A recent article in a respected medical journal speaks of diagnosing a condition called bacteremia. This term refers to the presence of bacteria in the blood of a patient and is a very problematic indicator of serious illness. There is a many-years old means of diagnosing bacteremia, based on a series of blood tests starting when the patient first presents to the hospital. Nevertheless, doctors have been looking for an alternate faster mechanism for diagnosing bacteremia so as to begin treatment even before the blood tests are drawn.
The data for this study came from 17,000 visits by adults to an emergency department in Denmark over a two-year period. Let me already point out that collecting all of the necessary information by hand from 17,000 patients would normally take a tremendous amount of time. On top of this, actually analyzing the data by hand would make such a study extremely difficult to do in the first place, and secondly to repeat by another investigative group so as to validate or challenge the original data and conclusions. So it is evident already that the availability of electronic medical systems makes it possible to do research and data analysis in a way that simply was not available even 30 years ago.
The authors conclude that there are a series of clinical markers, that if negative, effectively rule out the likelihood of bacteremia. What this means is that if the patient fails to manifest certain findings, then the chance that the patient has this serious illness is small. Therefore, such a patient with negative findings can be safely discharged home after a short period of observation.
Four of the criteria for evaluating the likelihood of bacteria in the blood were temperature, the rate of breathing, heart rate and the white blood count. The fifth criteria was a blood test that is done in some emergency facilities and is called CRP. So the combination of these five tests allow a physician to quickly determine the status of a patient. I should add that these tests require far less time than waiting for the result of the blood cultures. Blood cultures are a unique test that looks to identify the actual presence and type of bacteria in the blood. Such cultures can take 24 hours or more to be definitive. Therefore, even if alternative tests take a few hours, this is still far better than waiting for the blood culture results.
In addition to a physician remembering to do the five tests to rule out bacteremia, the physician must also remember what the problematic ranges of values are. So the doctor needs to remember what a normal CRP is and what range of temperature is problematic and what range of white blood counts is abnormal. This is in addition to all of the other information that a physician must remember. The doctor must remember the appropriate ranges for all of the tests that are done on patients for all of the various diseases that present to the hospital. I will let you in on a little secret, that many doctors do not remember normal result ranges and must look them up in a book or on their iPad that they carry in their coat.
Let’s imagine an alternate system where the ambulance driver or at least the receiving nurse in the emergency room, is able to indicate that there is a concern for bacteremia. The ambulance driver or nurse may require additional specific training to identify this condition as a possibility. But let’s assume that the computer system that oversees all of the patient’s care is already informed of the possibility of bacteremia before the doctor walks into the examining room. In this scenario, the expectation is that the computer system [if properly programmed] would begin analysis of blood tests already taken and suggest additional blood tests as needed, to rule out bacteremia and other dangerous and/or likely diagnoses. By the time the patient is seen by the doctor, all of the pertinent testing should already have been done and the doctor really only has to indicate which of the list of likely diagnoses is in fact the final confirmed diagnosis.
I am not over-reaching in this scenario, because strictly speaking, the computer will most likely be able to also flag which of the list of likely diagnoses is most probable. It will still be the role of the doctor to make the final decision, but in time, the rate of agreement between the computer and the human doctor will rise to 100%.
This degree of computer involvement in health care is already happening. In my last place of employment, I added limited diagnostic functionality to the electronic medical record system that I had written. My system will process over 600,000 patients in 2014. One of the tasks of the system is to look for indicators of serious illness or contrarily, to rule out serious illness based on various findings. Historically, each time a new module, specific for a disease, was added to the system, accuracy in diagnosis immediately increased. Unfortunately I was not able to stay to complete adding such modules for a whole further range of diseases, but that task has now fallen to the new systems manager of the company.
40 years ago, when large-scale studies were very difficult to run, and doctors depended far more on their medical school training and personal experience to diagnose disease, the concept of having a computer being so involved in every step of healthcare would have seemed impossible. In 40 years from now, young doctors will refuse to believe that medical care was ever effective without the constant oversight of computerized systems. Such is the nature of progress.
A special note to readers who would like to communicate with me on a one-to-one basis. I have been contacted about the content of my blog posts, and even about consulting on medical technology issues. The best way to contact me is not by phone or SMS, but by email at firstname.lastname@example.org.
Thanks for listening.