{
  "abstract": "Objective Loneliness and social isolation can have profound emotional and psychological impact on patients with cancer. Previous studies suggest loneliness may adversely impact cancer prognosis and survival. Potential mechanisms include a variety of biological, psychological and social factors, from impaired ability to access treatment to immune dysregulation. However, its impact is not clearly defined. We conducted the first systematic review and meta-analysis to investigate mortality in relation to loneliness and social isolation among cancer populations.Methods and analysis We systematically searched MEDLINE, Embase and PsycINFO until 13 September 2024, for studies reporting on the impact of loneliness or social isolation on all-cause and cancer mortality in patients with cancer. Three reviewers independently screened studies, extracted data and assessed bias. A random-effects meta-analysis was used to examine associations between loneliness/social isolation and all-cause and cancer mortality.Results Of 12 602 citations, 16 studies met eligibility criteria and 13 were included in the meta-analysis. Social isolation and loneliness were most frequently measured using the Social Network Index and UCLA Loneliness Scale, respectively. The median sample size was 6248, with a mean participant age of 63 years. Meta-analysis demonstrated loneliness/social isolation was associated with increased all-cause mortality (HR (95% CI)=1.34 (1.26 to 1.42), p<0.001) and cancer-specific mortality (HR (95% CI)=1.11 (1.02 to 1.21), p=0.014).Conclusion Loneliness and social isolation may be associated with increased all-cause and cancer-specific mortality in patients with cancer. If these findings are confirmed by future, more definitive studies, together they support the need to incorporate psychosocial assessments and targeted interventions into cancer care to improve patient outcomes.PROSPERO registration number CRD42024590482.",
  "authors": [
    {
      "affiliations": [
        "University of Toronto Temerty Faculty of Medicine, Toronto, Ontario, Canada"
      ],
      "name": "Samantha Cheng"
    },
    {
      "affiliations": [
        "University of Toronto Temerty Faculty of Medicine, Toronto, Ontario, Canada"
      ],
      "name": "Joshua Yi"
    },
    {
      "affiliations": [
        "University of Toronto Temerty Faculty of Medicine, Toronto, Ontario, Canada"
      ],
      "name": "Keiran Pace"
    },
    {
      "affiliations": [
        "Department of Biostatistics, University Health Network, Toronto, Ontario, Canada"
      ],
      "name": "Anna Santiago"
    },
    {
      "affiliations": [
        "Radiation Medicine Program, Princess Margaret Hospital Cancer Centre, Toronto, Ontario, Canada"
      ],
      "name": "Carlton Johnny"
    },
    {
      "affiliations": [
        "Department of Supportive Care, BC Cancer Agency, Vancouver, British Columbia, Canada"
      ],
      "name": "John-Jose Nunez"
    },
    {
      "affiliations": [
        "University of Toronto Temerty Faculty of Medicine, Toronto, Ontario, Canada",
        "Radiation Medicine Program, Princess Margaret Hospital Cancer Centre, Toronto, Ontario, Canada"
      ],
      "name": "Jane Jomy"
    },
    {
      "affiliations": [
        "University of Toronto Temerty Faculty of Medicine, Toronto, Ontario, Canada",
        "Radiation Medicine Program, Princess Margaret Hospital Cancer Centre, Toronto, Ontario, Canada",
        "Radiation Oncology, BC Cancer, Vancouver, British Columbia, Canada"
      ],
      "name": "Srinivas Raman"
    }
  ],
  "full_text": "WHAT IS ALREADY KNOWN ON THIS TOPIC Loneliness and social isolation can have significant emotional and psychological effects on patients with cancer, though its impact has not been clearly defined. We conducted the first systematic review and meta-analysis to investigate mortality in relation to loneliness and social isolation among cancer populations.WHAT THIS STUDY ADDS Our meta-analysis demonstrated an association between loneliness and social isolation with increased all-cause and cancer-specific mortality.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY With stronger evidence for a causal link, psychosocial assessments and targeted interventions for loneliness and social isolation could be investigated, and if they show benefit, should be integrated in cancer care to improve patient outcomes.Introduction Cancer remains a leading cause of morbidity and mortality worldwide, with nearly 20 million new cancer case diagnoses and 9.7 million cancer-related deaths in 2022. 1 These numbers are projected to rise to 35 million incident cases and 18.5 million cancer deaths by 2050.1 Despite advancements in treatment and supportive care, cancer continues to impose significant physical, emotional and economic burdens on patients and healthcare systems.2 3 Research has predominately focused on biological factors and medical therapies influencing cancer outcomes, but increasing evidence suggests that psychosocial determinants, such as loneliness and social isolation, may also play a critical role in patient outcomes.4 5Loneliness, defined as the subjective perception of social disconnection, and social isolation, defined as the objective lack of social relationships or engagement, are common among patients with cancer, affecting an estimated 16–47% of this population.6–8 Although conceptually distinct, these constructs are frequently studied together due to their shared risk pathways, overlapping measurement in research and similar clinical significance.9 10 For instance, loneliness has been linked to adverse health outcomes, including cognitive problems, sleep disorders, immune dysfunction and pain.11–13 Similarly, social isolation is associated with malnutrition, physical inactivity, elevated C-reactive protein and lipid levels and weakened immune function.14 Loneliness and social isolation were also found to be independent, contributing factors to increased mortality in a general adult population.10 However, the extent to which these psychosocial factors contribute to cancer mortality remains unclear, requiring further study. Therefore, the objective of this review was to examine the association between loneliness and social isolation and mortality among patients with cancer. This systematic review synthesises existing evidence to assess this relationship.Methods This systematic review was conducted in adherence with the 2020 Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines. 15 The protocol was registered on PROSPERO (CRD42024590482).Eligibility criteria We included studies describing the impact of loneliness or isolation on mortality among patients with cancer, where loneliness or social isolation was operationalised (eg, based on subjective feelings of loneliness, the diversity of social relationships and/or frequency of social interactions), using validated, predetermined questionnaires, such as the UCLA Loneliness Scale, Social Network Index and Social Integration Index. 16–18 Non-English publications, articles describing loneliness without associations to clinical outcomes, abstracts and review papers were excluded.Information sources and searches Systematic, structured literature searching was conducted by a trained clinical epidemiologist (JJ). MEDLINE, Embase and PsycINFO were systematically searched from inception to 13 September 2024, using the search strategy in online supplemental file 1. Boolean logic was used to search the terms “loneliness”, “isolation” and “mortality”. To identify any potentially missing studies, the electronic searches were supplemented by hand-searching reference lists of retrieved papers and review articles.SP110.1136/bmjonc-2025-000840.supp1Supplementary dataStudy selection Independent reviewers (SC, JY, KP) performed two levels of screening for eligible studies using Covidence (Veritas Health Innovation, Melbourne, Australia, available at www.covidence.org). All disagreements between reviewers were identified by Covidence and were resolved through joint discussion until mutual agreement. A third reviewer was involved for adjudication at all stages of screening. All studies removed at either level of screening had a documented reason for exclusion in accordance with PRISMA guidelines.Data extraction A standardised data extraction form was developed jointly by two investigators (JJ and SR) and consensus exercises were conducted in advance of data extraction. Data charting was conducted by three investigators (SC, JY, KP) and any disagreements were resolved through discussion until consensus was reached. The following data were extracted from the included literature, including general information (first author, publication year, countries and language), study population, participant demographics, baseline characteristics; details on how loneliness was operationalised; impact on mortality outcomes; mortality outcomes, times of measurement and adjusted covariates. In the event of missing data, corresponding authors were contacted directly to obtain clarification.Risk of bias assessment Using the Cochrane risk-of-bias instrument, 19 three investigators (SC, JY, KP) independently and in duplicate, assessed each trial for risk of bias, including random sequence generation; allocation concealment; blinding of patients, caregivers, data collectors and outcome adjudicators; selective reporting; and incomplete outcome data. We categorised a trial as being at high risk of bias overall if we identified high risk of bias among any of the eight items. Risk of bias assessments are presented using robvis software (https://mcguinlu.shinyapps.io/robvis/) per outcome in online supplemental file 2.SP210.1136/bmjonc-2025-000840.supp2Supplementary dataStatistical analysis Descriptive statistics and analyses of inter-rater agreement were conducted using IBM SPSS V.30, and random-effects meta-analyses of HRs were conducted using the ‘meta’ package in R V.4.4.1. 20–22 HR values extracted were drawn from the most isolated/lonely group, when applicable. We performed meta-analyses using a random-effects model using the inverse variance method to estimate the overall effect. Heterogeneity between studies was assessed by Cochran’s Q χ² test, I2 statistic and τ squared (τ2).23 The trim-and-fill method was used to estimate an adjusted overall effect in the presence of publication bias, using observed and imputed studies to generate a symmetric funnel plot.24 25 Meta-regression with random effects was used to investigate associations between study-level covariates such as publication year, study type (retrospective, prospective) and gender (both, female, male) with cancer mortality effect size. A subgroup analysis was done for study type to determine estimates of within-group effect size.Meta-regression and subgroup analyses were conducted to explore potential sources of heterogeneity, specifically assessing the impact of publication year, gender (men, women or both) and study design (prospective vs retrospective) on the association between loneliness/social isolation and cancer mortality. Mixed-effects meta-regression models were fitted using a restricted maximum-likelihood estimator for residual heterogeneity (τ²), and results were reported with regression estimates, SEs and corresponding p values. Subgroup analyses by study type were performed using random-effects models, and statistical significance for subgroup differences was evaluated using Cochran’s Q test.For studies not included in the meta-analysis, a narrative synthesis was conducted to qualitatively summarise the observed findings.Certainty of evidence We used the GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach to appraise the certainty of evidence. 26 With this procedure, evidence begins as high certainty but can be downgraded to moderate, low and very low, due to risk of bias, indirectness, imprecision, inconsistency or publication bias. ‘High certainty’ indicates high confidence that the true effect lies close to the estimate of the effect. ‘Moderate certainty’ indicates moderate confidence in the effect estimate and that the true effect is likely to be close to the estimate, though there is a possibility that it is substantially different. ‘Low certainty’ indicates confidence in the effect estimate is limited and that the true effect may be substantially different from the estimate. ‘Very low certainty’ indicates little confidence in the effect estimate and that the true effect is likely to be substantially different from the estimate of effect. We followed GRADE guidance for communication of findings and presented the overall certainty of evidence per meta-analysis in tabular format in online supplemental file 3.26SP310.1136/bmjonc-2025-000840.supp3Supplementary dataResults Our search identified 12 602 citations, of which we reviewed 148 full-text articles. Of these, 16 publications were included ( figure 1).4 8 27–40 Inter-rater agreement during the eligibility screening was (κ=0.76). The 16 publications reported on 1 635 051 patients across seven countries.Figure 1PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analysis) flow diagram of study selection.Study characteristics The baseline characteristics of the 16 studies included in this review are summarised in table 1, comprising 8 prospective and 8 retrospective cohort studies. These studies were conducted across seven countries, including Canada, England, Finland, France, Ireland, Japan and the USA, and recruitment periods spanned from the mid-1980s to 2020. Total patient sample sizes varied from 269 in smaller cohorts to over one million in large-scale national studies, with a median of 14 816 patients and a mean of 138 441. The studied populations comprised a wide spectrum of oncological diagnoses. Some studies focused on specific cancer types, such as head and neck,36 gastrointestinal cancers,37–39 lung cancer40 and breast cancer,29 34 while others included patients with all cancer types. The mean age across all studies was 63, ranging from a mean of 54–71 years and the mean proportion of female participants was 59%.Table 1Baseline characteristics of included studiesStudyCountryCohort typeYearsTotal patientsPatients analysedTypes of cancerMean ageFemale (%)Alcaraz et al27USAProspective1982–1983580 182580 182All5558Berkman et al28FranceProspective198920 62417 253All4625Canales et al29USARetrospective2007–2013171 6965207Breast75100Elovainio et al30FinlandRetrospective2000–2017137 25978 510All5946Ikeda et al31JapanProspective1993–200444 15244 152All5453Koh-Bell et al32USARetrospective1988–199417 2993360All54100Kraav et al33FinlandProspective1984–198926822570All700Kroenke et al34USAProspective1992–200232482385Breast65100Leung et al35CanadaRetrospective2011–201625 38225 382All7550Marcus et al4USARetrospective1988–1994, 2011, 1990580 00016 044All4553Mirza et al36EnglandRetrospective2002–201212 33312 333Head and neck6229Radwan et al37UKRetrospective2006–2014121120GI6454Sarma et al38USAProspective1992–2012896668GI69100Smith et al39EnglandIrelandProspective2001–200210 6137290GI7144Takemura et al40JapanProspective2018–2020269211Lung7179Zhao et al8USARetrospective2008–201833713371All6956GI, gastrointestinal.Table 2 details how social isolation and loneliness were operationalised in study, the definition of and number of patients in the exposed and reference groups and the incidence of all-cause and cancer-specific mortality. Adjusted covariates for each study are provided in a table in online supplemental file 4. Eight studies demonstrated a low risk of bias across all domains, indicating strong methodological quality (online supplemental file 2). Four studies exhibited a high risk of bias in one domain, while four are exhibiting a high risk of bias in two. According to the overall GRADE analysis, inconsistency of results, indirectness of outcomes, imprecision and publication bias are low across all studies (online supplemental file 3).SP410.1136/bmjonc-2025-000840.supp4Supplementary dataTable 2Operationalisation of loneliness and social isolation across included studiesStudyOperationalisationExposed groupReference groupPatients in exposed groupPatients in reference groupAll-cause deathsCancer-related deathsAlcaraz et al27Social Isolation Score (SIS)SIS=4SIS=03293251 16483 79829 720Berkman et al28Social Integration Index (SII)SII=ISII=IV466302925796Canales et al29Local and metropolitan statistical area measures of black isolationHigh local isolationLow local isolation1117104415361039Elovainio et al30Living aloneLiving aloneLiving with someone3202823326 89818 279Ikeda et al31Social Support Index (SSI)SSI=0SSI=5424912 840NR1561Koh-Bell et al32UCLA Social Network Index (SNI)SNI=0SNI=4487611NRNRKraav et al33Loneliness: 11-item scale; social isolation: 10-item scale; marital statusHigh isolation scoresLow isolation scoresNRNRNR283Kroenke et al34Berkman-Syme SNISNI=0SNI=4284539224107Leung et al35Psychosocial screen for cancerAny isolation indicatorNo indicators913816 244NRNRMarcus et al4UCLA SNISNI=0–1SNI=2–4230713 737NR1133Mirza et al36English Index of Multiple Deprivation (Q1–Q4)Q4–5Q1–2728250519495NRRadwan et al37Welsh Index of Multiple Deprivation (Q1–Q4)Q1Q43030NRNRSarma et al38Berkman-Syme SNISNI=1SNI=496299380167Smith et al39Townsend scoreQ4Q118231823489NRTakemura et al40Lubben Social Network ScaleHighest loneliness quartileLowest loneliness quartile73135102NRZhao et al8UCLA Loneliness ScaleSevere loneliness (score 20–33)Low/no loneliness (score 11–12)15431402686NRNR, Not reported.All-cause mortality The impact of loneliness on all-cause mortality was reported for 1 570 918 patients across 12 studies. Loneliness and social isolation were operationalised using the Social Network Index (n=3), 16 Lubben Social Network Index (n=1),41 Social Integration Index (n=3),17 UCLA Loneliness Scale (n=3),18 English Index of Multiple Deprivation (n=1),42 Social Network and Support Assessment tool used in the Epidemiological Study of the Elderly (n=1)43 and Metropolitan Statistical Area measures of local isolation (n=1).44 We found low certainty evidence according to the GRADE assessment that loneliness/social isolation is associated with all-cause mortality: HR=1.36 (95% CI=1.28 to 1.44, p<0.001) (figure 2), indicating a robust association between loneliness and social isolation on all-cause mortality (online supplemental files 3 and 4). Heterogeneity across the included studies was moderate, with I²=54.6% and χ²=19.84 (p=0.018). The adjusted estimate of the overall effect using the trim-and-fill method to account for possible small-study effects was HR=1.34 (95% CI=1.26 to 1.42, p<0.001) with measures of heterogeneity values τ2=0.004, I²=56.3% and χ²=25.15 (p=0.009), indicating moderate heterogeneity (online supplemental file 5 and figure S5a).SP510.1136/bmjonc-2025-000840.supp5Supplementary dataFigure 2Random-effects meta-analysis for all-cause mortality.Cancer mortality The impact of loneliness and social isolation on cancer mortality was reported for 1 635 051 patients in nine studies. Loneliness and social isolation were operationalised using the Social Network Index (n=4), 16 Metropolitan Statistical Area measures of local isolation (n=1),44 Social Support Index (n=1),45 Loneliness Scale (n=1)46 and Social Isolation Scale (n=2).47 We found low certainty evidence according to the GRADE assessment that loneliness/social isolation is associated with cancer mortality: HR=1.16 (95% CI=1.08 to 1.24, p<0.001) (figure 3), indicating a robust association between loneliness and social isolation on cancer mortality (online supplemental files 3 and 4). Heterogeneity measures (τ2=0.01, I²=61.7%) and the test for heterogeneity χ²=28.69, p=0.003 indicate moderate heterogeneity between the nine studies. The adjusted estimate of the overall effect using the trim-and-fill method to account for possible small-study effects was HR=1.11 (95% CI=1.02 to 1.21, p=0.013) with measures of heterogeneity values of τ2=0.015, I²=67.7% and χ²=43.39 (p<0.001) indicating substantial heterogeneity with some degree of between-study variability (online supplemental file 5 and figure S5b).Figure 3Random-effects meta-analysis for cancer mortality.Meta-regression analyses demonstrated that neither publication year (p=0.25; R²=26.6%) nor gender (p=0.50; R²=0%) significantly explained variability in the association between loneliness/social isolation and cancer mortality. However, study design was identified as a significant moderator, explaining nearly all observed variability between studies (R²=99.95%, p=0.0006), with prospective studies showing significantly smaller effect sizes compared with retrospective studies. Subgroup analysis confirmed this finding, revealing significantly higher pooled HRs in retrospective studies (HR=1.24, 95% CI=1.16 to 1.33) than in prospective studies (HR=1.10, 95% CI=1.10 to 1.10; subgroup difference p=0.0006).Three studies were not included in the meta-analysis due to different outcome metrics; hence, the results were unable to be pooled.28 37 39 Despite this, results were similar in our narrative synthesis. Berkman et al conducted a 1989–1999 prospective study of French employees at Electricity of France–Gas of France, examining the association between social integration and cancer mortality, and found that isolated men had a relative risk of 3.60 (95% CI=0.99 to 13.01) from dying due to cancer.28 Similarly, Radwan et al performed a retrospective review of patients in the UK requiring pelvic exenteration for pelvic cancer. They reported that the most socially deprived group had the lowest 5-year survival rates (53%), which was notably lower when compared with the other three quartiles.37 Lastly, Smith et al found that among patients undergoing surgery for colorectal cancer in Great Britain and Ireland, there was a significant association between social deprivation and high operative mortality, where each unit increase in deprivation was linked to a 2.9% rise in 30-day mortality (95% CI=0.5 to 5.2).39 Overall, these results further support the association between loneliness/social isolation and poorer cancer survival outcomes.Discussion Our systematic review and meta-analysis of the 16 studies, with 13 studies included in the quantitative synthesis, demonstrated that loneliness and social isolation are significantly associated with increased mortality among patients with cancer. Specifically, the estimate of the overall HR indicated a 34% increased risk of death among those experiencing loneliness or social isolation. Additionally, an estimated 11% increased risk of cancer-related death was observed among patients who are lonely or socially isolated. The three studies excluded from meta-analysis due to differing outcome metrics also consistently reported notable associations between social isolation and worse mortality outcomes. These findings collectively suggest that loneliness and social isolation may influence cancer outcomes beyond traditional biological and treatment-related factors.However, these results should be considered in light of methodological limitations. Our risk-of-bias assessment revealed substantial variability in methodological quality, with several studies exhibiting domains at high risk of bias, particularly relating to potential confounding factors and outcome measurement accuracy. Moreover, we identified moderate-to-substantial heterogeneity across the included studies. This variability partly reflects differences in how loneliness and social isolation were operationalised, depending on whether researchers aimed to assess general loneliness or cancer-specific social disconnection. For general loneliness, previous reviews recommend the UCLA Loneliness Scale (V.3) due to its strong psychometric properties and validation across diverse cancer types and demographics.48 For cancer-specific loneliness, the 7-item Cancer Loneliness Scale captures diagnosis-related isolation (eg, ‘How often does your cancer diagnosis make you feel isolated from others?’) with high reliability (α=0.94).49 Social isolation is commonly measured using the patient reported outcomes measurement information system (PROMIS-SF) V.2.0 or Berkman-Syme Network Index, which evaluate structural support networks.48 Additionally, variations in the timing of assessments across the cancer trajectory could further contribute to the observed heterogeneity, as social networks often diminish following diagnosis, potentially resulting in greater isolation among long-term survivors compared with newly diagnosed patients.50 51Our subgroup and meta-regression analyses sought to further clarify potential sources of the substantial heterogeneity observed in cancer-related mortality, examining publication year, gender and study design (prospective vs retrospective). While neither publication year nor gender significantly explained variability in effect sizes, study design emerged as a critical moderator, explaining nearly all between-study variability (R²=99.95%) for cancer-related mortality. Retrospective studies demonstrated significantly larger HRs compared with prospective studies, indicating a potential overestimation due to biases inherent in retrospective data collection, such as recall bias or differential ascertainment of exposure status.Given these considerations, the evidence regarding the impact of loneliness and social isolation on cancer-specific mortality in particular warrants cautious interpretation. Future research should prioritise high-quality prospective studies with standardised, validated measurement instruments and clearly defined assessment timing to provide more precise estimates of their impact.Despite these limitations, our findings are consistent with prior research linking psychosocial stressors to adverse health outcomes. Social isolation and loneliness are thought to increase mortality risk in patients with cancer through interconnected biological, psychological and behavioural mechanisms. Biologically, the stress response triggered by loneliness may lead to immune dysregulation and heightened inflammatory activity, ultimately contributing to disease progression.14 52 For example, several studies have shown that social isolation is characterised by a state of elevated inflammatory markers such as C-reactive protein and interleukin-6, which are associated with cancer risk and progression.53 54 In addition, a recent proteomic analysis of 42 062 UK Biobank participants identified five key plasma proteins mediating a relationship between loneliness and inflammation, cardiovascular disease, stroke and mortality.55 These findings highlight how social relationships may influence health outcomes through underlying biological mechanisms. Psychosocially, the unique burden of cancer survivorship often includes forms of isolation stemming directly from disease and treatment experiences, including the inability of loved ones to fully understand cancer-associated fears, stigma around visible treatment effects and survivorship-related anxieties.56 Treatment-induced physical changes (fatigue, cognitive impairments) may further limit social participation, while prolonged medicalisation of life can erode preillness identity and community connections.56 Socially disconnected individuals also lack practical support for managing symptoms or attending appointments, which is often exacerbated by lower economic status and can lead to a higher risk of adverse health outcomes.57 It is also important to note that loneliness is an established risk factor for psychiatric disorders, such as depression, anxiety and psychosis, which may also contribute to higher rates of mortality in our observed trends.58–62 At least 30–35% of patients with cancer experience psychiatric disorders, and an additional 15–20% face distressing meta-diagnostic conditions like demoralisation and existential distress.62 63 Taken together, the findings from our review reinforce the plausibility of a multifactorial pathway through which loneliness and social isolation exacerbates mortality risk in patients with cancer.The medical significance of isolation came to the forefront during the COVID-19 pandemic, where multiple studies documented worsening loneliness in the cancer population: for example, a 2021 study found that over half of 606 included oncology patients experienced loneliness, surpassing prepandemic baselines, while over one-third of patients reported acute isolation.7 These trends were linked to pandemic-related social distancing, disrupted care and heightened psychological distress.7 64 Postpandemic, there has been increased utilisation of virtual care in many jurisdictions,65 66 which is known to improve healthcare access but simultaneously reduce opportunities for in-person support.67 68 Thus, continued attention must be paid to ensure that these new paradigms of care do not exacerbate loneliness or social isolation in this vulnerable population.Although there is greater emphasis on incorporating psychosocial supports in oncology care, it continues to be under-recognised in patients with cancer.53 Despite emerging evidence for interventions that may mitigate loneliness among this population, evaluations of their effectiveness remain limited.69 Approaches that have demonstrated feasibility and acceptability include telephone or internet-based support groups, individual counselling and stress reduction programmes, particularly when incorporating cultural modifications to enhance relevance for diverse populations.69 For instance, interventions integrating patient education, coping strategies and community resource navigation show promise in addressing the social constraints and disconnectedness that exacerbate cancer-related loneliness.69–71 However, a 2020 systematic review of possible interventions identified only eight studies meeting inclusion criteria, most of which were lower-quality trials, underscoring a critical gap in robust intervention research.69 Notably, interventions tailored to specific demographic factors—such as age, gender or ethnicity—are nearly absent despite their potential to improve engagement and outcomes.69 While preliminary data suggest technology and/or non-technology-based strategies may reduce mortality risks linked to severe loneliness, large-scale randomised controlled trials are needed to establish efficacy, optimal delivery methods and long-term impacts on both psychosocial well-being and survival outcomes.69 72 73 Despite the lack of robust evidence in interventions helping to ameliorate the negative effects of social isolation, we believe our review supports that loneliness assessments should be included in standard oncology practice given the magnitude of this modifiable risk factor for mortality.A strength of our review is the comprehensive search strategy, which identified 16 studies that met the eligibility criteria with patients from multiple countries and various cancer diagnoses, contributing to the generalisability of our findings. Moreover, the evaluation of risk of bias supports a moderate quality of included reviews. Most studies used well-established social isolation and loneliness indices, ensuring reliable assessment of exposure variables. Nevertheless, several limitations must be addressed. First, the observational nature of the included studies limits our ability to infer causality, where it is possible loneliness and isolation is a surrogate for other risk factors. Social isolation and loneliness are correlated with other risk factors in older adults, including socioeconomic status and psychiatric disorders.74–76 However, the HR values were adjusted to account for other prognostic factors in most of the included studies, including age, sex, comorbidities and smoking status, which support a more causal relationship.4 8 27–29 34 36–38 40 Second, our search was restricted to English-language publications. This language restriction may have introduced publication bias, as studies with negative or null results conducted by non-English-speaking investigators may be less likely to be published in English-language journals. Sociocultural factors unique to non-English-speaking populations may be linked to unique experiences of cancer and loneliness, which is an area of further exploration in future studies.Consequently, while our results demonstrate a statistically significant association between loneliness/social isolation and all-cause and cancer-specific mortality, the effect sizes are modest and the lower bounds of the confidence intervals are near the null. Residual confounding, potential biases—including publication bias due to language restrictions—and methodological differences across studies may have attenuated or inflated the observed associations. and not viewed as definitive evidence of a strong causal relationship. Future research would benefit from consistent operational definitions, validated tools and standardised methodologies to improve comparability and external validity.Conclusions Our systematic review and meta-analysis provide evidence that loneliness and social isolation are associated with increased risks of both all‐cause and cancer-specific mortality in patients with cancer. However, given the modest effect sizes, methodological heterogeneity and overall low certainty of evidence, these findings should be interpreted with caution. Future research should focus on high-quality prospective studies using standardised and validated measures to better elucidate the causal pathways and inform the development of targeted psychosocial interventions in cancer care.",
  "title": "Impact of loneliness on cancer mortality: a systematic review and meta-analysis",
  "uid": "a69b6db2-f4c1-56c2-9a02-d670144c8bf8"
}