Interpreting Laboratory Test Results

Interpreting Laboratory Test Results

Summary:

Laboratory testing of body fluids and tissues is an essential part of medical care. It is generally accepted that about 70% of clinical decisions are based on laboratory test results. Patients can access their laboratory results by enrolling in the Patient Portal which all healthcare facilities are obligated to provide to patients. The requirement of releasing laboratory results to patients has altered the dynamic of doctor-patient interaction in that the patient may see the test results before the doctor does. Minor variations in results that are labeled as low/high/abnormal may cause unwarranted consternation to the patients. The number of laboratory results that are outside the “normal range” far exceeds the clinically meaningful abnormal results due to the usually accepted methodology for ascertaining “normal values”/reference ranges. An informal survey of the patients’ questions on HealthTap revealed that 21% of the questions have some element of laboratory test results.

This communication is designed to address patient concerns based on my personal experiences and may provide useful information to patients and practitioners in assessing laboratory results. Briefly, normal values are the central 95% of the range of results in healthy individuals, thus, 5% of the results from normal/healthy individuals would fall outside the normal range. As a corollary if 20 tests are done on a normal/healthy person, one of those results would fall outside the normal range. Additional factors, such as age, gender, ethnicity, test method, and changing notions of normal also need to be considered.

In the early stages of a disease the patient may not have any symptoms and laboratory test results may be the first indication of a health problem. For example, in the early stages of liver, kidney, bone, and heart diseases the person may feel healthy, and abnormal laboratory results provide the first indication of a pending problem. This is similar to many patients with abnormally high blood pressure being without symptoms/asymptomatic.

Neither ignoring the laboratory test results, nor over-reaction is appropriate. The results should be viewed in consultation with the doctor and in the context of other clinical information.

Who is healthy/normal?

World Health Organization (WHO) adopted the following definition of health in 1948; “A state of complete physical, mental and social well-being and not merely the absence of. disease or infirmity”. In a lighter vein, an intern’s definition of a healthy person, “Someone who has not been investigated enough” is just as appropriate. The search for a “Healthy” person is likely to yield a result, similar to that for the search for a “Happy” person in the poem by John Hay, “The Enchanted Shirt”! In practice, individuals without obvious illness are accepted as normal/healthy.

Source of “normal/reference” laboratory values

The population found to be suitable to donate blood is generally considered healthy and often specimens from this population are used to define “normal” values for laboratory tests. Under ideal circumstances, each laboratory is expected to determine its own “normal/reference” ranges by testing at least 120 individuals without discernible disease, from each relevant age, gender, and ethnic group in the population served by the laboratory. This is essentially impractical. In practice, reference ranges or normal values are adopted from historical standards, textbooks, and information provided by manufacturers of laboratory equipment and reagents. An added source of variation is added by the different methods used for various tests. For example, there is a 40-fold variation in the normal values for troponin I, depending on the instrument and method used. Only three tests/analytes (a chemical or molecule) have undergone international standardization, and harmonization, to mitigate the variation engendered by different testing methods. The three analytes are Cholesterol, Creatinine, and Hemoglobin A1C. The coagulation test result, INR, is not an individual analyte and reflects a calculated value that has been nearly standardized. The logistics for harmonization of any test/analyte are overwhelming, expensive, and subject to objections by various vendors of the test equipment and reagents.

It bears emphasis that the historical values were ascertained in metabolic units of research hospitals where patients were admitted and first morning (post-absorptive, basal) specimens were used to determine the normal/reference values. For many analytes, these values are lower than the values that may be seen in an ambulatory patient coming to the laboratory. For example, (a) blood levels of creatinine kinase, a muscle enzyme, go up with normal activities, similarly (b) blood levels of the enzyme alkaline phosphatase are higher after a meal.

Process for determining “normal/reference” range:

At the introduction of a new test for a new or old analyte, the regulations by the Centers for Medicare and Medicaid Services (CMS), Food and Drug Administration (FDA), and Centers for Disease Control (CDC) require following a prescribed process for validation of the test, including establishing normal/reference range. In its simplest form, 120 healthy individuals are tested. The specimen may be blood, a blood component, or another body fluid. The results from this group of people, referred to as a partition, are plotted. The lowest 2.5% and highest 2.5% of the values are excluded and the values comprising the central 95% are adopted as the normal/reference range. This convention of using the central 95% as normal is an important cause of “abnormal” laboratory results in healthy people. By definition, 5% of normal/healthy people have an abnormal result. Thus, it is imperative that all laboratory results be interpreted in the clinical context to ensure that a person is not labeled with a disease diagnosis based solely on an “abnormal” laboratory value. As a corollary of the central 95% convention for the normal range, if 20 tests are done on a normal/healthy person, one of the results is expected to fall outside the normal/reference range.

Exceptions to source and process for determination of “normal/reference” laboratory values:

1. 1. Expert opinion: The central 95% of values in an apparently healthy population are sometimes modified by experts based on additional information. For example, (a) Body mass index (BMI) of the extant population is not used as the standard because it is known that patients with higher BMI are susceptible to more diseases, (b) normal blood pressure readings have been lowered over the years, (c) thyroid stimulating hormone level range has been narrowed over the years due to increase sensitivity and preciseness of the test methods.
2. Age as a modifier: In an adult woman a hemoglobin level of 12.0 g/dL is considered normal, however, a newborn girl with that level of hemoglobin is considered to have anemia. Blood levels of prostate-specific antigen (PSA) increase with increasing age without the presence of prostate carcinoma. Blood levels of alkaline phosphatase are higher in growing children and pregnant women than in adult men.

The levels of some analytes/chemicals found in healthy blood donors do not reflect the usual levels in older individuals without a relevant disease. For example, the normal ranges for serum-free light chains noted in blood donors generate an abnormal value in 36% of adults presenting to a tertiary care hospital, without any evidence of a monoclonal gammopathy/bone marrow disease. It could be argued that normal/reference values should be derived from a “healthy” group of the same age as the one affected by a relevant disease.

1. Gender as a modifier: The lower limit of normal hemoglobin is 12.0 g/dL in adult women, and it is 14.0 g/dL in adult males. The lower level in women is not due to menstrual blood loss but due to the hormonal influence improving the delivery of oxygen to tissues. Transgender men and transgender women tend to acquire the values seen in cisgender men and women. Similarly, some hormone levels have gender-specific normal values.
2. Physiological state: As mentioned above, growing children have different values of some enzymes and hormones. Pregnant women have different normal levels for some laboratory tests than non-pregnant women.
3. Ethnicity: Some laboratory values vary by ethnicity, just like normal height differs between Dutch and Filipinos. Vitamin D levels are lower in persons of African origin even though they have a lower incidence of bone disease. The apparently lower levels of vitamin D in African-origin persons are due to genetically determined lower levels of vitamin D binding proteins, though bioavailable levels of the vitamin are normal. To avoid the stigma of race, newer ranges are sometimes described to avoid taking race into consideration, e.g., newer formula for calculating estimated glomerular filtration rate (eGFR).
4. 99^th percentile rather than central 95%: The normal lower limit of troponin, a marker of heart disease, is based on the 99^th percentile of the value in healthy subjects. There is no range, just one value. A person with the cutoff value or lower result is normal and a higher level is abnormal.
5. Only one normal/reference value, rather than a range: For some analytes/chemicals only values higher than a trigger point are cause for concern, and lower values have a different interpretation. For example, blood creatinine levels in men are 0.74 to 1.35 mg/dL, and for women, 0.59 to 1.04 mg/dL. Levels higher than the upper limit indicate kidney insufficiency. However, lower levels do not mean kidney hyper-efficiency but lower muscle mass or sarcopenia, as muscle is the source of creatinine.
6. Variation by method of testing: As mentioned earlier, testing methods for only three analytes/tests, Cholesterol, Creatinine, and Hemoglobin A1C have been standardized internationally. For all other analytes, results and normal values vary by the method of testing. The worst example of troponins has been alluded to earlier. Some analytes are tested by different methods even in a given hospital, for example, the blood hemoglobin measured by blood gas analyzers in the Emergency Department is 0.5 to 1.0 g/dL higher than the result from the main laboratory. Blood glucose values determined by portable glucometers are generally lower than the plasma glucose reported by the main laboratories. Values of vitamin D vary between immunological methods and mass spectrometry. It is important to address the normal values/reference ranges, particularly in the laboratory where testing was done. All accredited laboratories report their normal/reference ranges with results from patient specimens.
7. Essential nutrients: The recommended daily allowance (RDA) for essential nutrients and by corollary, their normal values, are geared to provide a value that prevents disease in 98% of otherwise healthy individuals. RDA is not designed for optimum health. Folic acid provides an excellent example of this conundrum. The National Institutes of Health (NIH) lists the normal level of serum folate/folic acid as about 3.5 ng/mL. This level prevents megaloblastic anemia in 98% of the population. A proportion of individuals with this level of serum folate have elevated levels of homocysteine and the values of this undesirable substance are normalized if sufficient folic acid is administered to raise serum folic acid levels to 7.0 ng/mL. FDA mandated that folic acid be added to cereals as the supplement is known to prevent neural tube defects in the fetus. However, the optimum serum folate level for maximum benefit is 13 ng/mL. Similarly, a serum folate level of >13.0 ng/mL has been shown to reduce strokes in hypertensive patients. Thus, there is a wide gap between the minimum essential level and the optimum level. A similar situation has also been observed for vitamin D. A serum level of 12.0 ng/mL prevents bone disease but a proportion of people with that level have elevated levels of parathyroid hormone. If sufficient vitamin D is administered to raise the serum levels to 20.0 ng/mL, parathyroid hormone levels get normalized. In most of human history, we were exposed to sun all day, extant populations with similar exposure have vitamin D levels of 50-80 ng/mL. Thus, it could be argued that normal serum levels of vitamin D should be pegged at 50-80 ng/mL.

Screening tests:

While there are reasons for using screening tests, it is pertinent that laboratory tests be ordered only if there is a clinical reason for doing so. CMS tries to force this by not paying for tests that do not have a clinical justification. Unwarranted tests add to not only the cost of healthcare but add to the safety risks of healthcare. Unwarranted tests will generate abnormal results even in the absence of any disease and incur additional tests and procedures to address the apparent abnormality. A wise suggestion for the use of laboratory tests can be summarized in the following quip, “Remember, ordering a diagnostic test is like picking your nose in public: You must first consider what you will do if you find something.”

It is by no means implied that laboratory tests are not a useful diagnostic tool. In contrast to the above caution for using laboratory tests judiciously is the well-founded expression that, “Without Laboratory Medicine, you are just guessing”, i.e., laboratory tests are essential for accurate diagnoses. It is generally accepted that about 70% of clinical decisions are based on laboratory results. This conundrum of using or not using laboratory tests is similar to the apparently mutually contradictory expressions, “It is never too late” and “It is later than you think”, both being correct and wisdom lies are the appropriate use of the expressions and laboratory tests.

Screening tests for “early” diagnosis have a variable history of usefulness and tests that seem to make sense for use in screening the general population, often fail to demonstrate usefulness when tested rigorously. Some examples of this concept are tests for prostate cancer, Prostate-specific antigen (PSA) colon cancer screening with carcinoembryonic antigen (CEA), and screening for diabetes with tests for fasting blood glucose. The recommendation of the National Preventive Services Task Force usually provides fact-based recommendations, though it takes years if not decades, to arrive at evidence-based conclusions.

Characteristics of a useful laboratory test:

A clinically useful laboratory test is expected to meet at least one of the following characteristics. Examples of representative tests are listed for each of the characteristics.

1. Detect disease or predisposition to disease:
   1. Elevated troponin levels detect injury to the heart muscle.
   2. Elevated cholesterol levels point to a predisposition to heart disease.
   3. Confirm or reject a diagnosis:
      1. Presence of HIV nucleic acids in blood confirms a diagnosis of HIV infection.
      2. Lack of HIV nucleic acid and antibody at 6 months after a risk event rules out HIV infection.
      3. Establish prognosis:
         1. Elevated hemoglobin/hematocrit in a patient with pancreatitis indicates a poor prognosis.
         2. Guide patient management:
            1. Monitoring Hemoglobin A1C concentration facilitates the management of diabetes mellitus.
            2. Monitor efficacy of therapy:
               1. Measurement of anion gap addresses the efficacy of treatment of diabetic ketoacidosis.

Repeat test results for monitoring:

Laboratory tests are often repeated at intervals to monitor the course of illness and response to treatment. Repeat test results have a certain amount of irreducible variation due to variations in the state of the patient, variability in specimen collection, and variability in the testing process. As a rule of thumb, a 10% variation is generally not clinically meaningful. For example, accrediting agencies allow a variability of 10% or 6 mg/dL for glucose results, whichever is larger, on repeat testing.

Clinical context:

All laboratory test results (and imaging findings) need to be interpreted in the clinical context and the doctor ordering the tests is generally in the best position to do that, therefore patients should consult the doctor before taking any action on laboratory results. It is prudent to treat the patient and not the laboratory results.

Respectfully,

Gurmukh Singh, MD, PhD, MBA, FCAP, CPE, FAAPL