Blogs

An article in the NEJM "Putting the Patient Back Together - Social Medicine, Network Medicine, and the Limits of Reductionism" Dec 21, 2017, page 2403, introduced new terms from emerging fields in a variety of areas. In the era of machine learning, artificial intelligence, network science and quantitative graphic methods there is a new relevance in putting the patient first.

The alternative payment, value based purchasing movement is realizing the potential for population metrics to influence health care costs and quality. However this potential to control costs is not meeting expectation. For CMS and health plans, the noticeable glacial response to cost containment incentives is not lost. There is now a belief that there is not and cannot be a transition from Fee For Service to global payment with its stronger incentives to control costs, without new innovation in the programs.

In this context the article by Loscalzo's and Barabasi's groups at Harvard and Northeastern Universities, radically changes the debate about the practice of medicine, and as a collateral issue, the policies to control health care costs.

Putting the patient first can be dauntingly complex particularly in this era of evidence based medicine and quality determined at the population level. The Loscalzo article places in the public domain a new way of thinking with a paradigm centered on holism. Mindfulness ideas such as a personal health and wellbeing tools are supportive of and promote holism. However there is no competition in the mindfulness arena with a century’s long tradition in medicine where the whole person has been reduced to its parts. This is not to say reductionism is not necessary for the advancement of medicine. Where we seem to be today is that when the parts need to be assembled into a whole person, the whole person does not really emerge from the parts. Instead these assembled parts result in averages from like parts.

First, to focus the discussion of holism to health care cost reduction requires understanding what holism and reductionism are, how they compare and how they differ. The contrast needs to be patient centered. Though this term appears obvious, and this centering is the cornerstone of many programs, plans, financial liability for cost sharing, and precision medicine, the complexity of the patient is imperfectly reflected as aggregates of information. Though these aggregated summaries are the foundation of communication of information from above the patient level, they are, at their core, not patient centered. These summaries are population centered, and are applied to the patient but cannot, as aggregates whether of individual metrics or collections of several metrics, define the uniqueness of individual patients.

Second, the essence of holism similarly appears obvious when the concept is used in many contexts, always descriptive of self, well-being, and mindfulness among other approaches to the uniqueness and improvement of the self in health and disease. However, the relationship of holism to medicine and as a corollary cost reductions in healthcare, forces the comparison to reductionism. The traditional use of reductionism in medicine is centuries old and its power comes from its long list of success in isolating small things such as bacteria, viruses, and other disease mechanisms. Without reductionism discovery in medicine and its application to service delivery in healthcare would come to a standstill. A rediscovered movement in the philosophy of science is "critical realism" and this approach can help to clarify the roles, necessity, and usefulness of holism vs. reductionism.

As a philosophy, critical realism is interested in understanding the process of discovery used by creative scientists. The concept of closed and open systems is one of the many facets of this remarkable philosophy and surprisingly critical in implementing cost reduction by framing these methods into the categories of holism and reductionism.

Third, specifically using examples of programs for cost containment that are emergent and trending toward being comprehensive, the differences between holism and reductionism can be highlighted. In addition to making these cost containment programs more effective, guidance can be offered by the philosophy of critical realism.

To relate this guidance from critical realism, general principles of this approach can be applied to current health care financing and emerging alternative payment programs. The two types of payment systems include fee for service (FFS) and accountable programs. The latter has episode based transactional accountability, and population based accountability in the form of accountable organizations, or global payment. For fee for service, the MIPS program of the MACRA law, is the largest CMS program, which speaks to the difficulty of transitioning to the accountable programs at large scale.

For patient centered methods, which are truly patient centered and not the application of population metrics to the patient level, understanding the difference between holism and reductionism is critical. Fee for service payment methods, both as MIPS and episode based, are transactional, one patient at a time, and are purely patient centered. The problem is that it is difficult to effect cost reduction at the patient level using metrics measured at the population. This is what happens with reductionism, the details of each service rendered to the patient remain transactional, and when this information is aggregated, the result is smoothing of patient data at the population level. This, as patient specificity is lost, even with narrowing the population by creating subsets, makes patient level interventions generic and not patient specific. This may be what is in operation to discourage providers from making commitments to assumption of full financial risk even in the most mature global payment models.

Holism can be an antidote to loss of patient specificity, when done in conjunction with the methods of critical realism. Holism defines patient specificity. There is nothing off limits to a holistic approach. Information about a patient needs to be meaningful relevant and complete. And not limited to population level objectives, this information can include psychosocial inputs, patient and provider preferences, interactions of multiple medical conditions, standard medical condition imperative in the form of guidelines, long term services and supports, access to care, environmental and work related stress; the list is long.

However the issue with holism is how to affect population level outcomes such as cost reduction and risk stratification. This is where critical realism comes in. To start there needs to be a distinction between open and closed systems. Next there needs to be a place for the unknown, from an analytic perspective. In the present era of massive data, the capacity to include as data all holistic patient centered information is possible. As the patient level data accumulates, there may not be a relevant population from the reductionist perspective to house holistic data. The realist approach will allow an unknown population to be assumed, if not defined. Once the population is postulated, queries about outcome level metrics such as cost containment can be formulated. To clarify what a postulated population is, the distinction between open and closed systems will help.

An open system is what holism is, unstructured, where many variables of uncertain relevance exist, in addition to many mechanisms driving services on behalf of the patient. A closed system will follow the open when there is a formulation about a service or an event in the journey of the patient. Here is where guidelines apply, methods for directing operations of services, standardization is possible, and where costs are defined and registered in the IT systems. The open system continues for patient level factors as the closed system engages. For example a heart failure hospitalization is triggered by an event, and the services to manage the event are well defined, including methods to optimize costs. But the event occurred in the context of a population at risk for a heart failure admissions, in the open space of the patient journey prior to the event. Here if the cost containment question of most importance is how to prevent the hospitalization in the first place, methods to explore this open space are needed. Here in the open space is the unknown; what patient level factors need to be discovered that predict an admission. This is a different question than what is attested by standard risk factors, which are medically centered. Realistically the variables offered by the holistic approach may hold the key to identifying what variables or combinations of variables are important. These are highly likely to be psychosocial. Looking back into the open space will then define a holistically centered population, where the important variables define the population of not only those that have a heart failure admission, but those that do not. This is a key point in designing an effective cost containment method.

At the heart of the critical realist interpretation of cost containment is to understand the difference in two patient centered approaches to data analytics. The reductionist approach is structured, and the holistic approach is agnostic to how the data is accumulated, or in realist terms the latter approach is open and unstructured. Putting the patient back together again requires a free form method where there is no reason not to draw conclusions from patient centered data because of concerns about statistical significance at the population level. Being unstructured the patient level data can create populations that reflect the holistic approach, in a bottom up method. This is an inversion of the top down approach in the reductionist method that characterizes patients as averages of the predetermined population.

Joseph's Loscalzo's introduction of holism to the mainstream medical literature opens a more robust paradigm for cost containment. The message is that patients can be defined as unique individuals, and when a population based approach is needed analytics can be fashioned whereby cost containment need not be held to the same statistical standard as medical evidence. This will broaden the range of possibilities for cost intervention.