young, healthy woman died this year because of a poor literature search.
Whose fault was it? And, more important, is she the first and the last?
The details of
the tragic event were covered in the national press, as well as a Newsbreak
written by Eva Perkins and published on Information Today Inc.'s Web site
["Johns Hopkins Tragedy: Could Librarians Have Prevented a Death?," http://www.infotoday.com/newsbreaks/nb010806-1.htm].
(In next month's issue of Searcher, searcher Perkins will do a follow-up
on that incident as part of a broader look at the subject.)
happened, however, is that a healthy, 24-year old woman named Ellen Roche
was given a large dose of hexamethonium as part of a project at Johns Hopkins
University, one of the nation's most prestigious medical research establishments.
The study was designed to show how healthy lungs coped with problems that
asthmatics face. Apparently, the amount of hexamethonium administered led
to lung and kidney failure. The principal investigator, Dr. Alex Togias,
had conducted a literature search, but his search failed to find the literature
that would have warned of the dangers from hexamethonium. Internal investigations
at Johns Hopkins included attempts by other physicians to locate relevant
literature. After the fact, medical librarians testing the same files searched
by Dr. Togias, along with others, quickly found relevant literature online.
Fortunately, at press time, we had heard that medical librarians had been
added to the committee set up at Johns Hopkins to establish protocols for
medical literature searching.
What went wrong?
How can we prevent this from ever happening again? What lessons can this
teach all of us, including those of us whose online searching only affects
lesser issues, such as money, jobs, or living conditions, not to mention
truth, justice, and the American way?
First, we must
all step up with our hands in the air to take our share of the blame. All
of us — information professionals, physicians and medical researchers,
database producers, and search services — had to fail for this to happen.
Professional searchers — in this case, medical librarians — apparently
have not made the potential and actual value of their contribution to the
quality of searches visible enough to their clients for the clients to
recognize the risks of working without them. Physician researchers overestimated
their effectiveness as literature searchers and didn't compensate for any
defects in their searching abilities by using professional networking to
double- and triple-check their research. Database producers failed to build
files and interfaces that would have found the needed information.
Bottom line: All
of us must recognize that we now live and will forever live in an end-user
searching world. We searchers can neither expect nor demand that people
come to us to perform intermediated searches to guarantee quality results.
We can — and should — urge and warn clients. We can — and should — train
clients in critical, effective searching techniques. But, ultimately, these
are all stopgap measures. The true solution to the problem lies in building
better end-user interfaces, finding the best data, and accommodating its
delivery to the needs of our users. As professionals, we must also plan
to share the work we do in creating better systems. Our professional ethics
require that we do our best to make information safe for as many people
as possible, even outside our own client constituencies.
The database industry
must come to the same realization. The industry has to stop giving us "not
what we need, but what they've got a lot of," in the immortal words of
Eugenie Prime, Hewlett-Packard's library director. Dr. Togias used PubMed,
the National Library of Medicine's Medline service aimed at end users.
Apparently, it wasn't enough. Did he know that it takes two searches to
push the search back to the mid-1960s? Some of the original research on
hexamethonium came from the 1950s. Did he know that other access points
for Medline, e.g., Medscape, take it back to 1960? Even if he did reach
the entire file, would he have known that the earliest literature would
require him to use the descriptor "Methonium Compounds"?
I doubt he knew
all this. And the answer to the problem is NOT to teach him. One cannot
expect the entire community of physicians, whether medical researchers
or practitioners, to acquire the exotica of Medline searching. And what
about the people searching the file who have no medical background, namely
patients or their concerned friends and families?
The database industry
face the fact that its data is being and will be used by people who know
little or nothing of the details of its construction. Nor will these end
users take instruction in those details. The industry must identify what
users seek when they approach the databases and arrange the data to meet
NLM is not the
only player on the line here. In the course of checking out hexamethonium
myself for curiosity's sake, I fooled around with PubMed first. Then I
turned to the PDR, the Physicians Desk Reference, a pharmacological
bible for identifying negative side effects. Not possessing a copy of the
printed text, I went online to Yahoo! and found a link to PDR.net. For
a mere $9.95, I bought a month's worth of access and searched on hexamethonium.
No results. I am not a medical or pharmaceutical searcher. So I poked around
on the files included in my $9.95 and finally figured out that these sources
covered mainly prescription drugs in use. Hexamethonium is mainly used
experimentally. PDR.net offered a set of databases like Medline, Toxline,
etc., which required premium access where I might have found the broader
literature sources and basic research. But once thrown into those files,
I was back to where I started when it came to tools. PDR.net would not
pre-slice the premium files it did not produce to pull out the "adverse
effects" or "side effects" or identify the experimental nature of drugs
not included in its main reference works. But in this age of end-user searching,
that is exactly what PDR.net must do to maintain its effective role as
an authority on negative side effects of drugs.
All database providers
must concentrate on establishing what their users expect from them and
then meet those expectations. The linkability of resources through the
Web has made end users believe that everything is available if only the
Web site they use has the will and wit to find it. Of course, we information
professionals — industry or consumer — all know that this is a lot harder
to do than it looks to amateurs. But this doesn't matter. We can no longer
ignore those false expectations. We must build systems to meet them. Ultimately,
the users — naïve though they may seem by our professional standards
— are right. It is possible to link just about everything off the Web.
There are no implacable technological barriers. It is a question of will
But first, we must
know what our users expect. Many years ago, I wrote an editorial called,
"Was It Good for You?" (Database Searcher, vol. 5, no. 7, July-August
1989, pp. 4-6; available in Gale Group's Trade and Industry database on
Dialog, Factiva, LexisNexis). In that editorial, I chided the online industry
for failing to have feedback forms attached to every search result, forms
that could help the industry assess user satisfaction and success in dealing
with their products. This is no longer a matter of improving customer relations
and tweaking system design. This has now become essential for database
providers to insure the effectiveness of their systems. Professional searchers
should design these feedback mechanisms into intranet interfaces, collect
and analyze the input, and then pressure vendors to respond, acting both
as representatives of their individual institutions and as a profession
through working and sharing with colleagues to present a united front.
It's no longer
just a matter of getting our money's worth. It's now a matter of life and
death. And now when it comes to the blame game, if any of us are not part
of the solution, we're part of the problem.