LEC 2 Types of epidemiological studies PDF

TYPES OF EPIDEMIOLOGICAL STUDIES  LEARNING OBJECTIVES At the end of this session, participants will be able to: Describe the different types of epidemiological studies approach Discuss measures of risks Describe disease occurrence  BASIC EPIDEMIOLOGIC APPROACH Observe Co...

TYPES OF EPIDEMIOLOGICAL STUDIES  LEARNING OBJECTIVES At the end of this session, participants will be able to: Describe the different types of epidemiological studies approach Discuss measures of risks Describe disease occurrence  BASIC EPIDEMIOLOGIC APPROACH Observe Count cases (events) Describe Descriptive ▪Time, place, person Epidemiology ▪Calculate rates, Compare rates Develop hypotheses Test hypotheses Analytical Epidemiology Implement actions (control, prevention)  DESCRIPTIVE VS. ANALYTIC ▪ Descriptive: Activities related to characterizing the patterns of disease occurrence in terms of person, place or time. Hypothesis Generating ▪ Analytic: Activities related to identifying possible causes for the occurrence of disease. Hypothesis Testing  DESCRIPTIVE EPIDEMIOLOGY STRENGTHS AND LIMITATIONS ▪ Inexpensive ▪ Time saving ▪ Describe disease patterns ▪ Formulate research questions ▪ Unable to test epidemiologic hypotheses  DISTRIBUTION ▪ How common is the disease? ▪ Who gets disease? ▪ Where does the disease occur? ▪ When does the disease occur?  HOW COMMON OR UNUSUAL IS DISEASE OCCURRENCE Endemic: The ongoing, usual level of or constant presence of a disease within a given population or a geographic area. Epidemic: The occurrence of disease at a higher level than normally expected in a population. Pandemic: An epidemic that is widespread across a country, continent or large populace (possibly worldwide).  DESCRIPTIVE EPIDEMIOLOGY - PERSON Disease does not occur at random. Not all persons within a population are equally likely to develop a particular disease.  DESCRIPTIVE EPIDEMIOLOGY - PERSON Variation of occurrence in relation to personal characteristics may reflect differences in level of: exposure to causal factors susceptibility to causal factors the need for some level of both susceptibility and exposure  DESCRIPTIVE EPIDEMIOLOGY - PERSON Personal characteristics that are commonly examined with respect to disease occurrence are: age race gender marital status education income occupation DESCRIPTIVE EPIDEMIOLOGY - PERSON  DESCRIPTIVE EPIDEMIOLOGY - PERSON Possible explanations for relationship between age and TB: 1. Long latent period between infection and clinical symptoms - skews toward later life. 2. Elderly persons are more likely to live in closed communal settings- easy spread of disease. 3. Older persons are more likely to have other illnesses, making them more susceptible. 4. Decline in immune function associated with normal aging may make them more susceptible. 5. Elderly individuals lived through time periods when TB was more common (birth cohort effect).  DISTRIBUTION ▪ How common is the disease? ▪ Who gets disease? ▪ Where does the disease occur? ▪ When does the disease occur?  DESCRIPTIVE EPIDEMIOLOGY - PLACE Where are the disease rates highest? Where are the disease rates lowest? Does the disease rate vary by country, region, etc? DESCRIPTIVE EPIDEMIOLOGY - TIME When is the disease common? When is the disease rare? Is the frequency of the disease in the present different from the frequency of the disease in the past? DESCRIPTIVE EPIDEMIOLOGY - TIME TYPES OF TRENDS AS RELATED TO TIME: 1. Secular trend(long term): occurrence of disease over a long period of time, decades, years. 2. Seasonal trend(Medium term): Variation in occurrence by seasons of the year, like diarrheal diseases which increase during summer. 3. cyclic trend(Short term): Like what happens during epidemics of diseases. TYPES OF EPIDEMIOLOGICAL STUDIES Types of Epidemiological studies Descriptive studies Analytical studies Other classification: Observational studies Non-observational studies Types of Analytic studies Cohort (Prospective) studies Case- control studies(Retrospective studies) Experimental studies(interventional studies) Observational studies: Descriptive studies Cohort (Prospective) studies Case-control (Retrospective) studies Non-observational studies: Experimental studies (interventional) studies  TYPES OF DESCRIPTIVE STUDIES Case reports and case series Descriptive incidence studies Cross-sectional studies (Descriptive prevalence studies) Ecologic (correlational) studies  DESCRIPTIVE STUDIES USES Display patterns of occurrence Focus on person, place, time Used for Program planning ▪Generating hypotheses  CASE REPORTS AND CASE SERIES Profile of a case or case series Generate new hypotheses Interface: medicine and epidemiology Numerator data only No measure of disease occurrence  CASE REPORT This is the most basic type of descriptive study of individuals consisting of a careful, detailed report by one or more clinicians of the profile of a single patient  CASE REPORT - EXAMPLE 1961: A case report was published of a 40-year old pre-menopausal woman who developed a pulmonary embolism (PE) 5 weeks after beginning to use an oral contraceptive (OC) preparation to treat endometriosis.  CASE REPORT - EXAMPLE PEs are not common in this age group, could OC use be the cause? However, OC use is not uncommon in this age range. Are women who develop PEs more likely to use OCs than are women who do not use OCs?  CASE SERIES A description of the characteristics of a number of patients with a given disease. (A series of case reports)  CASE SERIES - EXAMPLE 1980-1981: In a 6 month period, five young, previously healthy homosexual men were diagnosed as having Pneumocystis carinii pneumonia (PCP) at three hospitals. This clustering of cases was striking in that PCP was seen almost exclusively in the elderly. This unusual circumstance suggested that these individuals were actually suffering from an unknown underlying condition. (AIDS)  CASE SERIES - EXAMPLE However, to test these hypotheses, studies that evaluated if the risk of disease was different among individuals exposed or not exposed to a factor of interest was needed. Descriptive study Analytic study  DESCRIPTIVE INCIDENCE STUDIES Patterns in occurrence of incident cases (often from surveillance data) Defined population (denominators from census) Specified period of time Optionally, distribution of cases by factors of interest ECOLOGIC (CORRELATIONAL) STUDIES Exposure and disease at aggregate (e.g., country) level Data from groups not individuals Unit of observation is a population Limitation: no individual link of E-D Advantages: –Quick, inexpensive, data available Primary Disadvantage: –Aggregate association may not = individual  CORRELATIONAL STUDIES(ECOLOGICAL STUDIES) These studies provide a crude way of exploring associations between factors and disease. They are considered to be hypothesis generating rather than hypothesis testing. The group rather than the individual is the unit of comparison. Example 1: Comparison of the trend of saccharin usage to the trend in bladder cancer rates in the United States. example 2: During the period 1950-1969, The national cancer institute in USA published maps showing mortality rates for cancer by county. They found a clustering of cancer lung in the North east and Southeast and on the Gulf coast. An ecological study correlating these county rates with industry concentration data revealed that lung cancer mortality Was elevated in counties with paper, chemical petroleum and transportation industries.  This study hypothesized that lung cancer in certain coastal areas was associated with the ship-building industry. There was a need to test this hypothesis to prove the association. Blot et al., 1978 conducted a case control study and confirmed the association between shipbuilding and lung cancer, possibly as a result of asbestos exposure.  Although correlation studies might be an inexpensive mean for generating hypotheses, yet one should be cautious in drawing conclusions regarding individual risk based on group risk. Because data on individual behavior that may influence risk have not been collected. This may cause the ecological fallacy.  Example of Ecologic Analysis Age-adjusted incidence of rectal 25 20 cancer per 100,000 USA 15 10 5 0 0 20 40 60 80 100 120 Annual Per Capita Beer Consumption (liters)  CROSS-SECTIONAL STUDIES (DESCRIPTIVE PREVALENCE STUDIES) "Snapshot" of well-defined population Can classify exposures and diseases at same time Captures all existing disease (serosurveys capture asymptomatic cases) Advantages: − Quick, inexpensive, useful Disadvantages: − Uncertain temporal relationship − Survivor effect  CROSS-SECTIONAL SURVEY ▪Exposures and disease status are assessed simultaneously among individuals in a well-defined population. ▪Cross sectional surveys while easy and rapid to accomplish, they don't establish the temporal sequence of events necessary for drawing causal inferences.  CROSS-SECTIONAL SURVEY EXAMPLE In examinations carried out by the National Health Survey in US , the prevalence of coronary heart disease and level of serum cholesterol were determined at the same visit. The fact that those with CHD had a higher mean cholesterol level than those without CHD does not necessarily lead to the conclusion that elevated serum cholesterol increases the risk of CHD. This may well be so. But it is only by demonstrating increase CHD in people with previously elevated cholesterol that a causal inference about the relationship may be drawn.  CROSS-SECTIONAL STRENGTHS AND LIMITATIONS ▪ These studies generate a lot of data. ▪ Temporal relationship between the exposure and the disease cannot be defined. REMINDER: DESCRIPTIVE VS ANALYTIC EPIDEMIOLOGY Descriptive epidemiology deals with the distribution of disease Descriptive studies estimate disease frequency and time trends and (sometimes) generate etiologic hypotheses Analytic epidemiology deals with determinants of disease. Analytic studies test etiologic hypotheses in order to establish or reject causal links between Exposure & Outcome. ANALYTIC STUDIES – OVERVIEW Goal: to determine the relationship between exposure and disease with validity and precision ▪Assess determinants of disease ▪Focus on risk factors, causes Analyze distribution of exposures and disease Used for –Testing hypotheses –Looking for / quantifying associations ANALYTIC STUDIES – HALLMARK Q. What is the hallmark feature that distinguishes an analytic study from a descriptive study? A. Comparison Group  BASIC QUESTION Are exposure and disease linked? Exposure Disease  Additional questions To link exposure and disease: ▪What is the exposure? ▪Who are the exposed? ▪What are the potential health effects? ▪What approach to take to study the relationship between exposure and effect? Basic analytic studies Cross sectional studies Cohort studies Case control studies Experimental studies ANALYTICAL CROSS-SECTIONAL STUDIES Cross-sectional studies are sometimes considered as a type of analytic study and used to test epidemiologic hypothesis! WHEN? When current values of the exposure variables are unalterable over time EXAMPLE! Factors present at birth (eye color, blood group…) MEASURING ASSOCIATION IN ANALYTICAL CROSS-SECTIONAL SURVEYS ill Non ill Total Exposed a b a+b Unexposed c d c+d Prevalence Exposed= a / (a+b) Prevalence Unexposed = c/ (c+d) Prevalence exposed Prevalence ratio (PR)* = Prevalence not exposed Excess prevalence (EP)* = Prevalence exposed - Prevalence not exposed *PR and EP are the cross sectional analogs of relative risk and excess risk Cohort studies DESIGN OF A COHORT STUDY Subjects themselves determine their exposure status Synonyms Concurrent Follow up studies Incidence studies studies Cohort studies Longitudinal Forward-looking studies studies Prospective studies PROSPECTIVE COHORT STUDY Disease Exposure Study starts occurrence time When study starts, the relevant events may or may not have occurred, but the outcomes have certainly not yet occurred. Disease Study starts Exposure occurrence time Retrospective cohort study Disease Exposure Study starts occurrence time outcome All the relevant events (both the exposures and outcomes of interest) have already occurred when the study is initiated. COHORT STUDY ▪Investigator does not determine exposure ▪Enroll subjects on basis of exposure status (or enroll all members of a group, then classify by exposure) ▪Follow subjects over time and record occurrence of health event (outcome of interest) ▪Compare rates of disease occurrence among exposed and unexposed groups of persons  COHORT STUDIES – TIMING Exposure Disease Exposed ? Unexposed ? Prospective Retrospective  COHORT STUDIES – STEPS 1. Identify exposed group of concern, or enroll entire population 2. Identify appropriate unexposed comparison group, or characterize exposure 3. Document disease among exposed and unexposed groups 4. Calculate risks or rates of disease 5. Calculate risk ratios or rate ratios 6. Calculate tests of significance or confidence intervals COHORT STUDIES, STEPS 1 AND 2 — EXPOSED / UNEXPOSED GROUPS Entire population ▪ Classify as exposed or unexposed Exposed Group ▪ Need comparison unexposed group STEP 3 — FOLLOW-UP FOR OUTCOME Follow-up ▪ Major challenge for prospective cohort studies Outcomes ▪ Often multiple ▪ Progression of disease ▪ Spectrum of disease STEP 4 — MEASURES OF OCCURRENCE Cumulative incidence (attack rate, risk) ▪ Number of new cases at end of follow-up divided by number of disease-free persons at start of follow-up Person-time rate ▪ Number of new cases at end of follow-up divided by person-time at risk (e.g., person-years of disease-free follow-up)  COHORT – 2-BY-2 TABLE Ill Well Total Risk Exposed a b H1 a / H1 Unexposed c d H2 c / H2 RR = RiskExposed / RiskUnexposed  COHORT STUDY Cause Effect Disease No disease Exposed Disease No disease Population Unexposed  COHORT STUDY Cause Effect a Disease b No disease Exposed c Disease d No disease Population Unexposed COHORT STUDY, STEP 4 — MEASURE OF DISEASE OCCURRENCE Ate vanilla Ill Well Total % Ill ice cream? Yes 43 11 54 ____ No 3 18 21 ____ Total 46 29 75 ____ COHORT STUDY, STEP 5 — MEASURE OF ASSOCIATION Ate vanilla Ill Well Total % Ill ice cream? Yes 43 11 54 79.6% ____ No 3 18 21 14.3% ____ Total 46 29 75 ____ 61.3% COHORT STUDY, STEP 6 — TEST OF SIGNIFICANCE, CONFIDENCE INTERVAL Ate vanilla Ill Well Total % Ill ice cream? Yes 43 11 54 79.6% ____ No 3 18 21 14.3% ____ Total 46 29 75 ____ 61.3% RR = ____ 5.6  COHORT STUDY: MEASURE Incidence Rate (or Attack Rate) ▪Exposed group: Ie ▪Not Exposed group: Ine → How to compare these two risks? What can we calculate? 68  MEASURES IN A COHORT STUDY Comparisons : a b Risk Excess RE= Ie – Ine Relative Risk RR = c d Ie Ine a ( ) RR = a + b c ( ) c+d 69 ATTRIBUTABLE RISK (AR) AR = Re - Ru Outcome yes no a b a exposed a+b = Re a+b not c c d c+d = Ru exposed c+d a c Attributable Risk = a+b - c+d Attributable Risk = Re – background risk 70  Attributable Risk (AR) ▪Quantifies disease burden in exposed group attributable to exposure in absolute terms ▪AR = Re - Ru ▪Answers: ▪what is the risk attributed to the exposure? ▪what is the excess risk due to the exposure? ATTRIBUTABLE RISK (AR) AR = Re - Ru Outcome yes no a b a exposed a+b = Re a+b not c c d c+d = Ru exposed c+d a c Attributable Risk = a+b - c+d Attributable Risk = Re – background risk 72 COHORT STUDIES — ADVANTAGES In principle, allows for complete description of experience after exposure Clear temporal sequence of E and D Well-suited for rare exposures Can measure incidence of disease Can study multiple effects (risks and benefits) of an exposure Understandable by non-epidemiologists COHORT STUDY: ADVANTAGES If prospective, minimizes bias in: ▪ Selection, since enrollment is completed without knowledge of disease outcome ▪ exposure measurement COHORT STUDIES — DISADVANTAGES Many subjects needed for rare disease Follow-up: logistics, losses Exposure can change over time Prospective: time-consuming, costly, observation can influence behaviors Retrospective: requires suitable records Changes in practice, usage, exposures may make findings irrelevant COHORT STUDY: ISSUES Is exposure unchanging throughout the period of follow-up? If persons are lost-to-follow-up, before they develop disease and before the study is completed….can they be included in the analysis? COHORT STUDY: ISSUES Was exposure accurately measured? ▪ Were exposure data collected explicitly for the purpose of the study? Are data on potential confounders available for analysis? COHORT STUDY: ISSUES Measurement of exposure status or disease outcome should be similar for all compared groups COHORT STUDIES – SUMMARY Test hypotheses about disease risk factors / causes Cohort studies ▪Exposure → disease (cause → effect) ▪Prospective or retrospective ▪Measure of association = RR ▪Often large, expensive CASE-CONTROL STUDIES CASE-CONTROL STUDY – DEFINITION Observational analytic study that enrolls one group of people with a certain disease, chronic condition, or type of injury (“cases” or “case-patients”), and a group of people from the same population but without the health problem (“controls”), and compares exposures, behaviors and other characteristics to identify differences to identify and quantify associations, test hypotheses, and identify causes What is a case-control study? A case-control study is an observational analytic study that enrolls one group of people with a certain disease, chronic condition, or type of injury (the “cases” or “case-patients”), a group of people from the same population but without the health problem (“controls”). The investigator Collects information from both the cases and controls, and compares their exposures, behaviors and other characteristics to identify differences between the cases and controls, In order to identify and quantify associations, test hypotheses, and identify causes.  CASE-CONTROL STUDIES – TIMING Exposure Disease Exposed ? Yes (case) Unexposed ? No (control) Investigator CASE-CONTROL STUDIES – FLOW CHART Exposed Cases Unexposed Source Population Exposed Controls Unexposed CASE-CONTROL STUDIES – STEPS 1. Identify cases of disease of concern 2. Identify appropriate non-diseased comparison group (“controls”) 3. Document exposures among cases and controls 4. Calculate odds ratios 5. Perform statistical tests or calculate confidence intervals STEP 1 — IDENTIFY CASES In public health setting, cases often identified by surveillance Need case definition Incident (new) rather than prevalent (pre-existing) cases  STEP 2 — CONTROL SELECTION – GUIDELINES Critical design issue No optimal group for all situations Controls should... ▪ Not have the disease being studied ▪ Represent population from which cases arose ▪ Represent persons who, if developed disease, would have been a case in the study ▪ Be selected independently of exposure WHY HAVE CONTROLS? Provide estimate of prevalence of exposure in population “Expected” prevalence of exposure among cases if no association WHERE TO FIND CONTROLS Population-based Hospital- or clinic-based Neighbors Friends Other, such as ▪ Co-workers ▪ Classmates NUMBER OF CONTROLS Availability Ratio controls / cases Trade-off: cost vs. power Decision based on power calculation More than one control group? NUMBER OF CONTROLS PER CASE STEP 3 — DOCUMENT EXPOSURES Questionnaires Preexisting records Biomarkers STEP 4 — MEASURE OF ASSOCIATION Think Odds Ratio (OR) Odds ratio ▪ Good estimator of risk or rate ratio, especially for rare disease ▪ Odds of exposure among cases divided by odds of exposure among controls CASE-CONTROL – 2-BY-2 TABLE Case Control Exposed a b Unexposed c d V1 V2 Odds Ratio = (a/c) / (b/d) = ad / bc STEP 5 — STATISTICAL TESTING AND/OR CONFIDENCE INTERVAL Could elevated odds ratio be result of chance rather than true association? Statistical testing ▪ P-value from chi-square or Fisher Exact Test ▪ Compare with pre-set cut-off, e.g., 0.05 Confidence interval ▪ Shows range of values consistent with study ▪ CI that does not include 1.0 is statistically significant INTERPRET THE RESULTS (1) Cases Controls Total Exposed 12 7 19 Unexposed 3 8 11 15 15 30 Odds Ratio = (12 x 8) / (7 x 3) = 4.6 Chi-square = 3.59, P-value = 0.06 Confidence Interval = (0.7, 34.1) INTERPRET THE RESULTS (2) Cases Controls Total Exposed 12 21 33 Unexposed 3 24 27 15 45 60 Odds Ratio = (12 x 24) / (21 x 3) = 4.57 Chi-square = 4.55, P-value = 0.03 Confidence Interval = (1.01, 27.9)  In Case-Control Study the individuals are selected on the basis of whether they HAVE DISEASE (cases) or DON’T HAVE A DISEASE (controls). Then The groups are compared with respect to the proportion having a history of an exposure MEASURE OF ASSOCIATION IN CASE-CONTROL STUDY No direct calculation of risk Proportion of exposure RR estimation = Odds Ratio 99  WHAT ARE ODDS? Probability that an event will happen Probability that an event will not happen Odds Ratio (OR) case Controls Exposed a b Not exposed c d a+c b+d a ( ) Odds exposition among cases a *d Odds ratio (OR) = = c = Odds exposition among controls ( b ) c * b d 101 CASE-CONTROL – ADVANTAGES Quick and inexpensive Well-suited for rare diseases Better for diseases with long latency Can study multiple exposures Requires fewer subjects at entry Few ethical problems  CASE-CONTROL – DISADVANTAGES Usually cannot measure disease risk Determination, selection, and enrollment of appropriate control group may be difficult (potential selection bias) Relies on recall or records for info on past exposures (potential recall bias) May be difficult to determine that ‘cause’ preceded ‘effect’ Unsuitable for rare exposures Less familiar to non-epidemiologists Experimental studies Experimental Study, Defined Experimental Study = “study in which the investigator intentionally alters one or more factors and controls the other study conditions in order to analyze the effects of so doing. A study in which conditions are under the direct control of the investigator.” Defining feature (in contrast to observational studies): – Investigator assigns exposure to study subjects Although experiments provide the strongest evidence for testing any hypothesis, they are rarely possible in human population. EXPERIMENTAL STUDY Investigator determines participants’ Cause Effect exposure status Disease Intervention (“Exposed”) No disease Disease Study Comparison sample (“Unexposed”) No disease CLASSIFICATION OF EXPERIMENTAL STUDIES Based on Classification Type of intervention Therapeutic; Preventive Unit of randomization Individual; Community Phase Phase I, II, III, or IV Randomization Fixed; Adaptive; Blocking Design Parallel; Cross-over Masking Single, Double, Triple Characteristics of Experimental Studies Experimental study – Has at least 1 intervention and 1 comparison group – Allocation of exposures should be random, so groups are similar / comparable – Follow each group prospectively until well-defined end-point / outcome. Trials can be conducted on individuals or communities. Allocation of participants to the study group and control group should be by random allocation.  Masking or blinding is necessary in these trials to avoid bias: 1. Double blinded studies: Both experimenter and subject are blind regarding the group to which the subject is assigned. 2. Single blinded studies: Only the subject is blind 3. Triple blinded studies: The experimenter, the subject and person responsible for data analysis are all blind. Blinding is very important when the outcome is subjectively determined. If outcome is death or stroke, blinding become less essential. Implications of Experimental Studies If groups are similar at baseline, differences in outcome can reasonably be attributed to action of the intervention (Assuming study is well done), has high validity, high credibility in establishing causality. Types of Intervention(Experimental) Studies: 1. Therapeutic (Clinical trials) 2. Preventive(Prophylactic trials) 3. Community trials 4.Natural experiments: Snow on cholera, Atomic bombing in Japan in 1945. Therapeutic – “Clinical trial” – Enrolls patients with existing disease or disability – Purpose: to determine ability of agent or procedure to reduce symptoms, prevent recurrence, or decrease risk of death from the disease TYPES OF INTERVENTION STUDIES: THERAPEUTIC Therapeutic ▪“Clinical trial” ▪Enrolls patients with existing disease or disability ▪Purpose: to determine ability of agent or procedure to reduce symptoms, prevent recurrence, or decrease risk of death from the disease ▪Secondary or tertiary prevention TYPES OF INTERVENTION STUDIES: PREVENTIVE Preventive ▪Primary prevention ▪Individuals without the study disease ▪Purpose: to determine ability of agent or procedure to reduce risk of developing disease among disease- free individuals ▪Trial of individuals or communities  COMMUNITY TRIAL Sensible study design if intervention is ecological (mass media, water supply, etc.) Randomize communities No selection of individual subjects for study, so savings in costs of individual screening and enrollment Baseline and follow-up community surveys essential Can use surveillance systems already in place PHASES Phase I – Test drug in healthy volunteers to examine how drug is handled (pharmacodynamics, etc.) particularly to assess short-term safety Phase II – Test drug in patients with the disease; initial test of efficacy; dose-response curves Phase III – Test drug against placebo or standard therapy for efficacy and safety / side effects, for licensure Phase IV – Test drug while in marketplace to gather additional safety information from larger group of patients PHASES Animal Phase IV and/or FDA Phase I Phase II Phase III (post- laboratory Approval approval) studies 15-30 < 100 100s-1000s many subjects subjects subjects About 4–5 About 8–9 years About 1½ years years Experimental Studies – Steps 1. Develop research question 2. Write protocol 3. Enroll study sample 4. Assign participants to exposure and control groups 5. Monitor participants in each group for study outcome (first occurrence of disease, improvement, side effects, etc.) 6. Analyze data  ENROLL STUDY SAMPLE (EXPERIMENTAL POPULATION) Experimental population ▪ Actual group in which trial is conducted ▪ Should yield results generalizable to reference pop., but must yield valid results ▪ Must be large enough ▪ Must be stable / available over time to obtain complete and accurate follow-up for duration of trial  ENROLL STUDY SAMPLE (STUDY POPULATION / STUDY SAMPLE) Study sample = Participants After ▪ Invitation ▪ Information provided on purpose of trial, study procedures, possible risks and benefits, possibility of allocation to group receiving no treatment or usual care ▪ Screening for eligibility according to predetermined criteria, such as absence of previous history of study end points, definite need for study treatment, or contraindication ELIGIBILITY CRITERIA Eligibility criteria = requirements that determine whether an indivdual can be included in a study Inclusion criteria = factors that must be met for an individual to be included in a study Exclusion criteria = factors that prevent an individual from being included in a study Eligibility criteria help define the study population, ensure safe and ethical research, and promote scientific validity Often includes age, gender, medical condition, previous treatment history, and other characteristics unique to the protocol. INTERVENTION AND COMPARISON GROUPS Intervention Group receives ▪ Therapy ▪ Preventive ▪ Other (particularly for operations research) Comparison Group receives ▪ Nothing ▪ Placebo ▪ Usual care Groups are called “arms”  RANDOM ALLOCATION OF PARTICIPANTS TO STUDY ARMS Randomization = allocation of each participant to an exposure group (e.g., intervention or comparison / placebo / usual care) by chance Reasons to randomize ▪ To achieve baseline comparability between groups of both measured and unmeasured characteristics,” so difference in outcome can be attributed to difference in intervention) ▪ Each participant has same chance of receiving any of the treatments under study ▪ Removes investigator bias in assigning patients to groups ▪ Increases validity of statistical tests  BLINDING (MASKING) Investigator(s) and/or study participants are kept ignorant of the group to which participants are assigned Purpose — to avoid bias in ascertainment of outcome Single blinded — participant does not know which intervention arm he/she is in Double blinded — neither participant nor investigator(s) know Triple blinded — participant, investigator(s), data analyst(s) do not know  BASIC TRIAL DESIGNS Parallel Group = subjects randomized to one group, groups followed in parallel to determine effect in each (most common design) Crossover design = subjects randomized to different sequences of treatments, but all subjects eventually get all treatments in varying order; subject serves as his/her own control BASIC TRIAL DESIGNS: PARALLEL; CROSSOVER Parallel Intervention (A) Comparison (A′) Crossover A A′ A′ A Washout period FOLLOW-UP PERIOD: THREATS TO VALIDITY Non-compliance — not sticking with regimen Contamination — subjects receive component of another treatment arm, e.g., controls receive treatment Additional threat to validity: Ineligibility — subject(s) should not have been enrolled in the study in the first place Loss to follow-up which may lead to ▪ Loss of power ▪ Bias ▪ Or both  FOLLOW-UP PERIOD: DATA MONITORING AND INTERIM ANALYSES Interim analysis = analysis during data accumulation, should be specified in protocol Data Monitoring Committee / DSMB — independent group, conducts interim analyses Purpose: to protect participant safety DMC then recommends: ▪continue trial as planned; ▪stop early for hazard; ▪stop because efficacy is unequivocally established; ▪stop because continuing the trial is futile  END POINT(S) / OUTCOME MEASURES Targeted outcome of the study, usually disease, death, symptom, or other clinical feature Based on the research question / hypothesis Types of end points: occurrence of event, time to event, laboratory response, etc. Should be well defined, stable, reproducible, unbiased, ascertainable in all participants COMPARISON TYPE Superiority – new intervention is more effective than comparison (placebo or usual care); most common type Equivalency – new intervention is as effective as comparison; effect of new intervention is not statistically significantly different from comparison, with adequate sample size Noninferiority – new intervention is not significantly weaker than current approach ADVANTAGES OF EXPERIMENTAL STUDIES Avoids selection bias Ensures only one factor differs among study arms Randomization minimizes confounding With little bias and confounding, differences are likely to be causal Accepted by medical and scientific community as the “gold standard,” i.e., has great credibility “Attractive statistically” — many statistical methods assume random assignment  DISADVANTAGES OF EXPERIMENTAL STUDIES May be complex and expensive, especially for low incidence outcomes Ethical concerns May lack generalizability / representativeness Resistance to randomization by clinicians, patients Administrative complexity Statistical complexity for complex designs HUMAN SUBJECTS PROTECTION, ETHICAL CONSIDERATIONS Major issue for Experimental Studies Research question is appropriate for study Role of Institutional Review Board (Human Subjects Committee) All study subjects properly informed (informed consent) Trial is conducted ethically THANK YOU