|Home | About | Journals | Submit | Contact Us | Français|
The strengthening of health systems is fundamental to improving health outcomes, crisis preparedness, and our capacity to meet global challenges, such as accelerating progress towards the Millennium Development Goals, reducing maternal and child mortality, combating HIV, malaria and other diseases, limiting the effects of a new influenza pandemic, and responding appropriately to climate change. To meet these complex needs, the Association of Schools and Programs in Public Health, the World Health Organization, and the Institute of Medicine promote systems thinking as the only sensible means to respond to issues that greatly exceed the normal capacity of health and medical services. This paper agrees with the application of systems thinking but argues that health organizations have misunderstood and misapplied systems thinking to the extent that the term has become meaningless. This paper presents the basic constructs of systems thinking, explains why systems thinking has been misapplied, examines some misapplications of systems thinking in health, and suggests how the concept can be applied correctly to medicine and public health to achieve the reason it was adopted in the first place.
Effective and capable health systems are a vital component in our approach to managing a wide array of health challenges. However, global threats and even local health service problems can appear overwhelming and without cure due to their inherent complexity. In response, many have encouraged the use of systems thinking as a means of deciphering complex situations and preparing adequately for complex events, such as crises and disasters, that have the potential for adverse outcomes. In health, a systems approach was promoted by the World Health Organization as a “fresh and practical approach to strengthening health systems,”1 and by the Institute of Medicine as a means to respond to an incident that greatly exceeds normal demands on health and medical capacity.2 Systems thinking has even become enshrined as an essential competency for all public health graduate students in the United States.3 The concept is fairly straightforward with the systems approach being defined as a “management strategy that recognizes that disparate components must be viewed as interrelated components of a single system.”4 What makes systems thinking important? The famous philosopher, Edgar Singer, showed that all the known sciences and professions are engaged in the act of measuring the distance from one point to another.5 When the measurement becomes difficult, more observers are employed and they need to collaborate. As the process of measuring becomes more complex, more fields are involved, such as the social sciences, political science, psychology, mathematics, and others. Eventually, an overarching management plan is required to ensure that all endeavors are aligned to achieve the end goal. Applied to a public health issue, encouraging an obese population (Point A) to become a trim population (Point B), is a challenge that requires the involvement of many branches of knowledge that only a complex systems approach can adequately manage.
More often than not, our training teaches us to solve complex problems by breaking them up into discrete, manageable pieces and working to ensure each piece functions and performs as well as possible. However, much like the human body, it is not how well some parts are functioning that limits progress; it is how poorly other parts are doing. The success of some pieces does not equate to success of the whole and sometimes the alteration of one part can affect the performance of the whole. As our awareness of interactions and interdependencies grows, we increasingly recognize the limitations of reductionist approaches to understanding complex health problems that have many determinants. Understanding the whole system is more than being aware of the distinct components. It is about being aware of the relationships between the pieces and how they change when a single piece is altered. Most management textbooks briefly describe systems thinking as a recent historical trend along with the commonsensical contingency viewpoint and total quality management framework, and rarely is any further mention made on the topic or on systems approaches.6–8 Inquiry systems are the basic constructs of systems thinking. They are never mentioned in these textbooks, but are essential to understanding how to implement systems thinking correctly.
Going back to basics, there are five different types of inquiry systems that we use to know something.
The first is based on agreement and is called the Consensual inquiry system. “The objectivity of science depends wholly upon the ability of different observers to agree about their data”.9 In this case, inputs are processed by experts who all agree on a single answer. The Delphi Method provides a good example of this construct.10 In an iterative process, the viewpoints of various experts are obtained on a particular issue. Since consensus is the goal, the results are shared with participants in each round. Participants are influenced by the results to conform to the majority view and non-conformers (outlying views) are discarded as being invalid.
Analysis is the second inquiry system. In this case, the inputs are processed using an agreed upon formula and a single numerical answer results. The decision-making matrix provides a good example of this system. Mitroff and Linstone argued that “agreement” is an inductive-consensual inquiry system, “analysis” is a deductive inquiry system, and that both are only appropriate for simple, well-bounded, and well-structured problems or pieces of problems for which single numbers can serve as answers.11 The eminent Scottish philosopher David Hume clearly showed that the agreement and analysis constructs cannot be used to accurately establish causality.12 Hume argued that mere observations alone could never establish the logical necessity (causation) between two events because they have no insight into the inherent nature and intangible features of the events.
The third type of inquiry system is “multiple realities” which is based on the notion that the use of many models and observations will shed the best light on an issue because more angles will be covered.5 It moves from “analysis” to “meta-analysis” to produce explanations that individual analyses may not identify and is thus suitable for analyzing complex problems. The drawback of this system is that it assumes that each reality is unbiased.13 We can never collect information about a problem without first having some preconceived notion of what the problem is. Our initial sense, vague understanding, or intuitive conception of a problem is an internal model that is a source of bias called “Kant's problem”. Furthermore, the multiple realities inquiry system is reductionist because it frames its results and presents problems in a technical context. Thus, the intangible issues like ethics, values, choice, culture, and aesthetics are omitted because they cannot easily be reduced to numbers that are truly representative.
In stark contrast to the Consensus model, the fourth approach to complex problems, the Dialectic inquiry system, uses contrasting views from experts that are furthest from the average.14 This is another form of the Delphi called the Policy-Delphi that is especially common in law, pros vs cons assessments, and the use of devil's advocates. The assumption is that truth will emerge from the confrontation of opposites. Decisions are based on the stronger argument, which makes them dismissive of other points of view. This approach can be useful in dealing with complex problems because it can help to break down the faulty assumptions underlying the definition of a problem. However, it is simplistic given that intangible differences between presenters due to their inherent capacity and ability to convince others of their viewpoint can bias the entire process.
Mitroff and Linstone promoted a fifth type of inquiry system, which they called Unbounded Systems Thinking (UST) or the Multiple Perspectives inquiry model.11 It concerns the management of all modes of inquiry and it incorporates the other inquiry systems. The inputs are complex “messes” and may be viewed in a variety of ways, such as epistemic, ethical, aesthetic and spiritual. This approach is based on the idea that problems are the product of their interactions and not the sum of their parts. It is based on the belief that, because everything interacts with everything, the widest possible selection of branches of knowledge must be brought to bear on problems. The traditional hierarchy of the sciences and professions is abandoned in favor of equal weighting. This approach is recommended for complex, highly ill-structured, unbounded problems for which systemic views are essential even if they lead to confusion or stalemate (eg, global poverty or starvation).
The following outcomes show how to use these five systematic pragmatic inquiry systems by illustrating what happens when one moves from one model to another.
Thus, when initiating an investigation into a problem, it is most effective to begin with the UST inquiry model and only move downward to a less complicated model if the complexity of a situation warrants a downsize. Attempting to upscale from a simple model, such as Consensus or Analysis, which is based on a single truth, to a more complex model, that incorporates many variant truths, is not recommended because belief in a single truth cannot be up-scaled. For example, ramping up an antiretroviral therapy program from one local population to a large regional population can be expected to have profound systemic effects, especially in less robust health systems.15,16 The different populations might each have their own version of the truth that did not conform to the regional model.
By using this process of managing systems thinking, we can gain insight into the reasons underlying incorrect problem formulations and bad decisions.
Recognizing that a problem exists and taking action to solve a problem are two activities that many do well, but there is an in-between step that often receives inadequate attention. The source of all important management blunders can be traced back to a mistake in accurately discerning what is important in a situation and incorrectly formulating a problem to be addressed.17–19 Challenges to systems thinking come from individuals, institutions, and society which put overt and covert blockers in the way of our ability to discern, assess, formulate, and solve problems. These blocking features can become complex and form their own system. For example, many company executives promote rationalizations that are a mix of truth, half-truth, distortions, and lies. They shamelessly promote TV violence, unhealthy food advertising, unethical behavior, tobacco, and alcohol when both anecdotal and evidence-based data clearly show their detrimental effects on individuals and society. They argue that freedom of expression is a higher cause or they use denial that a problem exists because it means that they do not have to participate in the solution.
Mitroff attributes this oversight to fundamental differences between critical and uncritical thinkers.20 While uncritical thinkers tend to focus on minimizing Type I and II errors, critical thinkers focus on Type III errors. Type I and II errors are taught in statistics classes and are associated with there being significant differences when we think there are none or vice versa. These errors relate to actions that are taken to solve a problem after it has been formulated. Type III errors occur during problem formulation and are caused by picking the wrong stakeholders, selecting a limited set of problem-solving options, incorrect phrasing, incorrectly defining the boundaries of a problem, and/or failing to think systematically by focusing on a part of a problem and ignoring connections.
The bewildering divorce between problem formulation and ethical behavior in company executives leads to inappropriate actions, which in turn lead to serious systemic breakdowns or malfunctions between technology, people, and organizations or societies. The Consensus and Analysis models are useful in investigating technological systems because they are typically simple in nature. Personal inventories and 360 evaluations are commonplace now as people seek a deeper understanding of themselves and how they relate to others. Understanding organizational and societal systems is more difficult and there is no agreement on what approach to use. Since these approaches are generally used without regard for extenuating circumstances in health, here are four different perspectives that each demonstrate certain limitations. The first three are used widely in health and other disciplines and the last is the only one that comes close to true systems thinking.
1) Using an approach called Industrial or System Dynamics, Jay Forrester, an engineer at the Massachusetts Institute of Technology, developed an approach to problem investigation in the 1950s.21 Forrester used mathematical computer simulations to provide insight into the flows and materials in subunits and the unintended consequences of problems. His work has been broadly applied to understanding flow-on problems in business, government, and research.
2) A more recent and subjective “technique” that attempted to claim the label “Systems Thinking” was developed in the 1980s by Peter Senge22 and has no clear definition or set of instructions.8 Systems Thinking, as defined by Senge, involves understanding how the pieces influence each other within a whole and it views problems as pieces of an overall system. However, rather than focusing on the problem pieces, which might have unintended consequences, it highlights cyclical relationships and nonlinear causes and effects in organizations. It has been useful in drawing attention away from problematic people and enabling people to concentrate on problematic systems.23 Efforts to create a framework for this approach use the term systems thinking loosely, revert back to mathematics and do not cater to intangible elements.24,25
3) Niccolo Machiavelli observed how leaders related to their internal followers and how they dealt with external competitors and formulated a very different system to understand people and organizations.26 Although he recognized that some people were altruistic and moral, he found that most people were for the most part naturally and selfishly engaged in the ruthless pursuit of their own self-interest and that this activity was occasionally counter-intuitive and even irrational. Machiavelli also found that people naturally tended to exist within social or work groups in which they cooperatively exchanged resources or information. However, this cooperation was only seriously pursued when people believed that it would advance their own interests. If one accepts this interpretation of humanity, it is difficult to see how any methodology can accurately model systems that contain unknown personal interests and occasional irrational decision-making.
4) Unbounded Systems Thinking differs fundamentally in that it considers all inputs and accepts the messiness of all inputs.11 By doing so, it attempts to include, to whatever extent possible, knowledge of personal interests and counter-intuitive possibilities. Some of its core beliefs include recognition that:
- The designer of a system has a particular psychology that is important to know
- Strong consideration of ethics, values, judgments, and background experiences underlie the choice of which perspectives are brought to bear on a problem
- The combined use of multiple technical, individual, and organizational perspectives to view any problem yield insights that each perspective could not reach alone
- Some subjective and intangible problem variables cannot be determined as “correct” and therefore cannot be assigned a weighting or value
- Perspectives may change over time as new information is revealed
The need to understand complex systems with the aim of strengthening national and local health systems is well accepted and yet there have been few established frameworks or methodologies to implement.28 Less resourceful health systems lack the capacity to assess barriers to their success which provides little guidance to policy-makers. Even very simple health interventions can fail to achieve their goals due to the occasional unpredictable behavior of the system within which it operates. Every intervention affects its containing system, and every overarching system affects its interventions.1 For example, reducing the length of stay in one hospital ward may result in increased re-admission rates in another division which compromises overall quality and costs.28 Interactions at the systems level can be chaotic, with actions having unclear relationships with outcomes. For example, interventions that aim to increase the quality of care usually succeed initially, but can decrease the quality of care over time as skills reach a peak and workloads become unacceptable.29 For these reasons, systems thinking has been promoted in health. But, as can be seen from the literature thus far reviewed, it remains somewhat abstract and the methodology is impractical, simplistic, or omits key parts of systems. To further complicate matters, ignorance of how humans acquire knowledge has resulted in the wrong inquiry models being used under the label of “systems thinking.”
Take, for instance, the Association of Schools and Programs in Public Health's (ASPPH) inclusion of systems thinking as one of seven key interdisciplinary competencies for the nationally certified US Masters of Public Health degree.3 These competencies were developed using a 3-stage Delphi method.30 In the first round, there were 58 competencies relating to systems thinking, which were whittled down to 32 and 14 in subsequent rounds. In this case, a simplistic Consensus inquiry model was used to assess competencies for a very complex multidisciplinary program. Furthermore, many of the eleven systems thinking sub-competencies derived from Forrester's System Dynamics limited method of understanding organizational subunits. This is very unlike public health, which usually draws from multiple perspectives, leaving one questioning the validity of the competencies. From a consultant's viewpoint, the use of Systems Dynamics is favorable because its complicated implementation instills dependency that requires fee-for-service handholding. From a broader systems perspective, this approach to competencies began with the wrong inquiry model and uses a limited methodology. From a public health perspective, it is impossible to achieve many of the competencies beyond a superficial level because they are overly ambitious. For example, see #10. Analyze global trends and interdependencies on public health systems. Even attempts to apply the Delphi method to health disparities research resulted in simplistic models that presented only part of the whole picture.31 From a common sense perspective, it is unreasonable for the ASPPH to require every public health student and teacher in the United States to be trained in systems thinking methodology that was not derived from a logical and appropriate systems thinking process.
A broader approach to systems thinking has been developed and applied by the Alliance for Health Policy and Systems Research in the World Health Organization.1 They view systems thinking as a powerful approach for guiding investments in health systems and they believe that failing to take health system complexity into account hinders efforts to achieve equitable outcomes. They are interested in moving away from vague and limited approaches to providing health workers with a clear method to strengthen health systems efficiently and equitably. They outline a process that includes four steps during the intervention design phase and six steps during the evaluation design phase. The evaluation steps are project management and not systems thinking, but the intervention steps demonstrate careful consideration of systems thinking. They are: convene stakeholders, collectively brainstorm, conceptualize effects, and adapt and redesign, which are all necessary components of problem formulation.
Given that all evaluations are necessary simplifications of real-world complexity, systems thinking helps to determine how much and where to simplify. If one accepts that the priority is to minimize Type III errors in problem formulation, the reason for the systems thinking changes from the UST model of, “We need to understand everything about everything” to a more focused model of, “We need to know what is important.” This different perspective is far more focused, requires far less effort and is thus more achievable even in under-resourced health organizations. What is required for its success is a list of key activities (closely tied to the strongest reasons for Type III errors) and leaders who are capable of discerning what is important. This latter point is important because the ability to discern what is important was the highest ranked quality of excellent leaders in business, health, and emergency management professions across three countries.32
Addressing the complex problems in healthcare systems is never going to be easy, but some knowledge on how to proceed is better than none. Leaders have to lead, managers have to manage, and we all need to make decisions throughout our lives. It is better that we do so without being ignorant of the psychological constructs and constraints within which we choose to operate. The careful selection of inquiry models is fundamental to systems thinking, which, when used appropriately, can assist us in correctly identifying the underlying problems that prevent progress. Above all, we need to move away from the simple consensus and analytical inquiry systems favored by economists because they fail to recognize that social dynamics are fraught with counter-intuitive behavior, that events may have multiple effects on different time scales, and that effects may gain sufficient momentum to become independent from causes and in turn become causes of further events.
Funding for global health concerns and health systems has steadily increased over the past decade, but this is entirely due to private increases in the Global Fund because government contributions have decreased by a third in the plans for 2012–2014.33,34 Private funders are more discerning and need to know what works in what context. They understand that even simple health interventions are highly complex so knowledge of the full range of systemic effects is imperative to mitigating detrimental behavior and amplifying possible synergies. Mere knowledge of a health system is insufficient and an appropriate inquiry method and systems perspective are required. Leaders need to be able to discern what is important so that they can employ the correct inquiry system and formulate problems correctly. Only with this approach can they be certain that they have taken every reasonable precaution before implementing a program or change or new course of action.
Jay Maddock, Office of Public Health Studies at John A Burns School of Medicine.
Donald Hayes, Hawai‘i Department of Health.