Overuse of troponin? A comprehensive evaluation of testing in a large hospital system
Troponin assays are integral to the diagnosis of acute myocardial infarction (AMI), but there is concern that testing is over utilized and may not conform to published guidelines. We reviewed all testing performed at 14 hospitals over 12 months and associated troponin values with the primary and secondary diagnoses for each visit. Troponin was determined to be negative, indeterminate or elevated based on reference ranges. The majority of troponin measurements were single, not serial (64%). The rate of AMI was low, with only 3.5% of tested patients having a primary or secondary diagnosis of AMI. Sensitivity, specificity and negative predictive value were excellent, exceeding 90%. However, positive predictive value was low, suggesting testing of populations with diseases known to be associated with elevated troponin levels in the absence of AMI. The majority (79%) of elevated troponin values were associated with primary diagnoses other than AMI. Only 28% of elevated troponins were associated with a primary or secondary diagnosis of AMI. These data suggest possible overuse of troponin testing in our healthcare system. Journal of Hospital Medicine 2017;12:329-331. © 2017 Society of Hospital Medicine
© 2017 Society of Hospital Medicine
The ability of serum troponin measurement in the diagnosis of acute myocardial infarction (AMI) was validated in patients with at least a moderate pretest probability for the disease.1 The diagnostic yield of troponin testing in clinical trials has been between 20% and 50%, excluded patients thought unlikely to have AMI. In practice, physicians often encounter low-risk patients and patients in whom the diagnosis on initial presentation is unclear. Several noncardiac diagnoses, such as pneumonia and respiratory failure, are associated with an elevated troponin level in the absence of AMI, but patients can present with symptoms similar or identical to those of patients who present with AMI.2-4 Elevated troponin level in sepsis has been associated with worsened prognosis, though there is no evidence that this finding alters management. An American College of Cardiology Foundation opinion published in 2012 expressly recommends against troponin testing in patients with sepsis.4
The only guideline-based indication for troponin testing is the diagnosis or exclusion of AMI.5 We conducted a comprehensive review of troponin testing in our healthcare system to see whether testing might be used in clinical settings in which AMI was unlikely.
METHODS
We retrospectively obtained data on all visits to 14 hospitals in an integrated healthcare system in Texas between June 2013 and June 2014. We analyzed data for all hospital encounters during which a troponin assay was ordered and a troponin level reported—including qualitative point-of-care assays and quantitative laboratory troponin I measurements. We identified 93,436 visits. Quantitative measurements were divided into negative (<0.05 ng/mL), indeterminate (0.05-0.09 ng/mL), and elevated (>0.09 ng/mL), based on the reference ranges reported to physicians. We associated troponin levels with ICD-9 (International Classification of Diseases, Ninth Revision) primary and secondary diagnoses, grouping ICD-9 codes 410 (AMI), 411 (other acute or subacute forms of ischemic heart disease [IHD]), 412 (old myocardial infarction), 413 (angina pectoris), and 414 (other forms of chronic IHD) as representing IHD diagnoses.
To further evaluate troponin testing, we constructed 2 contingency matrices (Table).6 We included visits for which both primary and secondary diagnoses were available for review and for which quantitative troponin I measurements were available; 92,445 encounters met criteria for inclusion in matrix calculations. In the first matrix (part A of Table), a primary diagnosis of any AMI (ICD-9 code 410) was used as “positive” and all others “negative.” In the second matrix (part B of Table), “positive” includes any primary or secondary diagnosis of AMI.
RESULTS
We identified a total of 93,436 hospital visits associated with troponin testing; 179,239 troponin measurements were associated with these visits (an average of 1.81 per encounter). Of these visits, 59,897 (64.1%) were associated with a single measurement. Of the 179,239 measurements, 147,051 (82.1%) were negative, 21,881 (12.1%) indeterminate, and 10,307 (5.8%) positive. Primary diagnoses of hypertension, dizziness, abdominal pain, anxiety, dehydration, and headache associated with troponin testing comprised 6127 encounters and had no associated elevated troponin levels. Several non-cardiac primary diagnoses were associated with significant numbers of elevated troponin values including septicemia (27%), acute respiratory failure (28%), and cerebrovascular accident (10%). Seventy-six percent of encounters associated with troponin testing had no primary or secondary IHD diagnosis. Only 2% of 16,941 visits with a primary diagnosis of chest pain were associated with abnormal troponin levels (Figure).
