Clinical Research Unit Seminar UAB Ophthalmology

1
Clinical Research Unit SeminarUAB Ophthalmology

• How to Read a Paper
• Assessing the Methodological Quality of Published
  Papers
• October 15, 2003
• Gerald McGwin, MS, PhD
• Cynthia Owsley, MSPH, PhD

2

• Coming Attractions????

3
Tips from the start

• Not all papers are created equally.
• A paper can be in a respected journal and still
  have serious flaws.
• Famous clinicians are not necessarily good
  scientists.
• Just because a government agency funds the
  research does not mean that it should be
  uncritically accepted.

4
Approach

• Introduction
• Methods
• Results
• Discussion/Comment

5
Approach

• Introduction
• Methods
• Results
• Discussion/Comment

6
Introduction

• Does the Introduction succinctly describe the
  clinical question of interest?
• What is the problem? (e.g., physiologically,
  structurally, biochemically, genetically)
• Prevalence of problem.
• Person, place and time.
• What is the public health issue?
• Vision impairment/blindness/mortality.
• Health-related quality of life.
• Ocular pain discomfort.
• Why is our current understanding of issue
  inadequate?
• How do previous studies inform the issue?

7
Introduction Things to Watch For

• Is the ophthalmic condition or disorder
  adequately defined?
• Is the reader told key information on the
  conditions scope and treatment options?
• Is the answer to the question already known?
• Does the author incorporate the results of
  previous studies in framing the question?
• Beware of straw men used to motivate a topic.
• Clinically, does anyone care about this issue?
• If not, why should they?
• After youve read the Introduction, and youre
  not relatively clear on these issues, the paper
  probably has a serious problem.

8
Approach

• Introduction
• Methods
• Results
• Discussion/Comment

9
Methods

• Nitty gritty
• Provides important information about validity
• Setting
• Design
• Sample size calculation
• Data collection procedures
• Analysis

10
Collection of Data

• Context? (Why)
• Objectives? (Specific hypothesis)
• Primary exposure/condition of interest?
• Causes, prevents or treats outcome
• Accurately defined and measured?
  (misclassification)
• Primary outcome of interest?
• Accurately defined and measured?
  (misclassification)

11
Collection of Data

• Type of study?
• Experimental, cohort, case-control,
  cross-sectional, ecologic)
• Could there have been bias?
• Subject selection
• Sample size
• Recall, interviewer, LTF, misclassification
• What provisions made to minimize bias?
• Were they sufficient?

12
Analysis of Data

• Methods to control confounding bias?
• Measures of association?
• Measures of statistical stability?

13
Approach

• Introduction
• Methods
• Results
• Discussion/Comment

14
Results

• Population Characteristics
• Balance (randomized studies)
• Generalizability
• Findings
• Text, Tables and Figures
• Read the legends!

15
Approach

• Introduction
• Methods
• Results
• Discussion/Comment

16
Discussion

• Are the results discussed in the context of the
  question posed in the Introduction?
• Is there mention of how these findings fit with
  similar studies?
• If different, how so and potential reasons why.
• What are the clinical implications?
• For expectations on progression/prognosis.
• For treatment options.
• For prevention.
• Are the strengths and limitations of the study
  addressed?
• What questions remain unanswered?
• Directions for future research.

17
Discussion Things to Watch For

• Discussion section should not be a simple
  re-statement of the results.
• Over generalization.
• Unaddressed confounding factors and biases not
  frankly acknowledged.
• Clinical significance is overblown or not
  mentioned.
• Make note of any proprietary interests.
• Not a death blow, but do notice.

18
Future Seminars

• Please suggest other topics youd like to hear
  about in the CRU seminar.
• E-mail us at CRU_at_uab.edu with your ideas.