Accuracy and precision issues
Accuracy refers to the correctness of the result, with minimal difference from the actual value. Precision refers to the repeatability of the result. In practice, these two concepts are often confused, with precision attempted to substitute for accuracy. For example, when encountering an abnormal result, many colleagues prefer to re-test, observing its repeatability. However, this is unscientific, or at least incomplete. Good repeatability only indicates good precision, not good accuracy. If the testing method or instrument has a systematic error, all results will be higher than the actual value, and re-testing will still yield higher results. Some colleagues like to write on the test results: 'This result has been re-tested', which does not actually indicate the accuracy of the test result. On the contrary, it can only make the clinicians look down on the tests. Besides doing it again, there is no other way. To ensure the accuracy of the test results, we should pay attention to comprehensive quality control, pre-examination specimen collection time, method, preservation, transportation process, instrument maintenance, control of interfering factors, operational standardization, scientific nature, etc. The precision of the instrument itself is the performance of the instrument, which is irrelevant to the specimen tested and more irrelevant to the accuracy of the test results.
 
Sensitivity and specificity issues
Any diagnostic index has two basic characteristics: sensitivity and specificity. Sensitivity refers to the chance of missing a diagnosis (smaller is better). Specificity refers to the chance of misdiagnosis (smaller is better). For a single indicator, if its diagnostic sensitivity is increased, its diagnostic specificity will inevitably decrease. In other words, reducing missed diagnoses will inevitably increase misdiagnoses, and vice versa. For example, if we use AFP to diagnose liver cancer, and we stipulate that if AFP is greater than 10, liver cancer can be diagnosed, then the chance of missing a diagnosis will be very small. Obviously, many liver cancer patients have AFP greater than 10, although the chance of missing a diagnosis is small, the chance of misdiagnosis is large. Clearly, many patients with AFP greater than 10 do not have liver cancer. If we stipulate that AFP greater than 2000 can diagnose liver cancer, then there will be no or very few misdiagnoses, because no other disease has such a high AFP, but we will miss diagnoses because many liver cancer patients have AFP less than 2000. Therefore, we need to comprehensively evaluate the sensitivity and specificity of an indicator. The best method is to make a ROC curve, and determine the diagnostic efficacy of the indicator through the area under the curve, and determine the optimal diagnostic value, which balances the sensitivity and specificity of the indicator. An ideal indicator should have 100% diagnostic specificity and sensitivity, but such an indicator does not exist. Using a single indicator to diagnose a disease will inevitably lead to some false positives and false negatives, that is, missed diagnoses and misdiagnoses will inevitably occur.
 
Abnormal results and normal results issues
Many colleagues in daily work pay great attention to abnormal results, and take further measures to confirm abnormal results. This is necessary because false positive results can lead to misdiagnosis and cause suffering to patients. However, we ignore negative results. From another perspective, false negative results will lead to missed diagnoses, and missed diagnoses and misdiagnoses cause the same harm to patients. Therefore, I think that for negative specimens, especially those that do not match the patient's clinical manifestations, clinical diagnosis, or cannot be explained by the patient's pathophysiological process, as well as negative results (normal results) that affect the patient's diagnosis and treatment, we should pay attention. Of course, this requires close contact with clinicians, which I will discuss later.
 
Reference range and diagnostic value issues
Any laboratory index has a reference range. The establishment of the reference range is often obtained through large-scale surveys. If an index is normally distributed, then the 95% confidence interval is its reference range. If the indicators of blood routine are not normally distributed, we need to transform them into a normal distribution before determining their reference range, or rely on statistical processing to determine their reference range, such as AST, CK, etc. This actually implies two meanings: First, 5% of people have an indicator that falls into this 5% of the population, but they still belong to the normal population. Second, if a person's indicator exceeds the reference range, it only means that his indicator is different from that of normal people, but it cannot therefore determine that he has a certain disease. If a single indicator is used to diagnose this disease, it is not enough for the indicator to simply exceed the reference range. It also needs to reach a certain diagnostic value. For example, to diagnose acute pancreatitis, if the amylase is only slightly higher than the normal range, it cannot be diagnosed. If amylase is used alone to diagnose acute pancreatitis, the amylase must be more than 3 times higher than the normal range, or more than 500 units. Many laboratory indicators have specific diagnostic values when diagnosing certain diseases. Currently, laboratory personnel generally ignore the relationship between reference range and diagnostic value, and ignore the problem of sensitivity and specificity, believing that as long as the result is abnormal, a certain disease can be diagnosed.
 
Rationally viewing quality control
There is a bad idea in current testing, which is to try to use quality control results to explain the accuracy of test results. The use of target values provided by manufacturers in internal quality control results is a wrong practice. Quality control is a scientific and standardized operation, while routine specimens are operated improperly, and inter-laboratory quality control involves mutual comparison of results. When clinicians doubt the accuracy of test results, laboratory personnel often say that our quality control is qualified, so the test results must be accurate. This is actually an irresponsible approach. We should rationally understand quality control. First, quality control is an auxiliary means to ensure the accuracy of test results, not a decisive means. Second, quality control products have a certain one-sidedness and lack of representativeness. Specimens vary greatly, and various interfering factors exist. Smooth quality control does not necessarily mean smooth specimens, and accurate quality control does not necessarily mean accurate test results of specimens. Most importantly, our current so-called quality control is actually quality control in analysis, or in simpler terms, the instrument must be accurate. We should clearly recognize that the accuracy of a test result needs to be guaranteed from multiple aspects: pre-analysis quality control, selecting an appropriate item at an appropriate time, collecting specimens of appropriate quantity and quality in an appropriate way, preserving and transporting them to the laboratory in an appropriate way, and the laboratory personnel conducting tests with appropriate operating procedures. Coupled with good instrument performance and minimal interference, an accurate test result can be obtained. These are very important in our daily work. Do not think that quality control is the only thing, or the killer of testing. Qualified quality control is not an excuse for laboratory personnel to explain the accuracy of results and refute clinical doubts. To ensure the accuracy of test results, quality control is a necessary means, but not the only means, let alone a decisive means to ensure its accuracy.
 
The soul of testing is combining with clinical practice
The essence of laboratory testing is a controversial topic. Many colleagues believe accuracy is paramount, but I disagree. I believe it lies in clinical integration. Firstly, to elevate the status and compensation of laboratory professionals, we need clinical integration, increased interaction with patients and doctors, and better mutual understanding. Integrating laboratory work into clinical practice reduces misunderstandings and enhances our influence. Secondly, we are called 'Laboratory Medicine' or 'Clinical Laboratory Medicine'. Some hospitals even name their departments 'Laboratory Diagnostics', emphasizing its medical nature. A science that doesn't involve patients or disease diagnosis cannot be called medicine, and a professional who doesn't solve clinical problems cannot be called a doctor. Currently, many laboratory professionals remain isolated in laboratories, focusing solely on specimens and tests, significantly detached from clinical practice. Some argue that reporting cell morphology only, without providing specific diagnoses, avoids liability. However, I believe every profession carries risks. Fear only limits us, weakens our voice in the hospital, lowers our status and compensation, and exacerbates the disconnect between laboratory and clinical practices. To progress, we must be brave enough to take risks. Only by solving problems clinical teams cannot can we gain respect, status and a future. Growth often requires sacrifices. Don't let fear of those sacrifices prevent your growth. While the costs may sometimes seem high, we can mitigate or eliminate them by improving our skills and competencies.
 
7. Fully Understand the Role of Laboratory Indicators
Many believe laboratory indicators are solely for diagnosing diseases, which is incomplete. Their uses are broader: 1. Disease prevention and prediction of disease probability (e.g., CRP for cardiovascular diseases); 2. Disease prognosis prediction (e.g., troponin I for AMI diagnosis and prognosis); 3. Treatment monitoring (e.g., CKMB for evaluating the efficacy of thrombolysis); 4. Disease exclusion (e.g., absence of myoglobin elevation within 6 hours of chest pain largely rules out AMI). Many more examples exist.
 
8. Focus on Key Areas
Our energy is finite. We cannot be experts in everything. We must focus on key areas and trends in laboratory medicine, such as comprehensive laboratory management, integrated indicator assessment, the discovery of new indicators, the invention of new methods, and molecular diagnostics. Some professionals obsess over urine crystals, spending all their time researching their morphology and meaning. While clinically significant, this is limited. Instead of this, we should study mass spectrometry techniques, large-scale survey data, and gain in-depth understanding of indicators. Judging a laboratory professional based on urine crystal expertise is not appropriate. Excellence lies in understanding trends, possessing broad knowledge, and staying current. The world is constantly evolving. Testing methods and indicators may be replaced, updated, or improved. We must keep pace.
 
9. Emphasize Coordination with Clinical Teams
I heard a story: A hospital's laboratory had only one biochemistry analyzer, which frequently malfunctioned. When it broke down, the director led the team to manually test each indicator using a 721 spectrophotometer, ensuring all reports were issued the same day. Their dedication is commendable, but their method is impractical, akin to Foolish Old Man moving mountains. First, manual testing is less precise and accurate than instrumental testing. Reagent source, quality, and technician skills are not guaranteed. Secondly, not every patient requires urgent reports. Coordination with clinicians is key; urgent samples can be handled manually. Non-urgent samples can wait until the instrument is repaired or outsourced. While some might view this as irresponsible, explaining the advantages and disadvantages of both methods, as well as the impact on time and cost, to patients, may make them understand the necessity of the method. The most important aspect is instrument maintenance. A laboratory unable to manage its instruments effectively shouldn't rely on manual methods. Instead of that, the laboratory should improve conditions, or other issues. While instrument failure and delayed results are embarrassing, clinical coordination and preventative measures are crucial.