Amended 16 June 2002: I'm sorry I did this now. I urge you not to imitate me. See below.
We hear many stories these days about Internet nuisances (viruses, spam, inescapable ads for the wireless cameras), but today the Internet was actually helpful for me, so I want to offer my story as a counter-example. Reflecting on the story, taken as a real-life use case, also suggests some implications for design of information systems.
Background: In December 2001, my daughter Michal was in her second trimester in womb. A routine ultrasound exam produced an ominous finding -- echogenic bowel. This is often insignificant (and, in her case, it was, as learned much later), but at the time we had cause for concern. Our doctor ordered several tests that could exclude some of the possible pathologies. The tests went well, but when the bill for the lab work got to my insurance provider, they refused to pay it because "the diagnosis code did not match the procedure given" -- in other words, we had no justification for having done the test. The way things work, it was now to me to resolve the differences between the insurance agency, the lab, and my doctor.
First I called the insurance, and got them to tell me what diagnosis code was on the bill. Then I called my doctor's office, and the billing person told me that that particular code (V72.6) meant "generic lab test". Since it did not explain what was being tested for or why, it made some sense that the insurance balked. The billing person at the doctor then gave me a new code (V22.0), which translates to "pregnancy". At that moment, I realized that the diagnosis codes were actually an industry standard thing, and I asked the billing person what the coding system was called (the International Classification of Diseases), and whether the dictionary of codes was online. She told me it was.
Armed with this new code, I again called the insurance. This time, the adjuster told me the new code would justify all the tests except for one, a screening for the CMV virus. It was obvious from the way he was speaking that he wasn't deciding this for himself, but rather that he was entering the code into some computer program which decided whether the test was or was not justifiable.
At this point, I realized that I needed to find a different code, so I quickly started my web browser (One benefit of DSL is being able to use the web while on the phone at the same time) and searched on Google for the ICD. Google found one at http://cedr.lbl.gov/icd9.html. While still on the phone, I searched for a diagnosis code specific to CMV, and, under "Infections specific to the perinatal period" I saw "771.1 Congenital cytomegalovirus infection". I read this code to the adjuster, who entered it into his program, and it was accepted. So that solved my insurance problem.
Why this was a bad idea:
June 16, 2002
A correspondent wrote to me to ask how I could be certain that the code I gave had the meaning I intended. The meaning I wanted was "this patient may have CMV, and we want to do the test to exclude it". But for all I know, this code really means "This patient has been confirmed to have CMV". Now to be sure, this does not much sense, since why would you test for a disease if you are already certain the patient has it? But just because it does not make sense to me, a layperson, how can I be certain. For all I know, the insurance company now believes my daughter has this disease, when in fact she does not, and this might lead to problems with them in the future. For all I know, I did the right thing, but also, for all I know, I didn't. And that's what happens when you have an elaborate specialized taxonometric code (such as ICM, or Medline, or the AACR.) A layperson may not understand it. So I may have been taking a chance with my daughter's future insurability, if not with her life. I would not do this again.
Despite all this, the lessons below still seem valid to me.
Now one could complain about the fact that insurance operates in a climate of suspicion and cost containment, so that it's up to patients and their doctors to prove that a test is needed, or about the fact that the correct code (771.1) was not provided in the first place. But instead, I want to notice that, at least in this case, the technical and social structure of the Internet allowed me to overcome these problems.
But what made this possible? Two things:
While there seem to be a number of copies of the file on the net, in this particular case the file I used was hosted at Lawrence Berkeley Labs. To judge from the home page of the web server where I found it, it was placed online as a direct consequence of US Government support. To quote from the home page: "The Department of Energy (DOE) has developed the Comprehensive Epidemiologic Data Resource (CEDR) Program to provide public access to health and exposure data concerning DOE installations." So in this case at least, I was a direct beneficiary of a public program (paid for by tax dollars) intended to deliver information to the public. This case is thus one example showing that simply putting information on the web is a good thing, even if you know nothing of the users. And it also raises a political point, but this a fine example of how "government" can actually help citizens.
There's also a point to be made about information design. The ICD list is a plain text flat file. By flat, I mean it's one large (1.4 megabyte) piece with no formatting or hyperlinks, and there is no graphic user interface or search function for it. Why was this helpful? Because once I downloaded it I could search it very quickly in an ad-hoc manner to find the code I needed in the context I needed. The designers might have instead broken the file into smaller pieces. (The file is rather large for a text file - 1.4 mbytes - and had I been using a dialup modem instead of DSL it would have been painfully slow to load.) but had they done that, I would have had to find the piece I wanted. This would likely have been slower (though who can tell for sure? a good table of contents or an index might have helped, but I still would have had to find the index and understand it.) The irony is, a simple flat plain text file is often more useful just because one can search the whole thing in ways the designers might not have anticipated.
The story also makes clear one of the potential risks of the envisioned world of "seamless" interconnected medical records. Imagine an electronic medical record shared by doctor, lab, and insurance agency, such that the diagnosis code was entered exactly once, when the test was first ordered, and then traveled to the lab, and onto the insurance agency, through digital channels never again touched by human hands. While there would be much to gain from such a system, I fear we might also lose the flexibility which, in our case at least, made it possible to correct a mistake. It's already clear that the insurance adjuster is unable to make decisions for him or herself, but can only enter codes for the consideration of the rule-based system, with no provision for overrides. But at least he retains the ability to enter a new code. But I can easily imagine that in the seamless world, the adjuster would lose this ability -- the diagnosis code once entered becomes immutable. I would suppose that there would be some authority somewhere, that could correct errors, but I also imagine that authority being busy, hard to find, and harder to convince. I wonder whether the increased efficiency would compensate for the lost flexibility.
I do not intend this story to be read as a boast about the wonder of the "Internet", where "Internet" means the technology of web servers and browsers, search engines, and DSL. True, all that was necessary, but it was not sufficient. Alas, while I can point at the other factors, I can't analyze them, but I know they are there. First are the forces that motivate people publish useful information, and second are the economics that allow Google to remain in operation.
I also feel obligated to point out that while the Internet worked for me in this case, I am not at all a typical person, compared to almost everyone else in the world. For starters, I not only have a home computer, I have a DSL connection to the net. Next, I am accustomed to dealing with bureaucracies. I know how to talk to clerical workers in a way that elicits their cooperation. On the phone, I sound like what I am, an educated, polite, middle class person. Perhaps I also come across as an honest person. I don't know about that, but I know that the insurance adjuster was cooperative enough to accept a code from me, with no proof that this code came from my doctor. In fact he didn't even ask me where I got the code. (For all I know, he deliberately chose not to ask me.) I could have been trying to cheat. I was not, but how could he tell? And finally, I could use the information only because I have enough education to understand the medical terms.
I hope this story can, in some way, help us to make the Internet, and the society it is embedded in, more just, so that the benefit I got today will become available to everyone else as well.