Chest pain is a common presenting complaint among patients presenting to the emergency department (ED). Many need assessment to diagnose or exclude an acute coronary syndrome. Data from the USA suggest that 15–25% of patients who present to the ED with undifferentiated chest pain are diagnosed with an acute coronary syndrome within 30 days.1 Isolated symptoms suggestive of acute coronary syndrome (eg, left arm discomfort, shortness of breath, nausea or vomiting, epigastric pain, dizziness) may require a similar investigation. These have been shown to capture more than 90% of acute coronary syndromes in patients in community-based studies.2

Guidelines have been developed to identify and risk-stratify patients with possible acute coronary syndrome. The cornerstone of these approaches is to stratify patients into high-, medium- and low-risk groups according to their past history, risk factors, nature of the pain and investigations (eg, ECG and serial troponin).3 4 Chest radiography is a commonly ordered investigation for this patient group with one study reporting a rate exceeding 70%,5 despite the fact that chest radiography is not included in guidelines for risk stratification.

The evaluation of undifferentiated chest pain requires consideration of diagnoses other than acute coronary syndrome. The differential diagnosis of other potentially life-threatening conditions includes pulmonary embolism, aortic dissection, oesophageal rupture, pericarditis, spontaneous pneumothorax, pneumonia and certain acute abdominal conditions. It is often unclear which patients with undifferentiated chest pain require a chest radiograph in the ED.

In general, the American College of Emergency Physician policy on chest pain recommends the liberal use of chest radiography in patients in the ED with non-traumatic chest pain.6 However, the recommendation is not well supported by a study that examined the role of radiography in patients with chest pain in this setting.7 The study found that the radiographs were often not interpreted correctly in the ED and that this investigation did not change the outcomes of patients.7

A number of studies have evaluated the use of routine chest radiography in various patient groups, most of which concluded that there is little additional value to patient management.7^–11 However, the ordering of a routine chest radiograph is commonplace in many clinical practice settings, especially in the ED.

The aim of this study was to determine the prevalence and nature of abnormalities on the chest radiograph in patients presenting to the ED with suspected acute coronary syndrome but no history or physical findings suggestive of other pathology. We hypothesise that this selected patient group is unlikely to benefit from routine chest radiography and that this investigation rarely alters their subsequent management in the ED.

METHODS

This was a retrospective study of patients presenting to a single ED over a 12-month study period (July 2005–June 2006 inclusive). The ED is a tertiary referral mixed adult and paediatric centre in metropolitan Melbourne, Australia, with approximately 55 000 presentations annually. The study was authorised by the institution’s human research and ethics committee.

Patients with a suspected acute coronary syndrome but without symptoms or signs of any other significant pathology were enrolled in the study. The inclusion criteria were age ⩾18 years and suspected acute coronary syndrome (an ECG and cardiac monitoring undertaken and cardiac enzymes (troponin or creatine kinase) measured). The exclusion criteria were symptoms suggestive of other chest pathology (eg, cough, sputum, haemoptysis) and examination findings suggestive of other cardiac disease (eg, murmur, added sounds, pericardial rub, arrhythmia, differential arm blood pressure), other chest pathology (eg, wheeze, crepitations, abnormal air entry, dullness on percussion, pleural rub) or other significant pathology (eg, deep venous thrombosis, congestive cardiac failure, abdominal signs). Patients with an initial vital sign outside the normal range (eg, temperature 36.0–37.5°C, heart rate 60–100/min, systolic blood pressure 100–160 mm Hg, respiratory rate 12–20/min, oxygen saturation >90% on air or >95% on oxygen, Glasgow Coma Score >13) and those presenting after trauma were also excluded.

The primary study end points were the proportion of patients who had a chest radiograph in the ED and the proportion with an abnormal chest radiograph. The chest radiography rates for younger (<40 years) and older (⩾40 years) patients were compared since the younger patients, with less chance of significant chest pathology, most probably had less indication for screening. The secondary end points were the nature of the chest radiographic abnormalities and whether the radiograph changed ED management.

MedTrak, the ED Information System (EDIS), was screened for patients who presented to the ED during the study period with a triage diagnosis of suspected acute coronary syndrome. A master list was prepared after cross-referencing presenting complaint codes (eg, chest pain, epigastric pain, left arm pain, palpitations, atrial fibrillation, supraventricular tachycardia, post-arrest, collapse, short of breath, dizziness) with the written description of the presenting complaint. From this master list, patients were randomly selected using a computer-generated random number table to provide a workable sample likely to include sufficient patients to meet the sample size requirement. The medical records of selected patients were retrieved and further screened for those who met the enrolment criteria.

The principal investigator (JN) collected all data using a data collection document specifically designed for the study. The document was trialled and revised before data collection began. The medical record ED notes were reviewed for evidence of inclusion criteria: cardiac monitoring (nursing notes), ECG recording (hard copy in the notes) and cardiac enzymes taken (hard copy results). Patients who met the inclusion criteria were further screened for exclusion criteria. A second investigator (DT) reviewed a random selection of 10% of all records reviewed and determined 100% accuracy in data collection.

The institution uses an electronic radiology system (Weblog 2000) which informed whether a chest radiograph had been performed in the patients enrolled. The formal radiology reports were reviewed. Reports from earlier chest radiographs, if applicable, were also reviewed and new changes noted. Comments on the radiographs by the ED were extracted from the ED notes.

The medical records were examined in detail to determine if the chest radiographic findings affected patient management based on documentation available in both the medical and nursing notes (eg, treatment clearly prescribed or further investigation undertaken). This allowed a determination of whether there was documented evidence that the chest radiographic findings had changed management. If no documented evidence was apparent, a determination was made as to whether the chest radiographic findings may have changed management based on their extent and likely clinical significance (eg, florid pulmonary oedema). For potential cases where the chest radiographic findings may have changed management, both investigators reviewed the charts independently and a consensus determination was made.

It was expected that the prevalence of significant chest radiographic abnormalities in the patient group examined would be small (approximately 3%). To be 95% certain that the prevalence found in this study would lie ±3% of this expected prevalence (0–6%), approximately 136 patients meeting the enrolment criteria and who had undergone chest radiography had to be enrolled.

All results are presented descriptively with 95% confidence intervals (CI) fitted around important proportions. EpiCalc statistical software Version 1.02 (http://www.brixtonhealth.com/epicalc.html) was used for all analyses.

RESULTS

A total of 5764 patients were identified with a triage diagnosis of suspected acute coronary syndrome. The course of these patients is shown in fig 1. Of the 158 patients enrolled, the mean (SD) age was 55.8 (15.4) years and 102 (64.6%) were men.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Course of patient enrolment.

One hundred and thirty patients (82.2%, 95% CI 75.2% to 87.7%) underwent chest radiography. Most (n = 136, 86.0%) were aged >40 years. The rates of chest radiography in patients aged <40 years (17/ 22, 77.2%) and ⩾ 40 years (113/136, 83.0%) were similar.

Radiologists reported that 40 (30.8%, 95% CI 23.1% to 39.6%) chest radiographs were abnormal; some had multiple abnormalities (table 1). The most common abnormalities were evidence of previous cardiothoracic surgery, cardiomegaly and pulmonary oedema. All cases of cardiomegaly were pre-existing and no report suggested that this had progressed since the previous chest radiograph. Only 12 cases had potentially significant new abnormalities (not present on previous chest radiographs or not previously imaged).

Thirty-seven patients (28.5%) had their chest radiography findings documented by ED doctors, 32 of which (24.6%) were reported as normal. The remaining five abnormal cases were reported as having a raised hemidiaphragm (n = 2), pulmonary oedema (n = 1), suspected pulmonary consolidation (n = 1) and suspected widened mediastinum (n = 1). The last two cases were not substantiated by the formal radiologist report.

Given the result of the chest radiography, there was documented evidence that patient management was changed in three cases. One patient was administered a diuretic for pulmonary oedema and another was given an antibiotic for chest infection. The third patient had an incidental finding of suspicious hilar lymphadenopathy. He was contacted following discharge to arrange further investigations.

In two patients chest radiography may have resulted in a change in management. This was based on the abnormalities documented on the formal radiographic report. One patient had middle lobe consolidation and may have benefited from antibiotic administration and the other had evidence of pulmonary oedema and may have benefited from diuretics, but there was no documentation that these drugs were given.

For the remaining cases of significant abnormalities, there were no changes from previous chest radiographic findings so it was determined that the findings were unlikely to have altered management.

The most common final diagnosis was chest pain (not elsewhere classified), a diagnosis where the chest pain is not specifically diagnosed but not thought to be cardiac in nature (table 2). Overall, acute coronary syndromes accounted for less than one-quarter of the final diagnoses.

Of the 158 patients enrolled, 70 (44.3 %) were discharged from the ED, 50 (31.6%) were admitted to the ED observation unit and 38 (24.0%) were admitted to hospital. Six (5%) of the 120 patients discharged (directly or after observation) presented again within 4 weeks. All causes for re-presenting were irrelevant to the index ED presentation and none was related to the management of the index episode.

DISCUSSION

Health systems have come under pressure from an increasing demand for services in recent years. Emergency departments, as the interface between the community and inpatient care, have been particularly vulnerable and have experienced considerable increases in patient numbers, access block and resource pressures. There is therefore an increasing need for cost-effective and expeditious care in the ED. In this regard, it is sensible to examine the usefulness of investigations in the ED setting.

In this study the exclusion criteria were designed to ensure that the patients enrolled had no suggestion of significant relevant pathology (on history and/or examination) other than possible acute coronary syndrome. Accordingly, these patients had no specific indication for a chest radiograph other than as a screening tool for pathology that could otherwise be missed.

Despite a lack of evidence supporting routine chest radiography, this study found that it was a commonly ordered investigation. Approximately 80% of all patients underwent radiography, the rate being slightly less for younger patients. This rate is higher than that reported by Katz et al5 which exceeded 70% among patients presenting to the ED with chest pain and without a history of trauma. In comparison, Hubbell et al10 prospectively evaluated the impact of routine chest radiography on admission in 491 medical inpatients and found a rate of 60%. In another study they showed that the rate of routine chest radiography in patients aged >65 years in the ED was 71%.11

While the rate of chest radiography was high, the yield of significant pathology was low. Less than one-third of patients had an abnormal finding of any sort according to the radiologists’ reports. This rate was lower than those reported by others, which ranged from 33.0% to 50.5%.7–10 12 In this study most of the abnormalities were inconsequential and only 12 patients were thought to have anything potentially significant. The abnormalities found were consistent with a large prospective outpatient study which reported that the most frequent acute findings were cardiomegaly, pulmonary oedema/congestive cardiac failure, pleural effusion and pneumonia.8

Chest radiography altered the management of only three patients and may have altered the management of a further two. Thus, a possible maximum of five patients (3.8%) had their management altered by the chest radiograph. This proportion is similar to reports of 4.0% and 5.1% from studies in an inpatient medical ward10 and an acute stroke unit, respectively.12

This small proportion of patients who did/may have had their management affected by the chest radiography results is based on the radiologists’ reports. In practice it is the ED doctor who determines management and, ideally, this proportion should be based on their reading of the radiograph. This was not possible in this study owing to poor documentation by the ED doctors. Notably, only five patients had chest radiographic abnormalities documented in the ED notes and two of these “abnormalities” were not verified by the radiologist reports. It is not known if this discrepancy between the ED doctor and radiologist reports resulted from a level of inaccuracy in interpretation among the former or a lack of clinical detail available to the latter. It is conceivable, however, that findings on the chest radiograph may result in overtreatment or undertreatment.

It is interesting to speculate on the outcomes of the five patients for whom the chest radiographic results did/might have changed their management. Certainly, the patient with hilar lymphadenopathy (suspected malignancy) may have benefited as the radiographic findings initiated further investigation. The remaining four patients may not have been affected adversely had they not undergone chest radiography and had been treated on clinical grounds. Indeed, by definition, their vital signs were stable, they had no signs or symptoms of significant illness and none required admission to the hospital.

This study has important limitations. Its retrospective design was associated with data that were sometimes incomplete and possibly inaccurate. Signs or symptoms that would have excluded a patient from the study may not have been recorded, resulting in a potential overestimation of abnormal chest radiographic findings. Other measurement bias was probably limited by the use of a specifically designed and trialled data collection document, a single data extractor, verification of data extraction accuracy and consensus discussion for patients whose radiographs did change or may have changed management. While the consensus decisions that the chest radiographic findings may have changed management were speculative, they provide a measure of sensitivity analysis. The strict and explicit exclusion criteria considerably restricted the proportion of patients who could have been enrolled. However, the study specifically aimed carefully to select a specific group of patients in whom radiography was thought to be screening rather than diagnostic. The study was undertaken in a single ED where usual chest radiography practice may differ from other centres. Accordingly, the findings may lack external validity.

Only the first set of ED vital signs was used to determine entrance to the study. It is possible that some patients developed abnormal vital signs later in their ED stay (eg, fever). In addition, other indications for chest radiography may have become apparent subsequently—for example, signs and/or symptoms may have changed or important diagnoses may have been made (such as myocardial infarction). Hence, some enrolled patients may have developed a legitimate indication for chest radiography. Overall, however, the exclusion criteria appeared to have excluded those patients with significant chest pathology. Indeed, no final diagnosis included congestive cardiac failure, pneumonia or other significant cardiopulmonary pathology.

This study concentrated on the use of the chest radiograph to identify abnormalities that may have changed management through the initiation of treatment. The retrospective design did not allow an examination of cases of diagnostic difficulty where the chest radiograph was used for the exclusion—rather than inclusion—of pathology. In this respect, a normal chest radiograph can be valuable.

This study suggests that chest radiography is of limited usefulness in this defined patient group. Importantly, this finding is limited to this patient group and not to patients with acute coronary syndrome in general. Hence, for this defined patient group, it is recommended that careful consideration be given before a chest radiograph is ordered. It is also recommended that the study be repeated in a multicentre setting with a larger sample size. The study should be prospective with the “rolling” exclusion of patients who develop exclusion criteria during their ED stay. This will assist in better defining the usefulness of chest radiography in this patient group.

CONCLUSION

Chest radiography is a common investigation for patients who present to the ED with possible acute coronary syndrome but without signs or symptoms of other relevant pathology. The proportion of chest radiographs with significant findings is low. Chest radiography rarely alters the management of these patients. Routine chest radiography for this patient group is questionable and further investigation is recommended to better define its usefulness in this setting.