Statistics Lecture 11: Evidence, Epidemiology, Odds & Risk

Lecture 11: Evidence, Epidemiology, Odds & Risk What is Evidence? -​ Evidence in its broadest sense includes everything that is used to determine or demonstrate the truth of assertion -​ Types of evidence: research studies, journal articles, books, practical experience and knowledge, intuition, common sense, tradition -​ Sources of evidence can be found almost anywhere Evidence from Research -​ Quantitative: -​ Have an hypothesis (an idea) -​ Randomized controlled trials -​ Cohort studies and case-control studies -​ Systematic reviews (meta-analysis) -​ Qualitative: -​ Interviews -​ Focus groups Randomized Controlled Trials (RCT) -​ They may also be referred to as true experiments. They are the gold standard of evidence -​ The researcher controls or manipulates the independent variable and randomly assigns subjects to different conditions/interventions Quasi Experimental Studies -​ They compare 2 groups, but without randomization -​ Because of the lack of randomization or control group, there is a higher chance of bias in the research and it is not best practice Hierarchy of Evidence -​ Applies only to quantitative research -​ Looks at study designs that have reduced bias and confounding -​ Epidemiology and Epidemiologic Designs -​ -​ Epidemiology is the study of the distribution and determinants of disease or illness in a specified population, and the application of knowledge for disease prevention and control The designs can either be descriptive (pattern and frequency) or inferential (determinants and association/comparison) -​ Descriptive: 1.​ Case Report a.​ A description of an event based on the experience of one patient only, highly detailed. 2.​ Case Series a.​ A study of a large group of patients with a particular health condition. Often used to understand the clinical picture of the disease b.​ Example: a report of a series of 5 cases on pneumocystis pneumonia that occurred in young previously healthy homosexual men in 3 hospitals over a 6-month period 3.​ Cross Sectional a.​ Examines the relationship between diseases and other variables of interest as they exist in a defined population at one particular time. b.​ Population selected without regard to exposure or disease status c.​ Example: investigators conducted a cross sectional study to determine the relationship between low back pain and sedentary work among crane operators, straddle carriers, and office workers. All 3 groups had sedentary jobs that required prolonged sitting -​ Inferential: 1.​ Case Control (retrospective) a.​ A study design that starts with the outcome of interest (the disease) and looks back to see if people were exposed to a potential causative agent b.​ Data from case control studies can be used to calculate prevalence, odds ratios, and attack rates 2.​ Cohort (prospective) a.​ A study design in which a group of individuals are followed prospectively to see who develops the outcome of interest (eg. disease) b.​ Data from cohort studies can be used to calculate incidence, relative risk, and attributable risk 3.​ Interventional Incidence and Prevalence -​ Incidence is the number of new events or cases during a specific period of time -​ Reported as an incidence rate. It considers the cases that already have the disease and then counts how many more happened after that time -​ Prevalence is all the people who have a specific disease at a specific point in time -​ Point prevalence: specific point in time -​ Period prevalence: for a specific period of time -​ Prevalence doesn’t look at when they developed the disease, just that they have it during that time section Odds Ratio and Relative Risk -​ OR and RR both measure association between outcome and predictor variables -​ OR is the ratio of 2 odds, whereas RR is the ratio of 2 possibilities -​ Odds Ratios (OR): -​ An OR is an estimate of risk of BEING sick not BECOMING sick -​ OR = odds the cases were exposed / odds that the controls were exposed -​ OR = AD/BC or (A/C) / (B/D) -​ OR = 3.0 → positive association -​ OR = 1.0 → no association -​ OR = 0.5 → negative association -​ Relative Risk (RR): -​ The ratio of the probability of an outcome in a group exposed and the probability of developing the same outcome in a group NOT exposed -​ RR = incidence rate in the exposed sample / incidence rate in the non-exposed sample -​ If the RR is greater than 1 the group that was exposed has a higher incidence rate than the group that was not -​ RR = (A/A+B)/(C/C+D) -​ RR > 1 means increased risk -​ RR < 1 means reduced risk -​ RR = 1 means no apparent risk -​ Don't forget to check if the RR is statistically significant by checking the p-value of the associated chi-square test OR check the confidence interval (if it contains 1 it is not significant) Absolute Risk (AR) -​ Can be used to calculate: -​ Absolute risk difference -​ Excess risk associated with exposure -​ Numbers needed to treat or harm -​ But if your AR is not statistically significant don’t bother doing further calculations with it -​ Absolute risk is the chance of an outcome occurring with no comparison group, whereas RR is the risk in comparison of one group to another -​ AR = the # of people who have the outcome / all those who could have had the outcome -​ This would be A/A+B or C/C+D -​ You can then calculate the risk difference by (A/A+B) - (C/C+D) Absolute Risk Difference, NNT, NNH, and Absolute Risk Reduction -​ The absolute risk difference is the absolute risk in the exposed group - the absolute risk in the unexposed group (A/A+B) - (C/C+D) -​ The risk difference can be used to calculate the number needed to treat (NNT) or number needed to harm (NNH) -​ NNT or NNH is calculated by using absolute risk reduction (ARR) -​ Formula for NNT: -​ NNT = 1 / ARR -​ ARR = risk in control group - risk in treatment group (A/B) - (C/D)