Introduction

Improvements in cancer diagnosis and treatment mean that the number of individuals living with and beyond cancer (LWBC) is increasing globally. An estimated 21 million people are projected to receive a cancer diagnosis annually by 2030.1 Survival rates vary by cancer site and geographical location but are as high as 90% after a breast cancer diagnosis in the United States and Australia1; this highlights the increasing need to meet the health demands of long-term cancer survivors.

Although the evidence for reducing cancer risk in those LWBC is limited, the World Cancer Research Fund and the American Institute for Cancer Research recommend following diet, nutrition, and physical activity recommendations for primary cancer prevention.2 These guidelines specifically state that dietary supplements (DSs) are not recommended for cancer prevention2 because randomized controlled trials have generally shown no benefit, and in some cases, unexpected adverse effects have been found.3

DSs can play an important role in ensuring adequate nutrient intake when diets are deficient, which is more common in individuals LWBC than others. However, in the US National Health and Nutrition Examination Survey (NHANES), high levels of supplementation were associated with a higher proportion of cancer survivors having intakes of calcium and certain other micronutrients above the tolerable upper intake level.4 There is also the potential for DSs to compromise treatment efficiency4, 5 or interact with medications.6 Individuals LWBC may be taking medications not just to treat cancer and manage side effects but also for comorbidities. Although DS use is unlikely to present a risk or benefit for most individuals LWBC, there is the potential for inappropriate DS use because half of those taking supplements do so without consulting health care providers.4

Despite an absence of recommendations or evidence of benefits, DS use has been widely reported by those LWBC, and DS use has been found to increase after a cancer diagnosis.7 As many as 73% of individuals LWBC reported DS use in one US study,8 and although other studies of those LWBC have reported lower intakes, they are still generally higher in comparison with cancer-free individuals4, 9 (eg, 33% vs 21% in a Korean-based study9). The reasons for taking supplements are not fully understood, but a quarter of DS users LWBC report doing so to treat cancer6 or prevent cancer recurrence.10 Supplement users LWBC have also reported DS use as “something they could do to help themselves” and to boost their immune systems and gain energy.11 The most common reasons given by just under 2000 supplement users LWBC in NHANES were to improve overall health and to maintain health. However, because this questionnaire was for all participants, the majority of whom were cancer-free, it did not include options related to cancer recurrence.4 It is unclear whether beliefs about DSs playing a role in cancer recurrence are widespread among individuals LWBC and if such beliefs are associated with demographic factors and DS use. The aims of the current study were to identify and classify supplements taken by breast, prostate, and colorectal cancer survivors and to explore factors associated with their use, including body mass index (BMI), diet quality, and beliefs about the role of supplements in reducing the risks of cancer recurrence.

Materials and Methods

A cross-sectional study using baseline data from the Advancing Survival Cancer Outcomes Trial (ASCOT)12 was performed. Ten hospital sites across London and Essex sent a survey to patients 18 years old or older who had been diagnosed with breast, prostate, or colorectal cancer between 2012 and 2015. Respondents were asked to provide contact details if they were interested in learning more about a trial of a lifestyle intervention. Those expressing interest, meeting eligibility criteria, and providing informed consent were asked to complete additional baseline assessments, including two 24-hour dietary recalls. Participants were not required to complete all baseline assessments for randomization, so some participants did not complete the recalls. In the current analysis, we included eligible participants randomized within the trial (between August 2015 and November 2017) who completed the first 24-hour dietary recall requested. (ASCOT was conducted in 2 phases with additional participants recruited in 2019 when a 2-year follow-up assessment was added to the trial; 1151 participants were randomized in the first phase, with 102 not completing the first 24-hour dietary recall.) Ethical approval was obtained through the National Research Ethics Service Committee South Central–Oxford B (reference number 14/SC/1369). Described next are the measures included in the current analysis.

Survey

The survey included questions to assess the following: age (in years), gender (male or female), ethnicity (White or non-White), marital status (married/cohabiting or separated/divorced/widowed/single), highest level of education (no formal qualifications, General Certificate of Secondary Education/vocational or equivalent, A-level or equivalent, or bachelor's degree and higher or equivalent), number of different types of cancer treatments (including surgery, chemotherapy, and radiotherapy) for the most recent cancer, and cancer type (breast, prostate, or colorectal cancer). Although participants were asked to report their cancer stage, a very large proportion did not know, so cancer spread, assessed with “Has this cancer spread to any other parts of your body?,” was used. For the cancer spread variable, the response option “don't know” was recoded as missing data. The survey assessed the number of comorbid conditions by asking participants if they had any health problems from a list of 15 supplied health problems or other health problems not listed. Responses were categorized as none, 1, 2, or ≥3. BMI scores were calculated with self-reported height and weight and were categorized as underweight/healthy weight (<25 kg/m2), overweight (≥25 and <30 kg/m2), or obese (≥30 kg/m2).13 The underweight and healthy weight categories were combined because less than 1% of the participants fell into the underweight category (BMI < 18.5 kg/m2).

Participants were also asked for their thoughts about lifestyle and cancer. This included the following question: “Please rate how important each of the following are to you in reducing the chance that your cancer will come back (recur).” The question was followed by a list of different factors, including “taking dietary supplements (eg, vitamins).” Responses were collected via a Likert scale, with 1 indicating “not at all important” and 5 indicating “very important.” Responses were dichotomized as “not important” (1-2) and “important” (3-5).

24-Hour Dietary Recalls

Two 24-hour dietary recalls could be completed online or by telephone with the myfood24 program14: one on a weekday and the other on a weekend day. A weighted average daily intake was calculated ([5 × weekday intake + 2 × weekend day intake]/7). If 2 days' information was collected but this was not for a weekday and a weekend day, then the mean intake was calculated, and if only a single 24-hour recall was completed, this information was used. For this article, dietary information from the 24-hour recalls was used to assess adherence to the World Cancer Research Fund and UK dietary recommendations for fiber15 (≥30 g of fiber [AOAC] per day), fruits and vegetables2 (≥5 portions [400 g] per day), red meat2 (<500 g per week), processed meat2 (none consumed), and minimal intake of high-calorie foods2, 15, 16 (<33% of calories from fat and <5% of calories from free sugars per day).

At the end of the first 24-hour dietary recall, participants were asked, “Did you take any vitamins, minerals or other supplements during your day?” Fields labeled “dose,” “type,” and “brand” were completed. DSs recorded at the end of the first 24-hour recall were included in the analysis if they met the definition used in European Union law for DS data.17 This included vitamins, minerals, amino acids, enzymes, and herbal extracts in capsule or liquid form. Other items recorded by participants in the space for DSs (eg, medications or laxatives) were excluded.

Analysis

For the data analysis, the Statistical Package for the Social Sciences (version 25) was used. t tests and χ2 tests were used to compare demographic data, clinical characteristics, beliefs about supplements, and dietary factors between DS users and nonusers.

Multiple imputation was used to reduce the possible bias of missing data,18 with all the variables planned to be included in the regression analyses included. Logistic regression analyses were conducted on the imputed data set to explore factors associated with DS use. DS use was the dependent variable, and other variables listed in Table 1 were included as independent variables. However, “cancer type” was excluded to avoid the problem of multicollinearity because “cancer type” and “gender” were identical in the breast cancer sample (all female) and the prostate cancer sample (all male). First, a series of regressions was run for each of the independent variables individually, with no covariates included in each model. Then, 1 regression including all independent variables and controlling for covariates was run. The logistic regression analyses were repeated with completers to confirm if the pattern of results was similar.

TABLE 1. Demographic Data, Clinical Characteristics, Dietary Factors, and Dietary Supplement Beliefs and Use in Breast, Prostate, and Colorectal Cancer Survivors Total (n = 1049) Dietary Supplement Use Yes (n = 420 [40%]) No (n = 629 [60%]) Gender, No. (%) 1049 Male 398 (37.9) 118 (28.1) 280 (44.5) Female 651 (62.1) 302 (71.9) 349 (55.5) Age, mean (SD), y 1046 (64.4) 64.74 (10.97) 64.21 (11.61) Missing, No. (%) 3 (0.3) Cancer type, No. (%) 1049 Breast 571 (54.4) 267 (63.6) 304 (48.3) Prostate 264 (25.2) 75 (17.9) 189 (30.0) Colorectal 214 (20.4) 78 (18.6) 136 (21.6) Racial group, No. (%) 1044 White 982 (93.6) 381 (91.4) 601 (95.9) Non-White 62 (5.9) 36 (8.6) 26 (4.1) Missing 5 (0.5) Highest level of education, No. (%) 983 No formal qualifications 176 (16.8) 77 (19.3) 99 (17) GCSE/vocational 326 (31.0) 128 (32.1) 198 (33.9) A-level 138 (13.2) 52 (13.0) 86 (14.7) Degree or higher 343 (32.7) 142 (35.6) 201 (34.4) Missing 66 (6.3) Comorbidities, No. (%) 1046 0 366 (34.9) 130 (31.1) 236 (37.6) 1 347 (33.1) 141 (33.7) 206 (32.8) 2 190 (18.1) 86 (20.6) 104 (16.6) ≥3 143 (13.6) 61 (14.6) 82 (13.1) Missing 3 (0.3) Cancer spread, No. (%) 953 Yes 17 (1.6) 7 (1.9) 10 (1.7) No 936 (89.2) 371 (98.1) 565 (98.3) Don't know/missing 96 (9.2) No. of treatments, mean (SD) 1025 (2.12) 2.17 (1.1) 2.09 (1.13) Missing, No. (%) 24 (2.3) BMI, No. (%) 985 Underweight/healthy weight 370 (35.3) 161 (41.1) 209 (35.2) Overweight 408 (38.9) 162 (41.3) 246 (41.5) Obesity 207 (19.7) 69 (17.6) 138 (23.3) Missing 64 (6.1) Perceived importance of supplements for preventing cancer recurrence, No. (%) 928 Not important 756 (72.1) 258 (67.9) 498 (90.9) Important 172 (16.4) 122 (32.1) 50 (9.1) Missing 121 (11.5) Fiber recommendation, No. (%) 1049 Not meeting 937 (89.3) 362 (86.2) Meeting 112 (10.7) 58 (13.8) Fruit and vegetable recommendation, No. (%) 1049 Not meeting 558 (53.2) 196 (46.7) Meeting 491 (46.8) 224 (53.3) Red meat recommendation, No. (%) 1049 Not meeting 133 (12.7) 50 (11.9) Meeting 916 (87.3) 370 (88.1) Processed meat recommendation, No. (%) 1049 Not meeting 526 (50.1) 186 (44.3) Meeting 523 (49.9) 234 (55.7) High-calorie food recommendation, No. (%) 1049 Not meeting 1007 (96) 405 (96.4) Meeting 42 (4) 15 (3.6) Abbreviations: BMI, body mass index; GCSE, General Certificate of Secondary Education. Results

A total of 5835 questionnaires were completed; 1151 participants were randomized as they expressed an interest in the trial, were eligible, and completed any baseline measurements. The first baseline 24-hour recall was completed by 1049 participants, so data from these participants were included in the current analysis. The majority of the participants completed 2 recalls, but 7 participants completed only 1. A missing value analysis found that 2.96% of 17,833 values were missing, and 35.18% of the 1049 cases had at least 1 piece of missing data.

Participants were predominantly White (94%) and female (62%) with a mean age of 64 years. Forty percent of individuals LWBC reported DS use, and 32% of DS users believed that supplements (eg, vitamins) were important for reducing cancer recurrence risk (Table 1).

DSs containing fish oils were taken most frequently (Table 2). Calcium supplements and those containing calcium with vitamin D were grouped together (calcium with or without vitamin D) because some participants reported a brand name producing both. DSs marketed for joint health, including glucosamine and chondroitin, were classified as joint supplements. Turmeric and garlic were the most commonly used plant-based supplements; those taken less frequently, including pomegranate and kelp, were grouped as “other herbal supplements.” Likewise, vitamins C, D, and B complex were shown separately, and less frequently taken vitamins were shown as “other vitamins.” Supplements not falling into any other category, such as coenzyme Q10 and probiotics, were listed as “other supplements.” Two participants reported taking supplements but did not provide details. DS users took on average 1.8 different supplements. Forty percent of DS users took more than 1 supplement, and 9.8% took more than 3 supplements (range, 4-15). The most frequently reported DSs were fish oils, which were followed by calcium with or without vitamin D, multivitamins and minerals, vitamin D, and herbal supplements. Table 3 shows unadjusted and adjusted results from the binary logistic regression analyses with the imputed data set. When we adjusted for covariance, being female, believing that supplement use was important to recurrence, and not having obesity remained significant predictors of DS use. Repeating this analysis only with completers showed the same associations for these factors (see Supporting Table 1). When we considered associations between diet and DS use, unadjusted results showed that adherence to some dietary guidelines (fiber, fruits and vegetables, and processed meat) was positively associated with DS use. However, after adjustments for covariance, only adherence to fruit and vegetable guidelines was found to be associated with DS use (odds ratio, 1.36). An analysis with completers showed largely similar results except that adjusting for covariance resulted in adherence to fiber guidelines, but not adherence to fruit and vegetable guidelines, being associated with DS use.

TABLE 2. Type of Dietary Supplements Taken by Breast, Prostate, and Colorectal Cancer Survivors Dietary Supplement Type Total (n = 1049) Cancer Type, No. (%) Breast (n = 571) Prostate (n = 264) Colorectal (n = 214) Fish oils 137 (13.1) 73 (12.8) 35 (13.3) 29 (13.6) Calcium ± vitamin D 95 (9.1) 86 (15.1) 5 (1.9) 4 (1.9) Multivitamins and minerals 87 (8.2) 56 (9.8) 18 (6.8) 13 (6.0) Vitamin D 81 (7.7) 52 (9.1) 13 (4.9) 16 (7.4) Joint supplements 61 (5.8) 32 (5.6) 16 (6.1) 13 (6.1) B vitamins 31 (3.0) 19 (3.3) 4 (1.5) 8 (3.7) Vitamin C 27 (2.3) 11 (1.9) 6 (2.3) 10 (4.7) Turmeric 20 (1.9) 10 (1.8) 6 (2.3) 4 (1.9) Other minerals 18 (1.7) 13 (2.3) 3 (1.1) 2 (0.9) Garlic 13 (1.2) 6 (1.1) 4 (1.5) 3 (1.4) Zinc 10 (1.0) 7 (1.2) 2 (0.8) 1 (0.5) Iron 9 (0.9) 5 (0.9) 2 (0.8) 2 (0.9) Other herbal supplements 64 (6.1) 45 (7.9) 10 (3.8) 8 (3.7) Other vitamins 6 (0.6) 4 (0.7) 1 (0.4) 1 (0.5) Other supplements 51 (4.9) 33 (5.8) 11 (4.2) 7 (3.3) Not specified 2 (0.2) 2 (0.4) 0 0 TABLE 3. Logistic Regression Analyses for Dietary Supplement Use in Breast, Prostate, and Colorectal Cancer Survivors (n = 1049) Unadjusted Adjusteda OR CI P OR CI P Gender Male 1.00 — — 1.00 — — Female 2.05 1.58-2.68 ≤.001 2.48 1.72-3.56 ≤.001 Age (y) 1.004 0.99-1.02 .467 1.02 1.00-1.03 .046 Ethnicity White 1.00 — — 1.00 — — Non-White 2.23 1.32-3.76 .003 1.57 0.89-2.78 .122 Highest level of education No formal qualifications 1.00 — — 1.00 — — GCSE/vocational 0.85 0.60-1.23 .39 0.85 0.56-1.31 .464 A-level 0.80 0.51-1.25 .33 0.87 0.52-1.46 .607 Degree or higher 0.92 0.64-1.32 .63 0.83 0.54-1.28 .405 No. of comorbid conditions None 1.00 — — 1.00 — — 1 1.24 0.92-1.68 .16 1.19 0.85-1.66 .320 2 1.51 1.05-2.15 .03 1.31 0.88-1.95 .189 ≥3 1.36 0.92-2.02 .13 1.40 0.88-2.23 .152 Cancer spread No 1.00 — — 1.00 — — Yes 0.95 0.42-2.17 .91 0.76 0.32-1.81 .527 No. of treatments 1.07 0.96-1.20 .24 0.90