Public 1
100honestyworkCHAPTER 14
From Single Solutions to Systems Thinking—The Future of Population Health
© Elena Itsenko/ShutterStock, Inc.
LEARNING OBJECTIVES
By the end of this chapter, the student will be able to:
• explain how systems thinking differs from reductionist thinking.
• identify characteristics of a system.
• identify the steps in systems analysis using systems diagrams.
• explain the meaning of interactions between factors.
• explain the meaning of bottlenecks and leverage points.
• identify and explain seven uses of systems thinking in public health.
• explain the requirements for successful systems analysis and the limitations of the approach.
You are pregnant and have a 10-year history of cigarette smoking. You are surprised that at your first prenatal visit, there is a big sticker on your chart saying “Smoker.” Everyone in the doctor’s office asks you what they can do to help, and they quickly enroll you in special services for smoking cessation for which you were not eligible before you got pregnant. When you ask why so much time, attention, and money is now coming your way, they tell you pregnancy is a leverage point for stopping smoking. What do they mean by “leverage point” ? you ask yourself.
A patient with active TB is reported by the local hospital laboratory to the health department. The health department quickly connects with the patient to determine his close personal contacts. They also ask him if they can test him for HIV. He turns out to be HIV positive, and permission is then requested to get in touch with his sexual contacts. You consider how you would describe the relationship between TB and HIV, and wonder how knowledge of this relationship can be used to reduce the risks of both TB and HIV.
A new type of large research study looks at four FDA-approved treatments for myocardial infarction. The study includes over 50,000 patients and is designed to determine not whether one treatment is better than another but instead whether there is any clinically important difference between the outcomes of the treatments. What type of research is this, you wonder, and why is it being done?
An expert panel reviews recent data in an effort to predict the incidence of HIV in 2020. They take into account the current data on incidence and prevalence as well as the effectiveness of known interventions. Unfortunately, within two years after the effort is completed, it is apparent that their estimates are quite inaccurate. Why is prediction so difficult ? you wonder.
You hear that motor vehicle injuries, especially those due to automobile collisions, have been dramatically reduced in recent years. Was there a magic bullet that accomplished this, you wonder, or was this reduction accomplished through a more complicated process ?
You love rare hamburgers. “Just wave them over the flame,” you like to say. Recently, you have heard that ground beef is a high-risk food—even a health hazard. What does that mean, and what is being done about it ? you ask yourself.
© 3DProfi/ShutterStock, Inc.
“Over the last 100 years, public health has evolved from having primarily a science focus to incorporating a problem focus and, most recently, a systems focus.”
—Commission on Education of Health Professionals for the 21st Century 1
Let us take a look at how population health thinking is changing in the 2000s and gradually incorporating what we will call systems thinking. The focus on systems is so important that along with the population perspective, it is coming to define 21st century population health. The importance of systems thinking in public health education has been recognized by inclusion of systems thinking as a “cross-cutting” competency that is being included in the public health certifying examination. 2
We begin by taking a look at what makes systems thinking different.
WHAT MAKES SYSTEMS THINKING DIFFERENT?
Traditional thinking in public health, like most science-based disciplines, has used what is called reductionist thinking. Reductionist thinking looks at one factor or variable at a time. 3 That is, it reduces the problem to one potential “cause” and one potential “effect.” a
Reductionist thinking has often been used in public health and medicine to search for the one-and-only answer to the why, or etiology, and the one-and-only answer to what should be done to improve outcome. This approach may be called the magic bullet or miracle cure approach.
Reductionist thinking has been very useful for establishing specific factors as contributory causes of disease, such as cigarettes and lung cancer, high blood pressure and vascular disease, as well as aspirin and Reye’s syndrome. However, it is increasingly important that we look at the impacts of multiple factors and see how they work together as parts of systems. Systems thinking, or a systems approach, often utilizes data derived from reductionist thinking but goes beyond reductionist thinking to look at multiple factors that cause disease and disease outcomes. Thus, it is the focus on multiple factors and how they fit together that distinguishes reductionist thinking from systems thinking. 4 , b
WHAT IS A SYSTEM?
There are a large number of definitions of a system. We will define a system as an interacting group of items forming a unified whole. According to O’Connor and McDermott, the key to identifying a system is that a system maintains its existence and functions as a whole through the interaction of its parts. They write:
Your body is the perfect example. Your body consists of many different parts and organs, each acting separately, yet all working together and each affecting the others. Your thoughts affect your digestion and heart beat, the state of your digestion affects your thoughts—especially after a large lunch. The eye cannot see, nor the legs move without a blood supply, and the blood supply has to be oxygenated through the lungs. The movement of the legs helps pump the blood back to the heart. The body is a complex system. 6
It is important to appreciate the features of a system and the implications of a system. O’Connor and McDermott distinguish a system from what they call a heap or collection of pieces as follows:
• A system is a series of interconnected parts which function as a whole. A heap is merely a collection of parts.
• A system changes if you take away or add pieces; if you cut a system in half, you do not get two smaller systems: you get a system that will not function. A heap can be divided into pieces, each of which can function on their own.
• In a system, the arrangement of the pieces is crucial, while in a heap, the arrangement is irrelevant.
• In a system, the parts are connected to each other and work together, while in a heap, the arrangement of the pieces is irrelevant.
• The behavior of a system depends on its overall structure, while in a heap, size rather than structure determines behavior.
As we have seen, the term “system” may be used to describe complex biological relationships. “System” may also be used to describe organizations’ relationships or processes, such as a healthcare system, a public health system, or a research system. Alternatively, “system” may be used to describe the working of factors or influences that bring about disease and the outcome of disease. Each of these uses of the term “system” shares the intention of understanding how the pieces or items fit together in a coherent whole. In recent years, population health has increasingly turned to systems thinking to better understand the operation of organizations and processes, as well as the development and outcome of disease. c
Implementing or operationalizing systems thinking requires tools for analyzing the pieces and understanding how they fit together. This process is called systems analysis. There are a wide range of these tools. Often, systems analysis relies on diagrams or graphics that visually display the relationships between the parts and allow us to better understand how the parts fit together and work together.
Let us look at some important methods of systems analysis that are being used in population health.
HOW CAN SYSTEMS ANALYSIS BE USED TO UNDERSTAND THE HEALTH RESEARCH PROCESS?
In the research arena, population health research is now being viewed as part of an integrated continuum of research designed to focus on outcomes at the population as well as the individual level. This came about in large part because the National Institutes of Health has developed and used a systems analysis approach to develop what is being called translational research.
Translational research began as an effort to connect basic science with clinical care of individuals and was described as “from the bench to the bedside.” More recently, this movement has been extended to the community and the population and may be described as “from the lab to the living room.” 7
We can view translational research as an attempt to bridge the gaps between basic research, clinical applications, and population health implications by seeing health research as a system. Figure 14-1 outlines the key components of translational research, which are increasingly guiding the distribution of research funding through the National Institutes of Health and other health research funders.
Translational research is often thought of as having three components called T-1, T-2, and T-3. T-1 refers to the now traditional efforts to establish efficacy by using randomized controlled trials. T-2 distinguishes between efficacy under experimental conditions and effectiveness in public health or clinical practice. This concept of effectiveness includes not only how well an intervention works but also its potential harms and, at times, its costs. That is, it includes all of its outcomes.
FIGURE 14-1 Translational Research Phases T-1, T-2, and T-3
Adapted from Dougherty D, Conway PH. The “3T’s” road map to transforming US health care: the “how” of high quality care. JAMA. 2008;299:2319–2321.
T-2 research also now includes efforts to compare two or more interventions for the same conditions, or what is being called comparative effectiveness research. For instance, comparative effectiveness research may examine a series of FDA-approved treatments for diabetes or a series of drugs that are FDA-approved for anticoagulation but have different side effects and different costs.
A new type of very large randomized controlled trial called an equivalence trial or a noninferiority trial is increasingly being used to study the relative effectiveness of currently approved and utilized interventions. These studies are designed not to determine whether one treatment is better or superior to other treatments but instead to determine if their effectiveness differs in clinically important ways. If no important differences in effectiveness can be demonstrated, decisions can be made on other grounds, such as safety, convenience, and/or cost. 8
T-3 research looks at the delivery of clinical and public health services. It asks what factors influence the success of interventions that appear to work in T-1 and T-2 studies. T-3 research is increasingly being guided by the RE-AIM framework, which seeks to evaluate the impact of research in practice. RE-AIM is a mnemonic that stands for reach, effectiveness, adoption, implementation, and maintenance. You can think of the “RE” factors as evaluating the potential of the intervention for those it is designed to include or reach. The “AIM” factors examine the realities for application of the intervention in clinical or public health practice. The RE-AIM framework may help us better understand how to translate new interventions into improved outcomes as part of a system. In describing the RE-AIM framework, Laurence Green, one of its developers, has written, “If we want more evidence-based practice, we need more practice-based evidence.” 9
The translational research framework can be viewed as a basic systems analysis diagram. It identifies the pieces, looks at how they should connect, and attempts to put the pieces together as a coherent whole. Despite the simplicity of the diagram, thinking in systems and backing the process up with financial support is having a major impact on how the research process is viewed and how population health in particular is integrated into the research system.
In addition to being useful for better understanding processes such as research, systems analysis is increasingly being used to understand the factors or influences that bring about disease as well as the outcomes of disease.
WHAT ARE THE INITIAL STEPS IN SYSTEMS ANALYSYS?
When using systems analysis to understand disease and its outcomes, we need to start by identifying the most important influences on the outcome(s) of interest. Influences are factors or determinants that interact with each other to bring about outcomes, such as disease or the results of disease. d
Let us see how we might identify influences on smoking cessation. Using one-at-a-time reductionist studies, the following interventions have been shown to be effective: smoking cessation programs, prohibitions on smoking in public places, social marketing, and cigarette taxes. In addition, measures of the strength of a relationship such as relative risk obtained from reductionist studies often help us measure the magnitude of the influence that a factor has on an outcome. Thus, the first two steps in systems thinking are often built on data derived using reductionist thinking.
Rather than looking at one intervention at a time, however, systems thinking asks about the best combination of interventions and how they can be used together. Let us assume that smoking cessation programs, prohibition on smoking in public, social marketing, and higher taxes have been identified as the four most important interventions or influences on the rate of cigarette smoking. The question then becomes how they can be effectively and efficiently combined.
Reductionist thinking usually assumes a straight-line or linear relationship between influences, implying that increased levels of an intervention, such as increasing taxes on tobacco, will produce a straight-line decrease in the levels of tobacco use. However, it is possible that small increases in taxes have little effect, while somewhat larger increases have dramatic effects. In addition, reductionist thinking does not look at how the impact of one intervention may be affected by connecting it with other interventions, whereas systems thinking looks at these interactions. Thus, systems thinking would ask questions about how to most effectively utilize cigarette taxes by combining them with other approaches, such as using the taxes to support tobacco education programs, or reduce exposure to other causes of lung cancer, such as radon and asbestos.
The following summarizes the initial steps in a systems analysis:
• Step 1: Identify the key influences or interventions on an outcome such as disease or the outcome of disease.
• Step 2: Indicate the relative strength of the impact of each of the influences or interventions.
• Step 3: Identify how these influences or interventions interact—that is, how they work together or, alternatively, interfere with each other.
These three steps in systems analysis are basic to understanding the structure of a system.
WHAT ADDITIONAL STEPS ARE NEEDED TO COMPLETE A SYSTEMS ANALYSIS?
Increasingly, we are taking the process one step further. We are using systems analysis to better understand how systems function. Systems thinking requires not only an examination of multiple influences and their interaction at one point in time using a static approach but also encourages us to look at how these factors change over time. That is, systems thinking can lead to a dynamic approach. Let us see how this may be accomplished.
Systems analysis attempts to take into account changes in the overall system that occur over time due to changes in one or more of the factors or influences. These changes in a factor or influence are said to provide feedback into the process producing what are called feedback loops. Feedback loops can have either positive or negative impacts on the outcome. For instance, in developing a systems analysis, we might ask: Does the reduction in the percentage of people who smoke due to higher taxes lead to changes over time in social attitudes, which themselves may set the stage for greater enforcement of public smoking regulations? This would be a positive feedback loop. Alternatively, raising cigarette taxes might reduce the money available to individuals to pay for smoking cessation programs if these services are not paid for by health insurance. This would be a negative feedback loop. Systems thinking does not view the impact of interventions as static. Rather, it tries to develop dynamic models, incorporating the feedback processes that reinforce or accelerate the impacts or alternatively dampen or reduce the impact.
Systems analyses encourage us to identify feedback loops, including positive feedback loops that reinforce or accentuate the process and negative feedback loops that dampen or slow down the process. Feedback loops are key to understanding how a system operates or functions. Complex systems, such as the human body, rely heavily on feedback loops in order to maintain stability. When one component gets out of control, such as body temperature or hydration, other components of the system respond to maintain the body within a tolerable range. This requires positive and negative feedback loops. Similarly, communicable disease in a population is controlled to a certain extent by responses or feedback, including voluntary isolation of sick individuals, development of immunity, and, unfortunately, death of affected individuals. Understanding these feedback loops can help us improve on the natural systems that exist while utilizing the positive aspects of existing systems.
Looking at the dynamic nature of systems and the changes that occur over time allows us to identify bottlenecks that limit the effectiveness of systems and leverage points that provide opportunities to greatly improve outcomes. For instance, systems analysis might identify a bottleneck such as the need to train large numbers of clinicians in smoking cessation methods so that they can address the demand for smoking cessation services created by social marketing, increased cigarette taxes, and better drug treatments. A leverage point that might be identified is pregnant women who smoke but are highly motivated to quit due to the severe impact on their offspring.
Thus, the additional steps in systems analysis can be described as follows:
• Step 4: Identify the dynamic changes that may occur in a system by identifying the feedback loops that occur in the system.
• Step 5: Identify bottlenecks that limit the effectiveness of the system.
• Step 6: Identify leverage points that provide opportunities to greatly improve outcomes.
HOW CAN WE USE A SYSTEMS ANALYSIS TO BETTER UNDERSTAND A PROBLEM SUCH AS CORONARY ARTERY DISEASE?
Let us use each of the six steps we have identified in a systems analysis to better understand the problem of coronary artery disease.
Step 1: Identify influences—We know from reductionist research that there are multiple factors that increase the risk of coronary artery disease, including high blood pressure, high LDL cholesterol, low HDL cholesterol, abdominal obesity, diabetes, cigarette smoking, physical inactivity, family history, etc. Recognizing each of these factors has been an important part of addressing the problem of coronary artery disease. Further progress, however, requires us to think about how these interventions connect to each other.
Step 2: Estimate the relative strength of the influences—We need to estimate the relative strength or magnitude of the impact of each of the influences. We might estimate the relative risk for each of these factors, or we might classify their impacts as weak, moderate, or strong. In the case of coronary artery disease, each of these factors is considered of moderate importance with relative risks in the range of 2 to 4.
Step 3: Examine the interactions between factors—Examining the interaction between factors helps us understand what happens when two or more of the factors are present. Risk factors for disease may add together to increase the risk of disease, such as high blood pressure plus high LDL cholesterol and low HDL cholesterol. Alternatively, one factor, such as physical activity, may have a protective effect against coronary artery disease in and of itself. Interactions between factors may multiply the risk rather than resulting in an additive impact. Risk factors for coronary artery disease are usually assumed to add together rather than to multiply the impact. However, a combination of risk factors known as the metabolic syndrome has been shown to interact and greatly increase the risk. Metabolic syndrome includes increased waist circumference, low HDL cholesterol, elevated triglycerides, hypertension, and elevated fasting blood sugar. When all or a number of these risk factors occur together, they greatly magnify the probability of coronary artery disease as well as other large blood vessel diseases such as strokes.
Step 4: Identify feedback loops that lead to dynamic changes in the functioning of the system—Understanding how systems operate over time requires us to identify feedback mechanisms, or feedback loops, that alter the likelihood of disease or impact its outcome. For instance, increased weight, especially increased abdominal girth, may lead to increased LDL cholesterol, diabetes, reduced exercise, reduced HDL cholesterol, and increased blood pressure. Alternatively, multiple interventions focused on weight, exercise, blood sugar control, and treatment of hypertension may work together to have a surprisingly positive impact on the probability of coronary artery disease. e
Step 5: Identify bottlenecks—Bottlenecks imply that there are points in the system that need to be addressed in order for the other factors or influences to have their potential impacts. For instance, in coronary artery disease, if severe narrowing of the coronary arteries already exists, it is unlikely that interventions such as reducing blood sugar, reducing LDL cholesterol, increasing exercise, or stopping cigarette smoking are going to have a dramatic impact. If the bottleneck, the narrowed artery, can be addressed using angioplasty or surgery, attention to the other risk factors may have a much greater impact.
Step 6: Identify leverage points—The systems analysis that we have done so far suggests some leverage points where interventions may have greater than expected impacts. For instance, increasing exercise post angioplasty or surgery may be safer than when severe disease is present. Patients may also be highly motivated to exercise after having angioplasty or surgery. Exercise then might be effective in helping patients stop smoking cigarettes and reduce abdominal girth, as well as having an impact on HDL cholesterol and blood sugar.
Table 14-1 summarizes the steps in the process, the meaning of each step, and the examples from cigarette smoking and coronary artery disease.
At times, we may be able to use the results of a systems analysis to display the structure and function of a system using what is called a systems diagram. A systems diagram is a graphic means of displaying the way we understand systems to be structured and/or to function. Let us see how we can use a systems diagram to display the functioning of a system.
TABLE 14-1 Steps and Their Meaning in Systems Analysis
Step # |
Meaning |
Examples |
1. Identify influences |
Identify factors or determinants that are thought to affect or influence the probability of occurrence of a disease or the outcome of a disease. |
Coronary artery disease—High LDL cholesterol, cigarette smoking, increased abdominal obesity, etc., increase occurrence. Cigarettes—Taxation of cigarettes, smoking cessation programs, prohibitions on public smoking, etc., improve outcome. |
2. Estimate the relative strength of the influences |
Estimate the relative risks of each of the influences or at least the relative strength, such as weak, moderate, or strong. |
Coronary artery disease—Most important factors are of moderate strength with relative risks between 2 and 4. Cigarettes—Degree of addiction is a strong factor in determining outcome. Radon and asbestos each have relative risk of approximately 5. |
3. Examine the interactions between factors |
How is the occurrence of disease or the outcome of disease affected when two or more influences are present? Do the impacts of the influences add together, does one influence protect against another influence, does the presence of two influences multiply the impact? |
Coronary artery disease—The metabolic syndrome is an example of interactions between factors in which the presence of multiple factors has more than an additive impact. The impacts of radon and asbestos on lung cancer are multiplied in the presence of cigarette smoking. |
4. Identify feedback loops |
Identify ways that an influence increases or decreases the impact of other factor(s) over time. |
Coronary artery disease—Exercise can reduce weight, blood sugar, and blood pressure as well as having a protective impact in and of itself. Cigarettes—Reduction in the percentage of the population who smoke may encourage greater use of other interventions to further reduce the percentage who smoke. |
5. Identify bottlenecks |
Identify points in the system or constraints that need to be addressed in order for the other factors or influences to have their potential impacts. |
Coronary artery disease—Severe constriction of major artery often needs to be addressed by angioplasty or surgery to enable other interventions to work effectively. Cigarettes—Addiction often needs to be addressed directly in order for other influences to be effective. |
6. Identify leverage points |
Identify points in the system that present opportunities for interventions to have greater than otherwise expected impacts. |
Coronary artery disease—Exercise may have greater than expected impacts if used post angiography or surgery when exercise is safer and patients are motivated. Cigarettes—Interventions aimed at pregnant women may have greater than expected impacts short term and longer term because women are highly motivated to stop cigarette smoking during pregnancy. |
HOW CAN WE USE SYSTEMS DIAGRAMS TO DISPLAY THE WORKINGS OF A SYSTEM? 10 , 11
Let us use an example to illustrate the development and use of systems diagrams. We will take a look at the etiology and outcomes of motor vehicle injuries, especially automobile injuries. Box 14-1 presents the “facts” that we will use in developing our systems diagrams.
The development of systems diagrams begins with identifying the key factors that will be included in the systems. For each factor, we need to:
• Indicate the direction in which it operates, in other words, which way the arrow points
• Indicate whether the factor operates to reinforce or increase another factor or outcome, which is indicated by a (+), or operates to dampen or decrease another factor or outcome, which is indicated by a (–).
BOX 14-1 Background on Motor Vehicle Injuries as a Systems Issue
Overview
Motor vehicle injuries in the United States, and automobile injuries in particular, have been the leading cause of death for children and young adults for at least the last half century. Today, they remain a critical problem; however, the death rates from motor vehicle collisions, especially when measured as death per miles driven, have fallen so dramatically that the CDC classified highway safety as one of the 10 great public health achievements of the 20th century. This progress has not been due to any one intervention or magic bullet—it is the combination of systems thinking and coordinated interventions that have made the difference.
We might regard the dramatic fall in automobile-related deaths as a systems thinking success story. However, change brings new issues and new challenges. The widespread practice of texting while driving poses new safety hazards that need to be addressed.
We will use motor vehicle injuries as an example that allows us to illustrate principles of systems thinking. We will aim to analyze the issue of motor vehicle injuries from both an etiology and an outcome perspective. That is, we will look at both the reason for motor vehicle injuries and consequences of motor vehicle injuries. We will see how we can impact one, the other, or both of these. In using a systems thinking approach, we need to incorporate the following information. For purposes of this example, you need to act as if the following represents important factual information.
Etiology
In terms of etiology, motor vehicle collisions are greatly influenced by alcohol use, which has direct impacts on the risk of motor vehicle collisions but also leads to speeding, which itself strongly influences the chances of motor vehicle collision. Speeding greatly increases the likelihood of a collision as well as reducing available response times. Efforts aimed at both speeding and alcohol use have been especially effective in reducing motor vehicle injuries.
Motor vehicle collisions are also increased by texting, which greatly increases the response time when potential hazards occur. In addition, it may directly produce motor vehicle collisions by the movements or mechanical issues produced by texting, which disrupt safe driving.
Motor vehicle collisions can be reduced by road safety technology such as wider shoulders, barriers, and straighter roads. Vehicle collision prevention technology such as more visible brake lights, occupied blind spot notification systems, and out of lane notification systems may reduce the probability of collisions. Vehicle collision safety technology such as crumple zones, which absorb impacts, roll-over protections, and safety glass can reduce injuries when collisions do occur. Passenger restraint systems including safety belts and airbags can reduce the chance of injury from motor vehicle collisions.
Outcome
In terms of outcome, an emergency response system can take advantage of the “golden hour,” a period of time when emergency intervention can save many lives, which can reduce death and disability after injuries do occur. The emergency response system can be thought of as including first responders, emergency department preparedness, as well as a trauma triage system helping ensure that those injured get appropriate care as fast as possible. By eliminating the long delays in reaching care, the emergency response systems have been especially effective in reducing the rate of death and disability due to motor vehicle injuries.
A systems analysis of both etiology and outcome underlies the approaches that have been used to address motor vehicle injuries. The collaborative efforts of public health and clinical medicine have been an essential ingredient in this success.
Figure 14-2 looks at the direction of two factors, emergency response system and injury, as well as their type of impacts on death and disability. Note that both emergency response system and injury point toward death and disability because they presumably impact the frequency of occurrence of death and disability. However, emergency response has a negative sign because it hopefully reduces the frequency of death and disability. Injury itself increases the frequency of death and disability. f
FIGURE 14-2 Positive and Negative Impacts
In addition to indicating the direction of influence and whether the influence is positive or negative, systems diagrams often indicate the strength or magnitude of the impact. The strength or magnitude of the impact is indicated by the width of the arrow used. The thicker the arrow, the greater the impact. Figure 14-3 illustrates the strength or magnitude of two relationships. The thicker arrow between texting and slow response indicates a stronger impact while the thinner arrow between texting and motor vehicle collision indicates a weaker direct impact.
Texting has a strong impact on response time because those who are texting often fail to see and respond to hazards in a timely way. In addition, texting may in and of itself lead to collisions due to the movements or mechanical issues produced by texting, which disrupt safe driving, such as having one or both hands off the steering wheel while writing a text, or inadvertently swerving when reaching for the phone.
Figure 14-4 demonstrates what is called a positive feedback loop. In a positive feedback loop, one factor reinforces another to magnify the impact. Alcohol use reduces inhibitions and often leads to driving well beyond the speed limit. It also decreases response time, and together with the increased speed, greatly magnifies the effect of alcohol alone. A positive feedback loop can help us identify leverage points where extra efforts can have great benefits. Efforts to address both alcohol use and speeding, often together, have been especially effective in having a major impact on the reduction of motor vehicle collisions.
FIGURE 14-3 Strength of Response
Figure 14-5 demonstrates what is called a negative feedback loop. As shown in the figure, the occurrence of injuries leads to increased use of the emergency response system, which itself is intended to reduce the probability of death and disability. The increased use of the system may provide increased experience and increased competence for these health professions, so the increased use may actually improve outcome. A negative feedback loop can help us identify bottlenecks, which prevent later otherwise effective interventions from working. Efforts to eliminate long delays in reaching trauma care and taking advantage of the golden hour have largely removed this bottleneck. These interventions have been especially effective in reducing the deaths and disabilities due to motor vehicle injuries. g
Figure 14-6 represents a basic systems diagram putting together the impact of the positive and the negative feedback loops and indicating the direction of influence and the strength or magnitude of the impacts. This is the simplest type of full systems diagram.
FIGURE 14-4 Positive Feedback Loop
FIGURE 14-5 Negative Feedback Loop
FIGURE 14-6 Basic System Diagram
Often additional factors can be included in the systems diagram. Figure 14-7 diagrams the dampening or negative impact of a series of factors on the probability of motor vehicle collisions and also on the occurrence of injury once motor vehicle collisions occur. These factors are road safety technology, vehicle collision prevention technology, vehicle collision safety technology, and passenger restraint technology.
Figure 14-8 illustrates additional positive or accelerating/magnifying impacts of the combination of speeding and texting on motor vehicle collisions. It indicates how both speeding and texting have direct and indirect impacts that increase the probability of motor vehicle injury. Here it is assumed that texting has its major impact by slowing response time, while speeding has its greatest impact by directly increasing the chances of a collision.
Finally, Figure 14-9 attempts to put all of these components together to develop a systems diagram incorporating all the factors or influences that we indicated have an impact on the occurrence of collision or the outcome of collisions. h Figure 14-9 may be used as the basis for developing potential interventions or future research on the expected impact of interventions. i
FIGURE 14-7 Additional Negative Influences
FIGURE 14-8 Additional Positive Influences
FIGURE 14-9 Systems Diagram
HOW CAN WE APPLY SYSTEMS THINKING TO POPULATION HEALTH ISSUES?
Even when we do not have enough data to do a full systems analysis or develop a systems diagram, specific components of systems thinking can be applied to population health issues. Let us take a look at the roles that systems thinking can play in population health. In doing this, we will look at seven population health questions that systems thinking can help us answer.
• How can systems thinking help us incorporate interactions between factors to better understand the etiology of disease?
• How can systems thinking help take into account the interactions between diseases?
• How can systems thinking help us understand the impact of a disease over the life span?
• How can systems thinking help identify bottlenecks and leverage points that can be used to improve population health?
• How can systems thinking help us develop strategies for multiple simultaneous interventions?
• How can systems thinking help us look at processes as a whole to plan short-term and long-term intervention strategies?
• How can systems thinking help us predict the future frequency of diseases?
HOW CAN SYSTEMS THINKING HELP US INCORPORATE INTERACTIONS BETWEEN FACTORS TO BETTER UNDERSTAND THE ETIOLOGY OF DISEASE?
Understanding the interactions between factors, influences, or determinants has become central to population health, as illustrated by new approaches such as the social determinants of health. Let us examine a specific interaction that has received a great deal of attention in public health in recent years: the interaction between radon and cigarette smoking in causing lung cancer. 12 Radon is a naturally occurring radioactive gas. It is colorless and odorless. Radon is produced by the decay of uranium in soil, rock, and groundwater. It emits ionizing radiation during its radioactive decay. Radon is found all over the country, though there are areas of the country with substantially higher levels than other areas. Radon gets into the indoor air primarily by entering via the soil under homes and other buildings at the basement or lowest level.
Today, it is recognized based on high-quality epidemiological studies that radon causes lung cancer. Radon is the second most important cause of lung cancer after cigarettes and the most common cause of lung cancer among nonsmokers. The EPA estimates that radon accounts for over 20,000 cases of lung cancer, as compared with the over 100,000 cases attributed to cigarettes. The average indoor level in the United States is about 1.3pCi/L. The EPA has set a level of 2pCi/L as an attainable level and a level of 4pCi/L as the maximum recommended level. Approximately 15% of homes in the United States have basement radon levels above 4pCi/L.
Cigarette smoking and radon exposure are multiplicative; that is, when both are present, the hazard is multiplied. For instance, using the EPA’s figures, the relative risk of lung cancer for the average smoker is approximately nine times the risk compared to a nonsmoker. The relative risk from radon when the level is 10pCi/L compared to 2pCi/L is over 4.5. When both cigarette smoking and a level of radon exposure of 10pCi/L are present, the relative risk of lung cancer increases almost 40 times.
The recognition that radon multiplies the impacts of cigarette smoking has had a key impact on the approaches used to address these potential hazards. For smokers with exposure to these hazards, the risk can be greatly reduced by reductions in radon as well as by stopping smoking. Because radon is a measurable and controllable environmental exposure, there has been a great deal of attention and effort given to control of this hazard. Thus, the recognition of interactions that multiply or greatly increase the risk have become an important tool for setting priorities and developing approaches to risk reduction.
HOW CAN SYSTEMS THINKING HELP TAKE INTO ACCOUNT THE INTERACTIONS BETWEEN DISEASES?
The classic connection between diseases was Edward Jenner’s observation that children who develop cow pox were very unlikely to get smallpox even when exposed. This fundamental observation led to the concept and term “vaccination,” from the Latin word “vacca,” or cow. It also established that there is a relationship between diseases.
In recent years, it has been increasingly recognized that some diseases predispose to other diseases. In addition, there are patterns of risk factors or symptoms that tend to occur together. These are often called syndromes. As we have seen, the components of the metabolic syndrome frequently occur together and greatly increase the probability of coronary artery and other large blood vessel diseases. The recognition of the frequent occurrence of the metabolic syndrome has led to concerted efforts to identify individuals with the syndrome and make a multi-intervention approach to reducing the risk.
HIV provides a good example of the complex interactions that occur between diseases. A number of sexually transmitted diseases, especially those that interrupt the mucosal membranes lining the genital organs, such as syphilis and herpes genitalis, increase the risk of being infected with HIV if exposed. In addition, diseases such as gonorrhea greatly increase the level of the HIV virus that appears in semen, thus increasing the communicability of HIV.
HIV itself predisposes to a large number of infections, the most important of which from the public health perspective is tuberculosis. Finally, HIV is found in association with other conditions, including drug abuse and intimate partner violence, which greatly increases the burden of disease. These types of interactions of HIV with other diseases and conditions have been described as a syndemic. A syndemic is the occurrence together of two or more diseases that interact to magnify the occurrence and/or burden of disease. 13
Disease interactions are not always detrimental. At times, one disease may provide protection against other diseases. Early infection with bacteria and other pathogens in environments such as that which occurs on farms has been shown to be associated with reduced incidence of food and skin allergies.
Systems thinking can not only help us understand the relationship between diseases, but it can also help us understand the impact that a disease has over the life span.
HOW CAN SYSTEMS THINKING HELP US UNDERSTAND THE IMPACT OF A DISEASE OVER THE LIFE SPAN?
Varicella, or chicken pox, provides a good example of the importance of understanding the impact of a disease over the life span. Chicken pox is a highly contagious viral disease that until the 2000s affected approximately 90% of children in the United States. It was considered a mild disease among young children. Healthy individuals who develop chicken pox as adolescents or adults often have far more severe disease, often resulting in pneumonia and other life-threatening conditions. Because the disease is so contagious and generally mild among young children, parents in the United States often held “chicken pox parties.” Children were intentionally exposed to other children with chicken pox to try to ensure early infection.
Chicken pox can result in a range of complications, including life-threatening skin infections, pneumonia, and encephalitis, or infection of the brain, especially among immunosuppressed individuals and pregnant women. In addition, women who develop chicken pox in early pregnancy can pass it on to their fetus, producing a 1 to 2% probability of congenital deformities including physical deformities and learning disabilities.
Naturally acquired chicken pox provides long-lasting immunity, which prevents recurrent chicken pox, but the virus remains within the sensory nerve roots that supply areas of the skin surface called dermatomes. As immunity wanes with age, the virus can emerge as a disease called shingles in the area supplied by these nerves. An estimated 50% of individuals who live to be 85 years old, the most rapidly growing age group in the United States, will experience shingles. Shingles produces chicken pox–like sores and severe pain that can become chronic severe pain.
Thus, chicken pox is truly a disease that affects people and populations from the fetus to the elderly. When considering interventions to prevent or treat chicken pox, it is important to recognize that this is a disease that spans the life span. 14
As we have discussed, identifying bottlenecks and leverage points is a fundamental goal of systems analysis. Let us take a look at how these points can be identified and used to improve population health.
HOW CAN SYSTEMS THINKING HELP IDENTIFY BOTTLENECKS AND LEVERAGE POINTS THAT CAN BE USED TO IMPROVE POPULATION HEALTH?
Looking at the dynamics of systems helps us identify two types of points that benefit from special attention. The first of these is called a bottleneck or a constraint. A bottleneck is a point at which events are slowed down, presenting obstacles to success of an intervention. We have already identified some important bottlenecks. In the 1960s, it was recognized that after trauma, such as injuries from war or motor vehicle collisions, many victims are able to physiologically respond and temporarily tolerate blood loss and other injuries before rapidly deteriorating. This early period became known as the golden hour.
Few victims of motor vehicle injuries before the 1970s were reaching emergency care during the golden hour. To address this bottleneck, a sophisticated system of emergency response was put into place in the United States, which, as we have discussed, greatly reduced the response time and resulted in a large reduction in deaths and disabilities from motor vehicle collisions.
We identified another example of a bottleneck in the course of cigarette smoking. The vast majority of cigarette smokers start before age 18 and often many years earlier. These smokers often have a great deal of difficulty stopping smoking, even when they are intellectually committed to stopping in later years. Addiction to nicotine in cigarettes has been recognized as a key bottleneck to successful control of cigarette smoking. Recent interventions are addressing this bottleneck, including new authority for the FDA to regulate the quantity of nicotine in cigarettes. On the other hand, leverage points are points in systems in which successful interventions produce better than expected outcomes. We can see them as opportunities to make major improvements in outcomes. At leverage points, there is no bottleneck, but the conditions are right to take advantage of the interactions that exist between factors. For instance, with cigarette smoking, pregnant women who smoke are at greatly increased risk of delivering premature infants. In addition, they are highly motivated to stop smoking and often have encouragement to do so by family and friends. New efforts to put extra resources and extra efforts into smoking cessation for pregnant women are having a large payoff for their newborns and themselves. In addition to helping us identify bottlenecks and leverage points, systems thinking can help us develop a coordinated approach or strategy for combining multiple simultaneous interventions.
HOW CAN SYSTEMS THINKING HELP US DEVELOP STRATEGIES FOR MULTIPLE SIMULTANEOUS INTERVENTIONS?
As we have seen, the approach to coronary artery disease has successfully utilized multiple simultaneous interventions for several decades. Today, we are moving to a coordinated strategy of utilizing primary, secondary, and tertiary interventions. Primary interventions include control of high blood pressure, cholesterol, cigarette smoking, obesity, diabetes, and a growing list of other contributory causes of coronary artery disease. Secondary interventions designed to prevent heart damage and death, including interventions in the early hours of a myocardial infarction, have become an increasingly successful part of an overall strategy. Drug treatment and post myocardial exercise rehabilitation are now a standard part of medical care. Finally, tertiary interventions to prevent sudden death in public places have now become a population health intervention, with placement of automated defibrillators in places where people congregate, such as airports and sporting events.
New approaches to disease often combine primary, secondary, and tertiary interventions. For instance, efforts to address HIV increasingly include primary prevention through barrier protection, circumcision, precoital and intracoital treatment, and eventually vaccination. Postexposure treatments are being extensively investigated as well. Detection during the first few weeks, when transmissibility is greatest, is being investigated as an important new intervention. In addition, early and continuous drug treatment of HIV has been found not only to help the individual but also to reduce his or her infectivity. The success of these efforts among sexually active individuals remains to be seen, but for maternal to child transmission, there is already strong evidence of success. Box 14-2 discusses the success of the multiple intervention approach to prevention of maternal–child transmission of HIV.
The strategy of coordinated use of multiple simultaneous complementary interventions has become a highly successful population health strategy. For many years, interventions were studied and applied one intervention at a time, with little thought to how they interact or how they could be used in combination to produce the best results. In recent years, systems thinking and systems analysis approaches have contributed to the development of increasingly effective strategies that combine multiple interventions.
Fully developed systems thinking approaches, when feasible, can also be used to help us see entire processes to help us plan short-term as well as long-term intervention strategies.
BOX 14-2 Success of the Multiple Intervention Approach to Maternal–Child Transmission of HIV
Prevention of maternal–child transmission of HIV in the United States has been highly successful. Today, infection with HIV by the maternal–child route should be considered a failure of public health and medicine. HIV can be transmitted via the placenta during pregnancy. The higher the level of virus in the mother’s blood, the greater the probability of transmission. Thus, early testing and active treatment of pregnant women is fundamental to prevention of maternal–child transmission. In addition, there is an increased risk during vaginal delivery. Selective use of Cesarean delivery can reduce this risk. Early treatment of infants has been shown to reduce the risk still further. Finally, breastfeeding carries a small but important risk of transmission. Avoidance of breastfeeding or active maternal drug treatment during breastfeeding among women with HIV can greatly reduce this risk as well.
HOW CAN SYSTEMS THINKING HELP US LOOK AT PROCESSES AS A WHOLE TO PLAN SHORT-TERM AND LONG-TERM INTERVENTION STRATEGIES?
Efforts to see the entire processes rather than pieces of the pie have become key to planning interventions and have been incorporated into the “health in all policies” approach. The drug regulatory process has moved beyond a focus on premarket studies designed to establish efficacy and acceptable levels of safety in low-risk patients. Today, we are increasingly looking at postmarket studies examining safety in practice as well as longer term effectiveness and safety. In addition, comparative effectiveness studies of FDA-approved treatments conducted as part of comparative effectiveness research are asking questions about the effectiveness, safety, acceptance, and cost of the existing treatments.
Systems thinking approaches to food safety have developed in recent years. 15 Initially, systems thinking has focused on identifying interventions for high-risk food one type of food at a time. This process has been called the Hazard Analysis and Critical Control Points (HACCP) system. Box 14-3 takes a look at the HACCP approach and the efforts to control the hazard of ground beef.
New approaches to food safety build on the HACCP system and its big picture look at the process as a whole. In an emerging systems thinking approach, most food’s detailed location and time of production down to the level of the farm or factory are being identified on the label. This allows public health officials to trace the food back to where and when the problem occurred. Adding this approach to the HACCP provides a mechanism for quickly responding to early indications of foodborne outbreaks, regardless of the type of food involved. The combination of the HACCP system and food tracing provides the potential for a fully developed systems thinking approach to food safety.
BOX 14-3 The HACCP System and Ground Beef
Hazard Analysis and Critical Control Points (HACCP) is a systems approach that looks for key leverage or control points to manage food safety issues. It is built upon a series of prerequisite conditions designed to first ensure basic environmental and operating conditions. These prerequisite conditions might be viewed as efforts to remove the bottlenecks to an effectively functioning system. These include facilities that maintain sanitary conditions; proper equipment construction, installation, and maintenance; personal hygiene by employees; etc.
Once these basic conditions of food safety are accomplished, HACCP looks for options for interventions at multiple leverage or control points and institutes a series of safeguards at these specific points.
Meat safety issues reflect this approach. Ground beef, which often combines meat from leftover portions of multiple animals, has been identified as a high-risk product or hazard. Toxin-producing strains of E. coli are widespread in ground beef products and have been responsible for a number of fatal outbreaks of foodborne illness in the past. The threats to health have led to a more coordinated systems thinking approach using the HACCP system. Let us take a look at the HACCP process and the ground beef example.
The systems thinking approach known as HACCP attempts to understand, monitor, and quickly respond to breakdowns in the food safety system. This methodology, originally developed for the U.S. space program, is based on the principle that risks to food safety exist from the field to the fork. HACCP is increasingly being adopted for such products as seafood, meat, poultry, and fruit juices. HACCP includes the following seven steps:
• Analyze hazards. Potential hazards associated with a food and potential interventions to control those hazards are identified. The hazard could be biological, such as a microbe; chemical, such as a toxin; or physical, such as ground glass or metal fragments.
• Identify critical control points. These are points in a food’s production—from its raw state through processing and shipping to consumption by the consumer—at which the potential hazard can be controlled or eliminated. Examples are cooking, cooling, packaging, and metal detection.
• Establish preventive measures with critical limits for each control point. For a cooked food, for example, this may include setting the minimum cooking temperature and time required to ensure the elimination of any harmful microbes.
• Establish procedures to monitor the critical control points. Such procedures may include determining how and by whom cooking time and temperature should be monitored.
• Establish corrective actions to be taken when monitoring shows that a critical limit has not been met. An example is reprocessing or disposing of food if the minimum cooking temperature is not met.
• Establish procedures to verify that the system is working properly. An example is testing time- and temperature-recording devices to verify that a cooking unit is working properly.
• Establish effective recordkeeping to document the HACCP system. This would include records of hazards and their control methods, the monitoring of safety requirements, and actions taken to correct potential problems. Each of these principles must be backed by sound scientific knowledge—for example, published microbiological studies on time and temperature factors for controlling foodborne pathogens.
Key control points at which ground beef may be contaminated in the meatpacking process have been identified. Monitoring by testing now includes a random testing process on all batches of ground beef. The process uses rapid testing of a sample of the finished ground beef and holding up distribution until the results are available. Education of consumers about the danger of eating rare or raw ground beef is also a key component of this strategy. In addition, separating beef products from other food preparation, especially from food products eaten raw, is an important educational effort.
The HACCP process has already had a major impact on the incidence of disease associated with ground beef. It is not a cure-all but looking at the process as a whole has helped us come up with effective interventions.
Finally, let us take a look at the use of systems thinking to predict the future.
HOW CAN SYSTEMS THINKING HELP US PREDICT THE FUTURE FREQUENCY OF DISEASES?
We would like to be able to predict the future frequency and course of a disease. This could help us focus on the big problems of the future to find better approaches or at least get ready for what is coming. Systems thinking approaches are being used to try to make these estimates. For instance, considerable effort has gone into estimating the future prevalence of Alzheimer’s disease, including stages of severity. Researchers have then made estimates of the impact of better interventions and better ways to take care of Alzheimer’s patients. Similar efforts are underway to understand the AIDS epidemic.
Estimates of the future are always difficult and need to rest on the big assumption that only the factors we are focusing on will change in the future. Of all the goals we have in public health, and other applied sciences such as economics, politics, meteorology, as well as biological systems, the most difficult task is prediction, especially long-term prediction. Understanding the factors that influence current conditions and the feedback processes that affect the dynamics of a system, we can hope to predict short-term outcomes in the range of months or a few years. Long-term outcomes are particularly difficult to predict, as new factors come into play and old relationships and feedback loops change over time. Small current changes can make large future differences like a slight misalignment of an auto tire can have major implications after many miles of travel.
In addition, current models of systems are built on past data. Data by definition reflects the past state of affairs. Unless new data is constantly collected and incorporated into our analysis, the evidence we incorporate into our systems analysis will literally be obsolete the day we include it. For example, we often use life expectancy to reflect the health status of populations. Yet life expectancy reflects the population health status at one point or year in time in the past. When used to predict the future, it makes the big and unrealistic assumption that nothing will change either for the better or for the worse. Life expectancy may help us predict the short-term future, but the longer we look into the future, the less it helps us predict.
Thus, a systems thinking approach can help us think through and develop plans for a wide range of population health problems, but we should not count on it alone to solve our most difficult challenge, that of predicting the future.
Having reviewed the list of applications of systems thinking to public health, let us step back and try to put these approaches into three basic categories that allow us to see the contribution that systems thinking can make to public health as well as the limitations of systems thinking.
WHAT CAN SYSTEMS THINKING CONTRIBUTE TO PUBLIC HEALTH, AND WHAT ARE ITS LIMITATIONS?
The contributions of systems thinking to population health can be classified as assisting in explanation, operation, and prediction. While systems thinking helps us think through each of these issues, it also has substantial limitations, often related to inadequate data and incomplete understandings of how systems actually work.
Explanation asks: What are the factors that influence the development and outcome of a disease or other problem as part of a system? Multiple factors may influence the development and course of disease. The influences on disease and disease outcomes are complicated, and we need to consider the interactions between factors. Systems thinking brings to bear evidence from multiple sources and allows us to examine multiple causal relationships at the same time. In addressing issues of explanation, systems thinking is limited by our limited knowledge of the factors that actually cause disease as well as the interactions between factors.
Operation asks: How can an intervention be combined with other interventions to maximize the overall impact? Systems operate in dynamic and changing ways to produce disease and the outcomes of disease or other conditions. Looking at systems may allow recognition of bottlenecks that interfere with maximum outcome and leverage points that should be utilized as ways to facilitate optimal outcomes. Recognizing bottlenecks and leverage points requires an understanding of the dynamic ways that systems operate, including feedback loops producing changes over time. Once again, systems thinking is of limited use if we do not understand the way feedback loops act to influence the operation of the system. Taking advantage of these understandings still requires the adequate resources and well-developed collaborations.
Prediction asks: What changes in systems can be expected and how can they be channeled to improve future outcomes? Prediction of the future is the ultimate challenge of systems thinking and the most difficult to achieve. Linear or straight-line prediction based on current trends is commonly used but is very dangerous because it fails to take into account the multiple changes that are likely to occur over time. Prediction by necessity is based on past data, and by definition, past data is not the same as future reality. Our less than complete knowledge and understanding makes prediction very difficult. Short-term prediction measured in months or a few years is often more reliable than predictions that extend over multiple years or decades.