Efficacy and safety of non-pharmacological interventions for cancer-related insomnia: a study protocol for a systematic review and network meta-analysis

Abstract

Introduction

Cancer-related insomnia (CRI), a significant concurrent symptom of cancer, profoundly impacts patients. Non-pharmacological interventions include cognitive behavioural therapy, mindfulness-based stress reduction programmes, bright light, acupuncture, exercise and music therapies and tai chi. These approaches, unlike pharmacological treatments, exhibit minimal adverse effects, without drug–drug interactions. They are a promising treatment strategy for CRI patients. However, a comprehensive comparative study evaluating the efficacy and safety of all non-pharmacological interventions for CRI is lacking. Accordingly, we aim to conduct a relatively comprehensive systematic review and network meta-analysis.

Methods and analysis

We will conduct an extensive search across various databases, including Pubmed, Web of Science, Cochrane Library, Embase, Google Scholar, China National Knowledge Infrastructure (CNKI), China Biomedical Literature Database (CBM), Wanfang and Vip databases (VIP). The search will focus on non-pharmacological therapeutic interventions related to CRI in randomised controlled trials published from the inception of these databases until 15 May 2024. The primary outcomes of this study will encompass the Pittsburgh Sleep Quality Index (PSQI) and the Insomnia Severity Index (ISI), while the secondary outcomes will evaluate sleep parameters, fatigue levels, anxiety-depressive mood, quality of life and any potential adverse effects. Paired meta-analyses and network meta-analyses will be conducted utilising ADDIS V.1.16.8, Stata V.14.2 and V.R4.1.2. Bias risk will be independently assessed using the Cochrane Risk of Bias tool (ROB V.2.0), and the evidence quality will be evaluated according to Grading of Recommendations Assessment, Development, and Evaluation (GRADE) standards.

Ethics and dissemination

There are no ethical issues as this study did not conduct any experiments, surveys, or human trials. We will ensure that the findings are shared through pertinent channels.

PROSPERO registration number

CRD42023427752.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study will conduct a systematic review and network meta-analysis to assess the efficacy and safety of non-pharmacological interventions for treating cancer-related insomnia (CRI).

• This study will examine randomised controlled trials only.

• The study will employ the Cochrane Bias Risk Tool (ROB V.2.0) for an independent assessment of bias risk and utilise Grading of Recommendations Assessment, Development and Evaluation (GRADE) to evaluate the quality of evidence.

• This study will exclusively examine publicly available literature in both Chinese and English, which may introduce a language bias.

• Clinical heterogeneity may arise from variations in cancer types and stages among CRI patients included in this study.

Introduction

Malignant neoplasms stand as a leading global cause of mortality, profoundly affecting human health. In 2022 alone, almost 20 million individuals were diagnosed with cancer, with 9.7 million deaths.¹ It is predicted that by 2030, the number of cancer patients in the USA will exceed 22.1 million, based on population growth and ageing.² Cancer patients frequently experience a range of complications throughout their illness, stemming from the disease itself or diverse treatments. These complications encompass issues such as insomnia, depression, pain and fatigue. Notably, insomnia ranks as the second most prominent complication associated with cancer, following closely behind fatigue.³ Cancer-related insomnia (CRI), also recognised as cancer-related sleep disorder, constitutes a secondary sleep disruption impacting the daily life and the emotional and physical well-being of cancer patients postcancer onset.⁴ This disorder arises from inadequate sleep duration and quality that fails to meet the normal physiological requirements. Key manifestations include difficulty in initiating sleep, heightened nocturnal awakenings, early morning awakenings and diminished sleep quality.⁵ Studies have shown that CRI accounts for 52.6%–67.4% of cancer complications, which is two to three times higher than in the healthy population.⁶ Remarkably, up to 95% of cancer patients report ongoing sleep disturbances at the time of diagnosis, throughout treatment and even a decade into survivorship.⁷ Sleep deprivation amplifies the physical, psychological, social and existential suffering experienced by cancer patients.⁸ It diminishes the body’s functional status, intensifies pain and discomfort, induces fatigue, contributes to psychological disorders such as anxiety and depression, weakens the immune system and reduces patients’ adherence to conventional anticancer therapies.⁹ More critically, sleep disorders can impair both attention and memory, which can result in cognitive dysfunction.^{10 11} In some cases, they may even exacerbate or trigger cardiovascular and cerebrovascular diseases, adding a financial burden and significantly impacting the quality of life and long-term survival prospects of cancer patients.⁶ Consequently, enhancing the sleep quality of cancer patients has gained increasing focus, posing an urgent concern for patients, families, society and clinicians.

Patients and clinicians frequently overlook CRI, perceiving it as a common and transient reaction to either the cancer itself or cancer-related treatments. The reason for this is that on the one hand, the pathogenesis of CRI is complex. Some studies have identified advanced age, female gender, family history and neurological disorders as risk factors for CRI.^{12 13} Persistent factors contributing to CRI include prolonged bed rest, irregular sleep patterns, trepidation about disease treatment and fear of mortality.^{14 15} Moreover, postsurgery and radiotherapy-induced pain and fatigue further compromise the sleep quality of cancer patients.¹⁶ On the other hand, ambiguity remains regarding the pathogenesis of CRI. It has been suggested that primary or metastatic brain tumours may cause an imbalance between parasympathetic and sympathetic activity by interfering with neurohormonal sleep regulation and brainwave patterns during sleep phases.^{17 18} Advanced illness stages accompanied by muscular weakening and tumour growth in the upper and lower respiratory systems can cause dyspnoea, hypoxia and dyspnoea, which can disrupt sleep regulation systems and arouse patients.^{19 20} Additionally, there are correlations between the pathogenesis of CRI and cytokine dysregulation (including interleukin, interferon and tumour necrosis factor), dysregulation of the hypothalamic–pituitary–adrenal axis and the disruption of cortisol rhythms.21,23 Inflammation may be another potential mechanism for the pathogenesis of CRI, and patients with CRI often exhibit elevated proinflammatory factors IL-1β, TNF-α and IL-6.^{24 25}

Currently, clinicians primarily base the management of CRI on pharmacological interventions, supplemented by non-pharmacological interventions. Pharmacotherapy for CRI typically involves the use of benzodiazepines, non-benzodiazepines, antidepressants and antihistamines, among others.²⁶ Despite the widespread use of pharmacological interventions, studies have indicated that the efficacy of Western medications in treating insomnia often falls below 80%. Moreover, these medications can cause a range of side effects, including daytime dizziness and drowsiness, nocturnal cognitive fuzziness and long-term cognitive impairment. Prolonged use may also result in drug dependence, with rebound insomnia on discontinuation.26,29 Additionally, polypharmacy among cancer patients can bring about drug–drug interactions, creating long-term health risks. Consequently, relying solely on pharmacological interventions may not effectively alleviate insomnia symptoms in cancer patients. Conversely, non-pharmacological treatments have no serious adverse effects or drug–drug interactions and are expected to be clinical treatments to reduce insomnia symptoms in cancer survivors. Among clinical non-pharmacological therapies, cognitive behavioural therapy for insomnia (CBT-I) is widely acknowledged as the benchmark for treating persistent insomnia and is an exceptional first-line treatment. It not only treats sleeplessness, but it also addresses other cancer-related concerns such as daytime tiredness, depressive symptoms and inflammatory indicators, and is beneficial to general quality of life.30,32 However, due to the complexity and relatively high cost of its therapeutic procedures, patient compliance and acceptance are somewhat limited. Furthermore, relaxation therapy, physical activity and interventions based on positive thinking have demonstrated the potential to modulate the cancer microenvironment by influencing oxidative stress and immune-related functions, thereby enhancing sleep quality and alleviating daytime fatigue in cancer patients.^{33 34} In recent years, acupuncture, acupressure, tai chi, yoga and traditional Chinese medicine five-element music therapy have also been increasingly utilised by cancer patients and are evidenced to have varying degrees of effectiveness in improving CRI.^{35 36}

While some meta-analyses have been conducted to evaluate non-pharmacological therapies for CRI, they are typically limited to comparisons between two therapies or between different types of the same therapy, or they focus on a specific type of cancer only. For example, Chen et al designed a study protocol to compare the efficacy of different acupuncture therapies for CRI,³⁷ while Jin et al designed a protocol to assess the efficacy and safety of different non-pharmacological interventions for breast cancer survivors.³⁸ Our study aims to incorporate several common non-pharmacological treatment modalities and is not restricted by patient cancer type, so it will encompass a broader spectrum of affected cancers and enhance the generalisability of the findings. In addition, existing CRI-related studies typically employ only the Pittsburgh Sleep Quality Index (PSQI) and Insomnia Severity Index (ISI) as outcome indicators, while insomnia is not an isolated symptom. In reality, insomnia, along with fatigue, anxiety and depression, contributes to a feedback loop that increases the overall symptom burden for patients. Accordingly, our study expands the outcome measures to include the Fatigue Scale, Anxiety Depression Scale and Quality of Life Scale as significant indicators for assessing insomnia in cancer patients, aiming to comprehensively and multidimensionally evaluate patients’ symptoms.

In conclusion, this study aims to evaluate widely used non-pharmacological therapies in clinical settings by synthesising relevant clinical data employing both classic meta-analysis and network meta-analysis (NMA) methodologies. The goal is to assess the clinical efficacy and safety of various non-pharmacological therapies and subsequently rate them according to the strengths and weaknesses of their impacts. Ultimately, our goal is to provide a more comprehensive and reliable reference for selecting clinical treatments and care options.

Methods and analyses

Design and registration

This systematic review and NMA will follow the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines.³⁹ This study has been registered in the International Prospective Registry of Systematic Reviews (PROSPERO) database (no. CRD42023427752). The study is scheduled to begin on 15 May 2024, with modifications expected to be made on 11 August 2024, and expected to be completed on 10 May 2025.

Criteria for study selection

Types of studies

Only randomised controlled trials (RCTs) published in English or Chinese will be included, while cohort research, cross-sectional studies, meta-analyses, systematic reviews and case reports will not be included.

Types of participants

All participants will meet the following criteria:

1. Patients with pathologically or cytologically confirmed malignant tumours, regardless of tumour type and stage.

2. Patient classifications in insomnia based on the International Classification of Sleep Disorders-Third Edition (ICSD-3),⁴⁰ the Diagnostic and Statistical Manual of Mental Disorders-Fifth Edition (DSM-5)⁴¹ and other comparable criteria for identifying sleep disorders.

3. Patients (≥18 years) with a survival time exceeding 3 months, without restrictions on sex, nationality, race or education level.

4. PSQI score>7.

5. ISI score≥8 in the past 2 weeks.

6. Patients with full consciousness, no language or communication disorders and the ability to independently complete the questionnaires.

Types of interventions

Considering the results of the literature review, non-pharmacological interventions will include (1) CBT; (2) mindfulness-based stress reduction (MBSR); (3) exercise therapy: including aerobic exercise, walking, tai chi, qigong and yoga; (4) bright light therapy; (5) acupuncture treatment; (6) music therapy and (8) psychological therapy. The experimental group will be intervened by one or several non-pharmacological intervention methods. The control group will be treated with standard pharmacological therapy, a placebo or different exercise modalities from the intervention group.

Types of outcome measures

Primary outcome

The primary outcome will be to assess sleep outcomes in patients with CRI, incorporating subjective evaluation tools, such as PSQI and ISI, and objective sleep parameters measured through polysomnography.

Secondary outcome

The secondary outcomes will encompass measures of cognitive competence, fatigue, anxiety, depression, quality of life and adverse reactions. The assessment tools will include Montreal cognitive assessment, the Piper Revised Fatigue Scale, the Hospital Anxiety and Depression Scale and the 30-item Core Quality of Life Questionnaire (EORTC QOL-C30).

Data sources and search

This study will extensively search relevant RCTs using MeSH and Emtree principles in various databases. Search terms will include ‘Non-pharmaceutical intervention’, ‘Complementary Medicine’, ‘Cancer’, ‘Insomnia’, ‘Cognitive Behavioural Therapy’, ‘Mindfulness-based Stress Reduction therapy’, ‘Relaxation therapy’, ‘Exercise’, ‘Bright Light Therapy’, ‘Acupuncture therapy’, ‘Acupoint therapy’, ‘Music therapy’, ‘Yoga’, ‘Tai Chi’, ‘Qigong’, ‘Psychotherapy’ and ‘Randomised controlled trial’. The databases will include Pubmed, Web of Science, Cochrane Library, Embase, Google Scholar, China National Knowledge Infrastructure (CNKI), China Biomedical Literature Database (CBM), Wanfang and VIP. Searches will be ranging from the date of establishment of the database until 15 May 2024 and an updated study search will be conducted within 5 months of submission of the resulting manuscript (estimated to occur in August 2024). Languages will be limited to English and Chinese. In the case of Pubmed, a detailed search strategy can be found in the online supplemental materials. In addition, the references from the included studies will be manually screened to ensure that no relevant literature is disregarded or omitted.

Study selection

In line with the above strategy, two researchers (QC and XJ) will independently search the database and employ Endnote V.X9 software to manage the literature and remove duplicate studies. Initially, both researchers will perform a preliminary screening by reviewing the title and abstract of the literature. Subsequently, they will conduct a second screening by examining the full text, ultimately determining the final studies to be utilised based on the inclusion criteria, exclusion criteria and outcome indicators. In the event of disagreement between the two researchers, a third researcher (MK) will be consulted. The PRISMA flowchart will depict the selection process (figure 1).

Figure 1. Flowchart of study selection. CNKI, China National Knowledge Infrastructure; CBM, China Biomedical Literature Database; CRI, cancer-related insomnia.

Data extraction

Two researchers (QC and XJ) will read the full text of the screened literature while concealing the names of the authors, institutions and supporting fund projects, and will use Microsoft Office 2019 software to extract data from the final included literature. All relevant data from qualifying studies will be retrieved and organised into a standardised Excel file for significant information extraction.

The form will contain the following information:

1. General information: the first author, the publication year, the study country or region, the study type and the sample size.

2. Study characteristics: the study design, the random allocation method, the blinding method, the allocation concealment, the integrity of the outcome data, and the baseline characteristics.

3. Participant information: the age, the gender, the number of cases, the course of disease and the type and stage of cancer of patients in both the experimental and control groups.

4. Treatment: the type, time, frequency and intensity of the intervention in both the experimental and control groups.

5. Outcomes of interest: the primary and secondary outcome measures, the adverse events and the numbers of events, the duration of follow-up, the number of dropouts in each group and the reasons for dropouts.

Under the proposed approach, two researchers will extract the data independently, with any conflicts handled through discussion or negotiation with a third researcher (MK).

The risk of bias assessment

Two investigators (QC and XJ) will independently assess the risk of bias in the included studies employing the Cochrane Risk of Bias tool (RoB V.2.0).⁴² A cross-check will be conducted following completion of the evaluation. In case of disagreement, the third investigator (MK) will make the final decision. The degree risk of bias will be divided into ‘low risk of bias’, ‘high risk of bias’ and ‘uncertain risk of bias’. A summary of the risk of bias of RoB will be provided to assess the risk of bias of the included studies in a more intuitive manner.

Data synthesis

Pairwise meta-analysis

Before performing the NMA, a pairwise meta-analysis of directly compared interventions in RCTs using Stata V.14.2 software will be conducted. Subsequently, heterogeneity will be explored through Stata and the heterogeneity of studies will be tested using I² values.⁴³ If I²>50%, indicating excessive heterogeneity, the random-effect model will be employed for data analysis. Otherwise, the fixed-effect model will be used for analysis. For continuous data evaluated on the same scale, the mean difference (MD) and 95% CI will be utilised. When assessing data on different scales, the standard mean difference (SMD) and 95% CI will be utilised. For dichotomous data, the relative risk (RR) will be used with a 95% CI.

Network meta-analysis

ADDIS V.1.16.8, Stata V.14.2 and RV.4.1.2 will be used for NMA. Stata V.14.2 software will be used for network diagrams and comparison-adjusted funnel plots, RV.4.1.2 software for data quantitative analysis and ADDIS V.1.16.8 software for analysis of consistency. The NMA will be calculated and statistically analysed through the Markov Chain Monte Carlo method under a Bayesian framework.^{44 45} Random-effect models will be utilised for data pooling and analysis.⁴⁶ The initial chain will be set at four, with 50 000 iterations anticipated. The first 20 000 iterations will be used for annealing to reduce the influence of the starting value, with the remaining 30 000 iterations allocated to sample computation. The findings will be transformed to surface under cumulative ranking using R software, thus allowing for a visual comparison and evaluation of all interventions’ efficacy. Moreover, comparison-adjusted funnel plots will be drawn by stata software so as to detect publication bias.

Assessment of consistency and convergence

If the p value>0.05, the consistency between direct comparison and indirect comparison is good, and the consistency model will be employed for fitting. Conversely, if the p value<0.05, the inconsistency model will be used for fitting. Meanwhile, the Potential Scale Reduced Factor (PSRF) will be calculated using Brooks-Gelman Diagnostics to evaluate the convergence of the results. Convergence will be indicated as good if the PSRF tends to 1. Otherwise, the iteration parameters will need to be increased until the PSRF tends to 1 before data analysis can be conducted.

Assessment of reporting bias

We will employ a funnel plot to conduct publication bias analysis.⁴⁷ To further detect and quantify any potential publication bias, Begg’s test and Egger’s test will be carried out.⁴⁸ If p value>0.05, it indicates that there is no publication bias.

Sensitivity analysis

Sensitivity analysis of the primary outcome will be performed if the I² is 50% or higher in order to test the stability of the results.⁴⁹

Subgroup analysis

When data are available, subgroup analysis and meta-regression analysis will be conducted to investigate whether heterogeneity and between-group differences in results are influenced by tumour type (eg, central tumour and non-central tumour), TNM stage, patient age, gender and race.^{50 51}

Quality of evidence

The Grading of Recommendations Assessment, Development and Evaluation (GRADE) system will be applied to assess the quality of evidence for the primary and secondary outcomes of NMA.⁵² The downgrade factors include limitation, inconsistency, indirection, imprecision and publication bias. They are divided into four grades, in which no downgrade is ‘high quality’, a downgrade of one grade is ‘medium quality’, a downgrade of two grades is ‘low quality’ and a downgrade of three grades is ‘very low quality’. Two researchers will undertake the examination separately and then cross-check their findings. Any disagreements will be settled after consulting with a third researcher. GRADE proV. 3.6 software (http://www.gradeworkinggroup.org/) will be utilised to complete the evaluation process.

Patient and public involvement

None

Ethics and dissemination

Ethics committee approval is unnecessary for this review as the data here will be extracted from existing literature. The intention is to submit the study results for peer review and select a suitable academic publishing platform for dissemination to ensure broad sharing of the findings.

Discussion

Insomnia is a common clinical problem affecting more than 50% of cancer patients and survivors, with particular prevalence among lung cancer patients, breast cancer patients and head and neck cancer patients.¹⁷ Long-term insomnia is both detrimental to the body and is also harmful to mental health, aggravates mood disorders, reduces confidence in treatment and increases mortality and suicide rates. Therefore, seeking effective treatment methods to alleviate CRI is a clinical problem that urgently requires a solution. However, presently recommended pharmacological and CBT-I have not yielded satisfactory results in terms of efficacy or patient acceptability. This is due to the inability of patients to tolerate the side effects of pharmacological treatments, and the huge financial pressures associated with CBT-I.⁵³ There is evidence that cancer patients seeking relief from insomnia symptoms are increasingly utilising a variety of non-pharmacological treatments.^{32 54} These include MBSR, exercise therapy, relaxation therapy, acupuncture, tai chi and yoga. Nevertheless, there is no definitive evidence to compare the efficacy of these non-pharmacological treatments for CRI, and the quality of relevant literature is variable. Accordingly, this study will employ the NMA method to conduct a comprehensive quantitative analysis of RCTs investigating a range of non-pharmacological therapies for CRI. This approach will enable direct and indirect comparisons of different non-pharmacological interventions, thereby elucidating the efficacy of these treatments. Additionally, if data sufficiency allows, subgroup analyses will be conducted for concomitant symptoms such as fatigue, depression and anxiety, so as to explore the multifaceted efficacy of non-pharmacological treatments.

However, there are undeniably several limitations to this study. First, the non-pharmacological treatment of CRI is a very broad field of research, and the specific implementation methods of intervention time, duration, combined measures and types are not the same in all studies, and it is difficult to carry out blind so the effect of heterogeneity cannot be completely excluded. Second, this study will search publicly available Chinese and English literature, potentially leading to publication bias by excluding other languages and grey literature. Finally, the application of NMA has challenges in terms of data inconsistency, methodological differences and complexity of sensitivity analyses, which may affect the accurate assessment of the overall effect. The evidence supporting non-pharmacological therapies for treating CRI in clinical settings is equivocal, demanding a thorough examination and evaluation by clinicians of CRI patients.

This study will be a relatively comprehensive NMA for non-pharmacological therapies for the treatment of CRI. Integrating information from existing RCTs and ranking these non-pharmacological therapies based on their efficacy will offer valuable guidance for both clinicians and patients in choosing the best treatment modality and provide a new evidence-based rationale for non-pharmacological treatments for CRI.