Skip to Main Content

Course Support: Pediatrics Clerkship

Where to search: choosing databases

The most commonly-used databases for answering EBM questions are shown on the EBM Homepage tab of this LibGuide.

No two databases are exactly the same, and which one(s) you should use in a given search can depend on a variety of factors.

When using a database for the first time, look at the Help page, About Us page, or FAQs for information on:

Database coverage

  • Subject
  • Time
  • Geography
  • Type of publications included

Database search features

  • Subject headings
  • Advanced search features
  • "Related articles" links

In addition, there are other features that - while they shouldn't necessarily determine whether you use a particular database - can be of great use while searching:

  • An option for creating a personal account where you can save searches/search results, or create alerts
  • A facility for exporting search results to bibliographic management software, e.g. EndNote, RefWorks

What to search for

1. Bibliographic databases

The databases you will be most likely to search for EBM purposes are bibliographic databases: organized, digital collections of records of publications, including journal articles, reports, conference proceedings, etc.

Some databases search within the full text of items, but most search only abbreviated records containing the title, author(s), abstract, publication details, subject headings, etc. of an item.

The important thing to remember is that these records are composed of words.

 

2. Think about the words

When thinking about a search, think about the words that you are likely to find in the most relevant journal articles, reports, etc:

Identify the concepts in your research question and search for words describing those concepts

See Keyword Searching for more about this.

Don't just type in your research question

Unless the database has an option for doing this, such "natural language" searching can introduce unnecessary words that make a search less focused.

Use words that are likely to be used in formal scientific communication

For instance, research articles discussing obesity are unlikely to use the word "fat" to describe obese individuals. "Fat" is more likely to be used to describe a class of organic compounds or a food group.

If needed, use words appropriate to the kind EBM question you are asking

Diagnostic studies: search for sensitivity, specificity, etc.

Therapeutic studies: search for RCT, double-blind, etc.

 

3. Look at the results

Critically assess the the results you obtain from a search:

Are the results on-topic?

If not, which of your search terms is finding the wrong results?

Did you get 0 (zero) results?

Check your spelling, or remove one (or more) of your search terms (see Keyword searching for more).

Do you get thousands of results?

Check to see if they're on-topic; if they are, you may need to narrow your search (see Keyword searching for more).

Do the most relevant results use words that you haven't searched for?

Searching for them may improve your results.

 

4. Search again

It's normal to repeat a search several times, refining it each time based on the results from previous attempts.

What to search for: types of study

When searching, bear in mind what you would actually like to find: ask yourself what would the best kind of evidence to help you make a clinical decision in this situation?

The "evidence pyramid"

Some types of study generally provide better/stronger clinical evidence, based on their design. The classic summary of this is known as the "evidence pyramid:

EBM evidence pyramid

  • "Filtered information" consists of several kinds of review studies that "filter" previous individual clinical studies for quality and relevance, then combine their results. These are considered the best evidence.
  • N"Unfiltered information"refers to individual clinical studies. Generally, randomized controlled trials (RCTs) are considered the best design for answering clinical questions, followed by cohort studies then studies of a few or individual cases.
  • The type of evidence considered least reliable is the opinion of an expert - not because experts have nothing to contribute, but because they are prone to bias or limitations in their experience.

Types of clinical study and types of question

While RCTs are considered the best type of clinical design for testing a hypothesis, they are not always possible or ethical. In practice, the best clinical evidence for different types of clinical question (diagnosis, prognosis, treatment, or harm), is often supplied by a different type (or types) of study:

Type of question Suggested best type(s) of study in order of preference
Treatment RCT > Cohort study > Case control > Case series
Diagnosis Prospective, blind comparison to gold standard
Etiology/harm RCT > Cohort study > Case control > Case series
Prognosis Cohort study > Case control > Case series

 

More information:

For more information on study types, see Study Design 101 from Himmelfarb Health Sciences Library.