Research in Focus: Abstracts of Low Back Pain Trials Often Deemed Inaccurate

  •   July 5, 2019

Full Citation: Nascimento DP, Oliveira Pena Costa L, Zoldan Gonzalez G, Maher CG, Moseley AM. Abstracts of low back pain trials are poorly reported, contain spin of information and are inconsistent with the full text: An overview study. Archives of Physical Medicine and Rehabilitation 2019. doi:https://doi.org/10.1016/j.apmr.2019.03.024.

Background

Physiotherapists use the results of randomized controlled trials (RCTs) to guide their clinical decisions. However, these results can be exaggerated due to bias in how the trial was conducted or how it was reported. Although a review of the full report of trial results is necessary for critical appraisal of the evidence, clinicians often only have access to the abstract, or may only review the full study after first screening for relevance through the abstract. Therefore, the completeness and accuracy of the abstract is essential. Despite this, appraisal of abstracts in the areas of sports injury prevention, general medicine, pharmacy, and psychology have revealed that abstracts are often inaccurate due to the reporting of incomplete information, presence of spin in how the results are presented and inconsistencies in the information reported.

This study assessed the completeness, presence of spin, and inconsistencies of information in abstracts related to low back pain research, as compared with the full text versions of the studies in question.

Methods

  • 537 potentially eligible trial reports were identified in the PEDro database
  • 470 trials were deemed eligible for inclusion, from which 200 trials were randomly selected.
  • Inclusion Criteria:
    • Full-published trial report of RCT
    • Evaluation of at least one physiotherapy intervention for low back pain
    • Published between 2010 and 2015
    • Written in English, Spanish or Portuguese
  • Exclusion Criteria:
    • Trials in the recruitment stage
    • Published study protocols
  • The CONSORT-A (15 items) and SPIN-checklist (seven items) were used to assess the completeness of information and presence of spin in both abstract and full text of included studies.
  • Comparison between abstract and full-text scores on each scale were used to assess completeness and spin.
  • Kappa coefficients were used to evaluate agreement between abstract and full-text.
  • Trial and journal characteristics were collected to assess for associations with abstract completeness and spin.
  • A Kappa coefficient >0.61 was pre-specified as the criterion for “acceptable” agreement between abstract and full text.

Findings

CONSORT Abstract scores

  • Most abstracts failed to include complete information
  • Mean CONSORT-A score of 5.1 out of 15 possible points
  • The majority failed to specify group allocation or randomization, blinding, results for each group, and estimated effect sizes.

SPIN Abstract Scores

  • 98% of abstracts had at least one item of spin, with most overstating results
  • The mean SPIN-Abstract score was 4.9 out of 7 points.
  • Most abstracts failed to mention adverse events, demonstrated selective reporting of outcome and treatment recommendations.

CONSORT-Agreement between abstracts and full text:

  • The mean Kappa coefficient for CONSORT scores was 0.20 (indicating slight to moderate agreement), well below the pre-specified criterion.
  • No abstracts and only 3% of full texts reported all 15 CONSORT items.

SPIN-Agreement between abstracts and full text:

  • The mean Kappa coefficient for SPIN was 0.49 (fair to moderate agreement), below the pre-specified criterion.
  • Only 2% of abstracts and 17% of full texts scored zero for the SPIN score, meaning they were free of spin.

Associations with journal and trial characteristics:

  • Negative results were associated with the presence of spin in the abstract.
  • High CONSORT scores in abstracts were associated with:
    • Higher journal impact factors
    • Journal endorsement of CONSORT recommendations
    • Greater number of words in the abstract
    • Higher PEDro scores
    • Lower SPIN-Abstract scores
  • Low spin scores in abstracts were associated with:
    • Multicenter trials
    • Greater number of words in the abstract
    • Higher PEDro scores

In summary, the authors noted that the included trials demonstrated lower completeness of reporting in the abstract and more spin than trial reports from other areas of health care. “Spin of information can negatively impact health care as abstracts are commonly used by health-care professionals to inform treatment decision and are widely disseminated…. Whatever the cause of spin, this misinterpretation can potentially damage clinical practice and the integrity of research.”

Limitations

  • Assessment of spin is somewhat subjective and difficult to evaluate.
  • Abstracts that did not clearly define primary outcomes were automatically classified as containing spin, which may have over-estimated spin.

Relevance to physiotherapist practice in Alberta

Evidence-informed practice is an expectation of physiotherapists. To successfully integrate research evidence into treatment planning, physiotherapists must first critically appraise the evidence available to them. It is therefore essential that physiotherapists have the skills to critically appraise evidence and are cognizant of issues in academic publishing such as publication bias, incomplete reporting and spin. This study highlights the issue of spin and serves as a reminder that physiotherapists must review the research methods of studies as part of the critical appraisal process, rather than relying on the abstract or author’s conclusions to guide their clinical decision-making.

Disclaimer: The purpose of this summary is to highlight recently published research findings that are not openly accessible. Every effort is made to ensure accuracy and clarity of the summary. Readers are encouraged to review the published article in full for further information.