Perceived benefits of diet, physical activity, alcohol intake and smoking among cancer survivors: a cross-sectional study Authors: Rachel Drbohlav Ollerton, Nadine Bol, Gwenn Beets, Frans Folkvord, Emiel Krahmer Published: 2025 ABSTRACT BackgroundA healthy lifestyle can have various health benefits for cancer survivors, including positive effects for their general health and cancer-specific outcomes. In general, cancer survivors adhere more to smoking and alcohol recommendations, compared to diet and physical activity recommendations.MethodsTo better understand the differences in benefit perceptions among the four health behaviours, we conducted a cross-sectional study among 197 Dutch survivors with various cancer types.ResultsWe found a positive association between perceiving benefits and engaging in respective health behaviours. Furthermore, cancer survivors perceived these health behaviours differently, with low alcohol intake and not smoking being perceived as more beneficial for cancer outcomes and general health, compared to having a balanced diet and engaging in physical activity. Vulnerable groups, including those recently diagnosed and those with lower perceived control, perceived health behaviours as less beneficial.ConclusionsDiet, physical activity, not smoking and limited alcohol intake are perceived as differently beneficial by cancer survivors. These results hold important implications for patient education and communication of health behaviours to cancer survivors.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-025-26004-w. Introduction Adherence to health behaviours, in particular maintaining a healthy diet, engaging in regular physical activity (PA), reducing alcohol intake and not smoking, significantly impacts the development of several types of cancers [1]. More specifically, a healthy lifestyle is known to lower the risk of various cancers, and an estimated 30 to 50% of all cancer cases can be prevented by lifestyle factors [1, 2]. Furthermore, a healthy lifestyle after a cancer diagnosis has been found to positively influence both patients’ overall health and cancer-related outcomes, such as better therapy results and higher survival rates. It is also related to lower chances of cancer recurrence, and lower risk of developing other cancers [2–4]. Adhering to health behaviours is therefore important for cancer survivors – individuals living with or beyond cancer – in terms of their general health, but also in terms of cancer (treatment) outcomes. Cancer survivors are also recommended to adhere to these health behaviours if they can, according to the WCRF/AICR Third Expert Report on cancer protection and care recommendations [2], which are further expanded on in the method section. Nonetheless, cancer survivors tend to adhere more to smoking and alcohol recommendations, as compared to physical activity and diet recommendations [5]. Refraining from smoking and drinking alcohol are valuable lifestyle factors that contribute to health improvements in cancer survivors. However, research points to a necessity of adhering to multiple health behaviours, rather than single health behaviours, for cancer survivors to experience health benefits and reduce the risks of developing cancer [3, 5]. There are different reasons why cancer survivors may adhere to certain and not to other health behaviours. Behaviours such as smoking are widely recognised as a risk factor for cancer and socially less accepted, compared to other (un)health(y) behaviours [6]. Some behavioural patterns may relate to a built environment, such as limited access to healthy foods or facilities for exercise [4]. Importantly, cancer survivors often report not receiving (clear) recommendations regarding dietary changes, which may be a reason for their low adherence to diet-related behaviours [7]. In this study, we aimed to understand why differences in adherence to various health behaviours appear by looking at perceived benefits, a construct that refers to ones’ perception of engaging in a certain behaviour having positive outcomes. Perceived benefits are considered one of the predictors of health behaviour, proposed by several influential health models, such as, for example, the Health Belief Model [8, 9], the Transtheoretical Model [10], and Theory of Planned Behaviour [11]. Empirical research has supported the theoretical importance of perceived benefits, particularly in oncology. In non-patient populations, perceived benefits are positively linked to preventative cancer screening [e.g., 12, 13]. Among cancer survivors, perceived benefits are frequently associated with medication adherence [14], behaviour change (intentions) [e.g., 15, 16], and facilitators of behaviour change [17]. However, fewer studies have focused on associations between perceived benefits and actual behaviour performance, though positive associations with exercise levels have been reported [18–20]. We explore the relation of perceived general health and cancer-related benefits and adherence to different health behaviours. This will clarify the association between perceiving benefits and adherence while distinguishing the relative importance of general health versus cancer-related benefits. Therefore, to confirm the theoretical and empirical evidence, we formulated the following research questions (RQs): RQ11: To what extent are perceived benefits of health behaviours related to the performance of those behaviours?RQ1a: To what extent are the perceptions about the benefits of health behaviours (cancer-related and general health) related to the performance of those behaviours?RQ1b: Which type of benefit relates to the performance of the behaviour more? RQ1a: To what extent are the perceptions about the benefits of health behaviours (cancer-related and general health) related to the performance of those behaviours? RQ1a: To what extent are the perceptions about the benefits of health behaviours (cancer-related and general health) related to the performance of those behaviours? RQ1b: Which type of benefit relates to the performance of the behaviour more? RQ1b: Which type of benefit relates to the performance of the behaviour more? Considering the potential association of perceiving benefits and behaviour, it is important to understand whether cancer survivors indeed perceive the benefits of health behaviours. Experienced benefits may include positive outcomes for their physical health, social connection, or psychological well-being [21]. However, research suggests that cancer survivors may underestimate the impact of certain behaviours on cancer outcomes. For instance, nearly half of breast cancer survivors in one study were uncertain or disagreed about the benefits of diet on cancer recurrence [22]. Although studies on the perceived benefits of lower alcohol intake are limited, one study suggests cancer survivors identified as regular drinkers may not recognise its harms [23]. While physical activity is often recognised as beneficial for physical and mental health [24], many survivors underestimate its prognostic impact [25]. Similarly, smoking studies show mixed results, with some highlighting high perceived benefits of quitting [15], and others showing a lack of awareness among current smokers [26]. It is therefore unclear whether cancer survivors perceive the benefits of health behaviours and how these benefit perceptions differ between various health behaviours. Uncovering whether health behaviours are perceived differently could contribute to understanding why some behaviours are adhered to more than others. Besides health behaviours potentially not being perceived as equally beneficial, survivors with certain types of cancers might also differently perceive the benefits. For example, cancer survivors might perceive greater benefits of quitting smoking if they are diagnosed with a smoking-related cancer [15]. Therefore, we explored the following RQs: RQ2: To what extent do cancer patients perceive engaging in health behaviours as beneficial (i.e., beneficial for overall health, as well as for cancer-specific outcomes)?RQ2a: To what extent do the perceived benefits of health behaviours differ among cancer patients? Are some behaviours perceived as more beneficial for cancer, compared to others?RQ2b: To what extent do the perceived benefits of health behaviours differ between cancer patients? Do patients with certain types of cancer perceive health behaviours as more beneficial? RQ2a: To what extent do the perceived benefits of health behaviours differ among cancer patients? Are some behaviours perceived as more beneficial for cancer, compared to others? RQ2a: To what extent do the perceived benefits of health behaviours differ among cancer patients? Are some behaviours perceived as more beneficial for cancer, compared to others? RQ2b: To what extent do the perceived benefits of health behaviours differ between cancer patients? Do patients with certain types of cancer perceive health behaviours as more beneficial? RQ2b: To what extent do the perceived benefits of health behaviours differ between cancer patients? Do patients with certain types of cancer perceive health behaviours as more beneficial? Besides understanding differences at the group level (i.e., between survivors with different cancer types) and outcome level (i.e., perceptions of different health behaviours), individual differences should be considered. Studies have found that there are various individual differences associated with survivors’ behaviours [e.g., 17]. It is therefore important to identify cancer survivors that are less likely to perceive the benefits and adhere to health behaviours to gain a better understanding of (non)adherence and create possibilities for tailored behavioural interventions targeting those that are at highest risk. Although not much research has focused on factors influencing perceived benefits of health behaviours, various non-actionable (e.g., socio-demographic, cancer-related) factors have been found to influence survivors’ adherence. Studies have shown that survivors with lower educational levels, higher age and of male gender may be less likely to adhere to the health-related behaviours [17, 27, 28]. Furthermore, recently diagnosed survivors tend to have better adherence to health behaviours [5]. Along with these non-actionable factors, which are important to identify to devote particular attention to those survivors, other actionable factors may play a role [e.g., 29]. In this paper, we focus on those that have been understudied, namely perceived control over and acceptance of one’s health condition, proposed by the Subjective Experienced Health theory [30] as driving factors of patients’ behaviour. This study aims to expand knowledge on adherence correlates by identifying groups vulnerable to not perceiving benefits or adhering to health behaviours. Therefore, we explore: RQ3: Which groups might be particularly vulnerable to not perceiving benefits of health behaviours and not engaging in those health behaviours?RQ3a: To what extent does (not) perceiving benefits of health behaviours depend on perceived control and acceptance, medical and socio-demographic variables?RQ3b: To what extent does (not) engaging in those health behaviours depend on perceived control and acceptance, medical and socio-demographic variables? RQ3a: To what extent does (not) perceiving benefits of health behaviours depend on perceived control and acceptance, medical and socio-demographic variables? RQ3a: To what extent does (not) perceiving benefits of health behaviours depend on perceived control and acceptance, medical and socio-demographic variables? RQ3b: To what extent does (not) engaging in those health behaviours depend on perceived control and acceptance, medical and socio-demographic variables? RQ3b: To what extent does (not) engaging in those health behaviours depend on perceived control and acceptance, medical and socio-demographic variables? Using a cross-sectional design, we assess the relation of perceived benefits to health behaviours and to what extent cancer survivors perceive various health behaviours (diet, physical activity, low alcohol intake, and not smoking) as beneficial for their health. To our knowledge, this is the first study exploring differential health behaviour perceptions among cancer survivors. Lastly, we explore to what extent non-actionable (socio-demographic and medical) and actionable (control and acceptance) characteristics associate with benefit perceptions and adherence to health behaviours. With these research questions, we aim to contribute to understanding survivors’ perceptions of health behaviours and factors involved in (non)adherence behaviour. Methods The study was ethically approved by the Research Ethics and Data Management Committee of the Tilburg School of Humanities and Digital Sciences of Tilburg University (reference number: REDC2023.17) and preregistered on the Open Science Framework (https://osf.io/shx7p). Respondents Eligible respondents were adult (≥ 18 years) cancer survivors2. Respondents were recruited [in May and June 2023] through the largest platform for cancer survivors in The Netherlands, kanker.nl, which is frequently used for cancer-related research purposes. Respondents completed the survey in Dutch and participation was voluntary, meaning the respondents were not compensated for their time. We conducted an a priori power analysis using G*power for the analysis that required the largest sample size. To achieve a medium effect size, with power 0.80 (α = 0.05), for a Multivariate analysis of variance (MANOVA) with 4 groups and 2 response variables, G*Power suggested 118 respondents. We decided to oversample to 200 respondents, to account for incomplete responses, possible exclusion of respondents (completing the survey in under 5 min and straightlining), follow-up post-hoc tests, and exploratory analyses. Design and procedure We conducted cross-sectional survey research to assess perceived benefits and adherence to lifestyle behaviours in cancer survivors. After reading the information letter and consenting to participate in the research, respondents provided demographic and disease-related information. They were then asked about their current health behaviours, and afterwards about their perceived benefits of relevant behaviours. Respondents were given a definition of each health behaviour prior to answering these questions. They answered questions about their perceived acceptance and control, after which they could optionally provide information about their (health) communication experiences and preferences. Lastly, they were provided a debriefing letter, with sources about health behaviour change for cancer survivors. The median response time for completing the survey was 13.32 min (Q1 = 10.57, Q3 = 19.80). Measures The complete Dutch questionnaire (and a translation in English) can be found on the Open Science Framework (https://osf.io/8rndh/). Demographics Respondents completed baseline questions about their gender, age, and highest level of education they had completed. Medical background Medical background questions included time since diagnosis (respondents stated either the exact date or an estimation of the day/month/year of diagnosis), type of cancer, and whether they were undergoing or have undergone cancer treatment. Respondents were also asked to specify which treatment (if any) they had or were undergoing (including surgery, chemotherapy, radiotherapy, hormone therapy, and/or immunotherapy) and the purpose of the treatment (healing or life extension/relief of complaints/I don’t know). Based on the provided estimation of diagnosis date, the variable time since diagnosis was categorised into < 1 year ago, 1–5 years ago and > 5 years ago. Current health behaviours Health behaviours of interest were operationalised according to the latest Third Expert Report, published by WCRF/AICR on cancer protection and care recommendations [3]. Diet Diet was measured using an 8-item food frequency scale [33] developed for this study, including questions on 8 food categories – 4 healthy and 4 unhealthy3. According to WCRF/AICR [3], the chosen food categories relevant for cancer include: vegetables, fruits, wholegrains, legumes (healthy), red meats, processed meats, fast foods or highly processed foods, and sugar sweetened drinks (unhealthy). Respondents were asked on a 7-point scale, ranging from never to multiple times a day, how many times on average during a typical week they consume the mentioned food categories, with examples of foods for each category. The scores for healthy and unhealthy (reverse-coded) food categories scores were summed, with a higher total score indicating a more balanced diet or more guidelines-based dietary patterns. Physical activity In line with the WCRF/AICR recommendations [3], regular PA was defined as at least 150 min of moderate-intensity PA or at least 75 min of vigorous PA (or a combination) each week [3]. The World Health Organisations’ extended recommendation on replacing sedentary time with activity of any intensity (including light intensity) was added to this definition [34]. The Godin Leisure-time questionnaire [35] was used to assess PA. This scale has been widely used in cancer populations [36]. Respondents were asked how many times they perform strenuous, moderate, and light PA for more than 15 min, during a typical week. One question was asked for each type of exercise (i.e., strenuous, moderate, and light) and examples of each activity were given. The frequency of each type of exercise was multiplied with their metabolic equivalent values (strenuous exercise by 9, moderate exercise by 5 and light exercise by 3). The scores were summed, with higher sum scores indicating higher levels of physical activity [37]. Alcohol intake Limiting alcohol consumption was defined as no alcohol or drinking up to one alcoholic beverage per day, following the WCRF/AICR [2] recommendations. Alcohol consumption was assessed using two questions. Respondents were asked whether they drink alcohol, with the answer options being yes/no. Respondents who indicated that they drink alcohol were then asked to provide an estimated number of alcoholic drinks they consume during a typical week. A picture showing what is considered one alcoholic drink was provided and respondents provided their average weekly alcohol intake through an open-ended text box. Smoking Smoking status was assessed with one question asking whether the respondent smokes tobacco products. The answer options were yes; no, but I used to; no, and I never have, and respondents were categorised as current smokers, former smokers and never smokers [38]. Smoking status was used as a binary variable in which one group consisted of current smokers, and the other group consisted of former and never smokers. Perceived benefits of health behaviours Perceived benefits were measured using 7 items for each health behaviour separately, and respondents were given a definition of the health behaviour prior to answering these questions. The first 4 items were about the belief that engaging in a certain behaviour would lead to cancer-related health benefits. These items were adapted from previous research, in which they have been validated [15], and asked to what extent respondents agreed that a health behaviour can: (1) improve chances of getting the full benefit from their cancer treatment; (2) reduce chances of having side-effects and complications from treatment; (3) reduce chances of developing the same cancer again; and (4) reduce chances of developing a new type of cancer. The other 3 items, developed for this study, were about the belief that engaging in a certain behaviour would lead to general health benefits, and asked to what extent respondents agreed that a health behaviour can: (5) reduce the risk of other diseases; (6) prolong lifespan; (7) improve ability to do daily activities. Both cancer-related and health-related benefits were assessed on a 5-point scale ranging from strongly disagree to strongly agree. The scale was computed using mean scores, with higher scores indicating higher perceived benefits. The perceived benefits were used in the analyses in three different ways: as cancer-related benefits, containing the first 4 items (Cronbach’s α =.814.), general health benefits, containing the last 3 items (Cronbach’s α = 0.75.) and overall, containing all 7 items (Cronbach’s α = 0.85.). All scales were averaged. Control and acceptance Control over and acceptance of one’s health condition were each measured using 3 items from previous research [39]. A 7-point scale was used for these questions, ranging from strongly disagree to strongly agree. According to the literature [39], respondents were categorised into low (mean score ≤ 5.36) or high (mean score > 5.36) control (Cronbach’s α = 0.86), and low (mean score ≤ 4.96) or high (mean score > 4.96) acceptance (Cronbach’s α = 0.86). Data analyses All analyses were conducted using R software (ver. 4.2.3). Descriptive statistics were computed to summarise demographic and medical characteristics of the sample. Assumption checks were performed for all analyses and the details can be found in Supplementary file. There were a few missing values and the missingness was minor and deemed accidental, therefore missing values were excluded for relevant parts of the analysis. Outliers were checked using the identify_outliers function in R, with values three interquartile ranges below and above first and third quartiles respectively identified as extreme and removed for relevant analyses. To answer the three RQs, analyses were grouped into three stages. First, separate regression models were employed for the different behaviours, with linear regressions used for diet, physical activity, and alcohol as outcome variables, and the overall benefits as predictors; and a binary logistic regression for smoking behaviour as the outcome variable, and perceived overall benefits of not smoking as predictor (RQ1a)5. Furthermore, multiple linear regressions with cancer-related and general health benefits as predictors were performed for each behaviour, whereas for smoking, a binary logistic regression was performed (RQ1b). It should be noted that respondents consumed little to no alcohol and rarely identified as smokers, which presented limitations in analysing the alcohol and smoking data (as explained in the Supplementary file). To answer RQ2a, a within-subjects Analysis of Variance (ANOVA) was performed with the overall perceived benefits as the outcome variable and type of behaviour as the within-subjects factor. The analysis was performed in R using rstatix package, meaning that the Greenhouse-Geisser sphericity correction was automatically applied to factors violating it. Pairwise paired t-tests were performed with Bonferroni correction as a post-hoc test. To explore whether the behaviours were perceived differently in cancer-related or general health benefits, a follow-up within-subjects MANOVA with cancer-related and general health benefits as the outcome variables and type of behaviour (diet, PA, smoking, alcohol) as the within-subjects factor was conducted. Two within-subjects ANOVAs with Bonferroni correction applied6 were performed as follow-up analyses. Pairwise paired t-tests with Bonferroni correction were then performed for the behaviours that differed among each other significantly in the previous analysis (within-subjects ANOVA with overall benefits). As the sample consists of highly unequal groups of survivors that have the same cancer type, the analysis of possible differences of perceived benefits between cancer types (RQ1b) was not possible in the way that was pre-registered. A different approach was taken for this analysis, as reported in the section Exploratory analyses. Lastly, a multivariate multiple linear regression analysis was conducted with overall perceived benefits of the four health behaviours as the outcome variables and gender, age, education7, time since diagnosis, perceived control, and acceptance as predictor variables (RQ3a). Separate multiple linear regression analyses were conducted for each health behaviour as dependent variable (RQ3b), with the same predictor variables as included for RQ3a. Analogous binary logistic regression was conducted for the smoking behaviour. Results Descriptive data A total of 230 cancer survivors accessed the survey, out of which 197 completed it. No respondents fit the exclusion criteria, meaning that the final sample size consisted of 197 survivors (119 female, 78 male, Mage = 59.97, SDage = 10.84). Table 1 shows the socio-demographic and cancer-related characteristics of the respondents. Table 1Socio-demographic, medical and health characteristics of the respondentsVariablen (%)Age, M (SD)59.97 (10.84)Gender Women119 (60.4) Men78 (39.5)Education1 Primary school (basisonderwijs)1 (0.5) Pre-vocational education (VMBO)11 (5.5) Secondary education (HAVO/VWO)15 (7.6) Post-secondary vocational education (MBO)47 (23.8) Higher professional education (HBO)83 (42.1) University (WO)39 (19.7)Time since diagnosis < 1 year ago23 (11.6) 1–5 years ago94 (47.7) > 5 years ago79 (40.1)Type of cancer Breast cancer60 (30.4) Prostate cancer31 (15.7) Colon cancer15 (7.6) Ovarian cancer11 (5.5) Multiple cancers12 (6) Other268 (37.5)Treatment status Treatment completed112 (56.9) Still in treatment83 (42.1) No treatment1 (0.5)Healthy foods intake, M (SD) Fruits5.25 (1.49) Vegetables5.50 (1.02) Legumes2.73 (0.91) Grains4.41 (1.67)Unhealthy foods intake, M (SD) Red meat5.23 (1.18) Processed meat5.23 (1.33) Fast foods5.25 (1.26) Sugar sweetened drinks6.06 (1.43)Physical activity Active126 (64.0) Moderately active26 (13.2) Sedentary43 (21.8)Alcohol intake, 3M (SD)3.44 (6.14)Smoking status Current smoker14 (7.1) Former smoker107 (54.3) Never smoker76 (38.5)1There were certain missing values due to incorrect response input (education n = 1; time since diagnosis n = 1; treatment status n = 1; physical activity score n = 2)2Other cancers included melanoma (n = 8), bladder (8), uterus (7), lung (6), multiple myeloma (6), brain (6), kidney (4), Hodgkin lymphoma (4), oesophagus (4), non-Hodgkin lymphoma (3), throat (3), cervical (2), testicular (2), rectal (1), leukaemia (1), larynx (1), head and neck (1), and lymph node cancer (1)3A total of 108 (54.8%) respondents indicated that they do drink alcohol, and 89 (45.1%) indicated that they do not drink alcohol Table 1Socio-demographic, medical and health characteristics of the respondentsVariablen (%)Age, M (SD)59.97 (10.84)Gender Women119 (60.4) Men78 (39.5)Education1 Primary school (basisonderwijs)1 (0.5) Pre-vocational education (VMBO)11 (5.5) Secondary education (HAVO/VWO)15 (7.6) Post-secondary vocational education (MBO)47 (23.8) Higher professional education (HBO)83 (42.1) University (WO)39 (19.7)Time since diagnosis < 1 year ago23 (11.6) 1–5 years ago94 (47.7) > 5 years ago79 (40.1)Type of cancer Breast cancer60 (30.4) Prostate cancer31 (15.7) Colon cancer15 (7.6) Ovarian cancer11 (5.5) Multiple cancers12 (6) Other268 (37.5)Treatment status Treatment completed112 (56.9) Still in treatment83 (42.1) No treatment1 (0.5)Healthy foods intake, M (SD) Fruits5.25 (1.49) Vegetables5.50 (1.02) Legumes2.73 (0.91) Grains4.41 (1.67)Unhealthy foods intake, M (SD) Red meat5.23 (1.18) Processed meat5.23 (1.33) Fast foods5.25 (1.26) Sugar sweetened drinks6.06 (1.43)Physical activity Active126 (64.0) Moderately active26 (13.2) Sedentary43 (21.8)Alcohol intake, 3M (SD)3.44 (6.14)Smoking status Current smoker14 (7.1) Former smoker107 (54.3) Never smoker76 (38.5)1There were certain missing values due to incorrect response input (education n = 1; time since diagnosis n = 1; treatment status n = 1; physical activity score n = 2)2Other cancers included melanoma (n = 8), bladder (8), uterus (7), lung (6), multiple myeloma (6), brain (6), kidney (4), Hodgkin lymphoma (4), oesophagus (4), non-Hodgkin lymphoma (3), throat (3), cervical (2), testicular (2), rectal (1), leukaemia (1), larynx (1), head and neck (1), and lymph node cancer (1)3A total of 108 (54.8%) respondents indicated that they do drink alcohol, and 89 (45.1%) indicated that they do not drink alcohol Socio-demographic, medical and health characteristics of the respondents 1There were certain missing values due to incorrect response input (education n = 1; time since diagnosis n = 1; treatment status n = 1; physical activity score n = 2) 2Other cancers included melanoma (n = 8), bladder (8), uterus (7), lung (6), multiple myeloma (6), brain (6), kidney (4), Hodgkin lymphoma (4), oesophagus (4), non-Hodgkin lymphoma (3), throat (3), cervical (2), testicular (2), rectal (1), leukaemia (1), larynx (1), head and neck (1), and lymph node cancer (1) 3A total of 108 (54.8%) respondents indicated that they do drink alcohol, and 89 (45.1%) indicated that they do not drink alcohol Relation of perceived benefits to behaviour (RQ1) We found that perceiving overall benefits of health behaviours associated with engagement in the respective behaviours, meaning that survivors perceiving higher benefits were more likely to engage in those health behaviours. Specifically, perceiving benefits associated with having a more balanced diet, F(1, 195) = 10.49, β = 0.23, p = .001, R2adj = 0.05, higher engagement in physical activity, F(1, 191) = 9.54, β = 0.22, p = .002, R2adj = 0.04, and lower alcohol intake, F(1, 189) = 8.12, β = − 0.20, p = .005, R2adj = 0.04. Perceiving benefits of not smoking lowered the odds of being a current smoker (OR = 0.27, 95% CI [0.12, 0.60], p = .001). The evidence on which type of benefit (i.e., cancer-related versus general health) associates with the behaviour more strongly is inconclusive. Cancer-related and general health benefits were found to significantly explain the variation in dietary behaviour (F(2, 194) = 6.24, p = .002, R2adj = 0.05), physical activity, (F(2, 190) = 5.05, p = .007, R2adj = 0.04), alcohol intake (F(2, 188) = 4.07, p = .019, R2adj = 0.03), and smoking (p = .005). General health benefits were found to explain significantly more variance in diet-related behaviour (β = 0.20, p = .018), unlike the cancer-related benefits (β = 0.06, p = .455). However, neither the cancer-related nor general health benefits significantly explained the variance in physical activity scores (βcancer = 0.11, pcancer = .181; βgeneral = 0.15, pgeneral = .081), alcohol scores (βcancer = 0.16, pcancer = .091; βgeneral = 0.07, pgeneral = .479), or smoking status (ORcancer = 0.58, pcancer = .181; ORgeneral = 0.45, pgeneral = .081) beyond each predictor alone. Differences in perceived benefits (RQ2a) Survivors perceived the four health behaviours as differently beneficial (see Table 2). The mean value of overall perceived benefits was significantly higher for alcohol intake compared to diet (padj = .013, Mdiff = 0.12), smoking compared to diet (padj < .001, Mdiff = 0.49), smoking compared to physical activity (padj < .001, Mdiff = 0.45), and smoking compared to alcohol intake (padj < .001, Mdiff = 0.37). There was no statistically significant difference in the overall perceived benefits scores between alcohol intake and physical activity, and diet and physical activity (padj > .261). Follow-up analyses also showed differences between the behaviours regarding perceiving cancer-related benefits and general health benefits, and the results are visually presented in Fig. 1. The differences found for alcohol and diet were regarding cancer-related benefits (padj < .001), while differences for smoking and diet, smoking and physical activity, and smoking and alcohol regarded differences in both cancer-related (padj < .001) and general health benefits (padj < .002). Table 2Means and standard deviations for perceived benefitsCancer-related benefits1M (SD)General health benefits2M (SD)Overall benefits3M (SD)Diet3.47 (0.73)4.05 (0.57)3.72 (0.59)Physical activity3.47 (0.75)4.15 (0.54)3.76 (0.58)Alcohol3.69 (0.76)4.02 (0.68)3.84 (0.66)Smoking4.10 (0.78)4.36 (0.65)4.21 (0.67)1Model statistics: F(2.62, 513.97) = 92.61, p < .001, η2g=.102Model statistics: F(2.54, 496.94) = 29.36, p < .001, η2g=.053Model statistics: F(2.49, 488.88) = 78.54, p < .001, η2g=.09 Table 2Means and standard deviations for perceived benefitsCancer-related benefits1M (SD)General health benefits2M (SD)Overall benefits3M (SD)Diet3.47 (0.73)4.05 (0.57)3.72 (0.59)Physical activity3.47 (0.75)4.15 (0.54)3.76 (0.58)Alcohol3.69 (0.76)4.02 (0.68)3.84 (0.66)Smoking4.10 (0.78)4.36 (0.65)4.21 (0.67)1Model statistics: F(2.62, 513.97) = 92.61, p < .001, η2g=.102Model statistics: F(2.54, 496.94) = 29.36, p < .001, η2g=.053Model statistics: F(2.49, 488.88) = 78.54, p < .001, η2g=.09 Means and standard deviations for perceived benefits 1Model statistics: F(2.62, 513.97) = 92.61, p < .001, η2g=.10 2Model statistics: F(2.54, 496.94) = 29.36, p < .001, η2g=.05 3Model statistics: F(2.49, 488.88) = 78.54, p < .001, η2g=.09 Fig. 1Differences in two types of perceived benefits across health behaviours Differences in two types of perceived benefits across health behaviours Vulnerable groups (RQ3) Gender, age, education, time since diagnosis, control, and acceptance as predictors were found to significantly contribute to explaining the variation in perceiving benefits of diet, physical activity, lower alcohol consumption, and not smoking. Compared to those diagnosed 1 to 5 years ago, respondents that were diagnosed less than a year ago perceived less benefits of diet (β = − 0.60, p = .006), physical activity (β = − 0.45, p = .032), low alcohol intake (β = − 0.65, p = .003), and not smoking (β = − 0.75, p < .001). Respondents diagnosed more than five years ago also perceived less benefits of diet (β = − 0.31, p = .041) and physical activity (β = − 0.33, p = .023), but not lower alcohol intake (β = − 0.15, p = .336) and not smoking (β = 0.03, p = .823), compared to those diagnosed 1 to 5 years ago. Sense of control was consistently a significant predictor, indicating that respondents that have a low sense of control perceived less benefits of diet (β = − 0.66, p <.001), physical activity (β = − 0.82, p < .001), low alcohol intake (β = − 0.65, p < .001) and not smoking (β = − 0.39, p = .015), compared to respondents with a high sense of control over their disease. Gender, age, education, time since diagnosis, control and acceptance as predictors were also found to significantly explain the variation in diet behaviour, physical activity, alcohol consumption, and smoking status. Men had lower diet (β = − 0.55, p < .001) and higher alcohol intake scores (β = 0.61, p < .001) than women. Men also had higher odds of being current smokers (OR = 10.42, 95% CI [2.44, 58.52], p = .003) than women. Respondents with low sense of control had lower diet scores (β = − 0.50, p = .003) and lower physical activity scores (β = − 0.51, p = .004), compared to respondents with a high sense of control. There were certain nuances and predictors that were not significant across all behaviours. Coefficients for each of the predictor variables in each of the models can be seen in Table 3. Table 3Associations of demographic and medical variables with benefit perceptions and behaviours Perceived benefits 1 Behaviours 2 DietPhysical activityAlcoholSmokingDietPhysical activityAlcoholSmokingβ (p value)β (p value)OR (p value)Age.08 (.29).05 (.46).01 (.91)-.09 (.26).11 (.16) -.21 (.01) .01 (.94).76 (.40)Gender (man).01 (.95)-.02 (.87) -.29 (.05) .03 (.85) -.55 (<.01) .13 (.43) .61 (<.01) 10.42 (<.01) Education3 (VMBO)-.48 (.11)-.37 (.20)-.29 (.34) -.96 (<.01) -.30 (.34).02 (.96)-.18 (.57) 21.68 (<.01) Education (HAVO/VWO).39 (.14).28 (.27).14 (.60)-.16 (.54).06 (.82).23 (.42).25 (.37)3.59 (.34)Education (MBO)-.10 (.56)-.19 (.26).05 (.79)-.11 (.52) -.42 (.02) .003 (.98)-.11 (.54)1.37 (.76)Education (WO) .35 (.06) .58 (<.01) .47 (.01) .40 (.03) .20 (.28).08 (.69).01 (.97)1.81 (.53)Time since diagnosis4 (<1 year) -.60 (<.01) -.45 (.03) -.65 (<.01) -.75 (<.01) -.05 (.84)-.10 (.68).36 (.12)2.20 (.42)Time since diagnosis (>5 years) -.31 (.04) -.33 (.02) -.15 (.34).03 (.82)-.01 (.94)-.01 (.93).002 (.99).95 (.95)Control (low) -.66 (<.01) -.82 (<.01) -.65 (<.01) -.39 (.02) -.50 (<.01) -.51 (<.01) -.15 (.39)1.75 (.51)Acceptance (low)-.11 (.44)-.09 (.52)-.07 (.64)-.15 (.31).004 (.98)-.13 (.43)-.16 (.30)4.15 (.09)1Full model for perceived benefits of diet: F(10, 183) = 5.06, p < .001, R2adj = .17 Full model for perceived benefits of physical activity: F(10, 183) = 7.02, p < .001, R2adj = .24Full model for perceived benefits of alcohol: F(10, 183) = 3.93, p < .001, R2adj = .13Full model for perceived benefits of smoking: F(10, 183) = 4.25, p < .001, R2adj = .142Full model for diet behaviour: F(10, 183) = 3.31, p < .001, R2adj =.11 Full model for physical activity: F(10, 179) = 1.92, p = .046, R2adj = .05Full model for alcohol intake: F(10, 177) = 2.85, p = .003, R2adj = .09Logistic regression for smoking behaviour: p = .0013Comparison group for Education variable is HBO 4Comparison group for variable Time since diagnosis is 1–5 year Table 3Associations of demographic and medical variables with benefit perceptions and behaviours Perceived benefits 1 Behaviours 2 DietPhysical activityAlcoholSmokingDietPhysical activityAlcoholSmokingβ (p value)β (p value)OR (p value)Age.08 (.29).05 (.46).01 (.91)-.09 (.26).11 (.16) -.21 (.01) .01 (.94).76 (.40)Gender (man).01 (.95)-.02 (.87) -.29 (.05) .03 (.85) -.55 (<.01) .13 (.43) .61 (<.01) 10.42 (<.01) Education3 (VMBO)-.48 (.11)-.37 (.20)-.29 (.34) -.96 (<.01) -.30 (.34).02 (.96)-.18 (.57) 21.68 (<.01) Education (HAVO/VWO).39 (.14).28 (.27).14 (.60)-.16 (.54).06 (.82).23 (.42).25 (.37)3.59 (.34)Education (MBO)-.10 (.56)-.19 (.26).05 (.79)-.11 (.52) -.42 (.02) .003 (.98)-.11 (.54)1.37 (.76)Education (WO) .35 (.06) .58 (<.01) .47 (.01) .40 (.03) .20 (.28).08 (.69).01 (.97)1.81 (.53)Time since diagnosis4 (<1 year) -.60 (<.01) -.45 (.03) -.65 (<.01) -.75 (<.01) -.05 (.84)-.10 (.68).36 (.12)2.20 (.42)Time since diagnosis (>5 years) -.31 (.04) -.33 (.02) -.15 (.34).03 (.82)-.01 (.94)-.01 (.93).002 (.99).95 (.95)Control (low) -.66 (<.01) -.82 (<.01) -.65 (<.01) -.39 (.02) -.50 (<.01) -.51 (<.01) -.15 (.39)1.75 (.51)Acceptance (low)-.11 (.44)-.09 (.52)-.07 (.64)-.15 (.31).004 (.98)-.13 (.43)-.16 (.30)4.15 (.09)1Full model for perceived benefits of diet: F(10, 183) = 5.06, p < .001, R2adj = .17 Full model for perceived benefits of physical activity: F(10, 183) = 7.02, p < .001, R2adj = .24Full model for perceived benefits of alcohol: F(10, 183) = 3.93, p < .001, R2adj = .13Full model for perceived benefits of smoking: F(10, 183) = 4.25, p < .001, R2adj = .142Full model for diet behaviour: F(10, 183) = 3.31, p < .001, R2adj =.11 Full model for physical activity: F(10, 179) = 1.92, p = .046, R2adj = .05Full model for alcohol intake: F(10, 177) = 2.85, p = .003, R2adj = .09Logistic regression for smoking behaviour: p = .0013Comparison group for Education variable is HBO 4Comparison group for variable Time since diagnosis is 1–5 year Associations of demographic and medical variables with benefit perceptions and behaviours 1Full model for perceived benefits of diet: F(10, 183) = 5.06, p < .001, R2adj = .17 Full model for perceived benefits of physical activity: F(10, 183) = 7.02, p < .001, R2adj = .24 Full model for perceived benefits of alcohol: F(10, 183) = 3.93, p < .001, R2adj = .13 Full model for perceived benefits of smoking: F(10, 183) = 4.25, p < .001, R2adj = .14 2Full model for diet behaviour: F(10, 183) = 3.31, p < .001, R2adj =.11 Full model for physical activity: F(10, 179) = 1.92, p = .046, R2adj = .05 Full model for alcohol intake: F(10, 177) = 2.85, p = .003, R2adj = .09 Logistic regression for smoking behaviour: p = .001 3Comparison group for Education variable is HBO 4Comparison group for variable Time since diagnosis is 1–5 year Exploratory analyses Upon visual inspection of Fig. 1, we conducted an additional exploratory within-subjects MANOVA analysis with type of benefit as the within-subject factor, and diet, physical activity, smoking, and alcohol as outcome variables, to assess whether each behaviour individually was perceived as differently beneficial (i.e., as being more beneficial for general health or cancer). There was a significant difference of type of benefit in the different behaviours, F(1, 196) = 64.31, p < .001). Four separate within-subjects ANOVAs with Bonferroni correction applied8 showed that behaviours had statistically significantly higher scores for general health benefits, compared to cancer-related benefits (diet, F(1, 196) = 183.29, p < .001, η2g = 0.17; physical activity, F(1, 196) = 211.19, p < .001, η2g = 0.21, alcohol, F(1, 196) = 55.26, p < .001, η2g = 0.05, smoking, F(1, 196) = 41.16, p < .001, η2g = 0.03). This means that cancer survivors tend to perceive more benefits of the health behaviours for their general health, than cancer-related outcomes. Furthermore, based on higher generalised eta squared scores, this effect is stronger in perceptions regarding diet and physical activity. Cancer type differences were analysed in a way that cancer types were categorised based on cancer site, meaning respondents were categorised in breast (n = 60), male reproductive system (n = 33), female reproductive system (n = 20), digestive system (n = 20), blood and lymph system (n = 15), urinary (n = 12), multiple cancers (n = 12), skin cancer (n = 8), lung cancer (n = 6), brain cancer (n = 6), and head and neck (n = 5). Between-subjects MANOVA with the overall perceived benefits of diet, physical activity, alcohol, and smoking as dependent variables, and type of cancer as independent variable revealed no significant differences between survivors with different cancer types, F(10, 185) = 1.33, p = .087. Discussion As cancer survivors tend to adhere to some health behaviours, such as refraining from smoking and reduced alcohol intake, more frequently than to other behaviours, such as having a balanced diet and engaging in physical activity [5], we explored differences in perceiving benefits as a potential explanation for these behavioural differences. First, to understand the importance of benefit perceptions, we analysed their relation to survivors’ current health behaviours (RQ1a), and demonstrated that perceiving benefits of a certain health behaviour is related to the performance of that behaviour. This means that the higher the perceptions of the general health and cancer-related benefits the person has, the more likely they are to engage in that behaviour. This held true for perceptions about adhering to a balanced diet and increased physical activity. These results were also demonstrated for lower alcohol intake, and not smoking, however due to limitations related to low variability in the sample, they should be interpreted with caution. These results are in line with theoretical expectations, such as the Health Belief Model [8], and empirical studies demonstrating the association of perceived benefits and behaviour [16, 18, 19]. In general, they hold important implications for increasing engagement in health behaviours such as diet and physical activity, which are often not adhered to by cancer survivors [5]. Although perceiving benefits seems to explain a small proportion of the variation, it is an aspect that could be considered when designing interventions that affect adherence to health behaviours. Furthermore, the results suggest that perceiving general health benefits of diet relates to having a more balanced diet, while this does not hold true for perceiving cancer-related benefits. Due to insignificant results, such conclusions could not be made for the other health behaviours, meaning that we could not infer whether one type of benefit relates to adherence more (RQ1b). The results indicate that cancer survivors perceive not smoking as more beneficial for their general health and cancer than having a balanced diet, engaging in regular physical activity, and lowering alcohol intake (RQ2a). Furthermore, lowering alcohol intake is also perceived as more beneficial than having a balanced diet. Overall, diet and physical activity seemed to be perceived as least beneficial, while smoking tended to be perceived as the most beneficial for their general health and cancer. In our exploratory analyses, we also found that all health behaviours tend to be perceived as more beneficial for general health, in comparison to cancer-related outcomes. This effect is stronger particularly in scores regarding perceived benefits of diet and physical activity, meaning that diet and physical activity in particular are perceived as more beneficial for general health, rather than for cancer-related outcomes. These findings are in line with previous research demonstrating that survivors often underestimate the effect that some health behaviours may have on cancer [19, 22], and tend to agree more with the general health benefits of physical activity, compared to cancer-related benefits [18]. Considering these results, it is important to note that, irrespective of the differences between behaviours, this sample had particularly high scores for perceived benefits of all behaviours. This means that survivors perceived all four health behaviours as beneficial, albeit in different levels. Previous research indicated that having a cancer related to a certain behaviour associates with increased benefit perceptions and behaviour [e.g. 15]. With the idea that some cancers may be perceived as more or less related to certain lifestyle behaviours, we explored whether benefit perceptions may differ across survivors with different cancer types (RQ2b). In our exploratory analyses of cancer type differences, the perceptions did not differ between survivors with different cancer types, implicating that the type of cancer (by our categorisation) does not relate to different benefit perceptions for different health behaviours. Certain patient characteristics emerged as factors that have a role in survivors’ benefit perceptions and adherence (RQ3). Time since diagnosis has been shown to play a role in perceiving benefits of health behaviours, while a sense of control held a role in both perceiving the benefits and adhering to health behaviours. These results bring new insights into factors that may play a role in perceptions, while the results for factors related to adherence somewhat support what has previously been found [e.g. 28]. In line with studies demonstrating that men are more likely to engage in risky health behaviours and less likely to engage in health-promoting behaviours [e.g., 40], gender emerged as an important factor for adherence to health behaviours. This finding could be attributed to factors such social norms and cultural expectations [41]. Age and education did not seem to directly relate to adherence levels, which may suggest these variables influence health behaviours indirectly through factors not measured in our study, such as health literacy. Strengths and limitations This study provides important insights into cancer survivors’ benefit perceptions of various health behaviours, and offers a perspective into how these benefits relate to adherence to lifestyle behaviours. As the results revealed that perceived benefits of diet and physical activity tend to be lower, especially regarding cancer-related benefits, the study brings to light important implications about the need for communication and education about the cancer-related benefits of health behaviours, particularly diet and physical activity. Moreover, the study’s robustness and importance lies in its ability to provide these results even within a sample that already engaged in quite a healthy lifestyle. This study also had limitations which should be taken into consideration when interpreting the results. First, the data were limited by the self-report bias. Although self-reported measurements of health behaviours are widely used in this setting [e.g., 3, 33], objective assessments of health behaviours would provide a more accurate insight into the relation of perceived benefits and behaviour. Another limitation relates to the sample, consisting of survivors who are part of an online cancer community, and are not fully representative of cancer survivors in general. Members of online cancer communities are generally higher educated and younger [42], which implies that if our respondents do not perceive certain benefits, this is presumably even less likely for the general population of cancer survivors. Practical implications and future research Findings from this study suggest that it could be useful to communicate to cancer survivors about the benefits of health behaviours for their general health and cancer-related outcomes. Particularly, cancer survivors may benefit from learning the importance of diet and physical activity for their health. Furthermore, these results suggest that particular attention should be given to men and survivors recently diagnosed when considering which survivors might be in need of lifestyle recommendations. Tailored interventions could include gender-specific educational materials and consultations that address unique barriers and facilitators for health behavior changes in men. For recently diagnosed survivors, the recommendations could focus on increasing awareness of the connection between health behaviors and recovery, as well as long-term survivorship benefits. The findings highlight that there are also actionable factors, such as perceived control, that can be influenced and used in creating interventions to increase the benefit perceptions of health behaviours. Perceiving benefits of a health behaviour seems to be, although small, a component to adhering to that behaviour. The results of this study reflect cross-sectional associations, meaning that a causal relationship between perceived benefits and behaviour cannot be inferred. Future research should focus on exploring the relationship between perceived benefits and health behaviours in a causal manner, to emphasise the importance of communicating these benefits to survivors. As our study did not bring conclusive results on which type of benefit might relate more to adherence to health behaviours, additional research is needed to understand the unique contributions of different types of benefits on behaviour. Previous research has shown that perceiving benefits relating to cancer increases survivors’ chances of implementing positive health behaviours [16], therefore this is an avenue that should be further explored. This can be done either by exploring cancer-related benefits alone, or in comparison to other types of benefits. Exploring what benefits survivors perceive, and which type of benefits contribute more to their behaviour, would promote understanding what should be communicated to elicit a successful behavioural response. Lastly, different aspects and ways of communicating have been found to be associated with health outcomes in survivors [43]. Based on our findings, benefits (particularly benefits of diet and physical activity) should be communicated to cancer survivors, therefore it is pivotal to focus on understanding how these benefits can be communicated in an effective way, to increase their adherence. Conclusion Health behaviours are perceived as differently beneficial and cancer survivors tend to perceive lower alcohol consumption and not smoking as more beneficial for their cancer and general health compared to having a healthy diet and engaging in physical activity. Furthermore, survivors perceive health behaviours as more beneficial for their general health, than for preventing or dealing with cancer. These perceptions could hold an important role in cancer survivors’ adherence to a healthy lifestyle, as our findings show that perceiving benefits of a certain health behaviour associates with engaging in that health behaviour. When designing interventions to improve benefit perceptions and encourage health behaviours among cancer survivors, particular attention should be given to those who are recently diagnosed, men, and survivors with a low perceived control over their disease. These groups may be more vulnerable to not perceiving the benefits and not adhering to certain health behaviours. To encourage adherence, interventions should focus on providing tailored, targeted information about lifestyle changes, particularly related to diet and physical activity. Providing proper information about the benefits of health behaviours can empower cancer survivors to make informed and long-term changes to improve their health outcomes. Supplementary Information Supplementary Material 1. Supplementary Material 1.