Discussion
Previously, the association of depressive symptoms with poor cardiac prognosis, cardiac mortality and overall mortality is well established in the literature. Yet, there remained a need for thorough investigation of the patient characteristics including demographics, comorbidities and clinical characteristics associated with new-onset depressive symptoms. Thus, the current study examined the characteristics of patients with new-onset depressive symptoms in CR patients to a greater extent. The findings of our study demonstrated that baseline characteristics of patients with new-onset depressive symptoms determine the change in the depression outcome after CR. More specifically, patient characteristics such as having a higher number of comorbidities and the comorbidities of angina, diabetes, stroke, emphysema, chronic back problems, increased weight, higher anxiety scores, physical inactivity, smoking, presence of heart failure, CABG treatment and being single were significant predictors of depression outcomes following CR. However, age, gender and IMD were not able to determine depressive symptoms after CR in patients with new-onset depressive symptoms.
A finding of note was that having a higher total number of comorbidities was a significant determinant of depressive symptoms. A prior study employing RCT data in their analysis was unable to find an association between comorbidities and depressive symptoms,28 which perhaps could be explained by the younger population (mean age 59.1±19.8) compared with the current study (66.24±10.69). Indeed, a recent meta-analysis of CR trials recommends future trials involve patients who are more representative of the broader patients with CVD, including patients with comorbidities.9 In addition, patients who present with multiple comorbidities are found to be less likely to be referred to or uptake CR which is a primary challenge for healthcare providers and CR programmes.29 30 Yet, attending CR enables patients with multiple comorbidities to benefit in terms of improving their functional capacity and psychosocial conditions.30 31
Another finding was that the comorbidity of diabetes was associated with 17% reduced likelihood of improvement in depressive symptoms following CR (OR 0.830 95% CI 0.714 to 0.965). Patients with diabetes attending CR had reduced physical fitness and increased cardiac risk factors compared with non-diabetic patients.32 Individuals with diabetes who participate or complete CR were also significantly less likely than non-participants or no-completers to experience mortality.33 34 When the condition of patients with diabetes was taken into account in terms of having a greater cardiac risk profile and lower programme participation rate, recommendations have been made to CR programmes to target patients with diabetes and involve them in CR.35–37 In addition, considering the prevalence of diabetes has been steadily increasing over time,38 the medical management of diabetes may relatively be necessary and perhaps can reduce depressive symptoms.
The stroke comorbidity was influential on the depression outcome in patients with new-onset depressive symptoms after CR. It was associated with reduced odds of improvement in depressive symptoms following CR (OR 0.718, 95% CI 0.550 to 0.937). Previous studies have shown that patients with stroke comorbidity were less likely to be referred to29 and uptake CR.30 However, the study of Marzolini et al39 found that patients who had a stroke who undergo CR improve their functional capacity and cardiovascular fitness; for that reason, including patients who had a stroke in CR can be beneficial.
Emphysema is the comorbidity which had the highest impact, among the comorbidities, on the change in depressive symptoms following CR. Having emphysema at baseline was associated with 34% reduced odds of improvement in the depressive symptoms after CR. Patients with chronic obstructive pulmonary disease (COPD) and CVD experience problems of breathlessness and disability in the context of multi-morbidity, thus, cardiac rehabilitation services are recommended to take positive action by providing tailored intervention options for patients with COPD.40 In addition, angina and chronic back problems were other comorbidities that were negative determinants of depression outcomes following CR with OR 0.778, 95% CI 0.661 to 0.916 and OR 0.812, 95% CI 0.674 to 0.977, respectively. Overall, all of the comorbidities that were mentioned were negative determinants of improvement in CR patient’s depressive symptoms following CR in patients who present with new-onset depressive symptoms. This is the first study showing the influence of the variety of comorbidities on CR outcomes in patients with new-onset depressive symptoms.
Our finding of clinical relevance was that smoking, physical inactivity and weight were modifiable CVD risk factors having an unfavourable impact on depression outcomes after CR. These factors were negative determinants of improvement in depression in patients with new-onset depressive symptoms following CR (OR 0.755, 95% CI 0.599 to 0.952; OR 0.822, 95% CI 0.721 to 0.937; OR 0.992, 95% CI 0.989 to 0.996, respectively). These results were in line with previous systematic reviews conducted in the general population,41–43 and other cohort studies that involved patients with CVD.44 45 However, these studies were unable to include CR patients and specifically patients with new-onset depressive symptoms. The current study, on the other hand, included patients who undergo CR and assessed outcome of change in depressive symptoms following CR. Prior studies have tended to focus on CR utilisation and depression with little in the way of evidence regarding factors determining outcome for patients with new-onset depression.
Another finding of the current study is patients with new-onset depressive symptoms and who have higher anxiety score at baseline were less likely to improve their depression symptoms (OR 0.900, 95% CI 0.885 to 0.915). This finding demonstrates that anxiety and depression are interrelated and associated with poor outcomes which was in accordance with a previous study.46 In the USA-based study, 622 patients with HF were included and it was revealed that anxiety, measured by Brief Symptoms Inventory, was associated with depressive symptoms in both men and women. In our study, HADS anxiety score measurement was employed for the assessment instead and still confirmed this association. In addition, the current study is first to show that patients with HF were 25% less likely to improve their depressive symptoms following CR. This result, perhaps, may explain why a recent CR trial was unable to demonstrate improvement in depressive symptoms in HF population.47 Thus, patients with HF may need additional psychosocial support from CR teams which may require further investigation in future studies. Moreover, the cardiac treatment was also adjusted in the multivariate analysis in the current study and it has been found that patients who receive the treatment of CABG were 43% more likely to have improved depressive symptoms.
In the current study, considering patient demographics, being single was associated with 24% reduced likelihood of improved depressive symptoms after CR (OR 0.761, 95% CI 0.660 to 0.877). Having partner support may improve patients’ coping with their illness and thereby improve their depression symptoms. The association of being single with increased depressive symptoms has also been found in meta-analysis of 24 cross-sectional and 8 longitudinal studies conducted by Yan et al.48 This study included 52 803 individuals from the general population of people aged ≥55 years. In our study, other patient demographics of age, gender and IMD were unable to significantly predict the depression outcome. In the multivariate analysis, the adjustments were made for other cardiac risk factors and comorbidities which may have an impact on this result. Bachmann et al49 recently found an association between lower neighbourhood socioeconomic context and reduced likelihood of participation in CR. In addition, patients who live in socially deprived areas were more disadvantaged than ones who live in less deprived neighbourhoods in terms of inhabiting poor health behaviours such as smoking and being physically inactive.50 In the current study, although the IMD was found to be associated with depressive symptoms at baseline assessment, after adjusting for other covariates it has no longer reached statistical significance.
Finally, the model of logistic regression was chosen to match with other literature in this field; for comparisons to be made, future work may look to invoke a propensity-based linear model of change accounting for included and non-included variables.
Limitations
In order to examine the determinants of outcome, specifically in patients with new-onset depressive symptoms, patients with prior history of depression have been excluded from our population. However, looking at the characteristics of our sample, it was representative of all available patients during the study time scale (n=277 521), mean age was 66.24 compared with 65.06, 25% female compared with 27%, and other variables did not differ by more than 4%. The sample was nationally representative of patients with new-onset depressive symptoms in the UK, yet, it is important to state that not all CR programmes provide a full record of patients who complete CR and in fact, in NACR data, 36.6% did not have follow-up assessment which may have an influence on the representativeness of the sample.19 A strength of this study was the use of an observational approach by employing analysis of routinely collected clinical data which enabled us to generate real-world understanding. In addition, the data involved more patients with multi-comorbidities and higher female percentage than prior RCTs.9 However, as this is a registry-based study, it should be noted that observational studies do not allow to draw causal conclusions, only association. The analysis was not able to take account of the treatment with antidepressant medication or diagnosis of depression in the CR period, as this was not recorded in the NACR data set.