Disparities in pediatric cancer survivorship care: A systematic review

Abstract Background Childhood cancer survivors (CCS) experience many long‐term health problems that can be mitigated with recommended survivorship care. However, many CCS do not have access to survivorship care nor receive recommended survivorship care. We reviewed the empirical evidence of disparities in survivorship care for CCS. Methods This systematic review searched PubMed, CINAHL, and PsycINFO for studies on survivorship care for CCS (PROSPERO: CRD42021227965) and abstracted the reported presence or absence of disparities in care. We screened 7945 citations, and of those, we reviewed 2760 publications at full text. Results A total of 22 studies reported in 61 publications met inclusion criteria. Potential disparities by cancer treatment (N = 14), diagnosis (N = 13), sex (N = 13), and current age (N = 13) were frequently studied. There was high quality of evidence (QOE) of survivorship care disparities associated with non‐White race, Hispanic ethnicity, and being uninsured. Moderate QOE demonstrated disparities among CCS who were unemployed and older. Lower QOE was found for disparities based on cancer diagnosis, cancer treatment, age at diagnosis, time since diagnosis, sex, insurance type, income, educational attainment, and geographic area. Conclusions We found strong empirical evidence of disparities in survivorship care for CCS associated with race, ethnicity, and insurance status. Multiple other disparate groups, such as those by employment, income, insurance type, education, cancer diagnosis, age at diagnosis, time since diagnosis, cancer treatment, geographic area, sex, and self‐identified gender warrant further investigation. Prospective, multilevel research is needed to examine the role of other patient characteristics as potential disparities hindering adequate survivorship care in CCS.

disparities research; health services research and access to care for populations that experience disparities; and administrative and payer perspectives.The key informants reviewed the key question, our definition of disparities, and possible sources of care disparities in CCS.This work is part of a larger project on disparities and barriers to care in CCS and interventions to support CCS. 14

| Sources and searches
A librarian specializing in systematic reviews planned, completed, and documented the search strategy shown in Appendix S1.We searched the research databases PubMed, CINAHL, and PsycINFO for published research on survivorship care.We identified studies through two search strategies: (1) we identified publications that focused on disparities directly (e.g., addressed disparities in the title, abstract, or key word), and (2) we searched for empirical research studies on CCS patients (i.e., which did not address disparities in the title, abstract, or key word).We reference-mined identified literature, and subject matter experts provided additional input to confirm that relevant studies were included.

| Screening and abstraction
Table A-1 in the Appendix S1 displays the framework of PICOTSS (population, intervention, or independent variables, comparison groups, outcomes, timing, study setting, and study design, or other limiters).Study populations eligible for inclusion were required to have at least 50% CCS.We defined CCS as those diagnosed before age 21, who had received primary acute treatment for any cancer, were currently in remission, and currently receiving or eligible to receive survivorship care services, care plans, and/or models of follow-up care at time of study.We conducted both the literature screening and data abstraction steps in a database designed for systematic reviews using a web-based interface.Literature reviewers first screened all identified citations, and those determined to be potentially relevant by at least one reviewer were obtained and screened at the full-text level.Citations that were excluded were later screened with a machine learning algorithm to avoid missing any potentially relevant citations.Two reviewers screened full text publications independently with the prespecified eligibility criteria (i.e., PICOTSS), and we resolved any inconsistent decisions by group discussion and consensus at a team meeting.
We used the definition of populations or groups that experience health disparities endorsed by the National Institute of Minority Health and Health Disparities and the National Academies: racial/ethnic minorities (including those who are Black/African American, Hispanic/ Latino, American Indian/Alaska Native, Asian American, Native Hawaiian, and other Pacific Islander); socioeconomic status; underserved urban or rural areas; sexual and gender minorities; and educational attainment. 7,8We also used disparity categories suggested by key informants as additional potential sources of disparities, such as employment, age at study, age at diagnosis, time since diagnosis, insurance status, cancer diagnosis (e.g., type, grade, stage, relapse), and cancer treatment (e.g., type, intensity).We included all studies that evaluated potential disparities regardless of the result of the evaluation, meaning that we included all studies reporting on the presence or the absence of disparities in care in studies that addressed patient characteristics potentially associated with disparities.
For abstraction of data from included studies, one reviewer abstracted applicable data and an experienced content expert checked the data for accuracy.We considered all available publications relevant to a study (the included participants define a study) and consolidated information in the data abstraction form.This avoided counting the same study multiple times and instead provided a complete record of the results of the overarching, most pertinent data regarding disparities.In addition to abstracting results on disparities, we also recorded the study design, country of origin, study participant characteristics (i.e., cancer type, proportion of CCS compared to adult survivors), and type of analysis and outcome(s) assessed.We abstracted the results of assessments of potential disparities and documented reported effect size estimates, the direction of effects, and the statistical significance of results, documenting the presence or the absence of disparities reported in the study.

| Critical appraisal and synthesis
To assess risk of bias in individual studies, we used the Quality In Prognostic Studies (QUIPS) tool for prognostic factor research addressing six domains: study participation, study attrition, prognostic factor measurement, outcome measurement, study confounding, and statistical analysis and reporting. 15Comprehensive evidence and risk of bias tables document the presence and absence of disparities for each content category and report the risk of bias for the included studies.
We documented findings across studies in a Summary of Findings table.We assessed the quality of evidence (QOE) across studies using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework for prognostic research. 16In brief, although based exclusively on observational research designs, we did not use low QOE as the starting point to avoid floor effects and instead started at high and downgraded where findings were exclusively based on retrospective rather than prospective studies for study limitations.In addition, we downgraded for inconsistency of results across studies, indirectness of outcome measures, imprecision of estimates, and potential reporting bias.The quality could be upgraded for large effects, exposure-response gradient, and where plausible confounding would mask an association.We differentiated high, moderate, low, and very low evidence corresponding to the level of uncertainty associated with the summary evidence statement.

| RESULTS
We screened 7945 citations for relevance, we screened 2760 publications as full text against the prespecified eligibility criteria, and we identified 22 studies (as reported in 61 individual publications) examining disparities in survivorship care for CCS.  Figure 1 presents the PRISMA flow diagram that visually depicts the number of studies that were both included and excluded at each stage of the review process and the reasons for exclusion.

| Study characteristics
The identified studies described CCS in the United States, United Kingdom, Canada, Switzerland, and Netherlands.The most frequently studied areas of potential disparities (see Table 1) were cancer treatment (N = 14), cancer diagnosis (N = 13), biological sex (N = 13), current age (N = 13), insurance status (N = 11), age at diagnosis (N = 10), and race or ethnicity (N = 8).Less frequently studied potential disparity sources were geographic area (N = 7), income (N = 6), insurance type (N = 6), employment (N = 5), and education (N = 2).Figure 2 depicts assessed disparity categories mapped to survivorship care disparities in the included studies.Figure 3 displays the risk of bias in the included studies.In the Appendix S1, Table A-2 provides the evidence table for included studies and Table A-3 displays the risk of bias assessment.All studies had 100% CCS in their evaluated sample.

| Results of syntheses
The summary of evidence and the resulting GRADE category are documented in Table 1.The table is organized by the disparity categories and documents the presence and absence of disparities in all studies assessing the potential disparity source. 3,34,36,38 Of those 11 studies, similar results were found across 9 studies, such that, being uninsured was associated with a disparity in survivorship care (high QOE). 17,21,22,26,29,31,34,36,38For example, a prospective study provided evidence that those who were currently insured experienced a decrease in the odds of difficulty obtaining care (OR 0.18; CI 0.12, 0.26) compared to those who were uninsured, after adjusting T A B L E 1 Summary of findings.Berg, 2016 21 Berkman, 2019 22 Daly, 2019 23 Gardner, 2014 24 May, 2017 26 Milam, 2015 29 Oeffinger, 2004 31 Szalda, 2016 34 Welch, 2017 36 Zheng, 2016 38 Favors those who are insured: Those who were currently insured experienced a decrease in the odds of difficulty obtaining care (OR 0.18; CI 0.12, 0.26) compared to those who were uninsured, after adjusting for sex, age at survey, marriage, employment status, age at diagnosis, cancer diagnosis, and radiation therapy. 17 multivariable analysis, lack of health insurance was associated with less than annual healthcare provider visits (OR 0.04; CI 0.01, 0.33), after adjusting for current age, sex, and chemotherapy exposure. 21 a multivariable analysis, participants with self-pay status or uninsured were significantly more likely to have missed visits (+8.46%, p = 0.005) when adjusting for tumor grade and race.22   In multivariable analyses and in comparison with those with insurance, those who were uninsured (OR 3.4; CI 1.2, 9.2) were more likely to be lost to follow-up for more than 1000 days and not attend survivorship clinic, when adjusting for diagnosis, treatment exposure, SCT exposure, age group at diagnosis, and race (global overall p-value = 0.02). 26rvivors who had any health insurance were more likely (OR 3.40; CI 1.10, 8.41) to access survivorship care in comparison with those who were uninsured, after adjusting for sex, race/ethnicity, treatment intensity, family influence on healthcare decisions, having a regular doctor, health care self-efficacy, and post-traumatic growth. 29ose who were uninsured (in comparison with those who were privately covered) were less likely to have survivorship care (OR 0. and English-speaking non-Hispanic parent. 29,52 the multivariable model and in comparison with survivors who were not insured, those who were insured (OR 2.06; CI 1.28, 3.32) were more likely to access survivorship care, after adjusting for years since diagnosis, current age, sex, race and ethnicity, socioeconomic status, high levels of depressive symptoms, number of late effects, treatment intensity, received a written treatment summary, reported having a doctor for regular (non-cancer) care, discussed cancer-related follow-up care with a doctor in the last 2 years, knowledge of the need for life-long survivorship care, health care self-efficacy, and family influence of health care decisions. 29,46 comparison with survivors who were insured, those who were uninsured experienced an increased odds (OR 1.7; CI 1.3, 2.2) of having general care (rather than risk-based survivorship care), after adjusting for sex, race/ethnicity, current age, age at interview, annual household income, educational attainment, employment status, poor emotional health, cancer-related anxiety, cancer-related pain, poor physical health, and grade of chronic disease. 31,58 univariate analysis, having insurance was the only significant characteristic associated with engagement in adult-oriented follow-up care (p < 0.005); no survivors who were uninisured had follow-up care.Does not favor those who were insured: In univariable analysis, use of a mental health professional was less likely to be associated with private insurance coverage, in comparison with those with public insurance or uninsured (p = 0.040).24   Does not favor those who experienced a change in their insurance coverage: In the multivariable model, survivors who experienced a change in insurance coverage from diagnosis to survivorship resulting in a gain in coverage (b: −0.05, SE: 0.02) or a loss in coverage (b: −0.15; SE: 0.04) had a decrease in the predicted probability of having a survivorship care visit in the prior 2 years; however, those covered by Medicaid and/or Medicare at diagnosis had an increase in the probability of having a survivorship care visit in the prior 2 years (b: 0.05; SE: 0.02); all models were adjusted for current age, sex, race and ethnicity, marital status, educational attainment, receipt of supplemental income, insurance at diagnosis, insurance understanding, socioeconomic status, treatment intensity, children, employment status, difficulty with a referral to a specialist, ability to see a doctor when needed, friend/family influence of health decisions, cancer diagnosis, and self-rated health. 29,46ere was not a significant relationship for who were uninsured in predicting an Baedke, 2022 18 Barakat, 2012 19 Berkman, 2019 22 Daly, 2019 23 May, 2017 26 Milam, 2015 29 Oeffinger, 2004 31 Welch, 2017 36 Zheng, 2016  Does not favor those of non-Hispanic White race and ethnicity: Uninsured non-Hispanic White survivors were less likely to report survivor focused-care for a cancer-related visit (OR 0.82; CI 0.74, 0.90) and a cancer center visit (OR 0.79; CI 0.66, 0.94) compared to privately insured non-Hispanic White survivors. 31,61es not favor those of Hispanic ethnicity: In multivariable models and in comparison with non-Hispanic White survivors, those who were Hispanic (OR 0.69, 0.51-0.95)were less likely to access survivorship care, after adjusting for years since diagnosis, current age, sex, socioeconomic status, health insurance coverage, high levels of depressive symptoms, number of late effects, treatment intensity, received a written treatment summary, reported having a doctor for regular (non-cancer) care, discussed cancer-related follow-up care with a doctor in the last 2 years, knowledge of the need for life-long survivorship care, health care self-efficacy, and family influence of health care decisions. 29,46es not favor those of non-White race: Non-Whites were nearly two times more likely to be non-attenders at survivorship clinic (OR 1.88; CI 1.19, 2.97) compared to Whites, after adjusting for age, socioeconomic status, race, years from diagnosis, additional cancer event, insurance status, means of travel, distance from hospital, scheduled social work consultation, and type of clinic visit. 17,39 univariable analysis, patients of non-White race were more likely to have no shows for survivorship visits (OR 0.27, p = 0.001). 19 the multivariable model, Black race was a significant predictor of missed visits (+6.90%, p < 0.001) when adjusting for tumor grade and insurance status. 22 compared to White survivors, those of Black race (HR 0.64, 0.52-0.79)or other races (HR 0.70; CI 0.49, 0.99) were less likely to have an initial survivorhip visit, when adjusting for sex, therapeutic modalities, current age, insurance status, and distance from the clinic. 23 multivariable analyses and in comparison with White participants, individuals who were Black (OR 2.0; CI 0.9, 4.4) or of other races (OR 2.0; CI 0.4, 9.2) were more likely to be lost to follow-up for more than 1000 days and not attend survivorship clinic, where Hispanic individuals (OR 0.7; CI 0.4, 1.1) were less likely to be lost to follow-up for more than 1000 days and not attend survivorship clinic, when adjusting for diagnosis, treatment exposure, stem cell transplant exposure, age group at diagnosis, and insurance status (global overall p = 0.03). 26 a multivariable model and in comparison with non-Hispanic White survivors, those who were other races or ethnicities (OR 0.69, 0.48-0.99)were less likely to access survivorship care, after adjusting for years since diagnosis, current age, sex, socioeconomic status, health insurance coverage, high levels of depressive symptoms, number of late effects, treatment intensity, received a written treatment summary, reported having a doctor for regular (non-cancer) care, discussed cancer-related follow-up care with a doctor in the last 2 years, knowledge of the need for life-long survivorship care, health care self-efficacy, and family influence of health care decisions. 29,46 comparison with non-Hispanic White survivors, minority survivors were less likely (OR 0.79; CI 0.64, 0.96) to report absence of a cancer center medical visit, after adjusting for age at study, sex, educational attainment, health insurance, health status, concern for future heatlh, and high-risk treatment. 31 comparison with White non-Hispanic CCS, survivors who were Black (OR 2.1; CI 1.3, 3.3) were more likely to report general care rather than risk-based survivorship care and those who were of other races (OR 0.8; CI 0.7, 1.0) were less likely to report general care versus risk-based survivorship care, after adjusting for sex, current age, age at interview, annual household income, educational attainment, employment status, insurance coverage, poor emotional health, cancer-related anxiety, cancer-related pain, poor physical health, and grade of chronic disease. 31,58urvivors who were from the other race/ethnicity groups who reported some form of medical care at baseline (either survivor-focused or general) experienced an increased risk of reporting no care at follow-up (OR 2.1; CI 1.2, 3.7) compared to non-Hispanic White CCS, after adjusting for sex, annual household income, chronic disease status, and education. 31,59 multivariable analysis and in comparison with privately insured Whites, privately insured Blacks (OR 1.In multivariable analysis, patients who identified as Black (OR 0.47; CI 0.23, 0.90) were associated with greater likelihood of nonadherence to survivorship care guidelines, after adjusting for insurance type, age, sex, and cost of recommended procedures. 20ere was not a significant association for those of Hispanic ethnicity in predicting an initial survivorship visit, after adjusting for sex, therapeutic modalities, current age, insurance status, and distance from the clinic. 23ce was not a significant predictor of follow-up at 5 and 10 years (adjusting for clustering of institutions). 36rvivors who were non-Hispanic Black, Hispanic,  20 Berg, 2016 21 Daly, 2019 23 Johnson, 2004 25 McBride, 2011 27 Michel, 2011 28 Milam, 2015 29 Oeffinger, 2004 31 Reppucci, 2017 32 Streefkerk, 2019 33 vanLaar, 2013 35 Welch, 2017 36 Favors older current age: Those who were ages 6-11 (HR 1.55; CI 1.24, 1.93) or 12-17 (HR 1.44; CI 1.14, 1.83) were more likely to have an initial survivor clinic visit compared to those who were between the ages of 2 and 5, after adjusting for sex, race/ethnicity, therapeutic modalities, insurance status, and distance from the clinic. 23 comparison with those <20 years, survivors aged 20 years or older had a significantly higher number of contacts with their PCP (20-29 years contact rate ratio 1.In multivariable analysis, older age (OR 0.97; CI 0.94, 1.0) was related to greater likelihood of nonadherence, after adjusting for race, insurance type, sex, and cost of recommended procedures. 20e odds of difficulty obtaining care increased 3% (OR 1.03; CI 1.00, 1.06) for every year of age at survey, after adjusting for sex, marital status, employment, age at diagnosis, cancer diagnosis, and radiation therapy. 17 the multivariable model, older current age was associated with less than annual healthcare provider visits (OR 1.35; CI 1.11, 1.63), after adjusting for chemotherapy exposure, sex, and insurance coverage. 21 comparison with those currently under the age of 25, those of ages 25-29 (OR 0.66; CI 0.42, 1.03), 30-34 (OR 0.27; CI 0.15, 0.49), and 35+ (OR 0.24; CI 0.12, 0.49) were less likely to attend survivorship care, after adjusting for health beliefs (susceptibility, severity, benefits, barriers, health value, cues to action), demographics (sex, living in a relationship, education, employment, immigration status, language spoken), and medical variables (age at diagnosis, diagnostic category, treatment, SCT, relapse, medical report received, follow-up checklist received). 28r each year increase in current age at survey, survivors were less likely (OR 0.83; CI 0.79, 0.86) to access survivorship care, after adjusting for socioeconomic status at diagnosis, socioeconomic status at survey, age at diagnosis, education, treatment intensity, sex, race/ethnicity, insurance. 29,50urvivors 21 years of age or older were associated with decreased odds of having a recent survivorship visit (OR 0.32; CI 0.13, 0.79), after adjusting for treatment intensity, time since diagnosis, sex, health insurance status, Spanish-speaking Hispanic parent, English-speaking Hispanic parent, and English-speaking non-Hispanic parent. 29,52 the multivariable model and in comparison with those currently 18-20 years of age, survivors ages 21-25 (OR 0.65, 0.50-0.85),26-30 (OR 0.32, 0.22-0.48),and 31-39 (OR 0.35, 0.24-0.50)were less likely to access survivorship care, after adjusting for years since diagnosis, sex, race and ethnicity, socioeconomic status, health insurance coverage, high levels of depressive symptoms, number of late effects, treatment intensity, received a written treatment summary, reported having a doctor for regular (non-cancer) care, discussed cancer-related follow-up care with a doctor in the last 2 years, knowledge of the need for life-long survivorship care, health care self-efficacy, and family influence of health care decisions. 29,46 were more likely to report absence of a cancer center medical visit, after adjusting for sex, ethnicity, educational attainment, health insurance, health status, concern for future heatlh, and high-risk treatment. 31ere was not a significant relationship for those age 18 or older in predicting an initial survivorship visit, after adjusting for sex, race/ethnicity, therapeutic modalities, insurance status, and distance from the clinic, in comparison with those who were between the ages of 2 and 5. 23 In the multivariable analysis, current age was not significantly associated with attending survivorship visits, when adjusting for current age, employment, and socioeconomic status. 25rrent age at the time of follow-up was not significantly associated with 10-year follow-up (adjusting for clustering of institutions). 36st

Quality of evidence Reason for downgrading
For each 1-year increase in age at study, survivors experienced an increased odds (OR 1.03; CI 1.02, 1.04) of having general care (rather than risk-based survivorship care), after adjusting for sex, race/ethnicity, annual household income, educational attainment, employment status, insurance coverage, poor emotional health, cancer-related anxiety, cancer-related pain, poor physical health, and grade of chronic disease. 31,58ose of older current age had reduced odds of adherence as compared with younger subjects (10-year increase: OR 0.  21 Johnson, 2004 25 Michel, 2011 28 Oeffinger, 2004 31 Favors those currently employed: Those who were currently employed experienced a decrease in the odds of difficulty obtaining care (OR 0.52; CI 0.36, 0.75) compared to those who were unemployed, after adjusting for sex, age at survey, marriage, age at diagnosis, cancer diagnosis, and radiation therapy. 17 comparison with employed survivors, unemployed survivors were less likely to report general care rather than risk-based survivorship care (0.7; CI 0.6, 0.8), after adjusting for sex, race/ethnicity, current age, age at interview, annual household income, educational attainment, insurance coverage, poor emotional health, cancer-related anxiety, cancer-related pain, poor physical health, and grade of chronic disease. 31,58ployment status was not significantly associated with annual healthcare provider visits, after adjusting for current age, insurance coverage, and chemotherapy exposure. 21 Barakat, 2012 19 Berg, 2016 21 Daly, 2019 23 Johnson, 2004 25 May, 2017 26 McBride, 2011 27 Michel, 2011 28 Milam, 2015 29 Nathan, 2016 30 Oeffinger, 2004 31 Reppucci, 2017 32 Streefkerk, 2019 33 Welch, 2017 36 Zheng, 2016 38 Favors those treated with chemotherapy: In multivariable analysis, survivors who received any chemotherapy (RR 1.85; 1.0, 3.  27 Does not favor those treated with chemotherapy: In the multivariable model, exposure to chemotherapy was associated with less than annual healthcare provider visits (OR 5.73; CI 0.98, 40.30), after adjusting for current age, sex, and insurance coverage. 21avors those treated with surgery: Those treated with surgery only (contact RR 1.23; CI 1.00, 1.51) had more PCP contact during follow-up compared with those treated with chemotherapy only, when adjusting for sex and attained age. 33urvivors who received treatment with surgery only (HR 0.02; CI 0.00, 0.13) were less likely to attend survivorship clinic in comparison with those who received treatment with surgery, chemotherapy, and radiation, after adjusting for sex, age at diagnosis, and insurance status. 38es not favor those treated with surgery: Those who received surgery only (HR 0.04; CI 0.02, 0.08) were less likely to have an initial survivor clinic visit compared to those who received chemotherapy only, after adjusting for sex, race/ethnicity, therapeutic modalities, current age, insurance status, and distance from the clinic. 23 comparison with those treated with chemotherapy only, those treated with surgery only (OR 0.44; CI 0.20, 0.99) were less likely to attend survivorship care visits, after adjusting for health beliefs (susceptibility, severity, benefits, barriers, health value, cues to action), demographics (sex, age at study, living in a relationship, education, employment, immigration status, language spoken), and medical variables (age at diagnosis, diagnostic category, treatment, SCT, relapse, medical report received, follow-up checklist received). 28 multivariable analyses, survivors treated with surgery alone (in comparison with chemotherapy alone) were more likely to be lost to follow-up for more than 1000 days and not attend survivorship clinic (OR 6.70; CI 3.10, 14.90), after adjusting for cancer diagnosis, race, insurance, and age at diagnosis. 26vors those treated with radiation: In comparison with survivors who did not receive radiation, those who did receive radiation to the brain (OR 0.5; CI 0. Does not favor those treated with SCT: In multivariable analyses, those who received treatment with a SCT (in comparison with no transplant) were more likely to be lost to follow-up for more than 1000 days and not attend survivorship clinic (OR 2.0; CI 1.04, 3.70), after adjusting for diagnosis, race, insurance, and age at diagnosis. 26vors those treated with high intensity regimens according to the Intensity of Treatment Rating Scale: For each increase in the level of treatment intensity (from 1 to 4), survivors were more likely (OR 1.83; CI 1.09, 3.06) to access survivorship care, after adjusting for sex, race/ethnicity, health insurance, family influence on healthcare decisions, having a regular doctor, healthcare self-efficacy, and post-traumatic growth. 29eater treatment intensity was associated with increased odds of having a recent survivorship visit (OR 1.89; CI 1.07, 3.31), after adjusting for current age, time since diagnosis, sex, health insurance status, Spanish-speaking Hispanic parent, English-speaking Hispanic parent, and English-speaking non-Hispanic parent. 29,52creased survivorship clinic attendance was associated with treatment intensity score of 2 (RR 1. Barakat, 2012 19 Benedict, 2021 20 Berg, 2016 21 Daly, 2019 23 McBride, 2011 27 Michel, 2011 28 Milam, 2015 29 Nathan, 2016 30 Oeffinger, 2004 31 Streefkerk, 2019 33 Welch, 2017 36 Zanetti, 2022 37 Zheng, 2016 38 Favors females: In univariable analysis, females were twice as likely as males to attend a survivorship clinic visit 5 years post-diagnosis (OR 2.20, p = 0.077). 19 multivariate regression, male sex (OR 0.09; CI 0.03, 0.35) was associated with less than annual healthcare provider visits, after adjusting for current age, insurance coverage, and chemotherapy exposure. 21 multivariable analysis, female survivors were more likely to have 10 or more PCP visits (RR Barakat, 2012 19 Berkman, 2019 22 Daly, 2019 23 May, 2017 26 McBride, 2011 27 Michel, 2011 28 Nathan, 2016 30 Oeffinger, 2004 31 Reppucci, 2017 32 Streefkerk, 2019 33 Welch, 2017 36 Zheng, 2016 38 Favors those who experienced a relapse: In multivariable analysis, survivors who had relapsed (RR 2.11; CI 1.4, 3.3) had an increased likelihood of oncologist visits compared with survivors of acute lymphoblastic leukemia; controlled for sex, socioeconomic status, residence in a metropolitan/large community/small community/rural area, time since diagnosis, current age, age at diagnosis, relapse status, second malignancy, and treatment). 27 comparison with those who did not experience a relapse, those who did experience a relapse (OR 2.78; CI 1.70, 4.56) were more likely to attend survivorship care, after adjusting for health beliefs (susceptibility, severity, benefits, barriers, health value, cues to action), demographics (sex, age at study, living in a relationship, education, employment, immigration status, language spoken), and medical variables (age at diagnosis, diagnostic category, treatment, SCT, medical report received, follow-up checklist received). 28vors those who were diagnosed with a bone tumor or sarcoma: In multivariable analysis, survivors of bone tumors (RR 3.89; CI 1.9, 7. May, 2017 26 McBride, 2011 27 Michel, 2011 28 Milam, 2015 29 Nathan, 2016 30 Oeffinger, 2004 31 Reppucci, 2017 32 Szalda, 2016 34 Zanetti, 2022 37 Zheng, 2016 were more likely to attend survivorship care, after adjusting for health beliefs (susceptibility, severity, benefits, barriers, health value, cues to action), demographics (sex, age at study, living in a relationship, education, employment, immigration status, language spoken), and medical variables (diagnostic category, treatment, SCT, relapse, medical report received, follow-up checklist received). 28r each year increase in age at diagnosis, survivors were more likely (OR 1.12; CI 1.08, 1.16) to access survivorship care, after adjusting for socioeconomic status at diagnosis, socioeconomic status at survey, education, treatment intensity, sex, race/ ethnicity, age at survey, insurance. 29,50r each year increase in age at diagnosis, survivors were less likely (OR 0.88; CI 0.84, 0.92) to access survivorship care, after adjusting for current age, sex, race and ethnicity, socioeconomic status, health insurance coverage, high levels of depressive symptoms, number of late effects, treatment intensity, received a written treatment summary, reported having a doctor for regular (non-cancer) care, discussed cancer-related follow-up care with a doctor in the last 2 years, knowledge of the need for life-long survivorship care, health care self-efficacy, and family influence of health care decisions. 29,46 age at diagnosis increased, adherence with echocardiogram recommendations decreased (10-year increase: OR 0.59; CI 0.37, 0.95), after adjusting for sex, radiation exposure, anthracycline exposure, transplant, and age at procedure recommendation. 32urvivors who were older ages at diagnosis (OR 0.90; CI 0.82, 0.99) were less likely to access survivorship care for each year of age increase, after adjusting for knowledge of risk for second cancer, comfort discussing concerns, and motivation to take care of health. 34,69 Barakat, 2012 19 Daly, 2019 23 McBride, 2011 27 Nathan, 2016 30 Oeffinger, 2004 31 Zheng, 2016 38 Favors those living in non-metropolitan areas: In multivariable analysis, survivors from a small community (in comparison with a metropolitan designated area) were more likely to have a visit with an oncologist (RR 1.45, CI 1.0, 2.0; controlling for sex, socioeconomic status, sex, current age, time since diagnosis, diagnosis, age at diagnosis, relapse status, second malignancy, and treatment). 27es not favor those who travel far distances: In univariable analysis, patients who lived farther than 57.6 km to the hospital were less likely (OR 0.24, p = 0.003) to attend a follow-up or survivorship visit as patients who lived closer. 19ose who lived 25-50 miles from the clinic (HR 0.76; CI 0.63, 0.93) or those who lived more than 50 miles from the clinic (HR 0.67; CI 0.54, 0.82) were significantly less likely to have an initial survivor clinic visit compared to those who lived less than 25 miles from the clinic, after adjusting for sex, race/ethnicity, therapeutic modalities, current age, and insurance status. 23ose who traveled 25-49 km (RR 0.88; CI 0.76, 1.01), 50-99 km (RR 0.77, CI 0.65-0.91),>100 km (RR 0.48; CI 0.39, 0.60) had a decreased likelihood of survivorship clinic attendance, after adjusting for sex, age at diagnosis, socioeconomic status, cancer diagnosis, diagnosis prior to 1999, treatment intensity, cyclophosphamide equivalent dose, doxorubicin equivalent dose, radiation, secondary malignancy or relapse before index date, survivorship clinic model, and complete history/physical exam by a PCP. 30 multivariable regression and after adjusting for individual factors, the number of Childhood Cancer Survivor Study centers (OR 1.12; CI 1.04, 1.20) and the number of physicians/surgeons (OR 1.06; CI 1.01, 1.11) within the geographic area was associated with greater odds of receiving risk-based survivor-focused medical care among U.S. residents. 31,60es not favor those who travel by car: Those who traveled by a car were more likely to be non-attenders at survivorship clinic (OR 12.74; CI 3.97, 40.86) compared to those who traveled by bus, after adjusting for age, socioeconomic status, race, years from diagnosis, additional cancer event, insurance coverage, distance from hospital, scheduled social work consultation, and type of clinic visit. 17,39rvivors who traveled more than 15 min to the hospital (in comparison with those with travel times less than 15 min) did not experience a significant association with survivorship clinic attendence, after adjusting for sex, age at diagnosis, and insurance. 38erall, there seems to be an association between being from an underserved or rural area and experiencing a disparity in survivorship care engagement.

Low
Reason for downgrading: Inconsistency a Insurance type 6 Retrospective observational studies: Benedict, 2021 20   Daly, 2019 23 Gardner, 2014, 24 Milam, 2015 29 Oeffinger, 2004 31 Welch, 2017 36 Favors those who are publicly insured: In adjusted models, publicly insured survivors were more likely to report a cancer-related (OR 1.22; CI 1.11, 1.35) or a cancer center visit (OR 1.41; CI 1.18, 1.70) than privately insured survivors, when adjusting for age, sex, household income, highest level of educational attainment, race/ethnicity, and Grade 3 or 4 chronic condition. 31,61 univariable analysis, use of a mental health professional was less likely to be associated with private insurance coverage, in comparison with those with public insurance or uninsured (p = 0.040).24   Does not favor those who are publicly insured: Those who had Medicaid at diagnosis (HR 0.77; CI 0.64, 0.92) were less likely than those who had private insurance to have had an initial survivor visit, after adjusting for sex, race/ethnicity, therapeutic modalities, current age, and distance from the clinic. 23 5-and 10-year post-diagnosis, public insurance coverage was negatively associated with follow-up care (5-year follow-up OR 0.58; CI 0.17, 1.0; 10-year follow-up OR 0.48; CI 0.22, 0.73), when compared to those with private or military insurance coverage. 36oes not favor those who are publicly insured or uninsured: In multivariable analysis, being insured by Medicaid or uninsured (OR 0.59; CI 0.36, 0.96) was related to greater likelihood of nonadherence, after adjusting for race, age, sex, and cost of recommended procedures. 20 comparison with those who were privately insured, those who were uninsured were 4.3 times as likely to have no regular provider for cancer-related follow-up care (CI 1.9, 9.4), 3.3 times as likely to lack a regular provider for non-cancer care (CI 1.6, 6.9), 5.3 times as likely to lack both sources of care (CI 2.1, 13.5), 3.9 times as likely to have had no primary care visit (CI 1.8, 8.2), and 4.5 times as likely to have not seen a cancer specialist (CI 2.1, 9.5).In comparison with those who were privately covered, those with public insurance were 2.5 times as likely to report no regular source of primary care (CI 1.1, 5.4) and 2.8 times as likely not to have made a primary care visit in the past 2 years compared with those with private coverage.Both models adjusted for sex, ethnicity, age, socioeconomic status, and treatment intensity. 29,48ere was no systematic association of insurance type and survivorship care.Daly, 2019 23  McBride, 2011 27 Nathan, 2016 30 Does not favor those with less time passed since diagnosis:

Low
In multivariable analysis, survivors who were 10-14 years since diagnosis (RR 0.53; CI 0.3, 0.9) and those ≥25 years since diagnosis (RR 0.31, CI 0.2, 0.6) were less likely to have a visit with an oncologist than those diagnosed 5-9 years prior were (controlling for socioeconomic status, residence in a metropolitan/large community/small community/rural area, current age, diagnosis, age at diagnosis, relapse status, second malignancy, and treatment). 27creased survivorship clinic attendance was associated with a diagnosis prior to 1999 (RR 0.74; CI 0.

Quality of evidence Reason for downgrading
Nathan, 2016 30  Oeffinger, 2004 31 Zheng, 2016 38 In comparison with survivors reporting current annual household income more than $60,000, survivors who had income less than $20,000 (RR 1.6; CI 1.2, 2.3) or between $20,000 and $39,999 (RR 1.4; CI 1.0, 1.9) who reported some form of medical care at baseline (either survivor-focused or general) experienced an increased risk of reporting no care at follow-up, after adjusting for sex, race/ethnicity, chronic disease status, and education. 31,59oes not favor adherence to higher cost procedures: In multivariable analysis, patients with recommended procedures that exceeded a median cost of $400 were less likely to be adherent than those with recommended procedures that cost less than $400 (OR 0.32; CI 0. The prospective study shows a negative association with female sex, where retrospective studies report positive associations with female sex with obtaining survivorship care. T A B L E 1 (Continued) for sex, age at survey, marriage status, employment status, age at diagnosis, cancer diagnosis, and radiation therapy. 17n multivariable analysis adjusting for different variables in retrospective observational studies, being insured led to increased survivorship care engagement, with odds ranging from 1.7 to 4.2. 26,29,31,36Similarly, in multivariable analysis of CCS who were uninsured, disparities in survivorship care engagement were demonstrated with odds ranging from 0.04 to 0.28. 21,29However, two studies found contradicting results.In a subsequent publication of the prospective study, CCS who were privately insured were more likely to be nonattenders at survivorship clinics (OR 2.36; CI 1.98, 3.79) compared to CCS who were uninsured, after adjusting for age, socioeconomic status, race, years from diagnosis, additional cancer event, means of travel, distance from hospital, scheduled social work consultation, and type of clinic visit. 17,39By contrast, in a retrospective study, use of a mental health professional was less likely to be associated with private insurance coverage upon univariable analysis, in comparison with CCS with public insurance or who were uninsured (p = 0.040). 24n total, nine observational studies (one prospective, eight retrospective) assessed potential disparities in survivorship care for CCS by race or ethnicity. 17,19,22,23,26,29,31,36,38Across studies, we found high QOE for disparities in survivorship care engagement for survivors of non-White race and Hispanic ethnicity.Of the nine studies examining disparities by race or ethnicity, we found seven studies 17,18,20,23,26,29,31 F I G U R E 2 Variables mapped to survivorship care disparities by quality of evidence and number of studies.
that identified a disparity experienced by those of non-White race or Hispanic ethnicity with estimates of lack of engagement odds ranging from 0.47 to 0.79. 20,29,31A prospective study reported non-Whites were nearly two times more likely to not attend survivorship clinic (OR 1.88; CI 1.19, 2.97) compared to Whites, after adjusting for age, socioeconomic status, years from diagnosis, additional cancer event, insurance status, means of travel, distance from hospital, scheduled social work consultation, and type of clinic visit. 17Notably, there was no systematic association of race or ethnicity with insurance type (private, public) although the result was based on low QOE.However, one retrospective observational study found favorable outcomes among CCS of non-White race and Hispanic ethnicity. 31Publicly insured Hispanic survivors were more likely to report a cancer-related survivorship visit (OR 1.41; CI 1.12, 1.78) and a cancer center visit (OR 1.88; CI 1.28, 2.77) compared to privately insured Hispanic survivors, after adjusting for age, sex, household income, highest level of educational attainment, and Grade 3 or 4 chronic condition. 31,61Hispanic survivors were more likely to report a cancer center visit (females: OR 1.5; CI 1.1, 2.0; males: OR 1.7; CI 1.2, 2.3) in comparison with non-Hispanic White survivors, and Hispanic males were more likely to report cancer-related medical visit (OR 1.3; CI 1.0, 1.8) compared to non-Hispanic White male survivors, after adjusting for age, cancer diagnosis, health insurance, household income, and highest level of educational attainment. 31,613.2.22][33]35,36 Of the 13 studies, nine indicated that older age is associated with a disparity in survivorship care and general care engagement (moderate QOE) among CCS. 17,20,21,28,29,31,32,35,36For example, a prospective study reported that the odds of difficulty obtaining care increased 3% (OR 1.03; CI 1.00, 1.06) for every year of age at survey, after adjusting for sex, marital status, employment, age at diagnosis, cancer diagnosis, and radiation therapy exposure. 17cross five studies (one prospective, four retrospective), being unemployed was not systematically associated with a disparity in survivorship care engagement and there were mixed results for other care outcomes (moderate QOE) among CCS. 17,21,25,28,31Overall, three of the five studies found no association between employment and disparities in survivorship care engagement among CCS. 21,25,28However, a prospective study found a significant association, in which CSS who were currently employed experienced a decrease in the odds of difficulty obtaining care (OR 0.52; CI 0.36, 0.75) compared to CCS who were unemployed, after adjusting for sex, marital status, age at survey, age at diagnosis, cancer diagnosis, and radiation therapy exposure. 17

| Low quality of evidence
0][31][32][33]36,38 Across these studies, there was evidence that survivors who received less intense treatment (e.g., guided by the Intensity of Treatment Rating Scale 70 ) or single modality regimens were less likely to be engaged in survivorship care and general care 21,23,27,28,30,31,38 ; however, some conflicting evidence was identified, in which four studies did not find an association with treatment type or intensity (low QOE). 17,27,33,36n total, 14 studies (1 prospective, 13 retrospective) reported on potential disparities by biologic sex, six of which reported that male sex was associated with lower survivorship engagement (compared to female sex) 19,21,27,29,30,33 ; however, six studies showed no association 20,23,28,31,[36][37][38] or a conflicting direction of effects 17 (low QOE).For example, a prospective study found that the odds of experiencing difficulty obtaining care in the prior year decreased for males (OR 0.59; CI 0.41, 0.85) compared to females, after adjusting for age at survey, marital status, employment status, age at diagnosis, cancer diagnosis, and radiation therapy exposure. 171][32][33]36,38 Across studies, no clear pattern for these disparities was found.In accordance with one prospective study, cancer diagnosis (e.g., hematologic malignancies compared to solid tumors) may be associated with disparities, but the exact pattern remains unclear (low QOE). 1431][32]34,38 The evidence primarily indicated that older age at diagnosis is associated with a disparity both in survivorship care and general care engagement among CCS, as reported in 6 studies. 26,28,32,34,37However, some inconsistent results were found in six studies (low QOE). 17,27,30,32,37,38verall, seven studies (1 prospective, 6 retrospective) explored the effect of geographic area on potential disparities among CCS. 17,19,23,27,30,31,38Some evidence of an association between being from an underserved or rural area and experiencing a disparity in survivorship care engagement was reported in four studies 19,23,30,31 ; however, conflicting evidence where other three studies indicated no association or a positive association exists (low QOE). 17,27,38nly six retrospective studies assessed potential disparities in survivorship care engagement by insurance type (low QOE). 20,23,24,29,31,36There was some indication of disparities by insurance type identified in all six studies, but there was no systematic association with disparities in survivorship care engagement among CCS.For example, two studies identified a benefit for CCS who were publicly insured 24,31 ; while, two other studies identified disparities for those who were publicly insured 23,36 and two remaining studies identified disparities for those who were publicly insured or uninsured (all in comparison with those who were privately insured). 20,29nly three retrospective studies evaluated time since diagnosis.Of those, two studies indicated that CCS farther from diagnosis were less likely to receive survivorship care and had lower general care engagement (low QOE) than more recently diagnosed CCS. 27,30However, two studies showed no association and conflicting results. 23,27nly two retrospective studies assessed disparities associated with educational attainment. 28,31One study indicated that lower levels of educational attainment are associated with a disparity in survivorship care engagement among CCS, 28 while another study did not find significant results (low QOE). 30ix retrospective studies assessed potential disparities by income. 20,25,27,30,31,38However, conflicting results were found, with three studies finding a positive association with income, 25,30,31 one study finding less adherence to higher cost procedures, 20 and three studies finding no association or conflicting results. 27,31,38Across the available research it, remained unclear whether income was associated with disparities in survivorship care engagement among CCS (very low QOE).
No included studies that assessed survivorship care disparities examined those by self-identified gender (insufficient evidence).

| DISCUSSION
In this systematic review, we identified strong evidence of disparities associated with race, ethnicity, and insurance status and moderate evidence for employment status and current age.The review documents the current state of the disparities literature for CCS and opportunities for interventions to reduce disparities at multiple levels, and areas for future investigation.
There was high QOE of survivorship care disparities among CCS despite contemporary policy changes designed to propel the health care system toward achieving health equity (e.g., the US Patient Protection and Affordable Care Act's Medicaid expansion). 71,7219,21-24,26,29,31,34,36,38 There was high QoE for race, ethnicity, and insurance status, indicating that CCS who were of non-White race, Hispanic ethnicity, or uninsured are less likely to engage in survivorship care.17,[21][22][23][24]26,29,31,34,36,38 However, it is worth noting that existing studies adjust for different covariates, hence it is difficult to estimate effect size.Nevertheless, these results support future studies to further "unpack" potential underlying factors.For instance, to counteract these disparities, providers could target survivors who may experience insurance instability or churn in coverage (e.g., those covered by Medicaid as children) to ensure survivors and their caregivers understand the importance of continuous engagement in survivorship care and potential resources for where they can seek survivorship services when they lack insurance coverage or experience instability in insurance.Additionally, if innetwork survivorship care is not an option locally for a survivor, a primary care provider could provide the needed survivorship care in some cases, as clinically appropriate, with a comprehensive survivorship care plan and close communication with a survivorship care provider.
This review explored a variety of disparities in suvivorship care for CCS, but in most cases, the existing research base remains limited and shows conflicting results, which hinders our ability to make concluding statements for specific survivorship care disparities.For instance, although a substantial amount of research examined cancer treatment, type of cancer diagnosis, sex, age at diagonosis, and geographic area, the QOE is low.Disparities by income have also been examined in a small number of studies; however, the QOE is very low as it is unclear whether income is associated with disparities across studies.Even fewer studies have examined time since diagnosis, insurance type, and educational attainment, contributing to a low QOE assessment.There was insufficient evidence of disparities by gender because we did not identify studies that reported on self-identified gender (rather than biological sex).Taken together, these results underscore the gaps in the field that need to be addressed.Coordinated efforts to study survivorship care disparities among CCS have been implemented through the Childhood Cancer Star Act and other policy changes. 73Future investigations should use prospective, multilevel study designs across independent research groups and study cohorts to understand drivers of disparities and to inform strong evidence-based conclusions.
The review has several strengths and limitations.We reviewed a large quantity of empirical literature to identify relevant studies for this review.Our broad search was designed to identify studies that have evaluated potential disparities, regardless of the result of the evaluation, in order to objectively document the presence and the absence of disparities.This comprehensive review is nonetheless limited by the small number of studies that have empirically assessed disparities among CCS, our reliance on the methodological rigor of identified studies to detect and report disparities, and the restriction to studies reported in English.Furthermore, the vast majority of included studies are quantitative in nature.Qualitative or mixed methods studies would allow for more context about the intersectionality of identifies that contribute to or experience disparities.Future research should specifically prioritize exploring potential disparities among CCS to improve our understanding of health inequities and to inform how the diversity of CCS should best be supported.

| CONCLUSION
Across identified empirical research, we found high QOE that non-White race or Hispanic ethnicity and lack of insurance are associated with disparities in survivorship care among CCS.Nonetheless, multiple other variables, such as employment, income, insurance type, education, cancer diagnosis, age at diagnosis, time since diagnosis, cancer treatment, geographic area, sex, and self-identified gender, have been suggested in individual studies and reported effects warrant further investigation.Furthermore, CCS face many complex and multifactorial challenges, including those that interplay with social determinants of health, such as socioeconomic hardship (e.g., changes in income, increasing out-of-pocket medical expenses, rising debt), decreased educational attainment, detrimental employment factors (e.g., being unemployed, lack of benefits, limited or no paid time off and/or sick leave), inadequate insurance coverage (e.g., being under or uninsured, barriers due to type of insurance, transitions or changes in insurance coverage), and being from an underserved or minority group (e.g., racial or ethnic minorities, urban, rural, nonmetropolitan regions).These factors may lead to barriers to care, and exacerbate disparities, or, could independently act as disparate groups.Future prospective research is needed to fully understand the myriad of disparities faced by CCS, particularly using a multilevel approach, to understand how we may design studies to intervene to counteract these disparities.

Variable Number of studies by type: citation Disparities assessment findings: significant Disparities assessment findings: not significant Summary across studies Quality of evidence Reason for downgrading
38

of studies by type: citation Disparities assessment findings: significant Disparities assessment findings: not significant
a T A B L E 1 (Continued) (Continues) Variable Number

Variable Number of studies by type: citation Disparities assessment findings: significant Disparities assessment findings: not significant Summary across studies Quality of evidence Reason for downgrading
T A B L E 1 (Continued) 3), chemotherapy/surgery (RR 2.30; CI 1.3, 4.1), chemotherapy/radiation (RR 2.43, CI 1.3, 4.0), radiation/surgery (RR 3.5; CI 2.0, 6.0), or chemotherapy/radiation/surgery (RR 3.95; CI 2.2, 7.1) were more likely to have an oncologist visit compared with survivors who received only surgery (adjusting for sex, socioeconomic status, residence in a metropolitan/large community/small community/rural area, current age, time since diagnosis, diagnosis, age at diagnosis, relapse status, and second malignancy).

Variable Number of studies by type: citation Disparities assessment findings: significant Disparities assessment findings: not significant Summary across studies Quality of evidence Reason for downgrading
23,586), chest (OR 0.3; CI 0.2, 0.4), other (OR 0.5; CI 0.4, 0.6), or unknown sites (OR 0.5; CI 0.4, 0.8) were less likely to report general care versus risk-based survivorship care, after adjusting for sex, current age, and age at interview.31,58Thosewhoreceivedradiationtherapyhadgreaterodds of DXA scan adherence (OR 2.60; CI 1.39, 4.88) as compared with those who did not receive radiation, after adjusting for sex, anthracycline exposure, transplant, age at diagnosis, and age at procedure.32Thosetreatedwithradiotherapyonly(contactRR1.49;CI1.15, 1.93) had more PCP contact during follow-up compared with those treated with chemotherapy only, when adjusting for sex and attained age.33Survivors who received radiation to the head/neck (HR 1.88; CI 1.28, 2.77), chest (HR 2.72; CI 1.71, 4.34), or abdomen/pelvis (HR 1.78, CI 1.01, 3.13) were more likely to attend survivorship clinic in comparison with those who received no radiation, after adjusting for sex, age at diagnosis, and insurance status.38Doesnotfavorthosetreated with radiation: Those who received radiation only (HR 0.24; CI 0.15, 0.39) were less likely to have an initial survivor clinic visit compared to those who received chemotherapy only, after adjusting for sex, race/ethnicity, therapeutic modalities, current age, insurance status, and distance from the clinic.23Favorsthosetreatedwith SCT:In comparison with those who did not have a SCT, those who did have a SCT (OR 2.83; CI 1.21, 6.58) were more likely to attend survivorship care visits, after adjusting for health beliefs (susceptibility, severity, benefits, barriers, health value, cues to action), demographics (sex, age at study, living in a relationship, education, employment, immigration status, language spoken), and medical variables (age at diagnosis, diagnostic category, treatment, relapse, medical report received, follow-up checklist received).28

Variable Number of studies by type: citation Disparities assessment findings: significant Disparities assessment findings: not significant Summary across studies Quality of evidence Reason for downgrading
31, CI 1.26, 2.25), score of 3 (RR 2.21; CI 1.65, 2.96), or score of 4 (RR 1.69; CI 1.10, 2.58), compared to those with a score of 1 (least intense treatment); higher cyclophosphamide equivalent doses as those who received 4000-7999 mg/m 2 Models were adjusted for sex, age at diagnosis, socioeconoomic status, cancer diagnosis, cyclophosphamide equivalent dose, doxorubicin equivalent dose, radiation, secondary malignancy, or relapse before index date, survivorship clinic model, distance to survivorship clinic, and complete history/physical exam by a PCP.30In comparison with survivors who did not receive high-risk treatment, those who did receive high-risk treatment were less likely to report absence of cancer-related medical visits (OR 0.59; CI 0.52, 0.65) or cancer center medical visits (OR 0.45, CI 0.39, 0.51), after adjusting for age at study, sex, ethnicity, educational attainment, insurance coverage, health status, and concern for future health.31Incomparisonwith survivors who did not receive treatment with anthracyclines, those who received chest radiation (OR 0.4; CI 0.3, 0.6) and those who received anthracyclines and chest radiation (OR 0.5; CI 0.4, 0.6) were less likely to report general care versus risk-based survivorship care, after adjusting for sex, current age, and age at interview.In comparison with survivors who did not receive treatment with alkylating agents, those who received the highest dose in the third tertile (OR 0.6; CI 0.4, 0.9) were less likely to report general care versus risk-based survivorship care, after adjusting for sex, current age, and age at interview. 31urvivors who received surgery and chemotherapy (HR 0.54; CI 0.30, 0.97) were less likely to attend survivorship clinic in comparison with those who received treatment with surgery, chemotherapy, and radiation; however, survivors who received alkylating agents (HR 2.28; CI 1.58, 3.30), anthracyclines (HR 3.05; CI 2.09, 4.44), or lung toxic therapies (HR 1.89; CI 1.19, 3.00) were more likely to attend survivorship clinic in comparison with those who received no chemotherapy; after adjusting for sex, age at diagnosis, and insurance status. 38oes not favor those with more time lapsed since treatment completion: Patients who were off treatment for a longer period of time were about half as likely (OR 0.44, p (RR 1.29; CI 1.08, 1.54) or those who received >8000 mg/m 2 (RR 1.47; CI 1.25, 1.73), compared to those who did not receive cyclophosphamide; and radiation to the brain (RR 1.44; CI 1.23, 1.67), chest (RR 1.59; CI 1.28, 1.97), or other sites (RR 1.31; CI 1.13, 1.53), compared to those who did not receive radiation.T A B L E 1 (Continued)

Variable Number of studies by type: citation Disparities assessment findings: significant Disparities assessment findings: not significant Summary across studies Quality of evidence Reason for downgrading
17,52ocioeconomic status at diagnosis, socioeconomic status at survey, age at diagnosis, education, treatment intensity, race/ethnicity, age at survey, and insurance.29,50Femalesurvivorsweremorelikelytohavearecentsurvivorship visit (OR 4.26; CI 1.60, 11.40), after adjusting for current age, time since diagnosis, treatment intensity, health insurance status, Spanish-speaking Hispanic parent, English-speaking Hispanic parent, and English-speaking non-Hispanic parent.29,52Femalesweremorelikely to attend survivorship clinic (RR 1.18; CI 1.07, 1.31), after adjusting for age at diagnosis, socioeconoomic status, cancer diagnosis, diagnosis prior to 1999, treatment intensity, cyclophosphamide equivalent dose, doxorubicin equivalent dose, radiation, secondary malignancy or relapse before index date, survivorship clinic model, distance to survivorship clinic, and complete history/physical exam by a PCP.30Females had more contacts with their PCP (contact ratio 1.96; CI 1.72, 2.23) in comparison with males, after adjusting for sex, attained age, and treatment received.33Doesnotfavorfemales:The odds of difficulty obtaining care decreased for males in the prior year (OR 0.59; CI 0.41, 0.85) compared to females, after adjusting for age at survey completion, marital status, employment status, age at diagnosis, cancer diagnosis, and radiation therapy.17 1.78; CI 1.5, 2.1), more likely to visit specialists (RR 1.15; CI 1.0, 1.3), and more likely to have seen an oncologist (RR 1.40; CI 1.1, 1.7) than male survivors (controlling for socioeconomic status, residence in a metropolitan/large community/small community/rural area, current age, time since diagnosis, diagnosis, age at diagnosis, relapse status, second malignancy, and treatment). 27Female survivors were more likely (OR 1.34; CI 1.11, 1.62) to access survivorship care, after adjusting

Variable Number of studies by type: citation Disparities assessment findings: significant Disparities assessment findings: not significant Summary across studies Quality of evidence Reason for downgrading
30 had an increased likelihood of oncologist visits compared with survivors of acute lymphoblastic leukemia; controlled for sex, socioeconomic status, residence in a metropolitan/large community/small community/rural area, time since diagnosis, current age, age at diagnosis, relapse status, second malignancy, and treatment).27eradjustingforsex,age at diagnosis, and insurance status, survivors diagnosed sarcoma (HR 3.30; CI 1.47, 7.42) were more likely to attend survivorship clinic in comparison with those diagnosed with central nervous system tumors.38snotfavorthosediagnosedwith bone cancer: In comparison with survivors' who were diagnosed with bone cancer, survivors who were diagnosed with central nervous system (7.25;1.25,42.23),Wilmstumor(6.67;1.24,36.02),orneuroblastoma(8.30; 1.33, 51.96) experienced an increased likelihood of reporting physician-based skin exam, after adjusting for sex, education, age (years), race/ethnicity, skin type, age at diagnosis, diagnosis, chemotherapy (yes/no), highest CTCAE grade chronic condition, maximum radiotherapy dose, patient activation.31,74orsthosewhowerediagnosed with leukemia or lymphoma:In comparison with those who were diagnosed with a hematologic malignancy, those with a solid tumor experienced a increase in the odds of difficulty obtaining care (OR 1.84, CI 1.25, 2.71) compared to those who were diagnosed with a hematologic malignancy, after adjusting for sex, age at survey, marriage, employment status, insurance coverage, age at diagnosis, and radiation therapy.17Afteradjustingforsex,age at diagnosis, and insurance status, survivors diagnosed with leukemia (HR 3.36; CI 1.Does not favor those diagnosed with a brain tumor or other cancer:In univariate analysis, those diagnosed with brain tumors (OR 0.24, p = 0.055) were less likely to attend a survivorship visit compared to patients with leukemia or lymphoma.19Decreasedsurvivorshipclinic attendance was associated with a diagnosis of a brain tumor (RR 0.63; CI 0.50, 0.77) or other cancer (RR 0.67; CI 0.54, 0.84) compared to ALL, after adjusting for sex, age at diagnosis, socioeconomic status, diagnosis prior to 1999, treatment intensity, cyclophosphamide equivalent dose, doxorubicin equivalent dose, radiation, secondary malignancy or relapse before index date, survivorship clinic model, distance to survivorship clinic, and complete history/physical exam by a PCP.30

Variable Number of studies by type: citation Disparities assessment findings: significant Disparities assessment findings: not significant Summary across studies Quality of evidence Reason for downgrading
31,58ors older age at diagnosis:For each 1-year increase in age at diagnosis, survivors experienced a decreased odds (OR 0.97; CI 0.95, 0.97) of having general care (thus more likely to have risk-based survivorship care), after adjusting for sex, race/ethnicity, current age, age at interview, annual household income, educational attainment, employment status, insurance coverage, poor emotional health, cancerrelated anxiety, cancer-related pain, poor physical health, and grade of chronic disease.31,58 Does not favor older age at diagnosis: In multivariable analyses, older age at diagnosis (ages 5-9 OR 1.8; CI 1.1, 3.0; ages 10-14 OR 3.3; 1.8, 6.1; and ages 15 and above: OR 4.8; 2.1, 11.7; in comparison with those ages 0-4) was more likely to be lost to follow-up for more than 1000 days and not attend survivorship clinic, when adjusting for diagnosis, race, insurance, treatment exposure, and SCT exposure (global overall p-value <0.001). 26In comparison with those ages 0-4 at diagnosis, those of ages 8-11 (OR 2.06; CI 1.18, 3.61) or 12+ years of age (OR 4.41; 2.56, 7.60)

Variable Number of studies by type: citation Disparities assessment findings: significant Disparities assessment findings: not significant Summary across studies Quality of evidence Reason for downgrading
37ltivariable analysis, being age 10-19 years at diagnosis was associated with decreased odds of receiving a dual-energy x-ray absorptiometry (DEXA) scan (OR 0.32; CI 0.11, 0.95) and mental health visits (OR 0.28; CI 0.11-0.70),respectively,after adjusting for sex, sponsor rank group, and type of care (military vs. civilian).37

Variable Number of studies by type: citation Disparities assessment findings: significant Disparities assessment findings: not significant
30, 0.86) compared to those diagnosed after 1999, after adjusting for sex, age at diagnosis, socioeconoomic status, cancer diagnosis, treatment intensity, cyclophosphamide equivalent dose, doxorubicin equivalent dose, radiation, secondary malignancy or relapse before index date, survivorship clinic model, distance to survivorship clinic, and complete history/physical exam by a PCP.30 Reason for downgrading, inconsistency: studies report conflicting result or prospective study does not confirm effects reported in retrospective studies.Three out of five studies found no association, but a prospective study found that those who were employed experienced less difficulty obtaining survivorship care.
a b Reason for downgrading, study limitation: the evidence statement was exclusively based on retrospective studies.c Reason for downgrading, indirectness: the identified studies partially addressed the question of interest.d e