- Email: [email protected]
Interpretation of exercise stress test recordings: Concordance between nurse practitioner and cardiologist
Interpretation of exercise stress test recordings: Concordance between nurse practitioner and cardiologist Evelyn Maier, RN, MN,a Louise Jensen, RN, PhD,b Brian Sonnenberg, MD, FRCP(C),a and Stephen Archer, MD, FRCP(C)a
AIM: Cardiology nurse practitioners (NPs) conduct exercise stress tests (ESTs) for diagnosis of cardiac disease. The diagnostic concordance of NPs to cardiologists has not been assessed. The hypothesis was that an NP is as reliable as a cardiologist in determining ST-segment depression, detecting arrhythmias, and making a diagnostic assessment. METHODS: An NP and two cardiologists (C1 and C2) were provided with 100 consecutive, anonymized ESTs, consisting of three 10-second, 12-lead tracings obtained at baseline, peak-exercise, and recovery. Interpretation was based on baseline rhythm, baseline and maximal exercise ST levels, arrhythmias, and global diagnosis (positive, negative, or inconclusive for ischemia). Raters used uniform criteria to interpret ESTs and were blinded to prior EST interpretation and computerized ST-segment analysis. RESULTS: There was similar concordance between the NP and cardiologists as between the cardiologists, measured by Kappa coefficients (rhythm: NP vs. C1 ⫽ .92, NP vs. C2 ⫽ .84, C1 vs. C2 ⫽ .84; arrhythmias: NP vs. C1 ⫽ .77, NP vs. C2 ⫽ .73, C1 vs. C2 ⫽ .75; EST diagnosis: NP vs. C1 ⫽ .75, NP vs. C2 ⫽ .73, C1 vs. C2 ⫽ .75). Pearson correlations demonstrated concordance for baseline ST levels (NP vs. C1 ⫽ .86, NP vs. C2 ⫽ .86, C1 vs. C2 ⫽ .90) and peak exercise ST levels (NP vs. C1 ⫽ .58, NP vs. C2 ⫽ .48, C1 vs. C2 ⫽ .67). CONCLUSIONS: Concordance among raters, and with the computer-generated algorithm, was moderate to high for all parameters of EST interpretation. This study lends support to NPs interpreting ESTs. (Heart Lung® 2008;37:144 –152.)
T
he need to provide quality health care and improve access has led to examining roles for nurse practitioners (NPs). The Division of Cardiology at the University of Alberta Hospital implemented a role for NPs in supervising and interpreting exercise stress tests (ESTs). There is a paucity of research examining the efficacy of NPs in cardiology and none assessing the diagnostic accuracy of EST interpretations. Lem et al1 first examined nurse supervised ESTs and concluded they could safely conduct ESTs, as judged by low morbidity and mortality rates associated with testing. This conclusion was
From the aDivision of Cardiology, University of Alberta Hospital; and bFaculty of Nursing, University of Alberta, Edmonton, Canada. Reprint requests: Louise Jensen, RN, PhD, University of Alberta, Faculty of Nursing, 4-112 D Clinical Sciences Building, Edmonton, AB, Canada T6G 2G3. 0147-9563/$ – see front matter Copyright © 2008 by Mosby, Inc. doi:10.1016/j.hrtlng.2007.05.009
144
www.heartandlung.org
supported in a larger retrospective audit2 of 17,467 ESTs performed on 5054 patients over 12 years. Nurse supervised ESTs were considered safe, based on a major complication rate of .06%, defined as patients requiring urgent intervention (sustained VT, defibrillation). Although the safety of nurse supervised ESTs has support, a complete EST requires not only safe conduct of the procedure but also the correct diagnosis of changes in rhythm, ST-segment deviation, and the synthesis of findings into the likelihood of ischemia. Thus, the expanded role for NPs includes supervising as well as interpreting ESTs. In studies examining provision of certain aspects of care for specific patient populations, NPs and physicians have been shown to be comparable.3-9 NPs have also been shown to be thorough and comprehensive in evaluating and treating patients in a variety of clinical settings, including outpatient departments and physician’s offices,5,7,10 long
MARCH/APRIL 2008
HEART & LUNG
NP and cardiologist concordance on EST interpretation
term and rehabilitative care,8 emergency,9 critical care,11,12 orthopedics,13 surgical care,14 and pediatric and neonatal care.15 A large randomized controlled trial16 found no differences in diagnosis of functional capacity, and social or emotional function between groups of patients cared for by NPs and physicians. As the NP role evolves, the question of NP effectiveness in other areas of patient care arises.17-19
OBJECTIVE The objective of this study was to examine the reliability of an NP in evaluating ischemia on the basis of electrocardiogram (ECG) analysis during an EST. This evaluation was conducted by establishing the concordance among an NP and two cardiologists (C1 and C2), and to the computer-generated averaged measurements in a blinded study that provided each rater with identical ECG recordings and “real-world” variables available when conducting an EST. The hypothesis was that an NP would obtain results concordant with those of the cardiologists. Comparisons were made on (1) baseline rhythm and detection of arrhythmias and conduction abnormalities during exercise; (2) quantification of ST-segment levels at baseline and peak exercise; (3) diagnosis of the EST as positive, negative, or inconclusive for ischemia; and (4) ST levels determined by practitioners versus computer-generated at baseline and peak exercise.
METHODS The NP received training in EST interpretation from a Masters Degree in Nursing program, had been employed in the role for 2 years, and had 15 years experience in cardiology; C1 had 14 years experience, and C2 had 5 years experience in cardiology. A pilot study was done with five EST recordings to rehearse and standardize the criteria for interpretation of EST recordings. Several issues arose in the variability of interpretations of the EST recordings, such as which ECG lead to use for ST segment quantification, method for determining the PR segment, what to do in case of severe artifact at peak exercise, definition of left ventricular hypertrophy, and whether or not to determine ST levels for left bundle branch block (LBBB). Also, several variables were identified as being necessary to allow one to render a global interpretation (medications, blood pressure). For example, if the heart rate-systolic blood pressure product was less than 15,000 a “negative” EST would be interpreted as inconclusive, as would a “positive” EST in a patient taking
HEART & LUNG
VOL. 37, NO. 2
Maier et al
digoxin. Standard guidelines for interpretation of EST recordings were subsequently developed and used in the study. After ethical approval, consecutive ESTs were selected from those completed over 6 months at the University of Alberta Hospital. EST recordings were then obtained from Health Records, and the first 100 complete ESTs with interpretable ECGs (judged by the clinician who provided the final clinical diagnosis) that did not have an exclusion criterion constituted the sample. Exclusion criteria included (1) lack of at least one stable 12-lead tracing at baseline, at peak exercise, and during the first 5 minutes of recovery; (2) inability to obtain a clear, high-resolution photocopy of all ECG tracings; and (3) termination of the EST because of technical errors. The assistant photocopied the resting, peak exercise, and first two ECGs recorded in immediate recovery. Patient identification data were masked, and a numeric code was assigned. Raters were provided with the following information: patient’s age, medications, cardiovascular risk factors, and body mass index, as identified on the original EST requisition, as well as the reason for the EST request, protocol used during testing, test duration, reason for test termination, and presence of angina during the test. The EST recordings were then given to the raters who independently analyzed (1) baseline rhythm; (2) presence and type of arrhythmia or conduction abnormalities at baseline and during exercise; (3) baseline ST-segment level; and (4) STsegment level during peak exercise in the maximally abnormal lead. On the basis of this information and the clinical data provided, each provided an EST diagnosis of positive, negative, or inconclusive. The NP’s and cardiologists’ measurements of baseline and peak exercise ST-segment levels were later compared with computer-generated averaged measurements.
DEFINITION OF TERMS Measurement of ST segments Baseline and peak exercise ST-segments were measured in leads II, aVF, or V5-V6, using calipers or a transparent ruler with finely graded increments. Baseline ST-segment level was measured 80 ms after the J-point using the PR junction as a baseline.
EST diagnosis Positive EST. ESTs were positive for ischemia if during the exercise or in the immediate recovery period the subject developed 1 mm or more of flat or downsloping ST depression20 or 1.5 mm or more upsloping depression.21 ST levels were measured 80
www.heartandlung.org
145
Maier et al
NP and cardiologist concordance on EST interpretation
Table I Exercise stress test patient characteristics N
Cardiovascular risk factors None Hyperlipidemia High blood pressure Diabetes Family history Smoking, hyperlipidemia Hyperlipidemia, diabetes Hypertension, hyperlipidemia Smoking Hypertension, hyperlipidemia, diabetes Hypertension, hyperlipidemia, family history Hyperlipidemia, family history Medications None Beta-blocker ACEI Beta-blocker, ACEI Calcium channel blocker Nitrates Beta-blocker, nitrates Beta-blocker, calcium channel blocker, ACEI Beta-blocker, ACEI, digoxin Calcium channel blockers, nitrates Beta-blocker, calcium channel blocker ACEI, nitrates Beta-blockers, ACEI, nitrates
Percent
Table II Indications for exercise stress test and exercise stress test termination N*
Percent
46 45 19
46.2 45.5 19.2
7 5 2
7.1 5.1 2.0
1 1
1.0 1.0
1
1.0
1
1.0
1
1.0
47
47.5
25 8 4 4 3 3
25.3 8.1 4.0 4.0 3.0 3.0
2 1 1 1
2.0 1.0 1.0 1.0
1
1.0
43 22 9 9 4 3 3 2
43.0 22.0 9.0 9.0 4.0 3.0 3.0 2.0
1
1.0
1
1.0
1
1.0
Reason for EST Previous CAD Ischemia Ischemia, functional capacity Risk stratification Functional capacity Ischemia, arrhythmia Research Ischemic threshold, functional capacity Arrhythmias, syncope Medication response Motor vehicle license Reason for EST termination Fatigue and dyspnea Fatigue ST depression, pain Target heart rate Hypertension Angina Unable to keep up to treadmill Other pain Dyspnea Arrhythmia Paced rhythm
1 1
1.0 1.0
EST, Exercise stress test; CAD, coronary artery disease. *Some referrals had more than one indication.
54 18 6 6 5 3 2 2
54.0 18.0 6.0 6.0 5.0 3.0 2.0 2.0
1 1
1.0 1.0
1
1.0
ACEI, Angiotensin-converting enzyme inhibitor.
ms after the J point in a minimum of three consecutive heart-tracing beats. An EST was also considered positive for ischemia if ST-segment elevation 1 mm or more occurred in leads that did not have pathologic Q-waves, other than in leads aVR or V1.22 Negative EST. ESTs were negative when no markers of ischemia were noted and when the patient exercised more than 5 minutes and achieved 85% of maximal heart rate (calculated as 220 minus
146
www.heartandlung.org
age) or a heart rate-systolic blood pressure product greater than 15,000, provided no other confounding variables were noted that would make the test inconclusive (eg, the presence of an LBBB, the use of digoxin). Inconclusive EST. ESTs were deemed inconclusive when ST depression did not reach diagnostic criteria for ischemia, when the patient failed to achieve target heart rate-systolic blood pressure product greater than 15,000, or when the ECG was uninterpretable because of artifact. Otherwise pos-
MARCH/APRIL 2008
HEART & LUNG
NP and cardiologist concordance on EST interpretation
Maier et al
Table III Exercise stress test rhythm reported by raters Baseline EST rhythm Rhythm
NP (N)
C1 (N)
C2 (N)
Normal sinus rhythm Atrial fibrillation Right bundle branch block Left bundle branch block Paced rhythm Left ventricular hypertrophy Atrial fibrillation and left ventricular hypertrophy
87 1 4 2 1 4 1
87 1 4 3 1 3 1
85 1 4 3 1 5 1
Arrhythmias during EST Arrhythmias
NP (N)
C1 (N)
C2 (N)
PAC or PNC PVC Second-degree AV block PAC or PNC and PVC None
10 12 1 2 75
11 11 0 2 76
12 11 1 3 73
Diagnosis
Positive (N)
Negative (N)
Inconclusive (N)
NP C1 C2
8 20 12
53 54 56
29 26 32
EST diagnosis reported by raters Diagnosis
NP (N)
C1 (N)
C2 (N)
Positive Negative Inconclusive
18 53 29
20 54 26
12 56 32
EST, Exercise stress test; NP, nurse practitioner; C, cardiologist; PAC, premature atrial contractions; PNC, premature nodal contractions; PVC, premature ventricular contractions; AV, atrioventricular.
itive ESTs were considered inconclusive if the patient had taken digoxin within 1 week of testing, had left ventricular hypertrophy based on Estes criteria, had LBBB, or had 1 mm or more depression or elevation in lead V5 on the baseline ECG.
rhythm, arrhythmias during EST, and EST diagnosis. Pearson correlation coefficients with regression plots were conducted to assess agreement among raters on baseline ST and peak exercise ST-segment levels. High agreement was considered to be .80 or more, moderate .60 to .79, and low less than .6.
Data analysis Descriptive statistics (SPSS Version 12.0, Chicago, IL) were calculated for age, gender, cardiovascular risk factors, medications, EST protocol used, baseline rhythm, arrhythmias during EST, and ST segment levels. Kappa coefficients were calculated to assess agreement among raters for EST baseline
HEART & LUNG
VOL. 37, NO. 2
RESULTS Of the 100 ESTs, 99 used the Bruce Protocol; one used a Balke protocol. The mean age of patients was 58.9 ⫾ 2.05 years, and 68% were male. Cardiovascular risk factor profiles and current medications are listed in Table I. Mean EST length was 8.32 ⫾ .21
www.heartandlung.org
147
Maier et al
B
1.75 1.5 1.25 1 .75 .5 .25 0 -.25 -.5 -.75 -1.5
-1
-.5
.5
0
1.5
1
Y=.133+.722*X; R²=.732
Baseline ST Levels for NP (N=98)
Baseline ST Levels for C1 (N=97)
C
2
Baseline ST Levels for C2 (N=97)
Baseline ST Levels for C1 (N=97)
A
NP and cardiologist concordance on EST interpretation
1.5 1.25 1 .75 .5 .25 0 -.25 -.5 -.75 -1 -1.5
-1
-.5
0
.5
1
1.5
2
Y=.126+.767*X; R²=.744
Baseline ST Levels for NP (N=98)
1.75 1.5 1.25 1 .75 .5 .25 0 -.25 -.5 -.75
-1
-.75
-.5
-.25
0
.25
.5
.75
1
1.25 1.5
Y=.042+.855*X; R²=.812
Baseline ST Levels for C2 (N=97)
Fig 1 Concordance on Baseline ST Levels Between Practitioners.
minutes, with a range from 1 to 13.57 minutes. The most frequent indication for the EST was evaluation of ischemia for the patient with a known history of coronary artery disease (46.2%); the most frequent reason for EST termination was the combination of fatigue and dyspnea (47.5%) (Table II).
Concordance on baseline EST rhythm Baseline EST rhythms interpreted by raters were classified into eight categories (Table III). Of the 100 EST recordings, 87 had normal sinus rhythm, 4 had right bundle branch block, with a 100% concordance by raters. C1 and C2 noted LBBB for three EST recordings, whereas the NP noted LBBB for only two EST recordings. A discrepancy was also noted for left ventricular hypertrophy, with C2 choosing that category for five recordings, NP for four recordings, and C1 for 3 recordings. The Kappa coefficient was .92 (P ⬍ .001) between the NP and C1 for baseline rhythm; .84 (P ⬍ .001) between the NP and C2; and .84 (P ⬍ .001) between the C1 and C2.
148
www.heartandlung.org
Concordance on arrhythmias during the EST Five categories of arrhythmias during the EST were noted by raters (Table III). The majority of EST recordings had no arrhythmias; 75% noted by NP, 76% by C1, and 73% by C2. Second-degree heart block was identified only by the NP and C1, and not by C2. Kappa coefficient was .77 (P ⬍ .001) between NP and C1 on arrhythmias during the EST; .73 (P ⬍ .001) between NP and C2; and .75 (P ⬍ .001) between C1 and C2.
Concordance on ST-segment levels Baseline ST segment levels were not obtained if there was LBBB. The range of baseline ST-segment levels was from ⫺1.0 to 1.6 for the NP; from ⫺.70 to 1.6 for C1; and from ⫺1.7 to 2 for C2. The highest agreement was between C1 and C2 (r ⫽ .90, P ⬍ .001), followed by similar agreement between the NP and C1 and C2 (r ⫽ .86, P ⬍ .001, respectively) (Fig 1). The range of peak exercise ST-segment levels for the NP was ⫺2.5 to 2.5 mm. C1 had 3 missing values
MARCH/APRIL 2008
HEART & LUNG
NP and cardiologist concordance on EST interpretation
B
3
Peak ST Levels for NP (N=98)
Peak ST Levels for NP (N=98)
A
2 1 0 -1 -2 -3 -4
-5
-4
-3
-2
-1
0
1
2
3
Peak ST Levels for C1 (N=97)
Peak ST Levels for C1 (N=97)
C
Maier et al
3 2 1 0 -1 -2 -3 -4 -3
-2
-1
0
1
2
3
4
Y=-.061+.569*X; R²=.23
Peak ST Levels for C2 (N=96)
3 2 1 0 -1 -2 -3 -4 -5 -3
-2
-1
0
1
2
3
4
Y= -.027+.837*X; R²=.444
Peak ST Levels for C2 (N=96)
Fig 2 Concordance on Peak ST Levels Between Practitioners.
due to LBBB, with a range from ⫺4.0 to 2.5 mm; C2 had 4 missing values due to LBBB, with a range from ⫺2.50 to 3.40 mm. Highest agreement was between C1 and C2 (r ⫽ .67, P ⬍ .001), followed by NP and C1 (r ⫽ .58, P ⬍ .01), and NP and C2 (r ⫽ .48, P ⬍ .001) (Fig 2).
Concordance on EST diagnosis Table III shows the EST diagnosis chosen by each rater. Agreement was a Kappa coefficient of .75 (P ⫽ .000) between the NP and C1 on the EST diagnosis; .73 (P ⫽ .000) between NP and C2; and .75 (P ⫽ .000) between C1 and C2. In no case were ESTs read as positive by the cardiologists that were read as negative by the NP, and no ESTs were read as negative by the cardiologist that were read positive by the NP.
Concordance on computer generated-averaged ST segment data The computer-generated averaged measurements were analyzed for baseline ST-segment levels for all raters and for peak ST segment levels for the NP and
HEART & LUNG
VOL. 37, NO. 2
C1. Peak ST-segment levels were not available for a comparison with C2. Baseline ST-segment levels were unavailable or uninterpretable for 12 EST recordings. All recordings for baseline ST analysis were taken from the resting data for lead V5. A simple regression analysis was done for the following: computer-generated averaged baseline ST-segment levels versus levels recorded by the NP (r ⫽ .85, P ⬍ .001), versus levels recorded by C1 (r ⫽ .88, P ⬍ .001), versus levels recorded by C2 (r ⫽ .93, P ⬍ .001) (Fig 3). Peak exercise ST segment levels were obtained by each rater according to the lead he/she considered most significant in the first readable recording in recovery. Two raters indicated which lead was used for analysis; therefore a simple regression analysis was obtained for the NP and C1 and the computergenerated averaged levels. Not all computer-generated averaged levels were available to be analyzed (n ⫽ 84 for the NP and n ⫽ 76 for C1) because data were not recorded for that particular EST, records were not available on recall, or ST levels were uninterpretable, as in the case of an LBBB. Agreement was r ⫽ .80 (P ⬍ .001) for the peak ST-segment for the NP versus the
www.heartandlung.org
149
Maier et al
NP and cardiologist concordance on EST interpretation
A
B
2
2 1.5
Computer Baseline ST Levels (N=89)
Computer Baseline ST Levels (N=89)
1.5 1 .5 0 -.5
1 .5 0 -.5 -1
-1
-1.5
-1.5 -1
-1.5
0
-.5
1
.5
1.5
-.75
2
Y=.189+.72*X; R²=.719
-.25
0
.25
.5
.75
1.25
1
1.5
1.75
Y=.094+.903*X; R²=.775
Baseline ST Levels for NP (N=98)
C
-.5
Baseline ST Levels for C1 (N=97)
2
Computer Baseline ST Levels (N=89)
1.5 1 .5 0 -.5 -1 -1.5 -1
-.75
-.5
-.25
0
.25
.5
.75
1
1.25
1.5
Y=.095+.896*X; R²=.865
Baseline ST Levels for C2 (N=97)
Fig 3 Concordance on Baseline ST Levels Between Computer and Practitioners.
B
4
Computer Peak ST Levels
Computer Peak ST Levels (N=84)
A
3 2 1 0 -1 -2 -3
4 3 2 1 0 -1 -2 -3 -4
-4 -4
-3
-2
-1
0
1
2
Y=.29+.965*X; R²=.637
Peak ST Levels NP (N=98)
3
-5
-4
-3
-2
-1
0
1
2
3
Y=.357+1.013*X; R²=.823
Peak ST Levels for C1 (N=97)
Fig 4 Concordance on Peak ST Levels Between Computer and Practitioners.
computer algorithm and r ⫽ .91 (P ⬍ .001) for C1 versus the computer algorithm (Fig 4).
DISCUSSION An important finding is that there was high concordance between the NP and the cardiologists, both for global EST diagnosis and for quantitative rest and exercise ECG analysis. Findings on EST diagnosis showed moderate agreement among all raters. Agreement was high on ESTs that were clearly positive or clearly negative. The NP did not
150
www.heartandlung.org
misdiagnose a strongly positive or strongly negative EST recording, which could compromise the patient’s outcome. The variability that existed was a result of differences in the lead selected as demonstrating maximal ST depression and differences in interpretation of inconclusive ESTs. Concordance on baseline rhythm showed high agreement among raters. All raters were concordant on paced rhythm and atrial fibrillation as a baseline rhythm. Concordance was also noted on right bundle branch block rhythm. Agreement for arrhythmias
MARCH/APRIL 2008
HEART & LUNG
NP and cardiologist concordance on EST interpretation
during the EST was moderate for raters. The only variances were two failures by the NP to identify an LBBB; however, in one case this was missed by one cardiologist and in one case a cardiologist missed the second-degree atrioventricular block that was noted by the NP. The other variances between the raters were mostly due to differences in diagnosing PACs, PNCs, and PVCs. Agreement on baseline ST-segment levels was also high among raters. Agreement was enhanced because all raters were analyzing the same lead, V5, as per protocol. The minimal variance in reading ST-segment levels was due to difficulty in quantifying sub .2-mm differences in ST levels on photocopies of ECGs, which is a limitation of any noncomputerized assessment of ST segments. Agreement for peak exercise ST-segment levels was lower among the raters than the baseline ST-segment levels. Unlike the baseline ST-segment level measurements, the peak ST-segment level measurements were not read using the same ECG lead by all raters. Each rater chose the ECG lead closest to the peak performance reading that he or she thought could be read the most clearly. Each rater also chose the ECG lead that was thought to have the most ST depression or elevation. Variability in the ECG leads chosen accounted for the low agreement for peak ST-segment levels. Concordance between baseline ST-segment levels and computer-generated averaged measurements for the baseline tracing of V5 lead showed high agreement among the raters. Likewise, concordance on computer-generated averaged peak STsegment levels was moderate for the NP and high for C1. The computer-generated averaged measurements are not considered the “gold” standard but are a useful unbiased comparison. Computer-generated averaged measurements are subject to error, related to artifact or improper tracking of ECG segments or the J point. The computer-generated measurements are an average of the values selected at a precise point in time and frequently recorded approximately 10 seconds later than the ECG recordings. Changes in ST-segment levels can occur during this time period, accounting for some of the variance noted among raters. However, because we confirmed the accuracy of the tracking by examination of the location of the markers, and because all EST tracings were of acceptable quality, the computergenerated averaged analysis was useful. Unlike manual interpretation, this technique allows precise subdivision of the net ST depression in increments of
HEART & LUNG
VOL. 37, NO. 2
Maier et al
.1 mm, and it is encouraging how closely both NP and cardiologist assessment of rest and peak exercise ST segments correlated with the computer algorithm.
CONCLUSION An experienced NP and cardiologists had moderate to high agreement on all aspects important to the interpretation of ESTs, and there were no cases in which there were important differences in diagnosis. This lends support to NPs interpreting ESTs.
REFERENCES 1. Lem V, Kripokapich J, Child JS. A nurse supervised exercise stress testing laboratory. Heart Lung 1985;14:280-4. 2. Zecchin RP, Chai YY, Roach KA, et al. Is nurse-supervised exercise stress testing safe practice? Heart Lung 1999;28: 175-85. 3. Avorn J, Everitt DE, Baker MW. The neglected medical history and therapeutic choices for abdominal pain. Arch Intern Med 1991;151:694-8. 4. Everitt DE, Avorn J, Baker MW. Clinical decision-making in the evaluation and treatment of insomnia. Am J Med 1990; 89:357-62. 5. Mundinger MO, Kane RL, Lenz ER, et al. Primary care outcomes in patients treated by nurse practitioners or physicians: a randomized trial. JAMA 2000;283:59-68. 6. Moody NB, Smith PL, Glenn LL. Client characteristics and practice patterns of nurse practitioners and physicians. Nurse Pract 1999;24:94-103. 7. Hall JA, Palmer RH, Orav EJ, Hargraves JL, Wright EA, Louis TA. Performance quality, gender, and professional role: a study of physicians and nonphysician in 16 ambulatory care practices. Med Care 1990;28:489-501. 8. Hill J, Bird HA, Harmer R, Wright V, Lawton C. An evaluation of the effectiveness safety and acceptability of a nurse practitioner in a rheumatology outpatient clinic. Br J Rheumatol 1994;33:283-8. 9. Chang E, Daly J, Hawkins A, et al. An evaluation of the nurse practitioner role in a major rural emergency department. J Adv Nurs 1999;30:260-8. 10. Thompson RS, Basden P, Howell LJ. Evaluation of initial implementation of an organized adult health program employing family nurse practitioners. Med Care 1982;20: 1109-27. 11. Hanneman SKG, Bines AS, Sajtar WS. The indirect patient care effect of a unit based clinical nurse specialist on preventable pulmonary complications. Am J Crit Care 1993;2: 331-8. 12. Russell D, VorderBruegge M, Burns S. Effect of an outcomesmanaged approach to care of neuroscience patients by acute care nurse practitioners. Am J Crit Care 2002;11:353-64. 13. Overton-Brown P, Anthony D. Towards a partnership in care: nurses’ and doctors’ interpretation of extremity trauma radiology. J Adv Nurs 1998;27:890-6. 14. Hylka SC, Beschle JC. Nurse practitioners, cost savings, and improved patient care in the department of surgery. Nurs Econ 1995;13:349-54. 15. Bissinger RL, Allred CA, Arford PH, Bellig LL. A cost-effectiveness analysis of neonatal nurse practitioners. Nurs Econ 1997;15:92-9. 16. Spitzer WO, Sackett DL, Sibley JC, et al. The Burlington randomized trial of the nurse practitioner. N Engl J Med 1974;290:251-6. 17. Buppert C. Justifying nurse practitioner existence: hard facts to hard figures. Nurse Pract 1995;20:43-8.
www.heartandlung.org
151
Maier et al 18. Newbold D. An evaluation of the role of the nurse practitioner. Nurs Times 1996;92:45-6. 19. Daffurn K. Roles of acute care nurse practitioners, physician assistants, and resident physicians in acute care settings. Am J Crit Care 1998;7:253-4. 20. Froelicher VF. Manual of exercise testing. 2nd ed. St. Louis: Mosby-Year Book, Inc; 1994.
152
www.heartandlung.org
NP and cardiologist concordance on EST interpretation 21. Ellestad MH, Selvester RHS, Mishkin FS, James FW, Mazumi K. Stress testing: principles and practice. 4th ed. New York: Oxford University Press; 1996. 22. Gibbons RJ, Balady GJ, Beasley JW, et al. ACC/AHA Guidelines for Exercise Testing: A report of the American College of Cardiology/American Heart Association task force on practice guidelines. J Am Coll Cardiol 1997;30:260-315.
MARCH/APRIL 2008
HEART & LUNG
AIM: Cardiology nurse practitioners (NPs) conduct exercise stress tests (ESTs) for diagnosis of cardiac disease. The diagnostic concordance of NPs to cardiologists has not been assessed. The hypothesis was that an NP is as reliable as a cardiologist in determining ST-segment depression, detecting arrhythmias, and making a diagnostic assessment. METHODS: An NP and two cardiologists (C1 and C2) were provided with 100 consecutive, anonymized ESTs, consisting of three 10-second, 12-lead tracings obtained at baseline, peak-exercise, and recovery. Interpretation was based on baseline rhythm, baseline and maximal exercise ST levels, arrhythmias, and global diagnosis (positive, negative, or inconclusive for ischemia). Raters used uniform criteria to interpret ESTs and were blinded to prior EST interpretation and computerized ST-segment analysis. RESULTS: There was similar concordance between the NP and cardiologists as between the cardiologists, measured by Kappa coefficients (rhythm: NP vs. C1 ⫽ .92, NP vs. C2 ⫽ .84, C1 vs. C2 ⫽ .84; arrhythmias: NP vs. C1 ⫽ .77, NP vs. C2 ⫽ .73, C1 vs. C2 ⫽ .75; EST diagnosis: NP vs. C1 ⫽ .75, NP vs. C2 ⫽ .73, C1 vs. C2 ⫽ .75). Pearson correlations demonstrated concordance for baseline ST levels (NP vs. C1 ⫽ .86, NP vs. C2 ⫽ .86, C1 vs. C2 ⫽ .90) and peak exercise ST levels (NP vs. C1 ⫽ .58, NP vs. C2 ⫽ .48, C1 vs. C2 ⫽ .67). CONCLUSIONS: Concordance among raters, and with the computer-generated algorithm, was moderate to high for all parameters of EST interpretation. This study lends support to NPs interpreting ESTs. (Heart Lung® 2008;37:144 –152.)
T
he need to provide quality health care and improve access has led to examining roles for nurse practitioners (NPs). The Division of Cardiology at the University of Alberta Hospital implemented a role for NPs in supervising and interpreting exercise stress tests (ESTs). There is a paucity of research examining the efficacy of NPs in cardiology and none assessing the diagnostic accuracy of EST interpretations. Lem et al1 first examined nurse supervised ESTs and concluded they could safely conduct ESTs, as judged by low morbidity and mortality rates associated with testing. This conclusion was
From the aDivision of Cardiology, University of Alberta Hospital; and bFaculty of Nursing, University of Alberta, Edmonton, Canada. Reprint requests: Louise Jensen, RN, PhD, University of Alberta, Faculty of Nursing, 4-112 D Clinical Sciences Building, Edmonton, AB, Canada T6G 2G3. 0147-9563/$ – see front matter Copyright © 2008 by Mosby, Inc. doi:10.1016/j.hrtlng.2007.05.009
144
www.heartandlung.org
supported in a larger retrospective audit2 of 17,467 ESTs performed on 5054 patients over 12 years. Nurse supervised ESTs were considered safe, based on a major complication rate of .06%, defined as patients requiring urgent intervention (sustained VT, defibrillation). Although the safety of nurse supervised ESTs has support, a complete EST requires not only safe conduct of the procedure but also the correct diagnosis of changes in rhythm, ST-segment deviation, and the synthesis of findings into the likelihood of ischemia. Thus, the expanded role for NPs includes supervising as well as interpreting ESTs. In studies examining provision of certain aspects of care for specific patient populations, NPs and physicians have been shown to be comparable.3-9 NPs have also been shown to be thorough and comprehensive in evaluating and treating patients in a variety of clinical settings, including outpatient departments and physician’s offices,5,7,10 long
MARCH/APRIL 2008
HEART & LUNG
NP and cardiologist concordance on EST interpretation
term and rehabilitative care,8 emergency,9 critical care,11,12 orthopedics,13 surgical care,14 and pediatric and neonatal care.15 A large randomized controlled trial16 found no differences in diagnosis of functional capacity, and social or emotional function between groups of patients cared for by NPs and physicians. As the NP role evolves, the question of NP effectiveness in other areas of patient care arises.17-19
OBJECTIVE The objective of this study was to examine the reliability of an NP in evaluating ischemia on the basis of electrocardiogram (ECG) analysis during an EST. This evaluation was conducted by establishing the concordance among an NP and two cardiologists (C1 and C2), and to the computer-generated averaged measurements in a blinded study that provided each rater with identical ECG recordings and “real-world” variables available when conducting an EST. The hypothesis was that an NP would obtain results concordant with those of the cardiologists. Comparisons were made on (1) baseline rhythm and detection of arrhythmias and conduction abnormalities during exercise; (2) quantification of ST-segment levels at baseline and peak exercise; (3) diagnosis of the EST as positive, negative, or inconclusive for ischemia; and (4) ST levels determined by practitioners versus computer-generated at baseline and peak exercise.
METHODS The NP received training in EST interpretation from a Masters Degree in Nursing program, had been employed in the role for 2 years, and had 15 years experience in cardiology; C1 had 14 years experience, and C2 had 5 years experience in cardiology. A pilot study was done with five EST recordings to rehearse and standardize the criteria for interpretation of EST recordings. Several issues arose in the variability of interpretations of the EST recordings, such as which ECG lead to use for ST segment quantification, method for determining the PR segment, what to do in case of severe artifact at peak exercise, definition of left ventricular hypertrophy, and whether or not to determine ST levels for left bundle branch block (LBBB). Also, several variables were identified as being necessary to allow one to render a global interpretation (medications, blood pressure). For example, if the heart rate-systolic blood pressure product was less than 15,000 a “negative” EST would be interpreted as inconclusive, as would a “positive” EST in a patient taking
HEART & LUNG
VOL. 37, NO. 2
Maier et al
digoxin. Standard guidelines for interpretation of EST recordings were subsequently developed and used in the study. After ethical approval, consecutive ESTs were selected from those completed over 6 months at the University of Alberta Hospital. EST recordings were then obtained from Health Records, and the first 100 complete ESTs with interpretable ECGs (judged by the clinician who provided the final clinical diagnosis) that did not have an exclusion criterion constituted the sample. Exclusion criteria included (1) lack of at least one stable 12-lead tracing at baseline, at peak exercise, and during the first 5 minutes of recovery; (2) inability to obtain a clear, high-resolution photocopy of all ECG tracings; and (3) termination of the EST because of technical errors. The assistant photocopied the resting, peak exercise, and first two ECGs recorded in immediate recovery. Patient identification data were masked, and a numeric code was assigned. Raters were provided with the following information: patient’s age, medications, cardiovascular risk factors, and body mass index, as identified on the original EST requisition, as well as the reason for the EST request, protocol used during testing, test duration, reason for test termination, and presence of angina during the test. The EST recordings were then given to the raters who independently analyzed (1) baseline rhythm; (2) presence and type of arrhythmia or conduction abnormalities at baseline and during exercise; (3) baseline ST-segment level; and (4) STsegment level during peak exercise in the maximally abnormal lead. On the basis of this information and the clinical data provided, each provided an EST diagnosis of positive, negative, or inconclusive. The NP’s and cardiologists’ measurements of baseline and peak exercise ST-segment levels were later compared with computer-generated averaged measurements.
DEFINITION OF TERMS Measurement of ST segments Baseline and peak exercise ST-segments were measured in leads II, aVF, or V5-V6, using calipers or a transparent ruler with finely graded increments. Baseline ST-segment level was measured 80 ms after the J-point using the PR junction as a baseline.
EST diagnosis Positive EST. ESTs were positive for ischemia if during the exercise or in the immediate recovery period the subject developed 1 mm or more of flat or downsloping ST depression20 or 1.5 mm or more upsloping depression.21 ST levels were measured 80
www.heartandlung.org
145
Maier et al
NP and cardiologist concordance on EST interpretation
Table I Exercise stress test patient characteristics N
Cardiovascular risk factors None Hyperlipidemia High blood pressure Diabetes Family history Smoking, hyperlipidemia Hyperlipidemia, diabetes Hypertension, hyperlipidemia Smoking Hypertension, hyperlipidemia, diabetes Hypertension, hyperlipidemia, family history Hyperlipidemia, family history Medications None Beta-blocker ACEI Beta-blocker, ACEI Calcium channel blocker Nitrates Beta-blocker, nitrates Beta-blocker, calcium channel blocker, ACEI Beta-blocker, ACEI, digoxin Calcium channel blockers, nitrates Beta-blocker, calcium channel blocker ACEI, nitrates Beta-blockers, ACEI, nitrates
Percent
Table II Indications for exercise stress test and exercise stress test termination N*
Percent
46 45 19
46.2 45.5 19.2
7 5 2
7.1 5.1 2.0
1 1
1.0 1.0
1
1.0
1
1.0
1
1.0
47
47.5
25 8 4 4 3 3
25.3 8.1 4.0 4.0 3.0 3.0
2 1 1 1
2.0 1.0 1.0 1.0
1
1.0
43 22 9 9 4 3 3 2
43.0 22.0 9.0 9.0 4.0 3.0 3.0 2.0
1
1.0
1
1.0
1
1.0
Reason for EST Previous CAD Ischemia Ischemia, functional capacity Risk stratification Functional capacity Ischemia, arrhythmia Research Ischemic threshold, functional capacity Arrhythmias, syncope Medication response Motor vehicle license Reason for EST termination Fatigue and dyspnea Fatigue ST depression, pain Target heart rate Hypertension Angina Unable to keep up to treadmill Other pain Dyspnea Arrhythmia Paced rhythm
1 1
1.0 1.0
EST, Exercise stress test; CAD, coronary artery disease. *Some referrals had more than one indication.
54 18 6 6 5 3 2 2
54.0 18.0 6.0 6.0 5.0 3.0 2.0 2.0
1 1
1.0 1.0
1
1.0
ACEI, Angiotensin-converting enzyme inhibitor.
ms after the J point in a minimum of three consecutive heart-tracing beats. An EST was also considered positive for ischemia if ST-segment elevation 1 mm or more occurred in leads that did not have pathologic Q-waves, other than in leads aVR or V1.22 Negative EST. ESTs were negative when no markers of ischemia were noted and when the patient exercised more than 5 minutes and achieved 85% of maximal heart rate (calculated as 220 minus
146
www.heartandlung.org
age) or a heart rate-systolic blood pressure product greater than 15,000, provided no other confounding variables were noted that would make the test inconclusive (eg, the presence of an LBBB, the use of digoxin). Inconclusive EST. ESTs were deemed inconclusive when ST depression did not reach diagnostic criteria for ischemia, when the patient failed to achieve target heart rate-systolic blood pressure product greater than 15,000, or when the ECG was uninterpretable because of artifact. Otherwise pos-
MARCH/APRIL 2008
HEART & LUNG
NP and cardiologist concordance on EST interpretation
Maier et al
Table III Exercise stress test rhythm reported by raters Baseline EST rhythm Rhythm
NP (N)
C1 (N)
C2 (N)
Normal sinus rhythm Atrial fibrillation Right bundle branch block Left bundle branch block Paced rhythm Left ventricular hypertrophy Atrial fibrillation and left ventricular hypertrophy
87 1 4 2 1 4 1
87 1 4 3 1 3 1
85 1 4 3 1 5 1
Arrhythmias during EST Arrhythmias
NP (N)
C1 (N)
C2 (N)
PAC or PNC PVC Second-degree AV block PAC or PNC and PVC None
10 12 1 2 75
11 11 0 2 76
12 11 1 3 73
Diagnosis
Positive (N)
Negative (N)
Inconclusive (N)
NP C1 C2
8 20 12
53 54 56
29 26 32
EST diagnosis reported by raters Diagnosis
NP (N)
C1 (N)
C2 (N)
Positive Negative Inconclusive
18 53 29
20 54 26
12 56 32
EST, Exercise stress test; NP, nurse practitioner; C, cardiologist; PAC, premature atrial contractions; PNC, premature nodal contractions; PVC, premature ventricular contractions; AV, atrioventricular.
itive ESTs were considered inconclusive if the patient had taken digoxin within 1 week of testing, had left ventricular hypertrophy based on Estes criteria, had LBBB, or had 1 mm or more depression or elevation in lead V5 on the baseline ECG.
rhythm, arrhythmias during EST, and EST diagnosis. Pearson correlation coefficients with regression plots were conducted to assess agreement among raters on baseline ST and peak exercise ST-segment levels. High agreement was considered to be .80 or more, moderate .60 to .79, and low less than .6.
Data analysis Descriptive statistics (SPSS Version 12.0, Chicago, IL) were calculated for age, gender, cardiovascular risk factors, medications, EST protocol used, baseline rhythm, arrhythmias during EST, and ST segment levels. Kappa coefficients were calculated to assess agreement among raters for EST baseline
HEART & LUNG
VOL. 37, NO. 2
RESULTS Of the 100 ESTs, 99 used the Bruce Protocol; one used a Balke protocol. The mean age of patients was 58.9 ⫾ 2.05 years, and 68% were male. Cardiovascular risk factor profiles and current medications are listed in Table I. Mean EST length was 8.32 ⫾ .21
www.heartandlung.org
147
Maier et al
B
1.75 1.5 1.25 1 .75 .5 .25 0 -.25 -.5 -.75 -1.5
-1
-.5
.5
0
1.5
1
Y=.133+.722*X; R²=.732
Baseline ST Levels for NP (N=98)
Baseline ST Levels for C1 (N=97)
C
2
Baseline ST Levels for C2 (N=97)
Baseline ST Levels for C1 (N=97)
A
NP and cardiologist concordance on EST interpretation
1.5 1.25 1 .75 .5 .25 0 -.25 -.5 -.75 -1 -1.5
-1
-.5
0
.5
1
1.5
2
Y=.126+.767*X; R²=.744
Baseline ST Levels for NP (N=98)
1.75 1.5 1.25 1 .75 .5 .25 0 -.25 -.5 -.75
-1
-.75
-.5
-.25
0
.25
.5
.75
1
1.25 1.5
Y=.042+.855*X; R²=.812
Baseline ST Levels for C2 (N=97)
Fig 1 Concordance on Baseline ST Levels Between Practitioners.
minutes, with a range from 1 to 13.57 minutes. The most frequent indication for the EST was evaluation of ischemia for the patient with a known history of coronary artery disease (46.2%); the most frequent reason for EST termination was the combination of fatigue and dyspnea (47.5%) (Table II).
Concordance on baseline EST rhythm Baseline EST rhythms interpreted by raters were classified into eight categories (Table III). Of the 100 EST recordings, 87 had normal sinus rhythm, 4 had right bundle branch block, with a 100% concordance by raters. C1 and C2 noted LBBB for three EST recordings, whereas the NP noted LBBB for only two EST recordings. A discrepancy was also noted for left ventricular hypertrophy, with C2 choosing that category for five recordings, NP for four recordings, and C1 for 3 recordings. The Kappa coefficient was .92 (P ⬍ .001) between the NP and C1 for baseline rhythm; .84 (P ⬍ .001) between the NP and C2; and .84 (P ⬍ .001) between the C1 and C2.
148
www.heartandlung.org
Concordance on arrhythmias during the EST Five categories of arrhythmias during the EST were noted by raters (Table III). The majority of EST recordings had no arrhythmias; 75% noted by NP, 76% by C1, and 73% by C2. Second-degree heart block was identified only by the NP and C1, and not by C2. Kappa coefficient was .77 (P ⬍ .001) between NP and C1 on arrhythmias during the EST; .73 (P ⬍ .001) between NP and C2; and .75 (P ⬍ .001) between C1 and C2.
Concordance on ST-segment levels Baseline ST segment levels were not obtained if there was LBBB. The range of baseline ST-segment levels was from ⫺1.0 to 1.6 for the NP; from ⫺.70 to 1.6 for C1; and from ⫺1.7 to 2 for C2. The highest agreement was between C1 and C2 (r ⫽ .90, P ⬍ .001), followed by similar agreement between the NP and C1 and C2 (r ⫽ .86, P ⬍ .001, respectively) (Fig 1). The range of peak exercise ST-segment levels for the NP was ⫺2.5 to 2.5 mm. C1 had 3 missing values
MARCH/APRIL 2008
HEART & LUNG
NP and cardiologist concordance on EST interpretation
B
3
Peak ST Levels for NP (N=98)
Peak ST Levels for NP (N=98)
A
2 1 0 -1 -2 -3 -4
-5
-4
-3
-2
-1
0
1
2
3
Peak ST Levels for C1 (N=97)
Peak ST Levels for C1 (N=97)
C
Maier et al
3 2 1 0 -1 -2 -3 -4 -3
-2
-1
0
1
2
3
4
Y=-.061+.569*X; R²=.23
Peak ST Levels for C2 (N=96)
3 2 1 0 -1 -2 -3 -4 -5 -3
-2
-1
0
1
2
3
4
Y= -.027+.837*X; R²=.444
Peak ST Levels for C2 (N=96)
Fig 2 Concordance on Peak ST Levels Between Practitioners.
due to LBBB, with a range from ⫺4.0 to 2.5 mm; C2 had 4 missing values due to LBBB, with a range from ⫺2.50 to 3.40 mm. Highest agreement was between C1 and C2 (r ⫽ .67, P ⬍ .001), followed by NP and C1 (r ⫽ .58, P ⬍ .01), and NP and C2 (r ⫽ .48, P ⬍ .001) (Fig 2).
Concordance on EST diagnosis Table III shows the EST diagnosis chosen by each rater. Agreement was a Kappa coefficient of .75 (P ⫽ .000) between the NP and C1 on the EST diagnosis; .73 (P ⫽ .000) between NP and C2; and .75 (P ⫽ .000) between C1 and C2. In no case were ESTs read as positive by the cardiologists that were read as negative by the NP, and no ESTs were read as negative by the cardiologist that were read positive by the NP.
Concordance on computer generated-averaged ST segment data The computer-generated averaged measurements were analyzed for baseline ST-segment levels for all raters and for peak ST segment levels for the NP and
HEART & LUNG
VOL. 37, NO. 2
C1. Peak ST-segment levels were not available for a comparison with C2. Baseline ST-segment levels were unavailable or uninterpretable for 12 EST recordings. All recordings for baseline ST analysis were taken from the resting data for lead V5. A simple regression analysis was done for the following: computer-generated averaged baseline ST-segment levels versus levels recorded by the NP (r ⫽ .85, P ⬍ .001), versus levels recorded by C1 (r ⫽ .88, P ⬍ .001), versus levels recorded by C2 (r ⫽ .93, P ⬍ .001) (Fig 3). Peak exercise ST segment levels were obtained by each rater according to the lead he/she considered most significant in the first readable recording in recovery. Two raters indicated which lead was used for analysis; therefore a simple regression analysis was obtained for the NP and C1 and the computergenerated averaged levels. Not all computer-generated averaged levels were available to be analyzed (n ⫽ 84 for the NP and n ⫽ 76 for C1) because data were not recorded for that particular EST, records were not available on recall, or ST levels were uninterpretable, as in the case of an LBBB. Agreement was r ⫽ .80 (P ⬍ .001) for the peak ST-segment for the NP versus the
www.heartandlung.org
149
Maier et al
NP and cardiologist concordance on EST interpretation
A
B
2
2 1.5
Computer Baseline ST Levels (N=89)
Computer Baseline ST Levels (N=89)
1.5 1 .5 0 -.5
1 .5 0 -.5 -1
-1
-1.5
-1.5 -1
-1.5
0
-.5
1
.5
1.5
-.75
2
Y=.189+.72*X; R²=.719
-.25
0
.25
.5
.75
1.25
1
1.5
1.75
Y=.094+.903*X; R²=.775
Baseline ST Levels for NP (N=98)
C
-.5
Baseline ST Levels for C1 (N=97)
2
Computer Baseline ST Levels (N=89)
1.5 1 .5 0 -.5 -1 -1.5 -1
-.75
-.5
-.25
0
.25
.5
.75
1
1.25
1.5
Y=.095+.896*X; R²=.865
Baseline ST Levels for C2 (N=97)
Fig 3 Concordance on Baseline ST Levels Between Computer and Practitioners.
B
4
Computer Peak ST Levels
Computer Peak ST Levels (N=84)
A
3 2 1 0 -1 -2 -3
4 3 2 1 0 -1 -2 -3 -4
-4 -4
-3
-2
-1
0
1
2
Y=.29+.965*X; R²=.637
Peak ST Levels NP (N=98)
3
-5
-4
-3
-2
-1
0
1
2
3
Y=.357+1.013*X; R²=.823
Peak ST Levels for C1 (N=97)
Fig 4 Concordance on Peak ST Levels Between Computer and Practitioners.
computer algorithm and r ⫽ .91 (P ⬍ .001) for C1 versus the computer algorithm (Fig 4).
DISCUSSION An important finding is that there was high concordance between the NP and the cardiologists, both for global EST diagnosis and for quantitative rest and exercise ECG analysis. Findings on EST diagnosis showed moderate agreement among all raters. Agreement was high on ESTs that were clearly positive or clearly negative. The NP did not
150
www.heartandlung.org
misdiagnose a strongly positive or strongly negative EST recording, which could compromise the patient’s outcome. The variability that existed was a result of differences in the lead selected as demonstrating maximal ST depression and differences in interpretation of inconclusive ESTs. Concordance on baseline rhythm showed high agreement among raters. All raters were concordant on paced rhythm and atrial fibrillation as a baseline rhythm. Concordance was also noted on right bundle branch block rhythm. Agreement for arrhythmias
MARCH/APRIL 2008
HEART & LUNG
NP and cardiologist concordance on EST interpretation
during the EST was moderate for raters. The only variances were two failures by the NP to identify an LBBB; however, in one case this was missed by one cardiologist and in one case a cardiologist missed the second-degree atrioventricular block that was noted by the NP. The other variances between the raters were mostly due to differences in diagnosing PACs, PNCs, and PVCs. Agreement on baseline ST-segment levels was also high among raters. Agreement was enhanced because all raters were analyzing the same lead, V5, as per protocol. The minimal variance in reading ST-segment levels was due to difficulty in quantifying sub .2-mm differences in ST levels on photocopies of ECGs, which is a limitation of any noncomputerized assessment of ST segments. Agreement for peak exercise ST-segment levels was lower among the raters than the baseline ST-segment levels. Unlike the baseline ST-segment level measurements, the peak ST-segment level measurements were not read using the same ECG lead by all raters. Each rater chose the ECG lead closest to the peak performance reading that he or she thought could be read the most clearly. Each rater also chose the ECG lead that was thought to have the most ST depression or elevation. Variability in the ECG leads chosen accounted for the low agreement for peak ST-segment levels. Concordance between baseline ST-segment levels and computer-generated averaged measurements for the baseline tracing of V5 lead showed high agreement among the raters. Likewise, concordance on computer-generated averaged peak STsegment levels was moderate for the NP and high for C1. The computer-generated averaged measurements are not considered the “gold” standard but are a useful unbiased comparison. Computer-generated averaged measurements are subject to error, related to artifact or improper tracking of ECG segments or the J point. The computer-generated measurements are an average of the values selected at a precise point in time and frequently recorded approximately 10 seconds later than the ECG recordings. Changes in ST-segment levels can occur during this time period, accounting for some of the variance noted among raters. However, because we confirmed the accuracy of the tracking by examination of the location of the markers, and because all EST tracings were of acceptable quality, the computergenerated averaged analysis was useful. Unlike manual interpretation, this technique allows precise subdivision of the net ST depression in increments of
HEART & LUNG
VOL. 37, NO. 2
Maier et al
.1 mm, and it is encouraging how closely both NP and cardiologist assessment of rest and peak exercise ST segments correlated with the computer algorithm.
CONCLUSION An experienced NP and cardiologists had moderate to high agreement on all aspects important to the interpretation of ESTs, and there were no cases in which there were important differences in diagnosis. This lends support to NPs interpreting ESTs.
REFERENCES 1. Lem V, Kripokapich J, Child JS. A nurse supervised exercise stress testing laboratory. Heart Lung 1985;14:280-4. 2. Zecchin RP, Chai YY, Roach KA, et al. Is nurse-supervised exercise stress testing safe practice? Heart Lung 1999;28: 175-85. 3. Avorn J, Everitt DE, Baker MW. The neglected medical history and therapeutic choices for abdominal pain. Arch Intern Med 1991;151:694-8. 4. Everitt DE, Avorn J, Baker MW. Clinical decision-making in the evaluation and treatment of insomnia. Am J Med 1990; 89:357-62. 5. Mundinger MO, Kane RL, Lenz ER, et al. Primary care outcomes in patients treated by nurse practitioners or physicians: a randomized trial. JAMA 2000;283:59-68. 6. Moody NB, Smith PL, Glenn LL. Client characteristics and practice patterns of nurse practitioners and physicians. Nurse Pract 1999;24:94-103. 7. Hall JA, Palmer RH, Orav EJ, Hargraves JL, Wright EA, Louis TA. Performance quality, gender, and professional role: a study of physicians and nonphysician in 16 ambulatory care practices. Med Care 1990;28:489-501. 8. Hill J, Bird HA, Harmer R, Wright V, Lawton C. An evaluation of the effectiveness safety and acceptability of a nurse practitioner in a rheumatology outpatient clinic. Br J Rheumatol 1994;33:283-8. 9. Chang E, Daly J, Hawkins A, et al. An evaluation of the nurse practitioner role in a major rural emergency department. J Adv Nurs 1999;30:260-8. 10. Thompson RS, Basden P, Howell LJ. Evaluation of initial implementation of an organized adult health program employing family nurse practitioners. Med Care 1982;20: 1109-27. 11. Hanneman SKG, Bines AS, Sajtar WS. The indirect patient care effect of a unit based clinical nurse specialist on preventable pulmonary complications. Am J Crit Care 1993;2: 331-8. 12. Russell D, VorderBruegge M, Burns S. Effect of an outcomesmanaged approach to care of neuroscience patients by acute care nurse practitioners. Am J Crit Care 2002;11:353-64. 13. Overton-Brown P, Anthony D. Towards a partnership in care: nurses’ and doctors’ interpretation of extremity trauma radiology. J Adv Nurs 1998;27:890-6. 14. Hylka SC, Beschle JC. Nurse practitioners, cost savings, and improved patient care in the department of surgery. Nurs Econ 1995;13:349-54. 15. Bissinger RL, Allred CA, Arford PH, Bellig LL. A cost-effectiveness analysis of neonatal nurse practitioners. Nurs Econ 1997;15:92-9. 16. Spitzer WO, Sackett DL, Sibley JC, et al. The Burlington randomized trial of the nurse practitioner. N Engl J Med 1974;290:251-6. 17. Buppert C. Justifying nurse practitioner existence: hard facts to hard figures. Nurse Pract 1995;20:43-8.
www.heartandlung.org
151
Maier et al 18. Newbold D. An evaluation of the role of the nurse practitioner. Nurs Times 1996;92:45-6. 19. Daffurn K. Roles of acute care nurse practitioners, physician assistants, and resident physicians in acute care settings. Am J Crit Care 1998;7:253-4. 20. Froelicher VF. Manual of exercise testing. 2nd ed. St. Louis: Mosby-Year Book, Inc; 1994.
152
www.heartandlung.org
NP and cardiologist concordance on EST interpretation 21. Ellestad MH, Selvester RHS, Mishkin FS, James FW, Mazumi K. Stress testing: principles and practice. 4th ed. New York: Oxford University Press; 1996. 22. Gibbons RJ, Balady GJ, Beasley JW, et al. ACC/AHA Guidelines for Exercise Testing: A report of the American College of Cardiology/American Heart Association task force on practice guidelines. J Am Coll Cardiol 1997;30:260-315.
MARCH/APRIL 2008
HEART & LUNG