Welcome to Stat 507 - Epidemiologic Research Methods on Penn State's World Campus!
When you have completed studying the lesson notes and the assignments this week, you should be able to do the following:
Let's begin our study of epidemiology with some definitions.
The study of the distribution of disease and determinants of health-related states or events in specified human populations and the application of this study to the control of human health problems. (JM Last. Dictionary of Epidemiology. 2nd edition)
The science of making predictions about individual patients by counting clinical events in similar patients, using strong scientific methods for studies of groups of patients to ensure that the predictions are accurate. (Fletcher, Fletcher, Wagner. Clinical Epidemiology. 1996)
What is the difference in these two views of epidemiology? In the clinical setting, epidemiologic methods are used to make a prediction about a health outcome for an individual based on scientific studies of groups of similar patients. Clinical epidemiology is integral to evidence-based medicine. Epidemiology itself is the study of disease in a population.with goals of determining the frequency and distribution of the disease as well as risk factors for the disease. Although epidemiology is defined with regard to human populations, epidemiologic principles can be extended to study other problems, such as colony collapse disorder in honeybees or improving herd health for a dairy.
Following links in the table below, explore selected events in the history of epidemiology and population health.
Selected History of Epidemiology and Population Health
When designing epidemiologic studies, choices must be made about the role of the investigator, the purpose of the study, the hypothesis regarding exposure and the unit of analysis. Here are some examples:
Role of investigator:
Purpose of the study:
Hypothesized Effect of Exposure
Unit of Analysis
Data for a typical epidemiologic study may be summarized in a table comparing numbers of cases (those with the disease or condition) to non-cases in terms of their exposure to a risk factor or beneficial agent. (Fig. 1)
Figure 1: 2 × 2 Epidemiologic Table
A traditional model of infectious disease causation, known as the Epidemiologic Triad is depicted in Figure 2. The triad consists of an external agent, a host and an environment in which host and agent are brought together, causing the disease to occur in the host. A vector, an organism which transmits infection by conveying the pathogen from one host to another without causing disease itself, may be part of the infectious process.
A classic example of a vector is the Anopheles mosquito. As the mosquito ingests blood from an infected host, it picks up the parasite plasmodium. The plasmodium are harmless to the mosquito. However, after being stored in the salivary glands and then injected into the next human upon which the mosquito feeds, the plasmodium can cause malaria in the infected human. Thus, the Anopheles mosquito serves as a vector for malaria. Another familiar example of a vector are ticks of the genus Ixodes which can be vectors for Lyme disease.
In the traditional epidemiologic triad model, transmission occurs when the agent leaves its reservoir or host through a portal of exit, is conveyed by a mode of transmission to enter through an appropriate portal of entry to infect a susceptible host. Transmission may be direct (direct contact host-to-host, droplet spread from one host to another) or indirect (the transfer of an infectious agent from a reservoir to a susceptible host by suspended air particles, inanimate objects (vehicles or fomites), or animate intermediaries (vectors)).
Figure 2: Epidemiologic Triad of Disease Causation (Historical)
Further examples here: http://www.cdc.gov/osels/scientific_edu/ss1978/lesson1/Section10.html
Can the epidemiologic triad can be applied to a disease that not infectious? Consider a smoking-related disease (Figure 3). If smoking (or more specifically, a carcinogen in the smoke of the cigarette) causes the disease, those who manufacture, sell and distribute cigarettes are vectors, bringing the disease-causing agent to the susceptible host. Diagramming the epidemiologic triad also indicates potential interventions to reduce disease in the population. In this example, clean indoor air legislation, advertising potential harm from smoking or establishing workplace smoking cessation programs could change the environment and reduce the exposure of host to agent. Conversely, increased advertising from cigarette manufacturers or increased numbers of vendors would increase exposure of host to agent.
Thus, the traditional model of disease transmission can be useful to identify areas of potential intervention to reduce disease prevalence, whether infectious or non-infectious.
The objectives of epidemiology include the following:
Causality in Epidemiology
One objective of epidemiology is to identify the cause of a disease, with a desire to prevent or modify the severity of the condition. Consider Table 1. Would you agree that this table accurately portrays the true causes of death in the U.S. population? Why or why not?
Table 1: Estimated numbers by 'Cause' of Death
(From: McGinnis JM, Foege, WH. 1993 JAMA, 270(18): 2207-2212.)
As you may have noticed, the causes of death in Table 1 are all related to modifiable factors. The percentages do not total 100, but if these results are accurate, a large percentage of deaths can be postponed. The opportunity to prevent or ameliorate disease is an exciting component of epidemiologic study.
Epidemiologists follow pre-determined procedures in deciding whether to attribute a particular factor as a cause of a disease or condition. In the late 19th century, a German microbiologist, Robert Koch, devised a scheme for deciding whether or not a paricular microbe caused a disease.
Infectious Disease Model - Koch's Postulates: One organism leads to one disease. (one-to-one)
Do you see any problem with applying Koch's postulates to determine the cause of all diseases? Consider asthma or lung cancer: can one micro-organism be isolated as causing the development of these conditions?
Modern Epidemiology: Modern epidemiology accomodates multiple exposures contributing to increasing risk for one disease (many-to-one) and situations where one risk factor contributes to multiple diseases (one-to-many). Individual risk factors can be characterized as one of the following:
Figure 4 below is an example of a multifactorial model for disease.
Figure 4: Model for Analysis of Population Health and Health Disparities
From Warnecke RB, et al.. Approaching health disparities from a population perspective: the National Institutes of Health Centers for Population Health and Health Disparities. Am J Public Health 2008; 98: 1608– 1615
Considerations When Assessing Possible Causal Role of a Risk Factor for Specific Disease
Obviously, there are many factors to assess when considering whether a potential risk factor causes a disease or condition:
An epidemiologic hypothesis is a testable statement of a putative relationship between an exposure and disease. The hypothesis should be:
Look at the objectives in the following abstracts. Do these objectives meet the criteria for sound epidemiologic hypotheses?
Abstract 1: Robert J. Valuck and J. Mark Ruscin. A case-control study on adverse effects: H2 blocker or proton pump inhibitor use and risk of vitamin B12 deficiency in older adults. Journal of Clinical Epidemiology 57 (2004) 422–428.
Objective: Acid-suppressant drugs are commonly prescribed for elderly patients, a population in which vitamin B12 deficiency is a common disorder. The purpose of this study was to examine the possible association between use of prescription histamine H-2 receptor antagonists (H2RA) or proton pump inhibitors (PPI) and vitamin B12 deficiency in older adults.
Study Design and Setting: This was a case–control study in a University-based geriatric primary care setting. Among patients aged 65 years or older with documented serum vitamin B12 studies between 1990 and 1997, 53 vitamin B12-deficient cases were compared with 212 controls for past or current use of prescription H2RA/PPI according to information in subjects’ medical records.
Results: Controlling for age, gender, multivitamin use, and Helicobacter pylori infection, chronic (12 months) current use of H2RA/PPI was associated with a significantly increased risk of vitamin B12 deficiency (OR 4.45; 95% CI 1.47–13.34). No association was found between past or short-term current use of H2RA/PPI and vitamin B12 deficiency.
Conclusion: These findings support an association between chronic use of H2RA/PPI by older adults and development of vitaminB12 deficiency. Additional studies are needed to confirm these findings.
Abstract 2: Risk factors of gastroesophageal reflux disease methodology and first epidemiological results of the ProGERD study. M. Kulig, M. Nocon, M. Vieth, A. Leodolter, D. Jaspersen, J. Labenz, W. Meyer-Sabellek, M. Stolte, T. Lind, P. Malfertheiner and S. N. Willich J Clin Epi. June 2004, Volume 57, Issue 6, Pages 580-589
Objective: We describe the design and report the first results of the Progression of Gastroesophageal Reflux Disease (ProGERD) study, to our knowledge the largest prospective study of GERD patients.
Study Design and Setting: Patients were recruited at 1,253 centers in German, Austria, and Switzerland. Following an assessment of medical history, all patients were endoscoped and received esomeprazole for 2 to 8 weeks before entering the 5-year observational phase.
Results: A total of 6,215 patients (53% male, age 54 14) were included. Of these patients, 46% reported at least daily symptoms, 15% were unable to work at least once during the prior year, and 71% had visited a physician due to reflux symptoms. Barrett’s esophagus (BE) was found in 11% of our GERD patients. In polychotomous regression analysis, the main factors related to the occurrence of the three GERD subgroups (nonerosive, erosive disease, and BE) were age, gender, duration of GERD, body mass index (BMI), smoking, and previous PPI use. Factors associated with longer disease duration were increasing age, male gender, BMI, increasing symptom severity, presence of erosive GERD or BE, positive family history, and smoking.
2004 Conclusion: The findings indicate that GERD is a great burden for patients, and has significant socioeconomic implications. The long term follow-up period with further endoscopic and histologic evaluations, will help further our understanding of the natural course of the disease.
Labenz J, Nocon M, Lind T, Leodolter A, Jaspersen D, Meyer-Sabellek W, Stolte M, Vieth M, Willich SN, Malfertheiner P. "Prospective Follow-Up Data from the ProGERD Study Suggest that GERD Is Not a Categorial Disease." Am J Gastroenterol. 2006 Nov;101(11):2457-62. Epub 2006 Oct 4.
Results: "After 2 yr, 25% of patients who had NERD at baseline progressed to LA A/B and 0.6% to LA C/D; 1.6% of patients who had LA A/B progressed to LA C/D and 61% regressed to NERD; 42% of patients who had LA C/D regressed to LA A/B and 50% regressed to NERD (all figures exclude patients with confirmed BE at baseline). At 2 yr, 22% of patients had been off medication for at least 3 months. Patients with ERD-LA grade C/D were at greatest risk of developing BE: 5.8% compared with 1.4% for ERD-LA grade A/B and 0.5% for NERD."
2006 Conclusion: GERD does not seem to be a categorial disease. Progression and regression between grades was observed in this large cohort of patients under routine clinical care.
Both studies met their objectives yielding a suggestion of association between use of prescription histamine H-2 receptor antagonists (H2RA) or proton pump inhibitors (PPI) and vitamin B12 deficiency in older adults from the first study and description of GERD from the second that showed both progression and regression between grades as common occurrences.
The design of a study contributes to the strength of its findings. Below are types of studies, in order of increased strength for testing the relevant hypothesis.We will study some of these designs further later in this course.
Notice the difference is the possibility of a quasi-experimental study when considering whether or not an intervention is effective.
Target Population- Population to which inferences from the study are to be made.
Study Population- Population from which study subjects (cases and non-cases are selected).
Sample: Subjects (cases and noncases that provide data to the study).
A target population may be defined by geography, demography, health status, or some other factor. The animation below shows these terms applied to a study in Pennsylvania. Click on the map below:
Target Population - Population to which inferences from the study are to be made.
Study Population - Population from which study subjects (cases and non-cases) are selected; in the animation, only cases were shown
Sample - Subjects drawn from the study population (cases and noncases that provide data to the study).
In the animation, the sampled cases all came from eastern and central PA. Can we draw inference to the entire 12 million population of PA from these results? We should consider whether the sample was a result of random selection among all 15 PA cases or if some bias entered the sample selection process. Obviously the method for selecting the sample can greatly influence the study results.
Now that you have completed reading the Lesson 1 Notes, go back to ANGEL to complete the Week 1 assignments. (One assignment is to apply the terminology of the Epidemiologic Triad to an infectious disease, novel H1N1.)