COMMON TYPES OF BIOMEDICAL STUDIES 11in the same hospital in a time period just before the switch to the new technique. The
term historical controls is used because the control patients are treated earlier in time.
Inferences are more difficult to make in historical studies because other changes in
the treatment or the patients may occur over time.
The topics discussed in this chapter are conceptual in nature, but they can have
a profound effect on the final statistical analysis. The purpose of this book is to
provide information on how to analyze data obtained from the various study designs
so that after the analysis is completed, the data can be understood and inferences
made from it.
1.3.8 Rating Studies by the Level of Evidence
Medical studies are sometimes rated by what is called the level of evidence. One
obvious thing to rate is the type of study performed. Some medical journals favor
publishing results from studies that have a high level of evidence. There is no universal
rating system, but in general, level one studies tend to be either meta-analyses where
the results of well-designed randomized control trials are combined, well-designed
multicenter blinded randomized control studies, high-power randomized blinded con-
trol clinical trials with sufficient follow-up, or well-designed experimental animal or
bench studies. Somewhat lower-level studies may contain meta-analyses of well-
designed observational studies, randomized clinical studies with adequate power, or
well-designed observational studies with overwhelming evidence. Lower-level stud-
ies include well-designed observational studies, case reports without control, and
clinical examples. An even lower level could include expert opinion or reasoning
from first principles. In general, here power refers to the requirement that the sample
size is large enough that the conclusions reached are likely to be correct. It must
be emphasized that there is not one master list of how to do the ratings, and what is
written above may not coincide with how all journals or medical associations perform
their rating. But aspects of study design have an effect on the conclusions that can be
reached and how a study is rated. At the present time, many journals have set their
own levels and these levels are available from the editors of the journals.
1.3.9 CONSORTProper reporting of results of clinical research so that a reader can evaluate the in-
vestigator’s conclusions is an important issue. One set of guidelines is given by
CONSORT (Consolidated Standards of Reporting Trials). CONSORT provides on
their website a checklist and flow diagram to help improve the quality of randomized
clinical trials and meta-analyses. Use of this checklist has been endorsed by several
prominent medical journals. The checklist and flow diagram for randomized clinical
trials can be obtained by going to their website, www,consort-statements.org/ and se-
lecting statements to obtain the information. For meta-analysis, go to http://www.consort-
statements.org/Evidence/evidence.html#quorom for similar information. For more
information on CONSORT, see Moher et al. [2004].