Associations between Chest Pain, Diagnosis, and Clinical Outcome in Patients Hospitalized with Acute Dyspnea: Data from the ACE 2 Study

Dyspnea is referred to as a subjective experience of distressed breathing [1] and is defined as being acute if it arises within hours to days. Acute dyspnea is a common cause of hospitalization, and is associated with unfavorable prognosis due to high risk of readmission and mortality [2]. Acute dyspnea can be caused by a spectrum of diseases, with acute heart failure (HF) and acute exacerbations of chronic obstructive pulmonary disease (COPD) as two common etiologies [3]. Dyspnea is the cardinal symptom for both acute HF and acute exacerbation of COPD, but the prevalence and significance of other symptoms in patients hospitalized with acute dyspnea are not known.

Chest pain is another common symptom reported by patients being evaluated in the Emergency Department (ED). The etiology of chest pain can be cardiac or non-cardiac, and it is especially important to diagnose acute coronary syndrome (ACS) in patients with chest pain. However, there are also several more benign conditions that may lead to chest pain, like gastroesophageal reflux, esophageal spasm, musculoskeletal pain, and panic attacks [4]. Prognosis of patients with chest pain is dependent on the underlying cause [5], and with several benign conditions also causing chest pain, the clinical significance of having chest pain prior to hospital admission for dyspnea is not clear. The large number of etiologies resulting in chest pain symptoms may also result in chest pain being prevalent among patients hospitalized with acute dyspnea. Accordingly, in the current investigation, we wanted to examine the prevalence of chest pain symptoms in patients hospitalized with acute dyspnea, and whether concomitant chest pain associates with clinical outcome in patients with acute dyspnea.

We used data from the Akershus Cardiac Examination (ACE) 2 Study. The ACE 2 Study was a prospective, single-center study performed at Akershus University Hospital (Lørenskog, Norway) from June 2009 to November 2010. Akershus University Hospital is a teaching hospital affiliated with the University of Oslo and the catchment area was 460,000 during the time of patient inclusion. Details of the study protocol and study conduction have previously been reported [6].

All patients included into the ACE 2 cohort were considered having dyspnea as their primary complaint by the attending physician in the ED and fulfilling inclusion criteria and not excluded due to exclusion criteria. Briefly, the inclusion criteria were acute dyspnea as the main cause of hospitalization, age ≥18 years, and the ability to provide informed consent. The main exclusion criteria were dementia or other conditions that made informed consent challenging, myocardial infarction, coronary intervention, or any major surgical procedure within the last 14 days and known metastatic cancer or other somatic diseases in pre-terminal/terminal phase. Dedicated study personnel assessed eligible patients for inclusion into the study and filled a standard questionnaire where all clinical data, including information on self-reported chest pain, was obtained directly from the patients and/or the treating physicians in the ED. Study personnel used a standardized questionnaire and specifically asked the patients with self-reported chest pain about clinically relevant chest pain characteristics, which can be associated with certain etiologies of chest pain, both cardiac and non-cardiac. Patients were asked to localize the chest pain and inform if the chest pain was retrosternal/central or localized in other regions of the chest. Quality of chest pain was registered as squeezing, stabbing, burning or “other” in cases where patients were unable to characterize the quality of pain. Patients were asked if the chest pain radiated or not, and in cases there was radiation of chest pain, the site of radiation of chest pain was registered. It was registered if the chest pain was constant or there were certain factors such as physical exertion, deep inspiration, or palpation of the chest possibly associated with provocation of chest pain. All blood samples were obtained within 24 h of admission. Patient medical records were used to attain data on relevant clinical variables, such as heart rate, blood pressure, body temperature, electrocardiogram, and previous medical history.

Echocardiography was used to determine left ventricular ejection fraction and other clinically relevant echocardiographic indices. The echocardiogram was obtained as a part of the clinical routine; therefore, where HF was considered unlikely, echocardiogram may not have been performed. This approach is analogous to prior studies of unselected patients with acute dyspnea where not all included patients received echocardiographic examination [7, 8].

An adjudication committee, consisting of two senior physicians, established the final diagnosis after the index hospitalization by independently reviewing the patient electronic records, including results of additional examinations and follow-up data with median follow-up of 464 days (interquartile range [IQR] 304–705 days) from index hospitalization to adjudication. There was 95% concordance in the index hospitalization diagnoses adjudicated by the two senior physicians, and discrepancy was resolved by consensus.

The criteria outlined by the European Society of Cardiology [9] were used to establish the diagnosis of HF, and COPD diagnosis was based on Global Initiative of Obstructive Lung Disease (GOLD) [10]. Based on clinical findings, electrocardiogram results and biomarker concentrations, the adjudicators registered if patients had a likely underlying ischemic etiology as the cause of acute dyspnea. Survival status was determined from the patient’s electronic medical records, which are linked to the national death registry in Norway, with median follow-up of 823 (IQR 471-998) days.

Blood specimens were acquired within 24 h of hospitalization. After centrifugation, the serum was stored immediately at – 80°C. No preceding freeze-thaw cycles were performed prior to the conduction of the biochemical analysis, securing stability for the analytes, as previously reported [11]. A high-sensitivity assay (Elecsys TnT hs stat, Roche Diagnostics, Penzberg, Germany) and ProBNP II assay (Roche Diagnostics) on a Cobas 8000 Platform (Roche Diagnostics) at Akershus University Hospital were used for measuring cardiac troponin T (cTnT) and N-terminal Pro-B-type natriuretic peptide (NT-proBNP) concentrations, respectively. The estimated glomerular filtration rate was calculated using the Epidemiology Collaboration (CKD-EPI) equation [12, 13] as previously reported.

All continuous variables are presented as median [IQR] and compared with the Mann-Whitney U test, categorical variables are presented as absolute numbers and percentages (%) and compared with the chi-squared test. As a result of right skewed distributions, we transformed concentrations of cTnT and NT-proBNP by the natural logarithm before inclusion as continuous variables in regression models.

We assessed the prognostic accuracy in predicting all-cause mortality using receiver operating characteristic curve (ROC) analysis with the area under the curve (AUC). All-cause mortality stratified by chest pain on admission was illustrated using Kaplan-Meier plots and compared using the log-rank test. Cox proportional hazards models were used to assess predictors of survival. We compared the Cox proportional hazards models by using the Akaike information criterion and the relative likelihoods of the two models. The survival models were adjusted for sex, age, body mass index, systolic blood pressure, history of HF, coronary artery disease, atrial fibrillation, diabetes mellitus, COPD, and estimated glomerular filtration rate.

Logistic regression models were used to determine associations of demographic variables, co-morbidities, and biomarkers with self-reported chest pain, both in the total cohort and in patients with HF and COPD separately. We performed additional analysis with patients stratified according to chest pain on admission and concentrations of cTnT below and above the 99-percentile reported in a healthy population (<14 ng/L): (1) cTnT<14 ng/L and no chest pain, (2) cTnT<14 ng/L and chest pain, (3) cTnT>14 ng/L and no chest pain, (4) cTnT>14 ng/L and chest pain. We assumed statistical significance at p < 0.05. All analyses were performed using Stata 17 (StataCorp LP, College Station, TX, USA).

A total of 468 patients with acute dyspnea on admission were screened for inclusion, and 87 patients were excluded due to the exclusion criteria, 67 patients refused to participate, and 1 patient was excluded due to unknown chest pain status on admission (Fig. 1). Hence, the total cohort for the current study was 313 patients.

Adjudicated diagnoses for the index hospitalization were acute HF (n = 143, 46% of the population), 83 patients classified as hospitalized due to acute COPD (26% of the population), and 87 patients classified with non-HF and non-COPD related dyspnea (28% of the population). Median age of the patients was 73 (IQR 63–81) years, and 163 (52%) patients were male. History of CAD was present in 111 patients (36%), 101 patients had history of HF (32%), and 154 patients had prior diagnosis of COPD (49%).

Among patients with and without chest pain on admission, there were no statistically significant differences in age, sex, co-morbidities (HF, CAD, diabetes mellitus, AF, COPD, kidney failure), all-cause mortality and biomarker (cTnT and NT-proBNP) concentrations (Table 1). We found no associations between presence of chest pain on admission and patient demographics, co-morbidities, or biomarkers in unadjusted logistic regression analysis (Table 2). In adjusted logistic regression analysis, we found chest pain to be more prevalent in female patients compared to male patients: OR 0.53 (95% CI: 0.30–0.93).

We performed subgroup analysis for patients with HF (online suppl. Table 1A; for all online suppl. material, see https://doi.org/10.1159/000541897) and COPD (online suppl. Table 1B). In patients with HF as the adjudicated cause of admission, there were no associations between chest pain on admission and patient demographics, co-morbidities, or biomarkers. In patients with COPD as adjudicated cause of hospitalization, cTnT and age were associated with chest pain on admission in adjusted analysis.

During a median follow-up time of 823 days (IQR 471–998), 114 (36%) patients died. When stratifying patients by chest pain on admission for the total cohort, we found no statistically significant difference in survival (p = 0.09 by log-rank test; Fig. 2a). There was no significant association between chest pain on admission and all-cause mortality using univariate or multivariable Cox proportional regression models (Table 3). ROC-AUC for chest pain on admission for all-cause mortality was 0.54 (95% CI: 0.49–0.59; Table 3).

In patients with HF, chest pain on hospital admission was not associated with survival during follow-up (p for log-rank test = 0.45; Fig. 2b). Association with all-cause mortality was seen neither in univariate nor multivariable Cox proportional regression models (Table 3). We found similar results also for patients with adjudicated diagnosis of acute COPD (Fig. 2c; Table 3). We also did not find associations between specific chest pain characteristics and all-cause mortality in the total cohort (online suppl. Table 2; Fig. 1).

We tested the predictive ability of different subgroups according to presence or absence of chest pain on admission and concentrations of cTnT above (≥14 ng/L) and below (<14 ng/L) the 99-percentile in a healthy population. Patients with high cTnT concentrations had worse prognosis, independently of presence of chest pain (Fig. 3).

As previously reported, cTnT predicted outcome in our population with HR 1.30 (95% CI: 1.07–1.58) per log-cTnT unit increase [14]. The ROC-AUC for all-cause mortality for cTnT measurements alone was 0.70 (95% CI: 0.64–0.75) and ROC-AUC for cTnT and chest pain was 0.70 (0.64–0.76); p for comparison 0.46.

We classified all patients receiving a diagnosis of ACS by the original adjudication committee (2009-2010) as hospitalized with assumed myocardial ischemic etiology and the other patients as hospitalized with assumed non-ischemic etiology. We performed a subgroup analysis, where we classified patients according to either having chest pain with assumed myocardial ischemic etiology (n = 19) or chest pain of assumed non-ischemic etiology (n = 76). When the patients with chest pain were stratified according to etiology, there was no statistically significant difference in survival (p = 0.27 by log-rank test, online suppl. Fig. 2). There was no association of chest pain with mortality, neither in univariate nor multivariate cox-regression analysis (online suppl. Table 3).

The main result of this study is that concurrent chest pain on admission does not associate with specific groups of etiologies in unselected patients hospitalized with acute dyspnea, nor with clinical outcome during follow-up. Chest pain and acute dyspnea are common causes of admission at the ED, with dyspnea comprising 5–9% and chest pain 11–13% of total ED evaluations [15‒17]. The two symptoms may appear concomitantly as both symptoms are often associated with cardiopulmonary diseases. Patients hospitalized with dyspnea at the ED, have higher risk for in-hospital mortality, longer hospital stay, and higher risk for admission to the intensive care unit as compared to patients with chest pain as the cause of hospitalization [17, 18]. Similarly to dyspnea, chest pain has a broad spectrum of etiologies, varying from benign causes to life-threatening diseases, which are clinically often indistinguishable from each other. This can cause diagnostic challenges for physicians in the ED, and misdiagnosis can have a serious impact on the health of the patients. Intensity of chest pain does not always correlate with the severity of the underlying condition, and less intense chest pain should not be used to exclude life-threatening pathology [19]. The discharge rate in patients with unexplained chest pain, either directly from the ED or after hospitalization, has been reported as being 11.8–23% for men, and 8.1–25.3% for women. Patients discharged with unexplained chest pain have been reported at increased risk of cardiovascular mortality within the first-year post-discharge [20], but whether this relates also to patients with dyspnea is not known.

cTnT, the cornerstone in the evaluation of acute chest pain, has diagnostic and prognostic implications and is used to “rule in” or “rule out” ACS in patients presenting with chest pain [21‒23]. However, there are conditions not necessarily related to ischemia, such as cardiac toxicity, myocarditis, pulmonary embolism, and COPD where the presenting symptom can be chest pain and cTnT concentrations can be elevated [24, 25]. Therefore, when evaluating patients with acute chest pain, it is not advisable to rely just on clinical findings and cTnT, a thorough systematic evaluation is advisable, with a focus on medical history, risk factors, and associated symptoms, such as acute dyspnea [26]. The presence of acute dyspnea in patients presenting with chest pain is associated with poor prognosis [27]; however, to the best of our knowledge, it is not known if the presence of chest pain in patients presenting with acute dyspnea as the main complaint has prognostic implications. Compared to chest pain, guidelines for evaluation of patients presenting with acute dyspnea are less established [28] and we wanted to assess whether co-existing chest pain in patients with primarily acute dyspnea has diagnostic and prognostic implications.

In our cohort of 313 patients with acute dyspnea as the main cause for hospitalization, all-cause mortality was not higher in patients presenting with chest pain. Patients with acute dyspnea presenting at the ED generally have a worse prognosis than patients presenting with chest pain, and acute dyspnea is an independent prognostic marker. Dyspnea is often caused by serious conditions such as COPD and HF, and these patients have reduced organ reserves, for example, reduced lung capacity and reduced cardiac function, with increases also short- and medium-term mortality [28, 29]. The high overall mortality rate among patients with acute dyspnea could possibly explain why concomitant chest pain did not directly influence the outcome in our cohort, as any un-diagnosed, underlying coronary artery disease may more influence long-term mortality in this cohort.

We found concomitant chest pain on admission to be more prevalent in female than male patients. Prior reports have reported a higher prevalence of non-ischemic chest pain in women on hospital admission [30]. In our cohort, only a few patients were adjudicated with ACS, and it is likely that most patients had a non-ischemic cause of chest pain. We do not know whether the finding of more women having chest pain in our cohort also correlates with predominantly non-ischemic etiology of chest pain in our cohort, but this would also explain a lack of association between chest pain and clinical outcome in our patients. As previously reported, elevated cTnT concentrations were associated with mortality in our cohort [14, 31]. However, there was no statistically significant association of chest pain on admission with cTnT concentrations, which could indicate that elevated cTnT in our cohort was not directly associated with chest pain but was more a proxy of the underlying chronic conditions than any un-diagnosed coronary artery disease. ACS is associated with an adverse prognosis [32]; however, in our cohort, where acute dyspnea was the dominating symptom, the presence of chest pain and ACS was not associated with all-cause mortality.

The strengths of our study are that we explicitly asked the patients whether they had experienced chest pain during the last 24 h, and that all patient-related data and biological specimens were collected systematically and in a uniform manner by dedicated personnel. There was also excellent agreement between the adjudicators when diagnosing the cause of dyspnea, which should increase the validity of our results.

A major limitation of our study is the single-center design with a moderate sample size, which increases the risk of both type 1 and type 2 errors. We therefore encourage validation of our results in additional cohorts, possibly by using a larger sample size and involving multiple centers. Due to the relatively short follow-up, it is possible that the prognostic effect of chest pain in patients with acute dyspnea may have been underestimated. Patients in the cohort were dichotomized as either having chest pain or not having chest pain, but the underlying etiology of chest pain was not taken into consideration, which likely could have influenced the association between chest pain and adjudicated diagnosis and clinical outcome.

In patients hospitalized with acute dyspnea, chest pain did not associate with cTnT concentrations on admission or mortality during follow-up. Chest pain is a heterogenous condition, which could possibly explain the lack of prognostic value of chest pain when acute dyspnea is the dominating symptom.

This study involves human participants and was approved by the Regional Committee for Medical Research Ethics Southeast Norway; reference number for ethics approval: S-08824d 2008/21288 09/418. All study participants provided written consent before study inclusion.

The authors have no conflicts of interest to declare.

This work was supported by a research grant from the Norwegian Research Council (197992/B-07029) to T.O. and H.R. and by internal grants from Akershus University Hospital (41810/340104). Roche Diagnostics supported the study by providing reagents at a reduced price to Akershus University Hospital.

R.B., K.B., M.N.L., and H.R. had access to all of the data in the study; contributed to study conception and design, analysis and interpretation of data, and drafting of the manuscript; and take full responsibility for the overall content as guarantors. K.B., A.D.H., T.O., and H.R. contributed to acquisition of data. K.B., A.D.H., T.O., M.N.L., and H.R. contributed to critical revision of the manuscript for important intellectual content. R.B., K.B., A.D.H., T.O., M.N.L., and H.R. contributed to final approval of manuscript.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.