Introduction
Breast cancer is the most common of all cancers, with over 2.3 million diagnoses and 685,000 deaths worldwide in 2020 (World Health Organization, 2021). Breast cancer is an umbrella term used for tumours that are found in the breast and includes different types of cancer that originate from the breast and either remain there or spread across the body (National Institute for Health and Care Excellence, 2009, 2018). A 5-stage classification system is utilised to categorise different severities of cancer. Broadly speaking, stages 0–III are considered less invasive and non-metastatic (cancer is localised to the breast and might affect axillary lymph nodes), while stage IV is metastatic (cancer has spread to other areas in the body, for example the bones, liver or lungs). The treatment for stages 0–III focuses on removal of malignant tumours through a lumpectomy or mastectomy, which can be in conjunction with chemotherapy and/or radiotherapy. This can be followed with adjuvant treatment, which aims at reducing the risk of cancer recurrence. Adjuvant treatment usually lasts for 5–10 years (National Institute for Health and Care Excellence, 2018). Stage IV or metastatic cancer is incurable. The active treatment is similar to stages 0–III but additional treatment focuses on prolonging life (National Institute for Health and Care Excellence, 2009). Stages 0–III breast cancer sufferers experience few physical symptoms beyond localised changes such as lumps, skin irritation or discharge; stage IV sufferers might have additional symptoms such as pain, extreme fatigue, breathing problems and nausea, depending on location of metastases (National Institute for Health and Care Excellence, 2009, 2018). It is often the treatment and its side-effects that results in discomfort and distress. The treatment can result in dramatic bodily changes, possible amputation of breast, loss of hair and infertility. Furthermore, most patients experience serious side-effects such as nausea, joint pain, weight changes and tiredness (Binkley et al., Reference Binkley, Harris, Levangie, Pearl, Guglielmino, Kraus and Rowden2012). A lot of the younger patients’ adjuvant treatments include medically induced menopause, resulting in side-effects such as severe hot flushes, aches and impact on sexual functioning (Hunter et al., Reference Hunter, Coventry and Grunfeld2009).
It has been reported that with a prevalence of 15–20%, depression is likely to be more common in women with breast cancer compared with women without (Pitman et al., Reference Pitman, Suleman, Hyde and Hodgkiss2018). However, as breast cancer is a broad diagnosis, the prevalence varies: in one study, patients with stages I–III of breast cancer showed a lower prevalence with 6.5% compared with 9.6% in stage IV patients (Kissane et al., Reference Kissane, Ildn, Bloch, Vitetta, Clarke, Smith and McKenzie1998). In a later study, the prevalence of depression was shown to be 33% of patients on diagnosis of breast cancer, which reduced to 15% at one year follow-up, but which increased to 45% for patients diagnosed with recurrent breast cancer (Burgess et al., Reference Burgess, Cornelius, Love, Graham, Richards and Ramirez2005).
Depression is a combination of a number of symptoms over a minimum of a two-week period (American Psychiatric Association, 2013). Some of these symptoms mirror the side-effects of the breast cancer treatment, such as fatigue, insomnia, poor concentration and weight change; others, such as low mood, feelings of worthlessness and inappropriate guilt and suicidal ideations, do not mirror this. Depression in women with breast cancer leads to reduced quality of life, functioning and increased distress (Calderon et al., Reference Calderon, Carmona-Bayonas, Hernández, Ghanem, Castelo, Martinez de Castro and Jimenez-Fonseca2019; Purkayastha et al., Reference Purkayastha, Venkateswaran, Nayar and Unnikrishnan2017). In addition, untreated depression in this population is associated with increased access to health care services, less acceptance of the cancer treatment and increased mortality rates (Beatty and Kissane, Reference Beatty and Kissane2017). Depression accumulates the already devastating impact breast cancer (across all stages of the disease) has on women, and it is therefore important to identify effective treatment protocols for women with depression and breast cancer.
Despite evidence of the effectiveness of CBT for depression in general populations, the relatively limited evidence base in breast cancer patients is based on studies with substantial variation in what constitutes a CBT intervention. For example, one meta-analysis investigating the effectiveness of psychological interventions for depression in breast cancer patients included studies with very limited CBT components (Xiao et al., Reference Xiao, Song, Chen, Dai, Xu, Qiu and Guo2017). Within their analysis, studies were classed as CBT even if they were short term (2–4 weeks) and contained little cognitive interventions alongside other treatments such as supportive counselling. Similarly, other studies (Jassim et al., Reference Jassim, Whitford, Hickey and Carter2015; Matthews et al., Reference Matthews, Grunfeld and Turner2017) included a wide variation of interventions and approaches in their definition of CBT, such as cognitive existential therapy, memory and attention adaptation training, mindfulness stress management or psycho-education only. This lack of delineation as to what constitutes sufficient CBT leads to substantial, seemingly unrecognised heterogeneity within these reviews. This review therefore stipulates that the core intervention is CBT as per Beck’s (Beck et al., Reference Beck, Rush, Shaw and Emery1979) cognitive therapy (CT) protocol. This protocol has a strong evidence base (Beck and Dozois, Reference Beck and Dozois2011; Cuijpers et al., Reference Cuijpers, Berking, Andersson, Quigley, Kleiboer and Dobson2013; Dobson et al., Reference Dobson, Hollon, Dimidjian, Schmaling, Kohlenberg, Gallop, Rizvi, Gollan, Dunner and Jacobson2008; Hofmann et al., Reference Hofmann, Asnaani, Vonk, Sawyer and Fang2012; Hollon et al., Reference Hollon, DeRubeis, Shelton, Amsterdam, Salomon, O’Reardon, Lovett, Young, Haman, Freeman and Gallop2005; Tolin, Reference Tolin2010) for people with depression in the wider population and is recommended in the clinical guidelines for depression in the UK (National Institute for Health and Care Excellence, 2022).
Beck’s cognitive model (Beck et al., Reference Beck, Rush, Shaw and Emery1979) of depression identifies that child and young adulthood experiences lead to the development of enduring negative unconditional beliefs about ones-self, others, and the world. Conditional beliefs are formed to negotiate how to live with these unconditional beliefs, that are often rigid, unreasonable, and unmaintainable. Consistent, emotionally salient negative automatic thoughts arise due to the activation of this belief system influencing an individual’s mood and behaviour, leading to a self-perpetuating cycle of depression. Beck’s protocol (Beck et al., Reference Beck, Rush, Shaw and Emery1979) works directly and indirectly on all levels of thinking. Given that the co-morbidity of breast cancer and depression is complex, it is acknowledged that a standard protocol needs to be augmented to address this complexity. One protocol that has considered this for people with cancer is the adjuvant psychological therapy (APT; Greer et al., Reference Greer, Moorey, Baruch, Watson, Robertson, Mason, Rowden, Law and Bliss1992, Moorey and Greer Reference Moorey and Greer2012). APT is a cancer-specific CBT protocol, which enhances Beck’s cognitive model (Beck et al., Reference Beck, Rush, Shaw and Emery1979) with a number of techniques such as: relaxation techniques; a focus on emotional expression; communication skills and setting new life goals. This protocol has shown to be effective to alleviate some of the emotional distress associated with cancer and to be suitable for depression and anxiety (Moorey and Greer, Reference Moorey and Greer2012). Although it is not targeting breast cancer specifically, the APT protocol’s strength is in its flexibility, therefore allowing idiosyncratic adaptations. For example, for breast cancer patients APT might enhance Beck’s approach with techniques targeting insomnia or sexual dysfunction. APT therefore falls within the remit of this study and would be one example of a suitable augmentation to Beck’s protocol. Different approaches might enhance the protocol with a variety of interventions, for example relaxation techniques or life goal setting. In addition, cultural adaptations have been shown to be important for engagement and therapy outcome (Naeem et al., Reference Naeem, Sajid, Naz and Phiri2023; Rathod et al., Reference Rathod, Phiri and Naeem2019) and would therefore be considered important, additional adaptations to the treatment protocol.
To the authors’ knowledge there is no systematic review that investigates an evidence-based CBT protocol for depression for breast cancer patients. As outlined above, there is a need for a standardised CBT treatment for this patient group, based on the prevalence and complexity of the presentation. This systematic review, therefore, explores the effectiveness of CBT for depression in women with breast cancer.
Method
This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher et al., Reference Moher, Liberati, Tetzlaff and Altman2009).
Eligibility criteria
A PICO (population, intervention, comparison, outcome) framework (Liberati et al., Reference Liberati, Altman, Tetzlaff, Mulrow, Gøtzsche, Ioannidis, Clarke, Devereaux, Kleijnen and Moher2009) defining the inclusion and exclusion criteria (Table 1) was followed to aid with a methodological search process.
Table 1. Inclusion and exclusion criteria
Patients with all stages of breast cancer have been included as depression is common throughout all the stages (Kissane et al., Reference Kissane, Ildn, Bloch, Vitetta, Clarke, Smith and McKenzie1998). A minimum of six sessions, has been stipulated in line with NICE recommendations for depression in the UK (National Institute for Health and Care Excellence, 2022). For reasons detailed above, the primary treatment intervention is CBT for depression as per Beck (Beck et al., Reference Beck, Rush, Shaw and Emery1979) with allowance for any appropriate augmentations to the protocol to meet cancer-specific needs. The primary outcome for this review is depression symptoms/diagnosis, and only randomised controlled trials (RCTs) are included to ensure quality standards (Clancy, Reference Clancy2002).
Search strategy
The Cochrane Library, Embase, Medline, PsycInfo and PubMed online databases were searched using keywords for breast cancer, depression and CBT.
A combination of terms was employed, combined with the Boolean search modifier to facilitate relevant root/stem truncation search (see Table S1 in Supplementary material for search terms). The searches were conducted from 5 October 2020 to 20 December 2020, including the records of the mentioned databases from the time of their origin.
All records were imported into EndNote X9 (The EndNote Team, 2013) software. Screening of title and abstracts was undertaken independently by two authors (S.W. and S.M.). Once all potentially relevant studies were identified, full texts were independently appraised (S.W. and S.M.) for inclusion using the eligibility criteria. Disagreements on eligibility status of individual studies were handled by author T.C. Reference lists of included studies were hand searched to identify any further potentially eligible studies.
Data extraction
Data extraction was undertaken independently by two review authors (S.W. and S.M.), with any disagreements resolved by author T.C. Data were extracted into a pre-prepared Microsoft Excel document (Microsoft Corporation, 2018) (Tables 2 and 3) and covered the following categories: author and nationality of the study; participant demographics and characteristics; study design and duration; depression outcome; intervention; control group and study outcome. Methodological data were also extracted to advise the quality assessment.
Table 2. Study and sample characteristics
BDI, Beck Depression Inventory; BDI-II, Beck Depression Inventory-II; BLT, bright light therapy; CBT, cognitive behavioural therapy; HADS-D, Depression subscale of the Hospital Anxiety and Depression Scale; HAMA-14, Hamilton Anxiety Rating Scale; HDRS, Hamilton Depression Rating Scale; POMS, Profile of Mood States; NTC, non-therapy control; RCT, randomised controlled trial; SCM, self-care management; STG, supportive therapy group; UC, usual care; WLC, waiting list control; †Calculated from studies; ‡standard deviation as standard error not available; §main change reported only.
Table 3. Intervention and control condition characteristics
BLT, bright light therapy; CBT, cognitive behavioural therapy; CTSR, Revised Cognitive Therapy Scale; HDRS, Hamilton Depression Rating Scale; NTC, non-therapy control; SCM, self-care management; STG, supportive therapy group; UC, usual care; WLC, waiting-list control.
Data synthesis
Narrative synthesis
A narrative synthesis was employed to summarise and present the data. The primary outcome (depression) was narrated into summaries of the following: CBT in comparison with a TAU group; CBT in comparison with an active comparison group.
Meta-analysis
A random-effect meta-analysis (DerSimonian and Laird method) (Deeks et al., Reference Deeks, Higgins, Altman, Higgins and James2002) was conducted to assess the effects of CBT on depression symptoms compared with non-active/active controls (due to small number of studies) at post-intervention and 4–6 months follow-up. Standardised mean differences (SMD) depression symptoms were computed to Hedges’ g (as this corrects small number of comparisons). The SMD was interpreted using Cohen’s approach where effect sizes of .20, .50 and .80 are considered small, moderate and large, respectively (Cohen, Reference Cohen1992). Where standard deviations (SD) were not available, these were imputed from standard errors and sample size. Change scores from baseline to post-intervention per arm were used in one case, where post-intervention estimates per arm were not reported. In this case, the SMD was standardised by imputing a uniform post-intervention SD based on the mean difference between groups at post-intervention, its respective p-value and the sample sizes of both arms. Stata software (version 17) was used to conduct the meta-analysis. Due to the small number of studies (n < 10), a formal investigation of heterogeneity and publication/small-study effect bias was precluded.
Quality assessment
The PEDro risk of bias tool (Herbert et al., Reference Herbert, Moseley and Sherrington1998) was employed to assess the methodological quality and risk of bias of the included studies. Each study was independently assessed by two review authors (S.W. and S.M.) with any disagreements resolved by author T.C.
Results
Study identification
The database searches identified 1003 records, which was reduced to 605 following removal of duplicates. Following the initial title and abstract screening, 13 potentially eligible studies were identified. Full texts screened identified six eligible studies that were included in the narrative synthesis, and five in the meta-analysis (see Fig. 1).
Figure 1. PRISMA flow diagram.
Study design
The date of publication ranged from 1999 to 2019, and all were published in English. Two studies were conducted in Australia (Edelman et al., Reference Edelman, Bell and Kidman1999a; Edelman et al., Reference Edelman, Bell and Kidman1999b), two in China (Qiu et al., Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013; Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) and two in Canada (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018; Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006).
All studies undertook individual randomisation, with two stating that they used block randomisation (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018; Edelman et al., Reference Edelman, Bell and Kidman1999b).
Five studies (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018; Edelman et al., Reference Edelman, Bell and Kidman1999b; Qiu et al., Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013; Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019; Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006) compared CBT versus treatment as usual (TAU), with two studies (Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019; Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006) also having compared CBT versus an active control.
Participant characteristics
The sample sizes ranged from 45 to 392 participants. The mean ages of participants ranged from 47 to 57 years. All participants were recruited from either hospitals, cancer centres or through media, except for one study that recruited from a mental health centre (Qiu et al., Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013).Two recruited from multiple sites (Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019; Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006).
Participants’ stage of breast cancer diagnosis varied across the studies. Two studies only recruited stage IV/metastatic cancer (Edelman et al., Reference Edelman, Bell and Kidman1999b; Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006). One study recruited women at stages I and II (Edelman et al., Reference Edelman, Bell and Kidman1999a) and another study recruited women at stages 0–III/non-metastatic cancer (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018). The remaining two studies included all breast cancer diagnoses, with the majority in stages 0–III.
One study (Qiu et al., Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013) stated a diagnosis of ‘major depressive episode’ as per the DSM-IV (American Psychiatric Association, 1994) as an eligibility criterion. Two studies (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018; Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006) required a mild to moderate level of depression, measured on either the BDI/BDI-II or the HADS-D. One study’s eligibility criterion (Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) was to either score clinically on a depression or an anxiety measure; it was unreported what percentage of participants scored clinically on the depression measure. The remaining studies did not report any criterion relating to mental health status.
Outcome measures
Two studies used three validated depression measures each: HADS-D (Zigmond and Snaith, Reference Zigmond and Snaith1983), HDRS (Hamilton, Reference Hamilton1960) and either BDI (Beck et al., Reference Beck, Ward, Mendelson, Mock and Erbaugh1961) (Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006) or BDI-II (Beck et al., Reference Beck, Steer and Emery1996) (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018). Two studies (Qiu et al., Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013; Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) only used the HDRS (in these trials its alternative abbreviation ‘HAMD-17’ is used; for ease of this study it will be consistently referred to as HDRS). The remaining two studies used the depression subsection of the POMS (McNair et al., Reference McNair, Lorr and Droppleman1981).
Description of intervention and comparison conditions
Two studies conducted weekly individual CBT sessions over 8 weeks (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018; Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006). Four studies employed group CBT on a weekly basis with a duration of 8 (Edelman et al., Reference Edelman, Bell and Kidman1999b) to 12 weeks (Edelman et al., Reference Edelman, Bell and Kidman1999a).
All the studies stated that their protocol was based on Beck’s CBT for depression model (Beck et al., Reference Beck, Rush, Shaw and Emery1979). All studies were augmented with a variety of additional interventions. Four studies included sessions on assertiveness training/interpersonal communication and various relaxation techniques. Two studies added a session on re-defining life goals (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018; Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006).
Five studies had non-active control conditions that served to provide a variety of minimal support to the participants as follows: provision of community information (Edelman et al., Reference Edelman, Bell and Kidman1999b); educational booklet (Qiu et al., Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013); monitoring of mood and risk (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018); usual medical care (Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) and waiting list control (Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006). Two studies had a third trial arm alongside the non-active control condition: bright light therapy (BLT) (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018) and self-care management (SCM) (Edelman et al., Reference Edelman, Bell and Kidman1999a). One study used a supportive talking group (STG) as an active comparison. This unstructured talking group matched the experimental group in time and frequency.
Quality assessment
One study (Qiu et al., Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013) scored 9/11 on the PEDro scale (Herbert et al., Reference Herbert, Moseley and Sherrington1998), indicating an ‘excellent’ level of quality. Three studies scored ‘good’ and the last two studies scored a ‘fair’ with 4/11 (Edelman et al., Reference Edelman, Bell and Kidman1999b) and 5/11 (Edelman et al., Reference Edelman, Bell and Kidman1999a). None of the studies blinded participants, however, this would be difficult to achieve when delivering therapy in comparison with a waiting list control. Only one trial (Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) blinded the therapist to the outcome data and the study hypothesis. The two lowest scoring studies did not conceal the allocation schedule, the assessors were not blinded, their key outcome measures were collected for fewer than 85% of participants and they did not specify if their analysis was based on the intention-to-treat. Two studies did not have a similar baseline across the comparison groups (Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019; Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006).
Meta-analysis of depression outcome
The assessment of CBT compared with non-active/active controls was conducted using standardised mean difference (SMD). Post-intervention data were used from five RCTs (see Fig. 2). CBT resulted in a statistically significant improvement in depression symptoms compared with controls with an SMD of –0.93 (95% CI –1.47, –0.40, p<0.001, I 2 = 84.66).
Figure 2. Forest plot.
Meta-analysis of depression symptoms at longer term follow up (4–6 months) using data from four RCTs (Edelman et al., Reference Edelman, Bell and Kidman1999a; Edelman et al., Reference Edelman, Bell and Kidman1999b; Qiu et al., Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013; Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) found that CBT resulted in a statistically significant improvement in depression symptoms compared with controls with an SMD of –0.65 (95% CI –1.20, –0.10, p<0.05, I 2 = 83.2%).
Narrative synthesis of depression outcome
CBT versus treatment as usual
All studies with a TAU control reported a statistically significant group × time interaction for at least one of the clinical outcome measures used when CBT was compared with the control group. Savard et al. (Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006) showed a statistically significant interaction on the HDRS only, and not on BDI and HADS-D. Desautels et al. (Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018) showed that the interaction on the HDRS was non-significant, yet reported a statistically significant interaction on the BDI-II.
At 6-month follow-up, Qiu et al. (Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013) showed a statistically significant time × group interaction, while Edelman et al. (Reference Edelman, Bell and Kidman1999b) reported no difference at 6 months. For two studies (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018; Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006) the follow-up outcome and differences were not comparable due to reallocation of waiting list control participants to the intervention group at post-intervention.
CBT versus active control
Ren et al. (Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) reported a statistically significant lower HDRS score in the CBT group compared with their active condition. However, Edelman et al. (Reference Edelman, Bell and Kidman1999a) did not find a significant difference in HDRS scores between CBT and STG. Desautels et al. (Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018) found that the CBT group showed only statistically significant lower scores than their active condition on the HDRS but not on BDI-II or HADS-D.
At follow up, Ren et al. (Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) showed a statistically significant time × group difference between CBT and the active control at 3 months; however, this was not replicated in Edelman et al. (Reference Edelman, Bell and Kidman1999a) at their follow-up point of 4 months. Desautels et al. (Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018) offered no follow-up comparison due to reallocation of waiting list participants to the intervention group.
Attrition/engagement/protocol adherence
Attrition rates at post-intervention ranged from 0% (Qiu et al., Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013) to 36% (Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006). Attrition rate at post-intervention was unreported for one study (Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) but their attrition rate at 3-month follow-up was 8%.
Three studies (Edelman et al., Reference Edelman, Bell and Kidman1999b; Qiu et al., Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013; Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) did not report attendance data across the interventions. One study (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018) reported the drop-out of four participants in the active control and one in the TAU group, but no drop-outs in the CBT group. Edelman et al. (Reference Edelman, Bell and Kidman1999a) reported seven drop-outs in the CBT group versus six in the active group. Finally, Savard et al. (Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006) reported 10 drop-outs in the CBT condition versus six in TAU condition.
Half the studies (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018; Qiu et al., Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013; Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) reported adherence to protocol of which only two (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018; Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) reported an adherence outcome, both reporting satisfactory adherence.
Discussion
The aim of this study was to determine the effectiveness of a well-defined CBT protocol for depression for women with breast cancer. This is the first review that stipulates the CBT interventions in the included studies were delivered according to Beck’s depression protocol (Beck et al., Reference Beck, Rush, Shaw and Emery1979); a protocol with sound evidence base and theory (Beck and Dozois, Reference Beck and Dozois2011; Cuijpers et al., Reference Cuijpers, Berking, Andersson, Quigley, Kleiboer and Dobson2013; Dobson et al., Reference Dobson, Hollon, Dimidjian, Schmaling, Kohlenberg, Gallop, Rizvi, Gollan, Dunner and Jacobson2008; Hofmann et al., Reference Hofmann, Asnaani, Vonk, Sawyer and Fang2012; Hollon et al., Reference Hollon, DeRubeis, Shelton, Amsterdam, Salomon, O’Reardon, Lovett, Young, Haman, Freeman and Gallop2005; Tolin, Reference Tolin2010) which, in the UK, is also recommended in the clinical guideline for depression (National Institute for Health and Care Excellence, 2022).
CBT yielded a large treatment effect on depression symptoms for women with breast cancer. Importantly, the quality of most studies ranged from good to excellent, indicating high validity of the study and reliability in terms of their outcome. The two Edelman studies (Edelman et al., Reference Edelman, Bell and Kidman1999a; Edelman et al., Reference Edelman, Bell and Kidman1999b), however, were of low methodological quality, potentially increasing their risk of bias, especially allocation and detection bias. There was some variation in drop-out rates across the studies (ranging from 9 to 35% at post-intervention) which could imply attrition bias and potentially limit the generalisability of findings (Boland et al., Reference Boland, Cherry, Dickson, Angela Boland, Cherry and Dickson2017). However, the majority of studies reported drop-out rates at post-intervention of less than 25%, and generally reported equal attrition across trial conditions, suggesting that drop-out was not likely due to unacceptability of the intervention. Considering this, the relatively high drop-out rates observed in two of the studies could have been due to a variety of factors, such as change in cancer symptomology or disease progression, rather than issue with the intervention itself.
Sample size across the individual studies varied, with two of the studies included in the meta-analysis self-reporting that they likely lacked statistical power, a finding common in mental health trials (Brown et al., Reference Brown, Murphy, Kelly and Goldsmith2019). However, two of the studies included in the meta-analysis reported to be of sufficient power to detect statistical differences at the effect size that was found. Thus, despite the variation in sample size, it is felt that the pooled analysis is sufficient to make reasonable estimates of effectiveness. An important consideration for future trials, is that one study (Ren et al., Reference Ren, Qiu, Yang, Zhu, Zhu, Mao, Mao, Lin, Shen and Li2019) that reported a large sample size, perhaps achieved this by broadening their eligibility criteria. However, this likely resulted in a more heterogenic group of participants. It can be argued that increasing the sample size at the cost of broadening the inclusion range (by including all breast cancer stages as well as a mental health diagnosis of depression and/or anxiety) does not increase the reliability of the study (Porzsolt et al., Reference Porzsolt, Wiedemann, Becker and Rhoads2019). Based on this, it is recommended that future studies in this area include a depression severity eligibility criterion for study entry, as opposed to other symptomology such as anxiety.
The studies included some variation in the depression entry levels of their samples, ranging from non-clinical to mild to moderate depression. Furthermore, there was a range of depression severity within studies. Only one study (Qiu et al., Reference Qiu, Chen, Gao, Xu, Tong, Yang, Xiao and Yang2013) stated a DSM–IV diagnosis (American Psychiatric Association, 2013) of major depressive episode as an inclusion criteria. Another study (Desautels et al., Reference Desautels, Savard, Ivers, Savard and Caplette-Gingras2018) showed that only six of 62 participants would have met this criteria. Traditionally, research in CBT and depression in the wider population has found CBT as a stand-alone intervention to be most effective in mild to moderate depression (Elkin et al., Reference Elkin, Shea, Watkins, Imber, Sotsky, Collins, Glass, Pilkonis, Leber and Docherty1989). Conversely, more recent studies found good effectiveness in moderate to severe levels of depression, showing that depression severity did not affect treatment efficacy (DeRubeis et al., Reference DeRubeis, Gelfand, Tang and Simons1999; Furukawa et al., Reference Furukawa, Weitz, Tanaka, Hollon, Hofmann, Andersson, Twisk, DeRubeis, Dimidjian, Hegerl, Mergl, Jarrett, Vittengl, Watanabe and Cuijpers2017). Nevertheless, it is recognised that more severe levels of depression need more intense treatment (National Institute for Health and Care Excellence, 2022). While less severe depression often responds well to 6–8 weeks of guided self-help based on CBT, for moderate to severe levels the recommendation is 16 or more weeks of CBT. There was no accounting for this in the included studies, potentially resulting in some participants receiving a sub-therapeutic level of CBT, while others might have received a higher intensity of CBT than required, therefore making the treatment more burdensome. However, despite the variation, the majority of samples consisted of women with elevated symptoms of depression, as assessed through either validated scales or through structured clinical interview, thus overall, it is felt that our results do reflect the effectiveness of CBT on a depressed population of breast cancer patients.
Whilst the six studies were all based on Beck’s (Beck et al., Reference Beck, Rush, Shaw and Emery1979) protocol, each was augmented differently to address the complex needs of this population. For example, some studies added skills around assertiveness and communication as well as various relaxation techniques, while others focused on re-defining life goals. This seems to be broadly based on the adjuvant psychological therapy (APT) (Greer et al., Reference Greer, Moorey, Baruch, Watson, Robertson, Mason, Rowden, Law and Bliss1992; Moorey and Greer, Reference Moorey and Greer2012), a CBT-based treatment protocol for cancer patients. APT is based on Beck’s (Beck et al., Reference Beck, Rush, Shaw and Emery1979) approach and highlights the importance of working on the underlying beliefs system.
APT was initially developed almost 30 years ago, providing a protocol for cancer based on Beck’s CT model. While it does not detail a specific protocol for different types of cancer, it provides flexibility to adapt the protocol according to individual needs. It certainly provides valid possible enhancements to the Beckian protocol for women with breast cancer and depression. It is important to consider that the medical side of breast cancer is a rapidly developing area and over the last 30 years, the 5-year survival rate has increased from 70 to 85% (Office for National Statistics, 2019; Quinn et al., Reference Quinn, Cooper, Rachet, Mitry and Coleman2008). This is partly due to increased screening for breast cancer and advances in treatment, which result in more people living longer with breast cancer. This means women spend more years receiving intensive medical treatment and with it having to deal with side-effects. Women commonly suffer with early menopausal symptoms, loss of fertility, exhaustion, insomnia and joint pain. These side effects have been linked to symptoms of depression and reduction in quality of life (Hunter et al., Reference Hunter, Coventry and Grunfeld2009). A study has shown that a CBT protocol for menopausal symptoms is effective in improving mood and quality of life within women suffering with early onset menopausal symptoms due to breast cancer treatment (Hunter et al., Reference Hunter, Coventry and Grunfeld2009). Similarly, research has shown that CBTi, a protocol for insomnia, has been effective in breast cancer sufferers not only to improve sleep but also to improve depression scores (Peoples et al., Reference Peoples, Garland, Pigeon, Perlis, Wolf, Heffner, Mustian, Heckler, Peppone, Kamen, Morrow and Roscoe2019). The included studies do not fully consider the complexity of symptoms within breast cancer patients and their protocols do not reflect the importance of augmenting the Beckian protocol with flexibility, to meet the myriads of individual needs across breast cancer patients. It is important to remember that stage IV patients are likely to present with difference in symptoms including more pain and discomfort from the disease in the first place and challenges of coming to terms with the incurability of the disease at that stage. All studies were conducted outside of the UK and none of them reported any cultural adaptations. Although this is certainly a limitation, it is felt that the findings and interpretations of the studies remain valid for UK populations; however, future studies should consider cultural sensitivity when working with non-Western populations.
The two studies that focused on stage IV (metastatic) breast cancer showed a more complex outcome pattern than the others. Edelman et al. (Reference Edelman, Bell and Kidman1999b) showed a significant outcome at post-treatment but not at 3- or 6-month follow-up; however, this study also had several limitations. The other study showed a significant outcome only on one and not the other two outcome measures used. There is notably less research available that focuses on psychological interventions for metastatic breast cancer, yet this client group experiences high distress levels and low quality of life (Reed et al., Reference Reed, Simmonds, Haviland and Corner2012). Studies examining the effectiveness of CBT for depression in people with advanced cancer in general found CBT to be less effective than in people with less advanced cancer (Mustafa et al., Reference Mustafa, Carson-Stevens, Gillespie and Edwards2013; Serfaty et al., Reference Serfaty, King, Nazareth, Moorey, Aspden, Mannix, Davis, Wood and Jones2020). For example, the CanTalk study (Serfaty et al., Reference Serfaty, King, Nazareth, Moorey, Aspden, Mannix, Davis, Wood and Jones2020) provided 12 weeks of 1:1 CBT to people with advanced stage breast cancer and found CBT to be ineffective in treating their depression. Several reasons were identified as to why CBT might be less effective in this patient group. For example, stage IV cancer sufferers undergo more intensive medical treatment, they experience more pain, hospitalisation is more common and ultimately participants are more likely to die during the study (Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006; Serfaty et al., Reference Serfaty, King, Nazareth, Moorey, Aspden, Mannix, Davis, Wood and Jones2020). These factors could potentially lead to lack of engagement with CBT, which could result in increased drop-out rates and poorer outcomes. This was observed in the two stage IV studies (Edelman et al., Reference Edelman, Bell and Kidman1999b; Savard et al., Reference Savard, Simard, Giguère, Ivers, Morin, Maunsell, Gagnon, Robert and Marceau2006) included in this review; most participants dropped out from the studies due to it being too burdensome, illness or death. This raises the question whether metastatic breast cancer patients would benefit more from a different psychological intervention in comparison with non-metastatic breast cancer patients.
Study limitations
The depression levels between and within the included studies varied greatly. Some participants showed levels of severe depression while others did not have a clinical level of depression. This is problematic as it is difficult to generalise the outcome of the studies in that context. It has shown to be effective for women with depression and breast cancer, but it is unclear due to the within-study range of depression, whether CBT is more or less effective depending on the severity of depression.
The meta-analysis outcome should be considered with caution given the limited number of studies, most of which consisted of small samples. Additionally, there was substantial heterogeneity within the analysis, likely stemming from variation between studies in baseline depression severity and cancer status (metastatic/primary breast cancer), and from the inclusion of one study with an active control condition.
Recommendations for further research
Ongoing research needs to consider the drastic medical improvements in the field of breast cancer and the new demands this poses to breast cancer sufferers that now live longer with the impact of this disease. There is scope for the development of and research in a breast cancer-specific CBT protocol for depression, which further augments Beck’s (Beck et al., Reference Beck, Rush, Shaw and Emery1979) protocol, allowing for the flexibility to address specific symptoms such as menopausal symptoms, insomnia or sexual dysfunction.
Furthermore, the question was raised whether patients with metastatic breast cancer would benefit more from a different intervention or treatment approach. Studies focusing on different psychological approaches or incorporating third wave CBT approaches may show a more beneficial outcome for women with metastatic breast cancer. Third wave CBT approaches such as acceptance and commitment therapy (ACT) (Hayes et al., Reference Hayes, Strosahl and Wilson2011) or mindfulness-based cognitive therapy (Williams and Penman, Reference Williams and Penman2011) move away from Beck’s (Beck et al., Reference Beck, Rush, Shaw and Emery1979) original approach that attempts to correct the dysfunctional belief system. These approaches do not focus on the content of thoughts but on understanding and accepting cognitive processes and the transient nature of thoughts. There is also a focus on living life according to meaningful values and practising acceptance (Hayes et al., Reference Hayes, Strosahl and Wilson2011). These approaches may be more effective when dealing with the terminal nature of the illness. A feasibility study for a RCT employing ACT within advanced cancer patients was successfully completed (Serfaty et al., Reference Serfaty, Armstrong, Vickerstaff, Davis, Gola, McNamee, Omar, King, Tookman, Jones and Low2019), suggesting that this patient group will be able to meet the demands of this therapy. Similarly, mindfulness-based interventions are showing potential benefits for advanced cancer sufferers (Tan et al., Reference Tan, Chee, Ngai, Hii, Tan, Ng, Capelle, Zainuddin, Loh, Lam, Chai and Ng2022; Zimmermann et al., Reference Zimmermann, Burrell and Jordan2018).
Furthermore, despite the positive findings of the studies over a variety of cultural contexts, it is surprising that no explicit cultural adaptations were incorporated, especially given that CBT is traditionally based on western cultural values and ideas (Naeem et al., Reference Naeem, Sajid, Naz and Phiri2023). As such, it is likely that meaningfully adapting the protocols where appropriate to the needs and nuances of the cultures that are being tested would likely bolster the effectiveness of the intervention. To this end, future trials testing CBT for this population should consider this in the design of their interventions.
Clinical implications
This review provides evidence that CBT for depression (Beck et al., Reference Beck, Rush, Shaw and Emery1979), can be helpful for treating depression in women with breast cancer and should therefore be considered as an effective treatment for this patient group. The evidence was more robust for stages 0–III breast cancer patients, less so for stage IV. The protocol has to be augmented appropriately and individually, with additional interventions to meet the complex and varied needs of this population. Additionally, practitioners need to consider the impact of medication side-effects, alongside actual and perceived losses contributing to the depression diagnosis and adapt their practice accordingly. For women with stage IV breast cancer and depression, clinical judgement needs to be made and a more idiosyncratic approach should be considered, possibly including third wave approaches such as ACT or mindfulness. There is also emerging evidence of effectiveness of behavioural activation (a CBT technique) in the cancer population (Hirayama et al., Reference Hirayama, Ogawa, Yanai, Shindo, Tanaka and Suzuki2023; Hopko et al., Reference Hopko, Armento, Robertson, Ryba, Carvalho, Colman and Lejuez2011; Hopko et al., Reference Hopko, Funderburk, Shorey, McIndoo, Ryba, File and Vitulano2013).
Conclusion
There is evidence that CBT is likely to be effective in reducing depression symptoms for women with breast cancer. However, the evidence is less robust for patients with stage IV breast cancer, highlighting the need for further research in this patient group to inform clinical guidance.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1352465824000092
Data availability statement
All extracted data can be found in the cited articles.
Acknowledgements
None to declare.
Author contributions
Stephania Wieland: Formal analysis (equal), Methodology (lead), Writing – original draft (lead), Writing – review & editing (equal); Sarah Melton: Writing – original draft (supporting), Writing – review & editing (supporting); Anastasios Bastounis: Formal analysis (lead), Methodology (supporting), Writing – original draft (supporting), Writing – review & editing (supporting); Tim Carter: Formal analysis (supporting), Methodology (supporting), Supervision (lead), Writing – original draft (supporting), Writing – review & editing (supporting).
Financial support
None to declare.
Competing interests
None to declare.
Ethical standards
All authors have abided by the Ethical Principles of Psychologists and Code of Conduct as set out by the BABCP and BPS. No ethical approval was required.
Comments
No Comments have been published for this article.