In addition to purposefulness, thinking systems have a second unique characteristic: creativity, which can be defined as the ability to “figure out how to use what you already know in order to go beyond what you currently think.” (Bruner 1983; 183) This is both “man’s only basic virtue” (Rand 1957: 935) and at the same time the source of evil, where people poison themselves with cigarettes and illicit drugs or kill one another over philosophical, even semantic differences. The “thinking” in thinking systems refers to the ability to create new ideas, to make things never seen before, to behave emotionally and/or illogically, and to act purposefully, sometimes without regard to survival, selflessly such as the firefighters on 9/11 or destructively as in the 9/11 hijackers.

THINKING SYSTEMS ALWAYS LEARN

The Process of Learning

Thinking systems include humans and humans always learn. They may learn the wrong things or things that the organization prefers that they not learn, but humans by their nature learn. If the environment is risk-averse and mistakes are punished, then workers learn not to take chances, to follow rules and procedures rigidly, and to hide or minimize adverse outcomes or mistakes. (Edmondson 1996) Learning is a necessary precursor to change and hopefully to improvement. Activities that enable learning can enhance the quality of outcomes. (Baker 2001, Berwick 1989, Leape 1991 & 1994, Waldman, Yourstone & Smith 2003)

Learning—the acquisition of “knowledge or skill by study, instruction or experience” —can occur on individual, group, even the national level. (Senge 1990) Prerequisites for learning include appropriate incentives; discretionary energy; enabling environments, both internal and external; necessary substrate(s); a plan; and time. While the opposites of these prerequisites can be impediments to learning, there are at least four special constraints in service industries such as healthcare: 1) Lack of established truth or fact; 2) Barriers to testing; 3) Substrate issues; and 4) Regulatory/legal milieu. Possibly the strongest suppressor of individual learning is the difficulty associated with un-learning: changing or giving up the already-established mental model.

In healthcare, mental models (Leape 1994), individual labels, shorthands (Cameron & Freeman 1991), effective tags (Ashmos 1998) are called diagnoses. They are based on statistical probability, not mathematical certitude and should always be provisional. One must be prepared to revise or even reverse the diagnosis and corresponding treatment—to unlearn what we “know”—based on new information, change in conditions, and response (or lack of) to therapy.

In the natural sciences and the manufacturing world, learning leads to accurate information and ultimately to high quality, efficiently produced products. In the social and biologic sciences as well as service industries, learning should produce judgment, adaptability, problem-solving ability and coping skills.

Healthcare has unique impediments to learning. (Waldman & Schargel 2003; Waldman, Yourstone & Smith 2003) Learning means moving from a state of lesser to a state of greater and certainly different knowledge or skill. This is unnerving to a patient who assumes that there is an established answer for their problem and expects their physician to know it. In addition to the individual problems associated with learning and unlearning, there are cultural issues. Among doctors and nurses, admitting ignorance or incomplete understanding is considered weakness or a personal failing and therefore is avoided, even rejected. Furthermore, healthcare culture is highly risk-averse and strongly protective of the status quo: “Different is bad even when it is good.” (Waldman, Hood, Arora & Smith 2004)

Thinking Systems Learn With Intent

Machines do not learn. Complex adaptive systems can learn. Thinking systems always learn. and can do what other systems cannot: direct its own learning--both the process and the intended outcome. Computers (at present) cannot program themselves. Bees do not consider alternative, more aesthetically pleasing structures for the hive. Thinking systems can use stochastic reasoning, apply inductive as well as deductive logic, structure the learning process itself, apply hypothesis testing, and innovate ideas or objects that never existed before for a specific purpose.

Figure 1: Learning by Machines, Complex Adaptive Systems and Thinking Systems

Legend for Figure 1

A machine does what it is constructed to do. An airplane does not learn. It flies when turned on and piloted. Airplanes alone cannot improve. Complex adaptive systems engage in random trials until something improves their survival. Thus, by evolution, birds have learned over millions of years that those who fly in a rotating “V” [a paceline for bicyclists] are more likely to reach their destination and procreate.

Lance Armstrong and his team for the Tour de France reviewed films of previous races to see what worked and did not. They compared the effect of changing the order and the intervals between rotations of the riders. They tested body position, drag created by clothing and type of handlebars to optimize outcomes. In short, the thinking system called Team Discovery knew what it wanted to accomplish, set up methods to study how to achieve the desired outcomes, tested various alternatives and compared the results, and then refined the process with more tests. This was purposeful, structured learning, a capability unique to thinking systems. Note also that there is feedback — planned and unplanned — throughout the learning process as well as the actual race. The creative capacity of a thinking system offers unique opportunities but also poses special challenges: how to direct and organize the learning of self-directed and self-organizing system parts (called people); how to produce the desired outcomes (when system parts may prefer a different outcome); and how to diffuse innovations that are proven to be effective. [Coleman 1957; Rogers 1983; Ritchie & Hammond 2005]

What should result from the process of learning in a thinking system? The answer depends on the responder: the individual wants to learn how to maximize personal utility while the organization wants workers to learn how to achieve organizational goals.

Four hundred years ago, descriptors like reliable, resolute, respectful, constant, and unchanging were considered positive attributes. In the modern, generally chaordic 21st century, the following would be considered compliments: flexible, self-starter, competitive, innovative, and adaptable. To survive in today’s turbulent environment, one must be able to change, with intent and continuously. Two questions of immediate relevance follow: 1) How to effect change, and 2) Change to what?

Weick (1993) suggested that no one can survive without continuous adaptation. This demands an effective strategy, which in turn requires good sense-making. As the external environment is constantly changing, so too sense-making and strategy must be continuously revised. This translates into continuous learning. Weick also shows how the “non-disclosive intimacy” (1993: 644) of smoke-jumpers was adequate until the situation changed drastically and suddenly, the lack of trust between workers caused their deaths. “Non-disclosive intimacy” describes many situations in healthcare, such as the operating room or the ER where people who do not really know each other are literally intimate with someone else’s vital organs. As long as the circumstances are predictable and within the team’s past experience, the patient survives. However, a team that does not know and trust each other cannot adapt to unknown conditions: it does not learn. Finally, Weick showed how small events can have great or disproportionate [Butterfly] effect. These are all common features of learning by thinking systems.

To survive requires change and effective change requires learning. “Education is the most commonly used change facilitator, and is one of the least successful.” (Miller 1998: 369) Though thinking systems always learn, they often refuse to be educated (educate: “to provide schooling in or training for”) or to be taught (teach: “to impart knowledge of or skill in; to give instruction in.”). Education and teaching are passive activities of the student. People learn best heuristically, by trial and error, by doing. The “education” (Miller above) most commonly used as a change lever is persuasion by data or by authority. It is imposed from the outside, taught to but not necessarily learned by the intended students. Didactic learning plays little role in the co-evolution of the components of a thinking system. In the risk-averse culture of healthcare, trial-and-error and co-evolution are actively discouraged, sometimes with disastrous consequences. (Baker 2001)

HEALTHCARE: A PARADIGM OF A THINKING SYSTEM

When a process is rigidly structured with interactions exactly defined and outcome tolerances precisely measured, there is no need for creative thinking or judgment while doing the work, only while designing the system. This applies to most manufacturing activities and some service occupations. However, increasingly in modern activities, outcomes are variable; the interactions are emergent, (Hock 1999; Johnson 2001) chaotic but not random (McDaniel 1997) and subject to repeated adjustment. System components must deal with ambiguity and uncertainty. This last description is particularly apropos of modern medicine.

There is no established right way—proven to work in all patients—to treat obesity or heart disease or breast cancer or diabetes or asthma. Therefore, the nurse or doctor must constantly use judgment rather than simply apply a cookbook approach to care. Data is often ambiguous and decisions are always made using imperfect information. Furthermore, given the number of complicating factors and the lead-lag between cause and effect in medicine, it may be impossible to know what worked and what did not. Healthcare professionals live in a world of uncertainty, ambiguity and increasing frustration and hostility, largely due to unrealized expectations.

Both the internal and external environments of healthcare have become adversarial: “most systems have formed as a defense against an increasingly hostile environment.” (Shortell 1993: 447) Patients expect answers for their problems and expect properly trained healthcare professionals to have these answers. These expectations often remain unfulfilled. The provider may not have the answer for a specific patient’s problems because an absolutely correct answer with a guaranteed outcome for that patient does not exist.  “When the [doctor] has to defend himself against a charge of malpractice by a plea of ignorance and fallibility, his plea is received with flat incredulity; and he gets little sympathy.” (Shaw 1913, 13) What Ashmos wrote about business managers is even more applicable to doctors and nurses: “When managers try to impose order and eliminate ambiguity, it is because ambiguity is seen as a form of ignorance and acknowledging ignorance brings out insecurities.” (Ashmos 1998: 591) Finally, healthcare workers are socialized to do good things for their patients but if outcomes are unclear or delayed, how can one be sure that he or she is doing good, and worse, how can they improve in a culture that discourages innovation? (Deal & Kennedy 1982; Collins & Porras 1997)

Begin with the end in mind, a widely touted management adage (Covey 1989), warns us that we must know where we are going; otherwise, we will never get there or even know that we got there. It is instructive to examine this aphorism in the context of a thinking system. Those who take the adage literally expect to define in advance the precise and specific outcomes of a stragic plan. However, the actions of a thinking system are influenced by the system components (humans) and therefore results emerge rather than follow inexorably and predictably. Secondly, business adages cannot be simply applied to a thinking system without significant adjustment. All too often, companies or institutions take a management idea and apply it, without modification, to a thinking system. This rarely produces the result intended. Many companies jumped on TQM (Total Quality Management) as a solution to near-term fiscal crisis only to be disappointed with the results. TQM involves cultural transformation. Without such fundamental change, TQM cannot succeed. With such change, success will be seen over many years to decades, not the next budget cycle.

HEALTHCARE: A DYSFUNCTIONAL THINKING SYSTEM

We believe that the system-wide problems in healthcare can be linked to the fact that it is a thinking system acting on a thinking system, while both are viewed as and managed like machines. We list below some root causes for dysfunction. (Waldman & Schargel 2003)

1) Timeline of Causality: There is a long delay—a large lead-lag or stock-and-flow problem—between action and effect, especially in preventative medicine.

2) Improper outcome measures: Currently we track what we do not want (death and cost) rather than what we desire: life, function and productivity.

3) The Substrate: The primary input to the system—the raw material—is people: thinking, feeling, willful humans. Healthcare is a people-processing system. There are limited numbers of patients with rare but severe conditions and one cannot make more to study. Patients often behave in ways that are not in their own best interests.

4) Micro-economic disconnection: In medicine, payments are fixed (regulated); demand is variable and is not linked—micro-economically—to reimbursement.

5) Contradictory incentives: In 1975, Kerr asked what you get if you want “A” but reward “B”. (Kerr 1975) In healthcare, patients want care but the system rewards productivity (patients per hour); is anyone surprised that efficient doctors spend little time with their patients?

6) Organizational structure: Shortell et al have argued that healthcare systems have failed “to behave as systems”. (Shortell 1993: 447) Martin Hickey, former CEO of Lovelace, said, “There is really no such thing” [as a health care system], just “‘silos’—individual interest groups ranging from insurance companies to physicians fighting for increasing scarce resources.” (Quigley 2002: A2)

7) Cosmology Episode: The worldview of most healthcare workers is what Weick called a “cosmology episode” (1993: 633) “when people suddenly and deeply feel that the universe is no longer a rational, orderly system.” Weick related this phenomenon to “vu jàdé –the opposite of déjà vu: I’ve never been here before, I have no idea where I am, and I have no idea who can help me.” (Weick 1993: 633)

UNINTENDED CONSEQUENCES

Logical, well-intentioned actions often result in unintended and often adverse outcomes. Low flush toilets increase water usage. The cost for Medicare was more than 800% above projection and helped create the impression that health care is a Right. The recently implemented Health Insurance Portability and Accountability Act has guidelines without protections that will create lawsuits over confidentiality issues, either for withholding clinically important data or for inappropriate transmission of protected medical information. Unintended consequences are so common they have become a law. Dee Ward Hock, founding CEO of VISA International and author of the Birth of the Chaordic Age, (1999) memorialized “The Sheep’s First Institutional Law of the Universe: Everything has both intended and unintended consequences. Intended consequences may or may not happen; unintended consequences always do. (Peirce 2000, 11)

In the 1980s, managed care (or “managed cost” per Klienke 1998) was adopted as a means to reduce the escalation of health care costs. Managed care was based on the sound concept that it was cheaper to prevent ailments than to treat them, in the long-term. However, the timeline between cause and effect in medicine—years, even decades—was forgotten or ignored and the focus became next month’s budget, not patient well being in twenty or thirty years and expenses that were avoided. Short term, linear thinking led to the following rationale. Preventative medicine is very expensive in the short term and most patients change health care plans every 2-3 years. Why should we spend current dollars on a patient who will not be in our plan when the complications of the present condition and the attendant costs, viz., of diabetes or hypertension or smoking, are years in the future. Therefore, it is not in our economic best interest to pay for patient education or in-home screening or smoking cessation programs.

The least care is the best care (financially). Therefore, the system is efficient when it: a) Erects barriers for patients to enter the system, b) Prevents doctors from ordering tests and consultations, and c) Discourages expensive procedures from coverage. The net effect of managed care is to reduce access, restrict tests and procedures, and frustrate both care-providers and patients. Managed care has changed the doctor-patient relationship from fiduciary [“a person to whom power is entrusted for the benefit of another”] to “accidental adversaries”. (Aronson 1996-8)

“When managers take explicit actions to solve one problem they sometimes create another.” (Ashmos 1998: 44) Indeed, it seems to be the norm that healthcare managers and regulators create new, unexpected and worse problems when attempting to solve the current, immediate issue, as exemplified from an actual case study shown in Figure 2 below.

Figure 2: Fixing The Operating Room — A Systems Analysis

​

Legend for Figure 2

The Figure shows how silo (non-systems) thinking can induce actions that lead to an effect opposite of the desired one, viz., fixing the operating room (OR) budget can cause closure of the OR.

Panel A:  A hospital has finds an excess number of nurses in the OR for the number of surgeries done. As the hospital wants a balanced budget, there is a Gap (#1). Remedial action is taken by reducing the number of OR nurses thus reducing waste (intended effect).

Panel B: Reducing the number of OR nurses creates a delay or constraint in throughput of surgical patients. In turn, patients wait in hospital for operations (greater length of stay), doctors refer their patients to other facilities and patients seek care elsewhere. Hospital staff and physicians become frustrated and seek work at other hospitals increasing turnover. These effects would be called positive feedback or balancing loops.

Panel C: The downstream consequences of the effects seen in Panel B include fewer referrals to the index hospital, fewer surgeries producing less revenue, and higher turnover generating additional costs.

Panel D: The net result of the intermediate effects in Panels B and C is an imbalance, again, between number of OR nurses and the OR budget, creating a new Gap (#2). Using the original reasoning, more nurses would be terminated causing a repetition of the first series of interrelated events, ultimately leading, quite logically and inexorably, to closing the OR.

A New Mexico newspaper headline stated, “Governor Richardson’s administration is proposing a 3.4 percent reduction in payments to health care providers in order to help slow the growth of Medicaid spending.” (Albuquerque Journal, January 3, 2004) While this was intended to reduce costs, it will drive doctors away, force more people into emergency rooms and increase the need for acute care, and thereby ultimately increase costs.

Unintended consequences occur for several reasons. In addition to the long delay between cause and effect, many problem solvers believe that good intentions and logic are sufficient to achieve positive, predictable outcomes. There is no obligation among either managers or legislators/regulators to produce—in advance of action—evidence-of-desired-effect. (Axelsson 1998)  Silo thinking is common and many do not consider the broader context of narrowly drawn decisions. Most important, the nature of a thinking system is to innovate, to co-evolve and self-organize, to modify the process, and to produce results different from those initially intended. It is as though the auto assembly line decided to reorganize the car, trying the steering wheel in the trunk and the headlights on top of the hood.

WHAT IS SYSTEMS THINKING?

Systems thinking embodies an approach to understanding how things work first described by Ludwig von Bertalanffy. (Bertalanffy 1975; Davidson 1983) Numerous strategists have modified and expanded the concepts over the years (Ackoff 1999; Aronson 1996-8; Ashmos 2000; Beinhocker 1997; Kauffman 1980 & 1995; Lazlo 1972; McDaniel 1997 & 2001; Senge 1990; Sterman 2002). The central thesis is the effects or output of any system are dependent on the interactions of its parts and that studying the parts in isolation will not provide an accurate picture of the system. Network analysis [Wellman & Berkowitz 1988; Scott 1991; Ibarra 1993; Marsden & Friedkin 1994; Stephenson 1996; Rowley 1997], like systems thinking starts with a similar assumption that one must study the multiple, cross-level interactions over time rather than simple dyadic interplay. Measuring wheel size and crankshaft length provides no understanding about what a car can do. Furthermore, optimizing the function of parts in isolation (rather than in relation to all other parts) often degrades net system outcomes. (Ackoff 1999) Systems thinkers distinguish between analysis, with its root lysis (“to cut or break apart”) from anasynthis with its root in synthis (referring to make whole or put together; Aronson 1996-8) where one studies an intact whole rather than the system parts. Detailed study of hydrogen and oxygen separately provides no understanding of the wetness of water. Management theorists make an analogous distinction in how organizations deal with complexity. (Alexander 1986; Ashmos 2000; McDaniel 2001) Those using silo thinking and analysis seek to simplify, to reduce complexity. Those employing systems thinking absorb complexity and diversity, making the whole stronger (anasynthis).

Most current management approaches are short term, silo, linear, and binary (StSLB). Consider healthcare. St) We track 30 day survival after surgery rather than functional status ten years later; S) Silo thinking refers to solving problems in isolation, such as giving a drug to improve kidney function without regard to its effect on the liver; L) As elucidated by Sterman (2002), most analysts think in terms of flow charts and interactions in a straight, linear sequence rather than multi-layered feedback interactions over time (See Figure 3 below); and B) Only positive or negative effects are considered, rather than variable degrees of effect. While life and death are mutually dichotomous, they are not exhaustive. One can be alive normally or be incontinent or blind. Thinking in binary alternative outcomes is incomplete. Some examples of StSLB thinking in healthcare are shown in Table 2 below.