pages (from-to)e000228
JournalBMJ Open Sport and Exercise Medicine
Volume3
Issue number1
DOIs
StatePublished - Jun 1 2017

Fingerprint

Sports Medicine
Randomized Controlled Trials
Manuscripts
PubMed
Guidelines

Keywords

  • randomised controlled trial
  • sports medicine
  • statistics
  • testing

PubMed: MeSH publication types

  • Journal Article
  • Review

Cite this

Statistical testing of baseline differences in sports medicine RCTs : A systematic evaluation. / Peterson, Ross L; Tran, Matthew; Koffel, Jonathan; Stovitz, Steven D.

In: BMJ Open Sport and Exercise Medicine, Vol. 3, No. 1, e000228, 01.06.2017, p. e000228.

Research output: Contribution to journalArticle

@article{fafa9bf3b50e45d3a4622626bedb9a1b,
title = "Statistical testing of baseline differences in sports medicine RCTs: A systematic evaluation",
abstract = "Background/Aim: The CONSORT (Consolidated Standards of Reporting Trials) statement discourages reporting statistical tests of baseline differences between groups in randomised controlled trials (RCTs). However, this practice is still common in many medical fields. Our aim was to determine the prevalence of this practice in leading sports medicine journals.Methods: We conducted a comprehensive search in Medline through PubMed to identify RCTs published in the years 2005 and 2015 from 10 high-impact sports medicine journals. Two reviewers independently confirmed the trial design and reached consensus on which articles contained statistical tests of baseline differences.Results: Our search strategy identified a total of 324 RCTs, with 85 from the year 2005 and 239 from the year 2015. Overall, 64.8{\%} of studies (95{\%} CI (59.6, 70.0)) reported statistical tests of baseline differences; broken down by year, this percentage was 67.1{\%} in 2005 (95{\%} CI (57.1, 77.1)) and 64.0{\%} in 2015 (95{\%} CI (57.9, 70.1)).Conclusions: Although discouraged by the CONSORT statement, statistical testing of baseline differences remains highly prevalent in sports medicine RCTs. Statistical testing of baseline differences can mislead authors; for example, by failing to identify meaningful baseline differences in small studies. Journals that ask authors to follow the CONSORT statement guidelines should recognise that many manuscripts are ignoring the recommendation against statistical testing of baseline differences.",
keywords = "randomised controlled trial, sports medicine, statistics, testing",
author = "Peterson, {Ross L} and Matthew Tran and Jonathan Koffel and Stovitz, {Steven D}",
year = "2017",
month = "6",
day = "1",
doi = "10.1136/bmjsem-2017-000228",
language = "English (US)",
volume = "3",
pages = "e000228",
journal = "BMJ Open Sport and Exercise Medicine",
issn = "2055-7647",
publisher = "BMJ Publishing Group",
number = "1",

}

TY - JOUR

T1 - Statistical testing of baseline differences in sports medicine RCTs

T2 - A systematic evaluation

AU - Peterson, Ross L

AU - Tran, Matthew

AU - Koffel, Jonathan

AU - Stovitz, Steven D

PY - 2017/6/1

Y1 - 2017/6/1

N2 - Background/Aim: The CONSORT (Consolidated Standards of Reporting Trials) statement discourages reporting statistical tests of baseline differences between groups in randomised controlled trials (RCTs). However, this practice is still common in many medical fields. Our aim was to determine the prevalence of this practice in leading sports medicine journals.Methods: We conducted a comprehensive search in Medline through PubMed to identify RCTs published in the years 2005 and 2015 from 10 high-impact sports medicine journals. Two reviewers independently confirmed the trial design and reached consensus on which articles contained statistical tests of baseline differences.Results: Our search strategy identified a total of 324 RCTs, with 85 from the year 2005 and 239 from the year 2015. Overall, 64.8% of studies (95% CI (59.6, 70.0)) reported statistical tests of baseline differences; broken down by year, this percentage was 67.1% in 2005 (95% CI (57.1, 77.1)) and 64.0% in 2015 (95% CI (57.9, 70.1)).Conclusions: Although discouraged by the CONSORT statement, statistical testing of baseline differences remains highly prevalent in sports medicine RCTs. Statistical testing of baseline differences can mislead authors; for example, by failing to identify meaningful baseline differences in small studies. Journals that ask authors to follow the CONSORT statement guidelines should recognise that many manuscripts are ignoring the recommendation against statistical testing of baseline differences.

AB - Background/Aim: The CONSORT (Consolidated Standards of Reporting Trials) statement discourages reporting statistical tests of baseline differences between groups in randomised controlled trials (RCTs). However, this practice is still common in many medical fields. Our aim was to determine the prevalence of this practice in leading sports medicine journals.Methods: We conducted a comprehensive search in Medline through PubMed to identify RCTs published in the years 2005 and 2015 from 10 high-impact sports medicine journals. Two reviewers independently confirmed the trial design and reached consensus on which articles contained statistical tests of baseline differences.Results: Our search strategy identified a total of 324 RCTs, with 85 from the year 2005 and 239 from the year 2015. Overall, 64.8% of studies (95% CI (59.6, 70.0)) reported statistical tests of baseline differences; broken down by year, this percentage was 67.1% in 2005 (95% CI (57.1, 77.1)) and 64.0% in 2015 (95% CI (57.9, 70.1)).Conclusions: Although discouraged by the CONSORT statement, statistical testing of baseline differences remains highly prevalent in sports medicine RCTs. Statistical testing of baseline differences can mislead authors; for example, by failing to identify meaningful baseline differences in small studies. Journals that ask authors to follow the CONSORT statement guidelines should recognise that many manuscripts are ignoring the recommendation against statistical testing of baseline differences.

KW - randomised controlled trial

KW - sports medicine

KW - statistics

KW - testing

UR - http://www.scopus.com/inward/record.url?scp=85044623224&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85044623224&partnerID=8YFLogxK

U2 - 10.1136/bmjsem-2017-000228

DO - 10.1136/bmjsem-2017-000228

M3 - Article

VL - 3

SP - e000228

JO - BMJ Open Sport and Exercise Medicine

JF - BMJ Open Sport and Exercise Medicine

SN - 2055-7647

IS - 1

M1 - e000228

ER -