Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2014 Oct 15.
Published in final edited form as: Health Commun. 2011 Sep 20;27(4):380–388. doi: 10.1080/10410236.2011.586990

Seeking Cancer-Related Information from Media and Family/Friends Increases Fruit and Vegetable Consumption among Cancer Patients

Nehama Lewis 1, Lourdes S Martinez 2, Derek R Freres 3, J Sanford Schwartz 4, Katrina Armstrong 5, Stacy W Gray 6, Taressa Fraze 7, Rebekah H Nagler 8, Angel Bourgoin 9, Robert C Hornik 10
PMCID: PMC4197929  NIHMSID: NIHMS632349  PMID: 21932985

Abstract

Previous research suggests positive effects of health information seeking on prevention behaviors such as diet, exercise and fruit and vegetable consumption among the general population. The current study builds upon this research by examining the effect of cancer patients’ active information seeking from media and (non-medical) interpersonal sources on fruit and vegetable consumption. The results of this longitudinal study are based on data collected from a randomly drawn sample from the Pennsylvania Cancer Registry, comprising breast, prostate and colorectal cancer patients who completed mail surveys in the Fall of 2006 and 2007. There was a 65% response rate for baseline subjects (resulting N=2013); of those 1293 were interviewed one year later and 845 were available for final analyses. We used multiple imputation to replace missing data and propensity scoring to adjust for effects of possible confounders. There is a positive effect of information seeking at baseline on fruit and vegetable servings at follow-up; seekers consumed 0.43 (95% CI: 0.28 to 0.58) daily servings more than non-seekers adjusting for baseline consumption and other confounders. Active information seeking from media and interpersonal sources may lead to improved nutrition among the cancer patient population.

Keywords: Information seeking, Nutrition, Fruits, Vegetables, Prostate cancer, Breast cancer, Colorectal cancer

Introduction

There is some evidence to suggest a probable association between increased fruit and vegetable consumption and reduced risk for several cancers (Vainio & Bianchini, 2003). It has been suggested that this decrease in cancer risk could be attributable to the molecular components contained within fruits and vegetables (Aggarwal & Shishodia, 2006). Some studies have demonstrated clear and significant associations between fruit and vegetable intake and reductions in risk of breast (Gandini, Merzenich, Robertson, Boyle, 2000), prostate (McCann, Ambrosone, Moysich, Brasure et al. 2005), and colorectal (Wu, Dai, Shrubsole, et al. 2009; Michels, Giovannucci, Chan, et al. 2006) cancers. However, other studies have failed to provide definitive support for the link between fruit and vegetable intake and cancer risk. According to an expert report on the association between intake of fruits and vegetables and cancer (World Cancer Research Fund/American Institute for Cancer Research, 2007), studies conducted since the mid-1990s have “made the overall evidence, that vegetables or fruits protect against cancers, somewhat less impressive. In no case now is the evidence of protection judged to be convincing. However, in a substantial number of cases, a judgment of probable is justified... there is evidence that some types of vegetables, and fruits in general, probably protect against a number of cancers.” (Chapter 4.2, p. 75). Specifically, the report does find probable evidence that consumption of foods containing dietary fiber, found in plant foods (particularly when in whole or relatively unprocessed forms), probably protect against colorectal cancer. The report also found evidence which suggests that food containing selenium (such as brazil nuts, fish, whole grains, wheat germ, and sunflower seeds), and lycopene (most highly concentrated in tomatoes, but also found in watermelon, red (bell) peppers, pink or red grapefruit, pink-fleshed guava, and persimmons) reduces the risk of prostate cancer.

These findings, along with other evidence for the health value of fruit and vegetable consumption, has led to a general recommendation favoring increased fruit and vegetable consumption for everyone. Although research into this topic has yet to provide conclusive evidence to support a causal claim, the American Cancer Society (2009) has recommended that cancer patients increase their fruit and vegetable consumption. Efforts to promote fruit and vegetable consumption are still worthwhile even if data have not demonstrated that increased consumption of fruits and vegetables will have a major effect on cancer incidence. In recent studies among cancer patients, fruit and vegetable consumption has been shown to reduce risks of cardiovascular disease (Dauchet, Amouyel, Hercberg, & Dallongville, 2006; World Health Organization, 2003; Hung, Joshipura, Jiang, et al., 2004) and to reduce blood pressure (Appel, Moore, & Obarzanek, et al., 1997). Thus, whether fruit and vegetable consumption improves health generally, or may specifically prevent recurrence of some types of pre-existing cancer, advocacy of increased fruit and vegetable consumption among cancer patients is justified.

However, if increased consumption is justified, how is it to be produced? There are two likely complementary routes: (1) advocacy by health professionals or others engaging in explicit persuasive outreach, or (2) self-persuasion as people are exposed to information that is available to them through media or interpersonal sources (either because they are passively exposed to it through routine use of media and conversation with others, or because they actively seek such information). For cancer patients, in particular, diagnosis with cancer may motivate an active search for information about lifestyle behaviors that may serve to restore their health and reduce their risk of cancer recurrence. Indeed, in previous studies, we have shown that cancer patients report substantial levels of information seeking. (Nagler, Gray, & Romantan, et al., 2010.) In this paper we turn to the effects question: does such information seeking lead to changes in behavior?

Research on Information Seeking

In the current media environment, patients have access to a wide range of accessible sources of health information. During the year 2005, a conservative search found over 980 references1 to health benefits of fruit and vegetable consumption in major US broadcast, print and web media and over 3000 references to fruits and vegetables and health (Lexis Nexis Academic, 2005). In addition, national sources of health information which include the National Cancer Institute (NCI), the Food and Drug Administration (FDA) and the American Cancer Society (ACS) provided information about the health benefits of fruit and vegetables on their websites as well as other print publications.

The current study contributes to research on information seeking (Rees & Bath, 2001; Zanchetta, Perreault, Kaszap, & Viens, 2007) about cancer-related topics and its impact on cancer decisions and outcomes (Czaja, Manfredi, & Price, 2003; Talosig-Garcia & Davis, 2005). Specifically, we examine whether cancer patients’ seeking for cancer-related information leads to an increase in fruit and vegetable intake over time. This study adds to research showing cross-sectional positive and significant associations between information seeking and healthy lifestyle behaviors in a general population sample such as dieting to lose weight, exercise and fruit and vegetable consumption (Kelly, Niederdeppe & Hornik, 2009). Lagged evidence for a causal relationship, which demonstrates that information seeking leads to engaging in health behaviors, has been seen in other research (Ramirez et al. 2009). However, much of this research has been limited to studying information seeking and its influence on health behaviors among the general public.

The present study is one of the first to examine this question using a cancer patient population consisting of patients with one of three commonly diagnosed cancers (breast – women only, prostate – men only, colorectal), using longitudinal data to test the causal relation between information seeking and fruit and vegetable consumption. As noted above, this paper focuses on the relationship between information seeking specific to fruit and vegetable consumption and fruit and vegetable intake among cancer patients. The analyses reported here come from a broader study of cancer patients described below. Results of separate analyses examining the association between information seeking for other healthy lifestyle behaviors and cancer-related decisions are presented elsewhere (Lee, Gray, & Lewis, 2010; Gray, Armstrong, & DeMichele, et al., 2009). Past research shows an effect of seeking from non-clinical sources on fruit and vegetable consumption in the general population, adjusting for seeking from medical professional sources (Ramirez et al. 2009).

Why is it that seeking cancer-related information from non-clinical sources would influence fruit and vegetable intake? This process may in part be grounded in the shift toward a patient-centered model of health care, which places a higher responsibility on patients in general to be more actively involved in the decisions made about their health (Houston & Ehrenberger, 2001; Levine, 2000; Johnson, 1997). Under this new model of care, it is recognized that patients will be active in making decisions about their own health, and not entirely reliant on their physicians. Responsibility for making such decisions then likely leads to active engagement with available information sources. We then expect that many cancer patients will be actively looking for information to protect their health and, in that context, will locate information about fruit and vegetable intake. We further assume that available information is overwhelmingly positive about the protective benefits of such consumption, and also projects a sense of normative expectation for increasing such consumption. Following the logic of the Theory of Reasoned Action (Fishbein and Aizen, 1975, 2010; Ajzen & Fishbein, 1980; 2010) increased perception of valued benefits and increased perception of normative expectations will lead to more positive intentions and behavior. Thus the causal chain will run from information seeking to increased behavior. We test the following hypothesis:

H1: Seeking for cancer-related information from non-clinical sources at baseline increases the likelihood of consuming fruits and vegetables at follow-up, adjusting for baseline fruit and vegetable consumption as well as the propensity to seek cancer-related information, and other potential confounders.

Materials and Methods

Participants

A sample of patients diagnosed in 2005 (all patients received a cancer diagnosis within a year prior to their participation in this study) with any of three commonly diagnosed cancers (breast – women only, prostate – men only, colorectal) was drawn in the Fall of 2006. The sample, stratified by cancer, was randomly selected from the list of all diagnosed patients through the Pennsylvania Cancer Registry (PCR). By law, Pennsylvania requires that all cancer cases be reported within six months of diagnosis. The sampling frame for the present study was comprised of the approximately 95% of all cancer cases expected to be included in the PCR. Stage four cancer patients (those with metastatic disease) and African-American patients were over-sampled to maximize statistical power for some analyses, although in the weighted analyses reported here they are represented in proportion to their presence in the population in the PCR. In the Fall of 2007, a follow-up survey was administered to the sample drawn in 2006.

Procedure

Cancer patients residing in Pennsylvania received a 61-question paper survey in the mail. The survey asked questions regarding patient characteristics, lifestyle behaviors, and information-seeking about cancer and treatments. In addition to these questions, the survey included questions developed with the specific goals of the present study in mind. Results of interviews with cancer patients and a pilot test of the survey guided question wording and format.

Procedures for data collection followed procedures outlined in Dillman (2000) and have been described elsewhere (Martinez et al. 2009; Lewis et al. 2009). Briefly, patients received an introductory letter, and then a mail questionnaire with a small incentive in 2006 and then a follow-up mail questionnaire one year later, if they agreed to participate in the follow-up study. Participants received questionnaires tailored to their specific cancer but otherwise questionnaires were virtually identical. Response rate for baseline subjects (N=2013) was 65%; 85% of these baseline respondents agreed to be surveyed again and 76% of those who agreed completed the follow up questionnaire (N=1293).

Measures

Measures included in this study were based on self-report, measured at baseline and follow-up. Participants provided information about their level of fruit and vegetable intake by answering two questions: “In the past week, on average, how many servings of fruit did you eat or drink per day? Please include 100% fruit juice, and fresh, frozen or canned fruits.” “In the past week, on average, how many servings of vegetables did you eat or drink per day, not counting potatoes? Please include green salad, 100% vegetable juice, and fresh, frozen or canned vegetables.” Both questions included six response options, ranging from “Less than one serving per day” “one serving per day” “two servings per day” up to “5 or more servings per day.” As in Ramirez et al. (2009), we created a composite measure of fruit and vegetable intake, were we treated the response options for each question as interval, ranging from 0 to 5 levels, then generated a summed variable ranging from 0 to 10 levels. While the extreme categories (0 and 5) of each measure are ordinally related to the adjoining categories, the remaining categories retain meaning at the interval level. For the full 0-10 scale less than 5% of all respondents fall at those extremes. We felt confident then in treating this measure as interval. The one year test-retest correlation for the summed scale is high (0.64.)

Using a conceptualization of information seeking from previous research (Shim, Kelly & Hornik, 2006; Niederdeppe, et al. 2007) which defined seeking as any active effort to look for information related to cancer, we asked respondents to report their levels of seeking in the past year. Respondents were given three distinct opportunities to report seeking from non-clinical sources in the form of three questions: 1) “Think back to the first few months after you were diagnosed with your cancer. In making decisions about what treatments to choose, did you actively look for information about treatments from any sources?” 2) “What sources did you use when you were actively looking for any information related to your cancer?” 3) “Where have you actively looked for information about quality of life issues (i.e. physical problems, how to reduce cancer recurrence). Responses for each question included a list of various non-clinical sources (friends or family members, other cancer patients, television or radio, books or brochures, newspapers or magazines, support groups [face-to-face and online], and telephone hotlines) in which respondents were asked to check all that applied. Additionally, each question offered an option for respondents to indicate if they had not actively looked for information. The number of sources for each respondent was then tallied, and each respondent was assigned a seeking score, which was then mean-centered and summed to create an index of seeking. The variable was then dichotomized at the median split, where 0 represented no or lower levels of seeking and 1 represented higher levels of seeking2. This was our primary independent variable for seeking from non-clinical sources as it was entered into the propensity model.

A separate measure for assessing level of seeking from medical professional sources (including seeking from treating doctor and other doctors or health professionals) was included in analyses to establish that any effects were not an artifact of the association of non-clinical seeking and clinical information seeking. The variable was a summed six-level score for responses to the three seeking items (the same items that were used to create the dichotomous non-clinical information seeking variable). Patients who reported seeking from their doctor only or from another medical professional only were assigned a score of 1 for each of these items. Patients who sought from both medical sources were assigned a score of 2, and patients who did not seek from either source were assigned a score of 0 for that item. The ordinal level summed variable is the one used in the final model. As a note, because we examined a population already diagnosed with cancer, seeking from medical professional sources was an almost universal event. Since we were only truly interested in testing the effects of seeking from non-clinical sources over and above any seeking from medical professional sources, the very few patients who reported no seeking from medical professional sources across the three seeking items were excluded from the analyses.

In addition to these variables and ones mentioned earlier, the final propensity model included thirty-four demographics and patient characteristics that were incorporated as possible confounders for causal claims about the seeking – fruit and vegetable consumption relationship. These variables include participants’ age, gender, race, education, employment and marital status, insurance status, smoking status, height and weight, current exercise and diet behaviors and future intention to exercise or diet to lose weight. Other confounders include responses to questions about cancer-related treatments heard about and received, as well as self-reported satisfaction with treatments received and with clinical information provided. The model also accounted for patients’ responses concerning their experience of physical symptoms, preferences for medical decision-making and actual decision-making practices, concern about cancer recurrence, tendency to avoid health information, and to comply with lifestyle recommendations as well as self-reported feelings of confidence and optimism, and whether participants were assisted by others in their social network in seeking cancer-related information.

Data Analysis

As described above, our purpose in this study is to establish evidence that information seeking from non-clinical media and interpersonal sources influences fruit and vegetable consumption among a cancer patient population. Our purpose in this study is two-fold: to show through analysis of longitudinal data that the flow of causal order between seeking and fruit and vegetable consumption is one that runs from seeking towards fruit and vegetable consumption, and to capture an estimated causal effect of seeking from non-clinical media and interpersonal sources and fruit and vegetable consumption among a cancer patient population. Evidence for a causal effect of cancer patients’ seeking from non-clinical sources on fruit and vegetable consumption must satisfy two conditions. The first requires that a statistically significant association exists between seeking at baseline and fruit and vegetable consumption at follow-up, controlling for fruit and vegetable consumption at baseline. We tested this condition using a series of ordinary least squares regression models. The second condition requires that the observed association seen under the first condition is not attributable to other variables. We adjusted for any effect of potential confounders by using an approach known as propensity scoring. Propensity scoring attempts to replicate the strength of random assignment in experimental designs by comparing subjects matched on their probability of seeking conditional on potential confounders (Rosenbaum & Rubin, 1983). In this way, propensity scoring allows us to compare levels of fruit and vegetable consumption across subjects who are equally likely to seek at baseline while adjusting for effects of confounders.

This procedure allows us to then test the effects of seeking conditional on propensity score. The decision to use propensity scoring rests on the advantages conferred by this approach. Compared to conventional methods, through propensity scoring we are able to see if measured covariates are out of balance between conditions prior to calculating a final treatment effect (Rosenbaum, 2002; Yanovitzky, Hornik & Zanutto, 2008). A significant final treatment effect of seeking that adjusts for potential confounders would support our contention that cancer patients’ seeking from non-clinical sources influences their fruit and vegetable consumption. Like regression, propensity scoring cannot account for possible effects of unmeasured variables.

To generate the propensity score, we entered thirty-four confounders into a logistic regression model, with seeking from non-clinical sources as the dependent variable. Subjects with equal propensity for seeking at baseline were then compared for differences in fruit and vegetable consumption associated with baseline seeking. Using standard propensity modeling approaches, we excluded low- and high-seeking cases that did not overlap, thus dropping high-seeking cases with propensity scores higher than the highest low-seeking cases, and low seeking cases with propensity scores lower than the lowest high-seeking cases. This reduced our sample size for analyses involving propensity scoring from 852 (described below) to 845. The propensity scores were then divided into quintiles, with subjects in each quintile sharing equal probabilities of seeking. The final treatment effect was estimated using ordinary least squares regression, with fruit and vegetable consumption at baseline, seeking and propensity as independent variables. An additional test of the interaction of seeking and propensity was not significant and thus excluded from the final treatment effect model.

As with many longitudinal studies, the problem of missing data had to be addressed. Using the multiple imputation procedure as recommended by Allison (2001) and Little & Rubin (2002), we created ten datasets with imputed values using STATA 10 and the ICE program (Royston, 2005). We then used the MIM program (Galati, Royston, & Carlin, 2007) to average values across ten datasets, translating into parameter estimates for each imputed variable. Post-stratification weights based on age, sex, marital status, date of diagnosis, race-ethnicity, and stage of disease were also applied to reflect the distribution of cases in the PCR across cancers.

Results

Descriptive Statistics

The response rate for baseline respondents (N=2013) was 65%. By cancer the response rates at baseline were 61% for colorectal cancer participants, 68% for breast cancer participants, and 64% for prostate cancer participants. 1701 patients who had agreed to participate in a follow-up round, and who were not known to have died in the interim period, were sent follow-up questionnaires. Of these 1293 sent back follow-up questionnaires, a response rate of 76%. By cancer the follow-up response rates were 75% for colorectal cancer participants, 79% for breast cancer participants, and 77% for prostate cancer participants. Principal variables, demographics, and patient characteristics remained similarly distributed upon applying weights and imputing values. After running the multiple imputation procedure, we restricted our sample to include only subjects who responded to the questions about fruit and vegetable consumption at baseline and follow-up, which comprised a total of 964 subjects. We further excluded people (12% of the restricted sample) who did not report seeking from either their own physician or another medical professional producing a sample of 852, which was reduced finally to 845 under the propensity score procedures described previously.3 As distributions of sample characteristics did not significantly change after applying weights, Table 1 displays the only the weighted demographic characteristics of the sample following the imputation procedure. Table 2 shows how respondents varied from baseline to follow-up on primary variables. As seen in Table 2, fruit and vegetable consumption remained fairly constant from baseline to follow-up. At baseline, the mean number of fruit and vegetable servings consumed was 3.89 [95% CI: 3.75, 4.03], and 4.09 [95% CI: 3.95, 4.23] at follow-up. These are not significantly different.

Table 1.

Sample Demographics (N=852)

Demographics %
Age
    24-50 years 13.3
    51-60 years 25.4
    61-70 years 29.1
    71-80 years 23.0
    81-105 years 9.2
Gender
    Male 48.4
    Female 51.6
Race/ethnicity
    White 86.9
    Non-white 13.1
Education
    Eighth grade or less 2.3
    Some high school 9.8
    High school diploma 37.9
    Some college/2 year degree 23.2
    College graduate 26.8
Marital Status
    Married 72.6
    Not married 27.4
Cancer
    Breast 35.3
    Prostate 34.1
    Colon 30.7
Stage of Cancer
    In Situ and local 63.1
    Regional Spread 22.9
    Metastatic 14.1
Open in a new tab

Table 2.

Baseline and Follow-up levels of seeking and fruit and vegetable consumption (N=852)

Variables
Baseline Scores Follow-Up Scores
Seeking from Non-Clinical Sources (% seeking) 49.8% 46.8%
Fruit and Vegetables (mean (sd)) daily number of servings 0-10) 3.90 (2.08) 4.10 (2.14)
Open in a new tab

As with other measured variables, seeking from non-clinical sources remained constant across data points: 49% reported seeking from non-clinical sources at baseline and 46% reported doing so at follow up. Table 3 shows the correlations among the primary variables. All correlations among primary variables were statistically significant at the 0.01 level or less and were positive.

Table 3.

Correlation Matrix of Primary Variables (N=852)

Seeking (Baseline) Seeking (Follow-Up) Fruit and Vegetables (Baseline)
Seeking (Follow-Up) 0.422
Fruit and Vegetables (Baseline) 0.092 0.133
Fruit and Vegetables (Follow-Up) 0.126 0.085 0.638
Open in a new tab

All displayed variables are statistically significant at p < .01 or less.

Estimation of Results

Information seeking from non-clinical sources was positively associated with eating fruits and vegetables cross-sectionally at baseline (b=0.65, p<0.001) in Model 1 and at follow-up (b=0.37, p=0.005) in Model 2 of Table 4. Model 2 also shows that the effect of seeking at baseline on fruit and vegetable consumption at follow-up remains significant upon adjusting for the effect of fruit and vegetable consumption at baseline (b=0.65, p<0.001). The propensity score analysis adjusted for potential confounders (and excluded cases which had no overlap in propensity scores between seekers and non-seekers. appears in Model 3 (see Table 4.) This propensity-controlled model includes four dummy indicators for the seeking propensity quintiles. Interactions between seeking and propensity quintiles were tested using a likelihood ratio test. The test indicates these interactions did not make significant contributions (F (4, 1000) = 2.08, n.s.), and thus are excluded from the model. Two confounders (level of seeking from medical professional sources and insurance) were found to be out of balance in the generation of the propensity scores and were included in Model 3 as covariates. The propensity-controlled model (adjusting for effects of the confounders that were out of balance) shows that high seekers at baseline reported eating on average nearly half a serving of fruit and vegetables more per day one year later (B=0.43, p=0.01) compared to low seekers at baseline, adjusting for their baseline consumption (Table 4).

Table 4.

Cross-sectional and Lagged Relationships of Seeking from Cancer-Related Information from Non-Clinical Sources with Fruit and Vegetable Consumption

Model 1 (N=852) b (S.E.) Model 2 (N=852) b (S.E.) Model 3 (N=845) b (S.E.)
Seeking from Non-Clinical Sources (Baseline) 0.66(0.17)*** 0.37(0.13)** 0.43(0.17)**
Fruit and Vegetables (Baseline) 0.65(0.03)*** 0.65(0.03)***
Level of Seeking from Medical
Professional Sources 0.01(0.04)
Insurance 0.06(0.29)
R-Square 0.415 0.423 0.425
Open in a new tab

*p <.05

**

p<.01

***

p<.001.

Model 1: Cross-sectional association: baseline seeking and outcome behavior

Model 2: Lagged association, adjusting for baseline behavior

Model 3: Lagged association, adjusting for baseline behavior, propensity quintiles, and unbalanced confounders

Discussion

The central finding of this study, that cancer patients’ seeking cancer-related information from non-clinical sources predicts fruit and vegetable consumption adjusting for baseline fruit and vegetable consumption and an array of confounders, supports our hypothesis and is consistent with previous research examining similar questions among the general population (Ramirez et al. 2009, Kelly, Niederdeppe & Hornik, 2009). Our study is the first to show a controlled lagged association between information seeking about cancer-related information from non-clinical sources and fruit and vegetable consumption among a cancer population. This finding indicates that active information seeking about cancer-related information from non-clinical sources may lead to improved nutrition among the cancer patient population.

This finding suggests that, among a population afflicted with chronic disease such as cancer, information seeking from non-clinical sources matters with regards to engaging in healthy behaviors. Given the recent changes in the current model of healthcare and a continued focus on patients’ responsibility for their health, this may be particularly true for cancer patients who desire more information about prevention over and above what they receive from their treating physicians. This desire to engage in healthy behaviors may be particularly salient if cancer patients believe they are at high risk of recurrence and that a diet rich in fruits and vegetables may confer protective benefits. Despite having what could be reasonably expected to be a comparatively greater motivation to live a healthier life, most cancer patients in our study consumed servings of fruit and vegetables slightly below the daily recommended five servings encouraged by the American Cancer Society (2009). We find some support for this result in existing research suggesting that cancer survivorship in itself does not prompt considerable changes in related healthy lifestyle habits such as weight control and physical activity (Courneya, Katzmarzyk, Bacon, 2008).

Since our central finding shows that different levels of seeking cancer-related information from non-clinical sources matter for increasing fruit and vegetable consumption above and beyond the effect of seeking from clinical sources, some may wonder whether this is merely an artifact of engagement with medical professional sources. In the current analysis this was dealt with by limiting the analysis to those who reported some information seeking from clinical sources. In model 3 in Table 4, we also show that the effects of seeking from non-clinical sources remained even when level of seeking from medical professional sources was controlled.

The longitudinal design and propensity analysis demonstrated here offers substantial and novel evidence for the positive influence of seeking for cancer-related information from non-clinical sources on fruit and vegetable consumption among cancer patients. Given the use of propensity scoring procedures it also controls for effects of a wide range of potential confounding variables. The lagged design, with its control for baseline levels of fruit and vegetable consumption, does account for unmeasured variables which affect both seeking and fruit and vegetable consumption simultaneously. It also is not threatened by uncertainties about the causal order of the two variables, specifically the possible effects of fruit and vegetable consumption on seeking. Still, the present study has a number of limitations. Perhaps the most substantial threat to influence of seeking from non-clinical sources on fruit and vegetable consumption is the possibility of a non-measured third variable, which might drive both the likelihood of seeking as well as the increase in fruit and vegetable consumption but at different times. For example, our analysis does not include a measure of participants’ motivation to address their health issues; this is an example of a variable which might positively influence cancer patients’ motivation to seek information earlier but only affect their fruit and vegetable consumption later.

An additional limitation of the study is that reported frequency of information seeking and of fruit and vegetable consumption are based upon patients’ recall, and thus may potentially be an inaccurate reflection of the actual frequency of these events. A further limitation of the study is that the measures used have not been previously validated. However, a pre-test of the measures was conducted with cancer patients, and existing literature was used to guide the development of these measures. Given the high response rates, the sample likely represents the 2005 population of diagnosed breast, prostate, and colorectal cancer patients in Pennsylvania, but it does not represent patients in other states or with other cancers or in other time periods.

Another limitation of our study relates to the limitations of the statistical analysis used to answer the central questions of this research, which have not permitted us to perform a dose-response analysis. This raises an important question not addressed within this study - the examination of a dose-response effect of seeking on fruit and vegetable consumption. Examination of a dose-response effect would be of interest as it would identify a threshold point at which one could expect the effect of seeking to begin to influence fruit and vegetable consumption. Future research employing less confining statistical methods may explore this question

Finally, this study uses a sample consisting of patients with only three cancer types. However, it should be noted that breast cancer, colorectal cancer and prostate cancer are three of the most prevalent cancer types affecting the US population today (American Cancer Society [ACS], 2008). Consequently, our sample represents a sizeable population of patients diagnosed with cancer in the US. Breast cancer ranks second as a cause of cancer death in women (after lung cancer) (ACS, 2008, pg. 9). Prostate cancer is the most frequently diagnosed cancer in men. Colorectal cancer is the third most common cancer in both men and women. Finally, our measure of information seeking only assess breadth of seeking - whether information seeking occurred and from where, rather than depth - the frequency with which information seeking took place or sources were consulted. One potential avenue for future research may be to investigate reliance on specific non-clinical information sources and how it translates into engaging in healthy behavior such as fruit and vegetable consumption. Other further research may also examine the relation between information seeking from non-clinical sources among individuals in non-chronic or non-life threatening situations.

Summary Implications

The recent emphasis on empowering the patient in a clinical setting within the health care system has placed a greater responsibility on the part of the patient to be more aware of how their current health behaviors and choices translate into health outcomes. From the results of this study, it is clear that cancer patients look for cancer-related information above and beyond the information obtained through physicians, perhaps as a result of absorbing more responsibility for decisions related to their health behavior. This desire for supplemental information may become more acute over time as patients in general are expected to assume more active roles in their health care. The results presented here also show the influence that exposure to non-clinical information sources such as media and interpersonal sources can have on cancer patients’ engagement in health behaviors such as fruit and vegetable consumption. Based on these findings, physicians and health professionals may consider a dual-pronged approach to make the most of the potential health benefits resulting from seeking behavior. This entails encouraging cancer patients to seek cancer-related information from non-clinical sources and promoting more media coverage of healthy behaviors, thus shaping available cancer-related information in the media environment.

Acknowledgments

The authors are grateful to Susana Ramirez, Bridget Kelly, Chul-joo Lee, Susan Mello, Anca Romantan, Aaron Smith-McLallen, Sally Dunlop, Mihaela Johnson, Sara Parvanta, Dina Shapiro, Andy Tan, and Norman Wong for their insightful comments on earlier drafts of the manuscript and/or contributions to instrument development, data collection, and coding, and to Robin Otto, Craig Edelman and personnel at the Pennsylvania Cancer Registry for collaboration on sample development. The Pennsylvania Department of Health specifically disclaims responsibility for any analyses, interpretations or conclusions.

Financial Support

The authors wish to acknowledge the funding support of the National Cancer Institute's Center of Excellence in Cancer Communication (CECCR) located at the Annenberg School for Communication, University of Pennsylvania (P50-CA095856 and P20-CA095856).

Footnotes

1

Search term: ‘fruit or vegetable’ and ‘health’ and ‘benefit’.

2

We recognize that the use of a dichotomized measure restricts sensitivity to detect effects. However, in this case, the dichotomized measure permitted us to make use of propensity scoring procedures to fully account for the effects of confounders, and we thought the tradeoff worthwhile.

3

An analysis comparing the 845 respondents retained for the analysis with the remainder of the baseline sample showed no substantial differences. .

Contributor Information

Nehama Lewis, Department of Psychology Florida International University Miami, FL, USA.

Lourdes S. Martinez, Annenberg School for Communication University of Pennsylvania Philadelphia, PA, USA lmartinez@asc.upenn.edu

Derek R. Freres, Annenberg School for Communication University of Pennsylvania Philadelphia, PA, USA dfreres@asc.upenn.edu

J. Sanford Schwartz, Department of Medicine and The Wharton School Philadelphia, PA, USA University of Pennsylvania schwartz@wharton.upenn.edu.

Katrina Armstrong, Department of Medicine University of Pennsylvania Philadelphia, PA, USA karmstro@mail.med.upenn.edu.

Stacy W. Gray, Dana-Farber Cancer Institute Harvard University Boston, MA, USA StacyW_Gray@dfci.harvard.edu

Taressa Fraze, Healthcare Thomson Reuters Taressa.fraze@thomsonreuters.com.

Rebekah H. Nagler, Dana-Farber Cancer Institute Harvard School of Public Health rnagler@hsph.harvard.edu.

Angel Bourgoin, Annenberg School for Communication University of Pennsylvania Philadelphia, PA, USA abourgoin@asc.upenn.edu.

Robert C. Hornik, Annenberg School For Communication University of Pennsylvania Philadelphia, PA, USA rhornik@asc.upenn.edu

References

  1. Aggarwal BB, Shishodia S. Molecular targets of dietary agents for prevention and therapy of cancer. Biochemical Pharmacology. 2006;71(10):1397–1421. doi: 10.1016/j.bcp.2006.02.009. [DOI] [PubMed] [Google Scholar]
  2. Ajzen I, Fishbein M. Understanding attitudes and predicting social behavior. Prentice-Hall; Englewood Cliffs, NJ: 1980. [Google Scholar]
  3. American Cancer Society Nutrition after treatment ends. 2009 Available at: http://www.cancer.org/docroot/MBC/content/MBC_6_2X_Nutrition_after_treatment_ends.asp.
  4. American Cancer Society . Cancer Facts & Figures 2008. American Cancer Society, Inc.; Atlanta: 2008. [Google Scholar]
  5. Allison PD. Missing data. Sage; Thousand Oaks, CA: 2001. [Google Scholar]
  6. Appel LJ, Moore TJ, Obarzanek E, et al. A clinical trial of the effects of dietary patterns on blood pressure. DASH Collaborative Research Group. New England Journal of Medicine. 1997;336(16):1117–1124. doi: 10.1056/NEJM199704173361601. [DOI] [PubMed] [Google Scholar]
  7. Arora NK, Hesse BW, Rimer BK, Viswanath K, Clayman ML, Croyle RT. Frustrated and confused: The American public rates its cancer-related information-seeking experiences. Journal of General Internal Medicine. 2008;23(3):223–228. doi: 10.1007/s11606-007-0406-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Cecil DW, Killeen I. Control, compliance, and satisfaction in the family practice encounter. Family Medicine. 1997;29:653–657. [PubMed] [Google Scholar]
  9. Courneya KS, Katzmarzyk PT, Bacon E. Physical activity and obesity in Canadian cancer survivors: Population-based estimates from the 2005 Canadian Community Health Survey. Cancer. 2008;112(11):2475–82. doi: 10.1002/cncr.23455. [DOI] [PubMed] [Google Scholar]
  10. Czaja R, Manfredi C, Price J. The determinants and consequences of information seeking among cancer patients. Journal of Health Communication. 2003;8(6):529–562. doi: 10.1080/716100418. [DOI] [PubMed] [Google Scholar]
  11. Dauchet L, Amouyel P, Hercberg S, Dallongeville J. Fruit and vegetable consumption and risk of coronary heart disease: a meta-analysis of cohort studies. Journal of Nutrition. 2006;136(10):2588–2593. doi: 10.1093/jn/136.10.2588. [DOI] [PubMed] [Google Scholar]
  12. Dillman DA. Mail and Internet Surveys: The Tailored Design Method. Wiley; New York: 2000. [Google Scholar]
  13. Fishbein M, Ajzen I. Predicting and changing behavior: The reasoned action approach. Psychology Press (Taylor & Francis); New York: 2010. [Google Scholar]
  14. Fishbein M, Ajzen I. Belief, attitude, intention, and behavior: An introduction to theory and research. Addison-Wesley; Reading, MA: 1975. [Google Scholar]
  15. Galati JC, Royston P, Carlin JB. MIM stata module to analyze and manipulate multiply imputed datasets. 2007 Available at: http://ideas.repec.org/c/boc/bocode/s456825.html#author.
  16. Gandini S, Merzenich H, Robertson C, Boyle P. Meta-analysis of studies on breast cancer risk and diet: the role of fruit and vegetable consumption and the intake of associated micronutrients. European Journal of Cancer. 2000;36(5):636–646. doi: 10.1016/s0959-8049(00)00022-8. [DOI] [PubMed] [Google Scholar]
  17. Gerteis M, Edgman-Levitan S, Daley J, Delbanco TL. Through the Patient's Eyes. Jossey Bass; San Francisco, CA: 1993. [Google Scholar]
  18. Gray SW, Armstrong K, DeMichele A, Schwartz JS, Hornik R. Colon cancer patient information seeking and the adoption of targeted therapy for on-label and off-label indications. Cancer. 2009;115(7):1424–1434. doi: 10.1002/cncr.24186. PMID: 19235785. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Houston TK, Ehrenberger HE. The potential of consumer health informatics. Seminars in Oncology Nursing. 2001;17:41–48. doi: 10.1053/sonu.2001.20418. [DOI] [PubMed] [Google Scholar]
  20. Hung HC, Joshipura KJ, Jiang R, et al. Fruit and vegetable intake and risk of major chronic disease. Journal of the National Cancer Institute. 2004;96(21):1577–1584. doi: 10.1093/jnci/djh296. [DOI] [PubMed] [Google Scholar]
  21. Johnson JD. Cancer-related information seeking. Hampton Press, Inc.; Cresskill, NJ: 1997. [Google Scholar]
  22. Kelly B, Niederdeppe J, Hornik R. Validating measures of scanned information exposure in the context of cancer prevention and screening behaviors. Journal of Health Communication. 2009;14(8):721–740. doi: 10.1080/10810730903295559. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Laine C, Davidoff F. Patient-centered medicine: a professional evolution. Journal of the American Medical Association. 1996;275:152–56. [Google Scholar]
  24. Lee C-J, Gray SW, Lewis N. Internet use leads cancer patients to be active health care consumers. Patient Education & Counseling. 2010;81(S1):S63–S69. doi: 10.1016/j.pec.2010.09.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Levine SR. Consumer-driven health care: A path to achieving shared goals. Physician Executive. 2000;26:10–16. [PubMed] [Google Scholar]
  26. Lewis N, Gray SW, Freres D, Hornik RC. Cross source engagement with cancer-related information and its impact on doctor-patient relations. Health Communication. 2009;24(8):723–734. doi: 10.1080/10410230903264030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lexis Nexis Academic, search terms . fruit and vegetables and health. Major U.S. and World Publications, and Web Publications; (1/1/2005 to 12/31/2005), Retrieved: 1/17/2010. [Google Scholar]
  28. Little RJA, Rubin DB. Statistical Analysis with Missing Data. 2nd edition John Wiley & Sons; New York: 2002. [Google Scholar]
  29. Martinez LS, Schwartz JS, Freres D, Fraze T, Hornik RC. Patient-Clinician Information Engagement Increases Treatment Decision Satisfaction Among Cancer Patients Through Feeling of Being Informed. Patient Education and Counseling. 2009;77(3):384–390. doi: 10.1016/j.pec.2009.09.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. McCann SE, Ambrosone CB, Moysich KB, Brasure J, Marshall JR, Freudenheim JL, Wilkinson GS, Graham S. Intakes of Selected Nutrients, Foods, and Phytochemicals and Prostate Cancer Risk in Western New York. Nutrition and Cancer. 2005;53(1):33–41. doi: 10.1207/s15327914nc5301_4. [DOI] [PubMed] [Google Scholar]
  31. Michels KB, Giovannucci E, Chan AT, Singhania R, Fuchs CS, Willett WC. Fruit and vegetable consumption and colorectal adenomas in the Nurses’ Health Study. Cancer Research. 2006;66:3942–53. doi: 10.1158/0008-5472.CAN-05-3637. [DOI] [PubMed] [Google Scholar]
  32. Mooney KH. Cancer survivors information seeking behaviors: A comparison of survivors who do and do not seek information about cancer. Patient Education and Counseling. 2007;65:342–350. doi: 10.1016/j.pec.2006.08.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Nagler RH, Gray SW, Romantan A, Kelly BJ, DeMichele A, Armstrong K, Schwartz JS, Hornik RC. Differences in information seeking among breast, prostate, and colorectal cancer patients: Results from a population-based survey. Patient Education and Counseling. 2010;81:54–62. doi: 10.1016/j.pec.2010.09.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Niederdeppe J, Hornik RC, Kelly BJ, Frosch DL, Romantan A, Stevens RS, Barg FK, Weiner JL, Schwartz JS. Examining the dimensions of cancer-related information seeking and scanning behavior. Health Communication. 2007;22(2):153–167. doi: 10.1080/10410230701454189. [DOI] [PubMed] [Google Scholar]
  35. Niederdeppe J, Frosch DL, Hornik RC. Cancer news coverage and information seeking. Journal of Health Communication. 2008;13:181–199. doi: 10.1080/10810730701854110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Rains SA. Seeking health information in the information age: The role of internet self-efficacy. Western Journal of Communication. 2008;72(1):1–18. [Google Scholar]
  37. Ramirez SA, Martinez LS, Freres D, Lewis N, Hornik RC. Information seeking from media and family/friends increases the likelihood of engaging in healthy lifestyle.. Presented at the International Communication Association (ICA) annual meeting; Chicago: May, 2009. 2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Rao JK, Weinberger M, Kroenke K. Visit-specific expectations and patient-centered outcomes: a literature review. Archives of Family Medicine. 2000;9:1148–1155. doi: 10.1001/archfami.9.10.1148. [DOI] [PubMed] [Google Scholar]
  39. Rees CE, Bath PA. Information-seeking behaviors of women with breast cancer. Oncology Nursing Forum. 2001;28:899–907. [PubMed] [Google Scholar]
  40. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70(1):41–55. [Google Scholar]
  41. Rosenbaum PR. Observational studies. 2nd ed. Springer Verlag; New York: 2002. [Google Scholar]
  42. Roter D, Hall J. Physician gender and patient-centered communication: A critical review of empirical research. Annual Review of Public Health. 2004;25:497–519. doi: 10.1146/annurev.publhealth.25.101802.123134. [DOI] [PubMed] [Google Scholar]
  43. Royston P. Multiple imputation of missing values: update. Stata Journal. 2005;5:188–201. [Google Scholar]
  44. Shim M, Kelly B, Hornik R. Cancer information scanning and seeking behavior is associated with knowledge, lifestyle choices, and screening. Journal of Health Communication. 2006;11:157–172. doi: 10.1080/10810730600637475. [DOI] [PubMed] [Google Scholar]
  45. Talosig-Garcia M, Davis SW. Information-seeking behavior of minority breast cancer patients: An exploratory study. Journal of Health Communication. 2005;10(S1):53–64. doi: 10.1080/10810730500263638. [DOI] [PubMed] [Google Scholar]
  46. Vainio H, Bianchini F. Fruit and vegetables. IARC handbooks of cancer prevention. IARC Press; Lyon, France: 2003. p. 376. [Google Scholar]
  47. World Cancer Research Fund/American Institute for Cancer Research . Second Expert Report: Food, Nutrition, Physical Activity, and the Prevention of Cancer: A Global Perspective [report] World Cancer Research Fund; London: 2007. [Google Scholar]
  48. World Health Organization . Diet, Nutrition and the Prevention of Chronic Diseases: Report of a Joint WHO/FAO Expert Consultation (Report 916) World Health Organization; FAO; Geneva, Switzerland: 2003. [PubMed] [Google Scholar]
  49. Wu H, Dai Q, Shrubsole MJ, Ness RM, Schlundt D, Smalley WE, Chen H, Li M, Shyr Y, Zheng W. Fruit and vegetable intakes are associated with lower risk of colorectal adenomas. Journal of Nutrition. 2009;139(2):340–344. doi: 10.3945/jn.108.098889. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Yanovitzky I, Hornik R, Zanutto E. Estimating causal effect in observational studies: The propensity score approach. In: Hayes AF, Slater MD, Snyder LB, editors. The Sage sourcebook of advanced data analysis methods for communication research. Sage; Thousand Oaks, CA: 2008. [Google Scholar]
  51. Zanchetta MS, Perreault M, Kaszap M, Viens C. Patterns in information strategies used by older men to understand and deal with prostate cancer: An application of the modelisation qualitative research design. International Journal of Nursing Studies. 2007;44:961–972. doi: 10.1016/j.ijnurstu.2006.03.018. [DOI] [PubMed] [Google Scholar]

RESOURCES