Public HLT

100honestywork

chapter_2.docx

Home >Science homework help >Public HLT

CHAPTER 2

Evidence-Based Public Health

LEARNING OBJECTIVES

By the end of this chapter, the student will be able to:

• explain the steps in the evidence-based public health process.

• describe a public health problem in terms of morbidity and mortality.

• describe the course of a disease in terms of incidence, prevalence, and case-fatality.

• describe how the distribution of disease may be used to generate hypotheses about the cause of a disease.

• describe the approach used in public health to identify a contributory cause of a disease or other condition and establish the efficacy of an intervention.

• describe the process of grading evidence-based recommendations.

• use an approach to identify options for intervention based on “when, who, and how.”

• explain the role that evaluation plays in establishing effectiveness as part of evidence-based public health.

Tobacco was introduced to Europe as a new world crop in the early 1600s. Despite the availability of pipe tobacco and, later, cigars, the mass production and consumption of tobacco through cigarette smoking did not begin until the development of the cigarette rolling machine by James Duke in the 1880s. This invention allowed mass production and distribution of cigarettes for the first time. Men were the first mass consumers of cigarettes. During World War I, cigarettes were widely distributed free of charge to U.S. soldiers.

Cigarette smoking first became popular among women in the 1920s—an era noted for changes in the role and attitudes of women—and at this time, advertising of cigarettes began to focus on women. The mass consumption of cigarettes by women, however, trailed that of men by at least two decades. By the 1950s, over 50% of adult males and approximately 25% of adult females were regular cigarette smokers.

The health problems of cigarette smoking were not fully recognized until decades after the habit became widespread. As late as the 1940s, R.J. Reynolds advertised that “more doctors smoke Camels than any other cigarette.”

Epidemiologists observed that lung cancer deaths were increasing in frequency in the 1930s and 1940s. The increase in cases did not appear to be due to changes in efforts to recognize the disease, the ability to recognize the disease, or the definition of the disease. Even after the increasing average life span and aging of the population was taken into account, it was evident that the rate of death from lung cancer was increasing—and more rapidly for men than women. In addition, it was noted that residents of states with higher rates of smoking had higher rates of lung cancer. In the 1950s, the number of lung cancer deaths in females also began to increase, and by the 1960s, the disease had become the most common cause of cancer-related deaths in males and was still rising among women. 1 , 2

This type of information was the basis for describing the problems of cigarette smoking and lung cancer and developing ideas or hypotheses about its etiology, or cause. Let us take a look at how the evidence-based public health approach has been used to address the problem of cigarette smoking. There are five basic questions that we need to ask that together make up what we will call the evidence-based public health approach.3

1. Problem: What is the health problem?

2. Etiology: What is/are the contributory cause(s)?

3. Recommendations: What works to reduce the health impacts?

4. Implementation: How can we get the job done?

5. Evaluation: How well does/do the intervention(s) work in practice?

FIGURE 2-1 Evidence-Based Public Health: The P.E.R.I.E. Approach

Adapted from Riegelman R. Evidence Based Public Health and Cigarette Smoking. Available at www.teachprevention.org. Accessed August 16, 2013.

These five questions provide a framework for defining, analyzing, and addressing a wide range of public health issues and can be applied to cigarette smoking for the purposes of this chapter.4 We will call this framework the P.E.R.I.E. process. This process is actually circular, as illustrated in Figure 2-1 . If the evaluation suggests that more needs to be done, the cycle can and should be repeated. Thus, it is an ongoing process.

Using cigarette smoking as an example, we will illustrate the steps needed to apply the evidence-based public health approach.

HOW CAN WE DESCRIBE A HEALTH PROBLEM?

In describing a health problem, we need to address what we will call the burden of disease, the course of disease, and the distribution of disease. The first step in addressing a health problem is to describe its burden of disease, which is the occurrence of disability and death due to a disease. In public heath, disability is often called morbidity and death is called mortality. We will want to know the current burden of disease and whether there has been a recent change in the burden of the disease.

In addition to describing the burden of disease, it is important to describe what we call the course of a disease. The course of the disease asks how often the disease occurs, how likely it is to be present currently, and what happens once it occurs. Describing the course of a disease as well as the burden of disease requires us to use measurements known as rates. Box 2-1 discusses what we mean by “rates” and how we can use them to describe the burden and course of disease.

In addition to describing the burden and the course of a disease or other health problem, we need to ask: What is the distribution of disease? Distribution of disease asks such questions as: Who gets the disease? Where are they located? When does the disease occur? Let us see how understanding the distribution of disease may help generate ideas or hypotheses about the disease’s etiology (cause).

BOX 2-1 Rates and the Description of a Health Problem

The term “rate” is often used to describe any type of measurement that has a numerator and a denominator where the numerator is a subset of the denominator—that is, the numerator includes only individuals who are also included in the denominator. In a rate, the numerator measures the number of times an event, such as the diagnosis of lung cancer, occurs. The denominator measures the number of times the event could occur. We often use the entire population in the denominator, but at times, we may only use the at-risk population. For instance, when measuring the rate of cervical cancer, we would only use the population of women in the denominator, and when measuring rates of prostate cancer, we would only use the population of men in the denominator.a

There are two basic types of rates that are key to describing a disease.5, 6 These are called incidence rates and prevalence. Incidence rates measure the chances of developing a disease over a period of time—usually one year. That is, incidence rates are the number of new cases of a disease that develop during a year divided by the number of people in the at-risk population at the beginning of the year, as in the following equation:

We often express incidence rates as the number of events per 100,000 people in the denominator. For instance, the incidence rate of lung cancer might be 100 per 100,000 per year. In evidence-based public health, comparing incidence rates is often a useful starting point when trying to establish the etiology, or cause, of a problem.

Mortality rates are a special type of incidence rate that measure the incidence of death due to a disease during a particular year. Mortality rates are often used to measure the burden of disease. When most people who develop a disease die from the disease, as is the situation with lung cancer, the mortality rate and the incidence rates are very similar. Thus, if the incidence rate of lung cancer is 100 per 100,000 per year, the mortality rate might be 95 per 100,000 per year. When mortality rates and incidence rates are similar and mortality rates are more easily or more reliably obtained, epidemiologists may substitute mortality rates for incidence rates.b

The relationship between the incidence rate and the mortality rate is important because it estimates the chances of dying from the disease once it is diagnosed. We call this the case-fatality. In our example, the chances of dying from lung cancer—the morality rate divided by the incidence rate—is 95%, which indicates that lung cancer results in a very poor prognosis once it is diagnosed.

Prevalence is the number of individuals who have a disease at a particular time divided by the number of individuals who could potentially have the disease. It can be represented by the following equation:

Thus, prevalence tells us the proportion or percentage of individuals who have the disease at a point in time.5, 6

Despite the fact that lung cancer has become the most common cancer, the prevalence will be low—perhaps one-tenth of 1% or less—because those who develop lung cancer do not generally live for a long period of time. Therefore, you will rarely see people with lung cancer. The prevalence of chronic diseases of prolonged duration, such as asthma or chronic obstructive pulmonary disease (COPD), is often relatively high, hence you will often see people with these diseases.c

Prevalence is often useful when trying to assess the total impact or burden of a health problem in a population and can help identify the need for services. For example, knowledge that there is a high prevalence of lung cancer in a certain region may indicate that there is a need for healthcare services in that area. Prevalence is also very useful in clinical medicine as the starting point for screening and diagnosis.

When using rates to describe a problem, we often use the rates of mortality and morbidity to describe the burden of disease. We use the incidence, prevalence, and case-fatality as the three key rates that together provide a description of the course of disease. Together, these three rates address the key issues that we need to know in describing the course of a health problem: How likely it is to occur, how likely it is to be present currently, and what happens once it occurs. Thus, understanding the burden of disease and the course of disease require us to understand and use rates. As we will see, rates are also key to understanding the distribution of disease.

_______________

a When talking about the term “rate,” many epidemiologists also include a unit of time, such as a day or a year, over which the number of events in the numerator is measured. This may also be called a true rate. The term “rate” as used in this text includes true rates, as well as proportions. A proportion is a fraction in which the numerator is a subset of the denominator. A time period is not required for a proportion; however, it often reflects the situation at one point in time.

b This is an example of the pragmatic approach that is often taken by epidemiologists when they are limited by the available data. The question facing epidemiologists is frequently: Is the data good enough to address the question? Thus, epidemiology can be thought of as an approximation science.

c The relationship between incidence and prevalence rates is approximately the incidence rate x average duration of the disease = the prevalence rate. Both the incidence rate and the average duration affect the prevalence of the disease. Together, the incidence, prevalence, and case-fatality rates provide a population-based summary of the course of a disease. Incidence reflects the chance of developing the disease, prevalence indicates the chances of having the disease, and case-fatality indicates the prognosis or chance of dying from the disease.

HOW CAN UNDERSTANDING THE DISTRIBUTION OF DISEASE HELP US GENERATE IDEAS OR HYPOTHESES ABOUT THE CAUSE OF DISEASE?

Public health professionals called epidemiologists investigate factors known as “person” and “place” to see if they can find patterns or associations in the frequency of a disease. We call these group associations or ecological associations. Group associations may suggest ideas or hypotheses about the cause, or etiology, of a disease.

“Person” includes demographic characteristics that describe people, such as age, gender, race, and socioeconomic factors. It also includes behaviors or exposures, such as cigarette smoking, exercise, radiation exposure, and use of medications. “Place” implies geographic location, such as a city or state, but it also includes connections between people, such as a university community or a shared Internet site. When these types of factors occur more frequently among groups with the disease than among groups without the disease, we call them risk indicators or risk markers.a Box 2-2 illustrates how person and place can be used to generate hypotheses about the cause of a disease.

In looking at the distribution of lung cancer and the potential risk factors, epidemiologists found some important relationships. In terms of person, the increases in lung cancer mortality observed in the 1930s through 1950s were far more dramatic among men than among women, though by the 1950s, the mortality rate among women had begun to increase as well. It was noted that cigarette use had increased first in men and later among women. There appeared to be a delay of several decades between the increase in cigarette smoking and the increase in lung cancer mortality among both men and women. This illustrates that “time” along with “person” and “place” is important in generating hypotheses.

In terms of place, it was found that the relationship between cigarette smoking and lung cancer mortality was present throughout the United States, but was strongest in those states where cigarette smoking was most common. Therefore, changes over time and the distribution of disease using person and place led epidemiologists to the conclusion that there was an association between groups of people who smoked more frequently and the same group’s mortality rates due to lung cancer. These relationships generated the idea that cigarettes might be a cause of lung cancer.

It is important to realize that these mortality rates are group rates. These data did not include any information about whether those who died from lung cancer were smokers. It merely indicated that groups who smoked more, such as males, also had higher mortality rates from lung cancer. The most that we can hope to achieve from these data is to generate hypotheses based on associations between groups, or group associations. When we try to establish causation or etiology, we will need to go beyond group association and focus on associations at the individual level.

BOX 2-2 Generating Hypotheses from Distributions of Person and Place

An increased frequency of disease based upon occupation has often provided the initial evidence of a group association based upon a combination of “person” and “place.” The first recognized occupational disease was found among chimney sweeps often exposed for long periods of time to large quantities of coal dust and who were found to have a high incidence of testicular cancer.

The Mad Hatter described in Alice’s Adventures in Wonderland by Lewis Carroll made infamous the 19th century recognition that exposure to mercury fumes was associated with mental changes. Mercury fumes were created when making the felt used for hats, hence the term “mad as a hatter.”

The high frequency of asbestosis among those who worked in shipyards suggested a relationship decades before the dangers of asbestos were fully recognized and addressed. A lung disease known as silicosis among those who worked in the mining industry likewise suggested a relationship that led to in-depth investigation and greater control of the risks.

More recently, a rare tumor called angiosarcoma was found to occur among those exposed over long periods to polyvinyl chloride (PVC), a plastic widely used in construction. The initial report of four cases of this unusual cancer among workers in one PVC plant was enough to strongly suggest a cause-and-effect relationship based upon place alone.

An important example of the impact that place can have on generating ideas or hypotheses about causation is the history of fluoride and cavities. In the early years of the 1900s, children in the town of Colorado Springs, Colorado, were found to have a very high incidence of brown discoloration of the teeth. It was soon recognized that this condition was limited to those who obtained their water from a common source. Ironically, those with brown teeth were also protected from cavities. This clear relationship to place was followed by over two decades of research that led to the understanding that fluoride in the water reduces the risk of cavities, while very high levels of the compound also lead to brown teeth. Examination of the levels of fluoride in other water systems eventually led to the establishment of levels of fluoride that could protect against cavities without producing brown teeth.

Such strong and clear-cut relationships are important, but relatively unusual. Often, examinations of the characteristics of person and place in populations suggests hypotheses that can be followed up among individuals to establish cause-and-effect relationships.5, 6

Finally, epidemiologists take a scientific approach to addressing public health problems. They are often skeptical of initial answers to a question and ask: Could there be another explanation for the differences or changes in the distribution of disease?

HOW DO EPIDEMIOLOGISTS INVESTIGATE WHETHER THERE IS ANOTHER EXPLANATION FOR THE DIFFERENCE OR CHANGES IN THE DISTRIBUTION OF DISEASE?

Epidemiologists ask: Are the differences or changes real or are they artifactual? There are three basic reasons that changes in rates may be artifactual rather than real:

• Differences or changes in the interest in identifying the disease

• Differences or changes in the ability to identify the disease

• Differences or changes in the definition of the disease

For some conditions, such as HIV/AIDS, these changes have all occurred. New and effective treatments have increased the interest in detecting the infection. Improved technology has increased the ability to detect HIV infections at an earlier point in time. In addition, there have been a number of modifications of the definition of AIDS based on new opportunistic infections and newly recognized complications. Therefore, with HIV/AIDS, we need to be especially attentive to the possibility that artifactual changes have occurred.

Thus in describing the distribution of a problem, epidemiologists ask: Are the differences or changes used to suggest group associations and generate hypotheses artifactual or real?

Let us see how this applies to our lung cancer example. As we have seen, lung cancer is a disease with a very poor prognosis; therefore, the burden of disease is high as measured by its high mortality rate. This was the situation in the past and to a large extent continues to be the situation.

Mortality rates have been obtained from death certificates for many years. The cause of death on death certificates is classified using a standardized coding system known as the International Classification of Diseases (ICD). No equally complete or accurate system has been available for collecting data on the incidence rates of lung cancer. However, as we learned in our discussion of rates, the incidence rates and mortality rates for lung cancer are very similar. Therefore, we can use mortality data as a substitute for incidence data when evaluating the overall burden of lung cancer in a population.

By the 1930s, epidemiologists had concluded from the study of death certificates that lung cancer deaths were rapidly increasing. This increase continued through the 1950s—with the increase in lung cancer occurring two decades or more after the increase in consumption of cigarettes. Therefore, it was not immediately obvious that the two were related. In order to hypothesize that cigarettes are a cause of lung cancer, one needed to conclude that there was a long delay and/or a need for long-term exposure to cigarettes before lung cancer developed. There was a need for more evidence linking cigarettes and lung cancer.

From the 1930s through the 1950s, a large number of studies established that lung cancer deaths were increasing among men, but not among women. That is, there was a change over time and a difference between groups. Epidemiologists, therefore, considered whether the changes or differences in rates were real, or whether they could be artificial or artifactual.

With lung cancer, the diagnosis at the time of death has been of great interest for many years. The ability to diagnose the disease has not changed substantially over the years. In addition, the use of ICD codes on death certificates has helped standardize the definition of the disease. Epidemiologists concluded that it was unlikely that changes in interest, ability, or definition explained the changes in the rates of lung cancer observed in males, thus they concluded that the changes were not artifactual, but real.b

Box 2-3 discusses age adjustment, which is one additional step that epidemiologists frequently make when looking at rates.

WHAT IS THE IMPLICATION OF A GROUP ASSOCIATION?

Group associations are established by investigations that use information on groups or a population without having information on the specific individuals within the group. These studies have been called population comparisons or ecological studies. Having established the existence of a group association, we still do not know if the individuals who smoke cigarettes are the same ones who develop lung cancer. We can think of a group association as a hypothesis that requires investigation at the individual level. The group association between cigarettes and lung cancer was the beginning of a long road to establish that cigarettes are a cause of lung cancer.

BOX 2-3 Age Adjustment

Despite the existence of a real change in the rates of lung cancer between 1930 and 1960, it was still possible that the increased mortality rates from lung cancer were due to the increasing life span that was occurring between 1930 and 1960, leading to the aging of the population and an older population on average. Perhaps older people are more likely to develop lung cancer and the aging of the population itself explains the real increase in the rates. To address this issue, epidemiologists use what is called age adjustment. To conduct age adjustment, epidemiologists look at the rates of the disease in each age group and also the age distribution, or the number of people in each age group in the population. Then they combine the rates for each age group, taking into account or adjusting for the age distribution of a population.a

Taking into account the age distribution of the population in 1930 and 1960 did have a modest impact on the changes in the mortality rates from lung cancer, but large differences remained. As a result, epidemiologists concluded that lung cancer mortality rates changed over this period, especially among men; the changes in rates were real; and the changes could not be explained simply by the aging of the population. Thus, epidemiologists had established the existence of a group association between groups that smoked more cigarettes and groups that developed lung cancer.

_______________

a Adjustment for age is often performed by combining the rates in each age group using the age distribution of what is called a standard population. The age distribution of the U.S. population in 2000 is currently used as the standard population. Adjustment is not limited to age and may at times be conducted using other characteristics that may differ among the groups, such as gender or race, which may affect the probability of developing a disease.

Not all group associations are also individual associations. Imagine the following situation: the mortality rates from drowning are higher in southern states than in northern states. The per capita consumption of ice cream is also higher in southern states than in northern states. Thus, a group association was established between ice cream consumption and drowning. In thinking about this relationship, you will soon realize that there is another difference between southern and northern states. The average temperature is higher in southern states, and higher temperatures are most likely associated with more swimming and also more ice cream consumption. Ice cream consumption is therefore related both to swimming and to drowning. We call this type of factor a confounding variable. In this situation, there is no evidence that those who drown actually consumed ice cream. That is, there is no evidence of an association at the individual level. Thus, group associations can be misleading if they suggest relationships that do not exist at the individual level.

Epidemiology research studies that look at associations at the individual level are key to establishing etiology, or cause. Etiology is the second component of the P.E.R.I.E. approach. Let us turn our attention to how to establish etiology.

ETIOLOGY: HOW DO WE ESTABLISH CONTRIBUTORY CAUSE?

Understanding the reasons for disease is fundamental to the prevention of disability and death. We call these reasons etiology or causation. In evidence-based public health, we use a very specific definition of causation—contributory cause. The evidence-based public health approach relies on epidemiological research studies to establish a contributory cause. This requires that we go beyond group association and establish three definitive requirements.7

1. The “cause” is associated with the “effect” at the individual level. That is, the potential “cause” and the potential “effect” occur more frequently in the same individual than would be expected by chance. Therefore, we need to establish that individuals with lung cancer are more frequently smokers than individuals without lung cancer.

2. The “cause” precedes the “effect” in time. That is, the potential “cause” is present at an earlier time than the potential “effect.” Therefore, we need to establish that cigarette smoking comes before the development of lung cancer.

3. Altering the “cause” alters the “effect.” That is, when the potential “cause” is reduced or eliminated, the potential “effect” is also reduced or eliminated. Therefore, we need to establish that reducing cigarette smoking reduces lung cancer rates.

Box 2-4 illustrates the logic behind using these three criteria to establish a cause-and-effect relationship, as well as what the implications of a contributory cause are.

These three definitive requirements are ideally established using three different types of studies, all of which relate potential “causes” to potential “effects” at the individual level. That is, they investigate whether individuals who smoke cigarettes are the same individuals who develop lung cancer.6 The three basic types of investigations are called case-control or retrospective studies, cohort studies or prospective studies, and randomized controlled trials or experimental studies.

BOX 2-4 Lightning, Thunder, and Contributory Cause

The requirements for establishing the type of cause-and-effect relationship known as contributory cause used in evidence-based public health can be illustrated by the cause-and-effect relationship between lightning and thunder that human beings have recognized from the earliest times of civilization.

First, lightning is generally associated with thunder; that is, the two occur together far more often than one would expect if there were no relationship. Second, with careful observation, it can be concluded that the lightning is seen a short time before the thunder is heard. That is, the potential “cause” (the lightning) precedes in time the “effect” (the thunder). Finally, when the lightning stops, so does the thunder—thus, altering the “cause” alters the “effect.”

Notice that lightning is not always associated with thunder. Heat lightning may not produce audible thunder, or the lightning may be too far away for the thunder to be heard. Lightning is not sufficient in and of itself to guarantee that our ears will subsequently always hear thunder. Conversely, in recent years, it has been found that the sound of thunder does not always require lightning. Other reasons for the rapid expansion of air, such as an explosion or volcanic eruption, can also create a sound similar or identical to thunder.

The recognition of lightning as a cause of thunder came many centuries before human beings had any understanding of electricity or today’s appreciation for the science of light and sounds. Similarly, cause-and-effect relationships established by epidemiological investigations do not always depend on understanding the science behind the relationships.

Case-control studies are most useful for establishing requirement number one, that is, the “cause” is associated with the “effect” at the individual level. Case-control studies can demonstrate that cigarettes and lung cancer occur together more frequently than would be expected by chance alone. To accomplish this, cases with the disease (lung cancer) are compared to controls without the disease to determine whether the cases and the controls previously were exposed to the potential “cause” (cigarette smoking).

When a factor such as cigarettes has been demonstrated to be associated on an individual basis with an outcome such as lung cancer, we often refer to that factor as a risk factor.c

During the 1940s and early 1950s, a number of case-control studies established that individuals who developed lung cancer were far more likely to be regular smokers compared to similar individuals who did not smoke cigarettes. These case-control studies established requirement number one—the “cause” is associated with the “effect” at the individual level. They established that cigarettes are a risk factor for lung cancer.

Cohort studies are most useful for establishing requirement number two—the “cause” precedes the “effect.” Those with the potential “cause” or risk factor (cigarette smoking) and those without the potential “cause” are followed over time to determine who develops the “effect” (lung cancer).d

Several large scale cohort studies were conducted in the late 1950s and early 1960s. One conducted by the American Cancer Society followed nearly 200,000 individuals over 3 or more years to determine the chances that smokers and nonsmokers would develop lung cancer. Those who smoked regularly at the beginning of the study had a greatly increased chance of developing lung cancer over the course of the study, thus establishing requirement number two, the “cause” precedes the “effect” in time.

Randomized controlled trials are most useful for establishing requirement number three—altering the “cause” alters the “effect.” Using a chance process known as randomization or random assignment, individuals are assigned to be exposed or not exposed to the potential “cause” (cigarette smoking). Individuals with and without the potential “cause” are then followed over time to determine who develops the “effect.” Conducting a randomized controlled trial of cigarettes and lung cancer would require investigators to randomize individuals to smoke cigarettes or not smoke cigarettes and follow them over many years. This illustrates the obstacles that can occur in seeking to definitively establish contributory cause. Once there was a strong suspicion that cigarettes might cause lung cancer, randomized controlled trials were not practical or ethical as a method for establishing cigarette smoking as a contributory cause of lung cancer. Therefore, we need to look at additional supportive or ancillary criteria that we can use to help us establish the existence of contributory cause.e

Figure 2-2 illustrates the requirements for definitively establishing contributory cause and the types of studies that may be used to satisfy each of the requirements. Notice that the requirements for establishing contributory cause are the same as the requirements for establishing efficacy. Efficacy implies that an intervention works, that is, it increases positive outcomes or benefits in the population being investigated.

WHAT CAN WE DO IF WE CANNOT DEMONSTRATE ALL THREE REQUIREMENTS TO DEFINITIVELY ESTABLISH CONTRIBUTORY CAUSE?

When we cannot definitively establish a contributory cause, we often need to look for additional supportive evidence.7 In evidence-based public health, we often utilize what have been called supportive or ancillary criteria to make scientific judgments about cause and effect. A large number of these criteria have been used and debated. However, four of them are widely used and pose little controversy. They are:

FIGURE 2-2 Fulfilling Requirements for Establishing Contributory Cause or Efficacy

• Strength of the relationship

• Dose-response relationship

• Consistency of the relationship

• Biological plausibility

Let us examine what we mean by each of these criteria.

The strength of the relationship implies that we are interested in knowing how closely related the risk factor (cigarette smoking) is to the disease (lung cancer). In other words, we want to know the probability of lung cancer among those who smoke cigarettes compared to the probability of lung cancer among those who do not smoke cigarettes. To measure the strength of the relationship, we calculate what we call the relative risk. The relative risk is the probability of developing the disease if the risk factor is present compared to the probability of developing the disease if the risk factor is not present. Therefore, the relative risk for cigarette smoking is calculated as:

The relative risk for cigarette smoking and lung cancer is approximately 10. A relative risk of 10 is very large. It tells us that the chances or probability of developing lung cancer are 10 times as great for the average smoker compared to the average nonsmoker.f

In addition to looking at the strength of the overall relationship between smoking cigarettes and lung cancer, we can ask whether smoking more cigarettes is associated with a greater chance of developing lung cancer. If it is, then we say there is a dose-response relationship. For instance, smoking one pack of cigarettes per day over many years increases the chances of developing lung cancer compared to smoking half a pack per day. Similarly, smoking two packs per day increases the chances of developing the disease compared to smoking one pack per day. These examples show that a dose-response relationship is present.g

TABLE 2-1 Supportive or Ancillary Criteria—Cigarettes and Lung Cancer

Consistency implies that studies in different geographic areas and among a wide range of groups produce similar results. A very large number of studies of cigarettes and lung cancer in many countries and among those of nearly every race and socioeconomic group have consistently demonstrated a strong individual association between cigarette smoking and lung cancer.

The final supportive criterion is biological plausibility. This term implies that we can explain the occurrence of disease based upon known and accepted biological mechanisms. We can explain the occurrence of lung cancer by the fact that cigarette smoke contains a wide range of potentially toxic chemicals that reach the locations in the body where lung cancer occurs.

Thus, the ancillary criteria add support to the argument that cigarette smoking is a contributory cause of lung cancer. Table 2-1 summarizes the use of ancillary or supportive criteria in making scientific judgments about contributory cause and illustrates these principles using the cigarette smoking and lung cancer scenario. It also cautions us to use these criteria carefully because a cause-and-effect relationship may be present even when some or all of these criteria are not fulfilled.7

We have now summarized the approach used in evidence-based public health to establish a contributory cause. We started with the development of group associations that generate hypotheses and moved on to look at the definitive requirements for establishing contributory cause. We also looked at the ancillary or supportive criteria that are often needed to make scientific judgments about contributory cause. Table 2-2 summarizes this process and applies it to cigarette smoking and lung cancer.

TABLE 2-2 Cigarettes and Lung Cancer—Establishing Cause and Effect

WHAT DOES CONTRIBUTORY CAUSE IMPLY?

Establishing a contributory cause on the basis of evidence is a complicated and often time-consuming job. In practice, our minds often too quickly jump to the conclusion that a cause-and-effect relationship exists. Our language has a large number of words that may subtly imply a cause-and-effect relationship, even in the absence of evidence. Box 2-5 illustrates how we often rapidly draw conclusions about cause and effect.

It is important to understand what the existence of a contributory cause implies and what it does not imply. Despite the convincing evidence that cigarette smoking is a contributory cause of lung cancer, some individuals never smoke and still develop lung cancer. Therefore, cigarettes are not what we call a necessary cause of lung cancer. Others smoke cigarettes all their lives and do not develop lung cancer. Thus, cigarettes are not what we call a sufficient cause of lung cancer.

The fact that not every smoker develops lung cancer implies that there must be factors that protect some individuals from lung cancer. The fact that some nonsmokers develop lung cancer implies that there must be additional contributory causes of lung cancer. Thus, the existence of a contributory cause implies that the “cause” increases the chances that the “effect” will develop. Its presence does not guarantee that the disease will develop. In addition, the absence of cigarette smoking does not guarantee that the disease will not develop.

Despite the fact that cigarettes have been established as a contributory cause of lung cancer, they are not a necessary or a sufficient cause of lung cancer. In fact, the use of the concept of necessary and sufficient cause is not considered useful in the evidence-based public health approach because so few, if any, diseases fulfill the definitions of necessary and sufficient cause. These criteria are too demanding to be used as standards of proof in public health or medicine.

BOX 2-5 Words that Imply Causation

Often when reading the newspaper or other media, you will find that conclusions about cause and effect are made based upon far less rigorous examination of the data than we have indicated are needed to definitively establish cause and effect. In fact, we often draw conclusions about cause and effect without even consciously recognizing we have done so. Our language has a large number of words that imply a cause-and-effect relationship, some of which we use rather casually.

Let us take a look at the many ways that a hypothetical newspaper article might imply the existence of a cause-and-effect relationship or a contributory cause even when the evidence is based only upon a group association or upon speculation about the possible relationships.

Over several decades, the mortality rates from breast cancer in the United States were observed to increase each year. This trend was due to and can be blamed on a variety of factors, including the increased use of estrogens and exposure to estrogens in food. The recent reduction in breast cancer resulted from and can be attributed to the declining use of estrogens for menopausal and postmenopausal women. The declining mortality rate was also produced by the increased use of screening tests for breast cancer that were responsible for early detection and treatment. These trends demonstrate that reduced use of estrogens and increased use of screening tests have contributed to and explain the reduction in breast cancer.

While these conclusions sound reasonable and may well be cause-and-effect relationships, note that they rely heavily on assertions for which there is no direct evidence provided. For instance, the following words are often used to imply a cause-and-effect relationship when evidence is not or cannot be presented to support the relationship:

• due to

• blamed on

• result from

• attributable to

• produced by

• responsible for

• contributed to

• explained by

It is important to be aware of conscious or unconscious efforts to imply cause-and-effect relationships when the data suggests only group associations and does not meet our more stringent criteria establishing cause and effect.

By 1964, the evidence that cigarette smoking was a contributory cause of lung cancer was persuasive enough for the surgeon general of the United States to produce the first surgeon general’s Report on Smoking and Health. The report concluded that cigarettes are an important cause of lung cancer. Over the following decades, the surgeon general’s reports documented the evidence that cigarette smoking causes not only lung cancer, but also other cancers—including cancer of the throat and larynx. Cigarette smoking is also a contributory cause of chronic obstructive pulmonary disease (COPD) and coronary artery disease. Smoking during pregnancy poses risks to the unborn child, and passive or secondand smoke creates increased risks to those exposed—especially children.8 Based on the surgeon general’s findings, there is clearly overwhelming evidence that cigarette smoking is a contributory cause of lung cancer and a growing list of other diseases. Thus, let us turn our attention to the third component of the P.E.R.I.E. process: recommendations.

RECOMMENDATIONS: WHAT WORKS TO REDUCE THE HEALTH IMPACT?

The evidence for cigarette smoking as a cause of lung cancer, as well as other diseases, was so strong that it cried out for action. In evidence-based public health, however, action should be grounded in recommendations that incorporate evidence. That is, evidence serves not only to establish contributory cause, but is also central to determining whether or not specific interventions work.9, 10 Evidence-based recommendations are built upon the evidence from studies of interventions. Thus, recommendations are summaries of the evidence about which interventions work to reduce the health impacts, and they indicate whether actions should be taken. Evidence-based recommendations utilize the same types of investigations we discussed for contributory cause. In fact, the requirements of contributory cause are the same as those for establishing that an intervention works or has efficacy on the particular population that was studied. Evidence-based recommendations, however, go beyond efficacy or benefits and take into account harms or safety.

In the decades since the surgeon general’s initial report, a long list of interventions has been implemented and evaluated. The term “intervention” is a very broad term in public health. Interventions range from individual counseling and prescription of pharmaceutical drugs that aid smoking cessation; to group efforts, such as peer support groups; to social interventions, such as cigarette taxes and a legal restriction on smoking in restaurants.

Recommendations for action have been part of public health and medicine for many years. Evidence-based recommendations, however, are relatively new. They have been contrasted with the traditional eminence-based recommendation, which uses the opinion of a respected authority as its foundation. Evidence-based recommendations ask about the research evidence supporting the benefits and harms of potential interventions. In evidence-based recommendations, the opinions of experts are most important when research evidence does not or cannot provide answers.

Before looking at the evidence-based recommendations on cigarette smoking made by the Centers for Disease Control and Prevention (CDC), let us look at how they are often made and can be graded. Evidence-based recommendations are based upon two types of criteria: the quality of the evidence and the magnitude of the impact. Each of these criteria is given what is called a score.9, 10 The quality of the evidence is scored based in large part upon the types of investigations and how well the investigation was conducted. Well-conducted randomized controlled trials that fully address the health problem are considered the highest quality evidence. Often, however, cohort and case-control studies are needed and are used as part of an evidence-based recommendation.

Expert opinion, though lowest on the hierarchy of evidence, is often essential to fill in the holes in the research evidence.9, 10 The quality of the evidence also determines whether the data collected during an intervention are relevant to its use in a particular population or setting. Data from young adults may not be relevant to children or the elderly. Data from severely ill patients may not be relevant to mildly ill patients. Thus, high-quality evidence needs to be based not only on the research, which can establish efficacy in one particular population, but also on the effectiveness of the intervention in the specific population in which it will be used.

In evidence-based public health, the quality of the evidence is often scored as good, fair, or poor. Good quality implies that the evidence fulfills all the criteria for quality. Poor quality evidence implies that there are fatal flaws in the evidence and recommendations cannot be made. Fair quality lies in between having no fatal flaws and fulfilling all the criteria for quality.h

In addition to looking at the quality of the evidence, it is also important to look at the magnitude of the impact of the intervention. The magnitude of the impact asks the question: How much of the disability and/or death due to the disease can be potentially removed by the intervention? In measuring the magnitude of the impact, evidence-based recommendations take into account the potential benefits of an intervention, as well as the potential harms. Therefore, we can regard the magnitude of the impact as the benefits minus the harms, or the “net benefits.”i

The magnitude of the impact, like the quality of the evidence, is scored based upon a limited number of potential categories. In one commonly used system, the magnitude of the impact is scored as substantial, moderate, small, and zero/negative.9 A substantial impact may imply that the intervention works extremely well for a small number of people, such as a drug treatment for cigarette cessation. These are the types of interventions that are often the focus of individual clinical care. A substantial impact may also imply that the intervention has a modest net benefit for any one individual, but can be applied to large numbers of people, such as in the form of media advertising or taxes on cigarettes. These are the types of interventions that are most often the focus of traditional public health and social policy.

Evidence-based recommendations combine the score for the quality of the evidence with the score for the impact of the intervention.9 Table 2-3 summarizes how these aspects can be combined to produce a classification of the strength of the recommendation, graded as A, B, C, D, and I.

It may be useful to think of these grades as indicating the following:

A = Must—A strong recommendation.

TABLE 2-3 Classification of Recommendations

Data from Agency for Healthcare Research and Quality, U.S. Preventive Services Task Force Guide to Clinical Preventive Services Vol 1, AHRQ Pub. No. 02-500.

B = Should—In general, the intervention should be used unless there are good reasons or contraindications for not doing so.

C = May—The use of judgment is often needed on an individual-by-individual basis. Individual recommendations depend on the specifics of an individual’s situation, risk-taking attitudes, and values.

D = Don’t—There is enough evidence to recommend against using the intervention.

I = Indeterminant, insufficient, or “I don’t know”—The evidence is inadequate to make a recommendation for or against the use of the intervention at the present time.

Notice that evidence-based public health and medicine rely primarily on considerations of benefits and harms. However, recently issues of financial cost have begun to be integrated into evidence-based recommendations. At this point, however, cost considerations are generally only taken into account for “close calls.” Close calls are often situations where the net benefits are small to moderate and the costs are large.

The evidence-based public health approach increasingly relies on the use of evidence-based recommendations that are graded based on the quality of the evidence and the expected impact of the intervention. The recommendations are made by a wide array of organizations, as discussed in Box 2-6. It is important to appreciate the source of the recommendations, as well as the methods used to develop them.7

Let us take a look at some examples of how interventions to prevent smoking, detect lung cancer early, or cure lung cancer have been graded. The CDC publishes “The Guide to Community Prevention Services,” commonly referred to as “The Community Guide.”10 This guide indicates that the following interventions are recommended, implying a grade of A or B:

• Clean indoor air legislation, prohibiting tobacco use in indoor public and private workplaces

• Federal, state, and local efforts to increase taxes on tobacco products as an effective public health intervention to promote tobacco use cessation and to reduce the initiation of tobacco use among youths

• The funding and implementation of long-term, high-intensity mass media campaigns using paid broadcast times and media messages developed through formative research

• Proactive telephone cessation support services (quit lines)

• Reduced or eliminated copayments for effective cessation therapies

• Reminder systems for healthcare providers (encouraging them to reinforce the importance of cigarette cessation)

• Efforts to mobilize communities to identify and reduce the commercial availability of tobacco products to youths

Additional recommendations encourage clinicians to specifically counsel patients against smoking, prescribe medications for adults, encourage support groups for smoking cessation, and treat lung cancer with the best available treatments when detected.

Of interest is the grade of D for recommending against screening for early detection of lung cancer using traditional chest X rays. The evidence strongly suggests that screening using this method may detect cancer at a slightly earlier stage, but not early enough to alter the course of the disease. Therefore, early detection does not alter the outcome of the disease. Research continues to find better screening methods to detect lung cancer in time to make a difference.

BOX 2-6 Who Develops Evidence-Based Recommendations?

Evidence-based recommendations may be developed by a range of groups, including the government, practitioner-oriented organizations, consumer-oriented organizations, organized healthcare systems, and even for-profit organizations. Organizations developing evidence-based recommendations, however, are expected to acknowledge their authorship and identify the individuals who participated in the process, as well as their potential conflicts of interest. In addition, regardless of the organization, the evidence-based recommendations should include a description of the process used to collect the data and make the recommendations.

For-profit organizations may make evidence-based recommendations. However, their obvious conflicts of interest often lead them to fund other groups to make recommendations. Thus, the funding source(s) supporting the development of evidence-based recommendations should also be acknowledged as part of the report.

One well-regarded model for the development of evidence-based recommendations is the task force model used by the United States Preventive Services Task Force of the Agency for Healthcare Research and Quality (AHRQ), as well as by the Task Force on Community Preventive Services of the Centers for Disease Control and Prevention (CDC).9, 10 The task force model aims to balance potential conflicts of interest and ensures a range of expertise by selecting a variety of experts, as well as community participants, based upon a public nomination process. Once the task force members are appointed, their recommendations are made by a vote of the task force and do not require approval by the government agency.

Thus, as a reader of evidence-based recommendations, it is important that you begin by looking at which group developed the recommendations, whether they have disclosed their membership, including potential conflicts of interest, and the groups’ procedures for developing the recommendations.

Evidence-based recommendations are not the end of the process. There may be a large number of recommendations among which we may need to choose. In addition, we need to decide the best way(s) to put the recommendations into practice. Thus, implementation is not an automatic process. Issues of ethics, culture, politics, and risk-taking attitudes can and should have major impacts on implementation. A fourth step in the evidence-based public health approach requires us to look at the options for implementation and to develop a strategy for getting the job done.

IMPLEMENTATION: HOW DO WE GET THE JOB DONE?

Strong recommendations based upon the evidence are ideally the basis of implementation. At times, however, it may not be practical or ethical to obtain the evidence needed to establish contributory cause and develop evidence-based recommendations. Naturally occurring implementation itself may be part of the process of establishing causation, as it was for cigarette smoking in the 1960s when 100,000 physicians stopped smoking and their rates of lung cancer declined rapidly, as compared to other similar professionals who did not stop smoking.

Today, there are often a large number of interventions with adequate data to consider implementation. Many of the interventions have potential harms, as well as potential benefits. The large and growing array of possible interventions means that health decisions require a systematic method for deciding which interventions to use and how to combine them in the most effective and efficient ways. One method for examining the options for implementation uses a structure we will call the “When-Who-How” approach.

“When” asks about the timing in the course of disease in which an intervention occurs. This timing allows us to categorize interventions as primary, secondary, and tertiary. Primary interventions take place before the onset of the disease. They aim to prevent the disease from occurring. Secondary interventions occur after the development of a disease or risk factor, but before symptoms appear. They are aimed at early detection of disease or reducing risk factors while the patient is asymptomatic. Tertiary interventions occur after the initial occurrence of symptoms, but before irreversible disability. They aim to prevent irreversible consequences of the disease. In the cigarette smoking and lung cancer scenario, primary interventions aim to prevent cigarette smoking. Secondary interventions aim to reverse the course of disease by smoking cessation efforts or screening to detect early disease. Tertiary interventions diagnose and treat diseases caused by smoking in order to prevent permanent disability and death.

“Who” asks: At whom should we direct the intervention? Should it be directed at individuals one at a time as part of clinical care? Alternatively, should it be directed at groups of people, such as vulnerable populations, or should it be directed at everyone in a community or population?j

Finally, we need to ask: How should we implement interventions? There are three basic types of interventions when addressing the need for behavioral change. These interventions can be classified as information (education), motivation (incentives), and obligation (requirements).k

An information or education strategy aims to change behavior through individual encounters, group interactions, or the mass media. Motivation implies use of incentives for changing or maintaining behavior. It implies more than strong or enthusiastic encouragement—it implies tangible reward. Obligation relies on laws and regulations requiring specific behaviors. Table 2-4 illustrates how options for implementation for cigarette smoking might be organized using the “When-Who-How” approach. To better understand the “who” and “how” of the options for intervention when behavior change is needed, refer to Table 2-5 , which outlines nine different options.

Deciding when, who, and how to intervene depends in large part upon the available options and the evidence that they work. It also depends in part on our attitudes toward different types of interventions. In U.S. society, we prefer to rely on informational or educational strategies. These approaches preserve freedom of choice, which we value in public, as well as private, decisions. Use of mass media informational strategies may be quite economical and efficient relative to the large number of individuals they reach though messages, but they often need to be tailored to different audiences. However, information is often ineffective in accomplishing behavioral change—at least on its own.

Strategies based upon motivation, such as taxation and other incentives, may at times be more effective than information alone, though educational strategies are still critical to justify and reinforce motivational interventions. Motivational interventions should be carefully constructed and judiciously used, or they may result in what has been called victim blaming. For example, victim blaming in the case of cigarette smoking implies that we regard the consequences of smoking as the smokers’ own fault.

The use of obligation or legally required action can be quite effective if clear-cut behavior and relatively simple enforcement, such as restrictions on indoor public smoking, are used. These types of efforts may be regarded by some as a last resort, but others may see them as a key to effective use of other strategies. Obligation inevitably removes freedom of choice and if not effectively implemented with regard for individual rights, the strategy may undermine respect for the law. Enforcement may become invasive and expensive, thus obligation requires careful consideration before use as a strategy.

Understanding the advantages and disadvantages of each type of approach is key to deciphering many of the controversies we face in deciding how to implement programs to address public health problems; however, implementation is not the end of the evidence-based public health process.

EVALUATION: HOW DO WE EVALUATE RESULTS?

Public health problems are rarely completely eliminated with one intervention—there are few magic bullets in this field. Therefore, it is important to evaluate whether an intervention or combination of interventions has been successful in reducing the problem. It is also critical to measure how much of the problem has been eliminated by the intervention(s) and what is the nature of the problem that remains.

Traditionally, evaluation has asked before and after questions. For instance, studies of cigarette smoking between the mid-1960s, when cigarettes were first declared a cause of lung cancer, and the late 1990s demonstrated that there was nearly a 50% reduction in cigarette smoking in the United States and that the rates of lung cancer were beginning to fall—at least among males. However, much of the problem still existed because the rates among adolescent males and females remained high and smoking among adults was preceded by smoking as adolescents nearly 90% of the time. Thus, an evaluation of the success of cigarette smoking interventions led to a new cycle of the process. It focused on how to address the issue of adolescent smoking and nicotine addiction among adults. Many of the interventions being used today grew out of this effort to cycle once again through the evidence-based public health process and look for a new understanding of the problem, its etiology, evidence-based recommendations, and options for implementation as illustrated in Figure 2-3 .

TABLE 2-4 Framework of Options for Implementation

In recent years, this process of evaluation has been extended to attempt to address how well specific interventions work and are accepted in practice. A new framework, called the RE-AIM framework, is increasingly being used to evaluate these factors.12 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 as well as those it has the potential to reach in practice. It is important to recognize that interventions are often applied far beyond the groups for whom they have been designed or investigated. The “AIM” factors examine the acceptance of the intervention in clinical or public health practice in the short and long term. Table 2-6 defines the meaning of each of these components and illustrates how a new intervention for cigarette cessation might be evaluated using the RE-AIM framework.

Deciding the best combination of approaches to address a public health problem remains an important part of the judgment needed for the practice of public health. In general, multiple approaches are often needed to effectively address a complex problem like cigarette smoking. Population and high-risk group approaches, often used by public health professionals, and individual approaches, often used as part of health care, should be seen as complementary. Often using both types of interventions is more effective than either approach alone. Social interventions, such as cigarette taxes and restrictions on public smoking, are also important interventions to consider.

TABLE 2-5 Examples of “Who” and “How” Related to Cigarette Smoking

FIGURE 2-3 Evidence-Based Public Health: The Complete P.E.R.I.E. Approach

Adapted from Riegelman R. Evidence Based Public Health and Cigarette Smoking. Available at www.teachprevention.org. Accessed August 16, 2013.

TABLE 2-6 Evaluation: RE-AIM Framework

RE-AIM component	Meaning	Example
*How well does the intervention work in practice?*
Reach	Asks: Who is the intervention being applied to in practice? May be groups or populations that are different than those on which it was investigated or intended for, i.e., the target population.	New prescription smoking cessation drug along with behavioral intervention approved by FDA and given evidence-based rating of A for long-standing adult smokers. Adverse events include rare depression and liver disease that is reversible with cessation of medication. In practice, being used for short-term smokers and teenagers who experience increased incidence of suicidal ideas.
Effectiveness	Asks: What is the impact in practice on the intended or target population, including beneficial outcomes as well as harm?	When used for long-term adult smokers, follow-up studies demonstrate substantial long-term quit rates similar to those observed in randomized controlled trials with no serious adverse events not identified in preapproval studies. Benefits far exceed harms when used on intended target population.
*How well is the intervention accepted in practice?*
Adoption	Asks: How well is the intervention accepted by individuals and providers of services?	The drug is being widely used for long-term adult smokers. The drug is also being widely used for teenagers.
Implementation	Asks: How should the intervention be modified to reach target population and providers of services, but not those for whom the benefits do not exceed the harms?	A “black box” warning is placed on the prescribing information, warning clinicians of the potential suicide risk when used for teenagers.
Maintenance	Asks: How can we ensure long-term continuation of use and success of intervention among individuals and providers of services?	Long-term use of smoking cessation drug is needed and is encouraged by coverage by health insurance plans.

Data from Virginia Tech. RE-AIM. Available at http://www.re-aim.org. Accessed July 23, 2013.

Today, an enormous body of evidence exists on the relationship between tobacco and health. Understanding the nature of the problems, the etiology or cause-and-effect relationships, the evidence-based recommendations, and the approaches for implementing and evaluating the options for interventions remains key to the public health approach to smoking and health.4 Figure 2-3 diagrams the full P.E.R.I.E. approach. Table 2-7 summarizes the questions to ask in the evidence-based public health approach.

The P.E.R.I.E. process summarizes the steps in evidence-based public health. It emphasizes the need to understand the nature of the problem and its underlying causes. It also helps structure the use of evidence to make recommendations and decide on which options to put into practice. Finally, the circular nature of the P.E.R.I. E. process reminds us that the job of improving health goes on, often requiring multiple efforts to understand and address the problem.11

Now that we have an understanding of the basic approach of evidence-based public health, let us turn our attention to the fundamental tools at our disposal for addressing public health problems.

KEY WORDS

• P.E.R.I.E. process

• Burden of disease

• Morbidity

• Mortality

• Course of a disease

• Rate

TABLE 2-7 Questions to Ask—Evidence-Based Public Health Approach

1. Problem—What is the health problem?

• What is the burden of a disease or other health problem?

• What is the course of a disease or other health problem?

• Does the distribution of the health problem help generate hypotheses?

2. Etiology—What are the contributory causes?

• Has an association been established at the individual level?

• Does the “cause” precede the “effect”?

• Has altering the “cause” been shown to alter the “effect”? (If not, use ancillary criteria.)

3. Recommendations—What works to reduce the health impacts?

• What is the quality of the evidence for the intervention?

• What is the impact of the intervention in terms of benefits and harms?

• What grade should be given to indicate the strength of the recommendation?

4. Implementations—How can we get the job done?

• When should the implementation occur?

• At whom should the implementation be directed?

• How should the intervention(s) be implemented?

5. Evaluation—How well does the intervention work in practice?

• How well does the intervention work in practice on the intended or target population?

• How well does the intervention work in practice as actually used?

• How well is the intervention accepted in practice?

Adapted from Riegelman R. Evidence Based Public Health and Cigarette Smoking. Available at www.teachprevention.org. Accessed August 16, 2013.

• At-risk population

• Incidence rate

• Prevalence rate

• True rate

• Etiology

• Case-fatality

• Proportion

• Distribution of disease

• Epidemiologists

• Associations

• Group associations or ecological associations

• Risk indicators (or risk markers)

• Artifactual

• Evidence

• Age adjustment

• Age distribution

• Standard population

• Population comparisons

• Ecological studies

• Confounding variable

• Contributory cause

• Case-control or retrospective studies

• Cohort or prospective studies

• Randomized controlled trials or experimental studies

• Risk factor

• Reverse causality

• Randomization or random assignment

• Ancillary or supportive criteria

• Natural experiment

• Efficacy

• Strength of the relationship

• Relative risk

• Absolute risk

• Dose-response relationship

• Odds ratio

• Consistency

• Protective factor

• Biological plausibility

• Necessary cause

• Sufficient cause

• Recommendations

• Score

• Effectiveness

• Surrogate outcomes

• Attributable risk percentage (or the percent efficacy)

• Primary, secondary, and tertiary interventions

• Victim blaming

• RE-AIM