Monday, June 6, 2011

Scanning the Literature

It is certainly clear that we live in an information society, where access to information is pervasive, from books and journals to online websites, blogs, podcasts and more. But given this crush of information, how do we go about gleaning the critical information from a given piece of research? Rosser (1) provides a bit of guidance on how to assess an article for its applicability to the question that drive you to find the article in the first place. He suggests you consider the following:

1. The authors: knowledge of the authors and their work may influence you to select an article for assessment.

2. Abstracts: this allows you to get a sense of the value of the article without having to read the entire piece. This saves you time in a very busy schedule. Since most journals now use structured abstracts, look for the following information to be included.
a. Objectives: are these clearly stated? If not, the article may be of little value.
b. Methods: Methods need to be clearly and transparently described. I have referred to this as: could you repeat what they authors have done based on their description of their methods? In an abstract, look to see if the authors provide information about the study population (so you know if theirs is comparable to yours). Lack of strong methods may compromise the paper and suggest significant bias exists. You should also understand study design. Common designs you will confront include:
i. Descriptive reports: These have value because they allow a hypothesis to be generated in a more rigorous fashion.
ii. Case-control studies: These are typically used to understand disease causation. These are usually retrospective, and involve examination of medical records. The idea is to go back in time, look to see if there was an exposure to some risk factor in the two groups involved, the cases and the controls. From this information, one can calculate an odds ratio, or a ratio of the odds of the exposure in the disease group divided by the odds in the unexposed group.
iii. Cohort studies: Here, instead of going back in time, we go forward, with a group of individuals none of whom have the condition of interest. They are followed over time and then we look to see how an exposure correlates to the development of condition of interest. An example here would the famous Framingham study, looking at cardiovascular risk factors that developed in members of that community, many of whom were followed for more than 40 years. From this, we can calculate risk ratios, a ratio of the risk of developing the condition in the exposed group to the risk in the unexposed group.
iv. Randomized controlled trial: a study which, as we know, is designed to allow us to test an intervention in an experimental group and compare results to a control group of some sort.
v. Meta-analysis: a method where the data from several controlled trials are collapsed into a larger group for statistical analysis. This increases power and may reveal a more accurate effect size for an intervention.

c. Results: In the abstract, the results should report number of participants, and important group statistics. The important findings should be highlighted.

d. Conclusion: these should be specific and answer the question delineated in the objective section.

This is imply information from the abstract, but paying attention to this alone can help you scan articles quickly, locate those to read in more detail, and save you time and energy.

1. Rosser WW. Looking right down to the pores. Why it is important to learn how to read journals. In: Rosser WW, Slawson DC, Shaughnessy AF. Information mastery: evidence-based family practice. Hamilton, ON; BC Decker, Inc., 2004:76-82

No comments: