Effect of non-pharmacological interventions on symptoms and quality of life in patients with hematological malignancies – A systematic review

A B S T R A C T Background: Non-pharmacological interventions have the potential to enhance health-related quality of life (HRQoL) through symptom management. This systematic review aims to identify, collate, and assess randomized controlled trials investigating the effect of non-pharmacological interventions on symptoms and HRQoL within hematology. Methods: MEDLINE/PUBMED, EMBASE, CINAHL, PSYCINFO and COCHRANE were searched up to April 2021. Outcomes were changes in symptoms and HRQoL. Results: Sixty-five studies were categorized into five intervention types: Mind/body (n = 9), Web-based (n = 9), Music/art (n = 6), Consultation-based (n = 4), and Physical activity (n = 37). We found significantly reduced fatigue (n = 12 studies), anxiety (n = 8) and depression (n = 7), with 11 studies showing significant improvements in HRQoL. Conclusions: The evidence for non-pharmacological interventions shows substantial variation in efficacy and methodological quality. While specific symptoms and HRQoL outcomes significantly favored the intervention, no particular intervention can be emphasized as more favorable, given the inability to conduct a meta-analysis.


Introduction
Over the last few decades, there has been considerable advancement in the treatment of hematological malignancies, resulting in a growing number of survivors and individuals living with hematological malignancies (Pulte et al., 2020).Living with hematological malignancies is associated with a significant symptom burden from treatment-related side effects and late effects (Manitta et al., 2011;Geyer et al., 2017).Non-pharmacological interventions have been developed and investigated to provide supportive care to patients.This is important because relying solely on pharmacology may not always be sufficient to address the early and late effects of the disease and treatment (O'Connor et al. 2021).Any treatment not classified as a registered drug, such as physical activity, and psychosocial or psychological interventions, is regarded as non-pharmacological (O'Connor et al. 2021;Cramp et al., 2013).
Hematological malignancies constitute a diverse and complex group of diseases, including both acute and chronic conditions and their respective treatment trajectories.Acute hematological diseases, such as acute leukemia and lymphoma have high early mortality if not treated promptly, whereas chronic hematological diseases, such as chronic myeloid leukemia (CML), multiple myeloma (MM), myeloproliferative neoplasms (MPN) and myelodysplastic syndrome (MDS) have prolonged treatment trajectories (Rodriguez-Abreu et al., 2007;Kaifie et al., 2016).Therefore, the treatment protocols differ depending on the diagnosis, ranging from intravenous chemotherapy for acute leukemia or aggressive lymphoma to daily or weekly treatment regimens for MM, MPN or CML.Diagnoses such as chronic lymphatic leukemia (CLL) or MDS necessitate a "watch and wait" approach, involving regular outpatient monitoring for an extended period, potentially lifelong (Rodriguez-Abreu et al., 2007;Kaifie et al., 2016).Allogeneic hematopoietic stem cell transplantation (HSCT) is a curative treatment option for malignant hematological diseases, primarily recommended to patients with leukemias or myelodysplastic syndromes, however, patients with other diagnoses can also benefit from a HSCT (Passweg et al., 2020).
Patients experience symptoms that can be attributed to both the underlying hematological malignancy and their medical treatment, including fatigue, insomnia, somnolence, cognitive impairment, discomfort, and stress (Manitta et al., 2011;A. E. Hall et al., 2015).Additionally, patients report having several unmet supportive care needs, mainly concerning psychological and physical aspects of everyday life, resulting in a decreased health-related quality of life (HRQoL) (A.E. Hall et al., 2015;Allart-Vorelli et al., 2015;Boyes et al., 2015;Jacobs et al., 2019).Reduction in HRQoL negatively impacts the ability to return to work and engagement in everyday activities (Manitta et al., 2011;Behringer et al., 2016;A. Hall et al., 2016).
In a cross-sectional observational study, hematological patients report a high symptom burden, averaging 8.8 symptoms.This increases during treatment (symptom average 9.7), in newly diagnosed patients (10.5), in refractory (12.4) or relapsed disease (10.6) and during hospitalization (symptom average 11.4 inpatient vs. 8.0 outpatient) (Manitta et al., 2011).Moreover, 71% of hematological cancer survivors expressed a need for assistance in managing symptoms related to cancer or its treatment (A.Hall et al., 2016).
Aggregating data on the effect of non-pharmacological interventions on symptoms and HRQoL could optimize the management of disease and treatment trajectories for patients with hematological malignancies.To date, there is no comprehensive overview of non-pharmacological interventions specific to hematological malignant diagnosis groups.However, a critical review conducted in 2006 within the field of hematology and oncology identified a limited number of randomized controlled trials (RCT) for supportive care interventions (Joske et al., 2006).Currently, there is limited evidence available on non-pharmacological interventions.This emphasizes the need for further investigation of non-pharmacological interventions tailored for patients with hematological malignancies to provide evidence supporting symptom management and improvements in HRQoL.
This systematic review aimed to identify, collate, and assess RCTs that investigated the effect of non-pharmacological interventions on symptoms and HRQoL in patients with hematological malignancies.

Methods
This systematic review followed the PRISMA guidelines: 'Preferred Reporting Items for Systematic Reviews and Meta-Analyses' (Page et al., 2021).The protocol was registered at the international register of systematic reviews PROSPERO in March 2021 (Registry no.CRD42021237344).

Search strategy
Five databases were searched up to April 2021: Public Medline (MEDLINE/PUBMED), EMBASE, Cumulative Index of Nursing and Allied Health (CINAHL), APA PsycINFO (PSYCINFO), and the Cochrane Library (COCHRANE).The search strategy was developed using a matrix based on the principals from PICO (population, intervention, control, and outcomes), and compiled into three search focuses (population, intervention and outcomes).The matrix was consistently applied in all selected databases and included 1) Hematological malignancies (population), 2) Non-pharmacological interventions (intervention) and 3) Symptoms and/or HRQoL (outcomes).A combination of MeSH terms, thesaurus, indexed terms, and free text for the three search focuses was used.RCTs were identified by the Cochrane Collaboration's highly sensitive search strategy for identifying RCTs, without restrictions on the publication year.The search strategy was validated by an academic librarian before initiation.In addition, PROSPERO was searched for ongoing systematic reviews related to the scope of this systematic review (Appendix A: PICOsearch terms).

Eligibility criteria and outcomes
Studies were eligible for inclusion in this systematic review if they met the following criteria: 1) Enrolled patients ≥ 18 years with a hematological malignancy, 2) Investigated the effect of nonpharmacological interventions compared with no care, standard care, sham, or placebo.Non-pharmacological interventions included physical activity (e.g.exercise-based, high and low-intensity training, pilates, yoga), psycho-educational (e.g.meditation, cognitive-based interventions), psychosocial (e.g.nurse consultations, peer-to-peer support, group-based interventions) or multimodal interventions, 3) Included the following primary or secondary outcomes: disease and treatment-related symptoms and/or HRQoL measured on validated questionnaires, and 4) Were RCTs, including randomized cross-over and randomized two-armed pilot trials.

Outcomes
The outcomes assessed in this systematic review were changes in single symptoms, multiple symptoms, symptom clusters and HRQoL.Measures of symptoms and HRQoL included generic, cancer, and hematology-specific patient-reported outcomes (PROs) on validated questionnaires.The effect was reported as between-group differences at the end of intervention.In cases with multiple test time points postintervention, the time point nearest to the end of the intervention was reported.
Anticipated outcomes were registered in PROSPERO, but deviations occurred from the reported protocol.We planned to collect data on feasibility and safety, however, we chose to exclude these data due to a lack of systematic reporting in the original RCTs.Sub-analyses within diagnostic groups were initially planned but excluded because of heterogeneity within the diagnostic groups and across interventions.The data is descriptively presented at the diagnostic group level.

Study selection and quality assessment
Covidence was used to manage and screen all studies by two authors (MP & MSE).MP & MSE reviewed the studies considered relevant and read them in full.The studies were then categorized as either included, unclear (for further discussion), or excluded based on the inclusion criteria.Any disagreements between MP & MSE, were discussed with a third author (MJ).In case of overlapping data sets presented in different studies, we included the most recent publications.
Cochrane's risk of bias tool RoB2 was used to evaluate the methodological quality and assess risk of bias in the included studies.RoB2 ensured that five key domains were evaluated: 1) randomization process, 2) deviations from the intended intervention, 3) missing outcome data, 4) measurement of the outcome, and 5) selective outcome reporting (Sterne et al., 2019).Two authors (MP & MSE) independently conducted the RoB2 assessment and then discussed until consensus was reached.In case of any discrepancies, a third author (MJ) was involved.

Statistical analysis
A meta-analysis was not feasible due to the heterogeneity of outcomes and intervention types.However, results from included studies reporting statistically significant differences between groups are summarized in Table 2.
A small change in symptoms and HRQoL scores can result in a statistically significant result with a sufficiently large sample size.As an additional consideration, we suggest comparing the magnitude of between group differences with minimal clinically important differences (MCID).MCIDs are simple to interpret and provide information on the value of the intervention for the patients (Cleeland et al., 2011).As cut-off thresholds for MCID have not been established for all validated questionnaires used to measure symptoms and HRQoL, the MCID will be reported whenever possible to provide further context to statistically significant results and include the patients' perspective (Higgins and Cochrane Collaboration, 2020).
If reported in the studies, means and standard deviations (SD) of changes from baseline to post intervention time in both intervention group and control group were used to calculate Cohen's d to quantify effect sizes (J.Cohen, 1992).The formula √ was used.If 95% confidence intervals for within-group changes were reported instead of SDs, an approximate estimate for SD was calculated using the formula where L is the length of the confidence interval and n is the group size.Cohen's d <=0.2 is considered a small effect size; 0.5-0.8represents a medium effect size, and 0.8 and above is a large effect size (J.Cohen, 1992).A Cohen's d of size 0.8 implies that approximately 79% of individuals receiving the intervention will experience a change larger than the average chance from baseline to post-treatment in the control group.

Result from study selection
In total, 14,610 studies were identified through the initial search of the databases.After removing duplicates (n=2223), 12,387 studies remained for title and abstract screening (Fig. 1: PRISMA flow chart).
After full-text screening, 65 studies were included.The nonpharmacological interventions in the studies were categorized into five intervention types: i) Mind/body: ii) Web-based, iii) Music/art, iv) Consultation-based, and v) Physical activity (Table 1).
A broad range of hematological malignant diagnoses are included in the identified RCTs to assess the effect of the non-pharmacological intervention on symptoms and HRQoL.Some studies included one specific diagnosis, while others included several hematological malignant diagnoses (Clinical and sociodemographic characteristics are found in Table 1).

Records
Effect sizes could only be calculated for three statistically significant outcomes in one study, indicating a medium magnitude for anxiety, depression and HRQoL (Grossman et al., 2015).Determining the MCID for most outcomes was not possible due to insufficiently reported data or the absence of specific MCID cut-off values for the diverse range of outcomes.However, Deng et al. demonstrated MCIDs related to appetite, drowsiness and nausea (Deng et al., 2018).
The risk of bias assessment using Rob2 indicated some methodological concerns, classifying the studies as follows: low risk of bias (n=2), some concerns (n=3) and high risk of bias (n=4) (Table 1, Appendix B).

Web-based interventions
Web-based interventions (n=9) included 1421 participants testing computer or mobile phone delivered interventions (n=5) or the systematic use of online-administered PRO data (n=3) (Ruland et al., 2010;David et al., 2013;Braamse et al., 2016;Sagari et al., 2018;Syrjala et al., 2018;Bryant et al., 2020;Jim et al., 2020;Moore et al., 2020;Stevenson et al., 2020).Published between 2010 and 2020, the studies primarily took place in the USA (n=4), with contributions from the Netherlands, Norway, Australia, Germany, and Japan.Fatigue was measured in four studies, and significant improvements favoring the interventions were observed in two studies compared with control or waitlist groups (Bryant et al., 2020;Jim et al., 2020).Symptoms of depression or anxiety were measured in several studies, but no significant findings were reported (Braamse et al., 2016;Syrjala et al., 2018;Stevenson et al., 2020).However, recipients of an internet-based survivorship program after HSCT demonstrated a significant reduction in distress compared to controls (Syrjala et al., 2018).
HRQoL was investigated in four studies (Braamse et al., 2016;Sagari et al., 2018;Syrjala et al., 2018;Jim et al., 2020), and among them, two found a significant effect in favor of the interventions.
A large effect size was observed in one HRQoL outcome (Sagari et al., 2018), however, this was the only study in which effect sizes could be calculated due to limitations in the available data.In two studies, the between-group difference in HRQoL exceeded the MCID cut-off in favor of the interventions (Sagari et al., 2018;Jim et al., 2020) (Table 2).
An evaluation with RoB2 resulted in studies classified as follows; low risk of bias (n=0), some concerns (n=7) and high risk of bias (n=2) (Table 1; Appendix B).

Music and art interventions
Among the six studies identified, the majority investigated a music intervention (n=5) (Cassileth et al., 2003;Fredenburg and Silverman, 2014;Bates et al., 2017;Bro et al., 2019;Dóro et al., 2017), with one intervention involving art (McCabe et al., 2013).In total, 632 participants were included across the six studies varying from 13 to 199 participants.Studies were conducted in the USA (n=3), Brazil, Denmark, andIreland, andpublished between 2003 and2019.Two studies observed a significant effect on anxiety and mood in favor of the interventions (Dóro et al., 2017;McCabe et al., 2013).In one study investigating a music intervention, HRQoL was assessed, but no significant findings were reported (Bro et al., 2019).Effect sizes were considered large in relation to anxiety and mood (Dóro et al., 2017).Unfortunately, it was not possible to conclude on MCID because of missing outcome data and missing MCID cut-off values (Table 2).
An evaluation with RoB2 resulted in studies classified as follows; low risk of bias (n=0), some concerns (n=3) and high risk of bias (n=3) (Table 1; Appendix B).

Consultation-based interventions
Four studies involved interventions where a nurse or physician delivered face-to-face consultation-based palliative or survivorship care, encompassing both inpatient and outpatient care settings.The total sample across the four studies included 578 participants (Parker et al., 2020;K. Taylor et al., 2019;El-Jawahri et al., 2016;2021).In addition, Parker et al. also included 42 physicians assigned to a training program (5 hours) on survivorship planning or wellness rehabilitation consultations (2 hours) (Parker et al., 2020).All the studies, except one from Australia, were conducted in USA.
In two studies conducted by El-Jawahri, the interventions including palliative care, showed significant changes in symptoms and HRQoL, favoring the intervention groups (El-Jawahri et al., 2016;2021).On the contrary, studies on survivorship care planning did not find a significant impact on either HRQoL (Parker et al., 2020) or symptoms (K.Taylor et al., 2019).
Unfortunately, effect sizes could not be calculated because of missing data in the original studies.However, a MCID was present in the study by El-Jawahri et al. in relation to anxiety and fatigue (El-Jawahri et al., 2016) (Table 2).
The RoB2 assessment resulted in studies classified as follows; low risk of bias (n=0), some concerns (n=3) and high risk of bias (n=1) (Table 1; Appendix B).
An evaluation with RoB2 resulted in studies classified as follows; low risk of bias (n=0), some concerns (n=29) and high risk of bias (n=8) (Table 1; Appendix B).

Summary across non-pharmacological interventions
Across intervention types, significant improvements were found for symptoms of fatigue (n=12), anxiety (n=8), depression (n=7), pain (n=3) and sleep disturbance (n=7).Regarding fatigue, two out of the 12 studies exhibited large effect sizes (Furzer et al., 2016;Chuang et al., 2017), and five studies reduced fatigue above the MCID cut-off limit (Jim et al., 2020;El-Jawahri et al., 2016;Courneya et al., 2009;Chuang et al., 2017;Hathiramani et al., 2021).Physical activity interventions were most likely to reduce fatigue based on information on statistical significance, the effect size (Cohen's d) and MCID (Table 2).The studies included over 20 diagnosis-specific, cancer-specific, and generic questionnaires to quantify symptoms across intervention types and diagnoses (Table 1).HRQoL was measured in 35 studies across intervention types, and 11 of these studies found a significant between-group difference in favor of the intervention.In four of the 11 studies with positive HRQoL outcomes, the effect sizes were either large or medium (Grossman et al., 2015; Sagari et al., 2018;Furzer et al., 2016;Chuang et al., 2017).MCIDs were present in seven of the 11 studies (Sagari et al., 2018;Jim et al., 2020;Chuang et al., 2017;Furzer et al., 2016;Hathiramani et al., 2021;Jarden et al., 2016;Courneya et al., 2009).Web-based interventions and physical activity interventions were most likely to improve HRQoL considering statistically significant outcomes, effect size (Cohen's d) and MCID (Table 2).The most common questionnaires to measure HRQoL in this population were two cancer specific questionnaires; The European Organization for Research and Treatment of Cancer Core Quality of Life Questionnaire (EORTC QLQ-C30) and The Functional Assessment of Cancer Therapy -General (FACT-G), and one generic questionnaire; The Short Form Health Survey (SF-36).In 27 of 65 included studies, both symptoms and HRQoL were assessed following a non-pharmacological intervention.Among these, 10 studies reported significant p-values related to one or more symptoms as well as HRQoL.

Intervention type and diagnosis
More than half of the studies included patients with a mixed sample of hematological malignant diagnoses (n=33).Overall, leukemia was the most investigated patient population (n=15), followed by lymphoma (n=11), MM (n=5) and MPN (n=1) across types of nonpharmacological interventions.In 29 studies the treatment trajectory was pre, during or post HSCT (Tables 1 and 3).
Studies including patients with leukemia exclusively had most significant findings relative to the number of studies investigating symptoms as an outcome.HRQoL was measured in six studies including lymphoma patients, with significant p-values in half of the studies.

Discussion
In this systematic review, we identified, collated, and assessed 65 randomized controlled trials that investigated the effect of nonpharmacological interventions on symptoms and HRQoL in patients with hematological malignancies.Significant benefits were found in HRQoL and symptoms of fatigue, anxiety, depression, pain and sleep disturbance.
Several studies, including patients across various cancer types have investigated the effects of different types of non-pharmacological interventions.A meta-analysis, including 69 studies showed significant effects on cancer-related fatigue following non-pharmacological interventions within subgroups of yoga, mindfulness and psychosocialbased interventions (Haussmann et al., 2022).This suggests the potential for non-pharmacological interventions to positively impact symptoms and HRQoL in cancer.However, focusing on patients with hematological malignancies, the evidence is limited compared to studies that include various cancer diagnoses.
Patients with hematological malignancies experience a multitude of symptoms that may affect their HRQoL (Manitta et al., 2011;A. E. Hall et al., 2015;Boyes et al., 2015).Consequently, various non-pharmacological interventions have been investigated to reduce symptoms and improve HRQoL.In Section 3.2 fatigue was found to be reduced following mind/body interventions, aligning with findings in a systematic review and meta-analysis investigating the impact of mind and body practices on fatigue in children and adults with cancer and undergoing HSCT (Duong et al., 2017).
In two studies reported in Section 3.3 (Bryant et al., 2020;Jim et al., 2020) a web-based intervention reduced fatigue in patients, consistent with findings from a Cochrane review of telephone-delivered interventions (Ream et al., 2020).Ream et al. found symptom management improvement in patients with cancer, primarily related to fatigue but also depression, anxiety and distress (Ream et al., 2020).Our review did not find significant changes in symptoms of anxiety and depression following web-based interventions; however, this can be attributed to the heterogeneity within the diagnostic groups.
In Section 3.4, symptoms of anxiety and depression were significantly reduced in two studies following a music intervention (Dóro et al., 2017;McCabe et al., 2013), and HRQoL was only measured in one study, but with no significant findings (Bro et al., 2019).In contrast, a Cochrane review including cancer patients found that music interventions may have a positive effect on anxiety, and inconsistent with our finding, they found significant improvements in HRQoL (Chan et al., 2020).
Only two RCTs examining palliative care for hematological diseases could be identified in Section 3.5.The benefits from integrating palliative care into clinical practice are well recognized within solid cancer diagnoses, however, knowledge within hematology remains limited (Niscola et al., 2018).
In Section 3.6, fatigue was measured in more than half of 37 physical activity intervention studies, with seven demonstrating a significant effect.This is consistent with findings from a meta-analysis in a Cochrane Review on the effect of aerobic physical activity, revealing positive effects on fatigue, although the evidence regarding HRQoL remains unclear (Knips et al., 2019).The study by Knips et al. was published in 2019, whereas we included studies published between 2003 and 2021.A consensus statement on exercise guidelines for cancer survivors stated that there is good evidence of the benefits of aerobic exercise on anxiety, depression, and fatigue, as well as a combination of aerobic and resistance training related to HRQoL (Campbell et al., 2019).However, this refers to cancer survivors across diagnoses.The recognition of the effect of physical activity within hematological malignant diagnoses is still unclear in relation to several symptom and HRQoL outcomes, and more sufficiently powered studies are needed to support the evidence (Knips et al., 2019).
The utilization of PROs as a component of the intervention was found to reduce symptoms in two out of three studies in Section 3.3 (Ruland et al., 2010;Bryant et al., 2020).Across various diagnoses, PROs are used in diverse ways, serving both as a tool to provide data to determine effects or for longitudinally monitoring symptoms, but also as an active component of the intervention as in Ruland et. al. (2010) and Bryant et al. (2020).The use of PRO in clinical research has increased in recent years, and PRO measures are found to capture outcomes important to the patient (Howell et al., 2015).Moreover, PROs have been shown to improve patient-clinician communication, identify unrecognized symptoms, and enhance patient HRQoL (Howell et al., 2015;Maguire et al., 2021).This underscores the potential significance of PROs in the field of hematology, calling for further investigation.
To our knowledge, only a few systematic reviews investigating symptoms and HRQoL outcomes across various types of nonpharmacological interventions have been published in patients with hematological malignancies (O'Connor et al., 2021;Joske et al., 2006).A critical review on hematological and oncological patients summarized Hematological malignancies=Mix of hematological malignant diagnoses; HSCT=Hematopoietic stem cell transplantation; MPN=Myeloproliferative neoplasms; n=Number of studies the evidence of complementary therapies, such as massage, acupuncture, music and psychological interventions (Joske et al., 2006).The study by Joske et al. (2006) underscores the paradox, wherein many cancer patients have an increased interest in trying non-pharmacological interventions, with or without involving their clinicians.However, there is a lack of evidence regarding the effectiveness of such interventions (Joske et al., 2006).Thus, it is notable that only four of the 65 RCTs in this review were published before 2006, indicating a growing trend in the number of RCTs on non-pharmacological interventions within hematology being conducted over the last two decades.In 2021 a review of non-pharmacological interventions in acute myeloid leukemia concluded that there is a need for sufficiently powered studies (O'Connor et al. 2021).Despite the potential improvement of symptoms and HRQoL through non-pharmacological interventions, patients with hematological malignancies continue to experience a high symptom burden that affects HRQoL (Manitta et al., 2011;Geyer et al., 2017;Nielsen et al., 2017), highlighting the need for more research.
In 27 studies, both symptoms and HRQoL served as outcome measures.In 10 of these studies, a significant effect was observed for both outcomes, irrespective of the types of interventions.Although a causal relationship between symptoms and HRQoL cannot be inferred in this review, a correlation between symptoms and HRQoL is shown elsewhere (Tolstrup Larsen et al., 2018).
Across types of interventions, several studies in this systematic review recruited across various hematological diagnoses, possibly to achieve a larger sample size.Also, 29 studies included patients undergoing HSCT with similar treatment trajectories, but the underlying diagnoses are still divergent.Nevertheless, a mixed patient study population limits the possibility to draw specific conclusions about a particular patient population unless sub-analyses are conducted.However, 32 studies did investigate a specific diagnosis, where leukemia (n=15) was the most investigated, followed by lymphoma (n=11).Both leukemia and lymphoma are acute hematological diseases with high risk of early mortality if left untreated.Thus, chronic hematological malignant diagnoses are less represented in RCTs that investigate nonpharmacological interventions, despite the considerable symptom burden experienced by patients with chronic hematological malignancies (Geyer et al., 2017;Nielsen et al., 2017).We identified five studies including patients with MM and one including patients with MPN alone (Huberty et al., 2019;Emanuel et al., 2012).Hence, patients with chronic hematological diagnoses such as MM, MPN, or MDS should be considered in future trials when investigating the effects of non-pharmacological interventions, given the limited evidence available.
There are some limitations in this systematic review.Inclusion of studies only published in English or Danish may have increased the risk of publication bias (Guyatt et al., 2011).Another limitation of the findings is that we included studies measuring symptoms and HRQoL as both primary and secondary outcomes based on the pre-selected inclusion criteria presented in Section 2.2.As a result, not all studies had statistical power to detect an effect.This could potentially influence the outcome results in our findings, where a larger sample size could hypothetically enhance the possibilities for significant findings and reduce the risk of a type II error.This is supported by O′Connor et al. who emphasize the need for sufficiently powered studies within hematology (O'Connor et al. 2021).
Heterogeneity between the different intervention types and outcome measures did not support a meta-analysis.Instead, results are reported only for outcomes where statistically significant differences were observed between groups.However, it is important to acknowledge some limitations in relying solely on statistical significance to determine the effect of an intervention.Significant p-values do not provide information about the size of the effect or the clinical relevance of the result.In this review, the statistically significant results have been supplemented with data on effect sizes (Cohen's d) and MCID, wherever this could be computed from available data.Cohen's d quantifies the effect on a scale that is comparable across different outcomes (J.Cohen, 1992).However, limitations are associated with Cohen's d, as a large effect size does not guarantee clinical relevance and noticeable impact for the patients.To address this, MCID has been calculated to ensure the patients perspective is captured.However, it is a limitation that only a few symptom and HRQoL outcome measures have established cut-off values (Higgins and Cochrane Collaboration, 2020).In addition, Woaye-Hune et al. highlight disadvantages in the calculation of MCID, noting that the method of missing data imputation can affect and introduce bias into the result of MCID cut-off limits (Woaye-Hune et al., 2020).In summary, the calculation of Cohen's d was feasible in only a limited number of studies due to divergent reporting practices in the included studies, leaving several statistically significant outcomes with no opportunity to establish a MCID or Cohen's d, diminishing the transparency of the results.
Only few studies in this systematic review received a low risk of bias assessment, as blinding was not possible due to the intervention design.In just two studies investigating a mind/body intervention in Section 3.2, the blinding criterion could be met, as the control group received placebo treatment (Blackburn et al., 2017;Deng et al., 2018).It is seldom feasible to blind non-pharmacological interventions for the study participants and investigators.Therefore, knowledge of the assigned intervention could possibly influence the study participants' responses, introducing a higher risk of bias (Sterne et al., 2019).
To our knowledge, a systematic review collating types of nonpharmacological interventions and their effects on symptoms and HRQoL in hematological malignancies has not been published previously.Heterogeneity between the different types of interventions and outcome measures did not support a meta-analysis.A more focused investigation of a pre-selected type of non-pharmacological intervention might have increased the possibility of conducting a meta-analysis.Fatigue, the most frequently investigated symptom, showed significant findings in 12 studies, primarily in those involving a physical activity intervention.HRQoL was measured in 35 studies, with 11 studies significantly favoring non-pharmacological interventions, mainly those comprising web-based or physical activity interventions.However, interpretating the results requires caution due to varying methodological quality and missing data needed to establish effect sizes and MCIDs.

Non
Overall, the evidence of non-pharmacological interventions on symptoms and HRQoL has substantial variation in efficacy and methodological quality.While several symptoms and HRQoL outcomes showed positive effects, no superior non-pharmacological treatment can be highlighted as a meta-analysis was not feasible.
-pharmacological interventions were investigated in 65 RCTs measuring symptoms and HRQoL in hematological malignancies, as summarized in this systematic review.The interventions were categorized into i) Mind/body interventions, ii) Web-based interventions, iii) Music/art interventions, iv) Consultation-based interventions, and v) Physical activity interventions.

Table 1
Summary of non-pharmacological interventions, symptoms and quality of life outcomes.

Table 1
(continued on next page) M.Pedersen et al.

Table 1
(continued ) (continued on next page) M.Pedersen et al.

Table 1
(continued ) RCT = Randomized controlled trial.HSCT = Hematopoietic stem cell transplantation Abbreviation questionnaires: BDI: Beck Depression Inventory; BFI: Brief Fatigue Inventory; BSI: Brief symptom inventory; CES-D: Center for Epidemiological Studies-Depression Scale; CES-D-10: Short Form Center for Epidemiological Studies-Depression Scale; CTXD: Cancer and Treatment Distress; DASS21: The Depression, Anxiety, Stress Scale; EORTC-QLQ-C30: The European Organization for Research and Treatment of Cancer Core Quality of Life Questionnaire; ESAS: Edmonton Symptom Assessment System; FACT-An: The Functional Assessment of Cancer Therapy -Anemia; FACT-BMT: The Functional Assessment of Cancer Therapy -Bone Marrow Transplantation; FACIT-F: The Functional Assessment of Cancer Therapy -Fatigue; FACT-G: The Functional Assessment of Cancer Therapy -General; FACT-Leukemia: The Functional Assessment of Cancer Therapy -Leukemia; FSI: Fatigue Symptom Inventory; FQ: Fatigue Questionnaire; GAD7: Generalized Anxiety Disorder; HADS Hospital Anxiety and Depression Scale; ITPA: Interactive tailored patient assessments module; MDASI: The MD Anderson Symptom Inventory; MFI: Multidimensional Fatigue Inventory; MFIS: Modified Fatigue Impact Scale; MPN-SAF: The multifactor MPN Symptom Assessment Form; MYPOS: Myeloma Patient Outcome Scale; NIH PROMIS: National Institutes of Health Patient Reported Outcomes Measurement Information System; NRS: Numerical rating scale; PHQ-9: Severity of depression measured using the Patient Health Questionnaire; POMS-sf: Profile of Mood States short-form; PQoLC: Profile of health-related quality of life 2 Primary outcome.3Between-groupcomparisons are hampered due to sample size limitations.4Nop-values reported. 5Rob2 in Appendix B. * Significant effect (P≤0.05).

Table 2
Significant symptom and health related quality of life outcomes, effect sizes and minimal clinical important differences.
d(continued on next page) M.Pedersen et al.

Table 2
(continued ) Higher value indicates an improvement in symptom/HRQoL.↓ Lower value indicates improvement in symptom/HRQoL.
↑ * P-value comparisons changes over time.a Primary outcome at 6 months among participants impaired on CTXC at baseline.b Adjusted for covariates.c Group difference in changes over time (IG-CG).d Group difference at post-test (IG-CG).e IG = Therapist PT; CG=Wii PT.MCID: Minimal clinically important difference.

Table 3
Number of studies by diagnosis and type of intervention.