Epidemiology Study Designs PDF

EPIDEMIOLOGY STUDY DESIGNS Ellen M. Wells Fall 2024 Reminders Knowledge check #5 due September 29 Problem set #3 due September 29 Knowledge check #6 due October 6  Introduction to study designs Case reports/case studies Cross-sectional studies Ecologic studies INTRODUCTION...

EPIDEMIOLOGY STUDY DESIGNS Ellen M. Wells Fall 2024 Reminders Knowledge check #5 due September 29 Problem set #3 due September 29 Knowledge check #6 due October 6  Introduction to study designs Case reports/case studies Cross-sectional studies Ecologic studies INTRODUCTION  Term Description Type Distribution “Distribution” of health related states or events Descriptive Person, place, time Screening/surveillance Epidemiology Hypothesis-generating studies Determinants Identification of “determinants” of health related states Analytical or events Determinants could be causes, risk factors, or epidemiology treatments Hypothesis-driven studies Application Application of knowledge of distribution and Applied determinants of disease to improve health Methods development/diagnostics epidemiology Public health campaigns Evaluation of policy or public health interventions 5  STUDY DESIGNS ANALYTICAL “determinants” hypothesis testing DESCRIPTIVE APPLIED “distribution” “application” surveillance public health policy 6  STUDY DESIGNS EXPERIMENTAL OBSERVATIONAL Investigator controls Investigator does not control exposure exposure 7 EXPERIMENTAL OBSERVATIONAL Case study/case series Non-randomized trial Cross-sectional Randomized trial Case-control Cohort Advanced designs 8 EXPERIMENTAL OBSERVATIONAL Case study/case series Non-randomized trial Cross-sectional Randomized trial Case-control Cohort Advanced designs 9  Case report/ case series Cross-sectional studies Case-control studies Cohort studies Randomized trials  Case report/ case series Cross-sectional studies Observational Case-control studies Cohort studies Randomized trials Experimental  Case report/ case series Hypothesis Cross-sectional generating studies Case-control studies Cohort studies Hypothesis testing Randomized trials  Case report/ case series Cross-sectional Descriptive studies Case-control studies Cohort studies Analytical / applied Randomized trials  Case report/ Weaker designs case series Not as informative about causality Cross-sectional studies Case-control studies Cohort studies Stronger designs Randomized trials More informative about causality  Target Populations Source Target population Population the study hypothesis addresses Study Source population Population from which participants are recruited Study population The individuals who participate in the study Validity Accuracy How close a measurement is to the truth Internal validity Accuracy within the study population External validity Accuracy related to the source or target population Target Source Study Internal validity Target Source External validity Study CASE REPORT & CASE SERIES Case report Case Description of a single case of a disease Case Case Case Case Case Case series Case Case Case Case Case Description of a series of cases of a disease Case report, case series Observational Descriptive Describes the experience(s) of a one (or more) patients with a given diagnosis or characteristics No comparison group Typical characteristics No manipulation of exposure Often related to a clinical diagnosis Occurring over a short time period Novel disease/condition Data abstracted from medical charts All data collected at one timepoint No comparison group Mt. Sinai J Med 73(7), 2006.  26 MMWR 65(14);372–374 Strengths Limitations Identification of new No comparison group diseases or conditions Inexpensive Generally small numbers Cannot calculate most Hypothesis generating measures CROSS-SECTIONAL STUDIES Person 1 2 3 4 5 6 7 8 Time Cross-sectional studies Study Population Gather data on exposure and disease Exposed; Exposed; Not exposed; Not exposed; Has disease No disease Has disease No disease Characteristics Observational Descriptive All data newly collected at one time No manipulation of exposure Not longitudinal Assess association between exposure and outcome Temporality Measure exposure and disease at the same time Sometimes cannot clearly identify which came first Exposure → Disease Disease → Exposure Extent of issue will vary Incidence vs prevalence Cross-sectional studies typically assess prevalence If high case-fatality rate Survivors may be overrepresented Survivors may not accurately reflect all cases Incidence-Prevalence (Neyman) Bias Cross-sectional studies include survivors Prevalent cases instead of incident cases Sampling Random sample More representative of population More likely to be generalizable More challenging to recruit Convenience sample Not necessarily representative Less likely to be generalizable Easier recruitment ECOLOGIC STUDIES Ecologic study Uses aggregate or summary data Additional characterization of a study This does not refer to ecology or ecosystems 40 Ecologic study characteristics Unit of observation = populations or groups of people Often uses pre-existing data Different reasons for data collection Timing of data collection Validity, reliability may be concern Individual measurements Aggregate measurements Presence of disease vs. not Prevalence Cause of death Cause-specific mortality rate Alcohol consumption Alcohol sales Cell phone use Hands-free device law presence Passive smoking exposure Smoking bans Personal income Median household income Ecologic fallacy A form of bias Group-level correlations might not be the same as individual-level correlations Cannot determine subtle or complicated relationships Main limitation of ecologic data Deaths from cholera per 100,000 houses, according to water supply company for the district, London, 1853 Deaths from Deaths per Water Supply No. of Houses Cholera 100,000 houses Southwark and 192 167,654 114 Vauxhall alone Southwark and Vauxhill, and Lambeth 182 301,149 60 Combined Lambeth alone 0 14,632 0 Etiology of Lung Cancer R. Doll, Adv Cancer Res, 1955 France: smoking, tax and lung cancer rates in young males, 1980–2004 Jha, 2009 Ecologic studies Typically observational and descriptive Most often combined with cross-sectional designs Not considered to be very strong evidence However, can make strong contributions Hypothesis generating Some exposures are population-level