David Himmelstein of Cambridge Hospital and Harvard Med School (with co-authors) recently published a paper on the effect of computerization of hospitals.
The results, as Computerworld put it: Computers don't save hospitals money
This makes sense, especially when you look at the study. They focused on data collected reported by individual hospitals nationwide between 2003 and 2007. Computerization, especially at the clinical level, is incredibly disruptive. Thus, the efficiencies aren't likely to arise soon.
I've seen clinical automation disrupt work second hand: my wife is a practicing physician. They automated the medical records at her clinic about three years ago and physicians are still adapting to the change. Here are where various inefficiencies come in to play:
- Patient charts exist both electronically and on paper. Some charts are literally inches thick and it's not practical to digitize the whole chart in a useful manner. Everyone performs double the work when handling these double charts.
- Practitioners must retrain themselves to use the system efficiently. A lot of medical training implicitly teaches a physician how to conduct a patient visit. The process changes radically when integrated with computerized record handling. This slow things down as people adjust to the differences in the automated system.
- Some physicians, like my wife, are computer savvy and can type their notes with relative ease. Others never learned to type. Thus, the clinic often needs to maintain voice dictation and other traditional systems as well as the electronic system.
- Electronic medical records don't save time for physicians. After every patient visit the physician must update the patient's chart, and the electronic process is in general slower. I've talked to several physicians about this. Some spend as much time on the electronic charts as they spent on the paper system. Some spend more time to process the same number of patients. I don't think I've talked to one who believes the electronic process is dramatically faster.
Looking at nationwide numbers, the Himmelstein paper found little evidence of benefits in computerizing health care. There was a "weak correlation" between better health outcomes and increased computerization. But in general, hospitals that were increasing their level of computerization also saw increased administrative costs over the past few years.
Where are the efficiencies?
None of this really surprises me. Business machine vendors have touted the efficiencies of automated processes for most of the 20th century. Hollerith and IBM were pushing punched card automation since the gay '90s. Throughout the '50s and'60s, computer vendors asserted that automation would quickly pay for itself.
In fact, computerization did not visibly affect the US economy until very late in the century. When business researchers analyzed the economy and tried to measure the effect of automation, no large-scale benefit was obvious until the 1980s or -90s.
Clearly, computers made certain business activities more practical. Railroads couldn't keep up with traffic until punched cards helped them out in the early 1900s. Banks couldn't eliminate tellers until computerization led to ATMs and on-line banking. Clearly, today's plastic money wasn't practical without large scale computer-oriented organizations like Visa
I haven't seen anyone try to assess the impact of information security on health care automation. Originally, all health care security was physical security: you keep the charts locked up, and that's it. Every clinic had a cardinal rule: you never take the chart out of the clinic. To do so, you made a copy and gave the copy to an authorized person. Otherwise, charts never, ever left the clinic. Each medical facility had its own chart for each person. Sometimes the chart might include copies of information from another institution's charts. More often, there would simply be a letter from another physician focusing on some important detail or event in the chart.