Study design and setting

A prospective cohort study conducted from May 2012 to August 2013 in two teaching hospitals.

The Second Affiliated Hospital of Guangzhou Medical University (AHGZMU)is an academic hospital with 1500 beds affiliated with the Guangzhou Medical University and serves a population of approximately 1.56 million people in the Hai Zhu district, Guangzhou. The ED receives more than 150,000 new patients per annum and serves a local population of approximately 1,680,000 people. The Prince of Wales Hospital (PWH) is located in the New Territories in Hong Kong. It is a university hospital with 1400 beds. It sees approximately 150,000 new ED patients per annum and serves a local population of approximately 800,000 people.

Inclusion and exclusion criteria

Consecutive ED patient ≥18 years old, with a chief complaint of chest pain or discomfort were recruited. Patients were excluded if they clearly had a non-cardiac cause of chest pain. Confirmed STEMI patients at ED presentation were also excluded as they did not have undifferentiated chest pain. Patients were excluded if they were unable or unwilling to provide informed consent or unable to be contacted after discharge.

Measurements and data collection

Baseline patient characteristics, the characteristics of the chest pain (such as location, feature and radiation), vital signs, medical history, family history of coronary artery disease (CAD), ECG results and contact information to facilitate subsequent follow-up were collected and recorded in a computerized database. The HEART score was recorded and calculated by research staff. The details of the HEART score are shown in Additional file 1. Guidelines recommend that patients with possible ACS but with a normal initial ECG and cardiac markers should be observed in a chest pain unit, where continuous cardiac monitoring and repeated measurement of cardiac markers is available [12]. Point-of care troponin testswere used in GZ, whilst high sensitivity troponin tests were used in HK. Therefore, the results of troponin were divided into four groups (≤1× normal range, 1×-2× normal range, 2×-3× normal range and ≥ 3× normal range) for analysis. ST-depression on initialECG means ST depression ≥1.0mv in any lead of the initial ECG. The onset-ED time and ED-ECG time were recorded. Follow-up data were obtained from the Clinical Management System (CMS) in HK and Health Insurance Information Management System (HIIMS) in GZ.

Follow up

Subsequent visits to ED, hospital readmission for evaluation of chest pain and all cardiac procedures performed were retrieved from CMS in HK and HIIMS in GZ, and verified via telephone at 30-days and 6-months follow-up after initial presentation. Furthermore, death, myocardial infarction, readmission for ACS, and all cardiac testing and coronary revascularization procedures were also obtained via CMS and HIIMS.

Sample size calculation

According to our previous study [7], the 30-day MACE rates in Chinese patients presenting to ED with cardiac chest pain could have been as high as the upper limit of the 95% confidence interval in the low-risk group i.e. 13%, and as low as the lower limit of the 95% confidence interval in the high-risk group i.e. 23%. To achieve adequate power to address the objectives by using 2-tailed alpha of 0.05 and a power of 80%, the minimum sample size required per group is 220. We aimed to recruit an extra 30% in case of unforeseen circumstances and thus at least 230 (230 × 1.3 = 299) patients were required per group. Therefore, the minimum sample size in this study was 598 (299 × 2 recruiting sites = 598).

Outcomes

The primary outcomefor each patient wasall cause mortality within 30 days and 6 monthsafter initial ED presentation. The secondary outcomes were the contributory factors of primary outcomes.

Statistical analyses

Continuous variables were presented as median and interquartile range or mean ± standard deviation as appropriate. Categorical variables were expressed as frequencies. Statistical analysis was performed using SPSS v17.0 (SPSS Inc., IL, USA) and Medcalc v9.5 (MedCalc Software, Mariakerke, Belgium). Summary statistics were used to describe patient characteristics from the Hong Kong and Guangzhou groups. Chi-square analysis was used for categorical variables, whilst independent t-tests or Mann-Whitney U-tests were used for comparing data from continuous variables. An initial univariate analysis was performed on all variables with 30-day and 6-month mortalityas the dependent variables and presented as unadjusted odds ratios (OR) and 95% confidence intervals (95% CI). Variables with p ≤ 0.1 were entered into a multivariate ordinal logistic regression. Cumulative survival curves in relation to different hospitals were determined according to the Kaplan-Meier method with the use of log-rank tests for statistical assessment. Cox regression analysis was used to calculate unadjusted and adjusted hazard ratios (HRs) and 95% CIs for mortality rates after 30 days and 6 months in relation to different hospitals. A P value of < 0.05 was considered statistically significant and all probabilities were two tailed.

Recruitment of patients from the two centers

Patients were recruited from 17 March 2012 to 14 August 2013. The flow chart of the patient recruitment was shown in Fig. 1. There were 1025 eligible patients, with 589 cases in HK and 436 cases in GZ. Twenty-nine patients were excluded due to lack of follow-up data, leaving 996 patients (407 cases from GZ and 589 cases from HK) in the study. The mean age of the 996 patients was 65.1 ± 14.5 years and 55.3% were males.

Baseline characteristics and final diagnoses of chest pain patients

Comparison of clinical outcomes of chest pain patients in the two centres

The 6-month mortality of ischemia, non-ischaemic cardiac and non-cardiac causes in GZ were 1.7%, 1.2% and 1.7% respectively, whilst in HK these were 0.3%, 0.5% and 1.0% respectively. NSTEMI patients demonstrated a higher mortality than other ischaemic chest pain patients at 30-day and 6-month follow-up. Compared with cardiac aetiologies, higher mortality was observed for non-cardiac causes. Most diagnoses of non-cardiac chest pain were pulmonary embolism, pneumonia, ischemia stroke and cancer (see Table 2 for details).

Univariate and multivariate logistic regression for 30-day and 6-month mortality

Survival analysis in two hospitals

Figure 2 showed the corresponding Kaplan-Meier event-free survival curves for different hospitals. Kaplan-Meier analysis demonstrated a significantly increased probability of death ofchest pain patients in GZ after 30 days and 6 months (P < 0.0001 for 30-day mortality using the log-rank test, and P = 0.010 for 6-month mortality using the log-rank test).

After adjusting for age, hyperlipidemia, CHF, SBP, ST segment depression, NHYA classification, troponin, HEART score and serum creatinine, the adjusted HR for 6-month (HR = 1.77, 95% CI 0.73 to 4.26, P = 0.205) mortality in GZ was not significantly higher than that in HK.

Different distributions of chest pain patients between two hospitals

Over 50% of patients were diagnosed as non-cardiac chest pain in HK, a much higher rate than in GZ (37.1%). This may reflect the different healthcare systems. In Guangzhou, chest pain patients at low risk often present to clinics instead of EDs during office hours. The ED is the only choice for patients with chest pain in HK due to lower medical costs. So, the patients with urgent and severe chest pain/symptoms would likely to present to EDs in GZ. Compared to other studies, the proportion of non-cardiac chest pain patientsin this study was low. The proportions in GZ and HK were 50% and 37.1% respectively compared to 67.2%–70% in other studies [14–16]. A lower proportion of non-cardiac chest pain patients may allow earlier discharge from the ED and place less stress on ED and hospital services.

There were no significant differences in 30-day and 6-month mortality for non-ischemic cardiac chest pain and non-cardiac chest pain between the two hospitals. However, there was significantly higher 30-day and 6-month mortality rates for ischemic cardiac chest pain patients in GZ. ACS is a high-risk disease with significant mortality [9, 17]. The current study excluded STEMI patients but NSTEMI and UA patients were recruited. The combined proportion of NSTEMI and UA in GZ was 25.8%, which was higher than thatin HK (17.5%). The proportions of NSTEMI and UA were also higher than those in Cullen’s study (8.7%) whilst non ACS cardiovascular-related conditions were 20.8% [14].

Different clinical characteristics and outcomes of patients between two hospitals

Since the patients with severe chest pain/symptoms would likely to present to EDs in GZ while those patients with mild chest pains/symptoms would visit clinics, so the baseline clinical characteristicsof patients in GZ were worse than those in HK. The patients were older, had lower proportion of NHYA class I and higher creatinine level, higherproportion of ST-segment depression, lower proportion of negative TnT and higher HEART scoreat ED presentation. Those factors have also been shown to be associated with 30-day and 6-month mortality. The same parameters have been verified to predict short- and long-term risks of death from ACS [18–24]. Compared to HK, lower proportions of patients with risk factors were seen in GZ, whilst more patients with worse NHYA class were observed. This suggests that many chest pain patients in GZ did not have any knowledge of cardiac risk factors and were unaware of the presence of those risk factors, therefore the impact of those risk factors might be underestimated. Early detection and good control of risk factors for suspected cardiac chest pain patients may reduce mortality.

Furthermore, elevated serum creatinine level was an independent predictor for 30-day and 6-month mortality in our study.It has also been reported to be associated with worse outcomes for ACS and critically ill patients [18, 23]. Elevated serum creatinine can reflect vascular damage, renal impairment, endothelial dysfunction and impaired myocardial blood flow [24–27]. A multinational registry study by Tang et al. showed that SBP could predict the 6-month mortality of all subsets of ACS. As SBP increases by 10 mmHg, the hazard ratio would decrease by 0.95 [27].

Several studies have reported that the HEART score is not only related to the severity of chest pain in EDs but also predicts the occurrence of clinical endpoints [28–32]. Our study also demonstrated that the HEART score was positively associated with 30-day and 6-month mortality of suspected cardiac chest pain patients in the ED setting.

The dissimilarity of healthcare systems in the two cities is another reason for the different ED management. Limited capacity of hospital emergency care services, high out-of-pocket expenses with the need for up-front payment, prolonged discussions with the patient and families for both obtaining consent and pooling funds are likely to be major contributing factors for the long pre-hospital, in-hosptial delay and refused some expensive treatments [13, 32].

In ED chest pain pathways, guidelines recommend that an ECG should be recorded within 10 min and myocardial injury biomarkers should be measured as soon as possible in all suspected cardiac chest pain patients [4, 9]. The ED-ECG time and the proportion of patients having myocardial injury biomarkers measured in both hospitals met these guidelines.Myocardial injury biomarkers, such as troponin, CK and CK-MB are used to identify myocardial injury [33]. Many studies have shown that troponinis superior to CK and CK-MB as a biomarker for detection of myocardial injury [19, 20]. Therefore, troponin testing is recommended in most guidelines for its high sensitivity and specificity [17]. However in GZ, CK and CK-MB are still mainly used for the diagnosis of AMI and ACS due to their lower costs. High expenses with the need for up-front payment of troponin testing are likely to be a major contributing factor for the lower rate of use of troponin testing in GZ. The lower sensitivities and specificities of CK and CK-MB might contribute to the delayed diagnoses of AMI/ACS and later treatment of patients, possibly contributing to the higher early mortality in GZ. Except the mortaliy, readmissions after ACS are associated with higher long-term all-cause mortality [34].However, our study did not show the significant difference of 30-day and 6-month readmission rate in in HK and GZ. The reason was that the patients we enrolled were suspected cardiac chest pain patients instead of ACS.

Strengths and limitations

The strengths of this study include the fact it is the first comparison of outcomes in different healthcare systems in a similarly populated region of southern China, which effectively eliminates many confounding factors in any comparison. The current study also has some limitations, including the sample size of this study which met our precalculated size but may still be considered small by international standards. Secondly, patient risk factors (i.e. the past medical histories) may have been underestimated due to variations in history taking between the two centres. Thirdly, as a prospective observational study, propensity score with matching would attempt to reduce the bias due to confounding variables. However, this method requires large sample size. Also, the procedure only controls for observed variables, so any hidden bias due to latent variables may remain after matching.Logistic regression and Cox regression can meet the need of this study.