Background

The number of American cancer survivors continues to grow due to earlier detection and ever-improving therapeutic approaches.¹ Although many survivors overcome the initial acute threat of cancer-related mortality, they face an increased risk of second malignancies, cardiovascular disease, and accelerated functional decline—compounded by concerns about recurrence.²

In response, NCCN established a dedicated Survivorship Panel, joining >50 other NCCN Guidelines Panels that span the cancer continuum from prevention to palliative care.³ Since its formation in 2012, the NCCN Survivorship Panel has published guidelines addressing a broad range of survivorship care topics.^4–16 Recognizing the critical role of diet in managing comorbidities and late effects, one of the earlier published guidelines focused on nutrition and weight management,⁶ reinforcing extant recommendations from the World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) and the American Cancer Society.^17–19 These guidelines emphasize weight management (given that 70.3% of cancer survivors have body mass indexes [BMI] ≥25 kg/m²)²⁰ and advocate for increased consumption of fruits, vegetables, and whole grains while limiting refined grains and sugars (particularly sugar-sweetened beverages [SSBs]) and red and processed meats (RPMs). Subsequent updates to survivorship care guidelines have remained aligned with these nutrition-related guidelines^16,19 as well as the increased focus on dietary patterns (rather than individual foods) and the instruments to assess them, such as the Healthy Eating Index-2020 (HEI-2020).^21,22

However, it remains unclear to what extent these guidelines are followed. What are cancer survivors eating? What recommendations are met (or not)? Are there subgroups of survivors who require targeted dietary interventions? Addressing these questions will provide valuable insights for health care providers striving to improve care for the expanding population of cancer survivors who need dietary guidance. During the recruitment efforts for a nationwide diet and exercise intervention trial among cancer survivors,²³ an opportunity arose to explore these questions. The findings are reported herein.

Methods

A cross-sectional analysis was conducted among cancer survivors who self-referred or responded to recruitment solicitations from state and hospital-based cancer registries to enroll in a national web-based intervention trial. The intervention was designed to promote a healthy diet, regular exercise, and safe weight loss (ClinicalTrials.gov identifier: NCT04000880) among survivors aged ≥50 years with a history of 9 obesity-related cancers, each with a 5-year cancer-free survival rate of ≥70%. Eligible cancers included early-stage multiple myeloma and non-Hodgkin lymphoma, localized renal and ovarian cancer, and locoregional cancers of the colorectum, prostate, endometrium, thyroid, and female breast. Eligibility was further restricted to survivors who had completed primary cancer treatment (including active surveillance for prostate cancer or continuing chemotherapy as needed for nonsolid tumors). Data regarding cancer diagnoses were obtained from the registries or verified with providers for self-referred participants.

Cancer survivors self-reported data on cancer treatment, demographics (age, sex, race, ethnicity, education, and marital and employment status), as well as height and weight (which were used to calculate BMI because a BMI of 25–50 kg/m² was an eligibility criterion for trial participation). Survivors’ physical addresses were geocoded to obtain census tracts and blocks, which were then linked to the US Department of Agriculture Economic Research Service to classify rural-urban status, food access, and area deprivation.²⁴ Rural-Urban Commuting Area (RUCA) codes defined urban (codes 1–3) versus rural (codes 4–10) residence.²⁵ The Food Access Research Atlas categorized areas based on access to healthy foods, with higher access defined as at least one-third of the population having supermarkets, supercenters, or large grocery stores within 1 mile for urban settings or 10 miles for rural settings, and lower access applied to areas where less than one-third had similar availability.²⁶ The Area Deprivation Index (ADI) classified areas by socioeconomic disadvantage, with higher deprivation defined as ADI ≥50% and lower deprivation as ADI <50%.^27,28

Dietary intake was assessed using 2 nonconsecutive 24-hour recalls—one weekday and one weekend day—conducted by trained telephone interviewers using a multipass method and the Nutrition Data System for Research software.²⁹ A research dietitian reviewed the recalls for quality assurance, implementing corrective actions as needed. Output was then averaged, with diet quality assessed using the HEI-2020 (see Supplementary Table S1 for the scoring methodology, available online in the supplementary materials).²² Total and component HEI-2020 scores, along with output for SSBs and RPMs, were analyzed descriptively for the overall sample. Differences were explored by cancer type—breast, colorectal, urogenital (cervical, endometrial, ovarian, prostate, renal), and other (multiple myeloma, non-Hodgkin lymphoma, thyroid)—as well as by subgroups dichotomized by clinical and demographic factors. These subgroups included treatment history (eg, chemotherapy vs no chemotherapy, radiation therapy vs no radiation therapy), time since diagnosis (<5 vs ≥5 years), BMI category (overweight [25 to <30 kg/m²] vs obese [30–50 kg/m²]), age (<65 vs ≥65 years), race/ethnicity (non-Hispanic White vs racial/ethnic minority [including Non-Hispanic Black, Asian/Pacific Islander, Native American, mixed race, Hispanic]), sex (male vs female), marital status (married/stable union vs not married [single, divorced, separated, widowed]), education (high school diploma or less vs some college or more), employment status (employed vs unemployed, which included mostly retirees but also individuals with disabilities, and students), and area of residence (as defined earlier). Data normality was assessed using the Kolmogorov-Smirnov test, and subgroup analyses for nonparametric data were performed using the Kruskal-Wallis rank sum test, with post hoc comparisons conducted using Dwass-Steel-Critchlow-Fligner testing (α=.05).³⁰ Given the exploratory nature of this investigation, no adjustments were made for multiple comparisons. All statistical analyses were performed using SAS 9.4 (SAS Institute, Inc.).

Results

From March 2020 to September 2023, a total of 818 cancer survivors across 31 US states enrolled in the trial and completed the set of dietary recalls. Table 1 summarizes the sample characteristics, reflecting that participants were racially and ethnically representative of US adults aged ≥50 years,³¹ though most were female and more highly educated. Most survivors were within 5 years of diagnosis and approximately half were diagnosed with breast cancer. Most were treated with radiation therapy and approximately 40% received chemotherapy. Most participants had a BMI ≥30 kg/m² (vs 25 to <30 kg/m²). Nearly one-quarter resided in rural areas, and most lived in areas with an ADI that exceeded the 50th percentile, indicating higher socioeconomic disadvantage.

Table 2 presents overall diet quality and component scores for survivors in this sample, compared with ideal values and referent intakes among Americans of comparable age (with general population means and 95% confidence intervals [CIs] shown in footnotes).^32,33 The mean overall HEI-2020 score among survivors was 51.6 points (95% CI, 50.72–52.38)—well below the optimal score of 100, approximately 10 points lower than the average for comparably aged Americans, and appears significantly lower than population norms (as indicated by nonoverlapping 95% CIs). Seafood, plant protein, fruits, vegetables, and dairy were significantly underconsumed, whereas refined grains were overconsumed. Cancer survivors approached guideline levels for whole grains and sodium, added sugar, and saturated fat limitations.

Table 3 presents HEI-2020 subgroup comparisons by various clinical characteristics. Few differences were detected, with no differences observed by cancer type or treatment (data not shown). However, survivors with obesity had significantly lower overall diet quality compared with those who were overweight, with lower consumption of fruits and dairy products.

Figure 1 depicts the HEI-2020 radar plot for cancer survivors, along with comparative radar plots for survivors by subgroups that exhibited significant differences in at least 4 HEI components.

View Full Size

Healthy Eating Index radar plots of (A) cancer survivors versus the ideal, and comparisons by (B) weight status, (C) racial/ethnic status, (D) age, (E) education, and (F) residence in area of deprivation.

Abbreviation: NHW, non-Hispanic White.

Citation: Journal of the National Comprehensive Cancer Network 23, 6; 10.6004/jnccn.2025.7012

Table 4 presents diet quality comparisons by sociodemographic factors, revealing several more significant subgroup differences. Overall diet quality and specific component scores differed significantly by age, with survivors aged <65 years reporting lower intakes of fruits, vegetables, and whole grains and higher consumption of saturated fat compared with those aged ≥65 years. However, younger survivors reported significantly higher dairy product intake. Comparisons between Non-Hispanic White survivors and self-identified racial/ethnic minorities suggested that although White survivors had significantly lower fruit and total protein intake and higher saturated fat consumption, minority survivors had significantly lower whole grain and dairy intake but higher consumption of added sugars.

Male cancer survivors reported significantly higher intakes of protein and lower intakes of whole fruits than females. Unmarried survivors had significantly lower intakes of dairy, seafood, and plant protein foods compared with their married counterparts. Overall diet quality was significantly lower among survivors with a high school diploma or less, with lower component scores for total and specific vegetables and higher intakes of added sugars.

Few differences were observed between cancer survivors residing in rural versus urban areas, though rural residents reported significantly lower intakes of specific vegetables. In contrast, survivors living in higher- versus lower-deprivation areas had significantly poorer diets, as indicated by lower overall diet quality and reduced intakes of fruits and dairy foods. No differences were observed by employment status or between cancer survivors residing in areas with higher versus lower food access (data not shown).

Table 5 presents mean intake (ounces/week) of RPMs and SSBs, along with measures of variance. Footnotes also report population norms, which appear significantly higher than those of the sample, as indicated by nonoverlapping 95% CIs.^34,35 Several significant subgroup differences were observed for both RPM and SSB consumption. RPM intake was significantly higher among survivors within 5 years of diagnosis, those with urogenital cancers, males, rural residents, and individuals who were less educated. SSB consumption was significantly greater among survivors who were less educated, employed, younger, unmarried, of minority status, and living in areas of high deprivation.

Discussion

This study represents one of the largest cross-sectional analyses to explore diet quality in a national sample of survivors with clinically confirmed histories of cancer. The most remarkable finding is how poor their diets are compared with guidelines and with the general population. The mean HEI-2020 score of 51.6 (out of 100) falls nearly 10 points below the norm for older adults,³² a substantial gap given that score differences of just 5 to 6 points are considered clinically meaningful.³⁶ Moreover, comparisons between HEI-2020 scores from this study and those generated from the 2017–2018 National Health and Nutrition Examination Survey (NHANES) on the general population reinforces that diet quality among these cancer survivors was significantly lower, as indicated by nonoverlapping 95% CIs.

A higher HEI score of 55.6 was previously reported by Lee et al³⁷ in an analysis of 1,971 cancer survivors identified from NHANES (2005–2016); however, their study differs from this analysis in 3 key respects: (1) their cancer diagnoses were not clinically confirmed and included noninvasive skin cancers and lung cancer, with far less representation of obesity-related cancers—whereas our study focused exclusively on survivors of obesity-related cancers with BMI ≥25 kg/m² at the time of survey; (2) their sample was older (59% aged ≥65 years vs 42% in our study); and (3) they used a different HEI index (2015 vs 2020). Nonetheless, Lee et al³⁷ “sounded the alarm” that cancer survivors had poor diets, with diet quality decreasing significantly over time (P<.007), causing further concern. Their report builds upon an earlier NHANES analysis (1999–2010) by Zhang et al,³⁸ which compared the diet quality of 1,533 self-reported cancer survivors versus 3,075 age-, sex-, and race/ethnicity-matched controls and found significantly poorer diet quality among survivors (P=.03), despite a mean score difference of only 1.1 points at that time. Our recent data (including the 9.4-point difference in scores between our sample and population norms)³² align with these reports and suggest a heightened need to address this problem with interventions that promote better diet quality together with weight loss.

These earlier studies also identified disparities in diet quality across various clinical and sociodemographic factors, with findings that either contradict or corroborate the current results.^37,38 For example, both previous studies reported significantly higher diet quality among breast cancer survivors, whereas we found no differences by cancer type. However, our sample did not include lung cancer survivors, a subgroup renowned for poor diet quality.^37,38 Additionally, differences detected in subgroups dichotomized by time elapsed since diagnosis and the null results by treatment type observed in this current study are unsupported by previous studies, because these data were either not collected or structured differently in previous research.^37,38 However, both this study and Lee et al³⁷ found that cancer survivors with obesity reported significantly lower diet quality than those with lower body weight. Diet quality disparities were more pronounced across sociodemographic factors, with findings of this study corroborating previous reports that younger survivors and those with less education had significantly lower HEI scores.^37,38 Although overall diet quality did not differ by racial/ethnic minority status, significantly lower HEI component scores for whole grains, dairy, and added sugars were observed. Moreover, advances in geocoding enabled the current study to detect lower diet quality among cancer survivors residing in areas of higher versus lower deprivation—a novel finding that supports the premise that zip code supersedes genetic code in determining health.^39,40 Surprisingly, further analysis found no differences based on food access and few differences between rural and urban residents.

One of the few differences detected between rural and urban cancer survivors was a significantly higher consumption of RPMs among rural residents. Compared with the WCRF/AICR guidelines, which advocate ≤12 ounces per week for women and ≤18 ounces per week for men,¹⁷ mean intakes in this study exceeded these recommendations, and were significantly higher among males, individuals with less education, and survivors closer to diagnosis. This finding, along with higher diet quality scores among longer-term cancer survivors, suggests that cancer survivors seek and may adapt to healthier lifestyles over time. Nonetheless, barriers such as cultural preferences and limited access to whole-food, plant-based protein alternatives should be explored to guide dietary interventions. Despite exceeding recommended RPM intake levels, cancer survivors in this study consumed and average of 15.3 ounces per week—substantially lower than the 25.9 ounces per week consumed by the average American.³⁴ The difference was even more pronounced for SSBs, with cancer survivors consuming just 4.7 ounces per week compared with the national average of 21.6 ounces.³⁵ Hence, even though greater scientific evidence supports a direct causal link between RPM consumption and cancer risk compared with SSB intake,^41,42 the popular messaging that “sugar causes cancer” appears to have had a greater influence on dietary behavior.

Given that both diet quality and avoidance of RPMs and SSBs are associated with improved health outcomes among cancer survivors—including lower risk of chemotherapy-induced peripheral neuropathy,⁴³ physical decline,⁴⁴ and cancer-specific and all-cause mortality^45–49—it is important for survivors to receive dietary guidance. Findings from this study suggest that although survivors’ diets may include myriad deficits, suboptimal intakes are particularly notable for fruits, vegetables, dairy, and protein—particularly from fish, seafood, and plant sources (eg, legumes). Additionally, excessive consumption of refined grains presents a significant issue. Thus, if health care providers are to provide dietary guidance, prioritizing these areas is critical, rather than focusing on reducing added sugar and SSBs—areas in which most cancer survivors already appear to be adhering to recommendations. Furthermore, interventions should target those at highest risk, specifically cancer survivors who are more proximal to diagnosis and who meet at least one of the following criteria: (1) BMI ≥30 kg/m², (2) age <65 years, (3) education level of high school diploma or less, and (4) residence in an area of higher socioeconomic deprivation.

The current study has limitations that may constrain the generalizability of its findings. First, because participants were self-selected for a weight-loss trial, the results may overestimate dietary interest and underestimate the challenges faced by less motivated populations. Future research should include a broader, more representative sample, including individuals with normal weight and who are younger (ie, <50 years of age), because their diet quality remains largely unknown. Multiple testing is another limitation of this exploratory study, as is subgroup testing in samples that may offer inadequate power, such as comparisons among survivors from high versus low ADIs. Finally, a more complete dietary profile could be accomplished through recalls conducted over multiple days (rather than just 2). However, 2-day recalls often serve as the norm,^37,38 and they were conducted using state-of-the-art methods and an advanced database.⁵⁰ The fact that participants in our sample had excess adiposity and expressed an interest in weight loss highlights a critical need to leverage that interest to also promote a healthy diet. This interest may also extend to those seeking pharmacologic interventions, such as GLP-1 agonists.

Conclusions

In this national cross-sectional study of dietary intake among cancer survivors with a history of obesity-related cancer and with overweight or obesity, we found that although survivors adhered more closely to guidelines in some areas (eg, added sugars and SSBs), their overall diet quality was poor, and much lower than that of in the general population. Providers can use these findings to prioritize counseling on increasing fruits, vegetables, plant-based proteins, and seafood, particularly for survivors at higher risk for common comorbidities and second cancers, such as those with obesity. This need is greatest among survivors who are younger, more proximal to diagnosis, living with obesity, of lower educational attainment, and residing in areas of higher deprivation.