In my last two postings, I described how a physician’s task to stay up to date with medical knowledge in the field has increased in difficulty over the years, as the body of literature from which to cull the necessary knowledge has increased tremendously. I then described how physicians performed this necessary duty — initially by filing curated articles ripped from print journals, then later digital subscriptions and publications, attending tumor boards, meetings and selected seminars, keeping up with accreditations, and consulting with local experts. Even so, this methodology is hit-and-miss, and many discoveries and practice-changing developments could easily be missed. Academicians, who often have a lighter patient care load, are better suited to attend meetings and have more time for reading than the community physician, who is often saddled with an increasing amount of administrative paperwork and the extra time needed to deal with electronic medical records.
Hospitals, insurance companies, medical group administrators have long been aware that a doctor’s grasp of current developments is not always reliably comprehensive. However, these healthcare organizations know that better treatments often lead to better outcomes that improve survival, and better patient outcomes are what healthcare systems would prefer to tout to consumers selecting personal healthcare options.
Practice guidelines
One approach has been to develop clinical practice guidelines. In my field of oncology, the National Comprehensive Cancer Network (NCCN) has developed an extensive set of algorithmic guidelines that have proven to be a great resource to medical oncologists, especially to those who are relatively inexperienced, need a refresher, or are practicing in an environment with no easy access to expert consultants. Unlike the textbooks of yore, these algorithms more explicitly direct the oncologist to consider options that might be proferred by an expert consultant.
The goal has been to “continuously” update these guidelines, and in practice, they are updated “at least annually” although an important development could trigger an interim meeting. Changes are approved by a committee and then the guideline is updated. But it is possible that a practicing physician might not be made aware of late breaking developments for varying durations, possibly weeks to months, for example, if he or she unaware of changes made to the guidelines. What shall we do about doctors not checking guidelines regularly?
Clinical Decision Support systems
Recognizing that creating a nice reference tool is of little value if it is not used, another effort has been made in parallel with the deployment of guidelines. Rather than rely on the physician to read and follow the guidelines, what if the guidelines are put in front of their faces whenever the physician tries to order a treatment? What if the treatment that has been recognized as the most suitable, is promoted as the best option, when the physician is ready to place the order for a treatment? Brilliant, isn’t it? This is guaranteed to capture the attention of the ordering physician at the time of decision-making, to make sure that the right therapy is at least considered. This approach has been referred to as “clinical decision support”. No more dependence on doctors to educate themselves! The system will step in and educate the physician as to what to do. Physicians may be given the option to decline the suggestion, and enact the original decision, but the reason for the departure from the orthodoxy generally must be given in order to proceed to completion. The goal is to encourage physicians to modify their practice in accordance with what is considered standard therapy, and this is often done by utilizing the electronic medical record platform and “computerized physican order entry”, known as CPOE.
This approach has the backing of HealthIT.gov and at least one widely-used electronic medical record (EMR) system has implemented this. In the oncology world, the company behind many clinical journals, and which calls itself an information and analytics company, has a tool called ClinicalPath, which offers an algorithmic approach to cancer care similar to that of the NCCN guidelines, but which can be integrated into an EMR. While this will help guide an oncologist to select a treatment from a set of “reasonable” options, the implementations that I have had experience with have been similar to having to take the extra time to feed an automaton with pertinent clinical data, then having the algorithm present the treatment options to consider. If you agree, then you are done, but if you disagree, you ust state the reason behind your objection, and provide references and then detail the alternate regimen you recommend. This process needs to be repeated for every single patient who is getting treatment that day, and as you might imagine, get very time consuming. Proponents of CPS point to studies that suggest that these systems improve patient care and save lives. However, clinical decision support systems have not always been viewed with enthusiasm by physicians. Some of the reasons include interruption of workflow, and time consumption. Some have voiced low confidence with the recommendations presented and the evidence behind them. Some complain about the suboptimal implementation, with inadequate detail of the data input into the system, leading to simplistic and inappropriate recommendations. Some have said that the system is fraught with annoying and repetitive “advisories” that are sometimes irrelevant and are viewed as simply disruptive. CDS systems have been known to contribute to physician fatigue and burnout. Once implemented, there may be little effort taken to evaluate the effectiveness of the system, nor mechanisms to allow physicians to provide helpful feedback. These systems are very expensive, and once purchased and installed, it is highly unlikely that an organization will reverse course.
Yes, CDS systems are not going away, and substantial investments have been made to implement these systems already. The proponents of the concept have sold the idea to major healthcare systems, who see the benefit not only of providing evidence-based information to the employed physician, but also offers the ability to audit each physician. One can see which doctor ordered a treatment, and how that compares with peers, what the objections were, and even how much a treatment is costing the system.
Maintaining these systems, and keeping them current with latest evidence-based data requires an ongoing subscription investment as well. Committees of physicians will need to commit to the necessary update meetings to refine the guidelines on which the recommendations provided by the CDS will be based. Many aspects of medicine are not cookie-cutter straightforward and there are many clinical situations that do not lend themselves to rote algorithmic implementation. There is still the need for understanding the finer points of the studies cited behind the recommendation, and the nuances of the patient for which the treatent is being considered.
While CDS-directed systems have advantages (especially to the auditors, administrators and tech companies), it doesn’t address the fact that physicians are still not reliably being educated. For a while, it might feel that these systems get in the way of practicing medicine, and make the physician feel that he/she is reduced to just being an implementer of a committee’s decision. It takes more time and effort to deviate from recommended protocols, so the temptation may be to just “go with the flow” every single time. The physician may feel that his/her knowledge base is not as valued as before, but instead, he/she just needs to trust the software. I personally feel that this system is not what I envisioned medical practice would be like. I think that there is still a role for the informed, educated and insightful physician, willing to make bold and reasoned decisions.
In the next posting, I will discuss some efforts made to help guide physicians into making decisions for their patients.
Further reading:
https://www.nature.com/articles/s41746-020-0221-y
https://healthitanalytics.com/features/understanding-the-basics-of-clinical-decision-support-systems