Harvard Says Computers Don't Save Hospitals Money

Lucas123 writes "Researchers at Harvard Medical School pored over survey data from more than 4,000 'wired' hospitals and determined that computerization of those facilities not only didn't save them a dime, but the technology didn't improve administrative efficiency either. The study also showed most of the IT systems were aimed at improving efficiency for hospital management — not doctors, nurses, and medical technicians. 'For 45 years or so, people have been claiming computers are going to save vast amounts of money and that the payoff was just around the corner. So the first thing we need to do is stop claiming things there's no evidence for. It's based on vaporware and [hasn't been] shown to exist or shown to be true,' said Dr. David Himmelstein, the study's lead author."

Comments

Transferability

[oldhack]

Well, that's mouthful, but with electronic records you can at least switch doctors without having to take X-rays, tests, and other records again. No?

Re:Transferability

[arth1]

I would like to see him try to do0 the mapping of a genome with paper and pencil.

Yes, that would be as unbelievable as designing the Eiffel tower without a computer.
The number of pieces in the Eiffel tower is in the same order of magnitude as the number of human genes (the bolts can be compared to base pairs). Clearly not doable.

The old saying that "when all you have is a hammer, every problem starts looking like a nail" is true for computers too. We think of a computer solution first without even considering how something can be done other ways.

Re:Transferability

[MattSausage]

As a personal anecdote, computerized medical records most likely saved my father's life.

He woke up in the middle of the night with unbearable pain in his abdomen, and when he didn't fight my mom about going to the hospital she knew it was serious. Five hours later and untold scans and prodding later (not to mention a significant amount of morphine) they determined a blood vessel leading to my father's colon had been blocked and basically part of his colon was dead or dying. Not having the facilities or expertise to handle the necessary surgery in my mid-size town of Owensboro, KY, they sent him to the University of Louisville Medical center telling us that basically, even done by the best surgeons in the business there is a 3 out of 4 chance he wouldnt' make it through surgery unless it was done in the next few hours. The trip to Louisville takes two hours. Following the ambulance we arrived in Louisville in an hour and a half, and he was in his room and being prepped when the doctors realized they didn't have any scans of his abdomen, the EMTs had left them behind in their rush to get us all there. Faxing them was the only option, and to do that they would have to get in touch with someone in Owensboro, convince them who they were, have them look up the records, and then get them to a fax machine that could handle the scans. It could take another hour.

Except my mother (who, frankly, is the smartest person I know) insisted that she be given a copy of the CD with all the electronic scans and data the doctors had collected that morning. I thought the surgeon was going to kiss her. 20 minutes later my father is in surgery, and 5 hours after that (or so, it was a long day) he was back out and kept in ICU for two weeks before finally coming home. The doctor had said if he'd had to wait much longer chances would have shot up considerably that my father would have died. So, there is at least one example of how Electronic Medical Records did help a doctor save a life.

I work in a major hospital

[Anonymous Coward]

And have significant responsibilities for patient care and management. Computers have made my life much easier. With electronic charting I can follow all of my patients directly from a terminal that I carry with me. The charting software we have includes basic spreadsheet and summary functions that are highly customizable. I am able to track trends and make decisions for my patients based on sight and intuition rather than having to sort through paper charts and bad handwriting. Its all at my fingertips. I don't know where Dr. harvard did his research but maybe he just has bad software. My computer system is outstanding and I honestly don't know if I'll ever be able to work in another hospital.

The key being ...

[devloop]

"IT systems were aimed at improving efficiency for hospital management"

Doctors and other medical personnel do not typically hold much power
when it comes to IT.

Software vendors aim to please management, they are the ones who take
the purchasing decisions.

Your typical Lab software for example might not have a straightforward
way to cross-check isolates for emerging resistance trends,
run critical screens or automatically report to a global EPI database,
but it sure has 1,000 ways to generate Aging Reports and auto resubmit insurance claims.

Re:The key being ...

[malkavian]

Wow, in the hospital where I work, the doctors frequently turn up to the IT department saying how they've just bought in a new system and they need it supported. If they get told 'no', they complain to the directorate that IT aren't supporting a system based on IT. The directorate lean on IT (with not so veiled threats) until IT support a system they'd have vetoed if they'd be involved in procurement..

The problem that has been evaluated is that the research was done on an organisation with no true enterprise architecture (at the business silo stage at best). In other words, somewhere that hasn't invested in IT (and likely has the doctors doing what they feel like, with 'homegrown' Access databases and applications, trusting what the vendors say when they produce shiny pamphlets, and either not hiring people who understand how business and tech should map, or not giving them the clout to be able to change the way the organisation works to successfully be able to change things so that they do).

Re:The key being ...

[martyros]

Best quote from the article:

Himmelstein said that only a handful of hospitals and clinics realized even modest savings and increased efficiency -- and those hospitals custom-built their systems after computer system architects conducted months of research.

He pointed to Brigham and Women's Hospital in Boston, Latter Day Saints Hospital in Salt Lake City and Regenstrief Institute in Indianapolis as facilities with some success in deploying efficient e-health systems. That's because they were intuitive and aimed at clinicians, not administrators.

Programmers of the successful systems told Himmelstein that they didn't write manuals or offer training. "If you need a manual, then the system doesn't work. If you need training, the system doesn't work," he said.

In other words, computers are not a magic bullet. They only work well when you actually invest the time to find out what you need them to do, and then make them do that.

Computers need to be implemented correctly...

[mgchan]

And they usually aren't.

I'm a radiologist and computers have definitely improved patient care and saved the hospital money (or alternatively made the hospital more money) in our field. From digitized images and the ability to outsource to overnight coverage to voice recognition to get turnaround for finalized reports in an hour it has undoubtedly worked. And that's with in most cases only fair implementation of a computer system.

With most hospitals, the problem is that they like to do a piecemeal transition. Digitize a subset of notes and vital signs, half the time what you need isn't there so you have to look through the paper chart AND the computer chart. Or the vital signs are only half in the computer and half on a chart, so nurses double their workload. And when it's set up, they do it with an IT-centric interface that doesn't make intuitive sense to most users. When I use them I can see through my background in computer science and engineering why things are done a certain way, but it doesn't make any sense to physicians, nurses, etc.

Then they add in a new piece, such as more vital signs (but in a different section), some dictated notes, some linking to the outside. Outpatient notes are digitized, inpatient notes are still handwritten, etc. ED notes are separate, with their own system. It's a complete mess. This method is a waste of money and time, all for the sake of early deployment of a suboptimal system and minimal re-training of the staff to use a new system.

The VA had a decent attempt with CPRS. They digitized everything - from physician admission notes to clergy notes. At least everything is in one place, but people are overwhelmed with data and it's too easy to copy and paste incorrect or inaccurate information. The interface is also suboptimal (graphing lab values involves selecting a range of tests, building a worksheet, etc. much like you'd expect an engineer to make it for maximum flexibility, but minimal ease of use). And connecting to other VA systems is hit or miss.

Perhaps the best method is to build a new hospital from the ground up. All patient records get digitized (scanned, at least, if not run through some OCR). Have a tightly integrated medical record system developed in collaboration with health care practitioners. That would save the hospital money, in the long run, compared to them starting from scratch with paper records.

building bad clinical systems is harder

[r00t]

We are over ambitious. The more code we write, the more bugs we create.

The trouble with hospital data is that it is messy. You have to accept that.

It's tempting to design a hospital data system with specific fields for each item, every procedure enumerated, and every field validated. You want to normalize your data. You want it neat and tidy. You can work very hard trying to enforce this. You're screwed though, because life isn't like that.

You'd be better off with relatively "dumb" software, almost like a wiki, that lets you efficiently handle arbitrary text and arbitrary data blobs. It needs fast Google-style search. It needs to allow arbitrary associations so you can handle stuff like a patient claiming to have the same social security number as a different patient or a patient who claims to have a different identity than he did the last time he visited.

Then you need to keep medical staff away from both paper and computers. Data entry is for data entry specialists.