Recognizing the Potential for Overdiagnosis: Are High-Sensitivity Cardiac Troponin Assays an Example?

Abstract

Overdiagnosis occurs when persons are labeled with a technically correct diagnosis that does not improve health outcomes (1). It is a common unintended consequence of early disease detection in asymptomatic persons but also takes place in persons with mild or ambiguous symptoms. Despite initiatives to prevent overdiagnosis (2), new tests and changes to disease definitions may become established in clinical practice before their potential for overdiagnosis is recognized; as such, reverting to previous approaches may be difficult. In this commentary, we propose questions (Figure) to identify new practices with the potential for overdiagnosis and evaluate the example of using high-sensitivity cardiac troponin (hs-cTn) assays to diagnose type 1 myocardial infarction. Figure. Questions that help identify potential overdiagnosis. How the questions relate to subsets of the patient population who may be diagnosed with the condition is shown. Question 1 concerns patients who may be diagnosable with the new test or threshold, question 2 concerns patients who are newly diagnosed with the new test or threshold or already diagnosed using the old test or threshold, and questions 3 to 5 concern patients who are newly diagnosed compared with those already diagnosed. Questions Is There Potential for Increased Diagnosis? Overdiagnosis relies on the existence of a population reservoir of detectable disease and the availability of tests capable of unearthing cases from that reservoir (3). Increased diagnosis can take place when a new, sensitive test is added to the diagnostic pathway; a new, more sensitive test replaces a test in the diagnostic pathway; or the threshold of a test in the diagnostic pathway changes. Has Diagnosis Actually Increased? For overdiagnosis to take place, clinical action is needed to move from potential to actuality. If a new threshold for defining a disease is proposed but not widely adopted, additional cases will not be diagnosed. Overdiagnosis will not occur if more sensitive tests are rarely used, because additional cases in the reservoir will remain undetected. An indicator of overdiagnosis is therefore a higher rate of detection than that during a previous timeframe, in an alternative geographic region, or in another population where the new definition or test has not been used. Are Additional Cases Subclinical or Low Risk? To represent overdiagnosis as opposed to early and beneficial diagnosis, the additional cases detected must be subclinical orif symptomaticcharacterized by a low risk for important adverse clinical outcomes. An indicator that these cases are low risk is an apparent decrease in the case-fatality rate not clearly attributable to better treatment. This rate inevitably decreases in the presence of overdiagnosis, because the denominator (that is, the number of persons diagnosed) has been inflated with clinically insignificant cases (4). Have Some Additional Cases Been Treated? Overdiagnosis labels persons with diseases or conditions. Psychosocial harm may result, and unnecessary or potentially harmful treatment or invasive tests may be triggered. Assessing whether the additional cases have been treated helps determine the extent and type of harms from the overdiagnosis. Might Harms Outweigh Benefits? If the harms of detecting and treating the additional cases outweigh the benefits, then all conditions for overdiagnosis have been met. Establishing overdiagnosis may require evaluation of long-term outcomes that are important to patients. To allow an unbiased evaluation, keep in mind from the start that the harms of detecting additional cases may outweigh the benefits, and establish processes to collect data on both (harms as well as benefits). Example High-sensitivity cardiac troponin assays are used to assess patients with chest pain who are suspected of having myocardial infarction. They have been used in Europe and Australasia since 2010 and were approved for use in the United States in 2017. The Potential for Increased Diagnosis Troponin is a normal circulating biomarker, and high levels are sometimes found in asymptomatic persons. The threshold for elevated hs-cTn (the 99th percentile upper reference limit) was established primarily from populations of healthy persons without comorbidities. Approximately 2% of adults (and 5% of those aged 60 to 65 years) without symptoms of acute ischemia have a troponin level higher than the threshold, providing a reservoir of detectable cases (5). Increased Incidence of Disease A recent trial (High-STEAC [High-Sensitivity Troponin in the Evaluation of patients with suspected Acute Coronary Syndrome]) involved 48282 patients with chest pain. It showed that use of hs-cTn assays increased the detection of myocardial injury by 3.7% (from 8589 to 10360 persons) and the diagnosis of type 1 myocardial infarction by 1.2% (from approximately 5153 to 5738 persons) compared with contemporary cardiac troponin assays (6). Subclinical or Low Risk Elevated hs-cTn levels are associated with an increased risk for death. For persons diagnosed with type 1 myocardial infarction because of elevated hs-cTn levels whose hs-cTn levels were undetectable using the previous-generation cardiac troponin assay, the extent of this risk is uncertain. Overall, patients with myocardial injury in the High-STEAC trial who were reclassified by the results of hs-cTn assays had lower case-fatality rates from cardiovascular disease at 1 year than patients identified by the older test, indicating that the new test on average detects milder disease. The prognosis of the subset of reclassified patients who were labeled with type 1 myocardial infarction, and how their prognosis compared with that of patients with type 1 myocardial infarction identified by the older test, was not reported. Additional Treatment In the High-STEAC trial, patients reclassified with myocardial injury were more likely to have angiography in the implementation phase (when hs-cTn levels were reported) than the validation phase (when these levels were not reported). However, rates of percutaneous interventions did not differ. The proportion receiving preventive drug therapies (that is, antiplatelet and blood pressurelowering drugs and statins) increased. Balance of Harms and Benefits Current data are insufficient to determine whether the harms of reclassification with hs-cTn assays outweigh the benefits. In the High-STEAC trial, how reporting the results of these assays to clinicians affected the rates of subsequent myocardial infarction or cardiovascular death at 1 year was inconclusive (odds ratio for those reclassified in the validation versus implementation phase, 1.10 [95% CI, 0.75 to 1.61]). The average length of stay did decrease, suggesting that the test may be most useful to rule out myocardial injury rather than rule in myocardial infarction. Implications No randomized studies comparing health outcomes for patients with type 1 myocardial infarction diagnosed with and without hs-cTn assays were required or done before worldwide regulatory approval of the new test, which was based on its superior analytic performance. Our analysis underscores the need to better evaluate new tests before integration into usual care. The benefits from new tests may be smaller than expected, and there may be unanticipated harms from overuse and overdiagnosis. Our questions may help identify possible overdiagnosis sooner than standard surveillance of changes in clinical testing. As in our example, they may be unanswerable with current, publicly available data and further studies will be needed. These studies should focus on health outcomes for patients with discordant results between the old and new test (or definition) using robust designs with long follow-up (79). That the harms of detecting additional cases may outweigh the benefits must be considered so that processes to collect the data needed for unbiased evaluation of long-term outcomes are established early. Addressing the questions posed here may help identify changes in clinical diagnostic practice that warrant such an investment of time and resources.