Epidemiology and Sex

Epidemiology and Sex(ually Transmitted Diseases): The Basics Willard Cates, Jr., MD, MPH Family Health International Principles of STD/HIV Research University of Washington Seattle, Washington July 22, 2002 Objectives • To understand basic definitions of epidemiology •To describe components of descriptive, observational and experimental epidemiology •To know advantages and disadvantages of case-control and cohort designs Cholera in London, 1854 The Etymology of Epidemiology epi = upon demos logy = = people study of e.g. Population Level Science Definition of Epidemiology The study of the distribution and determinants of disease and health in human populations. Stedman’s Epidemiology The science of making the obvious obscure Clinician Epidemiology I0 = (480)(2)/106/yr (9.1 x 0.955) + 0.045 The science of long division Statistician Epidemiology The worst-taught course in medical school Anonymous Med Student Epidemiology The study of skin diseases Atlanta Native Epidemiology’s Fundamental Axioms • Non-randomness • Etiologic thinking • Preventability Epidemiology • Quantitative basic science • Method of causal reasoning • Vehicle for clinical and public health action W. Cates, 1982 Exposure Variable – “E” • Characteristic of interest • Risk factor • Predictor variable • Independent variable • Putative causal factor Outcome or Disease Variable – “D” • Health event of interest • Illness, injury, infection • Response variable • Dependent variable • Effect variable E-D Relationships – STD Examples • Gonorrhea • PID • Age • HPV – – – – PID ectopic pregnancy chlamydia infection cervical cancer • Alcohol – high risk behavior HIV infection • Circumcision – Categories of Study Design • Descriptive • Analytic • Experimental Knowledge Continuum Less More Most  Descriptive  Analytic   Experimental • Search for clues • Clues available Descriptive Studies • Patterns of occurrence • No comparison group • Generate hypotheses about E-D relationships Descriptive Studies: Examples from STD • Epidemiology of chlamydia in Norway • Prevalence of sexual behaviors among a sample of the general population • Trends in the first 20 years of AIDS in the US Analytic Studies • Test hypotheses about E-D relationships • Three main types: – Cohort – Case-control – Cross-sectional Cohort Studies - Overview • Subjects selected on basis of E • Directionality always forward – E  D • Timing – Prospective: “real time” – Retrospective: “historical time” Cohort Studies – Flow Chart Study Group E+ Source Population Study Group E D+ D D+ D Cohort Studies: Major Advantages • Logical temporal sequence • Can measure incidence of D • Well-suited for rare E • Can study many effects of one E Cohort Studies: Major Disadvantages • Many subjects needed for rare D • Follow-up: logistics, losses • Prone to selection bias • Prospective: time-consuming, costly, observation can influence behaviors • Retrospective: suitable records Case-Control Studies: Overview • Subjects selected on basis of D • Directionality is backward –DE Case-Control Studies – Flow Chart Cases D+ E+ E Source Population Controls D E+ E Case-Control Studies: Major Advantages • Quick and inexpensive • Can study multiple E • Well-suited for rare D and D with long latency • Requires fewer subjects at entry Case-Control Studies: Major Disadvantages • Design “backward” • Unsuitable for rare E • Usually cannot measure D incidence • Temporal E-D uncertainty • Prone to selection and recall bias Study Bias A Further Look • Selection bias: differential selection of participants on the basis of E or D • Information bias: differential collection or classification of E or D among participants – Recall bias: differential recall of E among cases and controls Experimental Studies (1) • Assign E randomly, follow for D • If placebo, blinding possible • Types: – Clinical trial – Community trial Experimental Studies (2) • Rolls Royce! • Randomization controls extraneous variables, both known and unknown • Limitations: ethical concerns, cost, length, not feasible for rare D Observational vs. Experimental: A Tale of 2 Studies Study Zekeng, 1993 Roddy, 1998 Design Observational/Cohort Experimental • Both were conducted in the same network of Cameroon sex workers • Both examined use of N-9 and HIV acquisition Initial Analysis of Observational Study Rate of HIV (per 100 women years) Inconsistent users Consistent users 16.3 3.5 Rate Ratio 0.2 (0.1 – 0.7) Source: Zekeng (1993) Reanalysis of Observational Study Data Source: Coital diaries from sex workers Measure: Efficacy per sexual episode Result: Condoms 92% (79-100%) N-9 Suppositories 100% (43-100%) Sources: Zekeng (1993), Wittkowski (1998) Observational Analysis of the 2 N-9 Studies Zekeng Analysis Observational 16.3 3.5 Rates of HIV: Inconsistent use Consistent use Rate Ratios Roddy Observational 15.6 5.0 0.3(0.1-0.7) 0.2 (0.1-0.7) Sources: Zekeng (1993), Roddy (1998) Observational vs. Experimental Analysis, Same N-9 Study Roddy Analysis Rates of HIV: Inconsistent use Consistent use 15.6 5.0 Placebo 4.3 N-9 5.3 Roddy Experimental Observational Rate Ratios 0.3 (0.1-0.7) 1.2(0.7-2.1) Source: Roddy (1998) Study Design: Concluding Remarks • Must consider: – – – – Objectives of study Current knowledge about E-D Ethical issues Time, money, human resources • Different approaches • Flexibility and creativity are KEY! Uses of Epidemiology: Levels of Evidence Quality of Evidence I. Good evidence - large RCT, - primary outcomes Strength of Recommendation A. Stronger - important benefits, - broadly applicable Weaker - smaller benefit, - limited generalizability Insufficient evidence - expert opinion II. B. Fair evidence - observational studies, - surrogate outcomes Weak evidence - anecdotes, - expert opinion C. III.