Linda Gorman, who runs the Health Care Policy Center at the excellent Independence Institute in Golden, Colorado, writes in with some important caveats over computerized medical records:

Medical records were begun as a simple reminder to clinicians about how they had been treating patients. That is why they exist. They then got embroiled in legal proceedings, Medicare billing schemes, and a number of half-baked quality improvement plans. The proper response to most of these plans should be something along the lines of “contact me when you’ve figured out an adequate quality metric.”
Medicare has had at least 3 big-shot commissions trying to define how a standardized medical record should look so that it can use them for billing and auditing purposes. These have failed, in part because what is useful for billing is not the same as what is a set of time saving reminders for clinical practice. [link] Some physicians have resorted to keeping multiple sets of records. An InformationWeek overview of progress in the US is attached.
IT is hugely, grossly expensive and is not secure. The IT introduction in the UK is failing. It doesn’t work. It has cost billions. [link] It is not secure [link].The VA patient record system has problem that are not widely discussed. [link] According to the GAO, the cost savings claims made for it cannot be substantiated. It is cumbersome, produces fat documents that waste physician time flipping through pages to get to the relevant data, and people get around the time consuming interface by cutting and pasting medical histories. This introduces errors. Legally, it is not clear who has the ability to change the record to correct those errors. This is a huge sticking point due to the legal process. An anti-Kaiser group in California routinely accuses Kaiser of losing or changing its medical records in medical malpractice cases. The costs and benefits of huge nurse time spent charging are also not clear.
People who study human factors are finding that computerized order entry changes work flows and takes time. It also introduces other errors. [for quickie examples, link, link]
Although companies like Walgreen’s have complete records of all the prescriptions they fill for you, they aren’t necessarily useful to physicians as they don’t tell the doctor which drugs the patient is actually ingesting. Billing records may or not reflect medical history due to the procedure coding games required by Medicare’s thousands of price controls. [sample link]
Electronic medical records are already available—concierge docs provide them all the time. You give the doc a flashdrive or a CD. He loads his acrobat files on you onto your media. Presto, health IT without big expensive centralized systems that can be hacked into. Plus, the storage is on a fairly robust medium that has been time tested by teenagers and can be easily converted to one of the most robust there is—ink on paper.
This has served physicians well for centuries. Parts of it will be computerized. Parts of it already are. But this is not considered adequate because patients would control access to the data. The real push behind EMRs is coming from vendors and people who want government to control medical care. Only with a centralized records system can the government rationing system snoop into everyone’s record and control physicians who are not abiding by its practice guidelines.
Since at least the Clinton administration, everyone has understood that this is central to running a national health care system.

Fair enough. I also understand that vendors are lobbying hard for the stimulus provisions, so this is likely just yet more corporate rent-seeking. And if the British NHS is doing it, that should be a big warning to America NOT to do it . . .