Panic disorder in chest pain patients referred for cardiological outpatient investigation

Ethics

The research protocol was accepted by the Regional Ethics Committee, Oslo, in November 1994.

Study population

Consecutive outpatients referred for evaluation of chest pain of unknown aetiology at four cardiology units from December 1994 to November 1996 were asked to participate in the study. Patients were included in the study according to the following inclusion criteria: (i) referred for investigation of a main complaint of chest pain; (ii) no prior documented organic heart disease; (iii) aged 18–65 years; (iv) no psychosis; (v) being able to understand and write the Norwegian language; and (vi) signed informed consent.

The fact that the referring physician had evaluated the patients' chest pain as sufficiently suspect to necessitate referral for investigation by a cardiologist was taken as adequate for study inclusion. Patients with documented organic heart disease prior to referral were excluded. Such documentation comprised: (i) a definite myocardial infarction on ECG or enzyme examination; (ii) previously positive exercise test, positive thallium scintigraphy or coronary arteriography; (iii) a history of documented mitral valve prolapse by auscultation and echocardiography; and (iv) arrhythmia documented by ECG.

Attrition. Three hundred and one patients were asked to participate in the study. Of these, 22 never met due to unknown reasons. Two patients died prior to study appointment. Thirteen patients were excluded during the study interview due to acute/known heart disease (five), cognitive dysfunction/dyslexia (four), language problems (two) and administration failure (two). Four patients who attended the psychiatric interview but did not return the self-assessment scales before the cardiological investigation were classified as non-participants. Altogether, the inclusion of 199 of the eligible 264 patients made up a participation rate of 75.4%. Reasons for not participating were obtained from 46.2% (n = 30) of the subjects and are described in Table 1.

Table 1. Reasons for not participating in the study

Participants versus non-participants. Participants and non-participants did not differ significantly in any variables such as age, sex, prevalence of coronary artery risk factors, prevalence of medical diseases/conditions, tenderness on chest wall palpation, chest pain characteristics (typical vs. atypical) or judgement of causation of symptoms due to GI disease, heart disease, musculo-skeletal disease or psychological distress. Significantly more patients were diagnosed as having coronary artery disease amongst the participants than amongst non-participants [16.1% (32/199) vs. 3.2% (2/63), P = 0.005)].

Possible selection bias due to differences in patient characteristics between study sites. Patients were included from four study centres: Ullevål Univer- sity Hospital (n = 144), Diakonhjemmet Hospital (n = 25), Aker University Hospital (n = 22), and Lovisenberg Hospital (n = 8). There were no statistically or clinically significant differences between the patients included from the principal investigation centre and the three other recruitment sites with regard to variables that may influence the outcome of the present study, such as age, sex or diagnosis of coronary artery disease.

Procedure

The head of each medical outpatient clinic screened all referrals according to the inclusion criteria. Patients considered eligible for study inclusion were mailed a letter with a proposed appointment hour with the first author together with information about the purpose of the chest pain study. They were informed that the purpose of the study was to evaluate psychological and stress-related factors amongst chest pain patients and to estimate the prevalence of psychiatric disorders in relation to the cardiological diagnosis.

Psychiatric assessments

Psychiatric diagnosis. Psychiatric disorders were assessed using the structural clinical interview for DSM-IV (SCID) [ 21]. The SCID is a semi-structured clinical interview which yields current and lifetime psychiatric state disorders (axis I disorders). It was administered by a trained interviewer and psychiatrist (TD) prior to the cardiological investigation. Thus, the interviewer and the patients were blind to the results of the cardiological investigation. The interviewer was also blind to the content of the letters from the referring physicians.

All DSM-IV axis I diagnoses were recorded. The symptom of chest pain was excluded when scoring criteria for somatoform disorders. The psychiatric diagnoses were recorded as current (diagnostic criteria met within 1 month prior to interview) or lifetime. All interviews were audiotaped. Thirty-five randomly selected tapes were rated by an experienced psychologist blind to the result of the first rater. The inter-rater reliability scores were high for all psychiatric diagnoses (κ= 0.69–1.0), including PD (κ= 0.88).

All patients suffering from panic disorder had explained to them the nature of the disorder by connecting their symptoms to the disorder criteria. By the end of the interview, the patients were asked open-ended questions about whether they had previously been informed that they might be suffering from PD and whether they wanted treatment for the disorder and a report, including treatment suggestions, to be sent to the referring physician. If they did not want treatment, they were asked to give the main reasons why.

Self-report measures. For the purpose of the present study, the following questionnaires were used: (i) demographic questionnaire for registration of sex, age, marital status, occupation, work status and income; (ii) Symptom Checklist-90-Revised (SCL-90-R) [ 22]; (iii) Eysenck personality questionnaire – neuroticism (EPQ-N) [ 23]; (iv) the 20-item Toronto Alexithymia Scale (TAS-20) [ 24]; (v) the Illness Attitude Scales (IAS) [ 25]; (vi) the Generalized Self-Efficacy Scale (GSE) [ 26]; (vii) Short Form McGill Pain Questionnaire (SF-MPQ) [ 27]; (viii) Somatosensory Amplification Scale (SSAS); [ 28] (ix) Agoraphobic Cognitions Questionnaire (ACQ) [ 29]; (x) Body Sensations Questionnaire (BSQ) [ 29]; (xi) Mobility Inventory for Agoraphobia (MIA) [ 30]; and (xii) Life Event Scale [ 31].

The questionnaires are widely used in clinical practice as well as in research. They display good to excellent psychometric properties. Questionnaires covering aspects of panic and agoraphobia are recommended for use in PD studies [ 32].

Intercorrelations between baseline variables. As high intercorrelations were found between the SCL-90-R subscales of anxiety, phobia and depression (r = 0.62–0.84), the phobia and depression subscales were excluded from further analysis. The somatization subscale was included for descriptive purposes. The relationship between subscales was also inspected by scatterplots to identify non-linear relationship.

Symptom attribution. Symptom attribution was rated on a global rating scale (from 1 =‘no degree’ to 7 =‘very high degree’) where patients were asked to what degree they considered cardiac disease, GI disease or psychological factors as being causative of chest pain.

Cardiological evaluation

A chest pain form was filled out by the cardiologist for each patient during the cardiological evaluation. The form was constructed by the investigators to obtain data of patients' previous or current medical diseases, current medication and risk factors for CAD (family history, smoking habits, diabetes, treated hypertension, hyperlipidaemia (total cholesterol > 8.0 mmol L⁻¹ and/or triglycerides > 2 mmol L⁻¹)). Chest pain was classified as typical/atypical according to the judgement of the cardiologists, taking into account the localization and the character of the pain, its appearance in relation to physical activity, psychological stress, large meals and cold weather, and its responsiveness to nitroglycerine. Palpation of the chest wall musculature was performed and rated as tenderness present or absent. Prior to the exercise test, the cardiologists rated on a global scale (from 1 =‘not at all’ to 7 =‘very high degree’) to what degree they judged coronary heart disease, gastrointestinal disease, musculo-skeletal disorder or psychological factors/stress as possible causes of chest pain.

In all patients a standard bicycle ergometer test was performed according to Nordenfeldt et al. [ 33]. The cardiologist conducting the test interpreted the result. The test was considered positive for coronary artery disease (CAD+) if ST-segment depression of ≥ 1 mm occurred in any of the ECG leads during exercise. In addition, the appearance of typical chest pain, increasing ventricular ectopic beats and lack of increase in systolic blood pressure of ≥ 30 mmHg during exercise would contribute to the diagnosis. With none of these signs present, the test was considered negative for coronary artery disease (CAD−). If inconclusive, the test was classified as such, and the patient was referred for further investigations such as thallium scintigraphy or coronary angiography. Thus, the referral to such investigations was guided by each cardiologist's clinical qualitative judgement. The cardiologists were not informed of the results of the psychiatric interviews. For the purpose of final diagnostic classification, all chest pain forms and results of the cardiological investigations were reviewed by an independent cardiologist (TH), blind to the results of the psychiatric interview. This review did not lead to any changes in the initial classification of diagnosis.

Data analysis

Comparison between PD patients and non-PD patients were performed by using a chi-squared or Fisher's exact test for dichotomous variables. Independent Student's t-test was used for normally distributed continuous variables and Mann–Whitney U-test for continuous variables without normal distribution. The latter variables were also analysed by Student's t-test. As both tests yielded basically similar results, data are presented as mean ± SD. Distributions were inspected by histograms. Bonferroni's correction was applied for repeated tests. When the statistical analyses consistently confirmed the hypotheses in the expected direction, corrections were not applied, as the results were then not considered to be at random. All tests were two-tailed. The difference was considered statistically significant when P < 0.05. Agreement between interviewers on psychiatric diagnoses was assessed using the kappa coefficient (κ). The statistical package SPSS/PC 7.5 was used for all data analyses.

PD prevalence

The prevalence of panic disorder was estimated to be 38.2% in the total chest pain population. In total, 16% of the study population suffered from CAD. The prevalence of PD was 41.3% in the 167 patients without coronary artery disease and 21.9% in the 32 patients with coronary artery disease (P = 0.038).

PD patients versus non-PD patients

Demographic data. Of the 199 included patients, 49.2% were women and 50.8% were men. Age ranged from 20 to 65 years with a mean of 50.4 years. Compared with non-PD patients, patients with panic disorder were significantly younger (mean age 38.3 ± 10.0 vs. 51.7 ± 8.8 years, P = 0.014), more likely to be female [59.2% (45/76) vs. 43.1% (53/123), P = 0.027)] and unemployed [25.0% (19/76) vs. 13.2% (17/123), P = 0.047)], had significantly fewer years of education (mean years 10.8 ± 3.0 vs. 12.2 ± 3.3, P = 0.003) and lower income (mean income (in 1000 NOK/year) 164.0 ± 87.9 vs. 252.6 ± 174.6, P < 0.001). There were no significant differences in marital status.

Psychiatric comorbidity. Specific phobia, generalized anxiety disorder, current major depression, pain disorder and hypochondriasis were significantly associated with PD. The results are shown in Table 2.

Table 2. Psychiatric comorbidity in patients with current panic disorder (PD) vs. patients without panic disorder (non-PD)

Psychiatric distress and self-assessment on pain scales. Overall, PD patients scored significantly higher than non-PD patients on all psychological and pain assessment scales with the exception of TAS-20, GES and life events ( Table 3). Patients' assessment of possible causation showed that the pain in PD patients compared with non-PD patients was rated as more likely to be caused by heart disease (mean score 3.6 ± 3.0 vs. 3.1 ± 1.2, P = 0.023). No statistically significant differences were shown between the two groups for scores on possible GI or psychological causation.

Table 3. Mean scores on psychological and pain self-assessment scales for PD and non-PD patients

Suicidal ideation. This was assessed with question 15 in SCL-90-R: ‘During the last 7 days, including today, how much were you distressed by thoughts of ending your life?’. Of the PD patients, 9% reported having suicidal thoughts, compared with 3% of the non-PD patients (P= 0.076). Furthermore, compared with non-PD patients, more PD patients responded positively to the PDQ-4 question (item 39) regarding whether they had tried to hurt or kill themselves during recent years [14.9% (11/74) vs. 5.1% (6/117), P = 0.034)– ( Table 4).

Table 4. Suicidal thoughts and self-destructive behaviour in PD and non-PD patients.

Medical assessment. Significantly more PD patients than non-PD patients were registered with a positive family history for coronary artery disease [68.4% (52/76) vs. 52.8% (65/123), P = 0.041)] and scored positively for current smoking [57.9% (44/76) vs. 34.1% (42/123), P= 0.001)]. A significant difference was also found for the mean number of risk factors (2.7 ± 1.6 vs. 2.2 ± 1.5, P = 0.021), whilst no statistically significant differences were found for single risk factors such as hypertension, obesity, diabetes, high glucose and lipid profiles. Migraine [17.1% (13/76) vs. 6.5% (8/123), P= 0.018)], fibromyalgia [13.2% (10/76) vs. 2.4% (3/123), P= 0.005)] and dyspepsia [30.3% (23/76) vs. 17.9% (22/123), P= 0.043)] were somatic diseases/conditions more commonly reported by PD patients than by non-PD patients. The mean number of registered somatic diseases/conditions was significantly higher in PD patients than in non-PD patients (1.2 ± 1.4 vs. 0.8 ± 1.2, P = 0.039). The cardiologists' assessment of possible causation showed that the pain in PD patients, as compared with non-PD patients, was rated as more likely to be caused by GI disease (mean score 2.3 ± 1.9 vs. 1.9 ± 1.0, P = 0.030) and psychological condition (3.8 ± 1.6 vs. 3.2 ± 1.5, P = 0.002). No significant differences were found for the assessments of atypical chest pain and pain on chest wall palpation.

Current PD treatment and experienced treatment need

Prior to study inclusion, only three PD patients (4%) were informed that they had panic disorder. Fifty-three per cent reported having been prescribed treatment for their symptoms during the course of the disorder. The most commonly prescribed medications were benzodiazepines (24%), nitrates (24%) and antidepressants (12%). Except for one patient, PD patients were prescribed low-potency benzodiazepines. Two of the patients being prescribed nitrates obtained a CAD diagnosis. No patients were receiving cognitive psychotherapy. Forty-five per cent of the patients reported that they experienced a need for treatment of PD symptoms and 71% of these wanted a written report to be sent to the physician responsible for follow-up. The most commonly stated reasons for not experiencing current treatment needs were clarification of cardiovascular illness experienced as primary need (29%) and satisfaction with current treatment (12%).

Prevalence

These findings confirm that patients referred for outpatient cardiological investigation because of chest pain frequently meet the criteria for panic disorder. Only 16% of the patients received a diagnosis of coronary artery disease, whilst 38.2% of the patients met the criteria for panic disorder with or without agoraphobia. As 9.2% of the PD patients also suffered from CAD, the presence of panic disorder does not exclude the presence of coronary heart disease. A recent primary care chest pain study reported extensive use of cardiological investigations, probably due to the physicians' focus on needing to exclude the possibility of cardiac disease, even when there was a low probability of its presence [ 34]. The results of the present study emphasize the importance of additional screening for PD in these patients.

Demographic characteristics

Patients with PD were significantly younger, more often female and less educated than those without the disorder. They also reported a lower income. The first finding is similar to that reported in pre-vious studies [ 12, 15–16], whilst other studies did not report differences in gender in chest pain populations, although PD is associated with being female in general population studies.

Psychiatric comorbidity

Overall, 64.5% of the PD patients also suffered from additional current psychiatric disorders. Agoraphobia, generalized anxiety disorder, current major depression, pain disorder and hypochondriasis were significantly associated with PD. A strong association between agoraphobia and panic disorder has been found in most studies conducted on psychiatric patients, whereas discrepant findings have been reported in cardiology patients. The high rate of comorbidity between panic disorder and agoraphobia (40.8%) found in this study contrasts with that reported by Beitman et al. [ 12] (6%) and Fleet et al. [ 16] (14.8%). However, Carter et al. [ 14, 15] reported that 50–68% met criteria for agoraphobia in their studies. The variability may be attributed to different sample selection.

High rates of comorbidity of panic disorder and both current and lifetime major depression have been reported [ 35]. In the present study, only 12% of the patients suffered from current comorbid major depression, whilst 25% of the PD patients had current or past major depression. These findings are similar to the reported prevalence of comorbid major depression in PD patients in other cardiology samples. Thus, it has been reported that 7–11% of the PD patients referred for chest pain suffered from current major depression [ 11, 16]. Other studies have reported prevalences of current major depression between 25 and 32%, and lifetime prevalences between 9 and 42% [ 14, 35, 36].

The present rates are somewhat lower than those reported in other PD populations. This may suggest that the high rate of major depression usually reported in panic disorder may not be representative of the true prevalence of depression in panic disorder, as depression may be an important element when patients decide to seek psychiatric treatment. In the present study, only few patients had sought such treatment and all of these suffered from comorbid depression. Pain disorder and hypochondriasis were also significantly associated with panic disorder. We found that 29% of the PD patients also suffered from pain disorder. We are not aware of any other study reporting somatoform disorders in a cardiology setting. However, one study conducted in an anxiety clinic reported that 40% of PD patients reported chronic pain [ 37]. It was further suggested that this subset of panic disorder patients was at risk for somatization, hypochondriasis and excessive health care utilization, and had more functional illness behaviours that may require specific treatment consideration. Thus, this comorbidity pattern needs replication, and further research is needed to clarify the impact of pain disorder on panic disorder morbidity and prognosis.

Some studies have suggested GAD to be associated with chest pain symptoms to a similar degree as panic disorder [ 15, 38]. This study did not confirm an independent association between GAD and chest pain, as only 6% obtained a diagnosis of GAD, and in the majority (75%) of the cases GAD was associated with comorbid panic disorder. This is in line with the results of a recent report [ 14].

Psychological distress and pain characteristics

PD patients scored significantly higher than non-PD patients on most psychological self-assessment scales and pain assessments, thus emphazising PD as a psychological distressful condition in this patient sample.

Suicidal ideation. Nine per cent of the panic disorder patients reported suicidal thoughts in the week preceding the psychiatric interview. This is less prevalent than what has been found in PD patients referred to emergency department (ED) for acute chest pain. Fleet et al. [ 16] reported that 25% of PD patients had suicidal thoughts 1 week prior to their ED visits. This questions the generalizability of the previous finding to cardiological settings and symptom acuity. We hypothesize that patients who seek emergency departments may experience more intense pain symptoms and experience more hopelessness about their condition than patients who are referred for outpatient investigation and endure standing on a waiting list for weeks. Furthermore, in the present study cardiological investigation was to take place only a few days after the psychiatric interview. This may have provided some hope for a symptom explanation and thus have minimized suicidal thoughts.

Also, different methods were used in the studies. Fleet et al. [ 16] estimated suicidal ideation using response to item 9 of the Beck Depression Inventory, whilst the response to item 15 of SCL-90-R was used in the present study. Significantly more PD than non-PD patients reported that they had tried to hurt themselves or commit suicide in recent years. The question did not discriminate between suicidal attempts and self-destructive behaviour, but the results may indicate that the PD patients in our sample are highly psychologically distressed during the course of the disorder, but the distressful periods are not necessarily just temporally connected to the cardiological investigation, as seems to be the case for PD patients in an emergency setting. Comorbid conditions such as mood disorder, substance abuse/ dependence and personality disorders are found to strongly influence whether panic disorder patients are at special risk of suicide attempts [ 39, 40]. Due to the low base rate of positive response to PDQ-item 39, regression analysis was not performed to explore the relationship between the base rates and possible predictors. Hence, suicidal ideation should be further elucidated in PD patients in cardiological settings.

Symptom attribution. PD patients were more likely to report that their chest pain was caused by heart disease than were non-PD patients. However, the cardiologists reported that the chest pain of the PD patients was more likely to be caused by a psychological condition than that of non-PD patients.

Medical comorbidity

PD patients were less likely to obtain a diagnosis of CAD than were non-PD patients. Only 9% of PD patients in the present sample suffered from comorbid CAD. PD patients were more likely to suffer from migraine, fibromyalgia and dyspepsia. Similar associations have been reported in other samples as well [ 41–43]. However, in the present study the prevalence of registered medical diseases was based on patient information reported to the cardiologists and was not confirmed by other specialists.

Current treatment and patients' experienced treatment needs

Only three of the 76 PD patients were previously informed that they were suffering from PD. The result is similar to those reported in previous studies, suggesting that physicians neither recognize nor correctly diagnose mental disorders in their patients [ 44]. As this is the basis for appropriate treatment, it may explain the under-utilization of effective PD treatment in the patients. Only 12% of the PD patients in the present study received effective pharmacological treatment such as selective serotonin re-uptake inhibitors (SSRIs), monamine oxidase inhibitors (moclobemide) and high potency benzodiazepines (clonazepam). None was prescribed cognitive/behaviour treatment, which is the best documented effective psychological treatment for PD.

All but two of the PD patients were positive to being informed about having panic disorder, and 45% of the patients who were informed communicated a need for treatment. Of those who did not, the reason most often stated was a wish to await the result of the cardiological investigation. This shows that screening for panic disorder at this stage in the investigation process was acceptable and gave treatment possibilities to a significant portion of the patients. It is possible that patients participating in the study are more positive towards psychiatric screening than non-participants. This result is supported by the finding that primary care patients want their physicians to perform psychiatric diagnostic evaluation [ 45]. This does not actually mean that they intend to start treatment, but rather that they are experiencing psychological distress which they want reduced or explained.

Study limitations

Some patients in the PD+/CAD– group may have had undetected CAD, as only one patient underwent coronary angiography and 21.6% of the patients underwent stress scintigraphy. The bicycle test may have been falsely negative and patients who were only investigated by this test could have suffered from CAD. However, the advantage of the study was that it was carried out in routine clinical practice and the investigations applied approximate clinical practice. For research purposes, all chest pain forms with test results and classification of patients into CAD+ or CAD– were controlled by an independent cardiologist. The disadvantage is the lack of systematic investigation such as angiography for all patients. Such a procedure is, however, not applied in routine clinical settings and could not always give the final answer as to whether the patients' chest pain was caused by CAD. Other possible causes of chest pain, such as spasm angina, oesophageal disease, chest wall syndrome and lung disease, were not systematically ruled out. Furthermore 22.6% of the patients reported symptoms of dyspepsia. This proportion is in the same range as that in other studies. A substantial overlap between psychiatric disease, oesophageal reflux and coronary disease has also been reported [ 46].

According to the DSM-IV criteria, a prerequisite for a diagnosis of PD is the exclusion of a medical condition such as CAD underlying the symptoms. As coronary risk factors were reported somewhat more often in PD than in non-PD patients, and not all PD patients underwent scintigraphy or angiography, one cannot exclude that some PD patients may have undetected CAD or spasm angina as an explanation of their symptoms.

As chest pain is both a symptom of coronary heart disease and a symptom of panic disorder, we excluded the symptom when assessing the diagnostic criteria for panic disorder. The prevalence remained similar after excluding this symptom. Patients with both PD and CAD were thoroughly investigated according to symptom characteristics and duration, and it was considered that these patients really had PD regardless of their medical condition.

As we studied a subset of patients, i.e. those referred to the cardiological outpatient clinics with no known cardiac disease, the generalizability of the PD prevalence to other cardiological settings and to all chest pain patients referred for cardiological investigation of chest pain must be questioned. As we expect the prevalence of CAD to be higher in consecutive chest pain patients, we would expect a somewhat lower PD prevalence for the total population. On the other hand, significantly more participants than non-participants were diagnosed as having CAD. In fact, only 3% of the non-participants obtained a CAD diagnosis.

Received 23 March 1998; accepted 13 August 1998.