Table 3 Summary of included studies

Daugaard et al. (2019) [40]

To examine the association between social and socioeconomic position and drug reimbursement due to terminal illness (DRTI) registration among patients who are terminally ill in Denmark

Population-based (Denmark)

Terminally ill patients who died 2006 – 2015 with cancer or seven other chronic illnesses (n = 307,188)

Quantitative

Database analysis

-No association between education and DRTI

-Higher prevalence of DRTI in middle/high income patients, however palliative care needs may be under-recognised in low income patients

-Lower prevalence of DRTI in employed patients

11

Ankuda et al. (2018) [41]

To identify what services are critical and why they matter to patients in a home-based palliative program

Hospice – home based program (US)

Enrollees in the program (n = 18)

Mixed

Individual interviews and survey

-Participants living in poverty had more non-routine and social work visits and found the social services provided by a home hospice service to be the most critical aspect of care

9

Azhar et al. (2018) [42]

To compare time of referral and characteristics (level of symptom distress) among uninsured (indigent), low-insured (Medicaid), and insured patients presenting to the supportive care center

Outpatient center (US)

Patients with advanced cancer (500)

Quantitative

Survey to participants

-Uninsured and low/partially insured patients had higher ESAS pain scores

-Indigent patients had:

-More encounters with palliative care team per month

-Higher levels of distress

-Higher risk for aberrant opioid usage and

-More frequently on opioids at time of referral to centre

-Insurance status did not impact timing of palliative care referral or follow-up

10

Brown et al. (2018) [43]

To examine the association between race/ethnicity, other key social determinants of health, and healthcare intensity in the last 30 days of life for those with chronic, life-limiting illness

Population-based (US)

All patients who died from 1 of 9 chronic conditions between January 2010 and December 2015 (n = 22,068)

Quantitative

Database analysis

-Lower income was associated with higher likelihood of receiving high intensity care (i.e., treatment in an intensive care unit, receipt of mechanical ventilation, receipt of cardiopulmonary resuscitation) in the last 30 days of life

11

Cervantez et al. (2018) [44]

To evaluate how insurance coverage and ethnicity impact distribution of symptom burden and how those factors influence palliative access for patients at a South Texas cancer center

Outpatient center (US)

Patients who attended clinics (n = 604)

Quantitative

Survey to participants

-Symptom burden was similar in insured and uninsured patients, but there was a four-fold delay in first palliative care visit in uninsured group

8

Huang et al. (2015) [45]

To test the hypothesis that men and those with lower levels of educational attainment would have less favourable attitudes towards palliative care

Community and hospital-based cancer clinics (US)

Patients with cancer (n = 383)

Quantitative

Secondary analysis of pre-existing VOICES dataset

-Most patients would desire palliative care if no further anti-cancer treatment would be helpful

-No difference in preferences between those with different levels of education

11

Cai et al. (2017) [46]

To examine the socioeconomic differences in the propensity and intensity of health service use for the three main home-based services (physician visits, nurse visits and personal support worker hours), and to explore the determinants of the use of home-based palliative care services

Home-based palliative care program (Canada)

Adult primary caregivers of patients who were diagnosed with malignant neoplasm (n = 181)

Quantitative

Survey administered to participants

-Patients with lower socioeconomic status were more likely to receive at least one home-based nurse visit or personal support worker (PSW) visit

-Lower socioeconomic status was associated with higher intensity of health service use

-Socioeconomic status is a significant predictor of propensity and intensity of home-based nurse and PSW visit even after controlling for health status

-Patients with higher levels of education were less likely to receive at least one home-based PSW visit

11

Koroukian et al. (2017) [47]

To gain a better understanding of the correlates of aggressive end-of-life care and hospice use in older adults dying from cancer, by accounting for both social determinants of health and multimorbidity

Population-based (US)

Participants < 66 years who died from cancer with complete Medicare claims history (n = 835)

Quantitative

Database analysis

-Percentage of patients receiving cancer-directed treatment increased with higher education and income

-No significant difference by income or education for admission to hospital

-Enrolment in hospice was generally higher with higher levels of education or income

-There was a U-shaped distribution between in-hospital death and income and education

10

Krishnan et al. (2017) [48]

To examine the family and facility factors that may influence the decision to transfer nursing home residents to hospital in the last month of life

Nursing homes (Canada)

Adult relatives of residents who had died in the past year and was self-reported as the most involved in their care in their last month of life (n = 119)

Quantitative

Survey to participants

-There was a higher prevalence of hospital transfers among deceased whose family members had lower income

-Low income family members were 3 times more likely to send patient to hospital

-No statistically significant relationship was noted between employment or education level of the family members and terminal hospital transfers

11

Carlucci et al. (2016) [49]

To elicit the patients’ end-of-life preferences in very severe chronic obstructive pulmonary disease

Hospital (Italy)

Patients with very severe COPD (n = 43)

Mixed

Individual interview

Survey to participants

-Patients with lower education were more likely to choose endotracheal tube intubation

-Higher education level is the strongest predictor of probability of choosing palliative care option with oxygen and morphine

9

Dhingra et al. (2016) [50]

To evaluate a diverse population served by an interdisciplinary model of community-based specialist palliative care and the variation in service delivery over time and identify subgroups with distinct illness burden profiles

Community palliative care program (US)

Patients referred to the program (n = 894)

Quantitative

Survey to participants

-Patients who reported very low illness burden more likely to reside in impoverished neighbourhoods

11

Karikari-Martin et al. (2016) [51]

To examine whether differences in hospice use and use of aggressive services in the last 12 months of life are influenced by race or any cancer diagnosis while adjusting for the covariates age, gender, marital status, education level, neighbourhood socioeconomic status, cognitive function, and physical function

A geographically-defined community in Chicago (US)

Participants < 65 years enrolled Medicare for at least 1 year before death (n = 2954)

Quantitative

Secondary analysis of pre-existing CHAP data

-Higher neighbourhood level socioeconomic status and higher income each significantly increased the likelihood of using hospice

-Race has a more powerful effect on hospice use than socioeconomic status neighbourhood

-Higher education was associated with a significantly decreased risk of hospitalization at the end of life

10

Khosla et al. (2016) [52]

To investigate the association between socioeconomic status (i.e., education and household income) and anticipatory care planning (ACP) over time using data from the Health and Retirement Study (HRS), a nationally representative survey of middle-aged and older adults living in the contiguous United States

Health and Retirement Study (HRS) (US)

Spouses, partners or proxy informants of HRS respondents (n = 6052)

Quantitative

Survey to participants

-Very limited support was found for the hypothesis that socioeconomic status would predict ACP

-Higher household income increased the odds of having a legally designated power of attorney for health care but had no impact on end of life discussions or written instructions

-Education level did not predict engagement in ACP behaviours

10

Lee et al. (2016) [53]

To understand the association of minority race/ethnicity and education with family ratings of the quality of dying and death

Hospital Intensive Care Units (US)

Families of patients who died (n = 1290)

Quantitative

Survey to participants

-Education was not significantly associated with ratings on the quality of death questionnaire for either patient or family models

11

Schou-Andersen et al. (2016) [54]

To investigate whether demographic and socio-economic factors were associated with preference for dying at home

Community (Denmark)

Relatives of adults who died from cancer (n = 282)

Quantitative

Survey to participants, database analysis

-At the beginning of palliative period, low income patients more likely to prefer dying at home than high income patients, however medium income patients less likely

-At the end of the palliative period, no association between preference for dying at home and income

11

Huang et al. (2015) [45]

To examine the association of individual income and end-of-life care in older cancer decedents in Taiwan

Population-based (Taiwan)

Adults aged over 65 years with cancer (n = 28,978)

Quantitative

Database analysis

-Low income was associated with more aggressive end of life care

-Older cancer decedents with low income were more likely to stay in hospital < 14 days and die in acute hospital

-Older cancer decedents with moderate/high income were more likely to have visited the emergency department and admitted to intensive care more frequently

10

Neergaard et al. (2015) [55]

To analyse associations between GP contacts at the end of life and socioeconomic and cultural characteristics of Danish cancer patients

Community (Denmark)

Relatives of adults who died from cancer (n = 584)

Quantitative

Secondary analysis of pre-existing dataset

-GP face-to-face appointments were higher for decedents with cancer who had low income than those with normal/high income

11

Tang et al. (2015) [56]

To describe longitudinal changes in post-traumatic growth (PTG) during the dying process and to identify determinants of PTG among terminally ill cancer patients

Hospital inpatient units (Taiwan)

Terminally ill cancer patients (n = 313)

Quantitative

Survey to participants

-Patients with at least senior high school education achieved higher post-traumatic growth scores than those with low education

11

Tucker-Seeley et al. (2015) [57]

To investigate the association between financial hardship and intensive care in the last week of life

Cancer Centers, as part of the Coping with Cancer (CwC) study (US)

Caregivers of deceased cohort (n = 281)

Quantitative

Secondary analysis of pre-existing CwC data

-Patients who reported financial hardship had higher odds of receiving intensive end of life care

-No association between educational attainment or health insurance status and aggressive end of life care

-When treatment preferences were included in the fully adjusted model, the association between financial hardship and intensive EOLC was slightly attenuated but remained statistically significant

11

Bhatraju et al. (2014) [58]

To evaluate the factors that were associated with palliative care consultation (PCC) utilization in patients who died in an urban municipal public hospital in the United States, and to examine the association of PCC utilization with symptom management and advance directives at the end of life

Hospital – Inpatients (US)

Patients who died in the units (n = 378)

Quantitative

Retrospective chart review

-Higher level of education was associated with palliative care consultation (PCC) utilisation

-Patients who had PCC were more likely to receive opiates in final 72 h of life, but there was no observed difference in benzodiazepine use

9

Chang et al. (2014) [59]

To assess the association between aggressiveness of end-of-life care and socioeconomic status in working-age terminal cancer patients in Taiwan between 2009 and 2011

Population-based (Taiwan)

Adult cancer patients (n = 32,800)

Quantitative

Database analysis

-High and moderate socioeconomic status patients had:

-Lower scores for aggressive end of life care

-Less chemotherapy

-Fewer emergency department visits

-Less intensive care unit admission

-Lower rates of dying in acute care hospitals

11

Lewis et al. (2014) [60]

To explore the nature of social capital in a socioeconomically disadvantaged group of palliative care patients and carers, using a social capital questionnaire to guide and frame discussions

Community (Australia)

Patients who were known to the palliative care service (n = 22)

Qualitative

Individual interviews

-Overall provision of informal care generally by sole caregiver with intermittent family support

-Established neighbour and inter-sectoral networks essential for sustaining care and social needs

-Did not feel very engaged with community

-Limited communication due to cultural differences

-Formalised community care support described as overall being somewhat inconsistent and unpredictable

-Patients and carers unsure of formal care being provided

-Contact with government agencies (welfare support and government housing) described as positive generally

11

Tang et al. (2014) [61]

To investigate the associations between accurate prognostic understanding and

end-of-life care preferences, and to identify correlates of accurate prognostic understanding among terminally ill cancer patients

Hospital (Taiwan)

Terminally ill cancer patients (n = 2452)

Quantitative

Survey to participants

-Patients with at least senior high school education were 1.28 times more likely to accurately know prognosis

-Accurate prognostic understanding associated with greater odds of preferring comfort-oriented care and increased preference for hospice care

11

Chang et al. (2013) [62]

To describe how much burden terminal cancer patients and their caregivers had experienced, what support they most needed, and the differences between them

Hospital (South Korea)

Patients with terminal cancer (n = 481) and their caregivers (n = 381)

Quantitative

Survey to participants

-Low level of education predicted satisfaction about overall care

11

Masucci et al. (2013) [63]

To examine the predictors of the propensity and intensity of five main health service categories in the last three months of life for home-based palliative-care patients

Community. Home-based (Canada)

Family caregivers of patients with malignant neoplasm (n = 109)

Quantitative

Individual interviews, database analysis

-Patients in the highest deprivation group had higher intensity of home-based nurse visits

11

Sahin et al. (2013) [64]

To evaluate the relationship between different demographic variables and hopelessness, depression and social support in end of life Turkish cancer patients

Hospital (Turkey)

Patients with cancer (n = 216)

Quantitative

Survey to participants

-Patients with lower education had higher hopelessness scores, but not depression or social support

11

Bergman et al. (2010) [65]

To assess the quality of end of life care in low income uninsured men prospectively enrolled in a specific public assistance program

Population-based (US)

All low-income, uninsured men in the program (n = 60)

Quantitative

Retrospective chart review, database analysis

-No patients had chemotherapy initiated within 3 months of death, and only 6% had chemo within 2 weeks of death

-Use of hospital resources (emergency department visits, inpatient admissions, intensive care) was uniformly low

-Hospice utilisation was comparable and timing of referral was better than population

8

Fergus et al. (2010) [66]

To identify key challenges and improvements to out-of-hours palliative care in a mixed urban and rural deprived area

Community (United Kingdom)

Mixed; Patients (n = 6), carers (n = 1) and healthcare professionals [29]

Mixed

Individual interviews, database analysis, observation

-Patients voiced reluctance to use Out of hours primary care phoneline due to:

-Stressful and cumbersome process of making initial contact

-Bad experiences previously

-Misunderstanding of its function

-Reluctance to speak to an unknown person

-Had bad experiences previously

-Did not realise had ‘special notes’ so assumed would be too complex

11

Maric et al. (2010) [67]

To estimate the frequency of anxiety and depression symptoms in patients with advanced lung cancer, and the associations of these with demographic, socioeconomic and clinical factors

Hospital (Serbia)

Patients with stage 3B and 4 non-small cell lung cancer (n = 100)

Quantitative

Survey to participants

-No difference in levels of psychological distress between education groups or rural versus urban groups

-However, unemployed patients had significantly less anxiety and depressive symptoms

9

Chochinov et al. (2009) [68]

To use the Patient Dignity Inventory (PDI), a novel, reliable and validated measure of end-of-life distress, to describe a broad landscape of distress in patients who are terminally ill

Population-based (Canada)

Patients in this program who received palliative care (n = 253)

Quantitative

Survey to participants

-Patients who were more educated were significantly more likely to report feelings of having lost control, feelings of unfinished business and not being able to perform tasks of daily living

10

Carr et al. (2016) [69]

To explore the extent to which socioeconomic status indicators are associated with attributes considered essential to the quality of one's death

Population-based (US)

Mixed; Bereaved spouses (n = 408), patients (n = 5276)

Quantitative

Secondary analysis of pre-existing datasets

-Limited evidence that socioeconomic status affects death quality, except in pain at the end of life

-Wealthier participants were less likely to report severe pain at the end of life, and are more likely to engage in advanced care planning

10

Adler et al. (2019) [70]

To explore the lived experiences of medically underserved women with advanced breast cancer

Community – clinic (US)

Patients with breast cancer and annual family income below 200% of the federal poverty level (n = 63)

Qualitative

Individual interviews

-Consistent issue of financial distress at the end of life, including difficulty meeting basic needs, inadequate financial resources, inability to work

-Patients had concerns about not being “useful” when unemployed and being financially burdensome

-Perceived bias in healthcare – some patients felt they were not treated as well by healthcare professionals

-Patients viewed fewer financial means, lack of access to private insurance and cancer centres, and minority status as barriers to high quality care

-They remained highly attentive to others’ needs and worked to maintain caretaking roles despite own struggles. They made effort to attend to practical matters

-Many downplayed desires to discuss dying due to pressure of staying positive

-They appreciated meaningful aspects of life (social relationships, creative outlets, spirituality) and felt altruism was important before and during illness

10

Bijnsdorp et al. (2019) [71]

To identify different types of home-based care networks of community dwelling older adults in the Netherlands, and to assess the association between the different types of home based care networks and the health status and sociodemographic characteristics of care recipients

Community (Netherlands)

Respondents who died within 12 months of their last interview, lived at home, and received personal and/or household care (n = 146)

Quantitative

Survey to participants

-Those with higher education were more likely to have informal care provided by partners whereas those with lower educational attainment more likely to require formal care

11

Saphire et al. (2020) [72]

To examine patterns of symptom management at end of life for older adults who died of lung cancer

Population-based (US)

Individuals who had at least one day in the outpatient non-hospice setting during the last month before death (n = 16,246),

Quantitative

Database analysis

-High poverty levels were associated with increased receipt of pain, dyspnoea and emotional distress medication

10

Yi et al. (2020) [73]

To compare health and social care costs, quality and their drivers in the last 3 months of life for older adults across countries

Hospital and Community (UK, Ireland, US)

Carers of patients who had accessed a participating palliative care team (n = 767)

Quantitative

Survey to participants

-Only 4% reported difficult/very difficult financial circumstances, 10% reported just about alright

-Higher care costs if “difficulty living on current income”

-Hospital care accounts for over 80% total health and social costs, community and palliative costs were low

11

Jacob et al. (2019) [74]

To assess health related quality of life (HRQoL) of advanced cancer patients in terms of general wellbeing (physical, functional, emotional, and social/family wellbeing), pain experiences, psychological state, and spiritual wellbeing, and determine the relationship between belonging to a disadvantaged group and HRQoL outcomes

Hospital (India)

Patients with cancer (n = 210)

Quantitative

Survey to participants

-Patients with higher financial difficulty scores reported lower functional wellbeing, lower emotional wellbeing, lower meaning/peace (subscale of spiritual wellbeing), and higher anxiety and depression scores. Level of education did not affect this

-Even in setting where cancer treatment is free, those with higher financial difficulty scores report lower quality of life outcomes

-Scores of this population generally lower than patients in higher income countries e.g., USA

11

Leng et al. (2019) [75]

To explore the prevalence, determinants and consequences of catastrophic health expenditure (CHE) among urban and rural end-of-life cancer patients in China

Community (China)

Families of deceased cancer patients (n = 792)

Quantitative

Face to face interviews

-There were very high levels of catastrophic healthcare expenditure (CHE) at end of life throughout

-1/3 of patients borrowed money from family and friends

-Households with higher income were less likely to incur CHE

-Rural, low income patients were least likely to access healthcare, inpatient care, and most likely to only use outpatient care. Use of inpatient and outpatient services in these patients increased the risk of CHE

-Rural patients allocated higher proportion of household income to healthcare and associated expenditures

11

Saeed et al. (2019) [76]

To study the effect of income and education on the completion of advanced directives

Community and hospital (US)

Patients with cancer (n = 265)

Quantitative

Secondary analysis of pre-existing VOICE dataset

-Nearly a third of lower income patients had not completed advanced directives (AD)

-Patients with lower income had lower AD scores, but there was no significant association between education or perceived financial strain and AD scores

11

Stajduhar et al. (2019) [77]

To identify barriers to assessing care among structurally vulnerable people at the end of life

Community (Canada)

Mixed; People experiencing structural vulnerability in a Western Canadian province (n = 25), their support persons (n = 25) and formal service providers (n = 69)

Qualitative

Repeated participant observation, individual interviews

-Structural vulnerability was defined in this study as people living in poverty, in unstable housing, while experiencing various forms of oppression and stigma, e.g., racism, ongoing or past trauma and violence, social isolation, mental and cognitive issues, behavioural issues, substance use, interactions with the criminal justice system, or physical disability

-These patients were found to have significant barriers in having their palliative needs met including:

-Survival imperative – focussing on basic care needs;

-Normalization of dying – often had been told they would die due to addictions

-Problem of identification – often not signposted to appropriate health, social benefits

-Professional risk and safety management – due to stigmatization from health care professionals

-Cracks of a silo-ed care system – complexity of the care systems made them difficult to navigate

10

Wales et al. (2020) [78]

To examine the association between socioeconomic status and other demographic factors on place of death in a population receiving home palliative care in Toronto, Canada

Community—Home palliative care service (Canada)

Patients who died with home-based palliative care service (n = 2066)

Quantitative

Retrospective chart review

-Rate of preference for home death was significantly lower in lowest income quintile

11