Errors in medicine and errors in laboratory medicine: what is the difference?

Clinicians do not like dealing with errors despite the importance of recognising and minimising errors in clinical practice. Nevertheless, it is clear that even the worst medical errors might be preventable if we move to another dimension, considering the system rather than individual errors and learning from them instead of always trying to find someone who is “guilty”.

In the seminal report “To Err is Human”, released by U.S. Institute of Medicine in 1999, it was stated that as many as 98,000 people die each year needlessly due to preventable medical harm, the equivalent of a national disaster every week of every year¹. A decade has passed, but the situation has not changed so much considering that the Consumers Union (i.e., the expert, independent, non-profit U.S. organisation whose mission is to work for a fair, just, and safe marketplace for all consumers) has recently released a report entitled “To Err is Human - To Delay is Deadly” with an alarming subtitle “Ten years later, a million lives lost, billions of dollars wasted”².

Remarkably, the terms “diagnostic” and “diagnosis” never appear in this document. This may be due to the fact that the entire diagnostic process (whether radiological or laboratory) is thought to be well standardised and controlled, or simply because radiological and laboratory errors are still largely underestimated by governments and healthcare systems worldwide². In fact, few other medical disciplines have such a prominent role to play in patients' safety as do laboratory medicine and radiology. As in other medical fields, however, thing can go wrong and, incidentally, do go wrong in diagnostics^3,4.

A diagnostic error can be defined as “a diagnosis that is missed, wrong or delayed, as detected by some subsequent definitive test or finding”⁵. Although it is difficult to obtain real estimates of diagnostic errors, approximately 3 to 5% of radiological interpretations made by radiologists in their daily practice contain errors^6,7, whereas the prevalence of laboratory errors can be as high as one every 330 – 1000 events, one every 900 – 2074 patients, or one every 214 – 8316 laboratory results⁸. Thanks to the introduction of quality assessment and monitoring, that is, internal quality control and external quality assessment programmes, laboratory medicine is far ahead of other clinical disciplines in introducing quality improvement initiatives. However, the real number of mistakes occurring in diagnostic testing is not fully appreciated because no widespread policies have been established either to determine how often mistakes occur or to systematically eliminate or minimize the latent conditions and active failure that ultimately generate an error⁸.

Laboratory activities/errors have been traditionally classified as pre-, intra- and post-analytical. Several efforts have been focused in the past on intra-analytical errors as well as on mistakes resulting in adverse events, but near misses, those that only apparently cause no harm, have been overlooked⁸.

Although considerable evidence has been provided that poor standardization in the extra-analytical phase of the total testing process has an adverse influence on test results, little information is yet available on the prevalence and type of pre-analytical errors involving specimens for routine haematological testing^4,8–10. Phlebotomists are not always aware of the variability introduced by a suboptimal blood-to-anticoagulant ratio. Under-filling blood collection tubes containing ethylene diamine tetra-acetic acid (EDTA) for haematological tests can also lead to spurious test results, including low blood cell counts and haematocrit, morphological changes to erythrocytes, and altered staining. On the other hand, over-filling the blood collection tube prevents the tube from being mixed properly, leading to platelet clumping and clotting. As such, it is currently recommended by the Clinical and Laboratory Standards Institute/ National Committee for Clinical Laboratory Standards (CLSI/NCCLS) that test-tubes for blood collection should be filled to ±10% of the nominal draw volume^11,12. The procedures that are used to handle samples before testing may also adversely affect the reliability of results; it is, therefore, important that procedures are closely followed to preserve sample integrity for the designated length of time under ideal storage conditions. Strict criteria for sample storage, handling, and transportation, established on the basis of documented stability tests, should be detailed, implemented in routine practice, and monitored thorough the entire pre-analytic processing by using ad hoc performance indicators¹³.

In the past, Lippi et al. described pre-analytical errors recorded on routine and stat patients' samples, focusing on whole blood specimens referred for complete blood count¹⁴. Since the occurrence of pre-analytical errors is much higher in those settings in which specimen collection is outside the laboratory's jurisdiction, such as the hospital wards^4,10,15, their analysis was concentrated on inpatient samples. With regards to 138,769 routine and stat specimens received in the laboratory, 500 (0.36%) pre-analytical mistakes were identified, a prevalence similar to that previously reported by Jones et al. (0.45%, p=0.881)⁹. The most frequent reasons for rejection were clotted specimens (76.8%), inappropriate containers (9.4%) and inappropriate volume (7,0%). The frequency distribution of specific errors was always statistically different between the hospital departments (p<0.001), indicating that varied and heterogeneous pre-analytical problems may occur in different hospital wards and might jeopardize patients' safety and complicate laboratory activities¹⁴.

Patient-related variables, such as physical exercise, stress and fasting status are additional sources of variability in laboratory testing. The CLSI currently recommends determining the patient's fasting status, especially for those tests that are more likely to be affected by food ingestion, such as glucose, lipoprotein fractions and triglycerides. For glucose alone, a fast of 4 hours is sufficient¹⁶. According to the National Cholesterol Education Program (NCEP), the fasting period for lipid measurements should be at least 9 to 12 hours¹⁷. Nevertheless, no clear recommendations about fasting requirements have been established for routine haematological testing, nor has the influence of meal ingestion been assessed. This prompted Lippi and colleagues to evaluate the influence of a regular, light meal on haematology tests. The results of their elegant study are published in this issue of Blood Transfusion¹⁸.

Laboratory tests are essential to the screening, diagnosis and follow-up of several haematological diseases. As previously highlighted, however, pre-analytical variability can affect results and jeopardize patients' safety as in other areas of diagnostic testing^13,19,20. Post-prandial variation of routine haematological parameters is probably relatively irrelevant in most circumstances, when blood is collected from non-fasting patients, for example in emergency situations that require a stat test. However, clinicians should be aware of post-prandial variations of the haematological profile in order to be able to appreciate and troubleshoot “spurious” variations, and to interpret test results adequately , especially in the longitudinal follow-up of the patient. In this context, the paper by Lippi and colleagues has relevant clinical implications, since it is the very first study in which post-prandial variations of routine haematological parameters have been investigated and reported. The results of their study clearly demonstrate that even a light meal can induce significant variations in the routine haematological profile of healthy subjects. As regards white blood cell counts, they observed a clinically significant increase in neutrophils (7–10%), as well as significant decreases in lymphocytes (up to −19%) and eosinophils (up to −23%) up to 4 hours after the meal. They also recorded a significant post-prandial decrease of red blood cell count, haemoglobin concentration and haematocrit, with the differences in red blood cell counts and haematocrit values being clinically significant 2 and 4 hours after the meal, respectively. The authors offered some pertinent explanations for their results, some original, some from a comprehensive review of the current scientific literature: in both cases, they are worth reading.

Over the past decades, laboratory medicine has been a front-runner in pursuing the issue of patients' safety, probably being much more proactive than other medical disciplines²¹. The study by Lippi et al., as well as other important ongoing projects (e.g., The International Federation of Clinical Chemistry Working Group on laboratory errors and patient safety)²², demonstrate that laboratory medicine is pursuing the right path for delivering safe, high-quality care.