Depression is a common disorder among cardiac patients, with a prevalence rate of 15%-30% [1,2]. It not only affects physical function, physical ability, and quality of life [3,4] but also has an undesired impact on medical treatment adherence, which is harmful for cardiac rehabilitation (CR) [5,6]. Capistrant et al. [7] showed that this psychological disorder is associated with a 69% risk of sudden death risk from coronary heart disease and a 55% risk of death from cardiovascular diseases (CVDs). The role of depressive symptoms is so outstanding in developing and stabilizing cardiac conditions [3,8] that 34% of patients indicate that psychological factors are the main cause of their illness [9,10]. In light of that finding, one of the methods that can help to reduce psychological complications is participation in a CR program [11,12]. CR is recommended as one of the most important measures to take after the occurrence of a cardiac event or cardiac surgery in order to reduce complications such as depression [1,13]. CR entails life learning along with chronic diseases or disabling conditions, and its aim is to restore patients to their highest possible levels of physical, mental, social, and occupational capacity [14]. CR programs aim to improve patients' psychological and social states, to limit the physical and psychological effects of CVDs, to minimize sudden death or repeated heart attack, and to control the symptoms of coronary artery disease [15]. These programs can decrease patients' psychological symptoms to 50%-70% [16], but some patients do show depressive symptoms at the end of the programs. The importance of this issue lies in the fact that apart from all of the problems that arise from depression [1], the mortality rate in depressed patients after CR programs is reported to be four times greater than that in non-depressed patients [16]. Therefore, in order to recognize the factors that increase patients' depressive symptoms and to control the future complications thereof, the present study investigated the predictors of aggravated depressive symptoms at the end of a CR program.

From the total of 615 patients (488 male and 127 female) who participated in the CR program and were entered into the analyses, 66 (10.7%) completed the program with increased depression levels; the remaining 549 (89.3%) showed decreased levels (Fig. 1). Table 1 displays the patients' pre- and post-intervention BDI scores and indicates that the CR significantly reduced depression among the total population. However, there was a significant difference between pre- and post-program depression scores among the patients who completed the course with higher scores. This finding suggests that the increased depression levels were not random but were likely affected by certain sociodemographic and clinical variables.

The sociodemographic and clinical characteristics of the study sample are shown in Table 2. At the baseline, there were differences between the patients with and without depression in terms of education, occupation, and smoking status. Those with less education and those who were self-employed were significantly more likely to complete the CR program with increased levels of depression compared with the better educated and non-self-employed (p<0.05). With regard to the clinical variables, patients who had quit smoking were less likely to complete the CR program with increased depression levels than were those who had not quit (p<0.05).

Table 3 displays the adjusted odds ratio (AOR), 95% confidence interval (CI), and p-value for each covariate that was included in the binary logistic regression model. The statistical assumptions needed for a valid binary logistic regression were met, and the model was statistically significant overall (χ²=39.812, p=0.008). Of the 24 variables that were included as covariates in the regression, 3 were found to be independently significantly associated with increased depression. Indicators of effect size showed only small explanatory power with respect to increased depression (Cox & Snell R²=0.063; Nagelkerke R²=0.127). Considering the significant differences in preintervention BDI scores between patients who left the program with increased vs. decreased depression levels (F_{(1, 613)}=38.443, p<0.0005), gender and age as well as the effect of the pre-intervention scores were controlled for. After adjustment for gender, age, and pre-intervention depression score, the table shows that the individuals who had a high school diploma (AOR=2.90; 95% CI, 1.16-7.22; p=0.022) or who had attended college (AOR=2.92; 95% CI, 1.01-8.44; p=0.047) and those who had quit smoking (AOR=4.32; 95% CI, 1.41-13.25; p=0.01) were less likely to complete the CR program with high rates of depression compared with other patients.

The present study was conducted to investigate the predictors of depressive symptom aggravation among patients at the end of a cardiac rehabilitation program. The results indicate that low education level and smoking were significantly related to increased levels of depression at the end of the program. It is predicted that participating in CR programs can possibly minimize psychological symptoms such as anxiety and depression in cardiac patients in the short term; thus, it is recommended that additional studies be conducted to explore the associations [9]. The present study found that not only had some patients' depressive symptoms not decreased at the end of the program but that they had in fact increased. In agreement with Ghasemzadeh et al. [18], one of the factors in higher depression levels is a lower education level. Based on the cumulative advantage theory in health, people with higher levels of education enjoy greater health resources (such as greater capacity to avoid acute stress factors and live healthier lifestyles). Each of these beneficial resources can have a cumulative positive impact on individual health [19]. More treatment cooperation [5,6], desired physical functioning and higher life quality [3,4], and cardiac symptom control [15] could be cited as positive effects of CR programs, particularly when patients show constant and active participation in the programs [12]. Illiterate patients have more trouble understanding the guidelines, and in many cases, they are unable to appropriately implement their treatment recommendations compared with patients with higher levels of education. Although it seems that an educated family attendant could solve many of these problems, limitations such as the shortage of training and practice areas make this an impractical and unlikely solution for patients.

Another characteristic of patients with low literacy was that they felt less perceived control over their diseases compared with patients who had completed high school or any higher levels of education. Perceived behavior control is related to depression among patients with heart problems, and it independently predicted mobility, physical fitness, and regular exercise among outpatients [20]. In other words, the less the perceived behavior control, the greater the likelihood that depression severity will increase [21]. Therefore, it could be concluded that low literacy first negatively affects patients' perceived behavior control, which is then followed by severe depressive symptoms. The results suggest that patients with higher levels of education are less affected by depressive symptoms because of their greater perceived behavior control.

Other data from this study showed that patients who had quit smoking were less likely to show increased levels of depression at the end of the program compared with those who actively continued to smoke. The results of one study [22] found a significant relationship between smoking and psychological disorders and also found that anxiety along with depression are the most common comorbid conditions of smoking. Based on cognitive inconsistency theory, individuals who have multiple thoughts that cannot all be true at the same time experience tension and disorientation [23]. From a more detailed standpoint, cardiac patients who, despite the recommendations of the CR team and with their own knowledge of the increasing dangers and effects of smoking, still smoke actively will suffer from a cognitive inconsistency between their thoughts and their actions that can ultimately lead to increased anxiety and stress at the end of a CR program. This pressure increases along with progress in the CR program, and the individual becomes helpless when he or she cannot resolve the existing conflict. Regarding the relationship between helplessness and depressive symptoms [24,25], increased depression levels at the end of a program would not be unlikely.

One of the limitations of this study was that the administrative data were gathered from only one hospital in western Iran, and another was the missing data that triggered the statistical software to exclude a large portion of the patients from the analyses. This missing data might have affected the generalizability of the results. Another limitation was the low number of female participants in the study. Overall, it is recommended that future studies investigate patients all around the country and also that more women be studied. Two more limitations were the retrospective nature of the data and the fact that we were not able to control for all possible confounding variables. Therefore, the impact of confounding variables must be controlled for in future studies. Finally, the present study introduced a small R square for its regression model that could have resulted from the variables that were included in the analyses, which could have resulted in incorrect predictions. It is suggested that other variables be analyzed in future studies.