Summary of main results

This systematic review examined the effectiveness of psychological therapies for the treatment of depression in patients with moderate to severe chronic obstructive pulmonary disease (COPD). The findings are based on 13 randomised controlled trials (RCTs) with 1500 participants. Ten trials, with a total of 1355 participants, used cognitive behavioural therapy (CBT)‐based interventions; two trials, with 95 participants, reported using psychotherapy sessions (one reported using elements of cognitive therapy and the second one did not provide any details); and one trial used mindfulness‐based body scan (50 participants). Therefore, 12 trials fit the category of cognitive and cognitive‐behavioural therapies by Kazdin 2000.

Overall completeness and applicability of evidence

This review's objective was to evaluate the effectiveness of psychological therapies in reducing symptoms of depression in patients with COPD. Thriteen RCTs from the global scientific literature search met our inclusion criteria and were available for investigation of the research question. Aside from two trials that examined group and one‐on‐one psychotherapy, the included studies used various therapeutic techniques of CBT. The pooled analysis showed that a psychological therapy (predominantly CBT‐based) may be effective in reducing COPD‐related depression when compared to 'no intervention', 'education' interventions, or added to and compared against a 'co‐intervention' (in this case, pulmonary rehabilitation). The quality of this evidence, however, is low. Our results are relevant to patients with moderate to severe COPD.

In terms of the applicability of CBT for depression in the general population, the evidence is strong (David 2018; Hofmann 2012), despite a debate on the potential decline of CBT's anti‐depressive effects (Cristea 2017; Johnsen 2015; Waltman 2016). There is also evidence for the effectiveness of CBT for patients with comorbid depression and chronic illnesses (Tyrer 2014). Meta‐analytic literature on the efficacy of cognitive and behavioural approaches for treatment of depression in the COPD population provides results in line with our findings, indicating that its psychotherapeutic approach is promising, and has antidepressant effects (Fritzsche 2011; von Leupoldt 2012). Coventry and colleagues, in a systematic review and meta‐regression of 74 RCTs, found that interventions that used psychological therapies (alone or in a combination with antidepressant treatment) were associated with greater reduction in depressive symptoms compared to studies that included antidepressant therapy only (Coventry 2014).

Out of the 13 included studies, eight did not include depression diagnosis or a cut‐off score for depressive symptoms in their participant eligibility criteria. As a consequence, some data came from participants who did not have clinically significant depressive symptoms (or had borderline symptoms) at their study entry. Out of the eight trials, only one study performed post‐hoc analysis to provide data for the subset of those participants who had clinically significant depressive symptoms. In order to improve applicability of this review, we performed sensitivity analysis with the four trials that tested their interventions only with participants who had clinically confirmed depressive symptoms.

The use of different instruments to measure depressive symptoms may have contributed to clinical heterogeneity across the studies: Beck Depression index (BDI (Hynninen 2010; Kunik 2008; Lamers 2010)); Hospital Anxiety and Depression Scale (HADS (Howard 2014; Jiang 2012; Schüz 2015)); Geriatic Depression Scale (GDS (Kunik 2001)); Patient Health Questionnaire‐9 (PHQ‐9 (Doyle 2017)); Center for Epidemiological Studies – Depression scale (CES‐D (Lee 2015)), Center for Epidemiological Studies–Depression scale (VAMD (Rosser 1983)). Although self‐reported screening tools for depression are validated and convenient to use, investigators question their suitability for chronically ill (older) patients and stress their inability to distinguish between overlapping symptoms of depression and physical illness (Meader 2011; O'Connor 2009), or between depression and anxiety (Norton 2013).

Also, it is important to note that the differences in disease patterns between males and females hold implications for suitable treatment type (Camp 2009; Morrow 2015).Traditionally, COPD was more commonly found in men (Buist 2007), but the condition is becoming increasingly prevalent in women, and manifests itself differently in disease mechanisms, comorbidities, and survival rates (Aryal 2013; Barnes 2016). Similarly, gender differences exist when it comes to manifestation of depression, and can be observed in prevalence rates, symptom profile, and treatment response (Altemus 2014; Johnsen 2015; Van de Velde 2010). Although in total, there were more male than female participants included in this review, the applicability of our results is not limited only to males, as a total of four studies had a greater (or equal) number of female participants compared to the number of males. The trials originate from various countries and cultural backgrounds, and different healthcare settings. Therefore, generalisability of the results is not a concern and the findings can be applied to various COPD populations and settings.

Despite the results showing improvement for depressive symptoms in the intervention groups, our review provides insufficient evidence to support the claim that CBT‐based psychological therapy is an effective treatment for COPD‐related depression. Given the variability in intervention delivery, it is not possible to provide conclusive, robust evidence supporting a specific therapy approach. Considering the low number of studies in each comparison and limited reporting for each of the secondary outcomes, further research is required to improve the overall completeness and acceptability of the evidence found.

Quality of the evidence

We included thirteen studies with 1500 participants in the review, out of which, 11 studies were used in the meta‐analyses. Given the variability in the designs of included studies, relatively small sample sizes, and overall high risk of performance, selection (allocation concealment) and detection bias, we rated the quality of evidence as very low. For this reason, confidence in the results is limited. High quality interventional experiments are designed to control for sources of bias (systematic errors) by using adequate randomisation, blinding, or allocation concealment. In the current review, the blinding of the participants, or personnel (or both) was often not feasible due to the nature of psychological therapies (Foster 2012). Lack of blinding or allocation concealment may inflate the effect size and produce (erroneously) a statistically significant result (Patsopoulos 2011). In a study that replicated the meta‐analysis of 70 systematic reviews, randomised trials without adequate allocation concealment have been shown to impact the conclusions drawn by overestimating the effectiveness of the interventions. Results from 48 of 70 pooled analyses no longer showed effectiveness when only the trials with adequate allocation concealment were included (Pildal 2007).

In terms of methodological designs, although we included only randomised controlled trials, and ten studies used CBT‐based interventions as a psychological therapy, the interventions varied. Therefore, the quality of evidence was downgraded due to inconsistency related to study designs, setting, means of delivery, and lengths of follow‐up.

Variability in baseline depression scores was a significant concern in this review. In Schüz 2015, only 21% of the sample had a score of 8 or higher on the HADS at baseline; in Kunik 2008, 53% of the sample had clinically diagnosed depression; and in Jiang 2012, the baseline mean for depression measured by the HADS was 7.16 in the intervention group and 7.08 in the control group. This indicates merely borderline depressive symptoms in the recruited participants. Also, Howard 2014 reported that only 32 out of 112 intervention participants had a score above 11 on the HADS, which means that only a small subgroup of the sample had clinically significant depressive symptoms. Moreover, a small proportion of these participants were taking medication for depression and anxiety, which may have confounded the results notably.

Aside from the fact that a number of studies included participants with unknown, borderline, or no depression, there are other noteworthy limitations reported across the studies that further explain the very low quality of the evidence. The fact that 70% of the participants in Doyle 2017 attended pulmonary rehabilitation programmes at the time of the trial, and 19% used mental health services, questions the reliability of their evidence. The dropout rate in Lee 2015 was 41% and was observed in older, less mobile participants, with more severe COPD. Also, the nurses who delivered the intervention via Skype reported difficulty in controlling the participants' environments, and reported that the participants were distracted by the circumstances around them. In Lamers 2010, the dropout rate was also high, despite the intervention being delivered via home visits. Again, the attrition was higher in older participants, with higher baseline BDI scores.

For the second primary outcome of adverse events, there were no data collected by any of the included studies. Duggan 2014 and colleagues examined 82 trials for the reporting of adverse events, and concluded that adverse events from experimental interventions were insufficiently documented in clinical trials in general, and in those assessing psychological treatments in particular. In the current review, four studies reported Patient Satisfaction scores, which may serve as useful feedback in the context of adverse events (Doyle 2017; Howard 2014; Hynninen 2010; Kunik 2001). The results from Kunik 2001 suggested that the participants had been equally satisfied with a two‐hour group session of CBT and a two‐hour group session of COPD education. In Hynninen 2010, the collected feedback indicated higher satisfaction with a manual‐based, group CBT intervention than with standard care and a brief telephone contact. In Howard 2014, where self‐guided, manual‐based CBT intervention was compared to information booklets, 78% of the participants were satisfied with the intervention, and found telephone conversations helpful. Also, group sessions were described as useful for improving the social aspects of life. In Doyle 2017, the participants were more satisfied with a telephone‐delivered CBT‐based intervention than with a telephone‐delivered attention placebo (befriending).

Few studies reported data on our secondary outcomes: quality of life, dyspnoea, lung function, exercise capacity, health service use, and cost effectiveness. Our results did not indicate improvement in the quality of life when CBT‐based intervention was compared to 'no intervention' or education. However, the sensitivity analysis of two studies that used the same quality of life measure, namely the SGRQ, showed that a CBT‐based intervention was more effective than 'no intervention'. However, this result is limited by a relatively small sample size (238 participants) since for continuous outcomes the sample size is not large enough to calculate a precise effect estimate if a total number of participants is smaller than 400 (a 'rule of thumb').

Paucity of trials reporting on dyspnoea, lung function, exercise tolerance, health service use, and cost‐effectiveness is a considerable gap in knowledge. Howard 2014 included health service use as their primary outcome, and compared a manual‐based CBT intervention to education (COPD booklets). Their results showed a reduction in hospital admissions (by 42% in the intervention group and by 16% in the control group) with significantly higher savings related to treatment costs in the intervention group compared to the control group. According to one Cochrane Review, it appears that interventions based on self‐management and education prevent hospital admissions, but do not have much effect on respiratory outcomes, and consequently, quality of life outcomes (Effing 2007). Another systematic review and meta‐analysis found that depressed patients with COPD were more frequently admitted and re‐admitted to hospital (Atlantis 2013), and that the costs associated with hospitalisation in this population are substantial (Torabipour 2016).

Two studies compared a psychological therapy added to a pulmonary rehabilitation programme, with the pulmonary rehabilitation alone, and showed a positive effect for the intervention. Current COPD guidelines basing on studies, such as Tselebis 2013, suggest that multidisciplinary pulmonary rehabilitation reduces symptoms of depression in this patient population. However, it is difficult to determine the extent to which cognitive and/or behavioural components benefit patients in pulmonary rehabilitation. Further investigation is needed to determine the most effective and efficient mode of delivery.

In summary, the reliability of our final results is limited, primarily due to indirectness related to diverse outcome measures, and inconsistency related to different designs of the studies.

Potential biases in the review process

The studies included in this review suffered from different degrees of bias. Most were at a high risk of performance and detection bias, due to lack of blinding of participants, personnel or both. In terms of outcome assessment, where the primary outcome was self‐rated by participants who were not blinded to treatment allocation, outcome assessor blinding was not achieved in these studies. Also, high risk of selection bias was present, due to lack of allocation concealment.

The inclusion of non‐depressed and depressed participants in a number of studies may have also led to biased outcomes and distorted results. Most of the self‐report rating scales for detecting depressive symptoms are biased by a large number of items addressing somatic symptoms, which may be the cause for confounded results.

Potential biases related to the selection of studies, data extraction, or data analyses were unlikely to be a concern, as we followed Cochrane guidelines. We obtained the searches from the registers of two Cochrane groups: Common Mental Disorders Group and Airways Group. We also searched trial registers and reference lists of relevant published manuscripts. Decisions about inclusion and exclusion of potential studies were made independently by two review authors, according to prespecified inclusion criteria defined in the protocol. Publication bias was not detected; however, there is a possibility of negative studies being missed before the publication of protocols became mandatory.

Agreements and disagreements with other studies or reviews

Other systematic reviews on comborbid COPD and depression based their findings on similar or a higher number of studies, however, they applied different eligibility criteria, and allowed the inclusion of randomised as well as non‐randomised trials (Baraniak 2011; Tselebis 2016). Moreover, most of the available systematic reviews published data related to the effect of psychological therapies on both depression and anxiety, rather than focusing exclusively on depression to minimise confounding of the results (Baraniak 2011; Tselebis 2016). The evidence from the current review is in line with the results from another systematic review focusing on the treatment of depression in COPD (Smith 2014). Our main finding indicates that a CBT‐based intervention, when compared to 'no intervention', 'education' interventions, and combined with a pulmonary rehabilitation compared to the pulmonary rehabilitation alone, may be effective in reducing symptoms of depression, but the effect sizes are small. This finding aligns with the findings from a systematic review and meta‐analysis that also reported only a small effect of CBT on improvement of depressive symptoms among patients with COPD (Coventry 2013). Meta‐analyses attempting to examine the impact of psychological interventions for depression in people with COPD have consistently identified a lack of methodologically rigorous evidence to support treatment efficacy (Baraniak 2011; Mikkelsen 2004). A recent review, assessing the management of depression and anxiety in COPD, highlighted the need for further research on CBT in this medical population (Panagioti 2014).

It is important to note that COPD is associated with cognitive dysfunction, regardless of the severity of disease; therefore, lower intensity and more respiratory symptom‐tailored therapies may be the best choice in the presence of COPD (Baraniak 2011). Studies investigating effectiveness of pulmonary rehabilitation in combination with a psychological therapy seem to support this hypothesis. Minimal interventions with a potential to avoid transport issues are of particular importance to patients with COPD. In this review, a number of included studies used phone‐based, nurse‐led, or minimal intervention approaches, with self‐help manuals. Telephone‐administered psychotherapy aiming to reduce depression was found to be effective in the general population (Stiles‐Shields 2014), as well as in patients with chronic illness such as cancer (Watson 2013). Nurse‐led programmes that do not require fully trained psychologists to administer the therapy have been shown to improve health‐related quality of life, and reduce hospital readmission rates (Wood‐Baker 2012). Admittedly, none of the trials included in this review investigated internet‐based interventions which have been shown to be effective for depression in the general population (Cuijpers 2010).

Future research is needed to more reliably determine the potential benefits of CBT approach for patients with COPD and comorbid depression. There is no published research evaluating the effectiveness of other psychological therapies in patients with COPD (for example Interpersonal Therapy, which is an evidence‐based therapy (de Mello 2005)). A study evaluating Supportive Therapy (ST) for oxygent‐dependant patients found that the ST enabled patients to participate in support groups and other social activities, while reducing the impact of COPD, and improving quality of life (Petty 1998). In another trial, pooled results showed statistically significant beneficial effects for both psychological well‐being and dyspnoea when using Relaxation Therapy in participants with COPD (Devine 1996).