Understanding-Odds-Ratios-in-Medical-Research.pptx

In regression analysis, our goal is to predict or infer an outcome variable (also known as the dependent variable) using one or more independent variables.

1. Understanding Odds Ratios in Medical Research A comprehensive guide to interpreting, calculating, and applying odds ratios in clinical studies

2. What is an Odds Ratio? An odds ratio (OR) is a measure of association between an exposure and an outcome. It represents the odds that an outcome will occur given a particular exposure, compared to the odds of the outcome occurring in the absence of that exposure. Primary Uses Most commonly used in case-control studies, but can also be applied in cross-sectional and cohort study designs (with modifications or assumptions). In Logistic Regression The exponential function of the regression coefficient (eb1) is the odds ratio associated with a one-unit increase in the exposure.

3. Interpreting Odds Ratios OR = 1 Exposure does not affect odds of outcome OR > 1 Exposure associated with higher odds of outcome OR < 1 Exposure associated with lower odds of outcome Odds ratios are used to compare the relative odds of an outcome (such as disease or disorder) given exposure to a variable of interest (such as health characteristic or medical history). ORs help determine whether a particular exposure is a risk factor for a particular outcome, and allow comparison of the magnitude of various risk factors.

4. Confidence Intervals and Statistical Significance The 95% confidence interval (CI) estimates the precision of the OR: Large CI indicates low precision Small CI indicates higher precision Unlike p-values, the 95% CI doesn't directly report statistical significance, though it's often used as a proxy if it doesn't overlap the null value (OR=1). Important Note: An OR with 95% CI that spans the null value (1.0) should not be interpreted as evidence for lack of association between exposure and outcome. Statistical significance is typically determined by the p-value, not solely by the confidence interval.

5. Understanding Confounding Confounding occurs when a non-causal association between exposure and outcome results from the influence of a third variable (the confounding variable). Confounding Variable Causally associated with the outcome of interest Relationship Non-causally or causally associated with the exposure Not Intermediate Not in the causal pathway between exposure and outcome Stratification and multiple regression techniques are two methods used to address confounding and produce "adjusted" ORs.

6. Calculating Odds Ratios: Two-by-Two Frequency Table Where: a = Number of exposed cases b = Number of exposed non-cases c = Number of unexposed cases d = Number of unexposed non-cases The formula for calculating the odds ratio is:

7. Example: Suicidal Behavior Study Using data from Greenfield et al. (2008) on previously suicidal adolescents (n=263): Study Data: 186 adolescents: non-suicidal at follow-up 77 adolescents: persistent suicidal behavior 45 suicidal adolescents had depression at baseline 86 non-suicidal adolescents had depression at baseline Calculating Values: a = 45 (exposed cases) b = 86 (exposed non-cases) c = 32 (unexposed cases) d = 100 (unexposed non-cases) OR = 45/32 ÷ 86/100 = 1.63 The odds of persistent suicidal behavior is 1.63 times higher given baseline depression diagnosis compared to no baseline depression.

8. Calculating Confidence Intervals For the previous example, confidence intervals are calculated using: Plugging in our values: Upper 95% CI = 2.80 Lower 95% CI = 0.96 Since the 95% CI (0.96 to 2.80) spans 1.0, the increased odds (OR 1.63) of persistent suicidal behavior among adolescents with depression at baseline does not reach statistical significance (p=0.07).

9. Key Insights from the Example Statistical Significance A positive OR does not necessarily indicate statistical significance. Consider confidence intervals and p-values to determine significance. Context Matters While depression is strongly linked to suicide in general literature, in this specific sample (with its particular size, composition, and other variables), the association was not significant. Comparative Risk Borderline personality disorder showed a stronger association (OR 3.8, 95% CI: 1.6–8.7) with persistent suicidal behavior than depression, and was statistically significant (p=0.002).

10. Summary and Resources Understanding odds ratios—how they're calculated, what they mean, and how to compare them—is crucial for interpreting scientific research in medicine and epidemiology. Key Takeaways: ORs measure association between exposure and outcome Interpretation depends on whether OR is <1, =1, or >1 Confidence intervals indicate precision Statistical significance requires consideration of CIs and p-values Confounding variables must be addressed Additional Resources: For more information about odds ratios and other statistics used in medicine, visit the Centre for Statistics in Medicine at Oxford University: http://www.csm-oxford.org.uk/index.aspx?o=1292 This site provides access to "Statistics Notes" published in BMJ by Doug Altman, Martin Bland, and others.