Management and outcomes of patients with chest pain and psychiatric disorders in the era of high‐sensitivity cardiac troponins

The management of patients with psychiatric disease and chest pain in the emergency department (ED) in the era of high‐sensitivity cardiac troponin assays is unexplored.


Introduction
Patients with psychiatric disorders have a high cardiovascular risk factor burden and subsequently a higher risk of cardiovascular events compared with the general population [1]. Prior studies have found that patients with severe mental illness who suffer from an acute myocardial infarction (MI) less often undergo both diagnostic and interventional procedures and are less often treated with guidelinerecommended cardioprotective medications [1][2][3][4][5].
The introduction of high-sensitivity cardiac troponin (hs-cTn) assays in clinical practice has improved the ability to detect and quantify cardiomyocyte injury, leading to enhancements in the low-end accuracy for the diagnosis of MI [6][7][8][9][10]. In addition, the use of hs-cTn assays has increased the proportion of patients with detectable troponin concentrations and subsequently allowed for an improved stratification of patients according to cardiovascular risk [11][12][13][14][15]. Prior studies on cardiovascular risks associated with mental illness in acute medical healthcare settings have mainly focused on population of patients actually diagnosed with MI. Hence, the implications of psychiatric disease in the emergency department (ED) among patients with symptoms suggestive of acute cardiac disease, particularly in the era of hs-cTn assays, is largely unknown.
Using a large observational cohort of patients seeking ED care due to chest pain, we sought to investigate management strategies and prognosis associated with psychiatric disorders.

Data sources
The study was based on data from seven large EDs in Stockholm and Göteborg, Sweden, between 9 December 2010 and 1 December 2016. Data from each hospital's administrative databases was used to collect information on all ED visits in patients >18 years of age. The local clinical chemistry databases were used to retrieve laboratory data. hs-cTn T (Hs-cTnT) (Roche Diagnostics) concentrations were analysed at all hospitals [16].
The dataset was thereafter linked with registers at the Swedish National Board of Health and Welfare to obtain information regarding comorbidities, medication use and cause-specific mortality. Record linkages were performed with the National Patient Register (NPR), the Prescribed Drug Register and the Cause of Death Register, respectively [17,18]. The NPR includes data on inhospital care and outpatient specialist care. Further data on MI diagnoses were obtained from the Swedish Web system for Enhancement and Development of Evidence-based care in Heart disease Evaluated According to Recommended Therapies (SWEDEHEART) register [19]. The study protocol was approved by the Regional Ethics Review Board in Stockholm, and the study complied with the principles of the Declaration of Helsinki.

Study population and definitions
All visits in patients aged >18 years with chest pain as the chief complaint, which is registered for all ED patients by the attending nurse at presentation, were identified in the study base. Comorbidities were defined as discharge diagnoses registered prior to the index visits coded according to ICD-10 in the NPR, starting from 1997 when the ICD-10 system was first introduced. Ongoing use of medication was defined as two or more dispensed medications during the year preceding the index date. One year was used to register large prescriptions of long-term medication usage, for example antihypertensive treatment, which would not have been captured if a shorter time period was chosen. Estimated glomerular filtration rate (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation [20]. Myocardial injury was defined as any hs-cTnT concentration >14 ng/L [21]. In patients with at least two hs-cTnT measurements, a dynamic change in hs-cTnT was defined as an increase of >3 ng/L from a first hs-cTnT concentration of <12 ng/L [11] or a relative increase of ≥20% from a first hs-cTnT concentration of ≥12 ng/L.

Exposure
The exposure was defined as any psychiatric disorder, categorised as an affective or a psychotic disorder, according to ICD-10 codes registered in the NPR at any visit prior to the index visit. Only diagnoses in the primary position were used to limit misclassification of the exposure as diagnoses registered in other positions were considered less accurate. Only diagnoses registered during inhospital care or in outpatient specialist care settings were used. Affective disorders were defined as any diagnosis of bipolar disorder, anxiety disorder or mood disorder (ICD-codes F30-F48) and psychotic disorders as any diagnosis of schizophrenia or a schizoaffective disorder (ICD-codes F20-F29). Patients with both affective and psychotic disorder diagnoses were allocated to the latter group. Patients without any psychiatric disease were used as the reference group.

Outcomes
The primary outcomes included patient management measures during the index visit. All visits, including repeated visits, were used in these analyses. New medical treatment usage was defined as at least 1 dispensed prescription within 180 days from the index visit which was not ongoing at the index date. A final diagnosis of MI was defined as a discharge diagnosis of I21 or I22 registered in the first position in the NPR and/or as an MI diagnosis registered in the SWEDEHEART register. We also identified all MIs diagnosed within 30 days that were not registered during the index visit. Coronary angiography and coronary interventions were defined according to ICD-codes registered in the NPR and/or registered investigations in the SWEDEHEART register.
The secondary outcomes were (i) all-cause mortality, which was ascertained using the Cause of Death Register, and for which follow-up started at the index date, and (ii) a first major adverse cardiovascular event (MACE) in patients who survived the index visit, for which follow-up started at the time of discharge in hospitalised patients and the day after the index date for all other patients and was defined as any of the following events: acute MI (ICD-codes I21 or I22 in the NPR), heart failure hospitalisation (ICD-code I50), stroke (ICD-codes I60-I64) or cardiovascular death (cause of death in the I-chapter, or R960-R961, in ICD-10). Only first visits in patients with at least 1 hs-cTnT measured at the time of the visit were analysed for the secondary outcomes. End of follow-up for cause-specific mortality and MACE was 31 December 2020 and 31 December 2019, respectively. Detailed definitions according to ICD-10 codes are provided in Table  S1.

Statistical methods
Logistic regression was used to estimate odds ratios (ORs) (hazard ratios [HRs]) with 95% confidence intervals (95% CIs) for primary outcomes, comparing patients with to patients without psychiatric disease. The diagnostic accuracy of the first hs-cTnT concentration for an MI was evaluated in receiver operational curve (ROC)-analyses, with the use of the following diagnostic thresholds: (i) the 99th percentile value (≥15 ng/L), (ii) the cut-off value recommended by the current ESC guidelines for identifying patients at a high risk of MI (≥52 ng/L) and (iii) an ROC-calibrated cutoff value that yielded a specificity of ≥90%. Cox proportional hazards models were used to estimate HRs with 95% CIs for secondary outcomes. Patients without any psychiatric disorders were used as the referent. All statistical models were performed in patients with or without myocardial injury and conducted unadjusted and adjusted for age (expressed as a restricted cubic spline with three evenly placed knots), sex (as a categorical variable), eGFR (as a categorical variable) and the following comorbidities: prior MI, heart failure, stroke, chronic obstructive pulmonary disease, atrial fibrillation, diabetes, and hypertension, and prior treatment with cardiovascular medications, all expressed as binary variables.

Study population
A total of 216,653 ED visits were included, of which 35,722 (16%) and 4332 (2.0%) occurred in patients with a history of affective and psychotic disorders, respectively (Table 1). Patients with psychiatric disease were younger and had a higher prevalence of diabetes, heart failure and chronic obstructive pulmonary disease compared to patients without psychiatric comorbidity. The use of cardiovascular medications was similar in all three groups. Details on prior psychiatric diagnoses are provided in Fig.  S1.

Analyses of hs-cTnT concentrations and hospital admission
Altogether, 183,410 (85%) patients had at least 1 cTn concentration analysed, of which 182,076 (99.7%) were measurements of hs-cTnT, with corresponding numbers being 80% and 83% in patients with psychotic and affective disorders, respectively, and 85% in patients with no psychiatric disorder ( Table 2). The adjusted risks of having any cTn measured and to have a second hs-cTnT measured (i.e. among patients with a first hs-cTnT analysed) was lower in patients with psychotic disorders, with ORs of 0.77 (95% CI: 0.71-0.83) and 0.87 (95% CI 0.81-0.94), respectively ( Table 2). The adjusted risk of being admitted to the hospital was reduced in patients with any psychiatric disorder, which was most notable when restricting the analyses to patients with myocardial injury (hs-cTnT >14 ng/L) (OR 0.49, 95% CI 0.42-0.57, and OR 0.65, 95% CI 0.61-0.69, in patients with psychotic and affective disorders, respectively) ( Table S2).

Cardiovascular medications and coronary intervention
The probability of having treatment initiated with a statin, a platelet inhibitor or any other cardiovascular medication after the index date were lower in both groups of patients with psychiatric disease, with the lowest probability observed in patients with psychotic disorders, with ORs of 0.65 (95% CI 0.56-0.75) ( Table 2). Similarly, the The probability of undergoing a coronary intervention in patients who had coronary angiography performed was similar in all groups (Table 2). We also observed lower probabilities of both medical treatment and coronary interventions associated with psychiatric disorders also among patients  Abbreviations: eGFR, Estimated glomerular filtration rate; hs-cTnT, high-sensitivity cardiac troponin T; MI, myocardial infarction; OR, odds ratio. a A diagnosis of schizophrenia or schizoaffective disorder coded in the primary position according to ICD-10 at any visit prior to the index visit. b A diagnosis of bipolar disorder, anxiety disorder or mood disorder coded in the primary position according to ICD-10 at any visit prior to the index visit. c Including cTn concentrations measured with conventional cTn assays. d Multivariable adjustment was made for age, sex, eGFR, prior myocardial infarction, heart failure, prior stroke, prior chronic obstructive pulmonary disease, atrial fibrillation, diabetes, and treatment with aspirin, P2Y12-inhibitors (clopidogrel, prasugrel, dipyridamol, and ticagrelor), oral anticoagulants (Warfarin and DOAC), beta-blockers, angiotensinconverting enzyme inhibitor/angiotensin receptor blockers, and statins. Any new treatment with aspirin, statins, betablockers or angiotensin-converting enzyme inhibitor/angiotensin receptor blockers.
with myocardial injury (Table S2), as well as among patients with dynamic hs-cTnT change where, however, statistical precision was limited (Table S3).

Diagnosis of myocardial infarction
A total of 11,677 (6.4%) MIs occurred within 30 days among patients who had any hs-cTnT concentration measured, of which the vast majority (89%) were MIs diagnosed in conjunction with the index visit (Fig. 1, Table 2 (Fig. 1).
The diagnostic accuracy of first hs-cTnT concentrations for an MI within 30 days was similar in all groups (p-values >0.05 in all pairwise comparisons of the area under the curve [AUC] in ROC-analyses) ( Fig. 2A), whereas the AUC for an MI at the index visit was slightly lower in the group of patients without psychiatric disease (pvalues 0.014 and 0.034 in pairwise comparison with patients with psychotic and affective disorders, respectively) (Fig. 2B). Using a first hs-cTnT concentration cut-off value of 15 and 52 ng/L, respectively, resulted in similar specificity for an MI diagnosis within 30 days in all groups (Table  S4). The application of ROC-optimised cut-off concentrations for a specificity ≥90% of 25, 21 and 20 ng/L in patients with no psychiatric disease, psychotic disorders and affective disorders, respectively, resulted in a higher PPV for MI in patients without psychiatric disorders, but the LR+ and efficacy, that is the number and proportion of correctly classified patients, were similar in all groups (Table  S4).

All-cause mortality and major adverse cardiovascular events
Altogether, 19,187 of 126,314 (15%) patients presenting with chest pain among whom at least one hs-cTnT measurement was performed died during a median follow-up of 6.3 (IQR 4.9-7.9)

Fig. 2 Diagnostic performance of first high-sensitivity cardiac troponin T (hs-cTnT) concentrations for myocardial infarction (MI). Panel (A) MI within 30 days. Panel (B) MI at the index visit. AUC, area under the curve.
years ( Table 3). The highest all-cause mortality was found in patients with psychotic disorders, in whom risks were 2.7-and 1.5-fold increased in patients without and with myocardial injuries, respectively, compared to patients without psychiatric disease (HR 2.71, 95% CI 2.32-3.16 and HR 1.56, 95% CI 2.32-3.16).
During a median follow-up of 5.1 (IQR 3.7-6.8) years, 18,622 (15%) patients experienced first MACEs (Fig. 3). Psychiatric disorders were associated with higher adjusted MACE risks among patients without myocardial injury, but risks were similar in patients with myocardial injury (Fig. 3). Patients with psychotic disease experienced higher overall risks of cardiovascular death and stroke, whereas the risk of MI was slightly increased in those without myocardial injury (HR 1.63, 95% CI 1.18-2.26).
Among patients without an MI diagnosis, the adjusted risk of death and MACE associated with a psychiatric disorder was increased in groups of patients with or without new cardiovascular medication treatments (Table S5) (Table S5).

Discussion
Using a large cohort of patients with a chief complaint of chest pain in the ED, we investigated the management and prognostic implications in patients with or without psychiatric disorders. We show that the presence of a psychiatric disorder was associated with a lower likelihood of patients being investigated and treated with cardiovascular medications and an almost 50% reduced risk of an MI being diagnosed at the index visit, a difference that persisted even among patients with biochemical evidence of myocardial injury. In contrast, patients with psychiatric disorders had a higher 30-day risk of an MI diagnosis after a visit with no MI diagnosis, and an increased long-term mortality. Several prior studies have elucidated differences in management strategies and survival after acute ischemic heart disease in patients with or without mental illnesses [3][4][5]22]. To our knowledge, studies investigating these aspects in large cohorts of ED patients with chest pain in the era of hs-cTn assays have not been performed.
We observed a lower MI risk within 30 days in patients with versus without psychiatric disorders, Fig. 3 Long-term risk of major adverse cardiovascular events and separate secondary outcomes. CI, confidence interval; HR, hazard ratio; MACE, major adverse cardiovascular events. Abbreviations: eGFR, Estimated glomerular filtration rate; HR, hazard ratio; hs-cTnT, high-sensitivity cardiac troponin T; CI, confidence interval. a A diagnosis of schizophrenia or schizoaffective disorder coded in the primary position according to ICD-10 at any visit prior to the index visit. b A diagnosis of bipolar disorder, anxiety disorder or mood disorder coded in the primary position according to ICD-10 at any visit prior to the index visit. c Multivariable adjustment was made for age, sex, eGFR, prior myocardial infarction, heart failure, prior stroke, prior chronic obstructive pulmonary disease, atrial fibrillation, diabetes and treatment with aspirin, P2Y12-inhibitors (clopidogrel, prasugrel, dipyridamol, and ticagrelor), oral anticoagulants (Warfarin and DOAC), beta-blockers, angiotensinconverting enzyme inhibitor/angiotensin receptor blockers and statins.
despite the fact that severe psychiatric disease is associated with an excess cardiovascular morbidity and mortality [23,24]. We speculate that these findings are attributable to a combination of factors, including fewer diagnostic investigations, difficulty interpreting symptoms and symptom presentation and a lower threshold for seeking ED care in these patient groups, all resulting in a lower diagnostic yield among patients with different psychiatric disorders who present with chest discomfort. Prior studies also indicate that MI patients with severe mental illness often present with atypical symptoms, for example dyspnea [3]. Moreover, in some patients with severe psychiatric disorders, the appropriate diagnostic evaluation may have been limited by both patient and clinician reluctance and/or opposition to further diagnostic procedures, which subsequently may have reduced the number of MI diagnoses. In addition, repeated visits were more common in patients with psychiatric disease. This observation may indicate a decreased threshold of seeking care in these groups, which in turn could limit some physicians' diagnostic and therapeutic aggressiveness. Still, patients with psychiatric disorders had an overall similar or higher risk of cardiovascular death and adverse cardiovascular events compared to patients without psychiatric comorbidity, which underlines the importance of improved ways to identify individuals in these groups who are susceptible for cardiovascular risk modification.
Patients with psychiatric disorders were less frequently treated with cardioprotective medications, including statins and platelet inhibitors, with the lowest adjusted risks of treatment observed among patients with psychotic disorders. Similar patterns were seen also in the presence of myocardial injury, and after accounting for registered MI diagnoses at the visit. These findings corroborate with observations from cohorts of MI patients, in whom concurrent severe mental illnesses have been associated with a lower likelihood of receiving appropriate treatment with secondary preventive medical treatment [2,3,5,22], and with a recent meta-analysis that found generally lower prescription rates of cardiovascular medications in patients with mental illness [25]. Recent studies suggest that improved use of such medical treatments may reduce cardiac mortality in patients with severe psychiatric disorders after an MI [2]. However, a more frequent use of cardioprotective medical treatments in this setting could prevent cardiovascular adverse outcomes is unknown and may be complicated by lack of adherence for example due to poor insight into physical illness, negative attitudes towards medication and associated cognitive impairment [25][26][27]. We also found that patients with psychiatric diseases' excess risks of MACE and cardiovascular death were high even in subgroups without an MI diagnosis and without new dispensed cardiovascular medications. Although associations may be confounded by contraindication or inappropriateness of treatment, findings emphasise the need of more sophisticated cardiovascular preventive strategies, including plans for successful maintenance of initiated cardioprotective medical use, which subsequently could improve prognosis in these vulnerable populations.
Patients with psychiatric disorders had a lower risk of an MI diagnosis at the index visit but conversely an increased 30-day risk of an MI diagnosis after a visit with no MI diagnosis. This was paralleled by a similar overall diagnostic performance of baseline hs-cTnT concentrations, with comparable efficacy and specificity for ROC-optimized hs-cTnT cut-offs, in all groups. Prior studies have indicated that unrecognised MI according to ECG signs consistent with a previous MI without a recorded MI in medical records is common among patients with severe mental illness [28], and that patients with schizophrenia who suffer an acute MI more often have pathological Q waves on ECG at presentation [3]. However, whether a considerable number of these MIs actually were 'missed' remains uncertain. Regardless, improved strategies for the early recognition of an evolving MI in patients with mental illness are still needed, which subsequently may improve outcomes.

Strengths
To our knowledge, this is the first large-scale cohort study investigating details of management strategies and their prognostic implications in patients with psychiatric disorders presenting with chest pain in the ED in the era of hs-cTn assays. We used validated national healthcare registers with complete nationwide coverage on background characteristics and outcomes, with virtually no loss to follow-up [17][18][19]. The study was conducted on data from several hospital sites in Sweden, and subsequently, we believe that the results are applicable to other healthcare settings with a similar standard of care.

Limitations
We assessed new medical treatment according to dispensed medications in the Prescribed Drug register, but we did not have information on prescription rates, indications or contra-indications for medical prescriptions, or data on adherence to treatment. This aspect may limit the interpretation of both data regarding medical treatment at baseline and the statistical analyses on new medical treatment usage.
We did not have information on diagnoses of psychiatric diseases registered in primary care. However, we believe that including patients diagnosed with psychiatric disorders in a primary healthcare setting would have resulted in a case mix with a larger magnitude of heterogeneity with regard to the severity of the disease, which consequently may have limited the generalizability and interpretation of the overall study findings.
We were not able to review or adjudicate medical records, which consequently impeded us to assess the impact of factors such as patient reluctance to follow recommendations, confounding by contraindications, or physician bias in deprioritising diagnostic procedures and/or treatments. The lack of further information from medical records also limited the ability to investigate the clinical circumstances and causes of any delayed MIs.
Lastly, although we adjusted for multiple confounders, residual confounding may have occurred. For instance, we did not have information on smoking habits or other lifestyle factors, nor of cholesterol levels.

Conclusions
In a large cohort of patients presenting with chest pain in the ED, we found that patients with psychiatric disorders were less likely to undergo cardiac investigations and to retrieve new cardiovascular medications compared to patients without such disease. Patients with psychotic disorders had an almost 50% reduced risk of an MI diagnosis within 30 days but an increased 30-day risk of an MI diagnosis after a visit with no MI diagnosis.