Article Text

Download PDFPDF

A descriptive analysis of patients with an emergency department diagnosis of acute pericarditis
  1. Andrew J Hooper1,
  2. Antonio Celenza1,2
  1. 1Department of Emergency Medicine, Sir Charles Gairdner Hospital, Nedlands, Western Australia, Australia
  2. 2Department of Emergency Medicine and Medical Education, University of Western Australia, Nedlands, Western Australia, Australia
  1. Correspondence to Winthrop Professor Antonio Celenza, Discipline of Emergency Medicine, Level 2, R Block, QE II Medical Centre, Hospital Avenue, Nedlands 6009, WA, Australia; tony.celenza{at}


Aim To describe clinical characteristics, assessment and treatment of patients diagnosed in an emergency department (ED) with acute pericarditis.

Methods A medical record review of patients with an ED diagnosis of pericarditis conducted in an adult tertiary hospital over a 5-year period. Variables collected included pain characteristics, associated symptoms, physical examination findings, investigation results, ED treatment and disposition.

Results 179 presentations were included, with 73.9% men and a mean age of 38.8 years. The majority of patients described pleuritic chest pain worse with inspiration with half characterising the pain as sharp or stabbing, with others describing tightness, dullness or cramping. Radiation to the left shoulder occurred in 2.8% and change of pain with posture occurred in 46.4%. A pericardial rub was documented in 19 presentations. All patients had an ECG recorded with ST segment elevation present in 69.3% and PR segment depression in 49.2%. Nearly 90% of patients had troponin testing but only 6.4% of these were positive. Only 8.1% of cases were treated with colchicine. No patients required pericardiocentesis. Patients with high-risk factors were more likely to have previous pericarditis, dyspnoea, nausea, abnormal investigation results, treatment with colchicine and admission to hospital. However, 16.9% of patients without risk factors were admitted, and 46.9% of patients with at least one risk factor were discharged.

Conclusions Pericarditis may not follow the classical clinical description. Admission and discharge decisions appear to relate to individual clinical characteristics rather than known risk factors. Use of colchicine for treatment in ED is infrequent.

  • cardiac care
  • ECG, interpretation
  • clinical assessment

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Acute pericarditis is an important differential diagnosis in the emergency department (ED) assessment of acute chest pain. Review articles have summarised understanding of the clinical presentation and management of acute pericarditis although based on minimal original data.1 ,2 There are few studies of the investigation or management of these patients with the exception of limited studies reviewing echocardiography, troponin use and some specific treatment modalities.3–12 Few of these studies involved ED patients.

Recognition of the ECG findings in pericarditis is important in the assessment of acute chest pain and typical ECG changes are frequently described.13 ,14 The significance of these changes in assisting diagnosis in the ED has not been described previously. Small studies have found that troponin rises are relatively common in acute pericarditis but are not associated with an increased risk of complication.6 ,7

Imazio et al3 describe a management programme for outpatient therapy of low risk acute pericarditis diagnosed in a day hospital setting, but do not describe the population presenting to an ED, or the management of higher risk cases. Acute pericarditis has traditionally been treated with non-steroidal anti-inflammatory drugs (NSAIDs), but recurrence rates have been estimated at up to 32%.9 The use of colchicine as a treatment for first and recurrent presentations of acute pericarditis has demonstrated a reduction in recurrence in both settings.8–10

The broad aims of this study were to describe the clinical features, ECG manifestations, investigation, management and disposition of adult patients who received an ED diagnosis of acute pericarditis. Comparison of patient groups was undertaken based on previously defined risk factors.


Study setting and subjects

This study was undertaken at a single adult urban tertiary referral hospital with an annual census of approximately 54 000 presentations per year with an admission rate of approximately 49%. Approval to conduct the study was obtained from the hospital's human research ethics committee.

A retrospective medical chart review was undertaken on all patients attending the ED who received an ED discharge diagnosis of pericarditis, over the years 2004–2008. Patients were identified from a search of the Emergency Department Information System (EDIS), a clinical and administrative database which records all ED patient presentations. The search used the term ‘pericard*’ in the ED diagnosis field which returns pericarditis; pericardial effusion; pericardial tamponade; and pneumopericardium as diagnoses.

Data acquisition

Following case identification, medical records were obtained and those not fulfilling the correct diagnosis were excluded. Data extracted included patient demographic characteristics, medical history, presenting symptoms, vital signs, physical examination findings, investigation results, treatment, subsequent disposition and follow up. These data were directly entered into a pre-formatted Microsoft Excel 2003 (Microsoft Corp., Redmond, Washington, USA) spreadsheet with variables designed after a review of pericarditis literature. Guidelines regarding parameters of data entry were established prior to data collection, in order to establish consistent coding.

Information was obtained from the medical notes, nursing notes and investigation results. As past medical history is usually well documented in the medical record, absence of a medical condition in the record was assumed to indicate a clinical absence. Other variables were analysed according to whether there was documentation of presence or absence of that variable. All patient symptoms and examination signs were extracted from medical and nursing documentation taken while the patient was in the ED. Investigation results were largely ED based, although some further investigations were performed on a hospital inpatient or outpatient basis and duly recorded.

Statistical analysis

Data were imported into SPSS V.17.0 (SPSS Inc., Chicago, Illinois, USA) and MedCalc V. (MedCalc Software, Mariakerke, Belgium) for analysis. Means with SD were calculated for continuous variables, and proportions with 95% CI for categorical variables. Based on evidence from previous studies, patients were divided into high and low risk groups for comparative analysis.2 ,3 Statistical significance was assessed using the t test or χ2 or Fisher's exact test, respectively. Significance was set at 0.05.

Inter-rater reliability was examined by both authors reviewing a random sample of 10 medical records. Where subjectivity was expected within the review, these variables were then divided into discrete groups (past medical history, symptoms, physical signs, investigation interpretation). Proportion agreement was calculated for all variables including those with multiple responses, and the κ statistic for each variable within those groups with bivariate responses, presented as the mean κ for the group.


Two hundred and forty presentations, comprising 220 patients, were identified by the EDIS search as an ED diagnosis of pericarditis, pericardial effusion, pneumopericardium or pericardial tamponade. Of these, 41 were excluded as they were incorrectly coded, had an ED diagnosis other than pericarditis or did not represent acute pericarditis on review of the medical record. A further 20 sets of medical records could not be obtained, leaving 179 episodes (161 patients) suitable for analysis.

Inter-rater reliability analysis of past medical history showed 97.7% agreement (11 variables) and mean κ 0.95; symptoms 75.8% agreement (15 variables) and mean κ 0.76; physical signs 89.1% agreement (eight variables); and investigation interpretation 83.3% agreement (12 variables).

Demographic characteristics and background

In those patients with an ED diagnosis of acute pericarditis, the mean age was 38.8 (SD 17.4) years and 119/161 (73.9% (95% CI 66.4% to 81.5%)) were men. Mean age of the index presentation for male patients was 37.0 years (SD 16.0) compared with 44.1 years (SD 20.2) for female patients (p=0.043). Of the 179 presentations, 32 (17.9% (95% CI 12.6% to 24.3%)) were representations following a previous diagnosis of pericarditis in the same hospital.

The patients as a group were relatively healthy with hypertension being the most common past medical illness (8.1% (95% CI 4.6% to 13.1%)), followed by ischaemic heart disease (3.7% (95% CI 1.5% to 7.6%)) and rheumatological disease (3.7% (95% CI 1.5% to .6%)). Nineteen patients (11.8% (95% CI 7.3% to 17.8%)) disclosed a past history of pericarditis and one patient had previous myocarditis.

Clinical presentation

Table 1 describes the clinical characteristics of each presentation. Most presentations had pleuritic chest pain worse with inspiration but only half of the documented pain was characterised as sharp or stabbing, with an equal number reporting dull or other chest pain (tightness, cramping). The classically described pain radiation to the left shoulder and arm was uncommon although a change of pain with posture was documented in nearly half the cases. Other clinical features were poorly documented.

Table 1

Clinical symptoms documented in the medical record

Duration of symptoms prior to ED presentation ranged from less than 1 h to 3 weeks although one patient had symptoms for 4 months, and another for 1 year. Eighty-three presentations (46.4% (95% CI 38.9% to 54.0%)) occurred within 1 day of symptom onset, 37 (20.7% (95% CI 15.0% to 27.4%)) between 1–2 days, 38 between 3 days and 2 weeks (21.2% (95% CI 15.5% to 27.9)), 11 presented greater than 2 weeks (6.1% (95% CI 3.1% to 10%)), and 11 did not have symptom duration recorded.

Physical examination

Only seven of 173 presentations with a documented temperature (4.0% (95% CI 1.6% to 8.1%)) had a temperature greater than 38°C in the ED, 2/169 (1.2% (95% CI 0.1% to 4.2%)) had a systolic blood pressure of less than 90 mm  Hg recorded, and 8/107 (7.5% (95% CI 3.3% to 14.2%)) recorded an elevated jugular venous pressure (JVP). Tachycardia with heart rate greater than 100/min (21.8% (95% CI 16.0% to 28.6%)) and respiratory rate greater than 20/min (15.7% (95% CI 10.7% to 21.9%)) were more common, but no hypoxia was recorded. A pericardial rub was noted in 19 presentations (19.4% (95% CI 12.1% to 28.6%)) and was documented as being absent in 79 presentations.


All presentations in the study had an ECG performed and all but one was available for review in the medical records. The record with no ECG available documented ST-segment elevation in the medical notes. Findings are presented in table 2.

Table 2

Electrocardiographic findings

Requested laboratory investigations varied considerably between patients as described in table 3. Troponin levels were abnormal in only 10 of 156 presentations tested initially (6.4% (95% CI 3.1% to 11.5%)) with a further two abnormal results on serial testing. Thyroid function testing demonstrated a single case of hyperthyroidism, and blood cultures showed only a single false positive result. The majority of viral serology and autoantibody testing was negative, but the most common finding was a positive rheumatoid factor in 10 (83.3% (95% CI 51.5% to 97.9%)) of 12 presentations tested.

Table 3

Laboratory investigation requests and results

Chest radiography was performed in 162 presentations (90.5% (95% CI 85.2% to 94.5%)), of which 121 (74.7% (95% CI 67.3% to 81.2%)) were reported as normal. Table 4 describes the reported abnormalities which included some diagnoses which may have been the cause for the patients' presentations, including pneumothorax, pneumomediastinum and lung consolidation.

Table 4

Chest radiograph results

Echocardiography reports were available for review in 57 (31.8% (95% CI 25.0% to 39.2%) presentations. Of those performed, 32 (56.1% (95% CI 42.3% to 69.2%)) were normal and a pericardial effusion was seen in 17 studies (29.8% (95% CI 18.4% to 43.4%)). Although most effusions were small or trivial in size, two cases had evidence of right atrial diastolic collapse.


Treatment details were recorded in 173 (96.6% (95% CI 92.8% to 98.7%)) presentations. The most frequent medications used were NSAIDs, which were given in 84.4% (95% CI 78.1% to 89.5%) of cases. Colchicine was used in only 14 cases (8.1% (95% CI 4.5% to 13.2%)), of which two were repeat presentations. Twelve of the patients treated with colchicine were admitted to hospital under the care of inpatient teams. A wide variety of other medications were used in 46 (26.6% (95% CI 20.2% to 33.8%)) presentations, including antacids, anticoagulants, antianginal agents and antibiotics. Three patients were given corticosteroids. No patients required pericardiocentesis in the ED.


The underlying cause of pericarditis was not defined, or was documented as idiopathic in 134 (74.8% (95% CI 67.8% to 81.0%)) of cases in this study. A presumed viral pericarditis was recorded in 31 patients (17.3% (95% CI 12.1% to 23.7%)), and a further 14 patients (7.8% (95% CI 4.3% to 12.7%)) had another cause for their disease documented in the medical record. These included post cardiovascular procedure, malignancy, radiation, autoimmune and drug related pathology.

Disposition and follow up

The majority of cases (70.4% (95% CI 63.1% to 77.0%)) were discharged directly from the ED with a further four (2.2% (95% CI 0.6% to 5.6%)) discharged after an admission to the ED observation ward. Recommended follow-up was recorded in 142 cases (79.3% (95% CI 72.6% to 85.0%)) and was predominantly with the patient's general practitioner (65.5% (95% CI 57.1% to 73.3%).

Comparative analyses

Further data analyses were performed, comparing patient disposition, sex, age and risk factors previously associated with more severe disease states.2 ,3 These risk factors are listed in the footnote of table 5.

Table 5

Risk factor analysis

Dividing the patients into two groups based on those with or without more than one risk factor demonstrated significant differences as listed in table 5. Using greater than one risk factor as a measure has 54.2% sensitivity for admission and 82.4% specificity for discharge. Alternatively, using any one risk factor has 77.1% sensitivity for admission and 35.1% specificity for discharge.

The analysis of admitted versus discharged patients revealed a number of significant differences in clinical and investigation findings as demonstrated in table 6. Treatment differences between admitted and discharged patients showed significant increased use of colchicine, antacids, paracetamol or opioid analgesia in the admitted group. This may represent an increased severity of disease, or alternatively, in the case of antacid and analgesic use, a diagnostic uncertainty which may have contributed towards admission.

Table 6

Comparison of admitted and discharged patients

Comparison of patients based on age showed those aged 40 years or older were the only patients with comorbid conditions and were more likely to be admitted. This patient subset was significantly more likely to have chest crackles on examination, describe retrosternal pain and an atypical tight or cramping character of pain. Colchicine or anti-anginal medications were more likely to be prescribed for the older group. Younger patients are more likely to be discharged than older patients (figure 1).

Figure 1

Comparison of admissions per decade of age.

Analysis of patient gender revealed that men were more likely to have an abnormal ECG (although less likely to have ST elevation), anti-anginal therapy and cardiology review, despite having a lower mean age than female patients.


This study describes the assessment and management of acute pericarditis when diagnosed in a major tertiary hospital ED. While many of the clinical features, investigations and treatment of the disease are in keeping with current understanding, there are several findings which are not typical.

Chest pain was always almost present, the character and radiation of pain was often atypical for the classical description of pericardial disease. Chest pain was equally recorded as sharp in nature or as a different character including dull, heavy, tight or gripping pain that may more commonly be associated with cardiac ischaemia. Radiation to the trapezoid ridge was not directly described in any patient records, and even including all documented radiation of pain to the neck, shoulder or arm, only 15.6% (95% CI 10.4% to 22.1%) of patients had such symptoms. Clinicians should be aware that the classical stabbing retrosternal chest pain with radiation to the trapezoid ridge is not typical of most patients in this study. Pleuritic or retrosternal pain with positional changes was common and should perhaps be given greater weight than other pain features when evaluating patients for acute pericarditis.

Examination findings were unhelpful in the diagnosis of pericarditis, with no abnormal findings having a high sensitivity. The presence of a pericardial rub was documented in only 19 patients and had a low sensitivity although previous studies suggest high specificity for pericarditis.

Investigations were inconsistently, although all patients had an ECG, and most had white cell count, renal function, troponin and a chest x-ray performed. None of these investigations were sensitive for pericarditis, but abnormal ECG, chest x-ray or troponin were significantly associated with admission to hospital. Imazio and colleagues demonstrated a 32.2% occurrence of elevated troponin I in idiopathic pericarditis which correlated with the extent of myocardial injury, but did not demonstrate a negative prognostic association.4 Elevated inflammatory markers, white cell count, d-dimer or thyroid function tests were not independently associated with inpatient admission.

Risk factors for severe disease have been based on several studies, with the presence of one or more risk factors suggesting an indication for admission.2 In this current analysis of risk factors, data reflecting immunosuppression or coagulopathy were not collected so a direct comparison with Imazio is not possible.3 Patients with more than one risk factor were significantly more likely to be admitted, although patients with elevated troponin, white cell count and fever above 38 degrees were frequently discharged. This may represent unfamiliarity with risk factors for severe disease, or that some patients with these features that can be safely discharged based on clinical impression. Each of the described risk factors may not be regarded as being equally sinister. Some risk factors such as elevated JVP and hypotension, are more easily recognisable as serious clinical features, while fever, raised white cell count, representation or subacute onset may be considered benign or routinely expected characteristics.

Electrocardiography was performed universally in this study, as would be expected in patients presenting to the ED with chest pain syndromes. Typical ST elevation and PR depression were the most common findings although were not universal. PR segment elevation in aVR was seen in 27.4% (95% CI 21.0% to 34.6%) of cases and has been described previously as a ubiquitous characteristic of acute pericarditis.13 ,14 Of the three patients with a low voltage QRS suggesting pericardial effusion, none required emergent drainage. Only a small number of patients in our study had echocardiographic analysis, making it an unreliable criterion for the assessment of acute risk in this study. Of the 17 patients with a documented effusion, 13 were admitted. All patients with a wall motion abnormality were admitted.

The high percentage of patients treated with NSAIDs reflects conventional practice, but the low number of cases treated with colchicine (7.8% (95% CI 4.3% to 12.5%)) is at odds with evidence that colchicine in addition to NSAIDs reduces recurrence rates of pericarditis following a first presentation.8

The wide variety of other drugs used by clinicians in our study may be due to initial diagnostic uncertainty in this patient group. Of note, however, three patients were treated with steroids despite good evidence that use of corticosteroids can favour recurrence.8 ,12


This study describes the patients with an ED diagnosis of pericarditis, and does not capture patients who were not diagnosed in ED but subsequently diagnosed with pericarditis. The lack of gold standard diagnostic criteria is also problematic, particularly in those patients with an ED diagnosis of pericarditis who may have subsequently had a different diagnosis. This was probably the case for those patients with pneumothorax or pneumomediastinum on review of their chest x-rays. This may also have been the case in patients with a wall motion abnormality on echocardiography, indicating a potential diagnosis of myocardial infarction with or without acute pericarditis.

As with any retrospective chart review, there are numerous methodological issues that limit our ability to draw conclusions from this study. The primary obstacle faced was the often poor and incomplete documentation in the charts of all aspects of patient presentations. The reported agreement between data extractors suggests a high level of accuracy in data extraction.

The use of EDIS to retrieve data also has inherent problems. The exit diagnosis is frequently entered by clerical staff, and may be erroneous if medical documentation in the notes is not adequate. In addition, patients diagnosed with pericardial effusion or tamponade may have had acute pericarditis as their underlying problem, but did not have this diagnosis entered into the EDIS database.


This study has demonstrated the clinical features, investigations and management of acute pericarditis in a real world ED setting. Classically described clinical and electrocardiographic features are not present in all patients. In particular, ST-segment elevation was present in approximately 70% of patients, pleuritic pain in approximately two-thirds, and postural changes in pain or PR-segment depression in less than half of the patients with the diagnosis. The criteria for in-patient management warrant further investigation in a prospective study, as patients with high-risk features were safely managed as out-patients in this review, and conversely patients without identified risk factors were admitted to hospital.

Routine use of colchicine in addition to NSAIDs, should be emphasised as a first line treatment for acute pericarditis, which may reduce recurrence and avoid ED re-presentation.



  • Contributors Both authors contributed to the study concept, planning and design, drafting and review of manuscript; AH provided database design and data entry; AC provided data analysis. Both authors are responsible as guarantor for the content of this study.

  • Funding None.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.