Adherence to Surveillance for Second Malignant Neoplasms and Cardiac Dysfunction in Childhood Cancer Survivors: A Childhood Cancer Survivor Study

Abstract

PURPOSE

To evaluate childhood cancer survivors’ adherence to surveillance protocols for late effects of treatment and to determine the factors affecting adherence.

METHODS

Between 2014 and 2016, 11,337 survivors and 2,146 siblings in the Childhood Cancer Survivor Study completed a survey ascertaining adherence to Children’s Oncology Group (COG) guidelines for survivors at high risk for second malignant neoplasms or cardiac dysfunction and to the American Cancer Society (ACS) cancer screening guidelines for average-risk populations. Adherence rates and factors affecting adherence were analyzed.

RESULTS

Median age at diagnosis was 7 years (range, 0-20.9 years), and median time from diagnosis was 29 years (range, 15-47 years). Among high-risk survivors, adherence to COG breast, colorectal, skin, and cardiac surveillance was 12.6% (95% CI, 10.0% to 15.3%), 37.0% (34.1% to 39.9%), 22.3% (21.2% to 23.4%), and 41.4% (40.1% to 42.7%), respectively. Among average-risk survivors, adherence to ACS breast, cervical, and colorectal screening was 57.1% (53.2% to 61.0%), 83.6% (82.7% to 84.5%), and 68.5% (64.7% to 72.2%), respectively. Twenty-seven percent of survivors and 20.0% of primary care providers (PCPs) had a survivorship care plan (SCP). For high-risk survivors, SCP possession was associated with increased adherence to COG breast (22.3% v. 8.1%; prevalence ratio [PR], 2.52; CI, 1.59 to 4.01), skin (34.8% v 23.0%; PR, 1.16; CI, 1.01 to 1.33), and cardiac (67.0% v 33.1%; PR, 1.73; CI, 1.55 to 1.92) surveillance. For high-risk survivors, PCP possession of a SCP was associated only with increased adherence to COG skin cancer surveillance (36.9% v 23.2%; PR, 1.24; CI, 1.08 to 1.43).

CONCLUSION

Guideline adherence is suboptimal. Although survivor SCP possession is associated with better adherence, few survivors and PCPs have one. New strategies to improve adherence are needed.

INTRODUCTION

Of the 420,000 childhood cancer survivors in the United States,¹ many are at an elevated risk of treatment-related adverse health outcomes, such as subsequent malignant neoplasms (SMN) and cardiac dysfunction.^2,3 Adherence to risk-adapted surveillance for these outcomes can reduce mortality.^4,5

First published in 2003, the Children’s Oncology Group (COG) Long-Term Follow-Up (LTFU) Guidelines for Survivors of Childhood, Adolescent and Young Adult Cancer⁶ recommend SMN and cardiac dysfunction surveillance in survivors at an elevated risk for these morbidities. Survivors without an elevated risk are advised to adhere to the American Cancer Society (ACS) guidelines for cancer screening in the general population.⁷ A Childhood Cancer Survivor Study (CCSS) questionnaire⁸ administered between 2002 and 2003 demonstrated that among 8,347 survivors, 46.2%, 11.5%, 26.6%, and 28.0% at high risk for breast cancer, colorectal cancer, skin cancer,⁵ or cardiac dysfunction (unpublished data; P. Nathan, MD), respectively, received the recommended surveillance. Since then, numerous initiatives have aimed to improve surveillance. In 2006, the Institute of Medicine (IOM) recommended that all childhood cancer survivors receive a survivorship care plan (SCP) that documents cancer treatment–related health risks and the recommended screening and surveillance.⁹ In 2012, the American College of Surgeons Commission on Cancer made SCP provision a requirement for cancer program accreditation.¹⁰ It is now common in pediatric oncology to provide a SCP to patients and their primary care providers (PCPs) at the completion of therapy. Although considerable resources have been dedicated to SCP development and distribution, it is unknown whether SCPs and other educational efforts have improved guideline adherence. This study’s purpose was to assess current adherence to COG and ACS cancer and cardiac surveillance guidelines, and to explore predictors of adherence. We analyzed changes in adherence between 2003 and 2016, and compared survivor adherence with survivors’ siblings and with general population data from the National Health Interview Survey (NHIS).

METHODS

Participants

The CCSS is a retrospective cohort study with longitudinal follow-up of 5-year cancer survivors diagnosed before age 21 years between 1970 and 1999 from 25 North American centers.¹¹ Each site’s institutional review board approved the study, and participants provided consent. Eligibility for this analysis was limited to participants (n = 11,337) who completed a questionnaire between 2014 and 2016. The CCSS has completed 3 trials aimed at increasing cardiac (ECHOS¹²), breast (EMPOWER¹³), and skin (ASK¹⁴) surveillance; participants were excluded from the analysis for the outcome evaluated in that study. Participants who developed a SMN or grade 3-4 (severe to life-threatening) heart failure based on the Common Terminology Criteria for Adverse Events (CTCAE) version 4.03¹⁵ prior to questionnaire completion were excluded from the analysis for the outcome that they developed.

Cancer and Cardiac Surveillance in High-Risk Survivors

For survivors at high risk for cardiac dysfunction or breast (females only), colorectal, or skin cancer, we assessed adherence to COG’s LTFU Guidelines, version 4.0 (www.survivorshipguidelines.org)⁶ published in 2013 (Data Supplement). For cardiac screening, survivors were asked, “When was the last time you had an echocardiogram (ultrasound of the heart to look at the heart muscles and heart valves) or a MUGA [multiple-gated acquisition] scan?” For breast cancer, survivors were asked, “When was the last time you had a mammogram?” and “When was the last time you had a breast MRI [magnetic resonance imaging]?” Both mammogram and breast MRI were recommended, whereas only mammogram was recommended in the previous guidelines (version 3.0, 2008). For colorectal cancer, survivors were asked, “When was the last time you had a sigmoidoscopy or colonoscopy to view the colon for signs of cancer or other problems?” For skin cancer, survivors were asked “When was the last time you had a skin exam for signs of cancer by a health care provider?” For each investigation, participants selected “never,” “less than 1 year ago,” “1-2 years ago,” “more than 2 but less than 5 years ago,” “5 or more years ago,” “I had one but I don’t recall when,” or “I don’t know if I ever had one.” For each guideline, survivors were classified as “adherent” if they completed the test within the period recommended in the Data Supplement.

Cancer Screening in Average-Risk Survivors

Survivors not at high risk for breast or colorectal cancer are advised to adhere to ACS screening recommendations (Data Supplement) for the general (average-risk) population.⁷ We assessed ACS guideline adherence among survivors at average risk for developing colorectal, breast (females only), and cervical (females only) cancer. Adherence was compared with a sex-, race- and age-adjusted cohort of siblings and with age- and race-matched aggregate population data from the 2015 NHIS,¹⁶ a cross-sectional survey of American health status and health care use.

Predictors of Screening and Surveillance

Sociodemographic data and surveillance practices were captured in the questionnaire, whereas cancer and treatment data were extracted from medical records. SCP receipt by the survivor and PCP was reported by the survivor. Survivors were defined as not having a SCP if they answered “no” when asked, “Do you currently have a cancer survivorship care plan and/or a summary of treatment for your cancer?” PCPs were defined as not having a SCP if the survivor answered “no” or “I don’t have a primary care doctor” when asked, “Does your local or primary care doctor have a copy of your cancer survivorship care plan and/or a summary of your treatment for your cancer?” Chronic health conditions were classified using the CTCAE version 4.03.¹⁵ Adverse health status was determined using 6 previously defined domains (general health, mental health, cancer-related pain, cancer-related anxiety, functional status, and activity limitations).^17-19 Cigarette and alcohol use were also analyzed.^20,21 Siblings were asked the same questions except for those addressing cancer-related pain and anxiety.

Change in Adherence to Breast, Colorectal, and Cardiac Surveillance Over Time

To assess changes in adherence to COG guidelines between questionnaires from 2002-2003, 2007-2009, and 2014-2016, survivors at high risk for a given outcome (Data Supplement) were identified at each time point. Using the COG guidelines current at the time of each questionnaire, survivors were classified as adherent to breast, colorectal, and cardiac surveillance if they completed the recommended surveillance for that outcome within the recommended time. Changes in skin surveillance adherence were not assessed because prior questionnaires did not assess this outcome.

Statistical Analysis

Descriptive statistics were calculated for risk group assignment and demographic, disease, and health status variables. Adherence was reported as a percentage. Adherence was compared between survivors and the general population using the Wald test. Survivors’ adherence between questionnaires and with siblings was compared using a log-binomial regression model. The generalized estimating equation was used to account for potential within-family correlation of survivors and siblings and correlation of multiple measurements within the same survivor. The prevalence ratio (PR) for adherence was estimated for each demographic and health status variable and compared in multivariable regression models. Variables with a P value < .2 in the univariable analysis were included in the multivariable analysis. Because of the small sample size in the high-risk breast analysis, covariables from the univariable analysis were further selected through backward elimination. Multivariable analysis results were used to determine predictors of adherence. Independent variable collinearity was evaluated by examining variance inflation factors and tolerance. Sampling weights were used to account for undersampling of acute lymphoblastic leukemia survivors. Variables that were highly correlated were not included in the same model. Analysis was completed with SAS version 9.4 (SAS Institute, Cary, NC).

RESULTS

Cohort Characteristics

Of the 18,043 survivors in the CCSS cohort, 11,337 (62.8%; 5,916 females) completed the questionnaire, along with 2,146 (1,245 females) siblings. Table 1 displays demographic, disease, and treatment data, and the Data Supplement displays health status data. Median age at primary cancer diagnosis was 7 years (range, 0-20.9, years), and the time from primary diagnosis to questionnaire completion was 29 years (range, 15-47 years). SCP possession was reported by 27% of survivors, and 20% reported PCP possession of their SCP. No differences existed in SCP possession by sex. Compared with survivors treated between 1970 and 1979, survivors treated between 1980 and 1989 (PR, 1.11; 95% CI, 1.00 to 1.23; P = .043) and between 1990 and 1999 (PR, 1.32; CI, 1.16 to 1.49; P < .001) were more likely to have a SCP. ECHOS¹² (n = 398), ASK¹⁴ (n = 728), and EMPOWER¹³ (n = 162) participants were excluded from the cardiac, skin, and breast analyses, respectively. Nonresponders were more likely to be male, younger at diagnosis, and have had leukemia (Data Supplement).

TABLE 1.

Demographic, Diagnostic, and Treatment Data of Survivors and Siblings

Cancer and Cardiac Surveillance in High-Risk Survivors

There were 625, 1,070, 5,125, and 4,220 survivors at high risk for developing breast cancer, colorectal cancer, skin cancer, or cardiac dysfunction respectively (Table 2). Adherence to breast, colorectal, skin, and cardiac surveillance was 12.64% (CI, 10.0% to 15.3%), 37.0% (CI, 34.1% to 39.9%), 22.3% (CI, 21.2% to 23.4%), and 41.4% (CI, 40.1 to 42.7), respectively (Data Supplement). In multivariable analyses, survivor-reported SCP possession was associated with increased adherence to breast (22.3% v 8.1%; PR, 2.52; CI, 1.59 to 4.01), skin (34.8% v 23.0%; PR, 1.16; CI, 1.01 to 1.33), and cardiac (67.0% v 33.1%; PR, 1.73; CI, 1.55 to 1.92) surveillance. PCP SCP possession was associated with increased adherence to skin surveillance (39.6% v 23.2%; PR, 1.24; CI, 1.08 to 1.43). Having a check-up related to past cancer in the past 2 years (regardless of location or provider) increased breast (PR, 7.94; CI, 1.99 to 31.74), skin (PR, 1.50; CI, 1.28 to 1.76), and cardiac (PR, 1.58; CI, 1.39 to 1.80) surveillance. Visiting a doctor more than 5 times in the past 2 years increased skin (PR, 1.47; CI, 1.28 to 1.69) and cardiac (PR, 1.06; CI, 0.99 to 1.14) surveillance. Visiting a special cancer survivorship clinic in the past 2 years increased cardiac surveillance (PR, 1.16; CI, 1.04 to 1.30), but visiting a cancer specialist did not increase adherence to any of the guidelines. There were no differences in adherence by sex. Table 3 and the Data Supplement show the relationship between other factors and adherence.

TABLE 2.

Risk Group Assignment of Childhood Cancer Survivors and Siblings

TABLE 3.

Multivariable Analyses of Predictors of Adherence to COG-Recommended Surveillance for High-Risk Survivors

Change in Adherence to Breast, Colorectal, and Cardiac Surveillance Over Time

Among high-risk survivors, adherence increased from 14.3% to 41.0% (P < .001) for colorectal and from 22.4% to 38.5% (P < .001) for cardiac surveillance and decreased from 37.9% to 13.1% (P < .001) for breast surveillance between 2007 and 2016 (Fig 1).

FIG 1.

Changes in adherence to the Children’s Oncology Group’s high-risk screening guidelines 2003, 2007, and 2016. (^a) When breast cancer screening was assessed in 2003, 2007, and 2016, there were 311, 339, and 382 eligible survivors at each time point, respectively. In 2003, 2007, and 2016, 118, 29, and 50 survivors were adherent to breast cancer screening, respectively. Between 2013 and 2016, adherence statistically decreased (P < .001). (^b) When colorectal cancer screening was assessed in 2003, 2007, and 2016, there were 468, 663, and 886 eligible survivors at each time point, respectively. In 2003, 2007, and 2016, 67, 189, and 363 survivors were adherent to breast cancer screening, respectively. Between 2013 and 2016 adherence statistically increased (P < .001). (^c) When cardiac dysfunction screening was assessed in 2003, 2007, and 2016, there were 1,386, 1478, and 1,545 eligible survivors at each time point, respectively. In 2003, 2007, and 2016, 310, 423, and 594 survivors, respectively, were adherent to cardiac dysfunction screening. Between 2013 and 2016, adherence statistically increased (P < .001).

Cancer Screening in Average-Risk Survivors

There were 627, 5,630, and 596 survivors at average risk for developing breast, cervical, or colorectal cancer, respectively (Table 2). Adherence to ACS breast, cervical, and colorectal screening was 57.1% (CI, 53.2% to 61.0%), 83.6% (CI, 82.7% to 84.5%), and 68.5% (CI, 64.7% to 72.2%), respectively. In multivariable analyses, survivor SCP possession was not associated with increased adherence to breast, cervical, or colorectal screening. PCP SCP possession was associated with increased adherence to breast (PR, 1.28; CI, 1.13 to 1.45) and colorectal (PR, 1.12; CI, 1.04 to 1.21) screening. Table 4 and the Data Supplement show the relationship between other factors and adherence. Compared with NHIS general population estimates, survivors were more adherent to ACS cervical (PR, 1.08; CI, 1.07 to 1.10; P < .001) and colorectal (PR, 1.15; CI, 1.09 to 1.22; P < .001) screening. Compared with siblings, survivors were less adherent to ACS cervical screening (PR, 0.95; CI, 0.93 to 0.97; P < .001; Data Supplement).

TABLE 4.

Multivariable analysis of predictors of adherence to ACS recommended screening for average-risk survivors

DISCUSSION

We evaluated the surveillance and screening practices of 11,337 childhood cancer survivors. In other groups at high risk for malignancy, surveillance decreases mortality,^22,23 and this is assumed to also be true for at-risk childhood cancer survivors. We demonstrated that fewer than half of high-risk survivors receive the recommended SMN and cardiac surveillance, which likely exposes them to preventable morbidity and mortality. SCPs are intended to improve adherence by providing follow-up information and by facilitating the transition from cancer treatment to survivorship and from pediatric to adult care. Despite the IOM’s recommendation⁹ and the Commission on Cancer’s mandate,¹⁰ few survivors and PCPs have a SCP. Survivors treated after 1990 were more likely to have a SCP, suggesting that dissemination is increasing, potentially because of these recommendations. In 2017, the Commission on Cancer lowered their standard regarding the percentage of survivors that must receive a SCP from 75% to 50%¹⁰ to reflect the reality that many centers lack the resources to comply. A recent systematic review evaluating SCPs concluded that “existing research provides little evidence that SCPs improve health”.^24(p1) In contrast, SCP possession by high-risk survivors in our analysis was associated with increased breast, skin, and cardiac surveillance. It is uncertain whether SCP possession leads to adherence or whether SCP possession is a marker of survivors who are focused on their health and thus likely to adhere to preventative health practices, including surveillance.

Despite the current suboptimal adherence to COG guidelines, adherence improved between 2003 and 2016. This was most evident for colorectal (from 14.3% to 41.0%) and cardiac surveillance (from 22.4% to 38.5%). Reasons for these increases are likely multifactorial. In 2012, only 12% of general internists²⁵ and 9% of family doctors²⁶ felt at least “somewhat familiar” with care guidelines for childhood cancer survivors. Physician awareness of COG guidelines may be increasing as the time from initial publication grows. As the number of patients with SCPs increases, PCPs may be becoming more comfortable with SCPs. Survivors can receive cancer-related care from PCPs or survivorship clinics. Survivorship clinics may be more familiar with guidelines and may provide additional education. In the 2 years preceding the questionnaire, 16.1% of survivors attended a survivorship clinic, but attendance only increased cardiac surveillance. Given that survivorship clinics use increased resources compared with PCP clinics,²⁷ we must revaluate survivorship care to create cost-effective programs that meaningfully affect survivors’ health.

Our study did not assess barriers to obtaining surveillance tests. ECHOS¹² participants reported not obtaining echocardiograms because of a lack of time, forgetting, a perception that screening is not important, concerns about insurance and cost, and because their physicians did not recommend testing. Many survivors report difficulty transitioning from pediatric to adult care.²⁸ Within the 2 years preceding the survey, 54.7% of survivors had a routine check-up related to their past cancer, and 92.5% had a physician visit, suggesting that survivors receive care, but that surveillance does not consistently occur at these visits. Previously identified psychosocial barriers include poor mental health, lower socioeconomic status, and lower educational level²⁸; we identified variable effects of these factors on adherence. Skin examination adherence was particularly poor (22.3%). Belief in self-capacity to screen, fear of body shaming, fear of facing a SMN diagnosis, and rejection of the benefit of examination may contribute to the poor uptake of skin examinations. However, given the ease of completing a skin examination, poor uptake suggests a lack of awareness of the need for surveillance, rather than concerns regarding the invasiveness, scheduling, or time commitment of the investigation. Patients and PCPs may also find the recommendations overly burdensome, given the high frequency of investigations required.

The CCSS has conducted 3 trials aimed at increasing surveillance. ECHOS showed that telephone counseling addressing barriers to surveillance increased echocardiogram adherence.¹² EMPOWER showed that mailed information coupled with motivational telephone interviewing increased mammography but not MRI adherence.¹³ ASK¹⁴ is aimed at increasing skin cancer surveillance and is ongoing. Web-based interventions have successfully altered survivors’ physical activity levels and health-related quality of life, and similar platforms exist to enhance late-effect awareness and surveillance.²⁹ Additional studies using these approaches, such as EMPOWER II (ClinicalTrials.gov identifier: NCT03435380), are underway. These strategies may improve surveillance and the health of survivors; however, scaling them to the larger community of survivors, many of whom are not engaged with survivorship programs, may be difficult.

There was better adherence to ACS than to COG recommendations, with 57.1%, 83.6%, and 68.5% adherent to ACS breast, cervical, and colorectal screening, respectively. There may be greater awareness of ACS guidelines because they are population-based guidelines that PCPs use more regularly. We found that survivors were less adherent to breast screening than the general population and less adherent to cervical screening than siblings. This is possibly because survivors avoid testing to escape worrying about the potential of an SMN.³⁰

Our study has several limitations. First, we used self-reported data regarding completion of surveillance and receipt of SCPs. There are no studies assessing the validity of childhood cancer survivors’ self-reported screening practices. In other populations, self-reported history of cancer screening has been validated,³¹ suggesting that the same is true for childhood cancer survivors. Many oncologists send SCPs directly to PCPs, which may result in survivors under-reporting PCPs’ SCP possession. Second, when breast surveillance was assessed in 2003, only mammogram was recommended, whereas now both mammogram and breast MRI are recommended, making comparison of change over time difficult. Having to obtain MRIs may explain why adherence dropped by 24.9% between 2003 and 2016. Many females (35.7%) had an MRI or mammogram within the recommended period, but were classified as nonadherent. In a model of adolescent Hodgkin lymphoma survivors, breast cancer mortality at 75 years of age was 16.3% with mammography and 15.4% with mammography and MRI, suggesting that having both tests may be unnecessary.⁴ Physicians may not recommend both tests if they feel that completing just one is adequate. It may also be difficult to obtain MRIs. Third, the CCSS regularly corresponds with participants, and this may increase late-effect awareness and surveillance compared with the general population of survivors, leading to an overestimation of adherence. Adherence may also differ between responders and nonresponders (Data Supplement). Fourth, survivors who participated in ASK, EMPOWER, or ECHOS were not included in the analysis for the outcome targeted in the study they participated in, but were included in the other analyses. It is possible that a “halo” effect exists, and participating in a study may increase adherence to guidelines not targeted in that study. In addition, adherence may be underestimated because of the use of strict adherence definitions. A survivor who had a test just beyond the recommended time would be classified as nonadherent, yet a short delay in testing is unlikely to adversely affect health outcomes. Finally, CCSS survivors were treated in the 1970s-1990s, and their health habits and adherence may not be representative of survivors treated more recently.

In summary, we demonstrated that survivors at increased risk for SMNs or cardiac dysfunction report suboptimal adherence to recommended surveillance for these outcomes. Survivor SCP possession was shown to be generally effective at increasing adherence; however, few survivors have a SCP. Greater awareness of SCPs may improve their value. Few survivors attended specialized survivorship clinics, and attendance did not improve adherence. These data underscore the importance of improving survivorship care, including increased SCP dissemination and developing and testing new interventions. Additional research assessing barriers to adherence is also needed to assist in developing programs for increasing adherence.

AUTHOR CONTRIBUTIONS

Conception and design: Adam P. Yan, Tara O. Henderson, Kevin C. Oeffinger, Melissa M. Hudson, Kirsten K. Ness, Leslie L. Robinson, Gregory T. Armstrong, Paul C. Nathan

Financial support: Leslie L. Robinson, Gregory T. Armstrong

Administrative support: Wendy M. Leisenring, Leslie L. Robinson, Gregory T. Armstrong

Provision of study materials or patients: Melissa M. Hudson, Joseph P. Neglia, Leslie L. Robinson, Gregory T. Armstrong

Collection and assembly of data: Adam P. Yan, Joseph P. Neglia, Leslie L. Robinson, Gregory T. Armstrong, Paul C. Nathan

Data analysis and interpretation: Adam P. Yan, Yan Chen, Tara O. Henderson, Kevin C. Oeffinger, Melissa M. Hudson, Todd M. Gibson, Joseph P. Neglia, Wendy M. Leisenring, Kirsten K. Ness, Jennifer S. Ford, Gregory T. Armstrong, Yutaka Yasui, Paul C. Nathan

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Adherence to Surveillance for Second Malignant Neoplasms and Cardiac Dysfunction in Childhood Cancer Survivors: A Childhood Cancer Survivor Study

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).

Tara O. Henderson

Research Funding: Seattle Genetics

Other Relationship: Seattle Genetics

Melissa M. Hudson

Consulting or Advisory Role: Oncology Research Information Exchange Network, Princess Máxima Center, SurvivorLink

No other potential conflicts of interest were reported.