Prognostic value of biochemical markers, 12-lead ECG and patient characteristics amongst patients calling for an ambulance due to a suspected acute coronary syndrome

Authors


Dr Leif Svensson, Division of Cardiology, South Hospital, SE-118 83 Stockholm, Sweden (fax: +46-8-6163040; e-mail: leif.svensson@sos.sll.se).

Abstract.

Objectives.  To evaluate whether a 12-lead ECG, together with a multi-marker strategy that used point-of-care measurements of myoglobin, creatine kinase (CK-MB) and troponin I, was able to predict patients at short- and long-term risk of death, when simultaneously considering age, gender, previous history, symptoms and clinical findings on arrival of the ambulance.

Design.  Prospective observational study.

Setting and subjects.  Consecutive patients (n=511) in ambulances in Stockholm and Göteborg in Sweden who called for an ambulance due to chest pain or other symptoms raising a suspicion of acute coronary syndrome.

Intervention.  In almost all patients, a diagnostic ECG, patient baseline characteristics and measurements of CK-MB, troponin I and myoglobin were recorded.

Results.  In univariate analysis, the highest 30-day mortality (17%) was found amongst patients with the combination of ECG signs of myocardial ischaemia and the elevation of any biochemical marker. The highest 1-year mortality (20%) was found amongst patients with ECG signs of myocardial ischaemia and the elevation of any biochemical marker. Increasing age (RR 1.07; 95 CI 1.02–1.13) lack of symptoms of chest pain and a previous history of hypertension (3.02; 1.08–8.79) were independent predictors of 30-day mortality. Myoglobin was the only biochemical marker independently associated with 30-day mortality (6.66; 1.83–22.3). Increasing age (1.11; 1.06–1.16), previous history of diabetes (3.42; 1.41–8.25) heart failure (2.64; 1.26–5.52) and other symptoms than chest pain and dyspnoea (5.23; 2.14–12.76) were independent predictors of 1-year mortality. In many of the variables the confidence limits were wide.

Conclusion.  Amongst patients with a clinical suspicion of acute coronary syndrome, those with the combination of ECG signs of myocardial ischaemia and the elevation of any biochemical marker on arrival of the ambulance form a group with a particularly high risk of death. However, age as well as aspects of clinical history and type of symptoms independently contribute to prognostic information.

Introduction

At the present time, there are several therapeutic options for acute coronary syndrome patients. The need for an early risk stratification in order to formulate a strategy and choose the appropriate treatment for each individual patient is therefore of great importance. The earlier high-risk patients can be defined, the faster we can start investigations and initiate the right kind of treatment. At the same time, it is important to delineate those at low risk in order to optimize the allocation of resources.

The ideal situation when dealing with patients raising a clinical suspicion of an acute coronary syndrome is to make a risk evaluation in the patients’ home or, at the very latest, during the ambulance journey.

Recent clinical trials indicate that an ECG and analysis of biochemical markers at the emergency department enable immediate risk stratification [1].

By reporting a simple protocol, using clinical data and physical examination from the ambulance, transmitting a 12-lead ECG and analysing biochemical markers, the doctor on call can be prepared for the patient's arrival and make an initial risk assessment.

The primary aim of this report was therefore to assess whether we could find early predictors of short- (30-day) and long-term (1-year) mortality in the ambulance amongst patients with a suspected acute coronary syndrome. Our hypothesis was that the combination of ECG signs of myocardial ischaemia and the elevation of biochemical markers would be associated with a particularly high risk of death.

A secondary aim was to attempt to define a group of patients with a particularly low risk of death.

Method

Study patients and ambulance organization

In all, the study population, 313 cases from Göteborg and 225 from Stockholm, contains a consecutive series of 538 patients who called for an ambulance due to chest discomfort or other symptoms raising a suspicion of an acute coronary syndrome.

In certain areas of Göteborg and Stockholm, the prehospital triage of patients with suspected acute coronary syndromes is performed by specially trained ambulance personnel, basically well-trained nurses. Three ambulances were used in this study. Doctors are very seldom on board the vehicle. When confronted by a chest pain patient, a prehospital ECG is generally transmitted from the ambulance or the patient's home to a doctor on call at the hospital coronary care unit.

Enrolment criteria

  • 1Chest pain or chest discomfort for more than 15 min within the last 6 h.
  • 2Dyspnoea without known lung disease such as asthma or bronchitis.
  • 3Any other condition causing the ambulance personnel to suspect acute coronary syndrome. There were no exclusion criteria.

Blood samples and analysis of biochemical markers

On arrival at the scene, the ambulance personnel made a rapid assessment of the patient by taking a short history and making a basic examination including measurements of blood pressure and heart rate. A blood sample was immediately obtained for a qualitative bedside test with a ‘3 in 1’ combination test (myoglobin, CKMB and troponin I) using the second generation of tests (Cardiac Status Spectral Diagnostics, Toronto, Canada).

The spectral qualitative test uses an immunochromatographic dry-strip method. If abnormal concentrations of any markers are present, immobilized antibodies bind the marker and a sharp band appears in the test window. A control band also appears, indicating that the test has functioned correctly. This type of test gives a yes (positive) or no (negative) answer. This cardiac status rapid bedside test is designed to give a cut-off for a positive test as follows: myoglobin >50 ng mL−1, CKMB >5 ng mL−1 and troponin I >0.1 ng mL−1. Before the start of the study, ambulance personnel participating in the trial were trained to follow a standard test procedure recommended by the manufacturer.

Venous blood was drawn in a lithium-prepared 1 mL syringe and 0.2 mL (approximately eight drops) was dripped into the test chamber. A special clock was used for a correct and exact 15-min reading of the test. During the 15-min blood test analysis, the ambulance personnel recorded and sent a prehospital ECG and established telephonic communication with the coronary care unit at the referring hospital.

System for sending and interpreting prehospital ECGs

All three ambulances were equipped with a portable ECG for transmission from the ambulance or from the patients’ home. All ECGs were transmitted either by the regular phone line system or by a mobile system using Global System Mobile, or a military system used by ambulances and fire departments (Mobitex).

Two different prehospital ECG systems were used, Mobimed-Ortivus (Stockholm) and Physiocontrol-Medtronic Life Pac 12 (Göteborg). There was always a doctor on call at the hospital to interpret the prehospital ECG and the doctor made the decision about treatment (thrombolysis–aspirin–heparin) and level of observation (directly to coronary care unit/cath laboratory or to the emergency department).

Definition of diagnosis at hospital discharge

Acute myocardial infarction was diagnosed if inclusion criteria plus troponin T >0.05 ng mL−1 when troponin was analysed or CK MB >10 ng mL−1 when troponin T was not analysed [2, 3]. (Troponin I was analysed prior to hospital admission but troponin T was usually analysed in hospital.) Diagnosis was based on ECG findings and analysis of biochemical markers whilst in hospital. CKMB was analysed 10 and 20 h after arrival in hospital.

Myocardial ischaemia was diagnosed when there were dynamic changes in ECG (ST-segment or T-wave changes) but no raised biochemical markers above myocardial infarction limit. Possible myocardial ischaemia was diagnosed when the patient had symptoms that strongly indicated myocardial ischaemia/ infarction, but the patient had no ECG-changes or raised biochemical markers.

A 12-lead ECG was recorded prior to hospital admission when arriving in hospital and then daily during the first 3 days in hospital.

Criteria for ECG changes

ST-elevation: ≥2 mm in leads V1–V4; ≥1 mm in leads V5–V6 or in any of the extremity leads.

ST-depression: ≥1 mm in any lead.

Further variables evaluated

Variables include previous history of myocardial infarction, angina pectoris, hypertension, diabetes, congestive heart failure and smoking. This information was retrospectively recorded and based on information from patients’ case records.

Symptoms of chest pain, dyspnoea or other symptoms were prospectively recorded by the ambulance personnel.

Clinical signs of paleness, cold sweat and nausea were prospectively recorded by the ambulance personnel.

Information on deaths was achieved through the National Registry of Deaths.

Statistical methods

Descriptive statistics

Distributions of variables are given as the median and the mean and percentages. Five per cent of the visits were repeat visits by the same patients. In this survey, only the first visit was included in the analyses.

Statistical analysis

For comparisons of dichotomous variables between groups, Fisher's exact test was used. This is a special form of Fisher's nonparametric permutation test [4].

Multivariate statistical analysis

A stepwise logistic regression was used to select independent predictors of dichotomous dependent variables. Only factors that showed some association with mortality (P < 0.2) in the univariate analysis were tested for inclusion in the model.

Results

Recruitment

A total of 511 patients took part in this survey on 538 occasions and all the patients were recruited between February and November 2000. This survey only dealt with the first visit of each patient (n = 511). In all, 31% had a final diagnosis of acute myocardial infarction and 60% had a diagnosis of either myocardial infarction or myocardial ischaemia (acute coronary syndrome). Information on death was missing in 10 patients (2%).

The median interval between onset of symptoms and analysis of biochemical markers was 90 min. ECG-recording and clinical examination took place a few minutes earlier.

Univariate analysis

Age, gender and previous history.  In overall terms, the mean age was 69 years. Elderly patients, above the median of 72 years, had a higher 30-day and 1-year mortality (Table 1). Women had a significantly higher 1-year mortality than men. A history of diabetes mellitus or heart failure was a predictor of a worse 1-year prognosis.

Table 1.  Mortality during 30 days and 1 year in relation to age, sex and previous history
 30 daysP*1 yearP*
n%n%
  1. *P-value denoted if <0.2; **Number of patients for whom information was available.

Age (years; median)
 ≤ 72 (252)**31.20.0006114.4<0.0001
 > 72 (247)197.7 5221.0 
Gender
 Men (293)134.4 289.60.022
 Women (207)94.4 3516.9 
Previous history of Myocardial infarction
  Yes (216)94.2 3013.9 
  No (282)134.6 3311.7 
 Angina pectoris
  Yes (276)93.3 3713.4 
  No (216)136.0 2511.6 
 Hypertension
  Yes (145)106.90.162114.5 
  No (343)123.5 3811.1 
 Diabetes mellitus
  Yes (81)67.4 1721.00.020
  No (408)163.9 4310.5 
 Heart failure
  Yes (111)65.4 2623.40.0003
  No (378)164.2 359.3 
 Smoking
  Yes (108)32.8 76.50.07
  No (352)174.8 4713.4 
Interval between onset of symptoms and blood sampling (median; min)
  ≤ 90 (252)124.8 3011.9 
  >90 (231)93.8 3113.1 

Symptoms, clinical findings, 12-lead ECG findings and biochemical markers prior to hospital admission.  No specific symptom or clinical finding found on admission by the ambulance crew was associated with prognosis (Tables 2 and 3).

Table 2.  Mortality during 30 days and 1 year in relation to symptoms, clinical findings, 12-lead ECG findings and biochemical markers prior to hospital admission
 30 daysP*1 yearP*
n%n%
  1. *P-value denoted if <0.2; **Number of patients for whom information was available.

Symptoms
 Chest pain
  Yes (473)**183.80.0225611.80.07
  No (26)415.4 726.9 
 Dyspnoea
  Yes (243)83.3 3112.8 
  No (236)125.1 2912.3 
 Other symptoms
  Yes (60)46.7 1220.00.18
  No (376)164.3 4712.5 
Clinical findings
 Pale
  Yes (281)113.9 4315.3 
  No (211)94.3 188.5 
 Cold sweat
  Yes (184)94.9 2111.4 
  No (299)124.0 4013.4 
 Nausea
  Yes (155)53.2 149.00.09
  No (299)144.7 4515.0 
ECG findings
 ST-elevation
  Yes (92)99.80.0221617.40.197
  No (402)133.2 4711.7 
 ST-depression
  Yes (152)74.4 1912.5 
  No (342)154.6 4412.9 
 T-inversion
  Yes (62)46.4 914.5 
  No (432)184.2 5412.5 
 Q-wave
  Yes (29)310.3 413.8 
  No (465)194.1 5912.7 
Biochemical markers
 Myoglobin
  Yes (32)618.8<0.0001825.00.07
  No (459)143.0 5311.6 
 CK MB
  Yes (42)511.90.020819.0 
  No (453)153.3 5311.7 
 Troponin T
  Yes (23)313.0 417.4 
  No (470)163.4 5611.9 
Table 3.  Mortality amongst patients with various signs of myocardial ischaemia in relation to whether there was any initial elevation of biochemical markers
 Mortality
30 days1 year
Any elev. markerNo elev. markerP*Any elev. markerNo elev. markerP*
  1. *P-value denoted if <0.2; **Number of deceased patients/number of patients evaluated; ***Percentage of deceased patients.

ST-elevation3/18**4/72 4/1810/72 
17***6 2214 
ST-depression3/203/131 3/2015/131 
152 1511 
Q-wave1/73/54 1/78/54 
146 1415 
T-inversion1/11/27 1/12/27 
1004 1007 

Ninety-two patients or 18% had ST-segment elevation on admission of the ambulance crew and they had a significantly higher 30-day mortality than those without. No difference was found when comparing patients with and without ST-depression.

Amongst biochemical markers, a positive myoglobin or CK MB was associated with a higher 30-day mortality. Troponin I was positive in 23 patients and did not predict 30-day or 1-year mortality.

As shown in Table 3, the combination of various ECG signs of myocardial ischaemia and the elevation of any biochemical marker tended to be associated with high early mortality. As shown in Fig. 1, patients with ECG signs of myocardial ischaemia and the elevation of any biochemical marker prior to hospital admission had an extremely high mortality within the first 30 days (17%). Due to the very low number of patients in this group (n = 29) it was not possible to adequately test this mortality rate versus the mortality in the remaining patients.

Figure 1.

Mortality during 30 days in relation to ECG-findings and elevation of any biochemical markers on admission of the ambulance crew.

However, patients with a nonpathological ECG and no elevation of any biochemical marker had a very low early mortality (2%).

Figure 2 shows that the 1-year mortality was also high in patients with ECG signs of myocardial ischaemia and the elevation of any biochemical marker (20%).

Figure 2.

Mortality during 1 year in relation to ECG-findings and elevation of any ECG findings and elevation of any biochemical markers on admission of the ambulance crew.

Multivariate analysis

Independent predictors of death during 30 days. Table 4 shows that increasing age, no symptoms of chest pain and a previous history of hypertension were independent predictors of 30-day mortality. Only elevation of myoglobin was an independent predictor of short-term outcome.

Table 4.  Independent predictors of death during 30 days
 Odds ratio95% confidence limitP
  1. *Age was analysed as a continuous variable, i.e. the odds ratio refers to the mean increase in risk per year of increase in age.

Age (years)*1.071.02–1.130.010
Previous history
 Hypertension3.021.08–8.790.036
Symptoms
 Chest pain0.180.05–0.750.010
Biochemical markers
 Myoglobin6.661.83–22.30.002

Independent predictors of death during 1 year.  As can be seen in Table 5, age, a previous history of diabetes and heart failure and symptoms other than chest pain and dyspnoea were the only predictors of 1-year mortality. Symptoms other than chest pain and dyspnoea included hypotension, cardiogenic shock, dysrhythmia and syncope.

Table 5.  Independent predictors of death during 1 year
 Odds ratio95% confidence limitP
  1. *Age was analysed as a continuous variable, i.e. the odds ratio refers to the mean increase in risk per year of increase in age.

Age (years)*1.111.06–1.16<0.0001
Previous history
 Diabetes3.421.41–8.250.006
 Heart failure2.641.26–5.520.010
Symptoms
 Other than chest pain and dyspnoea5.232.14–12.760.0003

Discussion

This study prospectively enrolled consecutive patients from the whole spectrum calling for an ambulance due to acute chest discomfort or other symptoms raising a suspicion of an acute coronary syndrome. This is a heterogeneous group. However, a high proportion still had a final diagnosis equivalent to an acute coronary syndrome (60%). ST-segment elevation was found in 18%, which is a higher figure than that reported in other surveys of ambulance-transported patients with acute chest pain [5, 6].

Furthermore, the fact that a patient calls for an ambulance in itself indicates worse symptoms [7] and a worse prognosis [8, 9].

This study is the first to attempt to define patients in a prehospital setting with an increased risk of an adverse short- and long-term prognosis. Previous smaller studies have evaluated biochemical markers in the prehospital setting in order to detect myocardial damage [10–12].

In this survey, we found that the elevation of myoglobin was an independent risk indicator of death during the first 30 days. However, confidence limits were wide. Furthermore, we found that the group of patients with a particularly high risk of death were those with any sign of ischaemia on ECG and the simultaneous elevation of any biochemical marker.

Initiating risk stratification in the prehospital phase by analysing ECG and biochemical markers and making the first tests in the ambulance and the next at the emergency department could be helpful. Evaluation in the emergency department might be of value amongst patients who were reached by the ambulance crew very early after onset of symptoms and therefore require one further examination. Further trials must be conducted to confirm this.

One might argue about the importance of evaluating the risk of death during the subsequent year prior to hospital admission. However, very early determination of the risk of death during the subsequent 30 days seems a reasonable priority.

A history of hypertension and the lack of symptoms of chest pain were independently associated with an adverse short-term outcome. For both the confidence limits were wide.

A history of hypertension was found to be associated with an adverse outcome amongst patients with acute myocardial infarction [13, 14]. The mechanisms behind this observation can only be speculated upon. Perhaps myocardial hypertrophy secondary to hypertension increases the risk of complications in the early phase.

The observation that a lack of chest pain amongst patients with any symptom raising a suspicion of an acute coronary syndrome is associated with an adverse prognosis is not new. Several previous studies have shown that symptoms other than chest pain in this patient population are an alarming sign [15–17]. For example, symptoms such as dyspnoea might indicate a deterioration in myocardial function.

Multivariate analysis showed that high age and a history of either diabetes and/or heart failure are predictors of increased 1-year mortality. Patients with these findings also form a well-known high-risk group whilst in hospital [18–21]. This paper confirms this risk profile and special attention must therefore be paid to this category of patients in the ambulance. These patients might be given priority when deciding who should be transported directly to the coronary care unit.

Neither ST-segment elevation nor ST-segment depression was an independent predictor of an adverse 1-year outcome.

ST-segment depression has been shown to predict an adverse prognosis [22, 23]. The reason why we were not able to confirm this might be multifactorial. The most important reason is probably the fact that the whole ambulance group of patients (including those with no ST-elevation) is a high-risk group with a relatively high long-term mortality [9]. Furthermore, the study population most likely represent a high-risk population amongst those being transported by ambulance as only patients reached by the mobile coronary care unit were included in the survey. This might also explain why, for example, patients with T-wave inversion, who have been shown in previous surveys to have a relatively good prognosis [24, 25], also had a relatively high mortality in this survey.

Biochemical markers are useful in the emergency department for both diagnostic and prognostic purposes [1]. This paper describes the prognostic value of the use of biochemical markers in the prehospital field. It is difficult to justify a single test for biochemical markers like the one performed in this survey. Biochemical markers must be studied over a period of time. Myoglobin suffers from a lack of cardiac specificity, but it is the marker that is released at the earliest stage in myocardial infarction and it appears to be an ideal marker to complement CK MB and troponins in the early prehospital phase. The risk of missing a myocardial infarction when only using a single troponin test upon arrival at the hospital emergency department is said to be about 10–15% [26]. When troponin is used, American and European guidelines recommend repeating tests up to 12 h after the onset of symptoms [2].

Myoglobin was the only biochemical marker in this study to predict early mortality. Although the odds ratio was high, the confidence limits were wide, indicating unsafe data. The marker results are somewhat confounding. However, in a recently published relatively large survey early elevation of myoglobin turned out to be a strong independent predictor of 6-month mortality amongst patients with non-ST-elevation acute coronary syndromes when CK MB and troponin I were simultaneously considered [27]. Hospital studies have shown that troponins are ideal prognostic markers. There are probably several reasons why myoglobin in this ambulance cohort emerged as the predictor marker for death rather than CK MB or troponin. The first is the time-dependent factor. Furthermore, the elevation of myoglobin might be associated with multiorganic failure, thereby delineating patients with the most extensive infarction in the early phase. Our data need to be confirmed in larger studies.

For risk stratification in the prehospital phase, both myoglobin and CK MB may play an important role as smoke detectors. When positive, they will place the patient at the right level of care. If negative within the first 4–8 h after symptom onset, they can be useful in ruling out acute myocardial infarction [28].

What is the ideal time window when using biochemical markers in prehospital and emergency department care? Hospital guidelines recommend repeated measurements 8–12 h after the onset of symptoms. However, recent studies have tested an interval of 6 h [1] and even 90 min [29] after presentation at the hospital to identify high-risk patients, the latter using a similar type of a point-of-care test to the one we used with the combination of myoglobin, CK MB and troponin I.

The potential clinical importance of a prehospital risk stratification in acute coronary syndrome is: (i) High risk patients could directly be allocated to coronary care unit or percutaneous coronary intervention centres. (ii) Early antithrombotic treatment (prior to hospital admission) might be most appropriate in high-risk patients.

Limitation

There were a few false positives with regard to biochemical markers analysed prior to hospital admission [30]. Furthermore, very early risk stratification might underestimate the risk of a serious coronary event and subsequent adverse outcome simply because the analysis were performed too early to detect any myocardial damage.

Conclusions

This ambulance study suggests that, amongst patients with a clinical suspicion of acute coronary syndrome, those with the combination of ECG signs of myocardial ischaemia and the elevation of any biochemical marker on admission by the ambulance crew form a group with a particularly high risk of death. However, age as well as aspects of clinical history and type of symptoms independently contribute to prognostic information.

Conflict of interest statement

No conflict of interest was declared.

Acknowledgements

This study was supported by grants from the Swedish Heart and Lung Foundation and Infinity Medical, Sweden/Spectral Diagnostics Toronto, Canada.

Ancillary