# Understanding Medical Tests And Test Results

Laboratory tests are not perfect and do some healthy people falsely as ill identify (false positive result), or diseased human disease free to report (false negative result). Whether a test can identify patients properly with a disease may depend on the probability with which already (a priori probability) it existed at the examinees, as well as the intrinsic characteristics of the test.

Test results can help to make (diagnostic tests) a diagnosis in symptomatic patients or to discover an occult disease in asymptomatic patients (screening tests). However, test results can interfere with clinical decision making when the test differs sufficiently between patients with or without disease when the result does not match the clinical picture or if the result is not properly integrated into the clinical context. Laboratory tests are not perfect and do some healthy people falsely as ill identify (false positive result), or diseased human disease free to report (false negative result). Whether a test can identify patients properly with a disease may depend on the probability with which already (a priori probability) it existed at the examinees, as well as the intrinsic characteristics of the test. Although diagnostic tests are often a critical factor in clinical decision making, testing can lead to undesirable or unintended consequences. Tests must be arranged carefully and with the expectation that the test result reduces ambiguity with regard to the findings of the patient and helps to improve their health. In addition to the risk of false results (and thus a delay in treatment initiation or execution of an unnecessary therapy), take laboratory tests important time and resources, and may themselves have side effects (eg. As a pneumothorax by lung biopsy) or lead to further unnecessary Testing. Definition of a positive test result Among the most common tests include those results along a continuous, quantitative scale supply (eg., Blood sugar, leukocytes). Such tests can provide useful clinical information within their areas, but often doctors use them to confirm or rule out a disease by using an established value or criterion as the limit (on a positive note here, from here negative). Such limits are usually set on the basis of statistical and conceptual analysis, the rate of false-positive results (and thus unnecessary, expensive and potentially dangerous tests or treatments initiating) and false negatives (and thus miss the correct diagnosis ) to improve. The identification of a limit also depends on a norm to identify the disease in question. Typically follow such quantitative test results (eg. As leukocyte count in suspected appendicitis) some type of distribution curve (not necessarily a normal curve, although often presented as such). The distribution of test results for patients with the disease is at a different point than that for patients without disease. Some patients with the disease have a very high or very low result, but most have a result on an average. Conversely, some healthy patients have very high or very low results, but most have a result on another average than patients with the disease. In most tests, the distributions overlap, so many of the possible test results occur in patients with and without disease. Such results can be displayed clearly if the curves are plotted on the same graph (distribution of the test results.). Some patients above and below the selected limit are mislabelled. more patients to identify with the disease adjusting a limit value (increasing the test sensitivity) increases the number of false positives (low specificity), while moving the limit value in the other direction to define no louder healthy patients as sick, the number of false negative results increased. Each limit is associated with a certain probability of true-positive and false-positive results. Distribution of the test results. Ill patients are displayed in the upper distribution, healthy patients in the lower distribution. In patients with disease of the area below the earnings distribution right of the limit with the true-positive test rate corresponds (this provides information about the sensitivity of the test), while the region corresponds to the left from the limit with the false-negative test rate. For patients without disease of the area to the right (h d.. Its specificity) corresponds to the limit of the false-negative rate and the area on the left side corresponds to the right-negative rate. When two overlapping distributions (z. B. patients with and without disease) refers to the shift of the threshold value, the sensitivity and specificity, but in opposite directions. When you change the limit from 1 to 2, the number of false-negative results from (the sensitivity is increased), but this also increases the number of false alarms (decreased specificity). Receiver operating characteristic (ROC) curves The graphical representation of the proportion of true positives (number of true-positive results or the number of patients) versus the amount of false-positive results (number of false alarms or number without disease) for a set of limit values ??generates a curve called ROC curve. The ROC curve is a graph showing the intersection between the sensitivity and the specificity, if the limit value is adjusted (Typical receiver operating characteristic (ROC) curve.). It is customary to arrange the true-positive amount on the y-axis and the false positive on the x-axis. The greater the area under the ROC curve is, the better the test differentiates between patients with or without disease. ROC curves make it possible to compare tests on a variety of limits. In the example, test A is meaningful across all areas as a test as ROC curves are also used for selection of the limit in order to maximize the benefits of a test. If a test is designed to confirm a disease, a threshold is selected with greater specificity and lower sensitivity. If a test is developed to screen for occult disease, a threshold is selected with greater sensitivity and lower specificity. Typical receiver operating characteristic (ROC) curve. Test Features Some clinical variables have only two possible outcomes (eg life – dead, pregnant – not pregnant.); Such variables as categorical and called dichotomous. Other categorical results to many individual values ??(z. B. Blood, Glasgow Coma Scale) and are referred to as nominal or ordinal. Nominal variables such as blood types have no particular order. Ordinal variables such as the Glasgow Coma Scale have individual values ??which are arranged in a specific order. Other clinical variables including many typical diagnostic tests are continuous and have an infinite number of possible outcomes (eg., Leukocytes, blood sugar). many choose doctors from a limit that has a continuous variable result can be treated as a dichotomous variable (z. B. patients with a fasting blood glucose> 126 m / dL are considered diabetic). Other continuous diagnostic tests have diagnostic utility if they have more limits or if the range of results has a different diagnostic value. If the test results can be defined as positive or negative, all possible outcomes can be recorded in a simple 2 × 2 table (see table: Breakdown of the hypothetical test results) from the key differentiators including sensitivity, specificity, positive and negative predictive value, and probability rate (LR) can be calculated (see Table: distribution of the test results of a hypothetical Leukozytenesterasetests in a cohort of 1000 patients with a prevalence is 30% for a UTI). Distribution of hypothetical test results results disease is present illness is not positive before the test true-positive false-positive test negative false-true-negative patients a total of all patients with the disease all patients without disease Sensitivity, specificity, and predictive values ??sensitivity, specificity, and predictive values ??are typically seen as characteristics of the test itself, regardless of the examined persons. Sensitivity is the probability of a positive test in patients with the disease (true-positive rate). A test is positive in 8 of 10 patients, has a sensitivity of 0.8 (also expressed as 80%). Sensitivity indicates how well a test detects the disease, a test with low sensitivity can not identify many patients with the disease, and a test with high sensitivity is useful to exclude a diagnosis if the results are negative. Sensitivity is the counterpart to the false-negative rate (i. E. The false-negative rate plus Sensitivity = 100%). Clinical calculator: sensitivity of an assay specificity is the probability of a negative test result in patients without the disease (true negative rate), a test that no disease is negative in 9 of 10 patients, has a specificity of 0.9 (or 90%) , Specificity indicates how correctly identified a test patients with disease because tests with high specificity have a low false positive rate. A test with low specificity diagnosed many patients with the disease, even if they are healthy. This is the opposite of the false-positive rate. Clinical Calculator: specificity of a test Positive predictive value (PPV) is the proportion of patients with a positive test who actually suffer if 9 out of 10 positive test results correctly (true-positive), and the PPV is 90%. Since all positive test results have a certain number of true positive and some false alarms, describes the PPV how likely it is that a positive test result for a particular group of patients is a true-positive test result. Clinical computer: Positive predictive value from raw clinical calculator: positive predictive value of a test Negative predictive value (NPV) is the proportion of patients with a negative test result which are virtually free of disease; if 8 out of 10 negative test results are correct (true negative), the NPV is 80%. Since not all negative test results are true-negative, some patients may still develop with a negative test result this disease. The NPV describes how likely it is that a negative test result in a particular patient group represents a true-negative result. Clinical Calculator: Negative predictive value from raw data Clinical Calculator: Negative predictive value of a test probability rates Unlike sensitivity and specificity that have nothing to do with certain patients probabilities, the probability rate may allow the physician to interpret the test results for a particular patient when a well-known (though often estimated) pre-test probability of disease is present. The probability rate describes the change in the pre-test probability of the disease when the test result is known, and answers the question: “How much is the post-test probability has now changed after the test result is known?” Many clinical tests are dichotomous, they are either above the limit (positive) or below the threshold (negative) and there are only two possible outcomes. Other tests give results that are continuous or occur over an area that are selected for the plurality of limit values. The actual post-test probability depends on the size of the probability fees from (depending on the flow of the test) and the estimate of the pre-test probability of disease. If the completed test is dichotomous and the result is either positive or negative, the sensitivity and specificity can be used to calculate a probability positive rate (LR b +) or negative likelihood rate (LR). LR +: The ratio of the probability of a positive test in patients with the disease (true-positive) to the probability of a positive test result in a patient who does not have the disease (false positive). LR: The ratio of the probability of a negative test result in patients with the disease (false-negative) to the probability of a negative test result in patients without disease (true negative) If the result is continuous or has several limits, the ROC curve and not the sensitivity or the specificity used to calculate a probability rate which can not be described as LR + or LR. Since the probability rate is a ratio of mutually exclusive events, and does not specify a proportion of the whole, it shows more chances than probabilities. For a given test, the probability rate for positive and negative results is different. For example, in a positive test result would a probability rate of 2.0 to opportunities of 2: 1 point (true positives to false-positive results) that a positive test result identified a patient as sick. Three positive tests were two occur in patients with the disease (true-positive) and one in a patient without disease (false positive). Because true-positive and false-positive results are components of sensitivity and specificity calculations, the LR + can also act as sensitivity to (1 – specificity) are calculated. The greater the LR + is, the more information has a positive test result; a positive test result in a test with an LR +> 10 is considered strong evidence for a diagnosis. In other words, the pre-test probability estimate moves sharply to 100% when a positive test + has a high LR. With a negative Trestergebnis a LR of 0.25 indicates that the chances be 1: 4 (false negative results), and that a negative test result identifying a patient as sick. Five negative test results of a patient with the disease (false negative) would occur and four in patients without disease (true negative). The LR can also be calculated as follows: (1 -Sensitivität): specificity. The smaller the LR is, the more information provides a negative test result; a negative test result in a test with a LR <0.1 is considered strong evidence for a diagnosis. In other words, the estimate of the pre-test probability moved sharply to a probability of 0% when a negative test has a low LR. Test results with a LR of 1.0 provide any information and did not influence the post-test probability of disease. Clinical Calculator: likelihood ratio (LR) MultiCalc clinical calculator: likelihood ratio (LR) of "negative" from raw data Clinical Calculator: likelihood ratio (LR) of "Positive" from raw data LR are useful for comparative tests and also be used for Bayesian analysis, to interpret test results. Just as sensitivity and specificity change when the limit changes, the probability rate (LR) change. As a hypothetical example, a high threshold for the number of leukocytes (. Eg 20,000 / ul) in a possible case of acute appendicitis would be more accurate and would have a high LR +, but also a high (and therefore not very informative) LR; lower the choice of a lot and very sensitive threshold (eg., 10,000 / ul) would cause a low LR, but also a low LR +. Dichotomous tests An ideal dichotomous test would have no false positives or false negatives; would all patients with a positive test result, the disease had (100% PPV), and all patients with a negative test result, the disease is not (100% NPV). In reality, all the tests have false-positive and false-negative results, some tests more than others. To illustrate the consequences of an erroneous sensitivity and specificity of the test results, hypothetical results can (see Table: Distribution of the test results of a hypothetical Leukozytenesterasetests in a cohort of 1000 patients with an assumed prevalence of 30% for a UTI) of urine dipstick in Leukozytenesterasetests at a group are considered of 1000 women. 300 (30%) of them have a urinary tract infection (as determined by a standardized test such as. For example, a urine culture test). This scenario assumes that the sensitivity of the test strip is 71% and a specificity of 85%. A sensitivity of 71% means that only 213 (71% of 300) were women with a urinary tract infection, a positive test result. The remaining 87 had a negative test result. A specificity of 85% means that 595 would (85% of 700) women without urinary tract infection, a negative test result. The remaining 105 had a positive test result. So were only 213 right of 318 positive test results (213: 318 = 67% PPV), a positive test result makes the diagnosis of urinary tract infection more likely than not a test, but is not sure. There were also 682 negative tests, one of which 595 are correct (595: 682 = 87% NPV), which is a lot less likely the diagnosis of urinary tract infection, but can appear as possible. 13% of patients with a negative test result actually had a urinary tract infection. Distribution of the test results of a hypothetical Leukozytenesterasetests in a cohort of 1000 patients with a prevalence is 30% for a HWI results disease exists disease is not available total number of patients test positive true positive (TP) 213 patients (71% of 300) False -positive (FP) 105 patients (700 -595) 318 patients with a positive test test negative false negative (FN) 87 patients (300 -213) true negative (TN) 595 patients (85% of 700) 682 patients with a negative test patients total of 300 patients with HWI (say) 700 patients without HWI (say) 1000 patients Positive predictive value (PPV) = TP: (all patients with a positive test) = TP (TP + FP) = 213: (213 + 105) = 67%. Negative predictive value (NPV) = TN: (all patients with a negative test) = TN (TN + FN) = 595: (595 + 87) = 87%. Positive probability rate (LR +) = sensitivity. (1 - specificity) = 0.71: (1 - 0.85) = 4.73. Negative probability rate (LR) = (1 - Sensitivity): Specificity = (1 - 0.71) / 0.85 = 0.34. However, the PPV and NPV derived therefrom may be used in this patient population not be used to interpret results of the same test, if the underlying morbidity (pre-test probability) is different. The impact of a change in disease incidence up to 5% is remarkable (see Table: Distribution of the test results of a hypothetical Leukozytenesterasetests in a cohort of 1,000 women with an assumed prevalence of 5% for a UTI). Now most positive test results are wrong, and the PPV only is 20%; a patient with a positive test result actually has rather no UTI. However, the NPV is now very high (98%), a negative result is basically a HWI from. Distribution of the test results of a hypothetical Leukozytenesterasetests in a cohort of 1000 women with a prevalence is 5% for a HWI results disease exists disease is not available patients total test positive true positive (TP) 36 patients (71% of 50) false positive (FP) 144 patients (950 -806) 180 patients with a positive test test negative false negative (FN) 14 Patientinn s (50 -36) true negative (TN) 806 patients (85% of 950) 820 patients with a negative test patients 50 patients with HWI (say) 950 patients without HWI (say) 1000 patients Positive predictive value (PPV) = TP: (all with a positive test) = TP (TP + FP) = 36: (36 + 144) = 20%. Negative predictive value (NPV) = TN: (all with a negative test) = TN (TN + FN) = 806: (806 + 14) = 98%. Positive probability rate (LR +) = Sensitivity: (1 - specificity) = 0.71: (1 - 0.85) = 4.73. Negative probability rate (LR) = (1 - Sensitivity): Specificity = (1 - 0.71) / 0.85 = 0.34. Clinical Calculator: false negative rate using sensitivity and prevalence of clinical calculator: false positive rate based on specificity and prevalence Note that do not change in both groups of patients, the probability rates, although PPV and NPV are very different. This is because the probability rates are determined only by the sensitivity and specificity. It is evident clear that a test result does not provide a definitive diagnosis, but only the probability of disease estimates (present-absent). This post-test probability (probability of the disease in a particular test result) varies greatly depending on the pre-test probability of the disease and on the sensitivity and specificity of the test (and thus its probability rate). The pre-test probability is not a precise instrument, but is based on clinical evaluation: How much can be symptoms and clinical picture a disease of the patient suggests that anamnestic indications and risk factors support the diagnosis, and how often the disease is in a representative population ? Many clinical evaluation methods have been developed to estimate the pre-test probability; adding points for different clinical features allows the calculation of an evaluation. These examples illustrate the importance of accurate pre-test prevalence estimate, since the prevalence of the disease in the population as dramatically affected the usefulness of the test. Validated published "prevalance-estimating" - tools should be used if they are available. For example, there are criteria for predicting a pre-test probability of pulmonary embolism (pulmonary embolism (PE): Clinical probability). Higher calculated figures suggest higher estimated probabilities. Continuous tests Many test results are continuous and may provide useful clinical information on a wide range of results contribute. Doctors often choose a certain limit to increase the usefulness of the test. For example, white blood cell counts> 15,000 can be characterized as positive; White blood cell counts <15,000 against it as negative. If a test provides continuous results, but a certain threshold is selected, the test works like a dichotomous test. You can select multiple thresholds. Sensitivity, specificity, PPV, NPV, LR + and LR can be used for single or multiple thresholds are calculated. Effect of changing the limit for the number of leukocytes in patients with suspected appendicitis shows the effect of changing the limit for the number of leukocytes in patients with suspected appendicitis. Effect of changing the limit for the number of leukocytes in patients with suspected appendicitis leukocyte limit * Sensitivity Specificity LR + LR> 10,500 84% 53.13% 1.79 0.3> 11,500 78% 62.5% 2.13 0 , 32 > 12,850 68% 75% 2.72 0.43> 13 400 61.33% 78.12% 2.86 0.45> 14 300 56.67% 81.25% 3.2 0.49 * Various limits are for a selected continuous variables such as leukocyte count; Results above the threshold are considered positive and results below as negative. LR = probability rate. Adapted from Keskek M, Tez M, Yoldas O, et al: Receiver operating characteristic analysis of leukocyte counts in operations for Suspected appendicitis. American Journal of Emergency Medicine 26: 769-772, 2008. Alternatively, it may be useful to sort the continuous test results in different groups. In this case, the results are not characterized as positive or negative, as there are several possible outcomes, so although a LR can be determined for each stage of the results, there is no longer a significant LR + or LR. Beispielsweise veranschaulicht Verwendung einer Gruppierung von Leukozytenzahlen zur Bestimmung der Wahrscheinlichkeitsrate einer Bakteriämie bei fieberhaften Kindern* die Beziehung zwischen Leukozytenzahl und einer Bakteriämie bei fieberhaften Kindern. Da die LR die Wahrscheinlichkeit eines gegebenen Ergebnisses bei Patienten mit der Krankheit geteilt durch die Wahrscheinlichkeit dieses Ergebnisses bei Patienten ohne die Krankheit ist,