Introduction To Epidemiology Studies PDF

Introduction to Epidemiologic Study Design Dr Muhammad Ahmed Alshyyab Lecture 2 Chapter 5 Lecture objective To design an epidemiological study in a manner that will connect the data to the generated hypotheses. To clearly define the study des...

Introduction to Epidemiologic Study Design Dr Muhammad Ahmed Alshyyab Lecture 2 Chapter 5 Lecture objective To design an epidemiological study in a manner that will connect the data to the generated hypotheses. To clearly define the study design elements To define and differentiated between common types of epidemiologic studies Selected study design elements Referent (‘‘control’’) group, Observational studies The unit of observation Longitudinal versus cross-sectional observations Cohort versus case–control samples Necessity of a referent (‘‘control’’) group The index group; is exposed to the factor thought to influence occurrence of the study outcome. The referent or control group; remains unexposed to provide a reference for comparison. Without the baseline of observation provided by the referent group, it is impossible to determine whether the exposure had a positive  comparative studies in epidemiology Experimental studies In experimental studies (‘‘trials’’), the investigator introduces or withholds an exposure in order to observe its effects. The experimental allocation of the study exposure can be: Randomized trials Nonrandomized trials. Non-experimental studies (observational studies). Observational studies do not assign treatments to study participants. In a simple observational cohort design, subjects are classified as either ‘‘exposed’’ or ‘‘nonexposed’’ to the study factor of interest and are then followed and assessed for the study outcome. Incidences in the two groups are then compared. Randomized Experiment vs. Observational Cohort Randomized Experiment Observational cohort  Observational Study findings. Classify women into those who take hormones (E+) and those who don’t (E−). Compare cardiovascular disease rates in the groups. Experimental Study. Random Assign of hormones supplementation to some women (E+) and gives others a placebo (E-). Compare cardiovascular disease rates in the groups.  The Findings Observational Study. women who took hormones at The ‘‘fairness’’ of menopause had a lower risk of comparison. cardiovascular disease than women who did not. Randomization helps achieve ‘‘like-to-like’’ Experimental Study. random use of hormone replacement therapy actually increased the risk of cardiovascular diseases. Unit of observation The unit of observation in an epidemiologic study refers to the level of aggregation upon which measurements are available. This can vary from person-level data to region-level data, and everything in between: Persons ↔ Families ↔ Social groups ↔ Neighbourhoods ↔ Regions ↔ Nations Unit of Observation studies example Question. Does cigarettes smoke (the exposure) cause lung cancer (the disease outcome)? Person-level data Classify individuals as smokers or non- smokers. Assess & compare rates of LungCA in exposed and nonexposed groups. Aggregate-level data Classify level of smoking in various regions. Assess & compare rates of LungCA according to regional smoking rates. 11 Longitudinal versus cross-sectional observations Longitudinal observations address individual experiences over time. Longitudinal data are preferable when conducting etiologic research, because for a factor to be causal, it must clearly precede the event it caused by a reasonable amount of time. In contrast, cross-sectional observations do not permit the accurate time-sequencing of events within individuals. Note that the study design feature that distinguishes longitudinal observations from cross-sectional observations is the ability to accurately place the events for individuals on a time-line. Example on the difference between longitudinal and cross sectional study A single serological ascertainment for HIV, for example, is cross-sectional even if collected prospectively, because it is unable to determine when an individual became seropositive. To derive longitudinal data for HIV status, one would need a multiple of serological measurements to be obtained over time starting with seronegative individuals. These longitudinal measurements could then derive the approximate dates of seroconversion for individuals. Longitudinal data are preferable when conducting etiologic research, because for a factor to be causal, it must clearly precede the event it caused by a reasonable amount of time. However, for characteristics that do not change over time (e.g., genetic factors), it matters little whether the measurement is longitudinal or cross-sectional, because if we know the status of this factor now, we also know its status in the past. In addition, many human habits that are potentially changeable, such as dietary choices, display some degree of long-term permanence. For stable characteristics such as these, the current status of the attribute serves as a suitable proxy for its longitudinal equivalent. Cohort versus case–control studies Cohort versus case–control studies Cohort studies begin by identifying disease-free individuals. Study subjects are then classified according to risk factors thought to be associated with future disease occurrence. A period ensues during which disease is monitored. Incidences of events are then tallied and compared among the exposure groups. case–control studies begin by identifying people with the disease being studied (the case series). They then select non-cases from the same population that gave rise to the cases (the control series). Exposures to risk factors thought to be predictive of the study outcome are then ascertained retrospectively in cases and controls. The key distinction between cohort and case–control studies is The way in which subjects are sampled for study: cohort studies begin with disease- free study subjects, while case–control studies begin with diseased (cases) and disease-free (controls) study subjects. Nevertheless, both cohort and case–control studies rely on the reconstruction of events in individuals over time and are thus longitudinal.  Case-Control vs. Cohort Selection of subjects based on exposure (cohort) or disease (case-control) status? Hypothesis. Does cigarette smoking cause lung cancer? Cohort. Identify smokers (E+) and non- smokers (E-)  assess and compare lung cancer rates Case-control sample. Identify lung cancer cases (D+) and non-cases (D-)  assess and compare smoking histories (E+/E-) Common types of epidemiologic studies