|
|
|
|
|
|
|
|
| Issue 1 1998 |  |
Welcome to InfoLink, the new look PaLMS Newsletter. It has been a hectic time for us over the last few months, but we can now see some great results coming about, and want to share with you just what has been happening and how it may affect you. ThePaLMS Informatics Team have now completed the AUSLAB launch across all Northern Sydney Area Health hospital campuses, which means patient results are now available on-line through the hospital computers. We have established a dedicated Service Centre as a central contact point for all your queries: – dial 1300 30 PALMS (or 9926 6066) or extension 66066 if you are calling internally. We are also in the process of connecting Royal North Shore Hospital and North Shore Private Hospital to PaLMS via a pneumatic tube system. Having been selected as the Preferred Provider of pathology services to North Shore Private Hospital we have also been working to ensure that we have services in place to meet the needs of the private hospital and the Clinicians who will be using the hospital when it opens in just a few weeks. Finally, I would like to thank all of you who have been involved in the projects being undertaken by PaLMS. Through this monthly Newsletter we plan to keep you informed of what's happening in the clinical and laboratory medicine world, and encourage you to provide us with feedback and suggestions on our Newsletter. I trust that you will find it a helpful and worthwhile addition to your reference library. |
|
| Kerry Heintze PaLMS Marketing Manager |
|
| IMPORTANT CONCEPTS IN LABORATORY UTILISATION |
|
James P. Isbister Head, Transfusion Medicine, PaLMS |
| The ever increasing range of laboratory investigations for diagnosis and treatment of disease can be daunting, but present the clinician with an important resource to practice scientifically based medicine. It is not possible for a clinician to be knowledgable in all aspects of the laboratory investigations, but some basic concepts are important if laboratory medicine is to be scientific and cost effective in patient care. |
Why do tests?
|
Uncertainty is an inherent part of the diagnostic process, a concept which also needs to be understood and accepted by patients, their relatives, the community, lawyers, bureaucrats and politicians. Investigations are performed to reduce this uncertainty, but ultimately a final diagnosis may not be definitive, nor can disease always be excluded. Individuals with disease may not be totally separable from individuals without disease. To be a successful diagnostician, appreciation of uncertainty, its quantitation, reduction and communication, is pivotal. The good diagnostician does the right test, on the right patient, at the right time, and knows what to do with the result. Patients rarely come into contact with pathologists or laboratory scientists responsible for the scientific study of their disease processes. It does not matter how skilfully specimens are requested and obtained for diagnosis, if they are not appropriately transported and studied in the laboratory, the whole exercise is pointless. |
| What is a "good" test?
|
| How well does a test perform?
There are a range of criteria by which an investigation can be judged as to its usefulness in diagnosis and management of patients. However, in order to evaluate a test we need to be able to make a perfect diagnosis. A test is compared with this gold standard, but even then it is only possible to determine if the test is as good as the gold standard, not better. Accuracy is the measure of how close the result is to the true value as determined by comparison with a known reference standard. A test with poor accuracy is one in which there is a systematic deviation from the true value, referred to as bias. Precision tells us into what range the test will fall if repeated by the same method in the same laboratory on the same sample, ie the standard deviation of a series of repeated measurements. This random scattering from the true value is referred to random error. Sensitivity measures the capacity of a test to identify patients with disease, ie. the percentage of patients with disease who have a positive test. Specificity measures the capacity of a test to identify people who do not have the disease, ie. the percentage of patients without the disease having a negative test. The ideal test would be 100% sensitive and 100% specific. Such a situation could only exist when molecular measurements are being made for quantitative normality. Such a test does not exist. There is thus a "trade off" in that sensitivity is inversely related to specificity, ie. the more likely a test is to detect patients with the disease, the more likely is it to produce false positives in patients without disease. |
| What does a positive result tell me? The likelihood of the patient having the disease in question is known as the "prior odds" (pre-test probability of disease) and will depend on the patient population with which the clinician deals and his/her ability as a diagnostician. Following the performance of the investigations the results will either support or refute the diagnosis by raising or lowering the probability, resulting in "revised odds" (post-test probability of disease). The predictive value of a positive test is the percentage of all positive results that represent disease. The predictive value of a negative test is the percentage of all negative results that exclude disease. The predictive value of a test is determined by the sensitivity, the specificity and the prevalence of the disease in the population tested. A good clinician is one who selects the patients on which to do the specific investigations and, as a result is more likely to come up with a positive result. Tests for nonselective population screening must fulfil more rigorous criteria than tests used for diagnosis in clinical medicine when the prior odds have been established by selecting the patients on whom the test will be performed after history and examination. Health screening can be a two-edged sword if not carried out with due attention to the likelihood of detected abnormalities being significant. The predictive value of any investigation is lower and even with a good test the majority of test positives in a low risk population will be false positives. This problem is partly overcome by selective screening of high risk groups. No information about the precision, accuracy, sensitivity, specificity or predictive value of test are usually included with a the reported result. As long as the result is accurate and precise the further interpretation of the result, in terms of its predictive value in diagnosing disease, rests with the clinician. It can thus be seen that the likelihood of an abnormal laboratory test result being clinically significant, increases or decreases when interpreted in the light of the clinical information. |
| What is a "normal" result of a test? Qualitative normality When the parameter/s in question cannot be quantified with any degree of objective accuracy the result is:
Quantitative normality When the result of the investigation can be numerically quantitated, the interpretation of normality depends on statistics and the knowledge of a reference range which may or may not be appropriate to the patient in question. The use of statistics has been invaluable in laboratory medicine, but caution must be exercised in the area where disease and non-disease overlap. Molecular normality Molecular normality is absolute and statistical methods are not needed for interpretation. There may be an amino acid substitution resulting in an abnormal protein. With molecular biology investigations now increasingly used in diagnostic pathology, more diseases will be identified by DNA analysis. Unfortunately, molecular biology is not the hoped for panacea in diagnosis. Not only do the techniques require a high level of laboratory quality control, but many so called molecular abnormalities do not represent disease and may only be a manifestation of genetic polymorphism or insignificant mutations. Detection of molecular abnormalities in the diagnosis of disease are most valuable when the disease entity (phenotypic expression of an abnormal gene) has been confirmed or there is compelling evidence. Such diseases usually have a single gene abnormality (eg haemochromatosis, cystic fibrosis) and in such circumstances the molecular tests are not so much of value in the diagnosing the patient in question but for genetic screening and antenatal diagnosis. In other diseases which have multifactorial pathogenesis (eg malignancy, vascular disease) molecular biology is only one of the tools for identifying patients with disease and risk groups for disease. |
|
Is this test result a "Laboratory Error" ?
Mistakes are made in laboratories, but caution should be exercised before a test result is attributed to laboratory error. Possible explanations when a laboratory result appears to be at variance with that expected, include:
|
