Biosta's'cs Board Review Parul Chaudhri, DO Family Medicine Faculty Development Fellow, UPMC St Margaret March 5, 2016
Review key biosta's'cs concepts Understand 2 X 2 tables
Objec'ves By the end of this session, ac've par'cipants will be able to: Iden'fy key biosta's'cs concepts Break down the ques'on into 2 X 2 tables Apply the key concepts to solve problems
It s all about the 2X2 Associa'on between Exposure and Outcome YES DISEASE NO EXPOSURE RR= OR= A/ (A+B) C/ (C+D) A/B = AD C/D BC
YES DISEASE NO TEST SpIN Sn TP (TP+ FN) Sp TN (TN+ FP) SnOUT
Which one of the following reflects the percentage of pa'ents with a disease who have a posi've test for the disease in ques'on? 1. Likelihood ra'o 2. Sensi'vity 3. Specificity 4. Posi've predic've value 5. Nega've predic've value
In a study to evaluate a test as a screen for the presence of a disease, 235 of the 250 people with the disease had a posi've test and 600 of the 680 people without the disease had a nega've test. Based on this data, the specificity of the test for the disease is 1. 235/250 = 94% 2. 15/250 = 6% 3. 600/680 = 88% 4. 80/680 = 12% 5. 15/80 = 19%
YES DISEASE NO TEST Sn TP (TP+ FN) Sp 600 (80+600)
YES DISEASE NO TEST PPV TP (TP+ FP) NPV TN (TN+ FN) +LR = -LR = Sn (1- Sp) (1- Sn) Sp
A study finds that PPV of a new test for breast cancer is 75%, which means 1. If 100 pa'ents with breast CA have the test, 75 (75%) will have a + result 2. If 100 pa'ents without breast CA have the test, 75 (75%) will have a test 3. 75% of pa'ents who test + actually have breast CA 4. 75% of pa'ents who test don t have breast CA
A home urine test is designed to detect a type of cancer. The gold standard for this cancer is a biopsy. The biopsy is more costly, invasive, and associated with lots of adverse side effects. To test the effec'veness of the home urine test, 104 people took the test and then agreed to a biopsy. When the study was concluded, 77 people tested nega've and 27 tested posi've on the urine test. Biopsies were posi've in 18 individuals, 8 of whom tested nega've on the urine test. What is the NPV of the home urine test, rounded to a whole number? 1. 20% 2. 37% 3. 56% 4. 80% 5. 90%
YES DISEASE NO TEST NPV= 69/ 77= 89.6 %
A 69 year-old female with postmenopausal bleeding. You consider whether to do a vaginal US to assess the thickness of her endometrium. In evalua'ng the usefulness of this test to either support or exclude a diagnosis of endometrial cancer, which one of the following sta's'cs is most useful? 1. Likelihood ra'o 2. Number needed to treat 3. Prevalence 4. Incidence 5. Rela've risk
Other Terms Prevalence is the existence of a disease in the current popula'on Incidence describes the occurrence of new cases of disease in a popula'on over a defined 'me period Rela've risk is the risk of an event in the experimental group versus the control group in a clinical trial The number needed to treat is useful for evalua'ng data regarding treatments, not diagnosis
It s all about the 2X2 Associa'on between Exposure and Outcome YES DISEASE NO TEST Type II error Type I error Power= 1- Type II error
The results of a given study are reported as achieving significance at a p-value of <0.05 (the 5% level). True statements about this finding include which one of the following? 1. 5% likelihood of the results having occurred by chance alone 2. If the study were replicated 100 'mes, 95 studies would repeat this finding and 5 would not 3. The confidence interval is 0%-10% 4. The null hypothesis has a 5% chance of being true 5. The β (type II) error is <5
A 95 % Confidence Interval Means 1. At least 95% of pa'ents with a disease have a posi've test for that disease 2. At least 95% of pa'ents without a disease have a nega've test for the disease 3. There is a 95% difference in risk between the treatment and control groups 4. It is 95% certain that the true value lies within the given range 5. At least 95% of the pa'ents need to receive an interven'on instead of the alterna've in order for one addi'onal pa'ent to benefit
95% confidence interval There is 95% certainty that the true value lies within the given interval range. P-value The probability of obtaining a result equal to or "more extreme" than what was actually observed, assuming that the null hypothesis is true OR Likelihood of achieving that result by chance alone
When a screening test iden'fies a cancer earlier, thereby increasing the 'me between diagnosis and death without prolonging life, this is called 1. Length-'me bias 2. Lead-'me bias 3. False-posi've screening test 4. Increasing the posi've predic've value of the screening test 5. Alributable risk
Bias Lead-Tme bias is when a screening test iden'fies a cancer earlier, thereby increasing the 'me between diagnosis and death without actually prolonging life. Length-7me bias is when a screening test finds a dispropor'onate number of cases of slowly progressive disease and misses the aggressive cases, thereby leading to an overes'mate of the effec'veness of the screening. AVributable risk is the amount of difference in risk for a disease that can be accounted for by a specific risk factor.
Results of a clinical study show a rela've risk reduc'on (RRR) of 33% and an absolute risk reduc'on (ARR) of 20%. There are 1000 pa'ents each in the treatment and control groups. To help determine the poten'al benefit of the treatment it is necessary to iden'fy the number needed to treat (NNT). Which one of the following is the NNT for this clinical study? 1. 3 2. 5 3. 13 4. 130 5. The number cannot be determined from the informa'on provided
Risk Reduc'on and Number Needed to Treat Number Needed to Treat (NNT): number of pa'ents necessary to treat in order for one pa'ent to benefit. Absolute Risk Reduc'on (ARR): Absolute adverse event rate for placebo minus the absolute adverse event rate for treated pa'ents Rela've risk reduc'on (RRR): omen quoted in the press or by those promo'ng a treatment, can be misleading to both the general public and to physicians. NNT= 1/ARR
Ques'ons
References: Fletcher RW, Fletcher SW: Epidemiology: The Essen1als, ed 4. Lippincol Williams & Wilkins, 2005 hlp://www.medpagetoday.com/lib/content/ Medpage-Guide-to-Biosta's'cs.pdf hlp://www.musc.edu/dc/icrebm/2x2table.html
Thank You Dr. Stephen Wilson UPMC St Margaret Faculty Development Fellows