The management of patients with chest pain is a major challenge for health services in developed countries.1^–3 On the one hand, expedited evaluation and treatment of patients with an acute coronary syndrome is highly desirable whereas, on the other, rapid and appropriate reassurance of patients with non-coronary pain is also important.1^–3 It is debatable whether current strategies for managing chest pain are effective.1^–3 While chest pain accounts for up to 30% of all emergency admissions, less than half of these patients have a discharge diagnosis of unstable angina or myocardial infarction. In recent years the number and proportion of patients discharged with non-coronary chest pain has increased dramatically.1 4 To help tackle this problem in England, the government has spent more than £20 million setting up outpatient, “one-stop”, rapid assessment chest pain clinics (RACPCs) to allow speedy evaluation of “people who develop new symptoms that their general practitioner thinks might be due to angina”.5^–7 One aim of these clinics is to prevent unnecessary admissions.1^–12 Whether RACPCs can reliably triage patients according to their risk of a future coronary event is, however, a concern, as highlighted in a recent Health Technology Assessment article.3 In other words, is the reassurance offered to patients considered to have non-coronary pain and to be at “low-risk” accurate and safe? Those patients constitute 30–70% of all attending RACPCs.8^–12 Although the operation of a number of clinics of this type has been described, most have reported only short-term follow-up (one year or less) with only one recent study reporting longer-term outcome (median follow-up 2.6 years).8^–13 We have, therefore, examined longer-term outcome a large cohort of patients evaluated in one of the first RACPCs established in the United Kingdom.12

METHODS

Operation of chest pain clinic

The Western Infirmary RACPC was established in April 1996 and is run by two staff-grade cardiologists.12 It operates Monday to Friday between 1 pm and 5 pm. General practitioners are invited to refer patients presenting with recent onset chest pain that might be angina pectoris (but not those thought to have an acute coronary syndrome) by telephone or fax and patients are usually assessed within 24 hours (except when referred on a Friday or the day before a hospital holiday).

Patient assessment

Patients attending the RACPC have a relevant medical history taken and examination performed. Usually an exercise as well as resting ECG is performed (the former using a full Bruce treadmill protocol). The only exceptions are patients with an obvious acute coronary syndrome or those unable to exercise because of another medical problem. All patient information is entered into a database at the time of presentation.12 Based on the overall assessment the patient is categorised as shown below and managed accordingly.12 A report of the findings of the assessment and recommended treatment is faxed or posted to the general practitioner.

Patient classification

Patients were categorised into one of three main groups:

1. “Low risk chest pain”: patients thought to have non-coronary pain. Usually these patients do not have ST-segment ECG change or symptoms suggestive of myocardial ischaemia on exercise. If any of these patients have previously identified coronary artery disease, their presentation must be thought not to reflect this. These patients were referred back to the GP for further management.

2. “Stable coronary artery disease”: patients thought to have symptoms caused by coronary artery disease. These patients must be thought to be clinically stable and not at a high risk of an early adverse coronary outcome. Anti-platelet and anti-anginal treatments were started immediately and the GP directed to treat patients according to local angina and secondary prevention guidelines; some patients were also referred directly to a cardiology clinic for further follow-up or for coronary angiography.

3. “Acute coronary syndrome”: patients in whom immediate hospitalisation is advised. This category includes patients with unstable angina, suspected acute myocardial infarction or a strongly positive exercise test thought to place the patient at high risk of an early coronary event. These patients were admitted for further investigation and treatment.

During the course of this study, some patients were categorised as “unclassified”. These were patients in whom a diagnosis could not be reached at the clinic, usually because of an inability to exercise on a treadmill or poor performance during the test.

Other patients were categorised as “inappropriately referred” because their primary complaint was not chest pain or any other symptom that might have been due to myocardial ischaemia.

Ethics

This audit used anonymous patient data and its conduct was approved by our local ethics committee.

Patients evaluated

All patients attending between April 1996 and February 2000 were evaluated.

Follow-up by electronic record linkage

Electronic linkage of morbidity and mortality records is possible for all residents of Scotland as previously reported.14^–16 The Scottish Morbidity Record Scheme was used to retrieve details of all hospital discharges (according to the tenth revision of the World Health Organization International Classification of Diseases, ICD-10) up to 30 June 2004. Audits have shown that these data are approximately 90% accurate in identifying the correct discharge diagnosis.17 Deaths and their certified cause (according to the ninth (ICD-9) and tenth revisions of the World Health Organization International Classifications of Diseases) were obtained from the General Registrar’s office, Scotland up to September 2004. We noted the occurrence and timing of discharges and deaths due to a cardiovascular cause (ICD-10 I00–I99, ICD-9 390–450), coronary heart disease (ICD-10 I20–I25, ICD-9 410–414), myocardial infarction (ICD-10 I21, I22, ICD-9 410), angina (I20), unstable angina (I20.0), CABG (ICD-10 K40–K46), PCI (ICD-10 K49, K50.1, K50.8), other chest pain (R07.3 and R07.4), heart failure (ICD-10 I50, ICD-9 428), cerebrovascular disease (ICD-10 I60–69, ICD-9 430–438), respiratory disease (ICD-10 J00–J99, ICD-9 460–519), cancer, all malignant neoplasms (ICD-10 C00–C96, ICD-9 140–208), thoracic cancer (ICD-10 C15, C16, C33, C34, C50, ICD-9 150, 151, 162, 164, 174, 175) and upper GI causes (ICD-10 K25–26, ICD-9 530–537).

Cardiovascular outcomes

We focused on adverse coronary outcomes that might have been prevented by treatment in a patient in whom coronary disease was missed. We pre-specified a hierarchy of composites similar to those used in clinical trials in patients with coronary heart disease (death from coronary artery disease with the sequential addition of non-fatal myocardial infarction, coronary revascularisation and a hospital discharge from angina).18

Comparison of outcomes with local population

We used government vital statistics tables to compare survival of patients in the low-risk group with survival in the overall Glasgow population.19 Also, with the assistance of the General Records Office and Information and Statistics Division of the NHS in Scotland, we calculated the coronary standardised mortality ratios (SMR) for our patient cohorts, relative to the local (Glasgow) population (setting the local population coronary SMR at 100).

Statistical analysis

Statistical analysis was carried out using SPSS version 11.5. Data are presented using frequencies and percentages for categorical variables and means and standard deviations for continuous variables. We compared continuous variables using one-way analysis of variance and categorical variables using the χ² test. We constructed age and sex adjusted event-free survival curves using Cox proportional hazard models to compare morbidity and mortality outcomes between the three risk groups. For each variable entered into a model, the lowest class was set at unity. The assumptions of the models were checked by looking at a log-log plot for each categorical covariate in the model.

RESULTS

A total of 3378 patients (51% male) attended the RACPC during the period of study. Of these, 2036 (60.3%) were categorised as “low risk”, 957 (28.3%) as having stable coronary artery disease and 214 (6.3%) as an acute coronary syndrome. Of the remainder, 156 (67 women), accounting for 4.6% of all referrals, were unclassifiable and 15 (0.4%) referred inappropriately. Record linkage was not possible in 339 patients (10%), of whom 77% were low risk compared to 58% in the group with complete follow-up. Follow-up data were available for 3039 patients over a median of 6.0 (mean 5.8) years (range 3 days to 8.2 years).

Baseline characteristics (table 1)

Women were older than men in all the main diagnostic categories. Patients categorised as having low-risk chest pain were an average of approximately 10 years younger than those thought to have stable or unstable coronary artery disease (p<0.001 for both comparisons). Low-risk patients were also less likely to have a history of, or risk factors for, cardiovascular disease.

Low-risk patients exercised for an average of 3.3 minutes longer than those thought to have stable coronary disease (p<0.001).

Coronary outcomes

Our pre-specified hierarchy of coronary outcomes is shown in table 2 and figure 1. Less than 2% of patients in the low-risk category died from a cardiovascular cause during follow-up and less than 10% experienced one or more components of the broadest coronary composite. Of 1773 patients deemed to be at low-risk following assessment at our RACPC, only 63 (3.6 %) went on to die from coronary heart disease cause or suffer a non-fatal myocardial infarction. Of these 63 patients, 12 (19 %) had previously recognised coronary artery disease (and 40% had recognised coronary artery disease, peripheral arterial disease, diabetes or hypertension).

Patients in the “stable coronary artery disease” category had a threefold to sixfold greater risk of an adverse coronary outcome and that risk was higher still in those identified as having an acute coronary syndrome.

Of the 156 “unclassified” patients, 12% (SD 8.4%) died from coronary heart disease or suffered a non-fatal myocardial infarction and 38% (SD 26.6%) experienced one or more components of the broadest coronary composite.

Figure 1 shows age and sex adjusted event-free survival for the pre-specified coronary composites.

Outcomes in the “low-risk” patients compared to the general population

Compared to the local Glasgow population (coronary SMR  = 100), our “low risk/non-coronary chest pain” cohort had a coronary SMR of 51 (95% CI 31 to 83), the “stable coronary artery disease” cohort 240 (187 to 308) and the “acute coronary syndrome” cohort 780 (509 to 1196). In other words, our low-risk cohort had a coronary SMR half that of the age-matched and sex-matched general population.

Other outcomes

During the period of follow-up after assessment at the RACPC, 16.0% of men in the low-risk category, 19.2% in the stable coronary artery disease category and 26.3% in the acute coronary artery syndrome category were discharged from hospital with a primary diagnosis of “other” (non-coronary) chest pain. These proportions were 13.8%, 28.2%, and 42.3%, respectively, for women.

One per cent of men in the low-risk had coronary artery bypass surgery and 2.3% had percutaneous coronary intervention. These proportions were 0.5% and 1.2%, respectively, in women.

DISCUSSION

The distinction between coronary and non-coronary chest pain can be difficult. Up to 12% of patients with an acute coronary syndrome are sent home form hospital because their problem is not recognised.9 20 Conversely, a much higher proportion of patients with non-coronary pain are admitted to hospital unnecessarily.3 RACPCs have been set up in order to improve the triage of patients with chest pain, although their long-term effectiveness and safety are not established.1^–3 In particular, the long-term outcome of patients labelled as having non-coronary pain or at low risk is unknown and of concern.1^–3 We found that of 1773 patients deemed to be at low risk following assessment at our RACPC, only 63 (3.6%) went on to die from a coronary cause or have a non-fatal myocardial infarction over 6 years of follow-up. Their standardised coronary mortality was half that of the local general population. We believe that this comparison shows that a RACPC can accurately and safely categorise patients as “low risk” although whether our findings can be generalised is an important question. Tenkorang and colleagues published the one year outcome of 425 patients followed for a year after referral to a RACPC at Charing Cross Hospital, London; there were no cardiac deaths among the 310 subjects considered to have non-cardiac chest pain.21 Colleagues from a Scottish district general hospital have presented preliminary 5-year follow-up data on almost 4700 patients referred to a general practitioner run RACPC, reporting a low risk of coronary events in subjects with a negative exercise ECG test.13 More recently, Sekhri described outcomes over a median follow-up of 2.6 years among 8762 referrals with non-cardiac chest pain (n = 6396) or angina (n = 2366) to RACPCs in six English hospitals.22 Patients thought to have non-cardiac chest pain had a 2.73% three-year risk of coronary death or an acute coronary syndrome (compared to 16.52% in the angina group), although coronary SMR was not much reduced from that in the general population. It is noteworthy that only 55% of patients with “non-cardiac” chest pain in that study undertook an exercise test compared to 97% of our “low-risk” patients, which may account for the better risk stratification in the present study.

It may be possible to make triage into a “low-risk” category more effective still. Of the 63 patients experiencing either a coronary death or myocardial infarction, about one in five had previously recognised coronary artery disease (and 40% had recognised coronary artery disease, peripheral arterial disease, diabetes or hypertension). By paying closer attention to patients with established arterial disease or atherosclerotic risk factors, it may possible to further refine triage.23

Although high short-term satisfaction with a RACPC has been reported by both patients and general practitioners, the cardiac-focused, “rule-out”, approach of RACPCs might lead to longer-term dissatisfaction with a lack of definite diagnosis and a high rate of hospital re-attendance.1^–3 24 25 We found that approximately 15% of patients in our low-risk category were subsequently discharged from hospital with a diagnosis of non-coronary chest pain over the 6 years of follow-up after initial assessment at the RACPC. This is a considerably smaller proportion than previous studies estimated would be admitted in the absence of a RACPC.3 8^–12

Six per cent of patients attending the RACPC were diagnosed as having an acute coronary syndrome and these patients had a high event rate in keeping with previous findings. Ideally, such patients would be sent directly to an accident and emergency department rather than to the RACPC, as recommended in our guidelines to referring GPs. This finding presumably reflects the difficulty in diagnosing some acute coronary syndromes.

One limitation of our findings is that 4.6% of patients could not be readily classified into a clinical management category. These individuals had an outcome intermediate between that of “low-risk” patients and those thought to have “stable coronary artery disease”. As has been noted before patients who cannot exercise or perform poorly on a treadmill or bicycle, even if they do not have symptoms or ECG changes during exercise, have a guarded prognosis.26 Women may be particularly liable to misclassification.26 Another limitation is that we did not have information on the application of recommended treatment strategies that could have influenced outcomes following attendance at the RACPC. In addition, we did not have qualitative information on patient satisfaction.

In summary, RACPCs appear to fulfil one of their aims, which is to safely reassure patients thought to have non-coronary pain and to be at low risk. Individuals categorised in this way have a small probability of a major coronary event and a relatively low rate of admission to hospital with non-coronary chest pain over the subsequent 6 years.