Epidemiology Lecture Notes 2023-2024 PDF

EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO Lecture 10: DESCRIPTIVE EPIDEMIOLOGY these clues can be turned into testable INTRODUCTION hypotheses. Every novice new...

EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO Lecture 10: DESCRIPTIVE EPIDEMIOLOGY these clues can be turned into testable INTRODUCTION hypotheses. Every novice newspaper reporter is taught that a story is complete if it describes the situation's COVERAGE OF DESIRABLE EPIDEMIOLOGY what, who, where, when, and why/how. TIME In the same way, epidemiologists strive for similar The occurrence of disease changes over time: comprehensiveness in characterizing an ○ Regular epidemiologic event, whether it be a pandemic of Health officials can anticipate and influenza or a local increase in all-terrain vehicle implement control and crashes. preventive measures Epidemiologists tend to use synonyms for the E.g., Influenza (Winter); West Nile 5W’s listed: virus infection ○ What- case definition (what exactly are (August–September) we talking about) ○ Any time (unpredictable) ○ Who- person Investigators can carry out ○ Where- place studies to identify the causes and ○ When- time methods of transmission and ○ Why- causes/risk factors/modes of develop targeted strategies to transmission control or prevent further Descriptive epidemiology covers time, place, and occurrences person. E.g., Hepatitis B; Salmonellosis ○ Seasonal occurrence ADVANTAGES OF COMPILING AND ANALYZING DATA Health officials can anticipate BY TIME, PLACE, AND PERSON their occurrence and implement The epidemiologist becomes very familiar with control and prevention measures the data. He or she can see what the data can or (e.g., vaccination campaigns for cannot reveal based on the variables available, its influenza, mosquito spraying, limitations (for example, the number of records etc.) with missing information for each important ○ Sporadic occurrence variable), and its eccentricities (for example, all Investigators can conduct studies cases range in age from 2 months to 6 years, plus to identify the causes and modes one 17-year old.). of spread, and then develop The epidemiologist learns the extent and pattern appropriately targeted actions to of the public health problem being investigated control or prevent further — which months, which neighborhoods, and occurrence of the disease. which groups of people have the most and least Examples: cases. ○ Diseases that occur during the same The epidemiologist creates a detailed season each year: Influenza (winter) and description of the health of a population that West Nile virus infection (August– can be easily communicated with tables, graphs, September) and maps. ○ Diseases that occur at any time: Hepatitis The epidemiologist can identify areas or groups B and Salmonellosis within the population that have high rates of It is critical for monitoring disease occurrence in disease. This information in turn provides the community and for assessing whether the important clues to the causes of the disease, and public health interventions made a difference. LSSM | BIOMED 121 1 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO Usually displayed with a two-dimensional graph: ○ Time scale may be as broad as years or ○ Vertical or y-axis- number or rate of cases decades, or as brief as days or hours. ○ Horizontal or x-axis- periods such as Examples: years, months, or days Chronic diseases- long-term Foodborne outbreaks- days/hours SECULAR (LONG-TERM TRENDS) vs SEASONALITY SECULAR (LONG-TERM) TRENDS Graphing the annual case or rate of a disease over period of years Advantages: ○ Assess the prevailing direction of disease The number or rate of cases is plotted over time. occurrence (increasing, decreasing, or Graphs of disease occurrence over time are essentially flat) usually plotted as line graphs (Figure 1.4) or ○ Help them evaluate programs or make histograms (Figure 1.5). policy decisions ○ Infer what caused an increase or decrease in the occurrence of a disease (particularly if the graph indicates when related events took place) ○ Use past trends as a predictor of future incidence of disease SEASONALITY Graphing by week or month over a year or more Seasonal patterns may suggest hypotheses about how the infection is transmitted, what behavioral factors increase risk, and other possible contributors to the disease or condition. The graph shows the timing of events related to disease trends being displayed. ○ Examples: A graph may indicate the period of exposure- this leads to insights into what may have caused the illness A graph may indicate the date control measures were implemented- show what impact, if any, the measures may have had on disease occurrence. ○ Time is plotted along the x-axis LSSM | BIOMED 121 2 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO Figure 1.6 shows the seasonal patterns of rubella, Table 1.3 displays SARS data by source of report, and influenza, and rotavirus. All three diseases display reflects where a person with possible SARS is likely to be consistent seasonal distributions, but each disease peaks quarantined and treated.33 In contrast, Table 1.4 displays in different months – rubella in March to June, influenza the same data by where the possible SARS patients had in November to March, and rotavirus in February to April. traveled, and reflects where transmission may have occurred. EPIDEMIC CURVE Shows the time course of a disease outbreak or epidemic ○ Y-axis- shows the number of cases ○ X-axis- shows the time as either the date of symptom onset or the date of diagnosis Depending on the incubation period and route of transmission: Broad as weeks Narrows as minutes Data are displayed as histograms (similar to bar charts but have no gaps between adjacent columns) The shape and other features of an epidemic curve can suggest hypotheses about the time and source of exposure, the mode of transmission, A map provides a more striking visual display of place and the causative agent. data. On a map, different numbers or rates of disease can be depicted using different shadings, colors, or line PLACE patterns, as in Figure 1.11. Provides insight into the geographic extent of the problem and its geographic variation Refers not only to the place of residence but to any geographic location relevant to disease occurrence ○ place of diagnosis or report, ○ Birthplace, ○ site of employment, ○ school district, ○ hospital unit, ○ recent travel destinations Spot maps generally are used for clusters or The unit may be as large as a continent or country outbreaks with a limited number of cases. A dot or as small as a street address, hospital wing, or or X is placed on the location that is most relevant operating room to the disease of interest, usually where each Sometimes refers to a place category such as victim lived or worked, just as John Snow did in urban or rural, domestic or foreign, and his spot map of the Golden Square area of institutional or noninstitutional. London (Figure 1.1). If known, sites that are relevant, such as probable locations of exposure LSSM | BIOMED 121 3 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO (water pumps in Figure 1.1), are usually noted on the map. Advantages of analyzing data by place: ○ Identify communities at increased risk of disease ○ Help generate hypotheses to test with additional studies PERSON Pertussis occurrence by standard 5-year age groups is Personal characteristics may affect illness, shown in Figure 1.13a. The highest rate is clearly among organization and analysis of data by “person” may children 4 years old and younger. But is the rate equally use: high in all children within that age group, or do some ○ Inherent characteristics of people (for children have higher rates than others? example, age, sex, race) To answer this question, different age groups are ○ Biological characteristics (immune needed. Examine Figure 1.13b, which shows the status), same data but displays the rate of pertussis for ○ Acquired characteristics (marital status), children under 1 year of age separately. Infants ○ Activities (occupation, leisure activities, account for most of the high rate among 0–4 year ○ Use of medications/tobacco/drugs) olds. Public health efforts should thus be focused ○ The conditions under which they live on children less than 1 year of age, rather than on (socioeconomic status, access to medical the entire 5-year age group. care). Epidemiologists begin the analysis of personal SEX data by looking at each variable separately. Males have higher rates of illness and death than Sometimes, two variables such as age and sex can females for many diseases. be examined simultaneously. Person data are For some diseases, this sex-related difference is usually displayed in tables or graphs because of genetic, hormonal, anatomic, or other inherent differences between the sexes, and AGE affects susceptibility or physiologic responses. The single most important “person” attribute, ○ For example: Premenopausal women because almost every health-related event varies have a lower risk of heart disease than with age men of the same age. This difference has Factors that vary with age: Susceptibility, been attributed to higher estrogen levels opportunity for exposure, latency or incubation in women period of the disease, and physiologic response Meanwhile, the sex-related differences in the (which affects disease development) occurrence of many diseases reflect differences in When analyzing data by age, epidemiologists try opportunity or levels of exposure. to use age groups that are narrow enough to ○ Example: Figure 1.14 shows the detect any age-related patterns that may be differences in lung cancer rates over time present in the data among men and women.34 The ○ Chronic diseases- The 10-year age group difference noted in earlier years has been is adequate attributed to the higher prevalence of ○ Other diseases- 10-year and even 5-year smoking among men in the past. age groups conceal important variations in disease occurrence by age LSSM | BIOMED 121 4 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO ○ variables that are easiest to measure may not accurately reflect the overall concept Epidemiologists commonly use occupation, family income, and educational achievement, while recognizing that these variables do not measure socioeconomic status precisely These patterns may reflect more harmful exposures, lower resistance, and less access to health care. A few adverse health conditions occur more frequently among persons of higher socioeconomic status. ○ Example: Gout was known as the “disease of kings” because of its ETHNIC AND RACIAL GROUPS association with the consumption Differences in racial, ethnic, or other group of rich foods. variables may reflect differences in susceptibility Other conditions associated with or exposure, or differences in other factors that higher socioeconomic status influence the risk of disease, such as include breast cancer, Kawasaki socioeconomic status and access to health care. syndrome, chronic fatigue Adverse health conditions increase with syndrome, and tennis elbow. decreasing socioeconomic status ○ Differences in exposure account for the Examples: differences in the frequency of these ○ Tuberculosis is more common among conditions. persons in lower socioeconomic strata ○ Associated with decreased income: Infant mortality and time lost from work due to disability ○ Gout- “disease of kings” due to its association with the consumption of rich foods ○ Other conditions associated with higher socioeconomic status: Breast cancer, Kawasaki syndrome, chronic fatigue syndrome, and tennis elbow (lateral epicondylitis) SOCIOECONOMIC STATUS Difficult to quantify Made up of many variables such as occupation, family income, educational achievement or census track, living conditions, and social standing LSSM | BIOMED 121 5 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO Lecture 11: DESCRIPTIVE STUDY DESIGNS Main objective: To provide a comprehensive and Why should a clinician be concerned with disease detailed description of the case(s) under etiology? How do we design and conduct studies to observation elucidate the etiology of and risk factors for human ○ This allows other physicians to identify disease? If we mount a preventive intervention, how do and potentially report similar cases from we know if it will be effective? Such studies are critically their practice, especially when they share important in both clinical medicine and public health geographic or specific clinical practice. characteristics. 1) Prevention is a major responsibility of the Examples: physician and the broader public health ○ 2015 witnessed an outbreak of Zika virus community; both prevention and treatment in Latin America. In 2016, infants are born should be viewed by the physician as essential with microcephaly in Zika virus-affected elements of his or her professional role. Most areas in Brazil. Another case report was opportunities to prevent disease require an published about the offspring of a understanding of the etiology or causes of Slovenian woman who lived and worked disease so that exposure to a causative risk in Brazil and became pregnant in factor can be reduced or the pathogenic chain February 2015. The Zika virus was found leading from the causal factor to the in the fetal brain tissue. development of clinical illness can be interrupted. Advantages: 2) Patients and their families often ask the physician ○ Hypothesis-generating questions about the risk of disease. ○ Simple 3) In the course of doing clinical work and making ○ Inexpensive bedside observations, a physician often “gets a ○ Easy to conduct hunch” regarding a possible relationship between Disadvantages: a factor and the risk of a disease that is as yet not ○ The lack of a comparison group is a major understood. Public health practitioners must disadvantage. understand how conclusions regarding health ○ External validity (generalizability) is risks to a community are determined and how a limited, given the biased selection of foundation for preventive measures and actions cases (all identified in clinical practice). is developed based on population-centered data ○ Any association observed in a case report that are properly interpreted in their biological or a case series is prone to potentially context. Only in this way can rational policies be unmeasured confounding unbeknown to adopted for preventing disease and enhancing the investigators. the health of populations at the lowest possible cost. CROSS-SECTIONAL (DESCRIPTIVE) Gathering data on one or more health-related CASE REPORTS AND CASE SERIES variables (e.g., exposures (risk factors) or Case report- individual-level observations outcomes (diseases)) as they exist in a specific describing a particular clinical phenomenon in a population at a particular point in time. single patient Data are analyzed solely to assess the distribution Case series- describes more than one patient of one or more variables with similar problems. Providing a "snapshot" of the frequency and Both case reports and case series are considered characteristics of a disease in a population at a the simplest of study designs, while some say specific time they are merely “prestudy designs” LSSM | BIOMED 121 6 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO ○ Effectively assess disease or risk factor utero in the first, second, or third prevalence trimester of the pregnancy during the ○ Help evaluate disease burden and outbreak. healthcare needs Problem: The authors themselves Advantages: stated, “The observed association ○ Simple is between pregnancy during an ○ Inexpensive influenza epidemic and ○ Typically, they do not raise significant subsequent leukemia in the ethical concerns offspring of that pregnancy. It is Disadvantages: not known if the mothers of any ○ Risk of bias (e.g., selection or of these children had influenza measurement bias)- results may not during their pregnancy.” What we represent the true situation in the are missing is individual data on population exposure (influenza infection). Why didn’t the investigators ECOLOGIC STUDIES obtain the necessary exposure A “study of group characteristics”- the first data? The likely reason is that the approach in determining whether an association investigators relied on and used exists birth certificates and data from a ○ The initial approach to identifying an cancer registry; both types of association may involve studying group data are relatively easy to obtain. characteristics through ecological studies #2: Ecologic fallacy ○ We may attribute characteristics to group members that they do not possess as individuals. ○ This problem arises in an ecologic study because data are only available for groups; we do not have exposure and outcome data for each individual in the population. ○ Examples: Problem: Lack of information on whether Nobel Prize winners in the country consumed high amounts of chocolate Available information: Average chocolate consumption and Nobel laureates per capita for #1: each country The table shows incidence data for Variability in chocolate children who were not in utero during a consumption among individuals flu outbreak and for children who were in in the country is not considered LSSM | BIOMED 121 7 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO Ecologic studies rely solely on grouped data. ○ For example: country-by-country data on average salt consumption and average systolic blood pressure Interestingly, when the variability of exposure is limited, ecologic correlations may provide a more valid answer about the presence of an association than studies based on individuals. ○ “If cases and controls are drawn from a population in which the range of exposures is narrow, then a study may yield little information about potential health effects (Wynder and Stellman).” ○ Example: The relationship between salt intake and blood pressure A strong, graded correlation has been observed using country populations as the units of analysis. Explained by the narrow range of salt intake among individuals within each country and significant variability in average salt intake between countries Are ecologic studies of value? ○ Yes, they can indicate promising research avenues for exploring etiologic relationships but do not definitively prove that a true association exists. ○ Example: The association between region-specific prevalence of antibodies to HCV and region-specific PAT exposure suggests that differences in seroprevalence of HCV antibodies across regions may be attributable to PAT exposure. LSSM | BIOMED 121 8 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO Lecture 12: CROSS-SECTIONAL STUDIES Advantages: INTRODUCTION ○ Simple Gathering and analyzing data on one or more ○ Inexpensive health-related variables (e.g., exposures (risk ○ Typically, they do not raise significant factors) or outcomes (diseases)) as they exist in a ethical concerns specific population at a particular point in time Disadvantages: Used to solely assess the distribution of one or ○ Duration is unknown (the time between more variables. the disease onset and "today") Used in initially investigating the association ○ No information on whether the exposure between a specific exposure and a disease of occurred before the outcome interest. ○ Risk of bias (e.g., selection or Also called a “prevalence study” because we are measurement bias)- results may not identifying prevalent (existing) cases rather than represent the true situation in the incident (new) cases. population ○ Example: Imagine that we have sliced through the population, capturing levels BIAS of cholesterol and evidence of CHD at the TEMPORAL BIAS same time. Note that in this type of The impossibility/inability of determining a approach, the cases of disease that we temporal sequence “exposure-disease” when it is identify are prevalent cases of the disease the disease that causes the exposure. in question, because we know that they Can we determine the order of events between existed at the time of the study, but we the exposure and disease? do not know their duration (the interval ○ Questionnaire between the onset of the disease and Can gather information about “today”), or whether the exposure exposure happened before the outcome. Can find out if certain behaviors The type of study design is called a cross-sectional (e.g., being sedentary, smoking, study because both exposure and disease or drinking too much alcohol) outcome are determined simultaneously for each were present before the disease study participant; it is as if we were viewing a started “snapshot” of the population at a certain point in time. SURVIVAL/SELECTION BIAS ○ Effectively assess disease or risk factor When exposure is related to the duration of the prevalence disease; identifying only prevalent cases would ○ Help evaluate disease burden and exclude those who died sooner after the disease healthcare needs developed but before the study was carried out. ○ Example: The relationship of increased ○ Example: serum cholesterol level (the exposure) to If individuals who develop electrocardiographic (ECG) evidence of disease due to exposure have a coronary heart disease (CHD). We survey shorter survival time than those a population, and for each participant, we who are not exposed, the cases determine the serum cholesterol level we observe (survivors) might and perform an ECG for evidence of CHD. show a lower rate of past The presence of CHD defines a prevalent exposure This is because we are case. only looking at those currently LSSM | BIOMED 121 9 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO living with the disease and not accounting for those who died earlier. High serum cholesterol levels → Coronary Heart Disease (CHD) However, in a cross-sectional study, the relationship might reflect both the likelihood of developing CHD and how long people survive after the onset of disease PREVALENCE-INCIDENCE BIAS b. When exposure is observed less frequently in 1) Used to determine whether there is prevalent than in incident cases evidence of an association between exposure and disease from a DESIGN OF A CROSS-SECTIONAL STUDY cross-sectional study Given its biases, the results of a cross-sectional 2) We can calculate the prevalence of study should be used to generate hypotheses that disease in persons with exposure and can then be evaluated using a study design that compare it with the prevalence of disease includes incident cases and allows for establishing in persons without exposure. the temporal sequence of the exposure and the 3) We can compare the prevalence of outcome. exposure in persons with the disease to Widely used and are often the first studies the prevalence of exposure in persons conducted before moving on to more valid study without the disease. designs. EXAMPLES a. 1) We define a population. 2) Determine the presence or absence of exposure and the presence or absence of disease for each individual at the same time. 3) Each subject then can be categorized into one of four possible subgroups. LSSM | BIOMED 121 10 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO LECTURE 13: COHORT STUDIES INTRODUCTION AND DESIGN OF A COHORT STUDY ○ We begin with an exposed group and an unexposed group. Of the (a + b) exposed persons, the disease develops in a but not in b. Thus the incidence of the 𝑎 disease among the exposed is 𝑎+𝑏. Similarly, in the (c + d) unexposed persons in the study, the disease develops in c but not in d. Thus the incidence of the 𝑐 OBSERVATIONAL STUDIES disease among the unexposed is 𝑐+𝑑. No experimental changes are made A hypothetical example of a cohort study is Look at the exposures of study participants either shown in Table 8.2. at a specific moment or over a period of time, and then observe their outcomes either at that same moment or at a later date Include several types: ○ Cohort studies ○ In this cohort study, the association of ○ Ecologic studies smoking with coronary heart disease ○ Cross-sectional studies (CHD) is investigated by selecting for ○ Case-control studies study a group of 3,000 smokers (exposed) and a group of 5,000 nonsmokers DESIGN OF A COHORT STUDY (unexposed), all of whom are free of heart disease at baseline. Both groups are followed for the development of CHD, and the incidence of CHD in both groups is compared. CHD develops in 84 of the smokers and 87 of the nonsmokers. The result is an incidence of CHD of 28.0/1,000 in the smokers and 17.4/1,000 The design may include more than two groups in the nonsmokers. (such as no exposure, low exposure, and high ○ Because we are identifying new (incident) exposure levels), although only two groups are cases of disease as they occur, we can shown here for diagrammatic purposes. determine whether a temporal Exposure and Disease Association: relationship exists between the exposure ○ If there is a positive link between the and the disease (i.e., whether the exposure and the disease, we would exposure preceded the onset of the expect that a higher percentage of the disease). Establishing this time exposed group would develop the disease relationship is crucial for considering the compared to the unexposed group. exposure as a potential cause of the The calculations involved are seen in Table 8.1: disease. LSSM | BIOMED 121 11 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO SELECTION OF STUDY POPULATION COHORT STUDIES Essential characteristic in the design of cohort Wait for outcomes to develop within a population studies: Comparison of outcomes in an exposed ○ Often requires a long follow-up period group and in an unexposed group (or a group until enough cases are present with a certain characteristic and a group without When using a second approach (where a that characteristic, such as older or younger population is selected based on a characteristic participants unrelated to the exposure), the exposure of There are two basic ways to generate such interest might not occur for a long time, groups: sometimes even many years after the population 1) Create a study population by selecting has been selected groups for inclusion in the study based on ○ Typically needs an even longer follow-up whether or not they were exposed (e.g., period than the first approach occupationally exposed cohorts However, in both cases, the core design of the compared with similarly aged community cohort study remains the same: we compare residents who do not work in those those who were exposed to those who were not occupations) (Figure 8.3) This comparison is a key feature of cohort studies TYPES OF COHORT STUDIES PROSPECTIVE COHORT STUDY 2) Select a defined population before any of its members become exposed or before their exposures are identified. We could select a population based on some factor not related to exposure (such as the Also called concurrent cohort or longitudinal community of residence) and take study histories of, or perform blood tests or At the beginning of the study, the investigator other assays on, the entire population. chooses the initial population and then tracks the Using the results of the histories or the subjects over time until either the disease tests, one can separate the population develops or not into exposed and unexposed groups (or Problems: those who have and those who do not ○ Take a long time to complete have certain biological characteristics) ○ Limited time for research grants (Fig. 8.4). (generally a maximum of only 5 years) ○ Thus, it proves unattractive to investigators who are contemplating a new research question RETROSPECTIVE COHORT STUDY Also called a nonconcurrent prospective study Use of historical data to telescope (reduce) the frame of calendar time for the study and obtain LSSM | BIOMED 121 12 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO our results sooner. It is no longer a prospective endpoints of interest during the design because we are beginning the study with a follow-up period preexisting population to reduce the duration of > 62 years- many would already the study. have established coronary However, as shown in Fig. 8.7, the designs for disease both the prospective cohort study and the Thus, it is not rewarding to study retrospective or historical cohort study are individuals at this age to identify identical: We are comparing exposed and coronary disease incidence. unexposed populations. The only difference between them is calendar time. IDENTIFICATION OF CASES ○ In a prospective cohort design, exposure Incident cases of coronary events were identified and unexposure are ascertained as they by: occur during the study; the groups are ○ Checking the study population every two then followed for several years into the years future and incidence is measured. ○ Monitoring hospitalizations daily at the ○ In a retrospective cohort design, exposure only hospital in Framingham is ascertained from past records, and the The study aimed to test several hypotheses: outcome (development or no ○ The incidence of CHD increases with age development of disease) is determined and occurs earlier and more frequently in when the study is begun. males It is also possible to conduct a study using a ○ People with hypertension develop CHD at combination of prospective cohort and a higher rate compared to those with retrospective cohort designs. With this approach, normal blood pressure exposure is ascertained from objective records in ○ High blood cholesterol levels are linked to the past (as in a historical cohort study), and a greater risk of CHD follow-up and measurement of outcome continue ○ Tobacco smoking and regular alcohol use into the future. are associated with a higher incidence of CHD ○ More physical activity is linked to a lower risk of developing CHD ○ Gaining weight increases the likelihood of developing CHD ○ People with diabetes are at a higher risk of developing CHD EXAMPLES OF COHORT STUDIES Methods: THE FRAMINGHAM STUDY ○ Study population selection based on their Framingham (town in Massachusetts) location or other unrelated factors ○ < 30,000 residents ○ Monitor the group over time for the ○ Stable population (facilitate followup) development or determine those who Eligibility already had the relevant exposures (e.g., ○ Between 30 and 62 years of age at study hypertension, smoking, obesity, or high initiation cholesterol) People < 30 years- unlikely to ○ Track those who developed manifest the cardiovascular cardiovascular outcomes LSSM | BIOMED 121 13 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO ○ Key advantage: allows investigation of INFORMATION BIASES multiple exposures and how they interact Created when the quality and quantity of with each other using advanced statistical information collected varies between those techniques exposed to a risk factor and those not ○ Unlike traditional cohort studies (i.e., ○ Likely to occur in historical cohort studies comparing one group with exposure to where information is obtained from PAST another without it), this approach records enables a broader assessment of various Inconsistent gathering of information for both factors influencing health outcomes groups Examples: A biased judgment regarding disease occurrence Breast Cancer Incidence and Progesterone among participants occurs if the person in-charge Deficiency in determining whether a disease has developed knows the exposure status of participants and is aware of the hypothesis being tested ○ Can be mitigated by “masking” the evaluator Unintentional skewing of data analyses and interpretations of the findings might occur if the Childhood Health and Disease epidemiologists and statisticians analyzing the data have strong biases WHEN IS A COHORT STUDY NECESSARY? POTENTIAL BIASES SELECTION BIASES Non-participation and non-response ○ Traits of those who do not participate might be quite different from those who do participate leading to incorrect conclusions regarding the association between exposures and outcomes ○ Losing participants during follow-up can be problematic If individuals with the disease To carry out a cohort study, we must have some drop out and they differ from idea of which exposures are suspected a priori as those who remain in the study, it possible causes of a disease and are therefore will be challenging to accurately worth investigating. interpret the incidence rates in Consequently, a cohort study is indicated when both exposed and unexposed good evidence suggests an association of a groups disease with a certain exposure/s (evidence obtained from either clinical observations or LSSM | BIOMED 121 14 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO case-control or other types of studies). Often, we collect biological specimens at the study baseline (enrollment), allowing testing of these samples in the future, often when new test methods are developed and/or new hypotheses are generated. SEVERAL REASONS WHY A COHORT STUDY MIGHT NOT BE PRACTICAL Often not enough strong evidence to justify conducting a large and costly study to investigate the impact of a risk factor on the development of a disease Even if evidence is available, it can be difficult to find a group of people with exposure to the risk factor and a comparable unexposed group Lack of access to historical records or data allowing for a retrospective cohort study. ○ Thus, a long follow-up period for monitoring the population after exposure is needed Very low rates of occurrence of many diseases Need to enroll a large number of participants ensuring that there are enough cases for meaningful analysis LSSM | BIOMED 121 15 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO LECTURE 14: CASE-CONTROL STUDIES disease (cases) than in those who do not INTRODUCTION have the disease (controls). Having a comparison group is essential in How is it conducted: assessing the importance of clinical observations 1) Selecting cases (with the disease) and in a group of cases reported by doctors controls (without the disease) Comparison is a crucial part of epidemiological 2) Measure past exposure by interview or by research and is clearly shown by the case-control review of medical or employee records or study design of results chemical or biological assays of blood, urine, or tissues. DESIGN OF A CASE-CONTROL STUDY 3) If the exposure is dichotomous—that is, General design: exposure has either occurred (yes) or not ○ Suppose you are a clinician and you have occurred (no)—breakdown into four seen a few patients with a certain groups is possible. disease. You observe that many of them have been exposed to a particular agent—biological or chemical. You hypothesize that their exposure is related to their risk of developing this disease. ○ To examine the possible relation of exposure to a certain disease, we identify a group of individuals with that disease (called cases) and, for purposes of comparison, a group of people without POTENTIAL BIASES IN CASE-CONTROL STUDIES that disease (called controls). We then SELECTION BIAS determine what proportion of the cases Sources of cases was exposed and what proportion was ○ Hospital and clinic patients not. We also determined what proportion ○ Registries of individuals with specific of the controls was exposed and what diseases proportion was not. Important considerations: Example: In the example of children with ○ When using hospitalized cases, it is cataracts, the cases would consist of children with desirable to choose cases from multiple cataracts, and the controls would consist of hospitals in the community children without cataracts. For each child, it ○ Clearly define and specify eligibility would then be necessary to ascertain whether or criteria not the mother was exposed to rubella during her In using Incident or Prevalent Cases pregnancy with that child. We anticipate that if ○ Incident cases (newly diagnosed) the exposure (rubella) is related to the disease Generally preferable (cataracts), the prevalence of a history of Need to wait for diagnosis of new exposure among the cases (children with cases cataracts) will be greater than that among the Exclude patients who may have controls (children with no cataracts). died before diagnosis ○ Thus in a case-control study, if there is an ○ Prevalent cases (with disease for some association of exposure with a disease, time) the prevalence of a history of exposure Availability of a large number of should be higher in persons who have the cases LSSM | BIOMED 121 16 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO Identified risk factors may relate infection, that a mother of a child without more to survival with the disease a birth defect may not even notice or may than disease development have forgotten entirely. This type of bias Underrepresentation of cases due is known as recall bias. to death soon after diagnosis Nonrepresentative group of cases OTHER ISSUES ○ Keep in mind the issues in case selection Major concern: Cases and controls might have when interpreting data and deriving different characteristics or exposures (aside from conclusions the specific factor being studied) ○ Take into account possible selection Approach: Match cases and controls ensuring biases introduced by the study design and their similarity in the factors that could be by the manner of conducting the study important to the study Selection Of Controls Sources of controls MATCHING ○ Nonhospitalized people Choosing control subjects who share similar ○ Outpatient clinics characteristics (e.g., age, race, gender, ○ Hospitalized patients socioeconomic status, and occupation) with the cases INFORMATION BIASES Two types of matching (affect statistical analysis Major problem: Recall of exposure history of the case-control study): Telescoping ○ Group matching ○ Recalling past events as if they occurred ○ Individual matching recently Hesitancy to admit behaviors that might be PRACTICAL PROBLEMS stigmatizing Difficult or impossible to identify a suitable Recall bias control when matching based on a lot of ○ Suppose that we are studying the characteristics possible relationship of congenital ○ E.g., matching each case for race, sex, malformations to prenatal infections. age, marital status, number of children, ○ We conducted a case-control study and ZIP code of residence, and occupation interviewed mothers of children with congenital malformations (cases) and CONCEPTUAL PROBLEMS mothers of children without Characteristics chosen as the basis of matching malformations (controls). controls to cases cannot be studied ○ Each mother is questioned about E.g. For using marital status as the basis of infections she may have had during the matching controls to cases, an equal percentage pregnancy. of married individuals in both the case and ○ A mother who has had a child with a birth control groups is created defect often tries to identify some ○ If 35% of the cases are married, 35% of unusual event that occurred during her the controls can be ensured to be married pregnancy with that child. She wants to ○ The same proportion of married people in know whether the abnormality was both groups is artificially ensured caused by something she did. Why did it happen? Such a mother may even recall an event, such as a mild respiratory USE OF MULTIPLE CONTROLS LSSM | BIOMED 121 17 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO Many case-control studies will have more controls than cases. ○ Controls of the Same Type ○ Multiple Controls of Different Types WHEN IS A CASE-CONTROL STUDY NECESSARY? Useful initial step when looking for the etiology of an adverse health outcome Often the initial step in determining whether an exposure is associated with an increased risk of disease Investigating rare diseases LSSM | BIOMED 121 18 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO Lecture 15: EXPERIMENTAL STUDIES 2. ALLOCATING SUBJECTS TO TREATMENT GROUPS INTRODUCTION WITH NO RANDOMIZATION Objective: Modify the natural history of the - “Are there some alternatives to randomization disease that could be used?” ○ Prevent or delay disability or death Studies without comparison ○ Improve the health of a patient or the ○ Possible alternative: Case study population or case series Challenge: Choose the best preventive or ○ Comparison is necessary in therapeutic measures to achieve the objective aiming to establish a Randomized trial: Ideal design for assessing both cause-and-effect relationship the efficacy and the side effects of new forms of between therapy and the intervention resulting outcome Studies with comparison DESIGN OF A RANDOMIZED TRIAL ○ Historical controls- records of patients with the same disease who were treated before the new therapy became available Disadvantage: We will not know whether the difference is a result of the drug we are studying 1. Begin with a defined population or of other changes that 2. Participants are randomized to receive either a take place in many other new treatment or the current treatment. factors that may be 3. Follow up with the subjects in each treatment associated with the group and see how many improved in the new vs outcome over calendar the current treatment time 4. At times, a group given new treatment may be ○ Simultaneous nonrandomized compared with an untreated group controls- an alternative approach 5. We may compare two or > two groups receiving to historical controls is to use different therapies simultaneous controls that are E.g., In assessing a newly developed not selected in a randomized treatment for acquired immunodeficiency manner syndrome (AIDS), we compare the new Disadvantage: therapy with a current regimen, which is The physicians could better than no treatment at all know what the assignment of the next METHODS patient would be 1. SELECTION OF SUBJECTS - “Who is eligible to study?” 3. ALLOCATING SUBJECTS USING RANDOMIZATION Inclusion and exclusion criteria with great Critical element: Unpredictability of the precision and in writing before the study next assignment is started How is randomization done? ○ No element of subjective ○ Random allocation usually done decision-making via computer programs LSSM | BIOMED 121 19 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO ○ Backup: Manual randomization ○Sometimes, treatment or ○ Main purpose: Prevent potential prevention successfully biases from assigning participants reducing deaths from a to different treatment groups specific disease may not improve the overall survival without events ○ E.g., A European study on prostate cancer screening (13-year follow-up). About a 27% decrease in Stratified Randomization deaths from prostate ○ Because randomization does not cancer. However, the ensure comparability, another overall death rates were option is to use stratified similar between the two randomization, an assignment groups. Indicates that the method that can help increase screening did not have the likelihood of comparability of any effect on overall the study groups. mortality 3. PROGNOSTIC PROFILE AT ENTRY DATA COLLECTION It is necessary to verify that Essential: Data collected should be of the same randomization created quality and complete for each of the study comparable groups regarding risk groups. factors identified for negative What are the variables about which data need to outcomes be obtained on the subjects? E.g., if age is a significant risk 1. TREATMENT (ASSIGNED AND RECEIVED) factor, a confirmation that both Needed data: groups are similar in age after ○ Treatment group randomization is needed assignment Gathering data on these ○ Therapy received by the important factors when patient participants enter the study and 2. OUTCOME then comparing the groups on Need for comparable these factors at the start (before measurements: Include both administering any treatment) desired effects and side effects should be done Clearly define the criteria for all Another means to evaluate outcomes that will be measured whether the groups are in a study comparable is to look at an Avoid more careful measurement outcome unrelated to the of those receiving a new drug treatment being tested than those receiving current ○ E.g., Cancer-related death therapy rates would be similar Prevented by blinding (masking) between the two groups “All-Cause Mortatlity Outcome” if the trial is evaluating a LSSM | BIOMED 121 20 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO new treatment for a participant in the placebo group migraines may switch to the active treatment (or vice versa) outside OTHER ISSUES IN CLINICAL TRIALS the study’s protocol, which can 1. MASKING (BLINDING) complicate results and data Definition: Masking, or blinding, is a analysis. technique used in clinical trials to prevent participants, and sometimes researchers, 3. FACTORIAL DESIGN from knowing which treatment group Definition: Factorial design is a study they are in (active treatment or placebo). design used to evaluate the effects of Purpose: This approach minimizes biases more than one intervention that could arise if participants or simultaneously, often by testing researchers are aware of the treatment combinations of treatments. assignments. Example: A study could examine two Example: To mask subjects, researchers interventions (e.g., drug A and drug B) often use a placebo—an inactive and use four groups: one receiving drug substance resembling the active drug in A, one receiving drug B, one receiving appearance, taste, and smell. However, both, and one receiving neither. This using a placebo alone does not ensure design helps in understanding not only complete masking, as some participants the individual effects of each treatment may attempt to deduce if they are but also any combined or interactive receiving the real treatment or the effects. placebo. 4. NONCOMPLIANCE 2. CROSSOVER Definition: Noncompliance refers to Definition: Crossover designs involve participants not adhering to the assigned participants switching from one treatment protocol. This could mean treatment to another at a certain point missing doses, taking the wrong dosage, during the study. or even discontinuing the treatment Types: entirely. ○ Planned Crossover: In a planned Impact: Noncompliance can affect the crossover, participants are validity of the trial results, as it can lead assigned to one treatment for a to deviations from the expected period, and then switched to outcomes, making it harder to assess the another after a specified true effect of the treatment. duration. This design allows researchers to observe effects CONCLUSION under both treatments in the The randomized trial is generally considered the same participants, helping to gold standard of study designs. reduce variability. Randomized trials are always at the top of the list ○ Unplanned Crossover: This when ranking study designs based on quality is occurs when participants done in descending order. unintentionally switch from one Components of randomized trials are designed to treatment to another, often due prevent biases. to external factors. For example, LSSM | BIOMED 121 21 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO Lecture 16: CAUSAL INFERENCE APPROACHES TO ETIOLOGY IN HUMANS 1. Clinical observations- observed that virtually every patient 2. Try to identify routinely available data, the analysis of which might shed light on the question 3. Carry out new studies to determine whether there is an association between exposure and a A variety of designs of epidemiologic studies are disease and whether a causal relationship exists used to determine whether an association exists Studies of group characteristics between exposure and a disease outcome. ○ Ecologic studies Studies of individual characteristics APPROACHES FOR STUDYING THE ETIOLOGY OF DISEASE ○ Cohort Example: If we are interested in whether a certain ○ Case-control substance is carcinogenic in human beings, a first ○ Other types step in the study of the substance’s effect might be to… TYPES OF ASSOCIATION Approaches: REAL OR SPURIOUS ASSOCIATIONS 1. Test in animals (controlled lab) If we observe an association, we start by asking Advantages the question, “Is it a true (real) association or a ○ Control exposure dose false (spurious) one?” ○ Control environmental Example: A study of coffee consumption and conditions and genetic cancer of the pancreas. The possibility was factors suggested that the controls selected for the study ○ Minimum loss to had a lower rate of coffee consumption than was follow-up found in the general population. Disadvantages ○ Problems of data INTERPRETING REAL ASSOCIATIONS extrapolation across If the observed association is real, is it causal? species Causal association- we observe an association ○ Variations in species’ between exposure and disease, as indicated by response the bracket, and the exposure induces the ○ Uncertainty of development of the disease, as indicated by the generalizability of animal arrow findings to humans Confounding- there is the same observed 2. In vitro (cell or organ culture) association of exposure and disease, but they are 3. Randomized trials exposing humans to associated only because they are both linked to a carcinogens third factor, which is called a confounding variable Unethical and designated here as factor X. 4. Non-randomized observations in the human population Cohort studies Case-control studies LSSM | BIOMED 121 22 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO More examples: Cigarette smoking by pregnant TYPES OF CAUSAL RELATIONSHIPS women is associated with low birth weight in A causal pathway can be either direct or indirect. their infants. ○ Direct causation- a factor directly causes ○ The effect is not just the result of the a disease without any intermediate step birth of a few low-birth-weight babies in ○ Indirect causation- a factor causes disease this group of women. Rather, the entire but only through an intermediate step or weight distribution curve is shifted to the steps left in the babies born to smokers. If a relationship is causal, possible types: ○ Necessary and sufficient ○ Necessary but not sufficient ○ Sufficient but not necessary ○ Neither sufficient nor necessary NECESSARY AND SUFFICIENT ○ The reduction in birth weight is also not a Necessary- in the absence of Factor A, the disease result of shorter pregnancies. The babies never develops of smokers are smaller than those of Sufficient- the disease always develops in the nonsmokers at each gestational age. presence of Factor A Rare ○ Example: Household members do not uniformly acquire infectious disease from the index case (same exposure dose but different immune status, genetic susceptibility, or other characteristics) NECESSARY BUT NOT SUFFICIENT ○ The more a woman smokes, the greater Requires multiple factors, often in a specific her risk of having a low-birth-weight sequence baby. Example: Carcinogenesis is considered to be a multistage process involving both initiation and promotion. For cancer to result, a promoter must act after an initiator has acted. Action of an initiator or a promoter alone will not produce a cancer. LSSM | BIOMED 121 23 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO SUFFICIENT BUT NOT NECESSARY EVIDENCE FOR A CAUSAL RELATIONSHIP The factor alone can produce the disease but so Postulates for Disease Causation- proposed by can other factors that are acting alone. Henle in 1840; expanded by Koch in the 1880s Example: Either radiation exposure or benzene ○ The organism is always found with the exposure can each produce leukemia without the disease. presence of the other. However, even in this ○ The organism is not found with any other situation, cancer does not develop in everyone disease. who has experienced radiation or benzene ○ The organism, when isolated from one exposure, so although both factors are not who has the disease and cultured through needed, other cofactors probably are. several generations, produces the disease (in experimental animals). GUIDELINES FOR JUDGING WHETHER AN OBSERVED ASSOCIATION IS CAUSAL 1. TEMPORAL RELATIONSHIP if a factor is believed to be the cause of a disease, exposure to the factor must occur before the disease develops. NEITHER SUFFICIENT NOR NECESSARY 2. STRENGTH OF THE ASSOCIATION A “sufficient cause” is formed by a constellation The strength of the association is of risk factors, termed by him “component measured by the relative risk (or odds causes.” In Rothman’s conceptualization of ratio). sufficient cause, a pie chart formed by several The stronger the association, the more “component causes” represents the “sufficient likely it is that the relation is causal. cause.” Thus Rothman’s “sufficient cause” is a Example: The relative risk for the cluster of “component causes.” relationship of high blood pressure (exposure) to stroke (outcome) is very high. High blood pressure levels cause stroke. 3. DOSE-RESPONSE RELATIONSHIP As the dose of exposure increases, the Probably the most accurate representation of the risk of disease also increases. causal relationships operating in most chronic If a dose-response relationship is present, diseases. it is strong evidence for a causal Example: Individuals may develop CHD if they are relationship. However, the absence of a exposed to smoking, diabetes, and low dose-response relationship does not high-density lipoprotein (HDL) or to a necessarily rule out a causal relationship. combination of hypercholesterolemia, In some cases in which a threshold may hypertension, and physical inactivity. Each of exist, no disease may develop up to a these CHD risk factors is neither sufficient nor certain level of exposure (a threshold); necessary. Interestingly, if not most, individual above this level, disease may develop. risk factors are neither sufficient nor necessary. Example: The dose-response relationship between cigarette smoking and lung cancer. LSSM | BIOMED 121 24 EPIDEMIOLOGY LECTURE (EPIDEMI) 4TH YEAR, 1ST SEMESTER | 2023-2024 | DR. RJ SIMANDO 4. REPLICATION OF THE FINDINGS data are consistent with what we would If the relationship is causal, we would expect if the relationship between expect to find it consistently in different smoking and lung cancer is established as studies and in different populations. a causal one. If an association is observed, we would also expect it to be seen consistently within subgroups of the population and in different populations. 5. BIOLOGIC PLAUSIBILITY Refers to coherence with current knowledge. Used to demonstrate that epidemiologic 9. SPECIFICITY OF THE ASSOCIATION observations have sometimes preceded An association is specific when a certain biologic knowledge. exposure is associated with only one Gregg’s observations on rubella and disease. congenital cataracts preceded any Example: Cigarette manufacturers have knowledge of teratogenic viruses. pointed out that the diseases attributed Similarly, the implication of high oxygen to cigarette smoking do not meet the concentration in the causation of requirements of this guideline because retrolental fibroplasia, a form of blindness cigarette smoking has been linked to lung that occurs in premature infants, cancer, pancreatic cancer, bladder cancer, preceded any biological knowledge heart disease, emphysema, and other supporting such a relationship. conditions. 6. CONSIDERATION OF ALTERNATE EXPLANATIONS The problem in the interpretation of observed association: Is the relationship causal or is it a result of confounding? Considerations: ○ Extent investigators took into account other possible explanations ○ Extent other explanations were ruled out 7. CESSATION OF EXPOSURE If a factor is a cause of a disease, we would expect the risk of the disease to decline when exposure to the