I am going to suggest an idea that you may not have thought of. The US NIH (National Institutes of Health) maintains a database of medical abstracts, PubMed. This LINK is the same article.

There are a huge number of tools to access this database. pubcrawler is an example.

What I am suggesting is that using an even bigger database of which the Karger articles would be a subset may be the way to go.

I've written 3 posts about PubMed: Marshall re: Pub Med. (Hit the search button when this screen appears). The node titles that I replied to are not informative to say the least! But I would suggest that you read these posts because I think that you will find some interesting pointers. Basically there are Perl tools and source code that will do what you want in an even more efficient and better way than scanning/parsing html pages.

I don't want to act as free advertisement for commercial products here, but there are professional level complete applications that do what you want and even more, but they will cost about $700.

Update:
Even this node title: "Re: web search for certain data" would not lead one to believe that what the real problem is to search medical databases. Again, the big, super big medical databases provide applications and Perl code to access them efficiently. One of my articles talks about how to be a "polite consumer".