Home Member Services Library Orthopaedic Informatics Finding Orthopaedic Information on the Internet - Part III Using Search Engines
Finding Orthopaedic Information on the Internet - Part III Using Search Engines PDF Print E-mail

Myles Clough, M.D., FRCSC
Clinical Instructor University of British Columbia
Kamloops, BC

In previous issues of the COA Bulletin, we have looked at two ways of finding orthopaedic information on the Internet. One article discussed using the PubMed interface to search the Medline database of journal articles (Part I). The second article explored using the textbook and index sites (Part II). Using search engines is a popular and seemingly easy way to find information on the Internet, however, it is much less predictable than the other two methods and is responsible for much of the bad press the Internet receives from serious users.  

A search engine is a database of links or addresses of pages on the Internet. In addition to the address, the database stores the page title and description, the keywords defined on the page and the words that appear on the page. When you submit a search, the search engine constructs a "relevance" numeric for every page which is a "hit". This number is based on where the search string is found in the page, how many times it occurs and what percentage of the text on the page is made from the search string words. If the search string has more than one word, the proximity of those words to each other also affects the "relevance". Each search engine has a secret "relevance algorithm" which it uses to put all the sites it finds in order. The average search turns up thousands of pages and no one looks much further than the first few. So, constructing a page which is attractive to search engines is both extremely desirable if you want to be "found" and very complex. That translates into "expensive" and pages that are consciously designed to turn up high in the search engines' lists cost a lot. Pages with large amounts of orthopaedic information posted by well intentioned but amateur Orthogeeks are low in the search engines' lists. Glossy pages selling snake oil and dubious ideas are higher. In this context, the most successful orthopaedic page is the American Academy of Orthopaedic Surgeons (AAOS) web site. It has a list of keywords that includes the highlights of orthopaedics (and orthopedics). Nearly all the text and links that appear on the page are derived from Javascript. The only text which would be scanned by a search engine is a short paragraph with the AAOS address and a copyright notice. The word "orthopaedic", which occurs 3 times in two sentences, is a large part of this short text. This, in turn, means that the site is scored highly in any search for "orthopaedic". Most of the providers of reputable orthopaedic information on the Internet do not take this sort of trouble and so their pages are not easy to find. However, there are ways to make your searches more successful. They amount to careful construction of the search string and a certain amount of trial and error.

Search Engines
Tables 1 and 2 give the addresses of popular search and metasearch engines respectively. The search engines scan a database which is sometimes only a small proportion of the Internet. Metasearch engines scan several search engines and provide a list of sites which score highly on more than one of these search engines. The tables also show the number of "hits" for a popular search string (orthopaedic). Table 1 shows that even abstruse search strings (osteopetrosis) occur on thousands of pages on the Internet.


Table 1. Search Engines
Search Engine visited 29 Sept 2002 Hits for "orthopaedic" Hits for "osteopetrosis"
AltaVista http://www.altavista.com/ 153244 1810
HotBot http://hotbot.lycos.com/ 234600 3000
Northern Light http://www.northernlight.com/ 26035 465
Google http://www.google.com/ 447000 4550
FAST http://www.alltheweb.com 709163 4628

Table 2. Metasearch Engines
Engine Name and URL Search for "Orthopaedic"
Excite Metasearch http://www.excite.com/ 70
Ixquick Metasearch http://www.ixquick.com/ 45
Meteor http://www.metor.com/ 80
Vivismo http://vivisimo.com/ 160
ProFusion http://www.profusion.com/ 18
Fossick.com http://fossick.com/ 72
Metacrawler http://www.metacrawler.com/index_power.html 48

If you use these search engines you will note that the coverage of the Internet is different for each one. The mathematical expression that each uses to decide its relevance algorithm is also different. The result is that each search engine provides a different "view" of the Internet. Since there may be thousands or hundreds of thousands of pages turned up by the search, it is tempting to abandon the whole concept as confusing and useless. It is impossible to believe that you are finding a complete list of the pages you want, and very obvious that you are finding many pages you do not want! However, it's not that the information is of poor quality; the problem is that there is so much information and we don't know the quality. The solution is to learn to refine your definition of the subject you are searching for.

Boolean Logic
The Boolean logic terms OR, AND and NOT are used to expand or contract a search. In a nutshell, the result is that OR is used to expand the search AND to look at the overlapping portions between two searches, and NOT to exclude a portion of the search. See Figure 1.

Figure 1. What Boolean terms mean
A OR B means all of A, all of B including the overlap
A AND B means only the overlap between A and B
A NOT B means all of A except the overlap with B
Note that some of the definitions in Table 3 are not intuitively what you might expect.

Table 3. Examples of Boolean terms and their meaning
Boolean Term Means Example Means
AND Together with Hip AND Avascular Necrosis AVN of the hip (only)
OR Either/or Osteonecrosis OR Avascular Necrosis Either Osteonecrosis or Avascular Necrosis
NOT Excluding Avascular Necrosis NOT Hip AVN of everywhere except the hip

If you search for Osteonecrosis AND Avascular Necrosis you might expect to get all the sites that refer to Osteonecrosis AND all the sites that refer to Avascular Necrosis. Instead you would get only those sites that refer to BOTH Osteonecrosis and AVN. To increase sensitivity you would use OR. To increase specificity you would use AND and NOT.

One of the concealed features of search engines is that they apply Boolean logic to your search string if it is longer than one word. If you search for Hip Fracture some search engines will search for Hip AND Fracture (which is likely what you want) but others may search for Hip OR Fracture and give you thousands of pages on hip hop and geological fracture lines.

Strategies for Successful Searching
  1. Get familiar with one search engine and one metasearch engine. Check out the features of the "advanced" search.
  2. Define your goal eg.:
    1. Find something/anything about an abstruse subject;
    2. Review a subject comprehensively;
    3. Find the contributions of an individual or department.
  3. Make the search string as comprehensive and specific as possible. If you want specialist orthopaedic information, use jargon terms that are unlikely to be directed at the public.
  4. Use metasearch if you want a small number of sites with greater relevance.
  5. If needed, expand the search using OR Boolean logic terms.
  6. Narrow the search using AND terms.
  7. Look at the titles of the first 50 pages at least. The longer the page - the lower it will come in the list.


1. Orthopaedic management of fractures in familial hypophosphatasia

  • The search string "familial hypophosphatasia fracture fixation" yields 13 sites on Google. The first one is specifically on the subject and the remaining sites are mostly about metabolic bone disease in general with hypophosphatasia included in the account.
  • Using the metasearch engine Vivisimo, the same search string yields 22 sites including most of the ones found by Google, the ICD-9 classification of diseases and an entry to the National Library of Medicine database of inherited conditions (http://www.ncbi.nlm.nih.gov/Omim/)

2. Ilizarov treatment for clubfeet

  • The technique here is to build up an "inclusive" search string for the overall subject (clubfeet) until you are reasonably certain that most pages to do with clubfoot have been included. This broadening is done by including OR terms. Then the search is narrowed down to the subject of interest by adding AND Ilizarov. (Table 4)
Table 4. Refining the search for Ilizarov treatment of Clubfeet
Search String Number of pages found
Clubfoot 10,400
Clubfoot OR clubfeet 11,500
Talipes Equinovarus 1,490
Clubfoot OR clubfeet OR "talipes equinovarus" 12,200
(clubfoot OR Clubfeet OR "talipes equinovarus") AND Ilizarov 254
TEV 203,000 203,000
  • Note that the "inclusive" search string combines clubfoot, clubfeet and talipes equinovarus. Significant numbers of pages are added with each new term. Pages that use the jargon term "talipes equinovarus" and avoid "clubfoot" may be more directed to an orthopaedic surgeon. Unfortunately, if you add TEV to the search string you get a huge number of off-subject pages which use the same acronym, such as Turk Egitim Vakfi!
  • The 254 pages found by the final refinement of the search string include case studies and presentations, the POSNA curriculum on clubfeet and several patient information pages. Several studies with two to five year follow-up are included in the list.

3. Canadian Orthopaedic Patient Information

  • This was a really difficult search and to a large extent, unsuccessful. You can use the Google Advanced facility to restrict your search to Canadian sites. In effect, this means sites with a .ca domain name - which many Canadian sites do not have. More critically, most Canadian sites which turn up with a search string like (orthopaedic OR orthopedic) AND ("patient information" OR topic) are not on the subject. "Patient Information" is commonly used to describe clinical or demographic information about patients not information written for patients.
  • Google also has a system which restricts the search to a named university site. If you use this for a search string like orthopaedic patient information you find a number of sites in each university. However, as before, most of them are not "Patient Information" pages. We are not much more successful if we use the search engine for a specific subject. I used the Google search engine to look for patient information pages on scoliosis at English language Canadian University sites.
Search String orthopaedic patient information scoliosis Number of Pages found (number relevant)
University of Alberta 2 (1)
University of British Columbia 3 (0)
University of Calgary 4 (0)
University of Manitoba 4 (0)
University of Ottawa 0
University of Toronto 3 (0)
University of Western Ontario 0
Dalhousie University 2 (0)
McGill University 0
McMaster University 0
Whole Internet 3670

There are hundreds or thousands of scoliosis patient information pages on the Internet, as a whole. What this search shows is that very little of it is produced by Canadian Universities. Below are a few of the more interesting and complete sites.

BC Healthguide

AAOS Scoliosis Brochure

Scoliosis Surgery (Iowa Health Book)

Review of Scoliosis (Scoliosis Research Society)

Patient Guide to Scoliosis (Johns Hopkins)

IScoliosis.com (Patient support site)

Scoliosis in children and Adolescents

Scoliosis - Orthopedic Surgery Health Guide (University of Maryland)

Classification of Orthopaedic Subject Headings
The difficulty we have with using search engines raises a very important subject. Searching is frustrating because defining the subject using ordinary language is so difficult. As we have seen, there may be many ways of saying the same thing and many words or phrases that mean more than one thing. The solution is to create an agreed set of orthopaedic subject headings. Then a code signifying the subject could be included in every page. If the coding system was detailed enough, we could then expect searches to be both accurate and comprehensive. Even a poor coding system would be better than trying to use slippery words.

The difficulty lies in getting agreement over what coding system to use and then getting people to use it. The Medical Subject Heading (MeSH) code is used by the National Library of Medicine to index the subjects of medical journals. Although it is not granular enough for orthopaedics, it would be an obvious starting point. The MeSH code for scoliosis is C05.116.900.800.875 which is obviously unique and won't be confused with any other subject on the net. You might think that webmasters who want people to find their pages would use such an obvious technique. But actually there is only one page in the Google database with that number on it which is in the OCOSH coding system (Orthogate Classification of Ortho-paedic Subject Headings http://www.orthogate.com/ocosh/). This system was presented to the COA in 2000 and incorporates the MeSH classification for orthopaedic subjects but expands it. It was put forward as a topic which orthopaedic organizations should pay attention to; as yet no one seems to have taken on that challenge.

Finding orthopaedic information on the Internet is a skill which takes practice and attention to detail. There are two main domains of information, the orthopaedic scientific literature and the material posted openly on the Internet. Searching the literature is best done through the Medline database maintained by the National Library of Medicine. Abstracts to the articles are accessible and it is increasingly possible to see full text versions (although not for free). This subject was treated in detail in Part I of the Finding Information on the Internet series. There are orthopaedic index sites and a number of "textbook" sites where orthopaedic material is freely available. Part II was a guide to these sites. This section (Part III) has described ways to use search engines to access the massive amounts of information elsewhere on the Internet. The most important piece of advice is to take several moments before you begin your search to decide exactly what it is you want to find and then to create a search string that is as specific as you can make it.

Further Reading
I have prepared workshops on the subject of searching the Orthopaedic Internet and posted them at http://www.orthogate.com/guide/workshops/search/default.htm


Last Updated on Tuesday, 05 September 2006 18:58