Publications from establishment of the database to October 2021 were systematically selected. A literature search was conducted in the following digital databases: PubMed, the Cochrane Library, EMBASE, Web of Science, MEDLINE, China National Knowledge Infrastructure (CNKI), and China Biology Medicine DISC (CBMdisc). The search strategy followed the PICO model: Population: informal cancer caregivers; Intervention: cognitive behavioral therapy; Comparison: routine nursing/health education, or blank control; Outcome: QOL, depression, or anxiety.

Keywords related to oncology (cancer OR neoplasm OR oncology OR palliative care OR palliative medicine OR malignancy) were combined with keywords related to the population (caregiver OR carer OR caregiving OR spouse OR relative OR partner OR family) and the intervention (CBT OR iCBT OR cognitive therapy OR behavioral intervention OR cognitive intervention OR coping skills OR psychosocial OR problem-solving OR cognitive restructuring OR exposure OR mindfulness OR meditation OR relaxation training OR cognitive behavior therapy OR cognitive behavioral therapy OR psychotherapy). In addition, a backward search (snowballing) of reference lists of identified studies was conducted, and earlier systematic reviews together with a forward search (citation tracking) until no additional relevant studies were found.

All studies selected for final inclusion met the following criteria: (1) interventions must include informal caregivers, alone or with patients with cancer. (2) The intervention content of the study should conform to the CBT content standard, include at least one of the following components, and be considered as CBT (15): cognitive recombination, imaginary or in-body exposure, coping skills training, problem solving, behavioral activation, structured work, reception-based cognitive intervention, and managing stress through relaxation or mindfulness. (3) Participants were randomly assigned to either the intervention group or the control group of the study. The randomized controlled trial (RCT) method. (4) Psychosocial health indicators including QOL, depression, or anxiety were included. (5) The studies were published in Chinese or English.

In addition, studies involving children with cancer were excluded as well as those involving drug interventions, because of the nature of the parent-child relationship. All previous studies were evaluated independently by two researchers, and disagreements over the inclusion/exclusion of the study were resolved by consensus. A literature search focused only on studies published in peer-reviewed journals to enhance the rigor of the methodology examined.

A pair of raters reviewed relevant studies and extracted data. In consideration of the heterogeneity of interventions and data, this study adopted the Cochrane Reviewer Handbook 5.1.0 (16) as the risk-of-bias (ROB) assessment tool to evaluate the overall quality of the study: random allocation method, allocation concealment, blinding (investigator-blinded and/or participant-blinded), integrity of result data, selective reporting of research results, and other sources of bias. All studies were scored as possessing (a) low risk of bias, (b) unclear, or (c) high risk of bias. Disagreements between researchers were resolved through discussion with a third reviewer.

Two researchers conducted data extraction by reading the full text to determine the indicators for analysis independently. Descriptive data were extracted based on the following four aspects: literature characteristics, participant characteristics, intervention plan, measurement indicators, and test tools. Literature characteristics included author and year of publication. Participant characteristics included types of cancer, number of participants, and average age. The intervention plan included intervention content and follow-up. Measurement indicators and test tools included testing tools for QOL, depression, and anxiety in patients with cancer. The outcome indicators included QOL, depression, and anxiety measurement results (see Table 1 for specific measurement scale).

To accurately extract the data, a researcher extracted the data, and a second researcher confirmed the extracted data to ensure accuracy. Using Stata 14.0 software for meta-analysis, we adopted the random-effects model because different measuring tools were used to measure the same outcome. For continuous data, the standardized mean difference (SMD) was selected as the effect scale index for statistics. The magnitude of effect indicated the degree of influence of CBT on informal cancer caregivers. The effect values were all expressed in a 95% confidence interval (CI). Heterogeneity was explored using Q and I² statistics. Q-tests were related to the probability that the results reflected systematic between-study differences. A P ≤ 0.10 was used to determine significant heterogeneity because of the generally low statistical power of heterogeneity tests. The I² statistic was an estimate of the degree of observed heterogeneity unexplained by sampling error and was unaffected by the number of studies. I² values of 0, 25, 50, and 75% were considered negligible, low, moderate, and high, respectively. Subgroup analysis, meta-regression and sensitivity analysis were conducted to explore the source of heterogeneity. Funnel plots and Egger's test were used to assess the presence of any publication biases.

First, the researcher preliminarily screened 1980 relative studies in the literature. After removing duplicates, screening titles and abstracts, and reviewing the full text, 17 studies met the criteria, with 2,348 informal cancer caregivers. All of 17 studies reported the changes in life quality, depression, or anxiety of informal cancer caregivers after the intervention. A summary of the results of the literature search and screening process is shown in Figure 1.

A summary of all study characteristics is shown in Table 1. A total of 17 RCTs meeting the inclusion criteria on CBT treatment of informal cancer caregivers published between 2000 and 2021 were included, with 2,348 cancer caregivers enrolled at baseline and the number of caregivers in each study ranging from 26 to 476. The CBT in the intervention group were based on different theories, mainly including experience support, equality theory, structured intervention, problem solving and counseling, skill learning, mindfulness, meditation, stress management, etc. The Intervention duration ranged from 4 weeks to 3 months, and specific intervention nodes could not be counted. In the control group, four studies (19, 22, 30, 32) described general psychological support and none of the others had specific interventions, among them, the control group of five studies (18, 20, 21, 27, 31) was set as the waiting or delayed intervention group. Among all studies, the overall participant attrition rate was 13.3% and the study attrition rate varied from 0 to 57.5%. The reasons for attrition included the worsening or death of patients and the business of cancer caregivers.

The average age of the caregivers was 57.25 and were primarily spouses, adult children, parents, siblings, friends or significant others of cancer patients. Except for studies that did not report cancer types, almost every study included spouses/partner. Three studies specifically focused on spouses or partners (17, 18, 22). In terms of cancer types, one studies specifically focused on patients with breast cancer (17), two studies were specifically designed for prostate cancer (22, 28), one on patients with oropharyngeal carcinoma (30), and one on patients with lung cancer (32). Furthermore, one study focused on a specific cancer stage of patients: terminal stage (29). while another study focused only on patients who received palliative care (27). The rest of the study consisted of patients with different types of cancer.

The results of the ROB assessment are shown in Figure 2. Eight studies reported the generation of randomized controlled sequences, including simple randomization of stratified module grouping, drawing from the envelope, and computer-generated randomization listing and grouping according to the time of admission (high risk of bias), while the others were not described. Only one study reported allocation concealment and blinded participants; no studies blinded the evaluation of results. The data of all studies were relatively complete, although their selective publication was unclear. One study (23) might have had bias due to the unbalanced baseline data. Another study (31) might have increased the risk of bias in the results because two researchers did not independently coded the post-intervention interview transcript.

This study applied a systematic and rigorous search strategy to retrieve relevant studies according to the research objectives. A meta-analysis was conducted on QOL, depression, or anxiety of informal cancer caregivers treated with CBT. The RCTs included were not well-controlled in terms of blinding and allocation concealment methods. Given the nature of the clinical environment and possible ethical issues in these studies, blind allocation of participants and key personnel was almost impossible. However, it was assumed that as long as the effects of the intervention were not interfered by intergroup contamination, the results were less likely to be affected by the lack of blinding.

Three outcome indicators were extracted: QOL, depression, and anxiety. Subgroup analysis, meta-regression, and sensitivity analysis were performed to eliminate the bias caused by confounders for QOL, but the results were not satisfactory. After excluding the RCTs that had a large impact on heterogeneity, the difference between the two groups was still not statistically significant. However, the result was stably reliable, that is, from the statistical sense, the improvement effect of CBT on the QOL of the informal cancer caregivers was not obvious. Regarding the effects of CBT on depression, the combined effect size showed significant differences between the two groups, but the results showed moderate heterogeneity. Subgroup analysis showed that the HADS subgroup had significant differences (P = 0.002), but with moderate heterogeneity, while the CES-D subgroup showed no significant differences in depression scores between the two groups, with no heterogeneity. Therefore, we suggested that the effect of CBT on depression in informal cancer caregivers might be influenced by different measurement tools, in addition to the specific intervention content of CBT. Carefully selecting measurement tools, such as HADS, to assess depression may be more meaningful in the future. The overall effect of CBT on anxiety was statistically different, but the difference was no longer significant after the literature was excluded by subgroup and sensitivity analyzes. These were not consistent with the results of previous studies. One study (34) involved the meta-analysis of eight RCTs. The results showed that CBT improved the QOL (P = 0.008), but the result was limited to the application of CQOL-C to assess the integration of the three studies. Depression were discussed by descriptive analysis. This study included more RCTs and extracted more outcome indicators, and CBT interventions used in RCTs were different. Therefore, this study from clinical sense could not completely deny the effect of CBT intervention measures.

By summarizing the included RCTs, we found that not all CBT interventions were effective in terms of QOL, depression, or anxiety among informal cancer caregivers (19, 30), and these interventions that were effective were not all consistently maintained during follow-up (32). Three RCTs evaluated the sleep status of the participants and showed good results (21, 32), but were not included in the meta-analysis due to the limited number. In addition, the CBT intervention aimed at couples were found effective and remained stable during 3 months (17, 18) and 4 months (22) of follow-up. RCTs for informal caregivers of patients with cancer in palliative care found that CBT interventions were promising and the earlier they were introduced, the better, especially for caregivers with severe emotional problems (27). In addition, some interventions were conducted via the Internet or mobile phones and all had positive effects on the participants, despite a lack of comparison with face-to-face interventions, which was significant in the rapidly developing era of Internet communication (22, 27, 33).

This study had some non-negligible limitations. The most important was the limited number of studies included. We only searched the published Chinese and English literature, excluding gray literature, non-indexed journals, or studies in languages other than Chinese and English, leading to reporting bias. In addition, the included RCT intervention plans, intervention time, measurement tools, outcome indicators, and so on were inconsistent. Therefore, it was difficult to compare the results between studies and the overall level of examination evidence, resulting in bias in the final results. Second, except for the CES-D subgroup of depression, the combined effect sizes of other outcome indicators were highly heterogeneous. Excluding the influence of publishing years and time span, the heterogeneity might be caused by the different research objects (cancer types, stages, and age were not the same), different CBT intervention contents (theoretical basis, implementation methods, intervention time, and so on were not the same), same ending index selection of different measurement tools, and many other factors. Third, this study did not account for the fact that different CBT intervention might bring about different effects on informal cancer caregivers. For example, the CBT interventions by face to face and mobile phones might be problematic in evaluating efficacy. Fourth, some studies did not carry out follow-up evaluation and the follow-up time and evaluation indexes were not consistent in the follow-up studies; hence, the overall long-term intervention effect could not be counted. Hence, it is difficult to provide specific recommendations for certain timings and frequency of the interventions.

This meta-analysis showed that CBT had a positive effect in terms of reducing depression and anxiety in informal cancer caregivers, did not reach statistical significance for QOL, but showed positive trends. The literature included in this study was limited, and hence its conclusions need to be further confirmed by more large-sample, multicenter randomized controlled studies. In the future, studies should be conducted according to the specific situation of the research object, including the type of cancer, cancer stage, relationship between patients and caregivers, sex of the caregiver, and so on. They should generate corresponding intervention content, have an appropriate evaluation time, and select reasonable measurement tools to enhance the QOL of informal cancer caregivers.