Separating Diseased from Disease Free Persons

Under ideal circumstances, sensitivity and specificity approach 100%. In reality, they are lower. The best currently available test to decide who is diseased or disease-free could be imperfect and have sensitivities and specificities in the 80% range. Moreover, discrepancies between a test's efficacy and its effectiveness are common. Efficacy is a test's performance under ideal conditions, whereas effectiveness is its performance under usual circumstances. Tests under development are evaluated under highly rigorous criteria, but in clinical practice, inadvertent error can be introduced into the technical performance or interpretation of the test results. Also, test values for the diseased and disease-free populations overlap.

A cutoff value may be chosen to separate "normal" from abnormal (Figure 15-1). This decision is arbitrary and involves selecting a balance between sensitivity and specificity. The receiver operating characteristic (ROC) curve is a graphic analysis used to identify a cutoff that minimizes false-positive and false-negative results (Figure 15-2). The sensitivity and specificity are calculated for a number of cutoff values, with the variables 1-Specificity plotted on the x axis and Sensitivity plotted on the y axis. Each point on the curve represents a cutoff for the test. A perfect test would have a cutoff that allowed both 100% sensitivity and 100% specificity. This would be a point at the upper-left corner of the graph. The most efficient cutoff for a single test is the one that gives the most correct results, represented by the value that plots nearest to the upper-left corner of the graph.

The optimal cutoff depends on the purpose of the test and essentially is a risk/benefit analysis. In situations where disease detection is most important, the cutoff may be chosen that maximizes sensitivity at the expense of decreasing specificity. If disease exclusion is the goal, sensitivity and negative predictive value need to be maximized. It is important that negative results be true negatives as opposed to false negatives, so that a negative test has correctly excluded the individual as having disease. Similarly, if disease confirmation is the goal, specificity and positive predictive value are critical. It is important that positive results are true positives and not false positives, so that healthy persons are not misidentified, especially when treatments (e.g., surgery) have serious risks.

The predictive value of a test is directly related to the pretest probability of disease. When the prevalence of disease is high in the population, a positive test result is expected and a negative result is not expected, because the disease is common. Similarly, when the prevalence is low, a negative test result is anticipated because few people have the disease. These characteristics of predictive value become clinically useful when one compares the outcome of a positive or negative test result with the pretest probability of disease (Figure 15-3). Prevalence (pretest probability of disease) is plotted against predictive value for a positive and negative test. Note that a test result loses its ability to discriminate those who have disease from those who do not at the extremes of prevalence. If disease probability is low, a positive or negative result does not change the post-test probability much— it is still low. On the other hand, if disease probability is high, the post-test results, whether positive or negative, do not substantially alter an already high probability of disease being present. The predictive value has the greatest power to discriminate those with disease from those who are disease free in the mid-pretest probability range, near 50%. A positive test result suggests a higher post-test probability of disease than a negative result.

Table 15-3 Diagnostic Test Performance Characteristics

Finding

Disease Present

Disease Absent

Test positive

True positive (TP)

False positive (FP)

Test negative

False negative (FN)

True negative (TN)

Sensitivity = TP/(TP + FN); Specificity = TN/(TN + FP).

Positive predictive value = TP/(TP + FP); Negative predictive value = TN/(TN + FN).

Increases sensitivity

Decreases specificity

Decreases sensitivity

Increases specificity

Value

Figure 15-1 Effect of changing a test's cutoff value on disease classification.

(Modified from Cebul RD, Beck LH. Teaching Clinical Decision Making. Westport, Conn, Praeger, 1985, p 4.)

Figure 15-1 Effect of changing a test's cutoff value on disease classification.

(Modified from Cebul RD, Beck LH. Teaching Clinical Decision Making. Westport, Conn, Praeger, 1985, p 4.)

1—Specificity

Figure 15-2 Receiver operating characteristic (ROC) curve showing the effect of changing the cutoff values for separating disease from no disease.

(From Tetrault GA. Laboratory statistics. in Henry JB (ed). Clinical Diagnosis and Management by Laboratory Methods, 20th ed. Philadelphia, Saunders, 2001.)

1—Specificity

Figure 15-2 Receiver operating characteristic (ROC) curve showing the effect of changing the cutoff values for separating disease from no disease.

(From Tetrault GA. Laboratory statistics. in Henry JB (ed). Clinical Diagnosis and Management by Laboratory Methods, 20th ed. Philadelphia, Saunders, 2001.)

Supplements For Diabetics

Supplements For Diabetics

All you need is a proper diet of fresh fruits and vegetables and get plenty of exercise and you'll be fine. Ever heard those words from your doctor? If that's all heshe recommends then you're missing out an important ingredient for health that he's not telling you. Fact is that you can adhere to the strictest diet, watch everything you eat and get the exercise of amarathon runner and still come down with diabetic complications. Diet, exercise and standard drug treatments simply aren't enough to help keep your diabetes under control.

Get My Free Ebook


Post a comment