Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Oct 1.
Published in final edited form as: Health Commun. 2022 Feb 16;38(9):1878–1886. doi: 10.1080/10410236.2022.2038866

Cancer Information Overload Across Time: Evidence from Two Longitudinal Studies

Helen Lillie 1, Rachael A Katz 1, Nick Carcioppolo 2,3, Elizabeth A Giorgi 1, Jakob D Jensen 1
PMCID: PMC9378766  NIHMSID: NIHMS1780339  PMID: 35172651

Abstract

A majority of U.S. adults report feeling overwhelmed by the amount of available cancer information, termed cancer information overload (CIO). Research has demonstrated CIO is prevalent and negatively related to health behaviors, but no study to date has examined this disposition across time. Two longitudinal studies – a colonoscopy intervention among older U.S. adults (N = 237) and an HPV vaccination intervention among young U.S. women (N = 411) – were utilized to examine CIO stability across time and its relationship to prevention intentions and indifference. CIO increased indifference for non-adherent individuals but had no effect on intentions. CIO was stable in study 1 but not study 2, suggesting CIO stabilizes across the life course. Results also support a five-item measure of CIO.

Keywords: cancer information overload, indifference, stability, CFA, prevention


Cancer information overload (CIO) is the perception that the amount of cancer information in one’s environment is overwhelming (Jensen, Carcioppolo, King, et al., 2014). A secondary data analysis of the health information national trends survey (HINTS) found that 75% of Americans reported experiencing CIO (Niederdeppe & Gurmankin Levy, 2007). Given its prevalence, and the possibility that it is negatively related to preventive health behavior (Jensen, King, Carcioppolo, et al., 2014; Occa et al., 2020; Niederdeppe & Gurmankin Levy, 2007), there is a pressing need for continued research on CIO (Khaleel et al., 2019).

In a recent review of the literature on health information overload, CIO was the most commonly represented topic; however, research is still somewhat sparse and critical questions remain (Khaleel et al., 2019). A key gap in the CIO literature is the question of CIO’s stability across time (Jensen et al., 2020). CIO is conceptualized as a stable disposition, but this assumption has not been tested. A test of CIO’s stability would require longitudinal data. Obamiro and Lee (2019) have advocated for increased longitudinal research on health information overload. Yet, in a recent review of research examining consumer need for cancer information, only three of the 117 studies identified were longitudinal (Jo et al., 2019).

To address this gap in the CIO literature, the current research provides a preliminary test of the stability of CIO and its effects across two longitudinal studies: (a) colonoscopy screening perceptions and intentions in a sample of older U.S. adults (ages 50 – 74), and (b) HPV vaccination perceptions and intentions in a sample of young adult U.S. females (ages 18 – 26). The stability of CIO and its impact on behavioral intentions and indifference are assessed in both studies. Additionally, the five-item and eight-item CIO measures are compared.

Cancer Information Overload as a Stable Disposition

CIO is conceptualized as an aversive motivation disposition wherein a person feels “overwhelmed by the amount of cancer-related material in the information environment” (Jensen, Carcioppolo, King, et al., 2014, p. 2). Recent research has focused on explicating CIO and its impact on the performance of cancer prevention and screening behavior, demonstrating that beyond being an unintended effect of intervention messaging (for more, see Cho & Salmon, 2007), CIO may be negatively related to health outcomes. Specifically, CIO has been associated with sub-optimal health decision-making, reduced perceived health, and confusion about and rejection of health recommendations (Khaleel et al., 2019; Kim et al., 2007).

A fundamental assumption of CIO research and theorizing is that CIO is a disposition which should be relatively stable across time (DeVellis, 2017; Jensen, Carcioppolo, King, et al., 2014; Kim et al., 2007). Yet, existing research has not tested this assumption (Jensen et al., 2020). Niederdeppe et al. (2014) found that cancer messaging can impact CIO, dependent on the amount of uncertainty depicted. That CIO can be impacted by a single message runs counter to the assumption that it is a stable disposition. Further, Jensen, Carcioppolo, King, et al. (2014) argue that “CIO is not a trait, but rather is cultivated by exposure to information about cancer from media, in conversations, and from healthcare providers” (p. 91). Understanding if CIO is stable is vital for health communication theorizing and cancer messaging. If CIO is stable, messaging could benefit from targeting the level of CIO. For example, Occa et al. (2020) found that CIO moderated the effect of a clinical trial information aid on attitudes towards clinical trials. Alternatively, if CIO is not stable, messaging could aim to reduce CIO. Therefore, we ask:

RQ1: Is CIO stable across time?

Impact of CIO on Indifference and Intentions

Indifference occurs when individuals have some understanding of a topic area but do not consider that topic to be important and therefore do put cognitive effort towards it (Kaplan, 1972; Song & Ewoldsen, 2015). Information management and processing theories explicate how individuals may reappraise a topic area (such as cancer) as unimportant to reduce negative feelings, generating indifference. According to the theory of motivated information management (TMIM; Afifi & Weiner, 2004), when efficacy around information seeking is low and anxiety is high, individuals may engage in cognitive reappraisal, including reducing the topic’s perceived importance. Reappraisal “serve(s) the purpose of neutralizing the source of negative emotion when reported efficacy to seek information is low” (Fowler et al., 2018; p 373). Similarly, the extended parallel process model (EPPM; Witte, 1994) claims that individuals will minimize the importance of fear or anxiety-inducing content if they do not have an effective method for addressing it. A key aspect of CIO is that individuals feel they do not have the means to sort through the overabundance of cancer information or to determine the accuracy or trustworthiness of information, leading to uncertainty and anxiety (Whitelaw, 2008). Developing a feeling of indifference towards cancer could protect against the anxiety and uncertainty caused by CIO.

The idea that information overload leads to indifference has been assumed by information overload scholars (e.g., Aikat & Remund, 2012; Magnini & Dallinger, 2018). Yet, to the authors’ knowledge, this supposition has not been tested. It follows both logically and from communication theory that information overload would lead to indifference as a defensive reaction to protect against the stress and uncertainty related to overload (Chae, 2016; Kim et al., 2007; Swar et al., 2017). Therefore, we propose that:

H1: CIO will be positively associated with indifference.

In their work connecting information overload with the theory of planned behavior, Liu and Kuo (2016) found that information overload impacted both attitudes and intentions. Therefore, in addition to impacting indifference, a type of attitude, CIO should influence behavioral intention. CIO has been associated with poor health behaviors in past cross-sectional research. For example, CIO has been related to smoking, tanning, sunburns, lack of exercise, and rare health checks, all of which are associated with heightened cancer risk (Breyton et al. 2021; Jensen et al., 2020; Niederdeppe & Gurmankin Levy, 2007).

Yet, longitudinal research connecting CIO to important outcomes is sparse (Jo et al., 2019; Obamiro & Lee, 2019). Jensen, Carcioppolo, King, et al. (2014) found that CIO was negatively related colonoscopy insurance claims across 18 months, and Chae (2016) found that CIO increased cancer information avoidance three months after measurement of CIO. This suggests that CIO can impact distant outcomes. The current research seeks to build this literature by assessing if CIO is related to prevention intentions over time. Given CIO’s negative relationship with other health behaviors, it is expected that CIO would be negatively related to cancer prevention and screening behaviors. Therefore, we propose that:

H2: CIO will be negatively related to prevention intentions.

Cancer Information Overload Measurement

A key component of explicating a construct is the identification of a valid and reliable measure (DeVellis, 2017). Currently, there are multiple operational definitions of health information overload generally (Khaleel et al., 2019), and CIO more specifically, indicating a distinct need for refinement. Originally, CIO was measured using a single item from the National Cancer Institute’s HINTS, “There are so many recommendations about preventing cancer, it’s hard to know which ones to follow” (Niederdeppe & Gurmankin Levy, 2007). Jensen, Carcioppolo, King, et al. (2014) proposed an eight-item measure to better assess CIO. In response, Costa et al. (2015) recommended removing three items and revising another three to transform the eight-item measure into a more lucid five-item version. They pointed out that the original eight-item scale only achieved high fit in specific contexts and models, which suggests that perhaps the measure could be further optimized (Costa et al., 2015). Subsequently, Jensen et al. (2020) utilized a five-item version of the scale but did not compare the effectiveness of the five-item measure to the eight-item measure.

Versions of the five-item measure modified for non-cancer contexts have been utilized successfully but have consistently lower reliabilities than eight-item versions (Obamiro & Lee, 2019; Ramondt & Ramirez, 2019). Further, Breyton and colleagues (2020) examined ten, eight, and five item versions of the scale in French and English. Support was found for all three versions with the five-item version demonstrating excellent factor structure but the lowest reliability (α = .83). Collectively, these findings provide some support for the shorter version of the scale, but the difference in reliabilities suggests that more research is needed to fully understand which measurement approach is optimal (Jensen et al., 2020). A shorter version of a scale is often preferable because it is more efficient, and saves the participant time (Obamiro & Lee, 2019), but a longer measure may offer optimal content validity (DeVellis, 2017). Therefore, the current research compares the five-item and eight-item versions, asking:

RQ2: Which CIO measure, the five-item or the eight-item, is supported by the data?

Study 1

The first study utilizes data from a longitudinal colonoscopy intervention among older U.S. adults. Measures were taken at three time points, time 1 (T1) before the intervention, time 2 (T2) immediately after the intervention, and time 3 (T3) two weeks after the intervention. For the intervention, participants watched one of two PSAs in which Karen Miller shared the story of her father, Doug Miller, and his death from colorectal cancer. The PSAs varied in whether Karen Miller explicitly stated that the outcome of her father’s cancer story could have been different if he had a colonoscopy or if that idea was implicit in the story. The analysis reported here focuses on T1 CIO’s relationship to T3 CIO, T3 colonoscopy intentions, and T3 indifference.

The subsequent analyses used a fixed-effects approach to account for clustering based on intervention condition (McNeish & Stapleton, 2016). Specifically, a dichotomous condition variable (0=implicit, 1=explicit) was entered into the analysis. A fixed-effects approach removes all group-level variance from analyses, meaning that it controls for all intervention condition effects. This includes effects from constructs not included in the analyses (McNeish & Stapleton, 2016) – for example if groups varied on cancer fear. As McNeish and Kelley state (2019, p 23), “researchers need not be concerned with including Level 2 predictors in the model, because variance attributable to all Level 2 variables (whether available in the data or not) is consumed by the cluster affiliation variables.” Because the current research is not interested in the effects of the intervention but rather the stability and effects of CIO across all participants, using a fixed-effects approach to account for intervention group differences seems most appropriate.

Participants

Participants were recruited and compensated via Qualtrics Panels to complete a longitudinal study, with 421 completing T1/T2 surveys and 237 completing all three waves (T1/T2/T3). Only those with complete data for all three waves are utilized in this analysis. The sample was nearly equal on gender distribution (male: 50.1%), predominantly White (96.9%) and not Hispanic/Latino (98.6%), about 62 years old on average (50 – 75 years; M = 61.99, SD = 6.9), and resided in the state of Utah. Education was distributed as follows: High school education or less (28.7%), Associates/Trade School/Technical Training (15.7%), 4-year degree (31.4%), and advanced professional degree (24.2%). The institutional review board (IRB) at the first author’s university approved and monitored the protocol.

Measures

Demographics

Participants reported age, race, education, sex, household income, and whether they were up to date on their colonoscopy screening.

CIO

For the substantive analysis, CIO is measured using a five-item scale (Jensen et al., 2020). (See the confirmatory factor analysis section under “Results” for comparison with the eight-item version). All CIO items are measured on a five-point scale ranging from strongly disagree (1) to strongly agree (5). The scale was reliable at T1 (M = 2.70, SD = .76, α = .85) and T3 (M = 2.64, SD = .86, α = .89).

Indifference

An eight-item measure of indifference was created by the research team. All indifference items were measured on a five-point scale ranging from strongly disagree (1) to strongly agree (5). Items asked participants how they felt about getting a colonoscopy, including “I don’t really care” (indiff1), “I’m indifferent” (indiff2), “I’m not that interested” (indiff3), “I’m apathetic” (indiff4), “I’m not motivated to pursue it” (indiff5), “It doesn’t really matter to me” (indiff6), “I have no feelings about it” (indiff7), and “I’m unconcerned” (indiff8). The new scale was examined using principal axis analysis with direct oblimin rotation. One factor emerged with an Eigenvalue greater than 1 (eigenvalue = 5.97; explains 74.57% of the variance). All items loaded at .74 or higher on that single factor. Accordingly, all eight items were retained and combined into a single scale. The scale was reliable at T1 (M = 1.93, SD = .96, α = .95) and T3 (M = 1.92, SD = .95, α = .95).

Intention to Screen in the Next Six Months

Intention to receive a colonoscopy was measured using a single-item from Tiro et al., 2005, “I intend to have a colonoscopy in the next 6 months,” on a seven-point scale, ranging from strongly disagree (1) to strongly agree (7). Intention was measured at T1 (M = 2.18, SD = 1.88), T2 (M = 2.37, SD = 2.01), and T3 (M = 2.46, SD = 2.15).

Results

Preliminary Analyses

Test of Multivariate Normality.

Micceri (1989) demonstrated that most datasets are non-normal. Given that, researchers should test whether data is multivariate normal and, if it is not, follow recommended courses of action for analyzing multivariate non-normal data. Lisrel 9.30 was utilized to test for multivariate normality among CIO items. Consistent with Micceri’s (1989) claims, the items exhibited significant multivariate abnormality, skewness = 914.20, Z-score = 77.59, p < .001, and kurtosis = 2838.52, Z-score = 23.65, p < .001.

For the current analysis, only confirmatory factor analysis (CFA) is vulnerable to violations of multivariate normality (Satorra & Bentler, 2010). When that is the case, the asymptotic covariance matrix should be utilized in CFA to calculate a Satorra-Bentler (S-B) χ2 (Satorra & Bentler, 2010). Lisrel 9.30 was utilized to conduct a CFA and five indicators were utilized to assess model fit: S-B χ2, CFI, RMSEA, SRMR, and Model AIC (Akaike, 1987; Hu & Bentler, 1999; Holbert & Stephenson, 2008).

Confirmatory Factor Analysis: Comparing 8- and 5-item CIO Measures (RQ2).

Jensen and colleagues originally developed an eight-item measure of CIO (henceforth, CIO8; Jensen, Carcioppolo, King, et al., 2014), but Costa et al., (2015) argued that several of the items should be revised and a five-item model examined (henceforth, CIO5). Accordingly, the current analysis examined and compared both a CIO8 and CIO5. From a comparison standpoint, Model AIC allows for the direct comparison of models; lower scores indicate superior fit (Akaike, 1987). We also examined CFI (.95 or higher indicates good fit, Hu & Bentler, 1999), RMSEA (.08 or lower indicates good fit, Holbert & Stephenson, 2008), and SRMR (.08 or lower indicates good fit, Hu & Bentler, 1999). Items from T1 CIO were utilized for this analysis.

CIO8 was not a good fit for the data, S-B χ2 (20, N = 237) = 76.71, p < .001, CFI = .94, RMSEA = .12 (90% CI: .10, .13), SRMR = .07, Model AIC = 873.78 (see Figure 1); however, CIO5 was a good fit for the data: S-B χ2 (5, N = 237) = 9.73, p = .08, CFI = .99, RMSEA = .08 (90% CI: .03, .14), SRMR = .03, Model AIC = 588.19 (see Figure 2). Compared to CIO8, CIO5 appears to be an optimal measurement approach. Accordingly, CIO5 is utilized as the primary measurement approach for CIO for the remainder of the article.

Figure 1.

Figure 1

Confirmatory Factor Analysis for CIO8

Figure 2.

Figure 2

Confirmatory Factor Analysis for CIO5

Substantive Analyses

CIO Stability (RQ1).

RQ1 asked if CIO is stable. Researchers have assessed construct stability using a variety of statistical tests, including correlations (with a significant correlation representing stability), t-tests (with a non-significant t-test representing stability), and repeated measures ANOVA (with a non-significant result representing stability) (see Vaidya et al. 2002). Correlation is less stringent than the other tests and may be more appropriate for assessing test-retest (focused on measurement) rather than for testing construct stability. As Cronbach (1947) explains, test-retest can show that a “measuring technique may be extremely accurate in reporting a biological instant in the life of an individual but not measure a stable characteristic of the individual” (p. 2). We are interested in if CIO is a stable disposition. Therefore, the current study tests CIO stability using a repeated measures analysis of covariance (ANCOVA), controlling for intervention condition. CIO will be considered stable if there is not a significant difference between T1 CIO and T3 CIO. Analysis found no difference between T1 CIO and T3 CIO, Pillai’s Trace = .01, F(1, 235) = 2.52, p = .11, indicating that CIO is stable.

CIO Effects on Prevention Indifference and Intention (H1 and H2).

H1 proposed that CIO would have a negative impact on prevention intentions, and H2 proposed that CIO would have a positive impact on indifference. To aid interpretation, participants were split into two groups for this analysis: (1) those who have not screened or are 5+ years removed from their last screening (n = 85; henceforth, non-adherent) and (2) those who have screened in the last 5 years (n = 145; henceforth, adherent). This split is necessary as interpretation of some measures – for example, intentions to screen – hinge on whether participants are behind on their screening.

The rmcorr package in R was employed to test the impact of T1 CIO on T3 intentions and indifference. Rmcorr is a tool to assess repeated measures correlation without violating the assumption of independence of observations, as it uses ANCOVA to adjust for inter-individual variability (Bakdash & Marusich, 2017). Separate analyses were conducted for the two outcomes. Regarding H1, CIO had a positive impact on indifference for non-adherent participants, rrm (87) = 0.25, 95% CI [0.039, 0.435], p = 0.02, but no impact for adherent participants, rrm (147) = −0.10, 95% CI [−0.257, 0.064], p = 0.229. Regarding H2, CIO had no impact on intentions for either adherent, rrm(147) = −0.05, 95% CI [−0.21, 0.113], p = 0.544, or non-adherent participants, rrm (87) = −0.10, 95% CI [−0.309, 0.108], p = 0.329.

Study 2

Study 1 supported the stability of CIO. It also confirmed that a revised five-item measure of CIO is superior to the eight-item version. Finally, T1 CIO scores were found to be positively related to indifference for non-adherent individuals after two weeks. Study 2 investigates similar questions, but this time with a national sample of young adult U.S. females (18 – 26) as part of a longitudinal HPV vaccination intervention.

The current data are extracted from a larger study that assesses the impact of narrative stimuli on intentions to vaccinate (Authors, under review). In-line with study 1, study 2 measures were taken at three time points, time 1 (T1) before the intervention, time 2 (T2) immediately after the intervention, and time 3 (T3) two weeks after the intervention. For the intervention, all participants read a cervical cancer narrative. Experimental conditions varied in whether the narrative was about cervical cancer survivorship or death. The analysis reported here focuses on T1 CIO’s relationship to T3 CIO, T3 HPV vaccine intentions, and T3 indifference. In-line with study 1, analyses use a fixed-effects approach to account for the intervention, including a dichotomous condition variable in the analyses (0 = survivor, 1 = death).

Participants

Qualtrics Panels enrolled 411 U.S. females (age range: 18–26, Mage = 22.60, SD = 2.55) from their national panel into an online survey experiment. Participants reported a range of education levels including high school education or less (33.7%), some college (26.6%), associates/trade school/technical training (22.1%), 4-year degree (16.1%), and master’s degree (1.5%). A majority of the participants (69.8%) identified themselves as Caucasian or White. Almost half of the participants were single (46%) and most had no children (68.9%) at the time of the study. Participants were compensated by Qualtrics Panels. An institutional review board (IRB) approved and monitored the protocol.

Measures

Demographics

Participants reported age, race, and education.

CIO

Based on support from study 1, the 5-item measure of CIO was used. It was reliable at T1 (M = 2.82, SD = .82, α = .78) and T3 (M = 3.03, SD = .83, α = .86).

Indifference

Indifference was measured identically to study 1, except that participants were asked how they felt about getting the HPV vaccine. The measure was reliable at T1 (M = 2.98, SD = .96, α = .92) and T3 (M = 3.02, SD = .95, α = .93). (For the confirmatory factor analysis, see the preliminary analyses section under “Results”).

Intention to Vaccinate in the Next Six Months

Intention to vaccinate was assessed with a single-item from Shepherd et al., (2000), “I intend to begin the HPV vaccination in the next 6 months”, measured on a 7-point scale ranging from strongly disagree (1) to strongly agree (7). Intention was measured at T1 (M = 2.67, SD = 1.67), T2 (M = 2.52, SD = 1.27), and T3 (M = 2.74, SD = 1.73).

Results

Preliminary Analyses

Analyses mirror study 1. Preliminary analyses include a test of multivariate normality, and a CFA of indifference. Consistent with study 1, the items exhibited significant multivariate abnormality, skewness = 112.42, Z-score = 47.18, p < .001, and kurtosis = 867.64, Z-score = 25.50, p < .001. Due to abnormality, the S-B χ2 is utilized in CFA (Satorra & Bentler, 2010).

The eight-item indifference measure exhibited excellent fit: S-B χ2 (18, N = 411) = 32.57, p = .02, CFI = .99, RMSEA = .07 (90% CI: .05, .10), SRMR = .03, Model AIC = 2443.71 (see Figure 3). Bentler (2010) noted that correlated error terms were almost inevitable in measurement models and advocated identification and engagement as a strategy for moving forward. Two sets of error terms were correlated: indiff2 (I’m indifferent) and indiff4 (I’m apathetic) and indiff3 (I’m not that interested) and indiff5 (I’m not motivated to pursue it). The first set may be correlated as they tap into a related concept (apathy) and the second set utilize similar sentence structure (i.e., “I’m not . . .”). In light of excellent fit, the eight-item indifference measure was utilized for analysis.

Figure 3.

Figure 3

CFA of Indifference – Study 2

Substantive Analyses

Identical to study 1, the current data is culled from a larger study that included an embedded message experiment between T1 and T2 data collection points. Experimental condition is not of interest to this analysis, so it is included as a covariate in all subsequent analyses. Unlike study 1, it is not necessary to split analysis for non-adherent/adherent individuals as all participants in this study were non-adherent by design. Partial correlations were calculated between all study variables (see Table 2).

Table 2.

Bivariate Correlations: Study 2

1. 2. 3. 4. 5. 6. 7. 8. 9.
1. CIO – T1 ----
2. CIO – T3 .52* ----
3. Indifference – T1 .15* .08 ----
4. Indifference – T3 .32* .35* .45* ----
5. Intention – T1 .09 .08 −.27* −.18*
6. Intention – T3 −.02 .05 −.24* −.15* ----
7. Age −.07 .03 −.05 −.08 −.02 −.09 ----
8. Race −.12 −.02 −.03 .02 −.15* .11 .10 ----
9. Education .04 .04 .01 .06 .04 .05 .09 −.05 ----

Note. Partial correlations between study variables.

p < .10

*

p < .05

CIO Stability (RQ1).

In-line with study 1, CIO stability was assessed using a repeated measures ANCOVA, controlling for experimental condition. Analysis did find a difference between T1 CIO and T3 CIO, Pillai’s Trace = .04, F(1, 197) = 8.93, p = .003, indicating that CIO is not stable. Comparison of the means reveals that CIO increased from T1 (M = 2.87, SD = .72) to T3 (M = 3.03, SD = .83).

CIO Effects on Prevention Indifference and Intention (H1 and H2).

Following study 1, the effects of CIO on HPV vaccination intentions and indifference were assessed using the rmcorr package in R. CIO had a positive impact on indifference that approached significance, rrm (198) = 0.14, 95% CI [−0.003, 0.271], p = 0.054, but had no impact on intentions, rrm (198) = 0.06, 95% CI [−0.081, 0.197], p = 0.406.

Discussion

The current research assessed the stability of CIO and its effects on behavioral intentions and indifference. Study 1 addressed these hypotheses in the context of colorectal cancer prevention for adults aged 50–75. CIO was stable, increased indifference for non-adherent individuals, and had no effect on colonoscopy intentions. Study 2 tested study hypotheses in the context of young women’s (aged 18–26) cervical cancer prevention. CIO was not stable, but rather increased, and had no significant effect on indifference or intention (although its association with indifference approached significance, p = .054). Findings suggest that CIO may stabilize with age, similar to other dispositions (e.g., Specht et al., 2013), and that the impact of CIO on intention and indifference may be contingent about other factors.

CIO has been conceptualized as a stable disposition (Jensen, Carcioppolo, King, et al., 2014; Kim et al., 2007). Findings from the current research provide partial support for this claim. Specifically, study 1 found CIO to be stable, but study 2 did not. As with other dispositions, CIO may be less stable at earlier points in the life course (Specht et al., 2013; Wu, 2016). Study 1 consisted of adults aged 50–75 whereas study 2 consisted of women aged 18–26. It makes sense that young adults would still be in the process of conceptualizing their cancer information environment. Greater exposure to cancer information compared to what they likely received as adolescents, alongside diminished reliance on parents as information gatekeepers, could cause an increase in CIO. Aside from age, another key difference between study 1 and study 2 is that study 2 only included female-identifying participants. The stability of CIO could vary depending on gender. However, further analysis identified no difference between the male and female participants in CIO stability for study 1, Pillai’s Trace = .01, F(1, 234) = 3.17, p = .08, making the life course explanation more likely. Future research should sample across age groups and test if CIO is stable for older groups and at what point in the life course CIO stabilizes. Additionally, studies should measure CIO at least at three time points to best assess stability (Lee et al., 2008).

Researchers have suggested that CIO may increase indifference (Aikat & Remund, 2012; Magnini & Dallinger, 2018). The current research provides some support for this idea. CIO led to increased indifference for non-adherent individuals in study 1 and had a nearly significant positive effect on indifference in study 2 (p = .054) in which all participants were non-adherent. That CIO increases indifference for already non-adherent individuals suggests that indifference could be a way that individuals cope with or justify their non-adherence in the face of cancer information. In both studies, all participants had been presented with a story demonstrating the importance of prevention, with most participants receiving a story about cancer death. Yet, rather than increase intention to take a preventative action, non-adherent individuals high in CIO responded with greater indifference. This finding presents a challenge for cancer communication – how can messages increase prevention intentions and decrease indifference for audiences that are high in CIO?

Further, the connection between CIO and indifference suggests that CIO may function like uncertainty, with CIO leading to cognitive reappraisal when efficacy to manage cancer information is low, a proposition supported by the TMIM (Afifi, 2004). This points to an important shortcoming in information overload research, the limited use of communication theory. In their health information overload scoping review, Khaleel et al. (2020) note only two studies utilizing existing theory. Specifically, Liu and Kuo (2016) used the theory of planned behavior, including information overload as a predictor of attitude, intention, and perceived behavioral control, and Ko et al. (2014) used social cognitive theory and information processing theories but measured cognitive load rather than information overload. Khaleel et al. (2020) recommend scholars use cognitive load theory. We echo their recommendation and add that future research should integrate CIO into information management and uncertainty theories, such as the theory of motivated information management and uncertainty management theory.

CIO did not impact prevention intentions in either study. It is possible that CIO is a less important factor in cancer prevention than previously theorized. However, existing literature would suggest that CIO does impact prevention (Occa et al., 2020; Niederdeppe & Gurmankin Levy, 2007). It is more likely that the impact of CIO on prevention intentions is dependent on other factors. CIO could function like uncertainty, where for some it prompts action and others avoidance (Afifi & Weiner, 2004; Fowler et al., 2018). Indeed, information overload has actually been linked to feelings of uncertainty (Whitelaw, 2008). Future research should assess if factors like cancer fear, dispositional worry, and locus of control moderate the relationship between CIO and prevention intention. Another possible explanation is that CIO requires a longer period of time than two weeks to have an effect. CIO did influence indifference for non-adherent individuals. It is possible that indifference would lead to reduced prevention intentions over time. Future longitudinal research should assess the longitudinal impact of CIO on prevention intentions across a longer period of time, testing indifference as a potential mechanism of effect.

The current research also supports the use of a five-item CIO measure. In addition to being better supported by the data, the five-item version is conceptually superior. The three removed items more accurately reflect constructs related to CIO but not necessarily CIO itself, as noted by Costa et al. (2015). Specifically, the removed items include not having the time or ability to follow all cancer prevention recommendations and believing that cancer information is exaggerated. The retained five items focus on feeling overwhelmed by cancer information and being unable to process or evaluate cancer recommendations because there are too many. Therefore, we recommend the use of the five-item measure in subsequent research. Consistency in measurement of CIO across studies will allow for more accurate comparisons of findings.

Limitations and Conclusion

The present research had several limitations. The studies covered two topics, but that is just a small subset of all possible health behaviors. Cancers that have received greater media attention, such as breast cancer, or are associated with blame, such as lung cancer, may produce different effects. Behavioral intention, rather than behavior, served as the outcome, and all measures were self-reports. Though longitudinal, there is only a two-week time difference between T1 and T3 for both studies, a relatively small time difference that could influence the size and nature of observed relationships. Similarly, both studies utilized only two time points. Future research should collect data across at least three time points to establish CIO stability (Lee et al., 2008). Study 1 and study 2 varied in terms of demographics. Study 1 focused on adults aged 50–75 who resided in Utah. Study 2 included women aged 18–26 across the United States. Differences in findings between the studies could be due to age, geography, or type of cancer. Further, neither study included adults aged 27–50 or those older than 75.

Despite these limitations, the present research engaged critical questions in the CIO literature. Over two longitudinal studies, we conclude that, like other dispositions, CIO becomes stable across the life course. CIO was also found to increase indifference to prevention for non-adherent individuals. Yet, CIO did not have an impact on prevention intentions. In light of past research, this suggests that CIO may impact intentions differently dependent on other individual difference factors. We also compared the five- and eight-item CIO scale, finding support for the five-item version. The results provide researchers with a stronger foundation for CIO scholarship and raise several compelling questions for future research.

Table 1.

Bivariate Correlations: Study 1

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
1. CIO – T1 ---- .52* .22* .27* −.12 −.16* −.11 −.09 −.22* −.08 −.08
2. CIO – T3 .72* ---- .30* .27* −.08 −.16 −.15 −.23* −.22* −.09 −.05
3. Indifference –T1 .43* .41* ---- .59* −.05 −.15 −.10 −.15 −.08 −.15 .01
4. Indifference – T3 .33* .45* .73* ---- −.06 −.05 −.13 −.09 −.13 −.10 −.10
5. Intention – T1 −.22* −.16 −.20 −.21* ---- .53* .20* −.05 .08 −.18* −.16*
6. Intention – T3 −.03 −.03 −.31* −.27* .71* ---- .06 .04 .01 −.05 −.06
7. Age −.07 −.04 .00 −.04 −.15 −.12 ---- −.07 −.03 −.14 .04
8. Sex −.06 −.05 .10 .12 −.25* −.12 −.09 ---- −.06 −.19* −.30*
9. Race .00 .03 −.06 −.04 .08 .08 .14 −.04 ---- .03 −.06
10. Income −.11 −.24* −.17 −.28* .13 .10 .04 .01 .05 ---- .47*
11. Education Level −.25* −.16 −.22* −.07 .09 .00 .02 −.15 −.15 .13 ----

Note. Partial correlations between study variables. Values above the dashed lines represent adherent participants; values below the dashed lines represent non-adherent participants.

p < .10;

*

p < .05

References

  1. Afifi WA, & Weiner JL (2004). Toward a theory of motivated information management. Communication Theory, 14(2), 167–190. 10.1111/j.1468-2885.2004.tb00310.x [DOI] [Google Scholar]
  2. Aikat D, & Remund D (2012). Of Time Magazine, 24/7 media, and data deluge: The evolution of information overload theories and concepts. In Strother JB, Ulijn J, & Fazal Z (Eds.), Information overload: An international challenge for professional engineers and technical communicators (pp. 15–40). Wiley & Sons, Inc. [Google Scholar]
  3. Akaike H (1987). Factor Analysis and AIC. Psychometrika, 52, 317–332. 10.1007/978-1-4612-1694-0_29 [DOI] [Google Scholar]
  4. Bakdash JZ, & Marusich LR (2017). Repeated measures correlation. Frontiers in Psychology, 8, 456. 10.3389/fpsyg.2017.00456 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bentler PM (2010). SEM with simplicity and accuracy. Journal of Consumer Psychology, 20, 215–220. 10.1016/j.jcps.2010.03.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Breyton M, Smith AB, Rouquette A, & Mancini J (2020). Cancer information overload: Association between a brief version of the CIO scale and multiple cancer risk management behaviors. Patient Education & Counseling, 104(5), 1246–1252. 10.1016/j.pec.2020.09.016 [DOI] [PubMed] [Google Scholar]
  7. Chae J (2016). Who avoids cancer information? Examining a psychological process leading to cancer information avoidance. Journal of Health Communication, 21(7), 837–844. 10.1080/10810730.2016.1177144 [DOI] [PubMed] [Google Scholar]
  8. Cho H, & Salmon ST (2007). Unintended effects of health communication campaigns. Journal of Communication, 57(2), 293–317. 10.1111/j.1460-2466.2007.00344.x [DOI] [Google Scholar]
  9. Costa DSJ, Smith A, Lim BT, & Fardell JE (2015). Simplifying the assessment of cancer information overload: A comment on Jensen et al. (2014). Patient Education & Counseling, 98(11), 1450. 10.1016/j.pec.2015.04.020 [DOI] [PubMed] [Google Scholar]
  10. Cronbach LJ (1947). Test “reliability”: Its meaning and determination. Psychometrika, 12(1), 1–16. 10.1007/BF02289289 [DOI] [PubMed] [Google Scholar]
  11. DeVellis RF (2017). Scale development: Theory and applications (4th ed.). Sage. [Google Scholar]
  12. Fowler C, Gasiorek J, & Afifi W (2018). Complex considerations in couples’ financial information management: Extending the theory of motivated information management. Communication Research, 45(3), 365–393. 10.1177/0093650216644024 [DOI] [Google Scholar]
  13. Holbert RL, & Stephenson MT (2008). Commentary on the uses and misuses of structural equation modeling in communication research. In Hayes AF, Slater MD, & Snyder LB (Eds.), The Sage sourcebook of advanced data analysis: Methods for communication research (pp. 185–218). Sage. [Google Scholar]
  14. Hu L, & Bentler PM (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1–55. 10.1080/10705519909540118 [DOI] [Google Scholar]
  15. Jensen JD, Carcioppolo N, King AJ, Scherr CL, Jones CL, & Niederdeppe J (2014). The cancer information overload (CIO) scale: Establishing predictive and discriminant validity. Patient Education & Counseling, 94(1), 90–96. 10.1016/j.pec.2013.09.016 [DOI] [PubMed] [Google Scholar]
  16. Jensen JD, King AJ, Carcioppolo N, Krakow M, Samadder NJ, & Morgan SE (2014). Comparing tailored and narrative worksite interventions at increasing colonoscopy adherence in adults 50 – 75: A randomized controlled trial. Social Science & Medicine, 104, 31–40. 10.1016/j.socscimed.2013.12.003 [DOI] [PubMed] [Google Scholar]
  17. Jensen JD, Pokharel M, Carcioppolo N, Upshaw S, John KK, & Katz RA (2020). Cancer information overload: Discriminant validity and relationship to sun safe behavior. Patient Education & Counseling, 103(2), 309–314. 10.1016/j.pec.2019.08.039 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Jo HS, Park K, & Jung SM (2019). A scoping review of consumer needs for cancer information. Patient Education & Counseling, 102(7), 1237–1250. 10.1016/j.pec.2019.02.004 [DOI] [PubMed] [Google Scholar]
  19. Kaplan KJ (1972). On the ambivalence-indifference problem in attitude theory and measurement: A suggested modification of the semantic differential technique. Psychological Bulletin, 77(5), 361–372. 10.1037/h0032590 [DOI] [Google Scholar]
  20. Khaleel I, Wimmer BC, Peterson GM, Zaidi STR, Roehrer E, Cummings E, & Lee K (2019). Health information overload among health consumers: A scoping review. Patient Education & Counseling, 103(1), 15–32. 10.1016/j.pec.2019.08.008 [DOI] [PubMed] [Google Scholar]
  21. Kim K, Lustria MLA, Burke D, & Kwon N (2007). Predictors of cancer information overload: Findings from a national survey. Information Research, 12(4). [Google Scholar]
  22. Ko LK, Turner-McGrievy GM, & Campbell MK (2014). Information processing versus social cognitive mediators of weight loss in a podcast-delivered health intervention. Health Education & Behavior, 41(2), 197–206. 10.1177/1090198113504413 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lee CJ, Hornik R, & Hennessy M (2008). The reliability and stability of general media exposure measures. Communication Methods and Measures, 2(1–2), 6–22. 10.1080/19312450802063024 [DOI] [Google Scholar]
  24. Liu CF, & Kuo KM (2016). Does information overload prevent chronic patients from reading self-management educational materials?. International Journal of Medical Informatics, 89, 1–8. 10.1016/j.ijmedinf.2016.01.012 [DOI] [PubMed] [Google Scholar]
  25. Magnini VP, & Dallinger I (2018). Consumer information overload and the need to prompt script deviations. Journal of Quality Assurance in Hospitality & Tourism, 19(3), 285–297. 10.1080/1528008X.2016.1230038 [DOI] [Google Scholar]
  26. McNeish D, & Kelley K (2019). Fixed effects models versus mixed effects models for clustered data: Reviewing the approaches, disentangling the differences, and making recommendations. Psychological Methods, 24(1), 20–35. 10.1037/met0000182 [DOI] [PubMed] [Google Scholar]
  27. McNeish D, & Stapleton LM (2016). Modeling clustered data with very few clusters. Multivariate Behavioral Research, 51(4), 495–518. 10.1080/00273171.2016.1167008 [DOI] [PubMed] [Google Scholar]
  28. Micceri T (1989). The unicorn, the normal curve, and other improbable creatures. Psychological Bulletin, 105(1), 156–166. 10.1037/0033-2909.105.1.156 [DOI] [Google Scholar]
  29. Niederdeppe J, & Gurmankin Levy A (2007). Fatalistic beliefs about cancer prevention and three prevention behaviors. Cancer Epidemiology, Biomarkers, & Prevention, 16(5), 998–1002. 10.1158/1055-9965.EPI-06-0608 [DOI] [PubMed] [Google Scholar]
  30. Niederdeppe J, Lee T, Robbins R, Kim HK, Kresovich A, Kirshenblat D, Standridge K, Clark CE, Jensen JD, & Fowler EF (2014). Content and effects of news stories about uncertain cancer causes and preventive behaviors. Health Communication, 29(4), 332–346. 10.1080/10410236.2012.755603 [DOI] [PubMed] [Google Scholar]
  31. Obamiro K, & Lee K (2019). Information overload in patients with atrial fibrillation: Can the cancer information overload (CIO) scale be used? Patient Education & Counseling, 102(3), 550–554. 10.1016/j.pec.2018.10.005 [DOI] [PubMed] [Google Scholar]
  32. Occa A, Morgan SE, Peng W, Mao B, McFarlane SJ, Grinfeder K, & Byrne M (2020). Untangling interactivity’s effects: The role of cognitive absorption, perceived visual informativeness, and cancer information overload. Patient Education & Counseling, 104(5), 1059–1065. 10.1016/j.pec.2020.10.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Ramondt S, & Ramirez AS (2019). Assessing the impact of the public nutrition information environment: Adapting the cancer information overload scale to measure diet information overload. Patient Education & Counseling, 102(1), 37–42. 10.1016/j.pec.2018.07.020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Satorra A, & Bentler PM (2010). Ensuring positiveness of the scaled difference chi-square test statistic. Psychometrika, 75, 243–248. 10.1007/s11336-009-9135-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Shepherd J, Peersman G, Weston R, & Napuli I (2000). Cervical cancer and sexual lifestyle: A systematic review of health education interventions targeted at women. Health Education Research, 15(6), 681–694. 10.1093/her/15.6.681 [DOI] [PubMed] [Google Scholar]
  36. Song H, & Ewoldsen DR (2015). Metacognitive model of ambivalence: The role of multiple beliefs and metacognitions in creating attitude ambivalence. Communication Theory, 25(1), 23–45. 10.1111/comt.12050 [DOI] [Google Scholar]
  37. Specht J, Egloff B, & Schmukle SC (2013). Examining mechanisms of personality maturation: The impact of life satisfaction on the development of the Big Five personality traits. Social Psychological and Personality Science, 4(2), 181–189. 10.1177/1948550612448197 [DOI] [Google Scholar]
  38. Swar B, Hameed T, & Reychav I (2017). Information overload, psychological ill-being, and behavioral intention to continue online healthcare information search. Computers in Human Behavior, 70, 416–425. 10.1016/j.chb.2016.12.068 [DOI] [Google Scholar]
  39. Tiro JA, Vernon SW, Hyslop T, & Myers RE (2005). Factorial validity and invariance of a survey measuring psychosocial correlates of colorectal cancer screening among African Americans and Caucasians. Cancer Epidemiology and Prevention Biomarkers, 14(12), 2855–2861. 10.1158/1055-9965.EPI-05-0217 [DOI] [PubMed] [Google Scholar]
  40. Vaidya JG, Gray EK, Haig J, & Watson D (2002). On the temporal stability of personality: Evidence for differential stability and the role of life experiences. Journal of Personality and Social Psychology, 83(6), 1469. 10.1037/0022-3514.83.6.1469 [DOI] [PubMed] [Google Scholar]
  41. Whitelaw S (2008). Health information: A case of saturation or 57 channels and nothing on?. The Journal of the Royal Society for the Promotion of Health, 128(4), 175–180. 10.1177/1466424008092233 [DOI] [PubMed] [Google Scholar]
  42. Witte K (1994). Fear control and danger control: A test of the extended parallel process model (EPPM). Communication Monographs, 61(2), 113–134. 10.1080/03637759409376328 [DOI] [Google Scholar]
  43. Wu CH (2016). Personality change via work: A job demand–control model of Big-five personality changes. Journal of Vocational Behavior, 92, 157–166. 10.1016/j.jvb.2015.12.001 [DOI] [Google Scholar]

RESOURCES