Sie sind auf Seite 1von 1

Journal of Clinical Epidemiology 74 (2016) 237e240

LETTERS TO THE EDITOR

Subjectivity is a strength: a comment on ‘‘an *Corresponding author. Tel.: 1-507-266-1754; fax: 1-507-284-
algorithm was developed to assign GRADE 0399.
levels of evidence to comparisons within E-mail address: gionfriddo.michael@mayo.edu
systematic reviews’’
Pollock et al. [1] recently published on their use of an
algorithm to facilitate GRADEing of evidence in system- Reference
atic reviews. This approach is problematic. Although the
[1] Pollock A, Farmer SE, Brady MC, Langhorne P, Mead GE,
Grading of Recommendations Assessment, Development
Mehrholz J, et al. An algorithm was developed to assign GRADE
and Evaluation (GRADE) process helps to identify aspects levels of evidence to comparisons within systematic reviews.
of a body of evidence which may increase or decrease our J Clin Epidemiol 2016;70:106e10.
confidence in the evidence, it does not by itself inform us of
the confidence one should have in the evidence. Confidence http://dx.doi.org/10.1016/j.jclinepi.2015.11.019
is a judgment and is subjective and sensitive to the needs of
the individual evaluating the body of evidence. Thus, the
rating of the overall body of evidence may differ depending Rating the quality of evidence is by necessity
on who is evaluating the body of evidence and for what pur- a matter of judgment
pose. What is important is that judgment and the reasoning
behind it be transparently reported. The subjectivity To the Editor:
involved in GRADEing of evidence is an essential feature
of the GRADE system as it allows flexibility and reflects We read with interest the article by Pollock et al. [1]
that different stakeholders have different values. The devel- about simplifying grades of recommendations, assessment,
opment of algorithms to facilitate GRADEing simplifies an development and evaluation (GRADE) and making it more
inherently complex process and does not do justice to the objective. Although this is a laudable goal, we have major
complexity of the judgments involved. For example, they concerns about the proposed algorithm and the subsequent
are using sample size to presumably assess precision misguided conclusions.
when evaluation of the confidence interval and judgment First, it needs to be understood that primary contribution
of whether it includes appreciable harm or benefit would of GRADE is the transparency of approach, which allows
be a more appropriate way to assess precision. Although readers to understand the rationale for the judgments
their goal is laudable, oversimplification is not the answer. made, even if they disagreed with the perspective taken.
End users of the evidence, whether they be clinicians, pa- That being said, GRADE has shown to lead to reproducible
tients, or guideline panel members, need to understand results when applied by independent assessors with suffi-
the confidence they can place in a body of evidence before cient competency in GRADE [2]. Second, the relative
applying it or making recommendations, but the precepts of importance of the various domains of quality of evidence
evidence-based medicine apply. Evidence alone is not varies depending on the situation. For example, lack of
enough. blinding may be important for certain outcomes (e.g.,
pain or quality of life) and less important for other out-
Michael R. Gionfriddo comes (e.g., all-cause mortality). Similarly, interpreting
Knowledge and Evaluation Research Unit the I2 as a measure of inconsistency requires careful,
Mayo Clinic content-specific judgment. For example, setting an arbitrary
200 1st Street SW cut-off for not rating down for an I2 value of less than 75%,
Rochester, MN 55905, USA as the authors suggest, is neither methodologically sound
Mayo Graduate School nor appropriate in all situations. Third, it is problematic
Mayo Clinic and misleading to include the quality of a systematic review
200 1st Street SW, Rochester, MN 55905, USA

DOI of original article: http://dx.doi.org/10.1016/j.jclinepi.2015.11.019.


Conflict of interest: The authors disclose that they are members of
DOI of original article: http://dx.doi.org/10.1016/j.jclinepi.2015.08.013. GRADE and the US GRADE Network. They have no financial conflicts
Conflict of interest: None. of interest.

0895-4356/Ó 2016 Elsevier Inc. All rights reserved.

Das könnte Ihnen auch gefallen