Measures of Effect: An Introduction Epidemiology Supercourse Astana, July 2012 Philip la Fleur, RPh MSc(Epidem) Deputy Director, Center for Life Sciences

Come to Ottawa, Canada and get “Out and About”

Canadian Agency for Drugs and Technologies in Health ( Emerg Med J 2003;20:

Objectives Understand how to calculate and interpret and articulate measures of effect and morbidity 1.2x2 Table Risk Odds Relative Risk Odds Ratio 2.Relative Risk Reduction and Absolute Risk Reduction 3.Number needed to Treat and Number needed to Harm

Summary Outcome Total ExposureYesNo Yesaba + b Nocdc + d Totala + c b + dN

1. 2x2 Table Outcome Total ExposureYesNo Yesaba + b Nocdc + d Totala + c b + dN

Risk Probability that an event will occur (Last 2001) E.g. that a person will die within one year Risk in Exposed = a/(a+b) Risk in unexposed, “Baseline risk” = c/(c+d) Outcome Total ExposureYesNo Yesaba + b Nocdc + d Totala + c b + dN

Odds The ratio of the probability of occurrence of an event to that of non-occurrence E.g. odds of smokers developing a chronic cough Odds in Exposed = a/b Odds in unexposed, “Baseline odds” = c/d Outcome Total ExposureYesNo Yesaba + b Nocdc + d Totala + c b + dN

Risk Versus Odds RiskOdds Conversion: Odds = Risk/(1-Risk) Risk = Odds / (1 + Odds) ⌂⌂⌂⌂/⌂ ⌂⌂/⌂ ⌂/⌂ ⌂/⌂⌂⌂⌂ ⌂/⌂⌂⌂⌂⌂⌂⌂⌂⌂

Would you swim here?

Develop a Question: PICO P opulation: children under 5 years of age I ntervention (exposure): Swimming in the Ishim River C omparator (control): Not swimming in the Ishim river O utcome: Otitis Media Question: What is the risk that a child under 5 will develop an ear infection after swimming in the Ishim river?

Odds Ratio Outcome Total ExposureYesNo Yesaba + b Nocdc + d Totala + c b + dN

Odds Ratio Odds of getting infection for swimmers: 40/60 = 0.67 Ear Infection Total Swimming in Ishim YesNo Yes No Total Odds of getting infection for non-swimmers: 5/95 = 0.052

Relative Risk Outcome Total ExposureYesNo Yesaba + b Nocdc + d Totala + c b + dN

Relative Risk Risk of getting infection for swimmers: 40/100 = 0.4 Ear Infection Total Swimming in Ishim YesNo Yes No Total Risk of getting infection for non-swimmers: 5/100 = 0.05

OR versus RR Key Messages Odds and Odds Ratios are difficult to conceptualize but statisticians prefer them in some situations because of their mathematical properties Odds Ratios always exaggerate the relative risk, but when baseline risk is low (e.g. <10%), the OR approximates the relative risk Relative Risk is a more intuitive measure and is becoming more common in medical literature

Objectives Understand how to calculate and interpret and articulate measures of effect and morbidity 1.2x2 Table Risk Odds Relative Risk Odds Ratio 2.Relative Risk Reduction and Absolute Risk Reduction 3.Number needed to Treat and Number needed to Harm

2. Relative Risk Reduction and Absolute Risk Reduction Objectives Learn how to interpret risk of events in the control (baseline group) and intervention groups (treatment group) from published studies. Understand the concepts of relative risk reduction and absolute risk reduction and how they usually differ from one population to another.

Trial 1: High Risk Patients – New drug for acute myocardial infarction to reduce mortality – First studied in a high risk population: 40% mortality at 30 days among untreated e.g., elderly, heart failure, anterior wall infarction Ref:

Trial 1: High Risk Patients – New drug for acute myocardial infarction to reduce mortality – First studied in a high risk population: 40% mortality at 30 days among untreated e.g., elderly, heart failure, anterior wall infarction 30% mortality among treated – How would you describe the effect of the new drug? Ref:

Trial 1: High Risk Patients Ref:

Trial 2: Low Risk Patients – New drug for acute myocardial infarction to reduce mortality – Later studied in a lower risk population: 10% mortality at 30 days among untreated e.g., younger, uncomplicated inferior wall infarction Ref:

Trial 2: Low Risk Patients – New drug for acute myocardial infarction to reduce mortality – Later studied in a lower risk population: 10% mortality at 30 days among untreated e.g., younger, uncomplicated inferior wall infarction 7.5% mortality among treated – How would you describe the effect of the new drug? Ref:

Trial 2: Low Risk Patients Ref:

Summary Points for Relative Risk Reduction and Risk Difference Relative risk reduction is often more impressive than absolute risk reduction. The lower the risk in the control group, the larger the difference between relative risk reduction and absolute risk reduction.

Objectives Understand how to calculate and interpret and articulate measures of effect and morbidity 1.2x2 Table Risk Odds Relative Risk Odds Ratio 2.Relative Risk Reduction and Absolute Risk Reduction 3.Number needed to Treat and Number needed to Harm

3. Number Needed to Treat Number Needed to Harm Objectives Learn how to calculate Number Needed to Treat (NNT) from an estimate of risk difference. Increase awareness of the range of NNTs associated with common interventions.

Definitions Number Needed to Treat (NNT): – Number of persons who would have to receive an intervention for 1 to benefit. Number Needed to Harm(NNH): – Number of persons who would have to receive an intervention for 1 to be experience a adverse event.

Calculating NNT If a disease has a mortality of 100% without treatment and therapy reduces that mortality to 50%, how many people would you need to treat to prevent 1 death?

Estimate NNT CER%EER%ARR%NNT How many 60-year-old patients with mild hypertension would you have to treat with diuretics for 5 years to prevent 1 stroke? How many people with myocardial infarction would you have to treat with ß- blockers for 2 years to prevent 1 death? How many people with acute myocardial infarction would you have to treat with streptokinase to prevent 1 person from dying in the next 5 weeks? Ref:

Estimate NNT CER%EER%ARR%NNT How many 60-year-old patients with hypertension would you have to treat with diuretics for 5 years to prevent 1 death? How many people with myocardial infarction would you have to treat with ß-blockers for 2 years to prevent 1 death? How many people with acute myocardial infarction would you have to treat with streptokinase to prevent 1 person from dying in the next 5 weeks? Ref:

Estimate NNT CER%EER%ARR%NNT How many 60-year-old patients with hypertension would you have to treat with diuretics for 5 years to prevent 1 death? How many people with myocardial infarction would you have to treat with ß- blockers for 2 years to prevent 1 death? How many people with acute myocardial infarction would you have to treat with streptokinase to prevent 1 person from dying in the next 5 weeks? Ref:

Calculation NNT= 100/ARR (where ARR is %) or NNT= 1/ARR (where ARR is proportion) NNH= 100/ARI (where ARI is %) Or NNH = 1/ARI (where ARI is proportion)

NNTs from Controlled Trials Control Event % Treatment Event % Absolute Risk Reductio n% NNT Population: hypertensive 60-year-olds Therapy: oral diuretics Outcome: stroke over 5 years Population: myocardial infarction Therapy: ß-blockers Outcome: death over 2 years Population: acute myocardial infarction Therapy: streptokinase (thrombolytic) Outcome: death over 5 weeks Ref:

Population: hypertensive 60-year-olds Outcome: stroke over 5 years Depiction of Results in Control Group Ref:

Population: hypertensive 60-year-olds Outcome: stroke over 5 years Depiction of Results in Treatment Group Ref:

Bottom Line It is easy to mis-estimate baseline risk and effects of therapy NNT is easily calculated from the absolute risk reduction (ARR) Awareness of threshold NNT can help anticipate the risk reduction to look for in a therapy.

Summary Outcome Total ExposureYesNo Yesaba + b Nocdc + d Totala + c b + dN

References/Slide Sources 1.Last JM. A Dictionary of Epidemiology, 4th ed. Oxford University Press, Guyatt G et al. Users’ Guides to the Medical Literature, 2nd ed. McGraw Hill, Guyatt G. Tips for Teachers of Evidence Based Medicine. Lecture on Odds and Risk. 4.Grimes and Schulz. Making Sense of Odds and Odds Ratios. Obs & Gyn 2008(111): Some Slides for Risk Reduction and NNT are from: Alexandra Barratt, Peter C. Wyer, Rose Hatala, Thomas McGinn, Antonio L. Dans, Sheri Keitz, Virginia Moyer, Gordon Guyatt, Robert Hayward, for the EBM Teaching Tips Working Group ( 6.Smiley Diagrams from: Dr. Chris Cates EBM Website: