sumattana glangkarn6 Evaluation of proficiency of test Precision (or Repeatability or Reliability) is the ability of a measurement to give consistent results on repeated trials Validity (or Accuracy) is the ability of a measuring instrument to give a true value. Validity can be evaluated only if there exists an accepted and independent (gold standard) method for confirming the condition.
sumattana glangkarn7 SENSITIVITY and SPECIFICITY Sensitivity is the ability of a test to give positive results in a group of persons with the disease (True positive) Specificity is the ability of a test to give negative results in a group of persons without the disease (True negative)
sumattana glangkarn8 Screening test
sumattana glangkarn9 SENSITIVITY and SPECIFICITY Sensitivity = TPx 100 = a x 100 TP + FN a + c Specificity =TNx 100 = d x 100 FP + TN b + d
sumattana glangkarn10 ACCURACY Accuracy = TP + TN x 100 Grand total =a + b x 100 a + b + c + d
sumattana glangkarn11 Conditions that requires high sensitivity test The disease is fatal if missed. If it is detected at an early stage, the patients would have high probability of surviving, or getting cured. The disease has the high potential of spread to other people if not detected. The confirmatory test is available for those who have screened as positive.
sumattana glangkarn12 Conditions that requires high specificity test The false positive will give fatal impression for the persons screened. The disease is not yet detected by other method, or the diagnosis has to be done through more painful or more complicated methods such as liver biopsy.
sumattana glangkarn13 Example: A pregnancy test is administered to 100 pregnant women and 100 non-pregnant women, the result are shown:-
sumattana glangkarn15 Example: Three screening test, A, B and C were applied to 1,000 patients with diabetes mellitus (diagnosed on the basis of glucose tolerance tests) and to 3,000 persons free of diabetes. Test A yielded positive results in 900 diabetics and 1,200 non diabetics. Test B gave positive results in 600 diabetics and 300 non diabetics. Test C was positive in 850 diabetics and 450 non diabetics.
sumattana glangkarn16 Results of Test A
sumattana glangkarn17 Results of Test B
sumattana glangkarn18 Results of Test C
sumattana glangkarn19 Competency of the Tests
sumattana glangkarn20 Test A : Best when we want a highly sensitive test Test B : Best when we want a highly specificity test Test C : Most valid of the three because it has high both in sensitivity and specificity. Competency of the Tests
sumattana glangkarn21 Ability to detect unrecognized disease and estimation of number of cases in the population Predictive value หมายถึง ค่าคาด ทำนายผลของการเป็นโรคที่น่าจะ เกิดขึ้น ตามผลการทดสอบที่ได้ Positive predictive value : ค่าที่ แสดงถึงโอกาสของบุคคลซึ่งผลการ ทดสอบเป็นบวก จะป่วยเป็นโรคจริง เท่าไร Negative predictive value : ค่าที่ แสดงถึงโอกาสของบุคคลซึ่งผลการ ทดสอบเป็นลบ จะปราศจากโรคจริง เท่าไร
sumattana glangkarn22 PPV & NPV Positive predictive value (PPV) =ax 100 a + b Negative predictive value (NPV) =dx 100 c + d
sumattana glangkarn23 Example: A test with sensitivity 95% and specificity 95% is applied to a population 0f 10,000 with estimated prevalence of a specified disease 10% Find :1. Positive predictive value 2. Negative predictive value 3. Efficiency of the test 4. % False positive of the positive test (to be used for mass screening purpose)
sumattana glangkarn24 Calculation : Total population= 10,000 Prevalence of disease=10% Estimate sick persons=0.1 x 10,000 =1,000 (a + c) Estimate non-sick= 10,000 – 1,000 =9,000 (b + d)
sumattana glangkarn25 Calculation (Cont.): From sick persons: Sensitivity of the test=95% True positive persons=0.95 x 1,000 =950 (a) False negative persons= 1, =50 (c)
sumattana glangkarn26 Calculation (Cont.): From non sick persons: Specificity of the test=95% True negative persons=0.95 x 9,000 =8,550 (d) False positive persons= 9,000 – 8,550 =450 (b)
sumattana glangkarn27 Result of Analysis of a Screening test
sumattana glangkarn29 Accuracy of the test = ,550 x ,000 =95.0% False Positive of the positive test =450x 100 1,400 =32.1% ( ) Calculation (Cont.):
sumattana glangkarn30 Example: From the previous example, suppose that the prevalence of the specified disease in the study population is 50% Find :1. Positive predictive value 2. Negative predictive value 3. Efficiency of the test 4. False positive of the test
sumattana glangkarn31 Calculation : Total population= 10,000 Prevalence of disease=50% Estimate sick persons=0.5 x 10,000 =5,000 (a + c) Then non-sick persons= 10,000 – 5,000 =5,000 (b + d)
sumattana glangkarn32 Calculation (Cont.): From the group of sick persons: Sensitivity of the test=95% True positive persons=0.95 x 5,000 =4,750 (a) False negative persons= 5,000 – 4,750 =250 (c)
sumattana glangkarn33 Calculation (Cont.): From the group of non-sick persons: Specificity of the test=95% True negative persons=0.95 x 5,000 =4,750 (d) False positive persons= 5,000 – 4,750 =250 (b)
sumattana glangkarn34 Result of Analysis of a Screening test
sumattana glangkarn35 Positive predictive value = 4,750x 100 5,000 =95.0% Negative predictive value =4,750 x 100 5,000 = 95.0% Calculation (Cont.):
sumattana glangkarn36 Accuracy of the test= 4, ,750 x ,000 =95.0% False Positive of the positive test = 250x 100 5,000 =5% ( ) Calculation (Cont.):
sumattana glangkarn37 Screening and Prevalence When the same test is used for screening in populations with higher prevalence of disease, the lower false positives would be obtained.
sumattana glangkarn38 จงหาค่า Positive predictive value เมื่อ นำการทดสอบที่มีคุณสมบัติต่อไปนี้มาใช้ใน ชุมชนที่มีอัตราความชุกของโรคต่างกัน Test 1 มี Sensitivity 95% และ Specificity 95% Test 2 มี Sensitivity 98% และ Specificity 98% Prevalence (%) Positive predictive value Test 1Test 2 0.1?? 1.0?? 2.0?? 5.0?? 50.0??
sumattana glangkarn39 Test 1 มี Sensitivity 95% และ Specificity 95% Test 2 มี Sensitivity 98% และ Specificity 98% Prevalence (%) Positive predictive value Test 1Test
sumattana glangkarn41 Results of screening test with Sensitivity 95% and Specificity 99% used in populations with various levels of prevalence Prev (%) Test resul t DiseaseTotalPPVEffici ency total ?? 2+ - total ??
sumattana glangkarn42 Prev (%) Test resul t DiseaseTotalPPVEffici ency total ?? total ?? total ??
sumattana glangkarn43 Results of screening test with Sensitivity 95% and Specificity 99% used in populations with various levels of prevalence Prev (%) Test resul t DiseaseTotalPPVEffici ency total total
sumattana glangkarn44 Prev (%) Test resul t DiseaseTotalPPVEffici ency total total total
sumattana glangkarn45 Criteria for screening program The condition should be an important health problem. The screening test should be acceptable to the general population. The test should have minimal (if any) side effects. Validity and reliability must be satisfactory. There must be a good confirmatory test.
sumattana glangkarn46 Criteria for screening program There must be an effective treatment. There must be resources and facilities for treatment. It must give a reasonable yield. There must be the evidence that early detection and treatment reduce morbidity and mortality The expected benefit from the program exceed the risk of adverse effect from the screening.
sumattana glangkarn47 Type of error that can happen after making a decision True positive (Sensitivity) Correct decision False positive, Type II error ( ) Error of commission Proportion of well persons diagnosed as sick False negative, Type I error ( ) Error of omission Proportion of sick persons diagnosed as well True negative (Specificity) Correct decision