Transcript How to Read Scientific Journal Articles

How to Read
Scientific Journal Articles
Major article sections
• Abstract -- brief summary of entire article
• Introduction -- includes literature review;
states why the research is relevant.
• Methods -- identifies how patients were
selected, what study procedures entailed,
and statistical methods used.
• Results -- presents objective results
• Discussion -- interprets results; states study
strengths/weaknesses; identifies future work
What population(s)?
• The study eligibility criteria define the
characteristics of the patients studied, and
therefore the population(s) to which the
results can be extended.
• The baseline characteristics define the
characteristics of the patients studied, and
therefore the population(s) to which the
results can be extended. (Often, Table I.)
Identification of groups
• Typically, one group is considered a control
group, while other groups are considered
new treatment groups.
• The groups should be similar, on average,
except with respect to the primary
treatment(s) of interest. (Often, Table I.)
• Ideally, groups will be created by randomly
assigning individuals to groups.
With respect to what are the
groups being compared?
• The main variables by which the groups are
compared are called the primary outcome
measures.
• A good study is designed with the primary
outcome measures in mind.
• The less important variables by which the
groups are compared are called the
secondary outcome measures.
How are outcome measures be
analyzed statistically?
• The type of data and how the study was
designed determine the appropriate analysis
methods.
• Categorical (binary) data are often
summarized as percentages, but also odds
ratios and relative risks.
• Numerical (measurement) data are often
summarized as means, but also medians.
Presentation of results
• Statistical results can be presented in words
or in a table.
• If a summary measure, like a percentage or
a mean, is reported without a P-value or a
confidence interval, the result only applies
to the sample of patients in the study.
Presentation of P-values
• In a table, P-values are usually indicated by
a footnote.
• In writing, P-values are usually indicated in
parentheses after the conclusion of a
hypothesis test is stated.
• It is the reader’s job to infer the null and
alternative hypotheses of the hypothesis
test.
Presentation of
confidence intervals
• Confidence intervals for a mean are
typically presented in one of three ways.
– Sample mean ± SD, (“standard deviation”)
– Sample mean ± SE (“standard error”)
– The lower and upper limits of the interval
• If “Sample mean ± SD”, the 95% margin of
error is 2×(SD÷sqrt(n))
• If “Sample mean ± SE”, the 95% margin of
error is 2×SE.
Presentation of
confidence intervals
• It is the reader’s job to know what value is
relevant to the confidence interval:
– If a confidence interval for a difference in two
means or a confidence interval for a difference
in two proportions, then is 0 in the interval?
– If a confidence interval for a relative risk, or a
confidence interval for an odds ratio, then is 1
in the interval?
– And so on ...
Evaluation of the
strengths/weaknesses of the study
• Is there a control group?
• Is the study a randomized experiment or
an observational study?
• Were the patients selected and assigned to
their treatments in an unbiased manner?
• Were the patients and physicians blinded to
treatment assignment?
Evaluation of the
strengths/weaknesses of the study
• Was the sample large enough to get useful
statistical results?
• If the sample was really large, were
confidence intervals created to show that a
rejection of the null hypothesis is actually
meaningful?
• Are the conclusions of the study overstated?
• Do the authors point out future necessary
research?