Relationship Between Individual Health Beliefs and Fruit and Vegetable Intake and Physical Activity Among Cancer Survivors: Results from the Health Information National Trends Survey

Abstract

Purpose:

The aim of this study was to describe diet and physical activity (PA) behaviors and health beliefs among cancer survivors and identify potential differences between adolescent and young adult (AYA) and adult/older cancer survivors.

Methods:

Cancer survivors (n = 1864) participating in the Health Information National Trends Survey (HINTS) provided responses regarding diet and PA and selected health beliefs related to general health and cancer (self-efficacy, attitudinal belief, normative belief, risk belief, intention, and self-regulation). Health belief associations with diet and PA were assessed using adjusted logistic regression models, and multiple linear regression was used for a computed Modified American Cancer Society Adherence score (0–10, higher score indicates higher adherence to recommendations); age at diagnosis was evaluated as a potential effect modifier.

Results:

Health behaviors between AYA and adult/older were not significantly different; a greater percent of AYA met fruit and resistance PA recommendations. Higher health self-efficacy was associated with meeting aerobic PA recommendations (odds ratio [OR]: 1.71; confidence interval [95% CI]: 1.13–2.60; p = 0.01). Higher intention was inversely related to meeting vegetable recommendation (OR: 0.58; 95% CI: 0.35–0.97; p = 0.04). Self-regulation was associated with higher odds of meeting each recommendation. Self-efficacy and self-regulation were associated with greater adherence (β = 0.52 ± 0.16, p = 0.001; β = 1.21 ± 0.24, p < 0.0001, respectively). Age at diagnosis was not an effect modifier.

Conclusion:

Health behaviors and beliefs among AYA and adult/older are similar. Self-efficacy and self-regulation through engagement with a mobile app support adoption of diet and PA recommendations among HINTS respondents. Future interventions should consider mechanisms to promote self-efficacy and self-regulation to maximize diet and PA behaviors in cancer survivors.

Introduction

It is estimated that nearly 1.9 million new cancer cases will be diagnosed in 2021, and cancer remains the second leading cause of death in the United States.¹ The American Cancer Society (ACS) Nutrition and Physical Activity Guidelines for Cancer Survivors provide an evidence-based framework for promoting health behaviors for cancer prevention.² Current guidelines include maintaining a healthy body weight, engaging in frequent physical activity (PA), and following an eating pattern with high fruit and vegetable (FV) intake. Higher adherence to ACS diet and PA guidelines is associated with lower cancer mortality³ and may positively impact cancer treatment and survivorship outcomes.⁴ Despite this, it is estimated that fewer than half of all cancer survivors meet PA recommendations and even fewer meet FV recommendations.⁵

Adolescent and young adult (AYA) cancer survivors, defined as individuals diagnosed with cancer between the ages of 15 and 39 years,⁶ comprise 5% of the annual incident cancer cases in the United States.⁷ While metabolic disruptions are common long-term consequences of cancer treatment,⁸ AYA experience greater risk than individuals diagnosed at a later age.⁹ Adverse metabolic late effects are potentially preventable and modifiable through health behaviors such as diet and PA. The limited studies in the AYA population indicate inadequate intake of fruits and vegetables and low PA levels after cancer treatment.^10–14 In an age stratified analysis of cancer survivors, approximately half of AYA consumed ≥5 servings FV and even fewer were physically active; prevalence which was greater compared to adult/older cancer survivors.¹⁵

Attainment of diet and PA guidance and related behavioral goals for cancer survivorship require attention to health behavior theory and the role of related constructs in driving behavior change.¹⁶ Frequently used health behavior theories include Social Cognitive Theory, Transtheoretical Model, Health Belief Model, and Theory of Planned Behavior.^17–20 These theories house similar constructs, which can be categorized as self-efficacy, attitudinal belief, normative belief, risk belief, intention, and self-regulation,²¹ which may predict likelihood of engaging in cancer preventive health behaviors such as diet or PA.

The relationship between individual health beliefs and achieving current health behavior recommendations in a large nationally representative sample of cancer survivors has yet to be investigated. The aim of this research was to (1) describe diet and PA behaviors of cancer survivors responding to a nationally representative sample survey; (2) assess health beliefs of the sample; and (3) examine the association between individual health beliefs and meeting ACS recommendations. Differences in health behaviors, beliefs, and associations by age of cancer diagnosis (AYA vs. adult/older) were explored.

Methods

Study design and population

The National Cancer Institute (NCI) Health Information National Trends Survey (HINTS) is a nationally representative, cross-sectional, validated population-based survey of the U.S. noninstitutionalized adult population that is conducted annually via mail, telephone, and web.^22,23 Data sets and methodology documentation are publicly available.²⁴ Administration of HINTS surveys has been approved by the Westat Institutional Review Board; this analysis is deemed exempt by the University of Arizona Institutional Review Board.

Three HINTS waves were combined according to NCI guidance²⁵ to ensure a sufficient sample size for reliable estimates in a subpopulation of U.S. cancer survivors and account for the complex survey design. The analytical sample was limited to individuals with a self-reported history of cancer and age at cancer diagnosis. Those without a history of cancer (n = 10,174), a pediatric cancer diagnosis (<15 years; n = 18), or missing age at diagnosis (n = 71) were excluded. The final analytic sample included 1864 cancer survivors from the three waves (HINTS 5 Cycle 1 [2017; response rate: 32.4%; n = 478] HINTS 5 Cycle 2 [2018; response rate: 32.9%; n = 564] and HINTS 5 Cycle 3 [2019; response rate: 30.3%; n = 822]). Age at diagnosis was categorized into AYA (AYA; 15–39 years) and adult/older (≥40 years) following NCI standard.

FV intake

FV intake were assessed using the two following questions: “About how many cups of fruit (including 100% pure fruit juice) do you eat or drink each day?” and “About how many cups of vegetables (including 100% pure vegetable juice) do you eat or drink each day?” Approximate serving sizes for one cup of fruit or vegetables were provided to respondents. Response options included seven categorical responses ranging from none to >4 cups per day. Following ACS recommendations (consuming1.5–2 cups of fruit and 2.5–3 cups of vegetables daily²), responses were dichotomized. FV intake were analyzed separately.

Physical activity

PA was assessed using the following questions: “In a typical week, how many days do you do any physical activity or exercise of at least moderate intensity, such as brisk walking, bicycling at a regular pace, and swimming at a regular pace (do not include weightlifting)?” and “On the days that you do any physical activity or exercise of at least moderate intensity, how long do you typically do these activities?” (aerobic PA) and “In a typical week, outside of your job or work around the house, how many days do you do leisure-time physical activities specifically designed to strengthen your muscles such as lifting weights or circuit training (do not include cardio exercise such as walking, biking, or swimming)?” (resistance PA). Continuous responses were dichotomized to meeting recommendations (≥150 weekly minutes for aerobic PA and ≥2 days weekly for resistance PA) according to ACS guidelines.² Aerobic and resistance PA were analyzed separately.

Modified ACS Adherence score

Our Modified ACS Adherence score was derived from ACS Nutrition and Physical Activity Guidelines for Cancer Survivors² following recommended cut-points (Supplementary Table S1).²⁶ Individuals were assigned a score of 0, 1, or 2 for each of the three components of the guidelines: FV, aerobic and resistance PA, and body mass index (BMI). For FV and aerobic and resistance PA, two points were allocated for exceeding the recommendation, one point was allocated for meeting the recommendation, and zero points were allocated for not meeting the recommendation. For BMI, points were allocated as follows: two points for normal weight (18.5–24.9 kg/m²), one point for overweight (25.0–29.9 kg/m²), and zero point for obese (≥30.0 kg/m²). Points were allocated for equal weight and summed to create a score ranging from 0 to 10, where 10 was exceeding all recommendations; this score was used for analyses.

Health beliefs

Health beliefs were ascertained from the following: “Overall, how confident are you about your ability to take good care of your health?” (self-efficacy); “It seems like everything causes cancer” (attitudinal belief); “There's not much you can do to lower your chances of getting cancer” (normative belief); “How worried are you about getting cancer?” (risk belief); “There are so many different recommendations about preventing cancer, it's hard to know which ones to follow” (intention); and “In the past 12 months has a smartphone or tablet helped you track progress on a health-related goal such as quitting smoking, losing weight, or increasing physical activity?” (self-regulation). Smartphone apps serve as a modern approach to self-regulation²⁷ through integration of various health promotion techniques.²⁸ Responses were dichotomized from a 4-point Likert scale to high/low or agree/disagree consistent with previous analyses.^29,30 Self-regulation data were available only for participants who previously answered “yes” to having a smart phone (68.5% of the analytical sample). In HINTS 5 Cycle 3 (n = 822), the variable used for risk belief was available only for individuals who reported no history of cancer, resulting in missing data. Therefore, analyses of self-regulation were restricted to respondents with a smartphone and analyses of risk belief were restricted only to responses from HINTS 5 Cycles 1 and 2.

Covariates

Sociodemographic characteristics included age, sex, race/ethnicity, geographic area (rural vs. metropolitan designation),³¹ occupational status, annual household income, household size, marital status, and education. Cancer history included age diagnosed, diagnosis duration (continuous-calculated from age at survey completion and age diagnosed and categorized), family history of cancer, and cancer type. Other health characteristics included general health (excellent to poor), having a primary care provider, health insurance coverage, comorbid conditions, and BMI (kg/m²).³²

Statistical analysis

Cross sectional analyses of health behaviors and health beliefs in cancer survivors were completed using three merged waves of HINTS data (HINTS 5 Cycles 1–3), representing collection years of 2017–2019. Outcome variables were evaluated before analyses to verify were no significant differences between cycles or delivery mode. Provided survey weights were applied using the Jackknife repeated replication method for population level estimates.

Differences in sociodemographic characteristics, cancer history, and health characteristics stratified by age at diagnosis were assessed using chi-square tests or t-tests. Adherence to recommendations and individual health beliefs were compared by age at diagnosis using Wald design-based chi-squared tests of independence. Age at diagnosis and associations with meeting recommendations were assessed using multiple logistic regression models. For all cancer survivors, the association between each individual health belief and meeting FV and aerobic and resistance PA recommendations were evaluated separately using adjusted multiple logistic regression models. For consistency between previous analyses utilizing HINTS data,²⁹ health beliefs were treated as separate variables in all statistical models. Adjusted multiple linear regression models were used to assess association between health beliefs and Modified ACS Adherence score. Age at diagnosis (AYA vs. adult/older) was evaluated as a potential effect modifier in all analyses. Models were adjusted for age, sex, race/ethnicity, marital status, number in household, cancer type, years since diagnosis, regular primary care provider, health insurance status, smoking status, HINTS cycle, and BMI (except for in Modified ACS Adherence score models, which included BMI in the score). Analytical samples were restricted to complete cases. An alpha level of 5% was considered statistically significant. All analyses were completed in STATA 16.1 (StataCorp LLC, College Station, TX, USA).

Results

Of 1864 cancer survivors, 16% (n = 310) were AYA (Table 1). The majority of cancer survivors were female, non-Hispanic white, and dwelling in a metropolitan area. AYA were more likely to be female (p < 0.001), Hispanic or non-Hispanic black (p = 0.05), and married (p < 0.001). Most cancer survivors were diagnosed ≥11 years before survey completion; AYA reported greater duration since their cancer diagnosis compared to adult/older. Gynecological cancers (including ovarian, cervical, and endometrial) were most prevalent among AYA, while skin cancer/melanoma were the most prevalent among adult/older cancer survivors.

Table 1.

Sociodemographics, Cancer History, and Health Status of Adult Cancer Survivors Participating in HINTS 5 Cycles 1–3 by Age at Diagnosis

Sample size, n (N)	Adolescent young adult (15–39 years)	Adult/older (≥40 years)	All ages
	310 (13,590,933)	1554 (48,087,354)	1864 (61,678,287)
Mean (SD) or weighted percent				p
Age (years)	53.0 (14.2)	68.5 (11.4)	65.05 (13.65)	<0.001
Sex
Male	30.1%	45.5%	42.1%	<0.001
Female	68.2%	53.1%	56.4%
Race/ethnicity
Non-Hispanic white	66.7%	75.4%	73.5%	0.05
Non-Hispanic black or African American	11.0%	4.8%	6.2%
Hispanic	10.9%	6.8%	7.7%
Non-Hispanic Asian	0.9%	1.7%	1.5%
Non-Hispanic other	5.0%	1.5%	2.3%
Geographic designation
Metro	88.8%	83.7%	84.8%	0.10
Rural	11.2%	16.3%	15.2%
Occupational status
Employed	43.8%	17.3%	23.1%	<0.001
Unemployed	4.2%	2.0%	2.5%
Homemaker	8.0%	2.6%	3.8%
Retired	11.7%	35.0%	29.9%
Disabled	4.6%	3.3%	3.6%
Other	0.7%	0.3%	0.4%
Annual household income
<$20,000	11.2%	13.2%	12.8%	0.49
$20,000 to <$35,000	11.8%	14.9%	14.2%
$35,000 to <$50,000	15.8%	11.6%	12.5%
$50,000 to <$75,000	17.9%	16.3%	16.7%
≥$75,000	35.8%	30.4%	31.6%
Household size
1	25.3%	17.3%	23.5%	<0.001
2	47.3%	28.6%	43.2%
3	12.5%	20.2%	14.2%
4	5.4%	17.5%	8.1%
≥5	9.4%	16.4%	10.9%
Marital status
Married	63.7%	61.7%	62.1%	<0.001
Divorced	10.3%	11.0%	10.8%
Widowed	7.2%	14.0%	12.5%
Separated	2.3%	0.6%	1.0%
Single	13.9%	10.3%	11.1%
Education
Less than high school	4.9%	5.2%	5.2%	0.26
High school graduate	20.8%	26.6%	25.3%
Some college	41.6%	38.0%	38.8%
College graduate	32.0%	27.9%	28.8%
Age diagnosed (years)	29.1 (6.5)	58.9 (11.6)	52.31 (16.28)	<0.001
Years since diagnosis	23.8 (14.5)	9.7 (9.2)	12.77 (12.09)	<0.001
Time since diagnosis (years)
<1	6.2%	15.7%	13.6%	<0.001
2–5	4.0%	26.7%	21.7%
6–10	10.4%	20.0%	17.9%
≥11	76.7%	35.1%	44.3%
Cancer type
Breast	6.5%	15.2%	13.3%	<0.001
Gynecological	30.0%	3.6%	9.4%
Colorectal	4.2%	4.9%	4.7%
Prostate	0.7%	8.4%	6.7%
Blood	7.6%	3.6%	4.5%
Skin/melanoma	19.7%	33.6%	30.5%
More than 1 cancer	19.2%	16.1%	16.8%
Other	11.9%	14.5%	13.9%
Family history of cancer	80.8%	80.0%	80.2%	0.89
General health
Excellent	9.1%	8.4%	8.5%	0.09
Very good	35.5%	28.9%	30.4%
Good	31.6%	36.1%	35.1%
Fair	17.5%	21.5%	20.6%
Poor	6.2%	3.6%	4.2%
Primary care provider	79.1%	83.2%	82.3%	0.004
Health insurance coverage	93.3%	96.5%	95.8%	<0.001
Comorbid conditions
Diabetes	16.3%	26.4%	24.1%	<0.001
High blood pressure	41.5%	58.0%	54.3%	<0.001
Heart disease	7.0%	18.5%	10.0%	<0.001
BMI (kg/m2)	27.9 (5.8)	28.3 (6.4)	28.21 (6.30)	0.31
BMI category
Normal (18.5–24.9 kg/m2)	31.0%	30.8%	30.8%	0.94
Overweight (25.0–29.9 kg/m2)	40.3%	36.0%	37.0%
Obese (≥30.0 kg/m2)	26.9%	31.2%	30.3%
Smoking status
Current	18.2%	9.2%	11.2%	<0.001
Former	23.1%	38.7%	35.2%
Never	58.3%	51.7%	53.2%

Values may not add up to 100% due to rounding or missing data. Missing data <10%.

BMI, body mass index; HINTS, Health Information National Trends Survey.

Overall, half met vegetable recommendations, and two-thirds met fruit recommendations. In addition, fewer than half of all cancer survivors met aerobic or resistance PA recommendations (Table 2). In adjusted logistic regression models, AYA had 50% lower odds of meeting aerobic PA recommendations compared to adult/older (odds ratio [OR: 0.50; confidence interval [95% CI] (0.26–0.96); p = 0.04). Age at diagnosis was not associated with any other recommendation (Table 3).

Table 2.

Weighted Percent of Cancer Survivors Participating in HINTS 5 Cycles 1–3 Meeting Fruit Vegetable Intake and Aerobic and Resistance Physical Activity and Recommendations by Age at Diagnosis

Health behavior	Adolescent young adult (15–39 years)	Adult/older (≥40 years)	All ages	p
Fruit intake	67.3%	65.7%	66.0%	0.71
Vegetable intake	46.4%	51.3%	50.2%	0.31
Aerobic physical activity	30.5%	34.4%	33.6%	0.38
Resistance physical activity	46.5%	40.4%	41.7%	0.21

Table 3.

Health Beliefs in Adult Cancer Survivors Participating in HINTS 5 Cycles 1–3

Health belief	Adolescent young adult (15–39 years)	Adult/older (≥40 years)	All ages	p
Self-efficacy	68.6%	67.5%	67.8%	0.57
Normative belief	26.5%	31.0%	30.0%	0.24
Attitudinal belief	68.9%	62.2%	63.7%	0.16
Risk belief	54.7%	53.8%	54.0%	0.90
Intention	70.0%	74.5%	73.5%	0.34
Self-regulation	44.1%	30.4%	35.8%	0.002

There was demonstrated similarity between AYA and adult/older cancer survivor health beliefs (Table 4). Self-regulation differed by age at diagnosis, with a greater percent of AYA reporting (44.1% vs. 30.4%; p = 0.002). Of the health beliefs assessed, only self-regulation was significantly associated with age at diagnosis.

Table 4.

Associations of Age at Cancer Diagnosis and Meeting Fruit and Vegetable Intake and Aerobic and Resistance Physical Activity Recommendations in HINTS 5 Cycles 1–3

Health behavior	Unadjusted OR (95% CI)	p	Adjusted OR (95% CI)a	p
Fruit intake	1.07 (0.73–1.59)	0.72	1.00 (0.42–2.35)	0.99
Vegetable intake	0.82 (0.56–1.21)	0.32	0.92 (0.39–2.18)	0.84
Aerobic physical activity	0.84 (0.55–1.27)	0.36	0.50 (0.26–0.96)	0.04
Resistance physical activity	1.28 (0.88–1.88)	0.20	1.00 (0.42–2.35)	0.99
Resistance physical activity	1.28 (0.88–1.88)	0.20	1.03 (0.52–2.02)	0.94

^a Logistic regression models using Jackknife replicate and survey weights. Models were adjusted for age, sex, race/ethnicity, marital status, number in household, cancer type, years since diagnosis, regular primary care provider, health insurance status, smoking status, HINTS cycle, and BMI. The referent group is adult/older cancer survivors.

CI, confidence interval; OR, odds ratio.

In adjusted logistic regression models, those reporting high self-efficacy had higher odds of meeting aerobic PA recommendations (OR: 1.71; 95% CI 1.13–2.60; p = 0.01). Those reporting a high intention had significantly lower odds of meeting vegetable recommendations (OR: 0.58; 95% CI: 0.35–0.97; p = 0.04). Engaging in self-regulation of health behaviors was significantly associated with increased odds of meeting each recommendation (Table 5). No moderation in these relationships by age at diagnosis was found.

Table 5.

Associations of Health Beliefs and Meeting Fruit and Vegetable Intake and Aerobic and Resistance Physical Activity Recommendations in Adult Cancer Survivors Participating in HINTS 5 Cycles 1–3

Health belief	Fruit intake		Vegetable intake		Aerobic physical activity		Resistance physical activity
	Adjusted OR (95% CI)a	p	Adjusted OR (95% CI)a	p	Adjusted OR (95% CI)a	p	Adjusted OR (95% CI)a	p
Self-efficacy	1.15 (0.74–1.80)	0.53	1.31 (0.78–2.19)	0.30	1.71 (1.13–2.60)	0.01	1.08 (0.75–1.57)	0.67
Normative belief	0.79 (0.51–1.22)	0.28	0.63 (0.36–1.10)	0.11	0.92 (0.61–1.39)	0.68	0.91 (0.64–1.28)	0.58
Attitudinal belief	0.95 (0.64–1.41)	0.81	1.15 (0.75–1.76)	0.51	0.86 (0.57–1.31)	0.49	0.80 (0.57–1.11)	0.18
Risk belief	0.98 (0.56–1.72)	0.94	0.94 (0.41–2.16)	0.88	0.84 (0.52–1.34)	0.45	0.96 (0.61–1.51)	0.86
Intention	0.76 (0.48–1.20)	0.24	0.58 (0.35–0.97)	0.04	1.01 (0.64–1.60)	0.96	0.80 (0.52–1.23)	0.31
Self-regulation	2.34 (1.44–3.80)	0.001	2.18 (1.26–3.76)	0.005	3.82 (2.40–6.08)	<0.0001	1.78 (1.15–2.75)	0.01

^a Logistic regression models using Jackknife replicate and survey weights. Models were adjusted for age, sex, race/ethnicity, marital status, number in household, cancer type, years since diagnosis, regular primary care provider, health insurance status, smoking status, HINTS cycle, and BMI.

A Modified ACS Adherence score was calculated for the 1487 cancer survivors participating in HINTS 5 Cycles 1–3. The average score was 4.3 (95% CI: 4.13–4.44). Self-efficacy and self-regulation were significantly positively associated with the score, whereas normative belief and intention were significantly inversely associated (Table 6). No moderation by age at diagnosis was identified.

Table 6.

Association of Health Beliefs and Modified American Cancer Society Adherence Score in Adult Cancer Survivors Participating in HINTS 5 Cycles 1–3

Health belief	Modified ACS adherence score
	Sample sizea	β Coef (SE)b	p
Self-efficacy	1247	0.52 (0.16)	0.001
Normative belief	1237	−0.43 (0.17)	0.01
Attitudinal belief	1234	−0.15 (0.17)	0.38
Risk belief	712	−0.07 (0.17)	0.69
Intention	1229	−0.52 (0.21)	0.01
Self-regulation	939	1.21 (0.24)	<0.0001

^a Unweighted, weighted sample sizes used in multiple linear regression models.

^b Multiple linear regression models using jackknife replicate and survey weights. Models were adjusted for age, sex, race/ethnicity, marital status, number in household, cancer type, years since diagnosis, regular primary care provider, health insurance status, smoking status, and HINTS survey year.

ACS, American Cancer Society.

Discussion

Overall, many adult cancer survivors in the United States do not meet FV nor aerobic and resistance PA recommendations. Proportions of cancer survivors meeting FV and aerobic and resistance PA varied by age at diagnosis; these differences were not statistically significant. Age at diagnosis did not act as an effect modifier. Our estimates of health behaviors are similar to those previously reported in age stratified or AYA-specific analyses.^10–15 A recent meta-analysis of health behaviors in adult cancer survivors found that 34% meet FV recommendations (≥5 cups a day combined) and 43% meet PA recommendations (≥30 minutes a day); adherence to these behaviors have increased over time, but decrease with survivorship duration.⁵ A similar finding of declining daily FV and PA over an average 8.8 years postdiagnosis follow-up was observed, especially among younger cancer survivors.³³ Between 41.7% and 66.0% of all cancer survivors met FV and PA recommendations in our analyses with an average survival duration of 12.8 years.

Select health beliefs were associated with adherence to recommendations for FV and resistance and aerobic PA. Self-regulation was consistently associated with meeting recommendations. High self-efficacy was associated with meeting aerobic PA recommendations. Self-efficacy and self-regulation were positively associated with Modified ACS Adherence score. Health beliefs have previously been observed as a contributing factor of variance in PA levels and dietary habits in cancer survivors.^34,35 Our findings suggest that high self-efficacy may contribute to health behaviors among cancer survivors and self-regulation through engagement with a mobile app can support these behaviors. Higher self-efficacy has previously been associated with maintained health behaviors in cancer survivors,^36,37 and evolving mobile technology may provide the opportunity for promotion of cancer preventive health behaviors in cancer survivors by offering support, motivation, and goal tracking.

In our analyses, use of mobile apps to achieve health goals was self-regulation proxy, previous studies have highlighted such technologies as stimuli for behavior change.³⁸ Majority of currently available mobile health apps include multiple components to support behavior change, including self-regulation,²⁸ and may be a beneficial delivery modality for health behavior support in survivorship. Mobile app use has been found to improve FV consumption as well as PA.^39,40 Converse to our findings, one study found that mobile app use, while associated with intentions to improve intake, was not significantly associated with meeting FV recommendations.⁴¹ The findings are not directly comparable, however, as the aforementioned study, meeting recommendations for FV was categorized as ≥4 cups of each, which is greater than current ACS recommendations, potentially diluting any effect.

Health behaviors and beliefs were similar between AYA and adult/older cancer survivors in our study. This was an unexpected finding, as AYA are a distinct group with differences in disease biology, diagnosis, treatments, and psychosocial outcomes.⁴² The lack of differences observed between AYA and adult/older cancer survivors may be, in part, related to AYA participating in HINTS were adult survivors of an AYA cancer many years beyond their diagnosis (mean: 23.8) and experienced lower comorbidity burden. Results from these analyses indicate that select health beliefs inform on health behaviors independent of age at diagnosis among adult cancer survivors. To date, few interventions targeting diet and PA report use of behavioral theories to guide interventions, especially among the limited interventions for adult survivors of AYA cancer.⁴³ Many health behaviors, which are modifiable in long-term survivorship,⁴⁴ indicating the importance of providing necessary support for health behaviors throughout the survivorship continuum (in particular for AYA, who are often establishing life-long habits in early diagnosis⁴⁵).

Strengths and limitations

HINTS data provided a nationally representative sample of U.S. adults, in which both health behaviors and beliefs could be evaluated in adult cancer survivors. The prevalence of AYA within the sample was comparable to U.S. estimates.⁷ The combination of three waves of HINTS data with consistent survey items contributed to more reliable estimates for our subpopulation of interest. The random sampling methodology of HINTS reduces risk of sampling bias. To our current knowledge, this study is the first to rigorously evaluate specific health beliefs in relationship to FV and aerobic and resistance PA in cancer survivors and investigate effect modification by age at diagnosis. Having a working knowledge of these relationships is imperative to developing effective interventions grounded in behavior theory. Distinctions between fruit versus vegetable consumption and between aerobic and resistance PA are often absent in literature evaluating these exposures. While no standard score currently exists, our Modified ACS Adherence score gives a more comprehensive picture of adherence to multiple recommendations, which can work synergistically. By evaluating recommendations separately and in aggregate, this study provides insight into which recommendations, in the case of these analyses, FV intake, are driving overall adherence among cancer survivors.

This analysis was limited by the reliance on self-reported data. The average 32% response rate across HINTS cycles potentially contributed to risk of nonresponse bias. In a previous iteration of HINTS, nonresponse was associated with demographic characteristics, which are appropriately accounted for with standard weighting.⁴⁶ There is also a possibility of overestimation of PA and FV. Even though example serving sizes were provided for FV to reduce over reporting, the response options were categorical rather than continuous. Subsequently, the findings presented herein are based on responses, which are best estimates and may not directly compare to other studies. Previous studies have identified characteristics associated with FV consumption and PA of adults in HINTS,^47–49 which the present study did not include. Cut points for AYA categorization vary and current NCI standard encompasses early young adulthood, young adulthood, and late young adulthood.⁶ At the time of survey completion, ∼12% of AYA were <40 years and 7% were <5 years since diagnosis, therefore, behaviors and beliefs during early diagnosis among AYA subgroups may differ from the presented findings. Future analyses with a larger sample size of AYA may consider comparing cancer survivor subgroups to those without a history of cancer. With respect to health beliefs, questions related to self-efficacy and self-regulation were not specific to cancer, but were in the context of general health. Missingness of data was related to risk belief and self-regulation, which could contribute to potential bias.

Conclusion

These findings provide valuable information on health behaviors in cancer survivors, and highlight the role of health beliefs among cancer survivors. Important demographic differences between AYA and adult/older cancer survivor populations were found, but age at diagnosis was not an effect modifier. These results emphasize the role of self-efficacy and self-regulation with meeting ACS recommendations for FV and aerobic and resistance PA. Interventions that promote self-efficacy and include self-regulation may support increased adoption of diet and PA behaviors in cancer survivors.

Disclaimer

The research presented was completed and archived as part of dissertation requirements toward a doctoral degree for MBS.

Author Disclosure Statement

L.M.K. is an employee of Arcus Biosciences. No competing financial interests exist for the other authors.

Funding Information

Health Information National Trends Survey (HINTS) is funded by the National Cancer Institute with support from the Health Communication and Informatics Research Branch of the Division of Cancer Control and Population Sciences. This research was supported by the Leonard A. Lauder gift in honor of Dr. David Alberts and his legacy in ovarian cancer research.

Supplementary Material

Supplementary Table S1