Introduction
Anxiety disorders affect up to 20% of patients across different stages of coronary artery disease (CAD) [
1‐
4]. Generalized anxiety disorder (GAD) is the most prevalent anxiety disorder with point prevalence rates ranging from 5% [
1,
2] to 12% [
3‐
5]. Other anxiety disorders are less common in CAD patients [
2‐
4]. Although, anxiety, when compared to depression, has received significantly less attention in CAD patients, but emerging data suggest that anxiety disorders are associated with increased risk for all-cause mortality and major adverse cardiac events independently from disease severity, depression and adverse health behaviors [
1,
2,
5]. Anxiety symptoms also predict poor clinical [
6,
7] and patient–centered outcomes [
8,
9].
The American Heart Association has recently recommended routine screening for depression in CAD patients [
10]. Screening for anxiety disorders was not included in these recommendations despite the fact that anxiety disorders are often under-recognized and untreated in cardiac patients [
11‐
13]. For example, a study in 74 patients admitted for acute myocardial infarction (MI) within 72 hours after symptom onset found that healthcare providers failed to identify 69% of patients with elevated symptoms of anxiety and 50% of patients with anxiety disorders [
11]. Another study in 158 stable heart failure patients found that 58% and 60% of patients with positive screening results for depression and/or anxiety disorder(s) on a telephone interview had a documented diagnosis of depression and/or anxiety and received mental health treatment, respectively [
12]. On the other hand, up to 92% of patients with a documented diagnosis of depression or anxiety received mental health treatment, suggesting that identification of psychiatric disorders in CAD patients improves mental health treatment availability that could potentially improve prognosis and quality of life of CAD patients [
12]. Poor identification of anxiety disorders can be attributed to the fact that healthcare providers are more concerned with management of physical symptoms and rarely used standardized screening scales [
13]. Furthermore, data regarding psychometric properties of anxiety self-rating scales remain limited in CAD patients [
13].
We and others have previously shown that Depression subscale of the Hospital Anxiety and Depression Scale (HADS-D) [
14] and Beck Depression Inventory-II [
15] had acceptable psychometric properties for screening of depressive disorders and for evaluation of depressive symptom severity across different stages of CAD [
16‐
19]. Recently, Frasure-Smith and Lesperance demonstrated that Anxiety subscale of the HADS (HADS-A) had acceptable sensitivity and specificity for screening of generalized anxiety disorder (GAD) in stable CAD patients two months after acute coronary syndromes [
2]. The Spielberger State-Trait Anxiety Inventory was designed to evaluate anxiety as personality trait (Spielberger Trait Anxiety Inventory or STAI) and as emotional state (Spielberger State Anxiety Inventory or SSAI) [
20]. In CAD patients, the STAI and SSAI are widely used for evaluation of anxiety symptom severity but screening properties of these instruments remain to be investigated [
21,
22].
The aims of this study were to establish the prevalence of anxiety disorders in stable CAD patients undergoing rehabilitation, and to evaluate the internal consistency and psychometric properties of the HADS-A, STAI and SSAI for screening of anxiety disorders.
Discussion
In patients with CAD undergoing cardiac rehabilitation, the prevalence of anxiety disorders, especially GAD, was high. Anxiety self-rating scales had comparable sensitivities, but the HADS-A had better specificities for screening of GAD and any anxiety disorders when compared with the STAI and SSAI. However, positive predictive values were low of all anxiety self-rating scales analyzed in the present study. Addition of anxiety screening to depression screening had incremental value for identification of anxiety disorders.
Prevalence of anxiety disorders in our cohort corresponds to previous studies. For example, similar prevalence rate of anxiety disorders was reported by Frasure-Smith and Lesperance in stable post-MI patients [
2]. Specifically, they reported that 5% of their patients had current GAD; 2% had panic disorder and <1% had social phobia according to the Structured Clinical Interview for the DSM [
2,
41]. Another recent study found that 10% of stable CAD patients met GAD criteria in the past year according to the Diagnostic Interview Schedule for the DSM-IV [
5]. Parker with colleagues, in patients with acute coronary syndromes, found higher prevalence rates of GAD (12%) and social phobia (9%) and similar prevalence rates of agoraphobia (2%) and panic disorder (2%) according to the Composite International Diagnostic Interview [
3,
42]. In patients awaiting coronary revascularization procedure, Tully and Penninx also reported greater prevalence rates of MINI diagnoses of GAD (10%), agoraphobia (4%) and social phobia (3%) [
4]. These data suggest that the prevalence of anxiety disorders can be different across different populations of CAD patients. Hence, large epidemiological studies evaluating the prevalence of anxiety disorders in CAD patients diagnosed using well-validated and reliable instruments are needed.
The HADS-A had superior psychometric properties for screening of GAD and any anxiety disorders when compared to the STAI and SSAI. Specifically, although at optimal cut-off values all three anxiety scales had similar sensitivities for screening of GAD and any anxiety disorders, but the HADS-A had greater specificities and PPVs when compared with the STAI and SSAI. However, confidence intervals for sensitivities were wide and can be explained by small numbers of cases. Hence, further studies in larger samples of CAD patients and meta-analysis of currently published anxiety screening studies in CAD patients should be undertaken in order to provide with more reliable estimates of optimal cut-off values of anxiety screening scales.
At cut-off value of ≥ 8 the HADS-A yielded optimal psychometric properties for screening of GAD and any anxiety disorders. Similar sensitivity (91%) and lower specificity (61%) of the HADS-A at cut-off value of ≥8 for GAD was previously reported in stable CAD patients [
2]. Marginal differences in psychometric properties of the HADS-A in the latter study when compared to our results can be partially explained by different gold standards used and by different time-frame of psychiatric evaluation with respect to acute coronary syndromes. Furthermore, in our study, at cut-off value of ≥8 the HADS-A had acceptable psychometric properties for screening of any anxiety disorder and GAD only, suggesting that CAD patients scoring ≥8 on the HADS-A should undergo psychiatric evaluation for all anxiety disorders, rather than for GAD alone.
Optimal cut-off values of the STAI and SSAI were ≥45 and ≥40, respectively, for GAD and for any anxiety disorder. A recent study in 100 pregnant women reported optimal cut-off value of the STAI and SSAI of ≥ 40 with sensitivities and specificities of about 80% for both scales for screening of current MINI diagnoses of panic disorder, agoraphobia, social phobia, GAD, and posttraumatic stress disorder [
43]. Lower sensitivities and greater specificities of the STAI and SSAI in pregnant women can be attributed to different patients’ populations and to different anxiety disorders evaluated across studies. Nonetheless, to the best of our knowledge, ours was the first study evaluating psychometric properties of the SSAI and STAI for screening for current anxiety disorders in CAD patients and our results remain to be replicated in independent samples of CAD patients.
It should be noted that the HADS-A, STAI and SSAI had low PPVs, suggesting that these scales are overly inclusive and the majority of CAD patients with positive screening results will not meet the DSM-IV-TR criteria for anxiety disorders. A study by Frasure-Smith and Lesperance in post-MI patients, found similar PPV (12%) of the HADS-A for screening of GAD [
2]. Also, low PPVs of the HADS-D for screening of major depressive disorder were previously reported in CAD patients [
17,
18]. These findings suggest that somatic symptoms can interfere with subjective evaluation of mental distress using self-rating scales and that significant proportion of CAD patients suffer from mental distress that does not reach the DSM-IV-TR severity threshold criteria [
44,
45]. Low PPVs of anxiety self-rating-scales can be partially explained by low prevalence rate of anxiety disorders and by possible residual and/or subclinical symptoms of mood or anxiety disorders in the subgroup of patients without current anxiety disorders. However, anxiety prevalence rates in our cohort correspond to the existing literature and subclinical symptoms of psychological distress are highly prevalent in CAD patients. The major goal of anxiety screening is differentiation of patients with anxiety disorder(s) from those with subclinical symptoms of psychological distress.
High false-positive rates suggest that mental health treatment should not be initiated based solely on screening results and patients with positive screening results should be always referred for detailed psychiatric evaluation by mental health specialists. High false positive rates also indicate that routine use of the HADS-A, STAI and SSAI for anxiety disorder screening purposes can overstretch healthcare resources. On the other hand, high specificities and high NPVs of the HADS-A, STAI and SSAI suggest that anxiety disorders are highly unlikely among CAD patients scoring below the recommended threshold values.
Our results also provided with evidence that anxiety screening can have incremental value to depression screening. Specifically, we found that depression screening can miss a substantial proportion of patients suffering from anxiety disorder, but these patients can be identified with anxiety screening. A two-step diagnostic algorithm can be particularly useful in the light of high false positive rates, is currently recommended for depression screening [
10] and should be investigated for anxiety screening purposes in CAD patients.
There remains a debate in the literature whether there is enough evidence to justify systematic screening for mental disorders in CAD patients. Specifically, although systematic screening for depression has been recommended in CAD patients [
10], but these guidelines were challenged mainly because there is not a single randomized controlled trial demonstrating benefits of such screening for clinical and mental health outcomes [
23,
46‐
49]. In addition, costs and safety of depression screening warrants additional research [
23,
47]. Similar limitations apply to anxiety disorder screening, since there are no studies demonstrating clinical benefits, safety and cost-effectiveness of such practice. Thus, prior to recommending routine screening of anxiety disorders in CAD patients there is a need for well conducted randomized trials clearly demonstrating that systematic anxiety screening is safe, cost-effective and carries clear clinical benefits, such as reduction of anxiety symptoms and improved clinical outcomes.
Exclusion of patients with advanced age, severe co-morbidities or unstable cardiovascular status can limit generalizability of our results. Also, our results cannot be applied to patients with acute coronary syndromes because we studied stable CAD patients admitted for cardiac rehabilitation. In addition, future studies should consider detailed evaluation for psychiatric histories since residual symptoms of mood and anxiety disorders in patients without current anxiety disorder can potentially impact the screening results. Finally, test-retest reliability of anxiety self-rating scales was not investigated. However, rehabilitation offers excellent opportunity for identification of mental distress when there are no life-threats that could potentially intervene with psychological assessments. The major strengths of our study include large sample size and the use of structured clinical psychiatric interview.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AB undertook statistical analyses, interpretation of findings and wrote the first draft of the manuscript. MS, JB and JN contributed to preparation of study protocol and were involved in the management of study protocol, data collection and management of the database. VJMP reviewed and commented on drafts, provided advice regarding interpretation of the data. RB designed the study, wrote the protocol, contributed to interpretation of findings and reviewed all drafts. All authors read and approved the final manuscript.