Vous êtes sur la page 1sur 5

1 2

We employ DESCRIPTIVE EPIDEMIOLOGY to describe the health-related conditions and behaviors in populations. To do this, we may conduct SURVEILLANCE on a sample of the population Descriptive epidemiology provides information on the distributions of health or disease in terms of PERSON, PLACE, and TIME (PPT) PPT information helps us FORMULATE HYPOTHESES about what might explain these distributions Hypotheses can be TESTED using various ANALYTCIAL EPIDEMIOLOGIC STUDY DESIGNS No matter which study design is used, a 2x2 TABLE can tally the information on exposure and disease for individual study subjects Calculations based on numbers in the 2x2 table yield RATES OF DISEASE in exposed versus unexposed groups, and a comparison of these rates yields a RELATIVE RISK. The relative risk measures the strength of the ASSOCIATION between the exposure and the outcome. Is the association causal? In other words, if an association is found, what does it mean? It is a causal association or is there some other explanation for observing the association? Is the association causal? In other words, if an association is found, what does it mean? It is a causal association or is there some other explanation for observing the association? This Lecture will explain how to think about why an exposure and an outcome turned up together. This is called interpretation of a study, or discussion of what the study results mean. In other words, how to we explain the observed association.

Ask:

What do we mean when we say that two things are tied, linked, related, or associated with each other? What does it mean to say that there is an association between two things? Tell students that one definition of association is that things that are associated are linked in some way that makes them turn up together.

In epidemiology, the 2x2 table is an important tool for calculating and demonstrating how things turn up together. In brief, the relative risk measure shows how strongly the exposure and the outcome turn up together. For example, a relative risk of 10 means that the outcome turns up 10 times more frequently among people with the exposure compared to people without the exposure. This is a strong association. A relative risk of only 1.5 is a much weaker association; the outcome does not turn up as often among the exposed versus the unexposed. The headlines that follow were found on the Internet. These headlines are from reports of investigations of possible associations between exposures and outcomes. These associations are based on single studies and are not necessarily causal or even true. Explain that this slide shows a cartoon about the headlines. The newscaster in the cartoon is spinning three wheels to decide which news he will report. What is the point of the cartoon? (Given the frequency with which we hear about associations on the news, it is hard to know which of these associations should be believed / taken seriously. We need to know more about the scientific basis for the reported associations; we need to know what explanations might explain associations) While The results of an epidemiological study may reflect the true effect of an exposure (s) on the development of the outcome under investigation, it should always be considered that the findings may in fact be due to an alternative explanation

Ask why these exposures and outcomes turned up together. In brief, the five reasons an exposure and an outcome might turn up together are: 1. Causal - the exposure is the underlying reason (or one of the underlying reasons) that the outcome (e.g., a disease) occurred 2. Chance - occurrence of the outcome is a coincidence unrelated to the exposure

3. Bias - a mistake in the study is the reason for the observed association 4. Confounding - a third factor (that is a cause) is the reason for the observed association

8 9 10 11

12 13

14

The first possible explanation is that they turned up together because the exposure causes the outcome. Review the definition of Cause. Example: a virus that causes a cold several days after exposure The second possible explanation is that an exposure and an outcome turned up together by chance. Review the definitions of chance with the class. Ask students to describe some activities in life that involve chance. (flipping a coin to see which football team gets to decide whether to kick off or receive; participating in a lottery; playing a slot machine) Explain to students that in epidemiology, chance plays a role in a similar way as it does in reporting results of an election poll. To avoid CHANCE as a possible explanation, pollsters try to poll a large enough sample so that a chance finding is unlikely. (It stands to reason that the more people you poll, the more likely this represents the real distributions of opinions of everyone (case in point is that if you poll everyone you will get the right result) Remind students of the phrase, leave nothing to chance. In epidemiology, CHANCE might explain an observed association if the number of people/health events in the study is so low that a result pops up accidentally, and it DOES NOT represent what is going on in the whole population (it does not mean anything). (See the next section on BIAS for discussion of other aspects of study conduct that can lead to an erroneous result / a result that does not mean anything.) The third possible explanation is that an exposure and an outcome turned up together because of bias. Review the definition of bias with the class. Teacher Note: Bias means systematic error, and in epidemiology it means error in how the study is set up or conducted. One source of bias is in selection of study participants; errors may occur in selecting a sample of study subjects that is representative of the whole population of potential subjects. Referring back to the example of an election poll (Slide #18), another task of pollsters (besides getting a large enough sample) is to sample carefully to obtain an accurate cross-section of the population. If this fails, the result of the poll will not be accurate. Similarly, in an epidemiology study, selection bias will lead to a mistaken result. Any association that is found will not really exist, it was only found because of a bias (systematic error) in the study. Another source of error/bias in an epidemiology study is measurement error. Unless the exposures and outcomes in a study are measured accurately, any associations that are found do not necessarily exist, but rather, were found because of measurement error. The fourth possible explanation is that an exposure and an outcome turned up together because of confounding. sometimes we are fooled because we observed an association between a non-causal factor and an outcome, in which that non-causal factor happens to be related to the real causal factor. The causal factor, acting as a confounder, can fool us into thinking that the non-causal factor is actually the cause Reinforce the idea that association is not necessarily causation. This is one of the central points in understanding epidemiology. Testing a hypothesis and finding that an exposure and a disease turn up together is an important step. However, interpreting what this association means is challenging because five explanations are possible for why the association has been found We are exploring four possible explanations for why a exposure and a health outcome turn up together. We will examine the third possible explanation; that they turned up together because of bias. Review the definition of bias. Bias does not mean that the investigator is prejudiced . Ask students: What does the word bias mean to you? What comes to mind when you think of bias? A line diagonal to the grain of a fabric Highly personal and unreasoned distortion of judgment Systematic error introduced encouraging one outcome over others

15

16

17 18

19

20

Several types of bias exist in research. Sackett et al have listed 60++ types of bias which exist in research. These biases can be classified either into SELECTION or INFORMATION BIAS. Bias is a result of an error in the design or conduct of a study. Efforts should therefore be made to reduce or eliminate bias or at the very least, to recognize it and take it into account when interpreting the findings of a study SELECTION BIAS Results from procedures used to select subjects into a study that lead to a result different from what would have been obtained from the entire population targeted for study Most likely to occur in case-control or retrospective cohort because exposure and outcome have occurred at time of study selection A. Response Bias - those who agree to be in a study may be in some way different from those who refuse to participate. Since in many studies no information is obtained from the nonresponders, nonresponse may introduce a serious bias that may difficult to assess. Thus they should be kept to a minimum or should be characterized as much as possible by using whatever information is available to determine ways in which they differ from responders and to gauge the likely impact of their nonresponse on the results of the study. B. Exclusion bias results from the investigators applying different eligibility criteria to the cases and to the controls in regard to which clinical conditions in the past would permit eligibility in the study and which would serve as the basis for exclusion. WHAT ARE THE SOLUTIONS? Little or nothing can be done to fix this bias once it has occurred. You need to avoid it when you design and conduct the study by, for example, using the same criteria for selecting cases and controls, obtaining all relevant subject records, obtaining high participation rates, and taking in account diagnostic and referral patterns of disease. Clear definition of study population Explicit case and control definitions Cases and controls from same population Selection independent of exposure Selection of exposed and non-exposed without knowing disease status EXAMPLE: Retrospective Cohort Study Outcome: COPD Exposure: Employment in tire manufacturing Exposed: Plant assembly line workers Non-exposed: Plant administrative personnel Bias: The exposed were contacted (selected) at a local pub while watching Monday night football; the non-exposed were identified through review of plant personnel files. Exposed persons may have been more likely to be smokers (related to COPD) EXAMPLE: Case Control Study Outcome: Hemorrhagic stroke Exposure: Appetite suppressant products that contain Phenylpropanolamine (PPA) Cases: Persons who experienced a stroke Controls: Persons in the community without stroke Bias: Control subjects were recruited by random-digit dialing from 9:00 AM to 5:00 PM. This resulted in over- representation of unemployed persons who may not represent the study base in terms of use of appetite suppressant products. INFORMATION BIAS also known as MEASUREMENT/OBSERVATIONAL BIAS An error that arises from systematic differences in the way information on exposure or disease is obtained from the study groups Results in participants who are incorrectly classified as either exposed or unexposed or as diseased or not diseased

21 22

23 24

Occurs after the subjects have entered the study Several types of observation bias: Bias may be introduced in the way that information is abstracted from medical, employment, or other records Bias from the manner in which interviewers ask questions y Interviewer s knowledge of subjects disease status may result in differential probing of exposure history y Similarly, interviewer s knowledge of subjects exposure history may result in differential probing and recording of the outcome under examination Investigator asks cases and controls differently e.g: soft cheese and listeriosis Investigator may probe listeriosis cases about hypothesis) about exposure

consumption of soft cheese (knows

25 26

Study group participants systematically differ in the way data on exposure or outcome are recalled Particularly problematic in case-control studies Individuals who have experienced a disease or adverse health outcome may tend to think about possible causes of the outcome. This can lead to differential recall Case-Control Study EXAMPLE: Outcome: Cleft palate Exposure: Systemic infection during pregnancy Cases: Mothers giving birth to children with cleft palate Controls: Mothers giving birth to children free of cleft palate Bias: Mothers who have given birth to a child with cleft palate may recall more thoroughly colds and other infections experienced during pregnancy Cases remember exposure differently than controls e.g. risk of malformation Mothers of children with malformations remember past exposures better than mothers with healthy children

27

Individuals with severe disease tends to have complete records therefore more complete information about exposures and greater association found This results in underreporting. Wish Bias occurs in subjects who developed a disease and who in attempting to answer the question Why me? seek to show, often unintentionally, that the disease is not their fault. Thus, they may deny certain exposures related to their lifestyle (such as smoking or drinking) Information bias: misclassification Measurement error leads to assigning wrong exposure or outcome category

28

29 30 the fourth possible explanation; that they turned up together because of confounding 31 To bring to ruin (archaic) Consume, waste To put to shame

32

Damn To throw into confusion To fail to discern differences between: mix up sometimes we are fooled because we observed an association between a non-causal factor and an outcome, in which that non-causal factor happens to be related to the real causal factor. The causal factor, acting as a confounder, can fool us into thinking that the non-causal factor is actually the cause. Note that a confounder has two characteristics: 1) it is a cause of the outcome; and 2) it is associated with, but not caused by, the exposure of interest. Scenario: We are studying lung cancer. Our exposure of interest is match-carrying (in one s pocket). Analysis of the data reveals an association between people who have lung cancer and match-carrying. But match-carrying is not a cause of lung cancer. A real cause is tobacco smoking. If the researchers do not ask about smoking, but only about match-carrying, the association they find is real but needs to be interpreted correctly (it is not a causal relationship, but rather is observed because of confounding by the real cause, smoking. In this example, smoking is a potential confounder that should have been measured.

33

Vous aimerez peut-être aussi