The AUC range: 0. However, other data using ROC curves to assess the ability of the tumor markers, prostatic acid phosphatase PAP and prostate specific antigen PSA , to differentiate prostate cancer from BPH and prostatitis at various cutoff values is illustrated in Figure 2. Once a clinical laboratory test with the appropriate diagnostic accuracy has been ordered, how are the results of the test interpreted? Decision level refers to a particular cutoff value for an analyte or test that enables individuals with a disorder or disease to be distinguished from those without the disorder or disease.
Moreover, if the diagnostic accuracy of the test and the prevalence of the disease in a reference population are known, then the predictive value of the decision level for the disorder or disease can be determined. Reference interval relates to the values for an analyte eg, PSA, glucose, etc. All quantitative assays have a finite lower limit of detection LLD , distinct from 0, that more precisely constitutes the lower limit of the reference interval when this lower limit encompasses 0.
Therefore, any PSA value less than 0. In addition, it is important to remember that reference intervals for an analyte are method dependent ie, the reference interval established using one method cannot automatically be substituted for that of a different assay that measures the same analyte. Thus, reference intervals are intended to serve as a guideline for evaluating individual values and, for many analytes, information on the limits of an analyte for a population of individuals with the disease or diseases the test was designed to detect is even more informative.
Also, it is important to recognize that values for some analytes in a population of healthy individuals may not be Gaussian distributed. Figure 2. The reference interval for this data must be determined using a non-parametric statistical approach that does not make the assumption that the data is Gaussian distributed. Example of a distribution of laboratory test values for an analyte ie, the liver enzyme, gamma-glutamyl transferase [GGT] for which the data are not Gaussian distributed. For example, alkaline phosphatase, an enzyme produced by osteoblasts bone-forming cells , would be expected to be higher in a healthy to year-old during puberty and the growth spurt ie, increased bone formation during lengthening of the long bones that normally accompanies puberty in adolescent males and females than those observed in a prepubertal or elderly individual.
Ideally, the best reference interval for an analyte would be individual-specific such that the value for the analyte, determined when the individual is ill, could be compared with the limits for this analyte, established on this same individual, when he or she was healthy or without the illness. For obvious reasons, it is difficult, if not impossible, to obtain such reference intervals. Thus, population-based reference intervals offer the most cost-effective and rational alternative. When using population-based reference intervals, however, it is critical that members of the reference population be free of any obvious or overt disease, especially diseases likely to affect the analyte for which the reference interval is being determined.
For example, when determining a reference interval for TSH also known as thyrotropin , it is critically important that the population of individuals tested be free of any pituitary or thyroid disease likely to affect the pituitary-hypothalamic-thyroid axis, which, under the action of the thyroid hormones tri- T3 and tetraiodothyronine T4 , exert regulatory control over circulating levels of TSH. Quantitative values for all analytes are affected by both imprecision ie, lack of reproducibility in the measurement of the analyte and intra-individual variation over time in the concentration of the analyte due to normal physiologic mechanisms ie, biological variation that are independent of any disease process.
For example, the analyte cortisol, a glucocorticoid produced by the adrenal cortex that is important in glucose homeostasis, normally displays diurnal variation. Blood cortisol levels begin to rise during the early morning hours, peak at mid-morning, and then decline throughout the day to their lowest level between 8 pm and midnight. There is a direct relationship between the magnitude of the CV and the degree of imprecision ie, the lower the CV, the lower the imprecision [or the higher the degree of precision].
The magnitude of analytical variation is given by CV a , while biological variability is defined by CV b. In addition, a change in values for an analyte that exceeds the change ie, reference change value [RCV] expected due to the combined effects of analytical and biological variation alone is due most likely to a disease process or to the affect of any therapy on the disease.
More recently, neural networks, a branch of artificial intelligence, have been used to evaluate and interpret laboratory data.
Product labeling and published guidelines reflect the importance of monitoring laboratory parameters for drugs with a risk of organ system toxicity or electrolyte. Authors investigted the analytical and biological influences on clinical laboratory parameters during administration of Dopamine, theophylline, Cordaron.
Neural networks have been applied to such diverse areas as screening cervical smears Pap smears for the presence of abnormal cells and the identification of men at increased risk of prostate cancer by combining values for PSA, prostatic acid phosphatase PAP , and total creatine kinase CK. The use of neural networks in clinical and anatomic pathology is likely to expand because of their ability to achieve a higher level of accuracy than that attained by manual processes.
Laboratory test results may influence up to 70 percent of medical decision making. Lopasata, to the sparse training in laboratory medicine provided in most United States medical schools. In the final analysis, it is important for clinicians and lab-oratorians to recognize that laboratory data, although potentially extremely useful in diagnostic decision making, should be used as an aid and adjunct to the constellation of findings eg, history, physical exam, etc.
Laboratory data is never a substitute for a good physical exam and patient history clinicians should treat the patient, not the laboratory results.
Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide. Sign In or Create an Account. Sign In. Advanced Search. Article Navigation. Close mobile search navigation Article Navigation.
Volume Article Contents. Diagnostic Decision Making. Medical Necessity.
Reasons for Ordering a Laboratory Test. Clinical Performance Characteristics of Laboratory Tests. Neural Networks. Laboratory Testing Paradox. Oxford Academic. Google Scholar. Readers should also be able to describe the general principles for selecting the most appropriate laboratory test based on its diagnostic performance characteristics. Chemistry questions and corresponding answer form are located after this CE Update article on page Cite Citation.
Permissions Icon Permissions. Abstract According to Dr. Figure 1.
Open in new tab Download slide. Table 1. Open in new tab. Figure 1A.
Table 2. Search ADS. Google Preview. Evaluation of laboratory data by conventional statistics and by three types of neural networks. Comparison of logistic regression and neural net modeling for prediction of prostate cancer pathologic stage. Download all figures.
View Metrics. Email alerts New issue alert. Advance article alerts. Article activity alert. The admissions included 88 emergencies, scheduled admissions and 32 transfers from other clinical departments or hospitals 45 not documented. The average length of hospital stay was 9. According to the Naranjo algorithm, 62 of the ADRs were categorized as possible and 47 as probable. All ADRs were judged to be of clinical relevance. In 50 cases, additional drug therapy and laboratory diagnostic measures were necessary.
coasiethrilatom.gq In 23 ADRs, the drug involved had to be withdrawn. In 40 cases, drug therapy was continued due to a positive benefit to risk ratio or a lack of therapeutic alternatives e.
In 10 ADRs, diagnostic procedures e. Opiate agonists caused gastrointestinal disorders, whereas anxiolytics, sedatives and hypnotics exerted liver or biliary tract toxicity. These ADRs resulted predominantly in metabolic and nutritional disorders. The largest numbers of alerts were those indicating hepatotoxicity and coagulation disorders Delta ALSs for sodium or triglyceride did not occur.
Of the ADRs, the attending physicians recognized 67 Overall, new ALSs were associated with Six ADRs were recognized by physicians only. Delta ALSs proved to be most efficient when based on electrolyte shifts e.
In this particular group, two of three alerts were indeed related to ADRs. Without the dynamic modification, only one of seven alerts was related to a real ADR. One explanation for this may be the high number of patients with pancreatic diseases and bile duct stones in this study. However, our system is still not optimally suited to achieve both an early detection of ADRs and a high predictive value of the alerts. Further improvement may be achieved by modification of the rule definitions. This finding indicates that it is feasible to detect ADRs by a computer monitoring system in real time, and alerted physicians may intervene before serious harm is inflicted.
This study shows that the computer monitoring system is a useful tool for the early and automated detection of ADRs in patients with hepatic or gastrointestinal diseases.
G Volume 19 , Issue 3. The full text of this article hosted at iucr. If you do not receive an email within 10 minutes, your email address may not be registered, and you may need to create a new Wiley Online Library account. If the address matches an existing account you will receive an email with instructions to retrieve your username.
Search for more papers by this author. Tools Request permission Export citation Add to favorites Track citation. Share Give access Share full text access. Share full text access.