The clinician as functional decision aid.

Healthcare decisions involve comparing the possible outcomes that could occur after choosing a treatment, diagnostic, or other management option. The best way to do this is by presenting information – both quantitative and qualitative – that compares and contrasts decision alternatives using a format that will support making these comparisons.

To foster accurate presentation of quantitative information about the risks and benefits of alternative management strategies in patient decision aids, the International Patient Decision Aid Standards (IPDAS) collaboration has published regular reviews and recommendations for best practices. [1] The most recent one was published in two parts, one for basic and one for more advanced topics. Both were published in 2021 and are freely available at the Medical Decision Making journal website. [2,3]

A total of 11 key recommendations are made between the two papers, starting with a set of overarching principles that include: 1) Avoid using only verbal terms to describe outcome likelihoods, use numerical formats instead, and 2) Use a presentation format that is appropriate for the intended audience.

These two principles capture the key take-home message of the reviews: that the goal is to accurately and effectively convey the differences in expected rates of benefits and risks among a set of decision options. The extent to which an expected outcome communication format is successful therefore depends on how well this is done.

The rest of the recommendations describe current best practices for accomplishing these two tasks and the areas where there is insufficient information currently to make firm recommendations. If you are interested in learning more, I suggest you take a look at the original papers.

Musings:

These two papers are intended for use by decision aid developers. Accurate information in patient decision aids is important. However I think a singular focus on patient decision aids ignores the larger picture that the vast majority of clinical decisions are made without the use of a patient decision aid. In these cases, shouldn’t we expect the clinician to serve as a “functional decision aid”, trusted to provide accurate, clearly presented comparative information just like patient decision aids should do?

It would be terrific if these two papers and the recent updated NICE risk communication guidelines discussed in the August 12, 2022 Musing [4,5] served as the basis for a concerted effort to ensure that all clinicians are adequately trained in proper comparative outcome communication practices.

High quality medical care depends on high quality medical decision making. If we know what should be done to provide the high quality information needed to make good clinical decisions, shouldn’t every effort be made to make sure appropriate support is always available?

References

1. International Patient Decision Aid Standards Collaboration: http://ipdas.ohri.ca/index.html

2. Bonner, Carissa, Lyndal J. Trevena, Wolfgang Gaissmaier, Paul KJ Han, Yasmina Okan, Elissa Ozanne, Ellen Peters, Daniëlle Timmermans, and Brian J. Zikmund-Fisher. “Current Best Practice for Presenting Probabilities in Patient Decision Aids: Fundamental Principles.” Medical Decision Making, 2021, 0272989X21996328.

3. Trevena, Lyndal J., Carissa Bonner, Yasmina Okan, Ellen Peters, Wolfgang Gaissmaier, Paul KJ Han, Elissa Ozanne, Danielle Timmermans, and Brian J. Zikmund-Fisher. “Current Challenges When Using Numbers in Patient Decision Aids: Advanced Concepts.” Medical Decision Making, 2021, 0272989X21996342.

4. Carmona C, Crutwell J, Burnham M, Polak L. Shared decision-making: summary of NICE guidance BMJ 2021; 373 :n1430 doi:10.1136/bmj.n1430.

Overview: Shared decision making. Guidance NICE [cited 2021 Jun 21]. Available from: https://www-nice-org-uk.ezpminer.urmc.rochester.edu/guidance/ng197

Incorporating shared decision making in routine clinical practice: The missing link?

In 2017, Sarina Schrager and colleagues published a paper describing how shared decision making (SDM) for cancer screening decisions can be implemented in primary care settings. [1] It is freely available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5697707/ As of Sept 22, 2022 it has been cited 26 times according to Google Scholar .

Schrager and her colleagues explain that learning how to incorporate SDM into practice is important because it is a framework for providing the type of high quality care advocated in the Intitute of Medicine’s Crossing the Quality Chasm report that is “…respectful of and responsive to individual patient preferences, needs and values”. [2] They then describe three SDM models: the AHRQ SHARE method, the 5 A’s method described by the US Preventive Services Task Force, and the IAIS (Invite, Acknowlege, Instruct, Summarize) model.

They conclude that: “Practicing clinicians will need to choose a model that works well for their practice.”

Musings:

I thought this paper was notable for two reasons.

1: It clearly and simply explains why routine SDM for screening decisions is an important goal to pursue.

2: The conclusion highlights the importance of adopting a SDM practice that fits in the local practice context.

The latter point is consistent with the Canadian Institute for Health Research (CIHR) Knowledge to Action (KTA) Framework, originally described by Graham and colleagues in 2006. [3] The original paper has been cited 4,821 times according to Google Scholar on September 22, 2022, so clearly, there is a lot to learn about this idea.

For now, however, I’d like to focus on the basic structure the KTA framework provides for thinking about how research findings can be effectively used to improve clinical patient care.

The figure below shows the original framework [3] :

The framework describes the path from research to practice as consisting of two separate and distinct processes: knowledge creation and knowledge application.

Knowledge creation is largely the work of academia. Clinical trials and other studies are done, summarized, and eventually packaged in a format intended to facilitate their application in clinical practice.

The knowledge application cycle takes place in a local practice context. It starts when a problem is identified and a decision is made to fix it. Steps are then taken to identify a research-developed solution – a product or tool – that could improve the situation, adapt it for local use, and monitor how well it works.

A key insight provided by the KTA framework, that can be easily overlooked, is that there is no direct connection between knowledge creation and the knowledge application cycle, another words there is a missing link:

Thus, one has to ask the question, how do research-based findings and products cross this gap?

It is easy when the research-based findings and/or products meet an urgent clinical need. A good example is the new vaccines and treatments for COVID-19. But most often, clinical practices in need of improvement are less obvious. In this more common situation, I think there are several ways research products could cross the gap that could work independently, additively, or even synergistically.

1. The products could be actively “pushed” from within the knowledge creation sphere. For example, clinical practice guidelines created by specialty societies and others. These guidelines summarize research information and, if done well, clearly state a rationale and method for adapting current practice to reflect the current research findings.
2. The products could be actively “pulled” by clinicians seeking to improve the quality of their practice. This was the original idea behind evidence-based practice.
3. In addition to these two mechanisms that operate within the health care context, there could be external forces seeking to push the information from research into practice. An example is the mandate by the US Center for Medicare Services (CMS) requiring elements of shared decision making before approving payment for lung cancer screening. [4] Another example is a health care insurer establishing quality of care targets that must be met to avoid some sort of practice penalty.

As discussed in recent blog posts, there have been many attempts trying to push SDM into clinical practice. Numerous articles have been published in academic journals, SDM is becoming increasingly included in clinical practice guidelines, and there are several mandates imposed by external agencies (such as CMS discussed above). The evidence to date suggests that none of these efforts has made much of a difference. In KTA terms, this situation suggests the way forward is not to continue to create more SDM tools/products but to identify and address issues that inhibit their use in practice starting with how to cross the gap between the academic world of knowledge creation and the applied world of clinical practice.

Perhaps the way forward depends on our ability to interest clinicians in learning how to incorporate SDM into clinical care and motivate them to do so. If so, the keys to improving SDM use would be: first helping local practices recognize they should incorporate SDM into routine patient care and then helping local practice leaders identify appropriate SDM products/tools and then adapting them for use in their setting.

Do we have methods for accomplishing these tasks? If not, are they being developed? Are these steps even possible or will they first require a more fundamental transformation of the current culture of clinical care? Food for thought.

References

1. Schrager S, Phillips G, Burnside E. Shared decision making in cancer screening. Family practice management. 2017 May;24(3):5.

2. IOM (Institute of Medicine). Washington, D.C: National Academy Press; 2001. Crossing the Quality Chasm: A New Health System for the 21st Century

3. Graham ID, Logan J, Harrison MB, et al. Lost in knowledge translation: Time for a map? J Contin Educ Health Prof. 2006;26(1):13-24. doi:10.1002/chp.47

4. Merchant FM, Dickert NW Jr, Howard DH. Mandatory Shared Decision Making by the Centers for Medicare & Medicaid Services for Cardiovascular Procedures and Other Tests. JAMA. 2018 Aug 21;320(7):641–2

---

Note: this post was originally published on September 23, 2022 on MDM Musings.

The anatomy and pathology of clinical decision making.

In the September 13, 2022 newsletter, I reviewed an article suggesting that the term shared decision making itself could be contributing to the difficulties implementing it in clinical practice. The proposed solution was to rebrand shared decision making as “contextualizing decisions”. Today, I’d like to expand on this idea further by posing the question whether shared decision making should be considered separate from regular clinical decision making. 

This question is addressed in David Eddy’s 1990 article Anatomy of a Decision. [1] His focus is improving the quality of healthcare which he defines as follows: 

“THE QUALITY of medical care is determined by two main factors: the quality of the decisions that determine what actions are taken and the quality with which those actions are executed – what to do and how to do it. If the wrong actions are chosen, no matter how skillfully they are executed, the quality of care will suffer. Similarly, if the correct actions are chosen but the execution is flawed, the quality of care will suffer.” 

In other words: 

Quality of care = Decision Quality (Knowing what to do) + Quality of care delivery (Knowing how to do it properly)

Eddy notes that up until 1990, much more attention had been paid to the quality of care delivery than decision quality.  A major goal of the article is to point out that decision quality is a major variable affecting the quality of healthcare:

“The importance of ensuring the quality of execution is well understood. In contrast, the medical profession has done much less to develop and evaluate its decision-making processes. If decisions are considered the command post and actions are considered the troops in the field, we have spent much more energy training and equipping the troops than providing intelligence and decision support systems to the commanders.”

As far as I can tell, this situation has not changed much in the past 30 years.

Eddy then uses a standard clinical format to describe clinical decision making starting with the anatomy of a clinical decision, then describing decision making pathology, and finally treatment. 

He depicts the anatomy of clinical decisions as follows: 

“In general, the goal of a decision regarding a health practice is to choose the action that is most likely to deliver the outcomes that patients find desirable. This identifies the two main steps of a decision … First, the outcomes of the alternative practices must be estimated; then, the desirability of the outcomes of each option must be compared.”

Eddy then relates that decision pathology results if the possible decision outcomes are incorrectly estimated or if patient preferences are not adequately ascertained and taken into account: 

“Misperceptions of patients’ preferences can also occur. Patients might misunderstand an outcome, the measure of the effect might be misleading, the outcomes can be presented in different ways that lead to different conclusions, the patient might not be consulted at all, or physicians might project their own preferences onto their patients.”

Eddy then goes to describe three principles of treatment to avoid poor quality clinical decisions:

“First, decisions should be based on outcomes that are important to patients. These are the ‘health outcomes’ that patients can experience and care about.”

“Second, the effects of a practice on outcomes should be estimated as accurately as possible, given the available evidence … and the information should be presented in a form that is meaningful and intelligible to patients.”

“The third principle is that the preferences assigned to the outcomes of an intervention should reflect as accurately as possible the preferences of the people who will receive the outcomes—that is, patients. Patients should be encouraged to participate in the decisions to the extent they want. If a patient chooses to delegate the decision, the person chosen to act as the agent must understand that the values he or she expresses will be projected onto patients.”

MUSINGS

Clearly, in Eddy’s view, shared decision making should not be considered separate from regular clinical decision making. Rather, it is an integral part of it. If so, perhaps there is no need to propose that clinicians and health care systems adopt a new way of doing business. They just need to focus making good decisions. I think this idea has a lot of merit. 

This is not a highly cited reference with regard to shared clinical decision making but it speaks to the topic directly. (In fact, I haven’t been able to find any references to it. If anyone knows of any, please let me know.) I think it should be added to the list of seminal articles about the importance of honoring patient preferences and values in decisions about their care. 

The other striking feature of Eddy’s article is that he treats decision making in a format familiar to clinicians. In my experience most articles discussing shared decision making do not. This raises the possibility that another problem implementing shared decision making is that the current literature is written in a format clinicians are not used to and non-clinical journals. Adopting clinical terms and formats in information aimed at clinicians could facilitate implementation efforts. 

References

1. Eddy, David. Anatomy of a Decision | JAMA | JAMA Network. JAMA. 1990;263(3):441–3.

I think we should ask the engineer.

Science is about knowing; engineering is about doing. ~Henry Petroski

At the most basic level, the goal of medical practice is to make good decisions regarding the care of individual patients.

The growing complexity of modern medicine, including the moral imperative to engage patients in shared decision making, has made it increasingly difficult for individual practitioners to achieve this goal consistently. In recent blog posts I raised the prospect of developing a new clinical decision support system to help clinicians cope with complex clinical decisions and engage patients in decisions about their care. The question is, however, how to go about it.

Proposed methods to promote shared decision making in clinical practice have been based on multiple conceptual frameworks. In a 2011 review, Glyn Elwyn and colleagues reviewed eight frameworks – expected utility theory, the conflict model of decision making, prospect theory, fuzzy-trace theory, differentiation and consolidation theory, the ecological rationality theory, the rational–emotional model of decision avoidance, and the Attend, React, Explain, Adapt model of affective forecasting – and concluded that none of them could effectively serve as the basis for clinical decision support interventions and added:

“… it was also clear that most work in theory-building has focused on explaining or describing how humans think rather than on how tools could be designed to help humans make good decisions.” [1]

Engineers design tools. Engineering has been characterized as:

… the application of science and math to solve problems. Engineers figure out how things work and find practical uses for scientific discoveries. Scientists and inventors often get the credit for innovations that advance the human condition, but it is engineers who are instrumental in making those innovations available to the world. [2]

The process of improving decision making using an engineering approach is called decision engineering. Decision engineering has been defined as:

… applying relevant knowledge to design, build, maintain, and improve systems for making decisions. [3] (see footnote)

A wealth of relevant knowledge is available. In addition to the traditional fields of clinical epidemiology, evidence based medicine and health care communication, information that can help improve clinical decision making can be found in the fields of cognitive psychology, decision and risk analysis, information visualization, and many others Moreover, recent advances in electronic communications and computing now make it possible to bring previously inaccessible information and tools into the clinical consultation room quickly and easily.

Musings

When I was a camp counselor, many years ago, the camp staff reported about two weeks before the camp opened and did all sorts of odd jobs to prepare the camp for the upcoming season. A local high school teacher was also hired to help. It was fairly common for a group of young college-age counselors to stand around wondering how to do something until someone said, “You know, I think we need to ask the engineer”, meaning the teacher. He always figured out a good way to do the job.

It seems to me that applying decision engineering principles to the problem of how to effectively facilitate the adoption of shared decision making into routine medical practice is well worth a try.

---

Footnote: It appears that the term “decision engineering” has largely been replaced by “decision intelligence”. I am not sure they mean the same thing and think using both terms makes sense.

---

References

1. Elwyn G, Stiel M, Durand M-A, Boivin J. The design of patient decision support interventions: addressing the theory-practice gap. J Eval Clin Pract. 2011 Aug;17(4):565–74.

2. What is Engineering? | Types of Engineering [Internet]. livescience.com. 2022 [cited 2022 Oct 20]. Available from: https://www.livescience.com/47499-what-is-engineering.html

3. What is Decision Engineering? [Internet]. Tim van Gelder. 2015 [cited 2022 Oct 18]. Available from: https://timvangelder.com/2015/01/07/what-is-decision-engineering/