NDM and Healthcare Guidelines: More Attention to the Current Status,Complexity,and Context Is Needed

Evidence-based medicine and guidelines cannot solve all problems in healthcare (Kemm, 2006). Indeed, it can be exceedingly frustrating for a clinician when the quality of his or her care gets measured based on adherence to guidelines that do not apply to a given patient (Boyd et al., 2005). Common examples of this are blood pressure and diabetes treatments for older adults. Lowering both blood pressure and glucose levels are important but can also be quite harmful for individual patients who are at higher risk for falls, incontinence, hypoglycemia, and cognitive impairment if either is controlled too aggressively.

In this issue, Dr. Falzer (2018) contributes an article titled “Naturalistic Decision Making (NDM) and the Practice of Health Care.” He argues that the “best practices regimen”—an approach based on evidence and guidelines—has not worked due to a fundamental fallacy that they are overly simplistic and do not account for the nuances of modern medicine in the way that NDM could. He further asserts that implementation science approaches have not helped because they only serve to support and perpetuate the flawed “best practices regimen” approach. His point is well taken that some of the evidence and some (generally older) guidelines fall far short of providing guidance for the complex patient. However, the argument has important weaknesses. The reasoning seems to begin with a conclusion that is supported by an argument based on older thinking about implementation science and guidelines that support the a priori conclusion. This type of reasoning is a classic example of confirmation bias—a common risk when people are left to NDM approaches (Nickerson, 1998).

One weakness of this paper is that it appears to be based on an outdated understanding of implementation science. Since the Lomas (Lomas et al., 1989) definition, the field of implementation science has evolved extensively and includes an understanding of how treatments reach the maximum number of eligible patients, how they are adapted to fit into different clinical contexts, how they are sustained, and how both changes in context and potential unintended consequences can be anticipated and avoided (Brownson, Colditz, & Proctor, 2017; Chambers, Glasgow, & Stange, 2013; Glasgow et al., 2012; Stirman et al., 2012). Since the articles referenced within the manuscript, there have been multiple advances in our understanding of both how to disseminate and how to implement interventions (Brownson et al., 2017; Brownson, Jacobs, Tabak, Hoehner, & Stamatakis, 2013; Chambers et al., 2013) and research on and understanding of guideline implementation and how to de-implement ineffective or harmful practices (Grimshaw et al., 2004). Indeed, current implementation science would design interventions and implementation strategies (two different things) (Kirchner, Waltz, Powell, Smith, & Proctor, 2018) that fit a given context. It would also consider the perspective of multiple stakeholders, including clinicians, patients, and others impacted by guidelines (Domecq et al., 2014; Selby, Beal, & Frank, 2012), rather than impose a best practices regimen.

The paper also presents a dated understanding of guidelines. Guidelines have evolved considerably since Woolf wrote the 1990 paper (Woolf, 1990). The United States Preventive Services Task Force is an excellent example of how guidelines can be constructed in a trustworthy way that applies to many patients (Harris et al., 2001). For the last several years, these guidelines, as well as others (Jacobs, Anderson, & Halperin, 2014; Kim, Puymon, Qin, Guru, & Mohler, 2009), are very nuanced and situation-specific and explicitly address uncertainty, complexity, and breadth of applicability—issues for which Falzer faults guidelines for not addressing.

Modern guidelines involve painstaking reviews of evidence and often include a grade of the strength of the evidence and specifically address applicability to help clinicians determine how rigorously they should apply these guidelines to the nuanced patient sitting in front of them. Also, evidence and guidelines are essential to and provide guidance for de-implementation of care that is harmful and has sometimes resulted from narrow-minded application of guidelines (May, Montori, & Mair, 2009) that do not take into account the totality of the patient and his or her medical and social-environmental situations (Levinson et al., 2015). Consider the example of hormone replacement therapy. Guideline developers understood that postmenopausal women were deficient in estrogen and that the deficiency put them at risk for osteoporotic fractures and heart disease (Nelson, Humphrey, Nygren, Teutsch, & Allan, 2002). Thus, it was intuitively obvious that replacing estrogen would be helpful. It turns out, many women were dying of stroke and breast cancer. It was the dissemination and implementation of evidence and guidelines that helped curtail a well-meaning initiative but public health disaster.

Finally, the Falzer article does not clearly acknowledge its underlying assumptions. Decision making must begin with clear goals. Well-meaning, expert clinicians using an NDM approach have driven aggressive utilization of health care in the name of perpetuating survival. Medicine has evolved with this implicit assumption that “outcomes” like survival must be maintained at all costs. Consequently, there is a large body of evidence that survival is often pursued without consideration of patient preferences or goals (Fagerlin et al., 2010; Sepucha et al., 2010; Zikmund-Fisher et al., 2010), and this is a major driver of rising healthcare costs. Patients, along with their preferences and understanding of healthcare decisions related to guideline application, must be critical players in informed decision making about guideline adoption. This is largely not addressed in the Falzer article, except for brief mention near the end. The entire field of shared decision making—which is essential to guideline implementation—is not discussed (Barry & Edgman-Levitan, 2012; Elwyn et al., 2012), and the patient perspective is hardly mentioned.

Science or NDM will not resolve many of these issues—a new theory based on NDM and proficiency will not help these problems and actually risks perpetuating them. Patient-centeredness is not based on behavioral theories, but rather, it is based on ethical, moral, legal, and human rights theories (King & Moulton, 2006). These are discussions that must be explicit and discussed at a societal level as well as between patients and clinicians. An important initial step toward addressing the issues Falzer raises would be discussion of the goals of NDM and the multilevel and contextual factors that influence healthcare decisions.

As written, it is difficult to see how the proposed application of NDM would help with this complex issue. By ignoring modern implementation theories, current approaches to guidelines, patient perspectives, and the field of shared decision making, the article falls short of providing viable solutions. At an extreme, one might interpret the author’s argument to suggest that NDM improves quality by continuing to support proficient experts to ignore patient preferences and drive up health care costs—an argument that is not only not helpful but also potentially harmful.

In summary, there certainly have been problems with narrow, unthinking adherence to a given guideline, especially in complex, comorbid patients, which have been discussed by numerous authors (Boyd et al., 2005; May et al., 2009). However, current guidelines, evolving criteria for guideline application, implementation science, and shared decision-making approaches have addressed and can further help with the problems described. NDM could potentially make relevant contributions to these issues, but to do this successfully, a deeper understanding of the issues involved, a recognition of the substantial and relevant literatures, and an awareness of current guideline application is needed.