Multiple studies have evaluated copeptin, a surrogate for arginine vasopressin, in the diagnosis of acute myocardial infarction (AMI) with mixed results. A systematic review and collaborative meta-analysis were performed for diagnosis of AMI and assessment of prognosis in patients presenting to the emergency department with chest pain. MEDLINE/PubMed, Cochrane CENTRAL, and EMBASE were searched for studies assessing copeptin in such patients. Study investigators were contacted, and many provided previously unpublished data. Random-effects methods were used to compare the data for copeptin, troponin, and their combination. There were a total of 9,244 patients from the 14 included studies. Mean age was 62 years; 64% were men; and 18.4% were ultimately diagnosed with AMI. Patients with AMI had a higher presentation copeptin level than those without AMI (22.8 vs 8.3 pmol/L, respectively, p <0.001). Although troponin had better diagnostic accuracy than copeptin for AMI, the combination of copeptin and troponin significantly improved the sensitivity (0.905 [0.888 to 0.921] vs 0.686 [0.661 to 0.710], respectively, p <0.001) and negative predictive value (0.97 [0.964 to 0.975] vs 0.93 [0.924 to 0.936], respectively, p <0.001) compared with troponin alone. Elevation in copeptin carried a similar risk of all-cause mortality to an elevation in troponin (odds ratio 5.84 vs 6.74, respectively, p = 0.67). In conclusion, copeptin not only identifies patients at risk of all-cause mortality, but its addition to troponin improved the sensitivity and negative likelihood ratio for diagnosis of AMI compared with troponin alone. Thus, copeptin may help identify patients who may be safely discharged early from the emergency department.
Acute myocardial infarction (AMI) remains a leading cause of morbidity and mortality. Exclusion of AMI is challenging because biomarkers such as troponin often have a delayed increase, which requires that the patients remain under observation in the emergency department until AMI can be confidently excluded. Arginine vasopressin plays a critical role in renal water retention and regulation of vascular tone. In the setting of increased blood osmolality or hypotension, the hypothalamus releases pre-provasopressin, which is broken down into arginine vasopressin and 2 larger peptides known as neurophysin II and copeptin. Although arginine vasopressin has a very short half-life, copeptin is a stable 39-amino acid glycopeptide arising from the C-terminal portion of provasopressin and provides a better surrogate measurement for arginine vasopressin. Copeptin is significantly elevated in patients with AMI, heart failure, shock, and other life-threatening conditions and also predicts outcome following AMI and heart failure. Although several studies have assessed copeptin in patients presenting to the emergency department with chest pain, uncertainty remains regarding the additional value of copeptin adds to diagnostic accuracy and prognostic risk stratification beyond the use of troponin alone. We therefore performed a diagnostic and prognostic systematic review and collaborative meta-analysis to assess the role of copeptin for AMI in patients with chest pain presenting to the emergency department.
Methods
Prespecified inclusion and exclusion criteria were established at the outset of the study. We included any study that (1) assessed patients who presented to the emergency department with nontraumatic chest pain and (2) measured copeptin levels. We excluded any study that (1) limited patients to those only with myocardial infarction or a specific subgroup of patients, (2) excluded patients with an initial positive troponin, and (3) utilized a case-control format. We included studies regardless of whether patients with ST-segment elevation myocardial infarction were included or excluded.
Two independent reviewers (MJL and ROE) systematically searched (September 2013) Cochrane CENTRAL, EMBASE, and MEDLINE/PubMed for studies that assessed copeptin in patients presenting to the emergency department with nontraumatic chest pain. In addition, we consulted experts, reviewed citations from eligible studies, and explored “see related articles” section for key publications in MEDLINE/PubMed. We limited our search to studies published in peer-reviewed journals and thus excluded trials presented only in the abstract form. No language restriction was applied. Our systematic review and meta-analysis were performed in accordance with Meta-analysis Of Observational Studies in Epidemiology (MOOSE) and Preferred Reporting Items for Systematic reviews and Meta-Analysis (PRISMA) guidelines. After obtaining full reports, eligibility was assessed from the full-text articles with divergences resolved after consensus. The retrieved studies were carefully assessed to exclude potentially duplicate or overlapping data. Studies were appraised in accordance with QUality Assessment of Diagnostic Accuracy Studies (QUADAS-2).
Data were abstracted by the same 2 investigators (MJL and ROE). Area under the receiver-operating characteristic (ROC) curves for individual studies and data regarding true-positive, false-positive, false-negative, and true-negative results were obtained. The clinical outcome of interest was all-cause mortality during follow-up. An attempt was made to contact the corresponding authors of all pertinent studies to obtain complete data, including data at the level of the subject. With previously unpublished data provided by the study investigators, we were able to look at variables, such as the use of different cut-points for copeptin and both conventional and high-sensitivity troponins.
Dichotomous variables are reported as proportions (percentages), whereas continuous variables are reported as means (SD) or medians. Measures of diagnostic accuracy are reported as point estimates (with 95% confidence intervals). Sensitivity, specificity, positive predictive values (PPVs), negative predictive values (NPVs), positive and negative likelihood ratios, and diagnostic odds ratios were computed using true-positive, true-negative, false-positive, and false-negative rates. Positive likelihood ratio (sensitivity/[1 − specificity]) provides an estimate of the probability of a positive test in a patient with disease divided by the probability of a person without disease testing positive. Negative likelihood ratio ([1 − sensitivity]/specificity) gives an estimate of the probability of a person with disease who tested negative divided by the probability of a person without disease who tested negative. Although sensitivity and specificity are impacted by disease prevalence, likelihood ratios are believed to be independent of prevalence rates. Furthermore, a positive likelihood ratio >10 and a negative likelihood ratio <0.1 provide evidence of satisfactory diagnostic performance. These variables can then be combined to form a single variable termed the diagnostic odds ratio (positive likelihood ratio/negative likelihood ratio), which represents how high the odds are of having the disease for a positive test compared with having the disease for a negative test.
Individual study statistics were calculated and then combined using random-effects methods, weighting each point estimate by the inverse sum of its variance and the between-study variance. We also generated weighted symmetric summary ROC plots with corresponding area under the summary ROC curves using the Moses-Shapiro-Littenberg method. Area under the ROC curves of individual studies were pooled using a random-effects generic-inverse variance method, providing summary point estimates with 95% confidence intervals. Sources of clinical and statistical heterogeneity were explored by means of subgroup analyses and metaregression with unrestricted maximum-likelihood metaregression (inverse variance-weighted regression) on sensitivity, specificity, and NPV separately for copeptin, troponin, and their combination using comprehensive meta-analysis.
Binary outcomes from individual studies were combined with random-effects model, leading to computations of odds ratios with 95% confidence intervals. Between-study statistical heterogeneity was assessed using the Cochran Q chi-square test. I 2 was calculated as a measure of statistical heterogeneity, with I 2 values of 25%, 50%, and 75% representing mild, moderate, and severe inconsistency, respectively. Small study or publication bias was explored with funnel plots, Egger’s test, and Peter’s test. Finally, sensitivity analyses (including exclusion of 1 study at a time) were conducted to explore heterogeneity. Statistical analysis was performed using Review Manager (RevMan) 5 version 5.1.7 freeware package (Copenhagen: The Nordic Cochrane Center, The Cochrane Collaboration, 2008), Meta-DiSc software (Department of Clinical Biostatistics, Ramón y Cajal Hospital, Madrid, Spain), SPSS 11.0 (SPSS, Chicago, Illinois), and NCSS 2007 (Kaysville, Utah), with statistical significance for hypothesis testing set at the 0.05 2-tailed level and for heterogeneity testing at the 0.10 2-tailed level.
Results
Of 322 citations on copeptin, we assessed 66 abstracts from which we performed detailed review of 35 full-text manuscripts. We excluded 31 abstracts because they pertained to heart failure, pulmonary embolus, or other disease states that cause chest pain but do not meet criteria for enrollment in our study. AMI was not the outcome of interest for the study. After evaluation based on prespecified inclusion and exclusion criteria, 22 articles were excluded because of the performance of the study on specific patient populations, inclusion of patients only with AMI, exclusion of patients with initially positive troponins, patient duplication, and a case-control design. However, the investigators of the COPeptin in Emergency Department (COPED) study (OM and PL), which excluded patients with initially elevated troponin, were contacted and subsequently provided comprehensive data. Thus, our systematic review and meta-analysis comprise data from 13 published studies and previously unpublished data from the COPED study, along with additionally unpublished data from other groups.
The details of our flow diagram are shown in Figure 1 . Study characteristics are listed in Table 1 , and appraisal of diagnostic accuracy study quality is listed in Table 2 . All studies defined AMI using the universal definition with the majority providing physician adjudication. The 14 studies included 9,244 patients totally (average of 660 patients [range 58 to 1,967]) with a weighted mean age of 62 ± 15 years; 64% of patients were men. The population also had a typical distribution of cardiovascular risk factors: 36% with known coronary artery disease, 27% with previous myocardial infarction, 63% with systemic hypertension, 51% with hyperlipidemia, 23% with diabetes mellitus, and 29% currently smokers. The average copeptin level on admission was 9.7 pmol/L, and the prevalence of AMI was 18.4% (range of 7.9% to 46.5%).
| Study | Year Published | Patients (n) | Multicenter | Inclusion Criteria for Chest Pain to Presentation (h) | Assay Cut-Points | Prognosis (Days Follow-Up) | |||
|---|---|---|---|---|---|---|---|---|---|
| cTnT (ng/L) | cTnI (ng/L) | Hs-cTnT or I (∗) (ng/L) | Copeptin (pmol/L) | ||||||
| Afzali | 2013 | 230 | No | None | Siemens Ultra ADVIA-Centaur XP, 40 | 14 | Yes (180) | ||
| Balmelli/APACE | 2013 | 1,247 | Yes | <12 | Roche 4th-gen, 35 | Siemens RxL, 140 (defined AMI) | Roche, 14 | 10 or 14 | Yes (360) |
| Charpentier | 2012 | 641 | No | <12 | Siemens ADVIA-Centaur, 100 | 14 | Yes (30) | ||
| Chenevier-Gobeaux | 2013 | 317 | Yes | <6 | Siemens Xpand HM, 140 or Beckman Coulter Access, 60 | Roche, 14 | 10.7 or 14 | No | |
| COPED | NA | 1,242 | Yes | <12 | Roche 4th-gen, 30 | 14 | No | ||
| Eggers | 2012 | 367 | Yes | <8 | Siemens Stratus CS, 70 | Roche, 14 | 14 | Yes (2,136) | |
| Folli | 2013 | 472 | Yes | <8 | NR | 14 | No | ||
| Giannitsis | 2011 | 503 | No | None | Roche, 14 | 10 or 14 | No | ||
| Keller | 2010 | 1,386 | Yes | None | Roche 4th-gen, 30 | Siemens RxL, 140 (defined AMI) | Siemens Sensitive Ultra, 40∗ | 13 | No |
| Lotze | 2011 | 142 | No | None | Roche 4th-gen, 100 | Roche, 14 | 14 | Yes (<10) | |
| Maisel/CHOPIN | 2013 | 1,967 | Yes | <6 | Siemens Ultra ADVIA-Centaur XP, 40 | 14 | Yes (180) | ||
| Meune | 2011 | 58 | No | <6 | Siemens Xpand HM, 140 | Roche, 14 | 10 or 14 | No | |
| Sebbane | 2013 | 194 | No | <12 | Beckman Access2, 40 | Roche, 14 | 13.1 or 14 | No | |
| Thelin | 2013 | 478 | No | None | Roche, 14 | 14 | Yes (60) | ||
| Study | Standard Copeptin Assay | Prespecified Copeptin Cut-Point | Consecutive Patient Inclusion | Withdrawals Reported | Copeptin Interpretation Independent of Troponin |
|---|---|---|---|---|---|
| Afzali | Yes | Yes | Yes | Yes | Yes |
| Balmelli/APACE | Yes | Yes | Yes | Yes | Yes |
| Charpentier | Yes | Yes | Yes | Yes | Yes |
| Chenevier-Gobeaux | Yes | No | Yes | Yes | Yes |
| COPED | Yes | Yes | Yes | Yes | Yes |
| Eggers | Yes | Yes | Yes | No | Yes |
| Folli | Yes | Yes | Yes | Yes | Yes |
| Giannitsis | Yes | Yes | Yes | Yes | Yes |
| Keller | Yes | Yes | Yes | Yes | Yes |
| Lotze | Yes | Yes | Yes | Yes | Yes |
| Maisel/CHOPIN | Yes | Yes | Yes | Yes | Yes |
| Meune | Yes | Yes | Yes | Yes | Yes |
| Sebbane | Yes | No | Yes | Yes | Yes |
| Thelin | Yes | Yes | Yes | Yes | Yes |
Baseline characteristics of the patients are listed in Table 3 . The number of true-positive, false-positive, false-negative, and true-negative values with corresponding sensitivities, specificities, PPVs, and NPVs for AMI is provided based on the cut-points for copeptin ( Supplemental Table 1 ), initial high-sensitivity and conventional troponin ( Supplemental Table 2 ), and their combination ( Supplemental Table 3 ). The area under the ROC curves for copeptin, initial high-sensitivity or conventional troponin, and their combination is listed in Supplemental Table 4 .
| Study | Age (yrs) | Men (%) | Previous CAD (%) | Previous MI (%) | HTN (%) | HLD (%) | DM (%) | Smokin g (%) | STEMI (%) | NSTEMI (%) | AMI (%) | UA (%) | Average Copeptin (pmol/L) | Copeptin in AMI (pmol/L) | Copeptin in No AMI (pmol/L) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Afzali | 65 ± 15 | 71.7 | 43.0 | 30.4 | 76.1 | 62.2 | 22.2 | 37.0 | 10.4 | 36.1 | 46.5 | NR | 20.8 ± 71.5 | 75.1 ± 208 | 12.2 ± 20.8 |
| Balmelli/APACE | 64 (51–76) | 66.3 | 36.7 | 24.5 | 63.6 | 44.7 | 20.6 | 24.4 | 4.0 | 11.9 | 15.9 | 14.0 | 6.8 (3.5–15.8) | 18.3 (7.1–45.3) | 5.7 (3.2–12.4) |
| Charpentier | 58 ± 16 | 66.1 | 30.7 | NR | 42.6 | 39.9 | 12.8 | 28.9 | 0 | 14.8 | 14.8 | 13.3 | 9.9 (5–20) | 22.8 (10.3–54.1) | 7.6 (5–14.8) |
| Chenevier–Gobeaux | 57 ± 17 | 65.0 | 26.0 | NR | 37.0 | 36.0 | 14.0 | 40.0 | 4.1 | 10.1 | 14.2 | 3.47 | 10.9 (5.7–20.7) | 23.2 (10.7–133.6) | 9.9 (5.2–17.2) |
| COPED | 67 ± 15 | 63.6 | NR | 30.7 | 67.1 | 51.6 | 28.9 | 29.3 | 0 | 26.7 | 26.7 | NR | 12.9 (5–20.1) | 19.9 (5–44.4) | 11.3 (5–15.7) |
| Eggers | 66 ± 12 | 65.6 | 45.3 | 37.5 | 42.8 | 38.3 | 18.3 | 18.1 | 0 | 34.9 | 34.9 | 18.5 | 7.6 (4.2–20) | 11.3 (5.3–31.6) | 6.1 (3.8–14.3) |
| Folli | 60 ± 15 | 65.0 | 30.0 | 19.0 | 52.0 | 35.0 | 13.0 | 22.0 | 5.9 | 5.9 | 11.9 | 9.1 | 10.5 (NR) | 27.4 (NR) | 8.2 (NR) |
| Giannitsis | 63 ± 16 | 63.0 | 44.3 | NR | 74.8 | 58.2 | 26.8 | 49.8 | 9.7 | 17.3 | 27.0 | NR | 8 (2.5–22.3) | 17.5 (7.5–59.3) | 6.5 (2.5–17) |
| Keller | 62 ± 13 | 66.4 | 37.2 | NR | 74.0 | 72.2 | 18.2 | 23.5 | 6.7 | 14.9 | 21.6 | 13.3 | 6.7 (3.3–16.5) | 18.5 (7–50.1) | 5.3 (2.9–11.9) |
| Lotze | 71 ± 14 | 76.0 | 27.5 | 15.5 | 73.9 | 16.9 | 28.9 | 7.7 | 6.3 | 2.8 | 9.2 | 2.1 | 15.7 (7.3–62.9) | 30.4 (5–139) | 14.9 (7.3–59.1) |
| Maisel/CHOPIN | 56 ± 13 | 56.8 | 38.5 | 27.9 | 69.8 | 55.9 | 28.8 | 33.1 | 2.0 | 5.9 | 7.9 | 6.4 | 9.7 (2.5–19) | 19.9 (9.2–63.5) | 9.2 (2.5–17.2) |
| Meune | 58 ± 14 | 63.8 | NR | 20.7 | 46.7 | 37.9 | 22.4 | 32.8 | 0 | 22.4 | 22.4 | 29.3 | 11.5 (6.2–23.8) | 15.3 (7.8–50.7) | 10.7 (5.5–20.6) |
| Sebbane | 61 ± 17 | 63.4 | 21.6 | 14.8 | 34 | 35.1 | 14.1 | 36.6 | 13.9 | 12.9 | 26.8 | 16.0 | 9.7 (5.3–25.7) | 30.4 (12.9–112.6) | 7.5 (4.4–16.0) |
| Thelin | 66 (55–76) | 63.0 | 39.0 | NR | 54.0 | 39.0 | 21.0 | 17.0 | 0 | 14.6 | 14.6 | 7.7 | 12 (7–21) | 17 (9–44) | 12 (NR) |
| Weighted mean | 62 ± 15 | 63.6 | 36.6 | 27.0 | 63.0 | 51.0 | 22.7 | 28.6 | 3.6 | 14.8 | 18.4 | 10.7 | 9.7 | 22.8 | 8.3 |
Patients with AMI had significantly greater copeptin levels on presentation than did patients without AMI (weighted average 22.8 pmol/L [22.2 to 23.5] vs 8.3 pmol/L [8.2 to 8.4], respectively, p <0.001). Use of a copeptin cut-point of 14 pmol/L for AMI had a significantly lower PPV compared with a cut-point of <14 pmol/L (0.304 [0.287 to 0.321] vs 0.348 [0.323 to 0.373], p <0.005, respectively). Otherwise, there were no differences in sensitivity (0.637 [0.610 to 0.663] vs 0.659 [0.623 to 0.693], respectively), specificity (0.675 [0.663 to 0.687] vs 0.690 [0.673 to 0.707], respectively), NPV (0.893 [0.883 to 0.902] vs 0.890 [0.876 to 0.902], respectively), positive likelihood ratio (1.993 [1.759 to 2.258] vs 2.015 [1.721 to 2.360], respectively), negative likelihood ratio (0.522 [0.440 to 0.619] vs 0.485 [0.402 to 0.585], respectively), diagnostic odds ratio (4.107 [3.046 to 5.538] vs 4.499 [3.536 to 5.725], respectively), or area under the summary ROC curves (0.727 [0.690 to 0.764] vs 0.735 [0.711 to 0.760]). These data suggest that a cut-point of <14 pmol/L, such as 10 pmol/L, could be adopted without compromising the diagnostic accuracy. Metaregression with weighted inverse variance found that the use of the different cut-points did not significantly impact diagnostic accuracy (p = 0.60).
The summary diagnostic characteristics of the 13 studies with cut-point data ( Supplemental Table 1 ) are listed in Table 4 , and they had significantly elevated heterogeneity and inconsistency (p <0.003 and >62% for all values, respectively), which is common among diagnostic meta-analyses. Neither Egger’s test nor Peter’s test demonstrated evidence of publication bias utilizing the diagnostic odds ratio (p >0.10). As listed in Table 4 , the initial measurement of both conventional and high-sensitivity troponins had significantly superior diagnostic performances for AMI in all categories compared with copeptin alone. When limited to studies assessing high-sensitivity troponin, copeptin had a sensitivity of 0.626 [0.595 to 0.657], specificity of 0.679 [0.664 to 0.695], PPV of 0.333 [0.311 to 0.356], NPV of 0.877 [0.864 to 0.889], positive likelihood ratio of 1.991 [1.764 to 2.248], negative likelihood ratio of 0.521 [0.423 to 0.642], diagnostic odds ratio of 4.140 [3.095 to 5.538], and an area under the summary ROC curve of 0.734 [0.700 to 0.768], again demonstrating that high-sensitivity troponin is statistically superior to copeptin in all diagnostic categories. Among studies that assessed high-sensitivity troponin presenting area under the ROC curves ( Supplemental Table 4 ), the initial high-sensitivity troponin had a significantly higher diagnostic accuracy than copeptin for AMI (pooled area under the curve 0.91 [0.87 to 0.94] vs 0.69 [0.63 to 0.75], respectively, p <0.001, I 2 = 97.5%).
| Initial Copeptin | Initial Conventional Troponin | Initial HS-Troponin | Copeptin and Conventional Troponin | Copeptin and HS-Troponin | |
|---|---|---|---|---|---|
| Pooled sensitivity | 0.641 (0.617–0.665) | 0.686 (0.661–0.710) | 0.878 (0.855–0.898) | 0.905 (0.888–0.921) | 0.957 (0.943–0.969) |
| Pooled specificity | 0.670 (0.659–0.681) | 0.947 (0.941–0.952) | 0.787 (0.773–0.800) | 0.655 (0.642–0.667) | 0.592 (0.575–0.608) |
| Pooled PPV | 0.310 (0.294–0.326) | 0.744 (0.720–0.768) | 0.513 (0.489–0.538) | 0.359 (0.343–0.376) | 0.377 (0.357–0.396) |
| Pooled NPV | 0.890 (0.881–0.898) | 0.930 (0.924–0.936) | 0.962 (0.954–0.968) | 0.970 (0.964–0.975) | 0.982 (0.975–0.987) |
| Summary positive likelihood ratio | 1.975 (1.776–2.197) | 15.425 (8.094–29.397) | 3.847 (2.752–5.378) | 2.584 (2.243–2.976) | 2.235 (1.913–2.611) |
| Summary negative likelihood ratio | 0.510 (0.435–0.598) | 0.322 (0.281–0.370) | 0.157 (0.109–0.225) | 0.148 (0.119–0.183) | 0.065 (0.032–0.135) |
| Summary diagnostic odds ratio | 4.150 (3.196–5.389) | 52.604 (33.824–81.813) | 27.551 (15.013–50.560) | 19.523 (15.042–25.339) | 35.826 (18.142–70.750) |
| Area under the summary ROC curve | 0.724 (0.692–0.756) | 0.885 (0.855–0.916) | 0.912 (0.870–0.954) | 0.856 (0.789–0.924) | 0.795 (0.648–0.941) |
The initial measurement of conventional troponin also had significantly superior diagnostic performance for AMI compared with copeptin in all categories ( Table 4 ). For studies that assessed conventional troponin, copeptin had a sensitivity of 0.633 [0.607 to 0.658], specificity of 0.678 [0.666 to 0.690], PPV of 0.308 [0.292 to 0.326], NPV of 0.891 [0.881 to 0.899], positive likelihood ratio of 1.975 [1.729 to 2.256], negative likelihood ratio of 0.521 [0.444 to 0.611], diagnostic odds ratio of 3.957 [3.023 to 5.180], and an area under the summary ROC curve of 0.719 [0.682 to 0.756], which confirm that conventional troponin is statistically superior to copeptin in all diagnostic categories. Among studies that assessed conventional troponin presenting area under the ROC curves ( Supplemental Table 4 ), the initial conventional troponin had a significantly higher diagnostic accuracy than copeptin for AMI (pooled area under the curve 0.86 [0.83 to 0.89] vs 0.71 [0.66 to 0.76], respectively, p <0.001, I 2 = 95.9%).
Also as listed in Table 4 , the addition of copeptin to either the initial high-sensitivity or conventional troponin significantly improved the sensitivity, NPV, and negative likelihood ratio compared with either the initial high-sensitivity or conventional troponin alone. However, the combination significantly reduced the specificity, PPV, and positive likelihood ratio compared with troponin alone. Additionally, there are no significant differences in the area under the summary ROC curves for the combination of copeptin and the initial troponin measurement compared with that of the initial troponin measurement alone ( Table 4 ), suggesting that the improvement in sensitivity and NPV is balanced by a trend toward overall lower diagnostic accuracy. The combination of copeptin and the initial conventional troponin had significantly better diagnostic accuracy for AMI as assessed by the pooled area under the ROC curves compared with conventional troponin alone (0.91 [0.89 to 0.94] vs 0.86 [0.82 to 0.90], respectively, p = 0.02, I 2 = 83%). However, the combination of copeptin and the initial high-sensitivity troponin did not significantly improve the diagnostic accuracy for AMI as assessed by the pooled area under the ROC curves compared with the initial high-sensitivity troponin alone (0.92 [0.88 to 0.95] vs 0.91 [0.87 to 0.94], respectively, p = 0.71, I 2 0%). Figure 2 shows the pooled NPV for copeptin alone, troponin alone, and their combination based on which troponin assay was employed, demonstrating the benefit of combining copeptin and troponin with the goal of safely discharging patients early from the emergency department.
