An observational study of the occurrence of anxiety, depression and self-reported quality of life 2 years after myocardial infarction

Most patients had “no problems” with any of the EQ-5D-3L dimensions (72-98%), but 48% (95% CI: 43%53%) self-reported at least “some problems” with pain/discomfort. In a multiple logistic regression model (EQ5D-3L) higher age (p<0.001) and female sex (p<0.001) were associated with more pain/discomfort. Female sex (p=0.047) and prior MI (p=0.038) were associated with anxiety/depression. History of heart failure was associated with worse mobility (p=0.005) and problems with usual activities (p=0.006). The median total health status of the patients (EQ-VAS) was 78 (95% CI: 75-80).

population [4,6]. In fact, up to two thirds of the patients with MI had mild depression during hospitalization [6]. Factors that increase the risk of CHD in depressed patients are enhanced platelet activation [12], endothelial dysfunction [13], in lammation [4], and non-adherence to lifestyle changes [4,6,14]. These factors are particularly important to consider in secondary prevention planning, especially the risk of nonadherence to behavior changes and poor medical compliance.
Patients with a MI experience worse quality of life (QoL) [15] within the irst year of the event compared with the normative population [16]. Several factors in luence QoL, including age, sex, hypertension, diabetes mellitus, prior MI and heart failure (HF) [17,18]; however, non-cardiovascular factors, such as anxiety [17] and depression [17,19,20], also negatively impact QoL.
Mood disturbances and QoL in patients with MI have been extensively evaluated [6,18,21,22], but few investigations have looked at the long-term effects of MI, especially in relation to both anxiety, depression and QoL. Therefore, this study aimed to examine the level and associated predictors of anxiety and depression, as well as patients' self-reported QoL 2 years after a MI.

Design
A descriptive and comparative design was adopted using two questionnaires.

Data collection and characteristics
This is a sub-study of the REBUS (The RElevance of Biomarkers for future risk of thromboembolic events in unselected post-myocardial infarction patients) trial [23,]. Brie ly, unselected patients with both non-ST elevation (NSTEMI) and ST elevation (STEMI) MI, treated strictly according to guideline recommendations (e.g., patients underwent percutaneous coronary intervention and echocardiography with evaluation of left ventricular ejection fraction (LVEF) when indicated) were included in this observational study and followed for 2 years between 2012 and 2014. Regarding medical treatment for the myocardial infarction all patients in the study were treated according to guidelines recommendations with a high prescription rate of antithrombotic treatment and statins during the whole study period. Information about medical history and current diseases (e.g., hypertension, diabetes, previous MI, heart failure) was collected at baseline.
During the 2-year follow-up, 14 patients died and 27 were excluded from the 2-year outpatient visit (the decision to exclude patients came from either the patients themselves or the principle investigator). However, the remaining patients who attended the 2-year outpatient visit agreed to complete the Hospital Anxiety and Depression Scale (HADS) (N=376) and the EuroQol 5-dimension (EQ-5D-3L) questionnaires (N=377).
Process: At the end of the 2-year follow-up, a research nurse (who was not a member of the research team) asked the patients if they were willing to complete the HADS and EQ-5D-3L questionnaires. The patients were left alone in a quiet room and instructed on how to answer the two questionnaires, which took about 10-15 minutes to complete. Immediately after completion, the research nurse returned to the room and at this time the patients had the opportunity to ask the nurse questions about the questionnaires (e.g., if the respondents had dif iculty answering or wanted clari ication).
Instruments: Swedish versions of the HADS and EQ-5D-3L questionnaires were used. Both questionnaires have been tested for validity and reliability [24,25]. The questionnaires were coded, but not anonymized.
The patients were asked to respond to each item based on their perceived state over the past week.
Anxiety and depression were calculated according to Zigmond & Snaith [26]. The scores were calculated separately for anxiety and depression and the maximum score for each subscale was 21. Missing items were imputed by the participant's subscale mean if at least half of the items were answered [26].
EQ-5D-3L: QoL was assessed by the EQ-5D-3L. The questionnaire contains ive dimensions of health-related QoL: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. For each of the ive dimensions (EQ-index), there are three scoring levels: no problems, some problems, and extreme problems. To assess the patient's total health status, the questionnaire also includes a vertical visual analogue scale (EQ-VAS) with endpoint values ranging from 100 (Best imaginable health state) to 0 (Worst imaginable health state). The patients were asked to indicate the number on the scale that best described their perceived total health state. All the answers included the patients' self-reported QoL during the past week.

Statistical analysis
The results are presented as numbers and percentages, median, mean values and standard deviations (SD) as appropriate. The proportion (with 95% con idence intervals [CIs]) of patients with a HADS score ≥8 and the proportion of patients with problems (some or extreme problems) in the EQ-5D-3L were reported. HADS, EQ-5D-3L and EQ-VAS levels were compared between groups using the Kruskal Wallis test and the Mann-Whitney U test. In addition, EQ-5D-3L and EQ-VAS were strati ied by age and sex and compared with the general population [28].
Binary logistic regression was used to identify variables associated with problems (some or extreme problems) in the different dimensions in the EQ-5D-3L 2 years after index MI. The variables in this logistic regression included age, sex, hypertension, diabetes, prior MI, prior heart failure, STEMI at index MI, percutaneous coronary intervention (PCI) treatment and left ventricular ejection fraction (LVEF) <0.45. Multivariable models, including all variables simultaneously, were estimated. Odds ratios (ORs) with 95% CIs are reported.
Spearman's rank order correlation coef icient was calculated to determine the association between the HADS, EQ-5D-3L anxiety/depression and EQ-VAS.
Missing HADS depression items were imputed for six patients and missing HADS anxiety items were imputed for eight patients.
All statistical tests were two-tailed and performed using a signi icance level of 0.05. No adjustments for multiple tests are reported. Analyses were performed in SAS version 9.4 (SAS Institute Inc. Cary, NC, USA).

Ethics:
The study was approved by the local ethical committee (Uppsala Sweden) and followed the regulations of the Helsinki declaration (DNR 2009/210).
The median value of the summarized anxiety and depression scores was 2.0 in both dimensions and the mean values were 3.5 (SD=3.4) for anxiety and 2.5 (SD=2.7) for depression.
Elderly patients had worse depression scores (p<0.003) and female patients had higher scores on anxiety (p<0.009) ( Table 2).
Feelings of tension or wound up were experienced "occasionally" in 32% of the patients and 22% did not enjoy things quite so much as before. Detailed information on each question is listed in table 3.
In multivariable analyses ( Figure 1) age was only associated with problems in the dimension of pain/discomfort (per 10 years, OR=1.53, p<0.001). Women experienced more problems with pain/discomfort (OR=0.33, p<0.001) and anxiety/depression (OR=0.58, p=0.047) than men. Prior MI was associated with more anxiety/depression (OR=1.97, p=0.038) and prior heart failure was associated with worse mobility (OR=4.14, p=0.005) and usual activities (OR=4.65, p=0.006).
The proportion of patients with some or extreme problems in the EQ-5D-3L dimensions were compared with the general population [28] (Figure 2). Problems with mobility were more common in the study population than in the general population.
In Figure 3 the study populations' EQ-VAS was compared with that of the general population [28] as a function of age and sex. Younger people had better total health status compared with the elderly, regardless of sex.
Correlations were observed between the HADS and the EQ-5D-3L dimensions anxiety and depression and EQ-VAS (Table 5).

Multivariable analysis
Number of patients with p roblems Figure 1: Proportion of patients with some or extreme problems in the EQ-5D-3L dimensions two years after myocardial infarction, and associated variables.   Figure 2: Proportion of patients with some or extreme problems in the EQ-5D-3L dimensions two years after myocardial infarction, compared to a general population*, by age and sex. *=Szende et al.

Discussion
Two years after a MI, experiences of anxiety and depression were normalized in our patient population. QoL was comparable with that of the general population [28], except for worse mobility in the study population.
Elderly patients and patients of the female sex had more problems with pain/ discomfort. Moreover, female patients experienced more anxiety than male patients.
Only a small proportion of the patients in this long-term follow-up experienced anxiety, depression, or both, which is in line with research showing that feelings of anxiety after a MI seem to decrease over time [29]. However, a Swedish study [30] showed that, whereas QoL (EQ-5D) after cardiac arrest improved over time, anxiety and depression (HADS) did not.
The probable explanation as to why patients with MI had few symptoms of anxiety and depression and a QoL similar to the general population 2 years after the event might be related to a time factor, but also that many individuals after a life-threatening event appreciate everyday things more and are grateful for being alive. However, it is important to inform patients about the fact that mood disturbances often dissipate with time.
The most frequent problem in the EQ-5D dimensions was pain/discomfort, which is similar to indings reported for the general population [28]. In Sweden, a large number of patients undergo PCI [31] and most of those patients present with no remaining symptoms. Therefore, the proportion of patients who reported problems with pain/ discomfort probably documented pain unrelated to the heart. The higher scores of pain/discomfort in the elderly and female patients might explain the increased symptoms of depression and poorer QoL in these groups.
Other studies [8,9] con irm our results regarding women's higher levels of anxiety and some studies have also reported worse depression and QoL in women [17,18,20,21]. The small differences in sex are probably a result of equally treated patients.
Prior MI was related to worse anxiety/depression in the EQ-5D-3L dimension but not in the HADS. Despite the correlations between the questionnaires, this result re lect some differences. EQ-5D-3L is a simpler way to measure anxiety/depression in comparison with the HADS. If the patients felt anxious or depressed and indicated "some" or "extreme" problems in the EQ-5D-3L, they were interpreted as anxious or depressed. In the HADS questionnaire individual item scores are summed to produce a reported score that must be above 7 to suggest the presence of a mood disorder.
The low frequency of mood disturbances and the relatively high QoL 2 years after a MI are better than expected indings. Despite the small number of patients who rated anxiety and depression, it is important to highlight this patient group. These patients who experience anxiety and depression are at higher risk of complications [4,10,32,] and mortality [6][7][8]. Furthermore, patients with depression are less likely to stop smoking [6,33] and be physically active [6,33]. Therefore, health care providers must give more time and effort in these patients.

Strengths and limitations
The study has some limitations. First, because of the rather small sample size and because the patients were recruited from only one hospital, the results should be viewed considering these shortcomings. The study only measured anxiety, depression and QoL on one occasion as opposed to several occasions over time. We do not have information of how many patients initiated medications for anxiety and depression during follow up before answering the questionnaires which were distributed at the last study visit.
To take part in the study the patients had to understand the Swedish language. Patients were excluded if they did not have the capacity to complete the questionnaires.
One strength was that the two questionnaires were easy to complete, taking only about 10 minutes in total.
No comparisons were performed for the general population and HADS because of dif iculties to ind a comparable age-related population. Finally, no multivariate models in HADS were available because of the small number of patients reporting anxiety or depression.
The EQ-5D-3L is a general questionnaire and not disease-speci ic. It measures all types of pain and not only pain related to the heart. Questionnaires related speci ically to patients with CHD are available [34][35][36], but these often focus on physical symptoms or conditions. A strength of the present study was that age and sex distribution in the study population was comparable to the national SWEDEHEART registry [31].
Other strengths are that the questionnaires have been tested for validity and reliability [24,25,37,38], are easy to administer and understand and can be completed simply and quickly. The questionnaires were not anonymous because of the eventual need to contact patients if they were suffering from severe anxiety or depression. Further strengths of the study were the small number of responsible research nurses and that they were available if parts of the questionnaires were ambiguous or dif icult to understand.

Clinical implications
Health ca re providers must give more attention to patients with mood disturbances and work continually to improve the support process, as well as to ask patients on a routine basis about their anxiety, depression and QoL.
A professional counseling team should be available and, when needed, patients should be offered psychosocial support. It is also important to inform patients that mood disturbances are common within the irst year after a MI event but that such disturbances often become less intense with time.