Skip to main content
Loading Explore guides…
Environmental Epidemiology · Acedra
Home
Explore
Medicine
Environmental Epidemiology
Sign in to save
Chat
If YouTube asks you to sign in, open the video on YouTube.
Watch on YouTube
Reload player
Transcript
52 segments
Copy
Hide video
0:01
It’s a pleasure to be with you. The title of this talk is “Environmental Public Health Practice, Part 2: Epidemiology.” There are no disclosures for this presentation. Objectives include: • Review the definitions of epidemiology and environmental epidemiology. • Review a brief history of epidemiology. • Define descriptive and analytic epidemiology. • Review key epidemiologic rates and terms. • Review observational and non-observational epidemiologic study designs.
0:48
• Review the concept of bias. • Review the 3 requirements for successful epidemiologic investigations. • Discuss major strengths and weaknesses of environmental epidemiology. The word epidemiology comes from Greek roots. Epi means upon. Demos means people. Ology means study of. So, epidemiology means the study of something that happens to people or populations. This definition gives more detail. Epidemiology is “the science concerned with the study of the factors determining and
1:39
influencing the frequency and distribution of disease, injury, and other health-related events and their causes in a defined human population for the purpose of establishing programs to prevent and control their development and spread, also including the sum of knowledge gained in such a study.” Let’s first review a brief history of field of epidemiology. Hippocrates who lived from 460-370 BCE emphasized reasoning vs. superstition, setting the basis for the scientific method.
2:22
Paracelsus,1493-1541 CE, initiated the field of toxicology by studying the dose response relationship of medications. John Graunt, 1620-1674 CE, was the first to employ quantitative methods to vital statistics. Ramazzini, 1633-1714 CE, founded the discipline of occupational medicine and was a pioneer in the field of ergonomics. Sir Percival Pott, 1714-1788 CE, described an environmental cause of cancer -- scrotal cancer in chimney sweeps.
3:11
Edward Jenner, 1749-1823 CE, developed a vaccine for smallpox using the cowpox or vaccinia virus that he administered first to a dairymaid, Sarah Nelmes. John Snow, 1800-1899 CE, was an anesthesiologist by training. He studied the cholera epidemic in London and identified the cause -- contaminated water, associated with the Broad street pump circled on the map. Each mark on the map is the physical
3:46
location of a case of cholera and helped Snow deduce that the pump was essentially the geographic center of the epidemic and could likely be the culprit. His method was what is termed in epidemiology a natural experiment that is still commonly used today. A natural experiment is defined as a “Naturally occurring circumstance in which subsets of a population have different levels of exposure to a supposed causal factor in a situation resembling an actual experiment, where human
4:21
subjects would be randomly allocated to groups.” This slide lists the major contributions John Snow made to the field of epidemiology including: The powers of observation and written expression; Description of Epidemiologic methods, including mapping and data tables to describe the outbreak; The use of the “Natural Experiment” epidemiologic method; Developing recommendations based on the data; Using that data to develop a practical solution
4:55
to the problem, in this case remove the pump handle, which Snow did. This picture is the famous Broad street pump with the handle removed. Let’s continue with a few more key historical considerations. William Farr, 1807-1883 CE, developed a more sophisticated system for codifying medical conditions. Robert Koch, 1843-1910 CE, developed the famous Koch’s postulates, an extremely important tool that helps determine infectious disease etiology. The postulates include:
5:38
• The organism must be observed in every case of the disease, • The organism must be isolated and grown in pure culture, • The pure culture must, when inoculated into a susceptible animal, reproduce the disease, and • The organism must be observed in and recovered from the experimental animal. There were several situations in the 20th century that groomed the field of epidemiology and its uses including; • The Spanish Flu epidemic of 1918-1919, • The Framingham study of 1948,
6:21
intensively studying coronary heart disease, • Development of the Centers for Disease Control and Prevention, Epidemiologic Intelligence Service (EIS) in 1951, • The Surgeon General’s Report in 1964 linking Tobacco to significant health risks including cancer, and • The successful smallpox eradication program that culminated with the last smallpox case seen in 1977 in Somalia. Epidemiologic study can be divided into two stages including Descriptive
7:03
epidemiology and Analytic epidemiology. Descriptive epidemiology is often the focus of state and local health departments, and often represents the first part or stage of an epidemiologic study. Descriptive epidemiologic data includes time, place, and characteristics of those affected. Descriptive epidemiology data leads to tentative theories regarding cause and effect or even just an association. The result of descriptive epidemiology is the generation of hypotheses.
7:40
Analytic epidemiology is usually the focus of academic institutions, the second stage of epidemiologic study where hypotheses are tested. Environmental epidemiology is the study of diseases and health conditions of a population that are linked to environmental factors generally outside the control of affected individuals. This slide lists some key quantitative epidemiologic terms. Incidence and prevalence is actually a
8:21
frequency providing a raw number of cases. These frequency or raw numbers often indicate a need for epidemiologic investigations. Sometimes one case is significant, like Ebola , smallpox, or one case of Zika in a pregnant woman in North Dakota, without a history of travel or an at-risk sexual partner, would raise the specter of possible local transmission; an appropriate vector (Aedes Aegypti or Aedes Albopictus) may be in the state.
8:58
To compare various population groups, statisticians divide the Number of cases -- the incidence or prevalence, the frequency -- by the Population size. Let’s discuss some additional uses of the Incidence Frequency number to develop an Incidence Rate. Incidence rate is defined as new cases during a period of time per population at risk, multiplied by a multiplier, often 100,000. These are new disease cases, conditions,
9:37
deaths, etc. in a defined period of time in a specified population, the population at risk. Population at risk means those capable of developing the disease or condition. Incidence rate is a rate since it is related to a period of time. Incidence rates are important in studying causation, like exposures affecting health outcomes. Incidence rates provide information about “…the probability of someone in
10:17
the population (at risk) developing a disease (condition, death, etc.) during a specified period, but, conditional on not dying first from another disease.” The condition of “not dying first from another cause or disease” brings in another concept of epidemiology: confounders that are extremely important in data analysis. A confounder is a variable that influences both the dependent and independent
10:55
variables, therefore causing a misinterpretation of the data. In mathematical and statistical modelling, there are dependent and independent variables. The models investigate how the former depend on the latter. The dependent variables represent the output or outcome. The independent variables represent inputs or causes. This slide is a visual representation of how a confounder may work. The independent variable or cause is smoking. The dependent variable or outcome is
11:36
lung cancer. Smoking has been shown to be associated with lung cancer. Smokers may also drink more coffee than non-smokers. That increased coffee drinking of smokers could appear to be a cause of lung cancer. Coffee drinking in this situation would be a confounder and, if not analyzed appropriately, could lead to misinterpretation and even poor policy development. Prevalence is defined as the number of existing cases in a population (at risk) at a
12:14
designated time. There are several types of prevalence including: Point prevalence or cases at one point in time (e.g., as of today). An example of this could be the number of parents who responded positive to a survey question, “Has a health professional every told you your child has asthma?” Period prevalence would be the number of cases within a parameter of time (e.g., a particular month, year, etc.). How many people in the US had asthma diagnosed over
12:52
the last 12 months? Lifetime prevalence is the number of cases during a whole lifespan. How many people ever had asthma diagnosed over their lifetime? This is a visual illustrating incidence versus period prevalence. All four of these examples are over a 1-month period. The maroon circles represents the onset of disease 1 and blue the onset of disease 2. This onset may be identified by laboratory data,
13:33
signs and symptoms or both. The ends of the black and green arrows represent death. In Example 1, the 1-month period for disease 1 gave a period prevalence of 4 and incidence of 3. In Example 1 for disease 2, the period prevalence is 3 and incidence is 1. The prevalence is the number of individuals during that period of time that have the disease. The onset may have been from before that period, as in case 5 in example 1,
14:15
disease 1, and continued into the study period; therefore, is counted in the period prevalence. In Example 1, disease 1, the incidence is the new cases identified during the study period, cases 1, 2 and 3; therefore, the incidence is 3 not four. In Example 2, disease 1 has a period prevalence of 4 for the study period and incidence of 0 since none of these cases were new. In disease 2 of Example 2, the period prevalence is 5 and incidence is 0 since once again no new cases of
15:00
disease 2 were identified during the study period. This slide lists some basic considerations of prevalence. Prevalence measures the distribution of health-related conditions and diseases in a population at risk. Prevalence data informs descriptive epidemiology by assessing variations in occurrence of disease, and aids in the creation of hypotheses (sets the stage for analytic studies). Prevalence data generally can’t be used
15:41
to compare populations of different sizes, unless it’s used to develop a prevalence proportion which then can be used to compare populations. Point prevalence is commonly used. Point prevalence proportion = number of cases at a point in time per population at risk. Note that point prevalence is a proportion not a rate, since it uses one point in time, not a period of time. New cases over a period of time would be an incidence rate. Some comments about the interrelationship
16:28
of incidence and prevalence. As incidence increases so does the prevalence. Other impacts on prevalence include duration of the disease or condition; immigration of new cases; and the ability to treat and cure diseases or mitigate deaths etc. Prevalence may increase without an increase in incidence. For example, longer life with HIV treatment would increase the prevalence of HIV cases. Crude death rate (CDR) is the number of deaths in a given year divided by the
17:11
reference population [during the midpoint of the year]) X 100,000 or 1000. For example, there were 2,448,017 deaths in the United States in 2005, with a population of the United States as of July 1, 2005 of 296,410,404. The crude death rate would then be (2,448,017 / 296,410,404) X 100,000 = 825.9 deaths per 100,000 population. This slide gives some common conventions regarding rates. Rates can be expressed in any unit; 100,
18:10
1000, 10,000, 100,000 etc. Generally, it’s desirable to express the number as an integer vs a decimal and therefore helps determine the unit to use. General expressions of rates include: Death rate per 100,000, Cancer rates per 100,000, and Infant mortality rates per 1000. Case fatality rates (CFR) are a measure of the lethality or a disease, exposure, or condition etc. Case fatality rates are often expressed as a percent. So, CFR % =
18:55
(Number of deaths due to a disease within a specified period of time / Number of people with the disease) X 100 For example, let’s consider the CFR%
19:11
for Ebola in the Republic of Congo in 2003. There were 143 cases of Ebola that year and 128 deaths. Therefore, the CFR % in 2003 for Ebola in the Republic of Congo was = (128 / 143) X 100, or 89.5%. Back to our example of Disease 1 and Disease 2. For disease 1, over a 5-month period, there were 5 cases and 3 deaths. Therefore, the CFR% = 3 deaths divided by 5 cases X 100 giving a CFR% of 60% for disease 1 over a 5- month period. For disease 2, there were 5 cases with
20:05
1 death over the five-month study period. The CFR% = 1 death divided by 5 cases X 100 giving a CFR% of 20% for disease 2 over a 5- month period. Epidemiologic study designs can be categorized as observational and nonobservational. Observational designs include cases series, cross-sectional, ecologic, case-control and cohort studies. Non-observational designs are experimental studies. A case series is a study of
20:50
individuals who share a common disease or health condition, and their information is generally gathered over time. This kind of study is not very helpful in determining causes but is often used to develop hypotheses that result in more complex investigations. Cross sectional studies make a one-time assessment of the prevalence of a disease or other health related condition of a population where a random sample is often utilized. For example, a cross-sectional study
21:30
of childhood asthma was conducted in Passiac, New York during the 1998-1999 school year. The cross-sectional sample included all third graders in the school system and their parents who were sent self-report symptom questionnaires. Of the 976 questionnaires sent, 818 were returned. A respiratory therapist completed spirometry or lung function tests on 615 children, which represented 58% of the target population.
22:08
Approximately 50% of the self-report questionnaires indicated asthma symptoms, but only 22% of the spirometry tests were abnormal with higher rates based on the child’s race and ethnicity. Forty-seven percent of Asian children had an abnormal spirometry, with 39.2% of Black children, 28.2% of children of Dominican ethnicity, and 27.8% of White children. Self-reported asthma symptoms were not a good predictor of asthma risk. The subject of an ecologic study is a population
22:55
or group of people (e.g., state census tract, county, tribe, school, rural community etc.) versus an individual. Ecologic studies have been used to examine correlations between air pollution, morbidity and mortality. A study reported in the New England Journal of Medicine in 2017 demonstrated increased mortality in a Medicare population exposed to PM2.5 and ozone at concentrations below current national standards. Other examples of ecologic studies include the study
23:35
of water quality on stroke and coronary artery disease and agricultural pesticides on childhood cancers. A study of water hardness and cardiovascular disease from Japan in 1995 showed no correlation. The pesticide propargite was correlated with childhood leukemia in a 1994 California Cancer Registry study. A problem for almost all ecologic studies is confounders, uncontrolled factors that influence the dependent and independent variables. For example, confounders and
24:16
uncontrolled factors that influence the study of air pollution on morbidity and mortality includes the prevalence of smoking, occupation exposures, demographic factors, genetics, length of residence, etc. As with most observational studies, ecologic studies can provide background information for more carefully designed studies. Case-control studies are very common in environmental work and provide more
24:51
complete exposure information versus cross-sectional studies. For case-control studies, subjects are chosen based on their having or not having an outcome. For example, subjects for a cardiovascular study would include individuals or cases who have coronary heart disease and control individuals who don’t. Cases and controls are usually matched on variables like sex, age, occupation, etc. dependent on
25:26
the study design. Once the case and control subjects are identified, exposures to factors are determined retrospectively often through interviews with study subjects. The measure of association between exposure and outcome in case-controlled studies is the odds ratio (OR) which is usually calculated from a 2X2 table. The Odds Ratio is (A/C)/(B/D), which can also be written as AD/BC. An Odds Ratio of more than
26:08
1 suggests a positive association of the exposure to the outcome (e.g., a disease). Cohort studies are commonly used in environmental health. A cohort study identifies subjects according to their factor exposure and observes them over time to identify new cases or incidence of various outcomes (e.g., diseases, health events, etc,). Cohort studies may be prospective or retrospective. To participate in a prospective
26:46
cohort study, subjects must be free of the study outcomes when entered in the study. A retrospective cohort study used existing records of subjects at some time in the past and then determines the current status for the study outcomes. An example of a retrospective cohort study would be the mortality among shipyard workers employed at a specific naval yard during a war (e.g., World War II). Cohort studies can identify many outcomes but are limited by the exposures chosen
27:27
(e.g., usually one). The measure of an association in cohort studies is relative risk (RR), which is a ratio of the incidence rate of a disease or health outcome in a group exposed to a factor versus the incidence of that disease or health outcome to a group not exposed. The formula for relative risk is RR = (A/A+B) / (C/C+D) where A/A+B is the incidence of a disease or health outcome in the exposed group, divided by C/C+D which is the
28:12
incidence of the disease or health outcome in the nonexposed group. A relative risk greater than 1 indicates that the exposed group has a greater chance of getting the disease or health condition than the nonexposed group. If the relative risk is less than 1, the exposure provides a protective effect for those exposed. Experimental studies are intervention studies used primarily in environmental health
28:46
research. An intervention study evaluates the result of an intentional change applied to subjects generally to test a hypothesis. These intentional changes may include the application of a preventive or therapeutic intervention to a group of people. Randomized controlled trials are a commonly used experimental study design where an intervention is applied to a treatment group and not to a control group. Group
29:19
assignment is random in these trials. For several reasons, including ethical concerns, the use of experimental studies like randomized controlled studies in environmental epidemiology is challenging; therefore, observational studies are more commonly used. A challenge for epidemiologic studies is bias. Bias is defined as “an error in the conception and design of a study, or in the collection, analysis, interpretation,
29:55
reporting, publication, or review of data, leading to results or conclusions that are systematically different from the truth.” Three major types of bias that impact environmental epidemiology are recall bias, selection bias, and confounding. Recall bias is often seen in case-control studies. It’s not uncommon for a person with a disease or health outcome to remember exposures more than controls. Selection bias is a distortion due to the process used to select subjects for study
30:37
participation, which may result in different characteristics between participants and non-participants. A good example of selection bias is the healthy worker effect. Individuals with significant disease or disability are less likely to be workers; therefore, choosing workers for a study often results in less morbidity and mortality in an occupational study versus the general population. Confounding, as previously discussed,
31:11
is a factor that influences both the dependent and independent variables, therefore causing a misinterpretation of the data causing bias. Let’s review the confounder example we previously discussed of smoking, lung cancer and coffee. The independent variable or cause is smoking. The dependent variable or outcome is lung cancer. Smoking has been shown to be associated with lung cancer. Smokers may also drink more coffee than non-smokers.
31:47
Coffee in this example is a confounder. Increased coffee drinking of smokers could appear to be a cause of lung cancer due to the potential bias of smokers drinking more coffee than non-smokers. Buffler identified 3 requirements for successful environmental investigations: 1. Direct and accurate estimates of exposures. 2. Direct and accurate determination of disease status. 3. Appropriate statistical analysis of study data.
32:26
The extent to which these requirements are not met define many of the limitations of epidemiologic studies. This table lists some common strengths and weaknesses of environmental epidemiology. A major strength is engaging in real world issues like a community’s exposure to environmental toxins that can significantly influence morbidity and mortality. Another strength is a unique environmental perspective on disease and health. Epidemiologic
33:03
environmental studies also often provide enough information to inform action even when the environmental mechanism is not fully understood. Weaknesses include insufficient sample sizes, uncontrolled important variables that can lead to study bias, and invalid exposure estimates.
33:30
In summary: Epidemiologic study can be divided into descriptive and analytic categories. Environmental epidemiology is the study of diseases and health conditions of a population that are linked to environmental factors that are generally outside the control of affected individuals. Epidemiologic study designs can be categorized as observational and nonobservational. Observational studies are more commonly used in environmental epidemiology. Three major types of bias that impact
34:12
environmental epidemiology are recall bias, selection bias, and confounding. A major strength of environmental epidemiology is engaging in real world issues.
Here to help you learn
Sign in to keep a private conversation with your course tutor.
Ask me anything about the material…
Sign in to chat