Evolving Information Needs among Colon, Breast, and Prostate Cancer Survivors: Results from a Longitudinal Mixed-Effects Analysis

There are currently close to 14 million cancer survivors in the United States (1), and the population of cancer survivors is estimated to increase to 18 million Americans by 2022 (2). For the oncology community, an important implication of this growing population of cancer survivors is the recognition that survivorship is a crucial phase in the continuum of cancer care. Accordingly, clinicians and researchers have called for greater efforts to understand and address the long-term medical and psychosocial issues that confront individuals diagnosed with cancer as they transition from their active treatment phase to cancer survivorship (3).

In the 2006 Institute of Medicine (IOM) report, From Cancer Patient to Cancer Survivor: Lost in Transition, an expert committee outlined four essential components of cancer survivorship care: (i) preventing recurrent cancers, new cancers, and late effects, (ii) surveillance for cancer spread, recurrence, second cancers, medical, and psychosocial late effects, (iii) intervention for consequences of cancer and its treatment, and (iv) coordination between specialists and primary care providers to meet the survivor's health needs (4). The committee further recommended several strategies to address each of the components of survivorship care. A more recent IOM report, Delivering High-Quality Cancer Care, further emphasized the importance of maintaining the health of survivors and managing long-term and late effects of cancer treatment as key components of delivering high-quality cancer care across the care continuum (5).

This study focuses on the first two components of survivorship care outlined in the 2006 IOM report. Specifically, the study aims to describe how cancer survivors cope with their concerns about recurrence risks, cancer risks of family members, and late effects by actively seeking information about these issues over the course of their survivorship. It is well documented that cancer patients frequently search for information about diagnosis and treatment (3, 6–8). However, less is known about whether they seek information related to survivorship when they transition beyond the active treatment period and when in the survivorship period these concerns become most salient.

Using longitudinal survey data from a population-based cohort of breast, colon, and prostate cancer survivors, we track the evolution of their information seeking, over the course of 3 years, for five distinct survivorship topics: (i) how to reduce the chance of cancer recurrence, (ii) how to reduce the chance of getting a different cancer, (iii) how to reduce the risk of family members getting the same cancer, (iv) whether family members are at risk of a different cancer, and (v) whether one is at risk of (other) health problems due to the cancer or its treatment. Earlier studies tended to conceptualize these concerns broadly as fears of recurrence or concerns about the future (4, 9). Recently, some researchers have argued that each of these concerns should be regarded as distinct survivorship concerns (10).

These five cancer survivorship topics are important because in prior studies survivors reported wanting information about how to manage their risk of recurrence, their family's risk of developing cancer, and long-term health problems resulting from their illness and treatment (11, 12). In addition, these concerns could be associated with adverse psychosocial outcomes, including cancer worry and reduced quality of life (13–18). Understanding whether certain survivors are experiencing concerns about these issues and whether they are coping with these concerns by looking for information would aid clinicians in providing survivors with timely and appropriate information on managing risk of recurrence and late effects of treatment. This could potentially alleviate survivors' psychologic distress about these concerns (19).

In addition, the article addresses two essential questions: whether seeking information about these concerns varies over time and across cancers. We investigate whether there are differences in the pattern of seeking information about the above five topics across survivors with different cancer diagnoses. Both the expected time course of cancer recurrence and the emergence of late effects vary widely across cancer diagnoses (20, 21). Thus, we hypothesized that survivors diagnosed with different cancers may experience concerns about risks of cancer recurrence, new cancers, or late effects at different time points during the survivorship period and therefore diverge in their need for information about these issues. However, there are a limited number of studies available to offer firm directional hypotheses for the associations between time from diagnosis and cancer type and information seeking about each of the five survivorship topics in our present study. From prior studies among cancer survivors (7, 15), we expected that seeking information about cancer recurrence and long-term effects of cancer and treatment would be more prevalent among breast and colon cancer survivors compared with prostate cancer survivors. Based on a study of breast cancer survivors (14, 22), we anticipated that breast cancer survivors' information seeking about risks of family members getting breast cancer would increase over time, whereas information seeking about recurrence would diminish over time. Knowledge about systematic differences in survivors' information seeking could aid clinicians in designing tailored communications including survivorship care plans for their patients that best meet these information needs.

We collected data from a longitudinal cohort of cancer survivors who were diagnosed with breast, prostate, or colon cancers between January and December 2005. These survivors were identified with permission from the Pennsylvania Cancer Registry (PCR). The sampling frame comprised 26,608 patients with these diagnoses from the PCR in 2005. Of these, 15% or 3,994 patients were randomly selected and invited to participate in the study, after stratification across the three cancer diagnoses. We oversampled patients who were diagnosed with stage IV disease and African-American patients to ensure sufficient statistical power for planned subgroup analyses. Further details about the study population and data collection procedure are described elsewhere (8).

Data were collected from participants using three annual mailed surveys following Dillman's procedure (23) in the fall of 2006, 2007, and 2008. The survey questionnaire was designed following a literature review, in-depth interviews with cancer patients who were not part of the longitudinal study (24), and expert consultation with oncology specialists.

A total of 2,013 participants completed the round 1 survey in the fall of 2006, roughly 9 to 21 months after initial cancer diagnosis. The American Association for Public Opinion Research (AAPOR) response rate 4 (RR4) was 64% for the round 1 survey. Of 1,758 respondents from round 1 who agreed to be recontacted, 1,293 (74%) completed the round 2 survey in the fall of 2007 (roughly 21 to 33 months after diagnosis) and 1,128 (64%) completed the round 3 survey in the fall of 2008 (roughly 33 to 45 months after diagnosis). Nonresponse in round 3 was attributed to refusal to be recontacted after round 1 (n = 255) or round 2 (n = 85), death during the study period (n = 66), or no response after repeated mailing of the survey (n = 479). Respondents were more likely to be white, married, diagnosed with breast cancer, and to have higher education compared with nonrespondents. The analyses described below adjust for these characteristics to account for potential nonresponse bias. All study participants provided informed consent, and our university's Institutional Review Board approved this study.

Drawing from earlier research, we asked respondents to indicate if they actively looked for information about each of the following topics: (i) “how to reduce the chance of my cancer coming back,” (ii) “how to reduce the chance of my getting a different cancer,” (iii) “how to reduce the risk of my children or family members getting my cancer,” (iv) “whether other members of my family are at risk for developing a different type of cancer,” and (v) “whether I am at risk of having other health problems as a result of my cancer or treatment” (11, 12, 15). In round 1, respondents were asked if they had actively looked for the above information after their initial diagnosis and treatment. In rounds 2 and 3, they were asked if they had looked for the above information in the past 12 months. The responses of these items were coded as “yes” or “no,” and each item was treated as a binary variable.

Rounds of survey and cancer diagnoses were the key predictor variables. Round of survey was coded into three categories (round 1, round 2, and round 3). Due to the conflating of cancer diagnosis and gender for respondents diagnosed with breast or prostate cancers, we categorized respondents into four cancer diagnoses (separated by gender): (i) male colon cancer, (ii) female colon cancer, (iii) breast cancer, and (iv) prostate cancer.

We adjusted for a number of covariates including age at diagnosis, education, race/ethnicity, marital status, stage at diagnosis (obtained from the PCR), treatment received at round 1 (surgery, radiation, and systemic therapy), health status, and frequency of doctor visits.

We analyzed the unadjusted proportion of survivors who sought about each topic at each survey round. Using the McNemar test, we compared the proportion of respondents who sought each topic against the proportion who sought other topics at each round (total of 10 comparisons across 5 topics for each round), utilizing a Bonferroni-corrected significance level of P < 0.005 for these comparisons. To demonstrate changes in information seeking about each topic over the three rounds, we utilized mixed-effects logistic regression (“melogit” command) in Stata version 13 that accounts for clustering of repeated observations within individuals (25). We tested whether information seeking varied over time by cancer diagnoses by including an interaction term between round of survey and cancer diagnosis. These regression models adjusted for the covariates listed above. There were 313 cases with missing values in one or more covariates who were excluded in these analyses based on listwise deletion. We conducted a sensitivity analysis that utilized multiply imputed data to assess potential bias due to missing data and compared the ORs and significance levels of the parameters with results presented here. The substantive results from the sensitivity analysis were virtually identical with the results presented here. Details of the sensitivity analysis are available from the corresponding author upon request. To assist in the interpretation of the regression results, we obtained the predicted proportion (“margins” command) of survivors who sought about each topic by cancer diagnosis across survey round, holding all other covariates at their respective means.

Table 1 shows the sample characteristics at the round 1 survey. Mean age of participants was 64.7 years, about half of the respondents were female, and the majority were white.

Table 2 shows the unadjusted proportions of respondents who reported looking for each of the survivorship topics in each survey round. Overall, cancer survivors sought information about how to reduce the chances of cancer recurrence significantly more frequently than all other topics; all pairwise comparisons between the proportion who sought about cancer recurrence risks and the other four topics were significant (P < 0.0005) at all three rounds. We further analyzed the overall percentage of respondents who did not look for any topic at each round: 62.1% of respondents in round 1, 68.1% in round 2, and 64.4% in round 3 did not search about any of the five survivorship topics.

Table 3 displays the adjusted ORs of information seeking about each of the five survivorship topics. The parameters of interest are the interactions between round of survey and cancer diagnoses. With the exception of one topic (how to reduce the risk of one's family getting the same cancer), we detected significant interactions between round of survey and cancer diagnoses. In other words, the patterns of information seeking about survivorship topics across the 3-year follow-up period differed significantly between survivors who were diagnosed with colon, breast, and prostate cancers.

To better visualize the differences in patterns of information seeking, we present the predicted proportions of information seeking for each topic by survey round and cancer diagnoses in Fig. 1. At round 1, breast cancer survivors had higher odds compared with other survivors to report having sought about three of the survivorship topics (how to reduce the chance of cancer coming back, how to reduce the chance of getting another cancer, and risk of late effects from cancer or treatment). In addition, breast cancer survivors had higher odds compared with prostate cancer survivors of information seeking about reducing the risk of family members getting the same cancer and whether other family members are at risk for getting a different type of cancer. [One reviewer commented that a few of the topics might not be directly relevant for a subset of the analyzed sample: those actively treated during the study period. We therefore conducted a sensitivity analysis, excluding those who had self-reported any treatment in rounds 2 and 3 (with the exception of hormonal therapy among some breast cancer survivors, which is indicated for up to 10 years for secondary prevention). The sensitivity analysis excluded 554 cases. We found that the substantive results were very similar to the original analyses with a few minor differences, although the error bars are wider (see Supplementary Table S1 and Supplementary Fig. S1 excluding these 554 cases for comparison). We opted to retain the original analyses; we think it would introduce bias to exclude those who reported receiving care because they may have higher risk, be more concerned, and be more likely to search for information.]

At round 2, which was one year after the round 1 survey, information seeking about these five topics declined for most cancer survivors. However, the sharpest declines were observed among breast cancer survivors; these declines from round 1 to round 2 were significantly sharper for four of the five topics (with the exception of information seeking about reducing risks of family members getting the same cancer) when compared with prostate cancer survivors, who reported less dramatic changes in their information seeking. The decline in information seeking about the risk of cancer late effects from round 1 to round 2 among breast cancer survivors was also significantly sharper than female colon survivors. Between round 2 and round 3, female colon cancer survivors reported more information seeking about how to reduce the risk of their family members getting colon cancer and whether their family members were at risk of other cancers. In comparison, other cancer survivors did not report significantly different levels of information seeking for these topics. By round 3, the proportion of information seeking about these topics did not differ substantially by cancer diagnosis.

In this study, we found that cancer survivors' information seeking varied over time depending on the survivorship topic, and the trends in information seeking differed by cancer type. A sizable proportion of survivors sought information about these topics at round 1. Over time, the proportion of survivors who sought information declined overall. Breast cancer survivors were more likely to seek information about these topics than prostate and colon survivors at round 1, and reported a significant decline in information seeking at rounds 2 and 3. Interestingly, at those later years, female colon cancer survivors tended to seek more information about how to reduce the risk of family members getting colon cancer or a different cancer. We further found that significant covariates of information seeking included younger age at diagnosis, higher education, and earlier cancer stage at diagnosis.

We highlight several of these findings and discuss potential explanations and implications for cancer survivorship care planning and delivery. First, it is notable that information seeking about recurrence occurs more frequently than information seeking about all other survivorship topics, corroborating smaller surveys that found cancer recurrence is the main concern among survivors (4, 9–11, 13–18). To our knowledge, this is the first large-scale population-based study showing cancer survivors actively seek most about recurrence information. This suggests a need for providing information to patients at the end of treatment (e.g., in survivorship care plans) about their residual risk of recurrence and how to reduce it (26). Given that active seeking about how to manage risks of recurrence was most prevalent in round 1, this study further suggests that the optimal “teachable moment” for health promotion among cancer survivors may be early on in the survivorship period (27). Clinicians could consider addressing issues including smoking cessation, physical activity, or other risk behaviors with their patients during the early survivorship period when patients are likely to be more receptive to information about managing risks of recurrence (28, 29). Moreover, across all diseases, patients at round 1 were about 2.5 times more likely to seek about recurrence than about late medical effects, putting into perspective the magnitude of this concern. Next, the frequency of information seeking about family members getting same cancer is about twice that of information seeking about family members getting a different cancer. This suggests that survivors think that all of these cancers tend to run in families.

Whereas results showed declining information seeking among other cancer survivors, female colon cancer survivors seem to seek more about certain survivorship topics over time. The reason for this divergent trend by topic across cancer diagnoses is unclear. We posit that this could be due to less attention and available media information focused on this population, or there could be qualitative differences in patient–physician communication among female colon cancer survivors compared with other cancer survivors. Alternatively, increased attention to survivors' information needs in oncology practice over the past decade could have prompted clinicians to be more proactive in addressing certain survivors' information needs (thereby reducing the need for active seeking) or, conversely, encourage more active seeking by enhancing the salience of survivorship issues for others.

Similarly, there are several plausible explanations for the initially high levels of seeking reported by breast cancer survivors. For example, higher round 1 seeking could be a function of (i) greater provider guidance, (ii) greater uncertainty about the long-term side effects of different treatment options, or (iii) greater motivation to learn about survivorship issues, inspired by a strong lay breast cancer community. Or, perhaps, results could reflect a halo effect whereby breast cancer survivors feel compelled to report greater seeking about survivorship, given the breast cancer community's culture of engagement (30). In addition, we can speculate why breast cancer survivors' initially high levels of information seeking dropped off in subsequent rounds. It is possible that, given a tremendous amount of readily available information (31), they simply had their information needs met. Alternatively, it could be that breast cancer survivors grew frustrated with or overwhelmed by the information they encountered (32, 33). Further research will be necessary to evaluate these hypotheses.

Among the significant covariates, the finding that cancer survivors diagnosed with lower stage disease were more likely to actively seek information would seem counter-intuitive; those with more invasive stage at diagnosis might have been expected to seek more information. It is plausible that patients diagnosed with higher stages of cancer undergo more intensive treatment and have more frequent interactions with their treating clinicians. They might already be receiving more information from their treating clinicians that addresses their concerns, thus obviating the need to conduct additional information seeking. Conversely, patients at higher stages may be relying on their family members to conduct information seeking, or are less likely to be interested in survivorship-related topics if they are still undergoing treatment over the course of the study.[We conducted a sensitivity analysis—excluding patients who have stage IV disease—to address (i) the concern that these patients do not consider these to be survivorship topics and (ii) the possibility that these patients were in active treatment during the study period. We found that the substantive results were largely identical to the original analyses (see Supplementary Table S2 excluding those with stage IV disease)] The finding that younger survivors were more likely to seek information than older survivors seems consistent with a generational shift toward online information seeking in general and greater engagement with health decisions among younger patients (6, 34, 35).

The present study contributes new insights to the body of literature describing survivorship information needs. Researchers have previously noted that there are gaps in the literature about survivorship information needs especially over the long term (6). Most studies have focused on the diagnosis and treatment phase of the care continuum. This study contributes to understanding not only the information needs at a single time point but also the overtime trajectory of these needs in a cohort sample of cancer survivors; it also considers factors that are associated with this trajectory. We monitored information needs about topics that relate to key components of survivorship care and examined whether survivors sought about these topics prospectively, whereas prior studies relied on cross-sectional and retrospective designs. The present study is also unique in involving a large population-based sample from across different health care providers within the state of Pennsylvania; prior studies were smaller and often involved convenience samples within single institutions.

The 2006 IOM report recommends that, where appropriate, first-degree relatives of cancer patients be informed about their increased risk and the need for cancer screening (e.g., breast, colorectal, and prostate cancers); this information should be included as a component of survivorship care plans. However, Salz and colleagues reviewed existing breast and colorectal cancer survivorship care plans of NCI-designated cancer centers and found that none had included information about family risk (36). In the present study, we found that some cancer survivors expressed the need for information about reducing family members' risk of getting the same cancer. The practical implication of this finding is that it would be important for survivorship care plans to include this information routinely to advise patients on how to manage cancer risks of their first-degree family members.

There are, however, some limitations to this study. First, we did not assess the specific content survivors sought. As an example, it would be highly informative to learn whether ER-positive breast cancer survivors were seeking information about hormonal therapy (tamoxifen) for reducing breast cancer recurrence in order to tailor information most salient to survivors. Second, we do not know where survivors sought for these topics. Previous analyses revealed that participants in this study actively sought treatment and quality-of-life information from their treating doctors, social network (e.g., family or friends), and print materials (e.g., books; ref. 8). Knowledge of the main sources of survivorship information would inform clinicians about delivery platforms or channels that would be most acceptable to survivors. Third, we followed cancer survivors for only 3 years after their diagnosis and could not assess longer-term concerns (e.g., late effects beyond 5 years). We recommend future studies assess information needs among long-term cancer survivors. Fourth, the results may not be generalizable to survivors with other forms of cancer or those from other geographic areas. Future research could consider including survivors with a broader range of cancer diagnoses and from a larger number of state-based cancer registries. We also note that the data are about 8 years old (or 6 for round 3). Trends in survivorship concerns and information seeking could have changed during the intervening years. Additional research would be needed to include other survivorship topics that have surfaced in recent years. For instance, Galarce and colleagues examined health information seeking among posttreatment cancer patients for a range of topics, including disease/treatment-related information (e.g., side effects of treatment), self-care management (e.g., diet and nutrition), health services (e.g., doctors or facilities), and work/finance (e.g., financial assistance and employment issues; ref. 37).

In sum, this study presented data about a population-based sample of breast, prostate, and colon cancer survivors seeking information about five survivorship topics over a period of 3 years after diagnosis. Among this list of survivorship topics, we recommend that clinicians consider providing information about managing the risk of cancer recurrence to patients early on in the survivorship period, when patients may be most receptive and find this information most salient. The study's findings could also inform efforts to tailor the content of survivorship care plans to include specific advice for survivors on how they could manage their risks of cancer recurrence (e.g., information about symptoms or signs of recurrence, recommendations for surveillance), late effects (e.g., describing potential late and long-term effects of cancer treatment, including the symptoms of such effects), and family members' cancer risks (e.g., recommendations for first-degree relatives to undergo cancer screening tests).

No potential conflicts of interest were disclosed.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Conception and design: A.S.L. Tan, R.H. Nagler, R.C. Hornik, A. DeMichele

Development of methodology: R.C. Hornik, A. DeMichele

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R.C. Hornik

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.S.L. Tan, R.C. Hornik, A. DeMichele

Writing, review, and/or revision of the manuscript: A.S.L. Tan, R.H. Nagler, R.C. Hornik, A. DeMichele

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R.H. Nagler, R.C. Hornik

Study supervision: R.C. Hornik

R.H. Nagler acknowledges support from the Building Interdisciplinary Research Careers in Women's Health Grant (2 K12-HD055887) from the Eunice Kennedy Shriver National Institutes of Child Health and Human Development, the Office of Research on Women's Health, and the National Institute on Aging, administered by the University of Minnesota Deborah E. Powell Center for Women's Health.

This study is supported by P20CA095856 and P50CA095856 from the NCI (PI: R.C. Hornik). A.S.L. Tan conducted this work while being a CECCR postdoctoral research fellow at the Annenberg School for Communication (supported by P20CA095856).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.