62Ana-Maria ŠimundićINTRODUCTIONDiagnostic accuracy of any diagnostic procedure ora test gives us an answer to the following question:"How well this test discriminates between certain twoconditions of interest (health <strong>and</strong> disease; two stages ofa disease etc.)?". This discriminative ability can bequantified by the measures of diagnostic accuracy:• sensitivity <strong>and</strong> specificity,• positive <strong>and</strong> negative predictive values (PPV,NPV),• likelihood ratio,• the area under the ROC curve (AUC),• Youden's index,• diagnostic odds ratio (DOR).Different measures of diagnostic accuracy relate tothe different aspects of diagnostic procedure. Somemeasures are used to assess the discriminative propertyof the test, others are used to assess its predictive ability[1]. While discriminative measures are mostly usedby health policy decisions, predictive measures aremost useful in predicting the probability of a disease inan individual [2]. Furthermore, it should be noted thatmeasures of a test performance are not fixed indicatorsof a test quality <strong>and</strong> performance. Measures of diagnosticaccuracy are very sensitive to the characteristicsof the population in which the test accuracy is evaluated.Some measures largely depend on the diseaseprevalence, while others are highly sensitive to thespectrum of the disease in the study population. It istherefore of utmost importance to know how to interpretthem as well as when <strong>and</strong> under what conditions touse them.SENSITIVITY AND SPECIFICITYA perfect diagnostic procedure has the potential tocompletely discriminate between subjects with <strong>and</strong>without disease. Values of a perfect test which areabove the cut-off always indicate the disease, while thevalues below the cut-off always exclude the disease.Unfortunately, such a perfect test does not exist in reallife <strong>and</strong> therefore diagnostic procedures can make onlypartial distinction between subjects with <strong>and</strong> withoutdisease. Values above the cut-off are not always indicativeof a disease since subjects without disease canalso sometimes have elevated values. Such elevatedvalues of a certain parameter of interest are called falsepositive values (FP). On the other h<strong>and</strong>, values belowthe cut-off are mainly found in subjects without disease.However, some subjects with the disease can havethem too. Those values are false negative values (FN).Therefore, the cut-off divides the population of examinedsubjects with <strong>and</strong> without disease into four subgroupsconsidering values of a parameter of interest:• true positive (TP) –subjects with the disease withthe value of a parameter of interest above the cutoff,• false positive (FP) –subjects without the diseasewith the value of a parameter of interest above thecut-off,• true negative (TN) –subjects without the diseasewith the value of a parameter of interest below thecut-off,• false negative (FN) –subjects with the disease withthe value of a parameter of interest below the cutoff.The first step in the calculation of sensitivity <strong>and</strong>specificity is to make a 2x2 table with groups of subjectsdivided according to a gold st<strong>and</strong>ard or (referencemethod) in columns, <strong>and</strong> categories according to test inrows (Table I).Table I. 2x2 tableSubjects with the disease Subjects without the diseasepositive TP FPnegative FN TNSensitivity is expressed in percentage <strong>and</strong> definesthe proportion of true positive subjects with the diseasein a total group of subjects with the disease(TP/TP+FN). Actually, sensitivity is defined as theprobability of getting a positive test result in subjectswith the disease (T+|B+). Hence, it relates to the potentialof a test to recognise subjects with the disease.Specificity is a measure of a diagnostic test accuracy,complementary to sensitivity. It is defined asa proportion of subjects without the disease with negativetest result in total of subjects without disease(TN/TN+FP). In other words, specificity represents theprobability of a negative test result in a subject withoutthe disease (T-|B-). Therefore, we can postulate thatspecificity relates to the aspect of diagnostic accuracythat describes the test’s ability to recognise subjectswithout the disease, i.e. to exclude the condition ofinterest.Neither sensitivity nor specificity are influenced bythe disease prevalence, meaning that results from onestudy could easily be transferred to some other setting
Measures of diagnostic accuracy: basic definitions 63with a different prevalence of the disease in the population.Nonetheless, sensitivity <strong>and</strong> specificity can varygreatly depending on the spectrum of the disease in thestudy group.PREDICTIVE VALUESPositive predictive value (PPV) defines the probabilityof having the state/disease of interest in a subjectwith positive result (B+|T+). Therefore PPV representsa proportion of patients with positive test result intotal of subjects with positive result (TP/TP+FP).Negative predictive value (NPV) describes theprobability of not having a disease in a subject with anegative test result (B-|T-). NPV is defined as a proportionof subjects without the disease with a negative testresult in total of subjects with negative test results(TN/TN+FN).Unlike sensitivity <strong>and</strong> specificity, predictive valuesare largely dependent on disease prevalence in examinedpopulation. Therefore, predictive values from onestudy should not be transferred to some other settingwith a different prevalence of the disease in the population.Prevalence affects PPV <strong>and</strong> NPV differently. PPVincreases, while NPV decreases with the increase ofthe prevalence of the disease in a population. Whereasthe change in PPV is rather substantial, NPV is somewhatless influenced by the disease prevalence.LIKELIHOOD RATIO (LR)Likelihood ratio is a very useful measure of diagnosticaccuracy. It is defined as the ratio of expectedtest result in subjects with a certain state/disease to thesubjects without the disease. As such, LR directly linksthe pre-test <strong>and</strong> post-test probability of a disease in aspecific patient [3]. Simplified, LR tells us how manytimes more likely particular test result is in subjectswith the disease than in those without disease. Whenboth probabilities are equal, such test is of no value <strong>and</strong>its LR = 1.Likelihood ratio for positive test results (LR+) tellsus how much more likely the positive test result is tooccur in subjects with the disease compared to thosewithout the disease (LR+=(T+│B+)/(T+│B-)). LR+ isusually higher than 1 because it is more likely that thepositive test result will occur in subjects with the diseasethan in subject without the disease. LR+ can besimply calculated according to the following formula:LR+ = sensitivity / (1-specificity).LR+ is the best indicator for ruling-in a diagnosis.The higher the LR+ the more indicative of a diseasethe test is. Good diagnostic tests have LR+ > 10 <strong>and</strong>their positive result has a significant contribution to thediagnosis.Likelihood ratio for negative test result (LR-)represents the ratio of the probability that a negativeresult will occur in subjects with the disease to theprobability that the same result will occur in subjectswithout the disease. Therefore, LR- tells us how muchless likely the negative test result is to occur in a patientthan in a subject without disease. (LR-=(T-│B+)/(T-│B-)). LR- is usually less than 1 because it isless likely that negative test result occurs in subjectswith than in subjects without disease. LR- is calculatedaccording to the following formula: LR- = (1-sensitivity) / specificity.LR- is a good indicator for ruling-out a diagnosis.Good diagnostic tests have LR- < 0,1. The lower theLR- the more significant contribution of the test is inruling-out, i.e. in lowering the posterior probability ofthe subject having the disease.Since both specificity <strong>and</strong> sensitivity are used tocalculate the likelihood ratio, it is clear that neitherLR+ nor LR- depend on the disease prevalence inexamined groups. Consequently, the likelihood ratiosfrom one study are applicable to some other clinicalsetting, as long as the definition of the disease is notchanged. If the way of defining the disease varies, noneof the calculated measures will apply in some otherclinical context.ROC CURVEThere is a pair of diagnostic sensitivity <strong>and</strong> specificityvalues for every individual cut-off. To constructa ROC graph, we plot these pairs of values on thegraph with the 1-specificity on the x-axis <strong>and</strong> sensitivityon the y-axis (Figure 1).Fig. 1. ROC curve
- Page 8:
8Wojciech J. Baranowskirating) move
- Page 13 and 14: The benefits resulting from introdu
- Page 15 and 16: The benefits resulting from introdu
- Page 17 and 18: Medical and Biological Sciences, 20
- Page 19 and 20: The history and the present of hern
- Page 21 and 22: The history and the present of hern
- Page 23: The history and the present of hern
- Page 26: 26Anna Budzyńska et al.Gram-dodatn
- Page 32 and 33: 32Piotr Kamiński et al.Streszczeni
- Page 34 and 35: 34Piotr Kamiński et al.both in Pom
- Page 36 and 37: 36Piotr Kamiński et al.hemoglobin
- Page 39 and 40: Medical and Biological Sciences, 20
- Page 41 and 42: Impact of mandatory vaccination pro
- Page 43 and 44: Medical and Biological Sciences, 20
- Page 45 and 46: Bone turnover during pregnancy 45ra
- Page 47: Bone turnover during pregnancy 47pl
- Page 50 and 51: 50Jan Styczyński, Anna Jaworska(p
- Page 52 and 53: 52Jan Styczyński, Anna JaworskaRES
- Page 55 and 56: Medical and Biological Sciences, 20
- Page 57 and 58: Analysis of immunophenotype at seco
- Page 59: Analysis of immunophenotype at seco
- Page 64 and 65: 64Ana-Maria ŠimundićThe shape of
- Page 67 and 68: Medical and Biological Sciences, 20
- Page 69 and 70: Quantitative anatomy of the aortic
- Page 71 and 72: Quantitative anatomy of the aortic
- Page 73 and 74: Medical and Biological Sciences, 20
- Page 75 and 76: Volumetric growth of various aortic
- Page 77 and 78: Volumetric growth of various aortic
- Page 79 and 80: Medical and Biological Sciences, 20
- Page 81 and 82: Effect of Low Level Laser Therapy a
- Page 83 and 84: Effect of Low Level Laser Therapy a
- Page 85 and 86: Medical and Biological Sciences, 20
- Page 87 and 88: Body weight support during treadmil
- Page 89 and 90: Body weight support during treadmil
- Page 91 and 92: Medical and Biological Sciences, 20