Computers were just being introduced in a significant way in the late 1980s when I started medical school. Essentially I began my career with entirely paper charts for documenting and ordering and now these functions are entirely electronic. Back in school if I wanted to find a published article I would have to find the reference, then go to the library and make a photocopy of it. Today I go to PubMed and can search a massive database in seconds to find the articles I need.
These advances may make it seem like the medical profession is at the cutting edge of information technology, but actually we are nowhere near where I thought we would be. Let’s look at a few different areas of information technology, where we are, and where I think we should be.
Access to information
Let’s start with the area in which I think we have made the most progress – access to information. This has benefited from the internet, and the development of medical databases like PubMed. In this way the medical profession has benefited from the internet the same way everyone else has. Since we are a very information-heavy profession, we may have benefited more than some other professions, but this is not due to any application specifically developed for medical practice.
Both medical research and practice have benefited from the information revolution, and I think such applications are keeping pace with other industries. In clinical practice it is incredibly useful to be able to look up laboratory and other study results on a patient and to have instant access to many of their records. What is lacking is a universal database. If a patient had an MRI scan at an outside facility I may not have access to it. They would have to mail or fax a paper copy of the report, which is then scanned as an image file into the patient’s record. This introduces a delay, does not always happen, and produces an image which cannot be searched or handled like text.
There is also untapped potential here for doing clinical research. Some nations outside the US, especially those with socialized medicine, have comprehensive databases of patient information, but that is not the case here. Such databases could be a goldmine for asking clinical questions, such as correlating patients who take certain drugs with specific outcomes. This would essentially be observational research, but could be extremely useful in exploring possible connections.
There are medical databases where this kind of research does happen, and we report on them all the time. But they are not universal or necessarily easy for a legitimate researcher to access or to search.
Electronic medical records
Electronic medical records (EMRs) were supposed to improve the practice of medicine, increase efficiency, improved outcomes, reduce errors, and decrease costs. The actual benefits have been modest, however. finds that EMRs are just barely cost effective, with a benefit-cost ratio of 1.23 and a payback period of >6 years. However, this analysis does not include the price of data breaches, which is since 2009.
The potential of a well designed and implemented EMR is huge, but we are not close to realizing this potential. An ideal system would be integrated thoughtfully into the workflow of a hospital or office. The system should minimize the time it takes to document medical care and shift as much documentation labor as possible to support staff. A good EMR can also facilitate communication among all the people involved in patient care.
An EMR can also be used to give the health care provider critical information at the point of care. For example, warnings about patient allergies, duplicate prescriptions, contraindications, or other critical information. Such a system could also reduce errors by flagging orders that seem unusual. Several studies indicate that EMRs , but only modestly.
Health care providers could also be given evidence-based nudges – have you considered anti-platelet therapy to reduce vascular risk in this patient?
All of these features exist to some extent in various systems, but they are far from being realized to the extent that they could. I work in a major health system with a popular EMR. The system, in my opinion, is a dismal failure compared to what an ideal EMR could be. It is inefficient, has a terrible user interface, and provides minimal clinical support.
It is not clear where the disconnect is, but in my experience (being involved in implementing EMRs at my own institution) the problem is in communicating between IT experts and medical experts. IT experts don’t understand what health care providers need, and the providers don’t necessarily understand what an EMR can and should do. Key individuals with dual expertise are needed to bridge the gap.
Further, the quality of available EMR applications just lags behind more popular applications. With the EMR I currently use there seems to have been zero investment in designing the user interface. This leads to unnecessary mouse clicks, increased cognitive work, decreased efficiency, and increased errors.
Expert systems are rapidly developing, and are being implemented to a limited degree in medicine. But again – we are behind where we could be given the current state of technology. These systems could provide critical information to health care providers at the point of patient care, dramatically improving the quality of care given. Every parameter would benefit from this, with improved outcome and reduced cost.
The potential of expert systems derives partly from the fact that the amount of information being generated by medical science is literally overwhelming. No one can hope to keep up with all of the advances being made, with every new study that is published. This forces narrower and narrower specialization, which then introduces new challenges to health care.
found that 46% of doctors frequently, and another 32% occasionally, use Google or other all-purpose resources to find medical information they will use to treat patients. I do this myself, because Google is a great search engine and there are deep resources online with authoritative medical information. But this also reveals the absence of access to expert systems optimized for medical decision-making.
Further, Google and similar resources are not expert systems, they are just search engines. Expert systems can potentially sift through a patient’s symptoms, signs, medical history, and previous test results and then suggest possible diagnoses. Or, given a diagnosis, they could provide specific information about sensitivity and specificity of diagnostic tests, or give evidence-based management guidelines.
All this information exists, but there isn’t enough time in the day to do extensive research on every clinical decision made on every patient. Obviously there is a lot of routine and repetition in clinical practice, but every patient is unique, and there is endless new information to incorporate. The more support practitioners have, the more compliant they can be with the latest evidence-based standard of care.
I have felt for a long time that we are not tapping into the potential for computer learning in training health care providers and maintaining their expertise. We are starting to see some of this, but it is tiny compared to the potential.
shows that video-game style computer learning is highly effective – more effective than traditional methods. Video games have the potential to confront students with simulated medical problems, give them real-time feedback, and individualize training to the user. This is exactly the kind of training that maximizes memory and learning.
Essentially a could be highly effective training that will also improve all medical outcomes. Players could track their patient outcomes, cost of care, error rate, compliance with evidence-based standards, and other variables and play to maximize all outcomes.
Of course, developing such games would take a lot of work, they would need to be updated, and the potential audience is small relative to games that appeal more broadly. But I suspect they would be highly worth the investment in terms of reduced health-care costs.
A worthy investment
On all fronts we are making slow progress, and electronic medicine is already a net gain, but a very modest one. We are falling far short of where we could be. Further, early evidence suggests that heavy investment in all these aspects of electronic medicine and training would be paid back many times over. cost $3.2 trillion in 2015. By all accounts there is a tremendous amount of waste and inefficiency built into this massive cost.
Even just a 1% improvement in efficiency would save $32 billion a year. Think of the software applications we could develop for just a fraction of that investment. There is enormous room for improvement, we know how to do it, and it would be worth the investment by many orders of magnitude.