There are instances in which the results of a laboratory test must be reported, but should be ignored by the clinician. Some are accurate findings without clinical significance, but most are spurious results caused by an error somewhere in the testing cycle. Any abnormal laboratory result should raise 2 basic questions: First, what is the clinical relevance of the result? Second, what actions should be taken? This article examines the range of reasons why some laboratory results should be ignored. It examines the causes of spurious results and examines mechanisms to improve communication to convey such reports to the clinician in a more meaningful fashion. The ultimate goals are improved patient outcomes and better utilization of resources.

Specific issues to be addressed in this commentary include the need to expand quality programs to address the entire testing cycle, root causes of spurious lab results in clinical and anatomic pathology laboratories, the critical involvement of clinicians in the quality improvement process, and the role of interpretative reporting to reduce medical errors related to spurious results. Many spurious results will not be identified by quality programs that address only the analytic phase. Joint efforts by clinicians and laboratorians can improve detection of errors and reduce the potential for patient harm and waste.

The testing cycle extends from the decision to perform the test through the final interpretation and response to the laboratory result (Figure 1). For decades, quality programs in laboratories have focused almost exclusively on the analytic phase of the testing cycle. Quality efforts that focus on the precision and accuracy of laboratory assays are important, but errors are more likely to arise in the preanalytic and postanalytic phases of testing.[1–4] Significant gains in quality can be achieved by focusing on these pre- and postanalytic phases.[5–9]

Figure 1 The testing cycle.

Common sources of error in the preanalytic phase include unintelligible requests, inadequate samples, errors in patient identification and data entry, and errors during collection, transport, and accessioning of specimens. In the postanalytic phase, common errors fall into 3 major categories: (1) turnaround time (result reported too late), (2) routing (report not received by the correct clinician), and misinterpretation (correct result reported, but wrong clinical action taken).[10] Because these types of error involve both the laboratory and clinician, both must be involved in efforts to reduce errors.

Figure 2 demonstrates the portion of the testing cycle that has been traditionally addressed by laboratory quality programs. Figure 3 gives a sense of the opportunity for gains in quality if the entire testing cycle is addressed. Errors in the pre- and postanalytic phases may be attributable to the clinician, the laboratory, or the system. An effective quality program relies on active participation from all participants in the testing process. Data entry errors remain a major cause of spurious laboratory results. Every requirement for manual data entry or transfer creates an opportunity for error. A specimen may be labeled at the bedside, with the same critical information entered as a separate process on the request form and again in the laboratory data system. Failure mode analysis of the processes involved often identifies separate requirements for manual entry of 15 or more digits for the test, 5 digits for the operator, and an additional 12 for patient billing information. Bar coding at the point of care can reduce medical errors related to data entry.[11,12] Even when bar coding is instituted, scanning errors may be related to multiple wristbands and incorrect codes related to hospital transfer or previous hospitalizations. Verification of the scanned information is still required. Manual data entry may still be required when bar-coded scans are unsuccessful. Information technology systems must ensure the integration of communication throughout the testing cycle. Common standards are required for data reporting systems to ensure that they communicate with each other with seamless electronic transfer of data and the ability to capture asynchronous data. Clinical decision support can be integrated into these systems to enhance effective communication.[13] Errors in communication can also be reduced through the use of bar-coded phrase sheets and keyboard macros.[14]

Figure 2 Quality efforts have traditionally focused only on the analytic phase.

Figure 3 The opportunity for improvement.

Although most laboratory errors relate to the pre- and postanalytic phases, the analytic phase should not be ignored as a source of errors. Modern laboratories use highly specific, sensitive methods that generally result in reliable results. Because of this, clinicians place great faith in the results that they receive, and false-positive results frequently result in a cascade of unnecessary testing and treatment. One widely publicized example involved a 22-year-old woman who underwent both surgery and chemotherapy on the basis of multiple false-positive human chorionic gonadotropin results.[15] The methodology employed was a sandwich immunoassay that utilized 2 specific monoclonal antibodies. In this assay, the target protein forms a link between the capture and signal antibodies. The assay is known for its high sensitivity and specificity. Unfortunately, heterophilic antibodies can also form links between the capture and signal antibodies, resulting in a false-positive result. Despite test modifications to minimize antibody interference, heterophilic antibody interference still occurs in a wide variety of immunoassays, and false-positive results have led to bad clinical outcomes.[16–20] Because this type of interference is patient-specific, it usually escapes detection by quality assurance systems that focus mainly on the analytic phase of testing. A broader system that addresses the entire testing cycle can improve quality.

False-positive results that result in medical errors occur in a wide variety of laboratory assays. A study of 6370 specimens analyzed by a widely used bacterial latex agglutination test found that only 11 pathogens were accurately detected, whereas there were 13 false negatives and 59 false-positive results. In this study, none of the true positives had a measurable effect on patient outcomes, but several of the erroneous reports resulted in unnecessary treatment.[21] Nonculture tests for detection of Chlamydia trachomatis have advanced the detection of this serious sexually transmitted disease, but have also been complicated by false-positive results. Implicated assays include both enzyme immunoassays and DNA probe tests.[22] The issue of false-positive nucleic acid detection tests for Neisseria gonorrhoeae has prompted recommendations for the reporting of such tests.[23] A major source for such errors is horizontal genetic exchange among Neisseria, resulting in commensal Neisseria acquiring N gonorrhoeae genes.[24] As with other studies, false-positive results are more likely in populations with low disease prevalence.[25] Published recommendations with regard to reporting of nucleic acid detection tests for N gonorrhoeae include the following: Assays for the cppB gene should not be used; all positive screening assays should be confirmed by a reliable supplemental assay before a positive result is reported; and the test combination should yield a positive predictive value of at least 90% in a population with a 1% prevalence.[26] Polymerase chain reaction (PCR) assays offer sensitivity and specificity similar to culture and can be used as confirmatory tests, but may not be suitable for pharyngeal specimens.[27–29]

A falsely reported outbreak of methicillin-resistant Staphylococcus aureus was linked to spurious results with an automated test system. Independent laboratory testing with oxacillin agar screen plates, broth microdilution minimal inhibitory concentrations, and assays for the presence of the mec A gene proved that almost three quarters of the organisms initially identified as methicillin-resistant S aureus were actually susceptible to oxacillin. The misclassification resulted in unnecessary vancomycin use.[30] An apparent outbreak of typhoid fever was related to false-positive tests with a hemagglutination method. Most laboratories issuing spurious reports performed the rapid slide agglutination test as opposed to the tube agglutination test.[31] Each of the above examples shares one important feature: Quality improvement programs that focused on the analytic phase of testing failed to detect unacceptable rates of false-positive results. Analysis of patient outcomes ultimately uncovered the errors.

Efforts to classify anatomic pathology errors and assess their influence on clinical outcomes have been hampered by marked differences in reporting practices and lack of a uniform classification scheme for errors in anatomic pathology.[32] As with clinical pathology, quality improvement efforts have traditionally centered on the analytic phase, with an emphasis on peer review of a percentage of cases, correlation of frozen-section and permanent-section diagnoses, case conferences, and consultations. Quality measures of the pre- and postanalytic phases have generally been limited to patient and site verification and turnaround time. Additional gains in quality may be achieved by focusing on the entire testing cycle and patient outcomes.

Interpretative anatomic pathology reports can be of value. An example is the cytologic atypia that is characteristic of so-called “ancient schwannomas.” Although the nuclei are strikingly atypical, the tumors behave in a completely benign fashion.[33] The lack of clinical significance must be clearly communicated in order to avoid unnecessarily extensive surgery. Perineuriomas represent another group of benign tumors with a wide spectrum of histologic appearance. Atypia and hypercellularity showed little correlation with clinical behavior.[34] Similar “ancient” atypia occurs in a host of other neoplasms.[35–39] Unambiguous communication of the lack of significance of the finding is essential to avoid overly aggressive treatment.

Interpretative reports are crucial when a laboratory result should be ignored, but even accurate results may be misinterpreted, and interpretative reports have dramatic potential to reduce medical errors and unnecessary follow-up testing.[40–42] Accurate interpretation of test results depends on patient history, experience of the interpreter, and local disease prevalence. When the pretest likelihood of disease is small, there will be a larger number of false positives.[43] In the preanalytic phase, redirecting test ordering has the potential to reduce false-positive results stemming from testing patients in low-prevalence populations. In the postanalytic phase, computer graphs have been used to show post-test probability of disease as a function of pretest probability.[44]

Interpretative reports vary from graphic representation of normal ranges to elaborate mathematic models. Computer systems have facilitated interpretive reporting, especially for high-volume and high-complexity laboratory tests.[45–48] Automated laboratory analyzers can present laboratory results in graphic form relative to population-based data. Graphic reports with manually added interpretative text can be of additional value in conveying results effectively.[49] Critical differences (as well as those differences that are clinically insignificant) must be clearly communicated in graphic reports. The reporting interval or bin size that is used to report numerical results influences interpretations of the critical difference between results. As the reporting interval size increases, there is a roughly linear increase in the appearance of a critical difference.[50] Standard reporting intervals must be integrated with data in regard to what difference is clinically relevant.[51]

When reporting culture results, the growth of organisms that usually represent normal flora is complicated by the fact that the same organisms can sometimes act as opportunistic pathogens.[52,53] The line between normal flora, colonist, and pathogen may be poorly defined.[54,55] It has been traditionally left to the clinician to determine the clinical significance of the organism, but as our understanding of the role of virulence factors in opportunistic infection and immune-mediated diseases evolves, laboratories will have a more defined role in communicating the possible clinical significance of various organisms. Interpretive reports are also being used to communicate bacterial susceptibility test results effectively.[56]

The wording of interpretative reports is critical, and it may be helpful to have standard comments reviewed by panels of clinicians. In one study of the quality of interpretative comments in clinical chemistry, the study authors found that a significant proportion of well-intended comments were considered to be misleading by a panel of clinicians.[57] A study of computer-based interpretative reporting of endocrine studies found that interpretative reports were valued more highly by clinicians when the interpretative comments were specific to the test result, stated as considerations rather than recommendations, and accompanied by an expanded list of differential diagnoses and a listing of drugs that are known to affect the test result.[58]

There are noteworthy models of whichspecialty societies representing both laboratory and clinical medicine have collaborated to create standards for reporting that define laboratory results that should be ignored, those that should be followed, and those that require action. One example is the reporting of atypical squamous or glandular cells in gynecologic cytology specimens.[59–61] Another example is the interpretation of weakly or moderately positive antinuclear antibody (ANA) tests.[62] As treatment options and the technology that is used to prepare specimens for interpretation continue to evolve, such published standards must be dynamic in order to remain clinically useful.[63,64]

Novel diagnostic tests represent a particular challenge. Longitudinal algorithms that compare the new test with previous methods are required to validate new markers. Longitudinal algorithms have been successfully employed for prostate-specific antigen and CA125 assays for the detection of prostate and ovarian cancer, and the parametric empirical Bayes screening algorithm has been proposed as a method for evaluating new tumor markers.[65]

Laboratory results that should have been ignored commonly trigger a cascade of unnecessary testing. At the worst, they result in unnecessary surgery or the inappropriate use of a medication. Improvements in quality can be gained by addressing the entire testing cycle, both in clinical and anatomic pathology. Unambiguous communication with the clinician and common standards of language are essential in order to avoid bad outcomes. During training, clinicians are often admonished not to ignore laboratory results, and clinicians are naturally cautious when dealing with conditions that are potentially serious. A commonly cited motive for ordering additional laboratory tests is the perceived need to reassure the patient through further testing.[66] Interpretative reports and clinical decision support pathways can reduce unnecessary testing without adversely affecting patient outcomes.[43, 67–74]

The College of American Pathologists continues to address questions of sensitivity, specificity, and predictive value of laboratory tests.[73] Clinical specialties have begun to address the undesirable consequences of incidental test results.[74] The time is ripe for joint efforts to improve quality. The establishment of standards will require significant investment by all stakeholders, but represents an opportunity to improve patient outcomes and reduce healthcare costs.