Abstract
There is concern that public education about testicular cancer (TC) may cause unnecessary anxiety. Psychological theory suggests that if threat (eg, TC) information is accompanied with threat control strategies (eg, testicular self-examination; TSE) anxiety is less likely. Male students (N=443) were randomized to either a TC or TC +TSE information group or a no information control group, and assessed at three time points. Anxiety levels did not differ between the groups and exposure to TC+TSE resulted in greater perceived message benefit, increased intention to self-examine and lower message denigration. This suggests TC information is not anxiogenic, but inclusion of TSE information may improve acceptance of disease awareness information.
Introduction
Low public awareness of symptoms and negative beliefs about cancer outcomes are considered key to late presentation of cancer and more advanced disease at diagnosis (Richards, 2009). The National Awareness and Early Diagnosis Initiative (NAEDI) was launched in 2008 in the UK by the Department of Health and Cancer Research UK to promote early diagnosis of cancer and potentially to save several thousands of lives each year. However, there is continuing concern over the possibility that efforts to promote early presentation could increase anxiety about cancer.
Men have poor knowledge about testicular cancer (TC) symptoms and rarely engage in testicular self-examination (TSE) (Cummings et al., 1983; Evans et al., 2010; Thornhill et al., 1986; Wardle et al., 1994). Even symptomatic men delay seeking help for testicular cancer (Smith et al., 2005). This has resulted in educational campaigns designed to increase awareness of testicular cancer in the healthy male population and help-seeking among those with symptoms (eg, Cancer Research UK, 2005; Trumbo, 2004). However, while some evidence suggests that these campaigns have been successful at increasing knowledge and reducing help-seeking delays (Khadra and Oakeshott, 2002; Vasudev et al., 2004), they have also attracted criticism and controversy (Law, 2004).
Particular concern has focussed on the recommendation to carry out TSE (Meffan et al., 1991). Given the lack of formal evidence for efficacy, some authorities have expressed the view that it is unethical to expose young men to the unnecessary anxiety that is assumed to be caused by information on TSE or TC. In fact, evidence on the anxiogenic effects of TC information is far from clear. Weist and Finney (1996) found that scores on a measure of state anxiety (STAI) after TSE exposure were within normal limits in two samples (N=29, N=30) of male students. However they did not measure baseline anxiety and did not include a control group. In contrast, Morman (2000) reported moderate levels of fear on a 5-item mood adjective scale in 80 male students after exposure to TC information. However, no data was provided on the duration of the fear response or whether it impacted on other aspects of emotional well-being, for example, anxiety or other measures of distress and again there was no control group.
Psychological theory suggests that inclusion of TSE advice alongside TC information might actually diminish negative emotional responses and could therefore have value beyond any effects on early detection. According to the Extended Parallel Process Model (EPPM; Witte, 1998), there are two possible response pathways following exposure to threat; ‘danger control’ or ‘fear control’. ‘Danger control’ (behaviour designed to reduce the objective threat) is elicited when high threat is combined with high perceived efficacy for dealing with threat. This could manifest itself in increased intention to self-examine. Attitudes towards the health information message tend to be positive in these circumstances. When efficacy information is absent, the EPPM predicts that people make efficacy judgments based on prior experience and in this case, men may rely on generalized perceptions of the controllability of cancer, which tend to be fatalistic (Powe and Finnie, 2003). High threat and low efficacy elicit a ‘fear control’ response, designed to reduce the aversive emotional state. This may involve derogating the message or turning attention away from sources of information about cancer. On this basis, it could be argued that inclusion of TSE advice as part of TC information would not only decrease anxiety, but also promote message acceptance and a positive attitude towards receipt of the information. We therefore hypothesized that inclusion of TSE advice in TC information would lead to lower anxiety, higher self and response efficacy, more positive message appraisal and higher intention to self-examine, compared with TC information alone.
Methods
Design and procedure
The study was a randomized controlled trial with three groups (see Appendix Figure 1: Study Profile). Participants were randomized using simple random numbers on Minitab to either: 1) control group: no information; 2) TC information (disease incidence, risk factors and signs and symptoms); and 3) TC + TSE information. The study was longitudinal with a baseline phase, intervention phase (seven days from baseline) and follow-up phase (five to seven days from the intervention). Three assessment points; T1 (baseline), T2 (post-intervention) and T3 (post follow-up) assessed whether effects were observed immediately after the intervention stage and examined whether they attenuate over time. The numbers recruited satisfied prior power calculations that specified a minimum of 64 participants in each group to detect a clinically significant difference in state anxiety, defined as a 5-point shift on the STAI measure of state anxiety (Spielberger, 1983) using the significance and power parameters α = 0.005 and β = 0.80. Ethical approval for the study was obtained from University College London (UCL) Research Ethics Committee.
The recruitment email, which went to all students at the university, briefly described the study, assured confidentiality and compliance with the Data Protection Act, gave details of a financial incentive for participation (a prize draw to win a top prize of £500 and two smaller prizes of £100), and provided a web-link to the study website. A website was created specifically for this study, with a first page consisting of a combined participant information and consent form, followed by the baseline questionnaire. Participants were asked to supply an email address to be used to contact them at T2 and T3. They were contacted seven days after baseline to complete the information intervention and then 5–7 days later with a request to complete the follow-up. Participants were sent daily reminders until they had either completed each questionnaire or 14 days had elapsed. Due to an error, a small number of students (N=24) were invited to complete follow-up (T3) before the planned 5–7 day interval. The time taken to complete T2 ranged from 6–24 days (median = 7), and T3 from 1–20 days (median = 7 days). Time taken to complete each phase did not differ between the three groups. Data were automatically captured by the computer program running the website and could be downloaded for analysis.
Once data collection had been completed, participants were debriefed via email about the intentions of the study and provided with details of where to access additional information about testicular cancer should they want it.
Participants
Participants were male university students (N=457) because they are a key demographic group for receiving information on testicular cancer. Participants were recruited to the study during April and May 2006. The mean age of students completing both baseline and intervention stages was 24 years (range = 23–25 years) and they were predominantly white (68%; N=214), undergraduates (57%; N=181) and studying science or engineering (41%; N=129). There were no significant differences on these characteristics between the three randomized groups.
Intervention materials
TC and TSE information was developed using existing materials published by Cancer Research UK and sponsored by Department of Health to approximate real-world cancer education. Both ‘TC’ and ‘TC + TSE’ texts scored 60 points or above on the Flesch Readability Score (within the acceptable range for a general population sample [Flesch, 1949]). The text was presented without graphics using simple san-serif type face with a high contrast background to maximize visibility. The TC information stated that young men (15–45 years) were most at risk, described established risk factors (eg, undescended testicles) and outlined the main signs and symptoms (eg, painless testicular lump). The TSE section described the benefits of early detection in terms of increased treatment effectiveness and detailed procedural advice on doing TSE.
Measures
Manipulation check
Knowledge was used as a manipulation check confirming that exposure to information led to the anticipated cognitive impact (knowledge gain) in the two experimental groups. Participants completed a knowledge questionnaire at baseline and immediately after information exposure. The questionnaire asked whether they knew anything about TC and presented 13 statements to which they could answer true, false or unsure. Nine of these statements were obtained from a previous study (Steadman and Quine, 2004), with four additional statements devised for this particular study. The number of correct responses was summed to give each individual a knowledge index score out of 13. For some analyses, participants were classified as either low or high on knowledge based on a median score of seven.
Demographic measures
Age, ethnicity, educational status (undergraduate and postgraduate) and subject studied (science, social science, medicine, humanities, arts and engineering) were recorded. Ethnicity was later recoded into two categories (white and not-white). Subject studied was reduced to four categories by combining science and engineering, and humanities and arts.
Previous awareness of TC and prior TSE frequency
At baseline, participants were asked whether they had previously heard of TC (yes or no). They were also asked to indicate on a 5-point response scale ranging from ‘never’ to ‘once a day or more’, whether they checked their testicles for signs of cancer; these were categorized into ‘never’ or ‘ever’ practicing TSE.
Anxiety
There is inconsistency in the literature about how anxiety as a construct is described and its relation to cancer surveillance behaviour (see Consedine et al., 2004 for a review). Anxiety is sometimes construed to be largely cognitive, and in other instances affective with physiological manifestation. In this study we have used measures based on the latter conceptualization. State anxiety and cancer specific anxiety were measured at each time. The brief form of the Spielberger State Trait Anxiety Inventory (STAI: Marteau and Bekker, 1992) was used to measure state anxiety. Scores were pro-rated to the long form for comparison with other studies and ranged from 20–80, with 43 representing the lower end of clinically significant anxiety. The Breast Cancer Fear Scale (Champion et al., 2004) was adapted to measure anxiety about TC. This 8-item measure assesses the physiological experience of fear/anxiety when thinking about cancer (e.g. ‘when I think about [testicular] cancer I get jittery’). Participants indicated agreement with the statements on a 5-point scale ranging from 1= strongly disagree to 5= strongly agree, and these were summed to give a score ranging from 8–40 (8–15 = low TC anxiety, 16–23 = moderate anxiety, 24–40 = high anxiety). Both measures were included to assess emotional well-being as there is debate about which approach is more appropriate. Prior research measuring the emotional impact of cancer screening reports that specific measures are more sensitive to distress. However it has been suggested that this distress is only an artefact of asking questions about cancer which serve to increase the salience of cancer worries (see Miles et al., 2009 for a review). In the absence of clarity on this issue we thought it prudent to include both measures.
Threat and efficacy perceptions
Threat and efficacy perceptions were measured at each time point. The scales were modified versions of those described by Witte (Witte et al., 1996) to make them specific to TC. Each construct consists of two sub-constructs. Threat comprized perceived seriousness of, and perceived susceptibility to, TC. Efficacy comprized response efficacy and self efficacy related to TSE and help-seeking for TC symptoms. Response efficacy items measured how effective TSE or help seeking was perceived to be at reducing the threat from TC, and self efficacy items measured how easily respondents felt they could carry out these behaviours. Each of these sub-constructs was measured using three items on a 7-point scale ranging from 1= strongly disagree to 7 = strongly agree. Scores for each scale were summed and averaged so that scores ranged from 1–7 to enable comparison with other similar studies.
Message appraisal
Participants in the two information groups were asked to evaluate the message along a number of positive and negative dimensions. Positive dimensions were message understanding and benefit from message (eg, ‘ I am pleased I have read this message’). Negative dimensions were message derogation and manipulative intent of the message (eg, ‘this message was misleading’). Each dimension was measured using scales with between two and four items and participants responded using a 7-point scale (1=strongly disagree to 7=strongly agree). These were summed and averaged so that scores ranged from 1–7. All items were derived from Witte et al. (1996). A 3-item fear scale was used to assess how frightening the message was perceived to be (eg, ‘this message made me feel frightened’).
TSE intention
Intention to perform self-examination in the future was assessed after information exposure, using a 5-item measure (Steadman and Quine, 2004). Participants were asked to rate agreement on a 5-point scale (from 1 = totally disagree to 5 = totally agree) to items such as: ‘It is likely that I will perform self-examination in the next week’. Responses were summed to produce scores that ranged from 5–35.
Scale reliability
Internal reliability coefficients were assessed for each measure and were within the acceptable range (α =0.61[perceived susceptibility] to α = 0.96 [TSE intention]) with the exception of the perceived understanding scale which fell to α = 0.54 at Time 2. Reliability improved to α=0.74 with the deletion of the item: ‘I learnt a lot about testicular cancer from reading the message’, and this item as well as the remaining 2-item scale were entered into the analyses separately.
Statistical analyses
The main outcome was level of anxiety. Analysis of co-variance (ANCOVA) was used to assess group differences on this and other variables at T2 and T3, controlling for baseline values. Data were analysed using SPSS Version 14.0 for Windows. Planned contrasts compared the control group with each of the information groups and compared the two information groups. Analyses were adjusted for age and academic status. The number of days since exposure was controlled for in T2 and T3 analyses.
Results
Demographic and baseline findings
Based on the inclusion and exclusion criteria, 14 participants were excluded; six on the basis of age and eight due to a reported cancer diagnosis. The remaining 443 students were randomized to one of the two information groups or the control group. Of these, 316 completed the information exposure phase and T2 questionnaire and 272 completed the follow-up (T3) questionnaire (refer to Appendix Figure 1 for a participant flow diagram through the study). Students who ceased study participation after baseline (N = 127) were significantly younger (mean age = 22 versus 24 years, F(1,443) = 8.25, p=0.004) and were more likely to be undergraduates than postgraduates (73.2% (N=93) versus 57.3% (N=181), X2(1, 443) = 9.77, p = 0.002). There were no differences in demographic characteristics between those who dropped out of the study or continued after T2. There was also no difference in baseline state anxiety between those who dropped out of the study before T2 and those who continued to participate. Attrition rates varied between groups at T2 (range 22 % to 36%) and T3 (range 9% to 13%). The difference was significant, X2(2, 443) = 6.408, p = 0.041, with a higher completion rate in the control group (78%) than in the TC + TSE group (64 %).
There were no significant group differences in baseline measures except that participants allocated to the TC + TSE group had significantly lower state anxiety than either the control group, F(1,308) = 5.10, p = 0.025 or the TC group F(1,308) = 4.1, p = 0.025. The majority of participants (93%; 293 out of 316) had previously heard of TC and 60 percent (190 out of 316) reported that they had carried out TSE at least once in the last 12 months. The mean knowledge score for the whole sample was 6.8 out of a possible 13 (CI = 6.4–7.1).
Immediate impact of information exposure (T2)
Knowledge (manipulation check)
The highest knowledge score following information exposure was observed in the TC + TSE group, with a mean of 8.7 (CI = 8.34–9.14) followed by the TC group with a mean of 8.1 (CI = 7.7–8.5). These scores were both significantly different from the control group mean of 6.8 (CI = 6.4–7.1), F(2,307) = 29.60, p = 0.000, and from each other, F(1,307) = 4.94, p = 0.027.
Anxiety
Mean scores and differences between groups for anxiety and threat and efficacy perceptions following information exposure are presented in Table 1. Neither general anxiety nor TC anxiety differed significantly between groups.
Adjusted a anxiety and perceptions of threat and efficacy at T2
Adjusted for age, educational status, baseline measurements and interval (days) between T1 and T2.
Means followed by different letters differ significantly. TC= testicular cancer, TSE = testicular self-examination
Threat and efficacy perceptions
After information exposure, participants in the TC group reported higher perceived severity of TC than either the control, F(1,307) = 5.80, p = 0.016, or TC + TSE groups, F(1,307) = 7.53, p = 0.007. Perceived susceptibility did not differ significantly across groups.
Response efficacy for TSE was higher in the TC + TSE group than the control group, F(1,307) = 7.17, p = 0.008. Self efficacy for TSE in the TC + TSE group was also higher than in the control group F(1,307) = 11.15, p = 0.001 and the TC group F(1,307) = 3.91, p = 0.049. Efficacy for help-seeking did not differ significantly between the groups.
Message appraisal
Mean scores and differences between the two information groups for message appraisal and TSE intention following information exposure are presented in Table 2. After information exposure, participants in the TC + TSE group reported greater learning from the message F(1,192) = 8.636, p = 0.004, perceived more benefit from having read it F(1,192) = 9.086, p = 0.003, and were less likely to derogate the message compared to the TC group F(1,192) = 6.323, p = 0.013. There was no difference between the TC and TC + TSE groups in terms of perceived manipulation from the message, or how frightening the message was perceived to be.
Adjusted a mean response scores for message appraisal and TSE intention after information exposure intervention (T2)
Adjusted for age, educational status, baseline measurements and interval (days) between T1 and T2.
Mean (CI) TSE intention for control group: 24.2 (22.78-25.63). TC = testicular cancer, TSE = testicular self-examination
We based our hypothesis that the TC group would be more likely to engage in fear processing than the TC + TSE group, on the assumption that TC knowledge and experience would be low, which was not the case. Previous studies (Nabi et al., 2008; Steffen, 1990) suggest prior knowledge and experience influence responses to threatening information, and we therefore entered baseline knowledge scores and past TSE practice into the ANOVA analyses as either a co-variate (prior knowledge) or a fixed factor (prior TSE practice). We found evidence of a moderating effect of prior knowledge on the likelihood to derogate the information, with participants with low prior knowledge (scores below 7) more likely to derogate the information (group mean for derogation: 2.64) than those with higher knowledge (group mean: 2.20, F(1,188) = 7.69, p = 0.006). This effect was consistent across both information groups.
TSE intention
Intention to perform self-examination was highest in the TC + TSE group, which was significantly different from the control group, F(1,308) = 12.80, p = 0.000 and the TC group F(1,308) = 4.26, p = 0.040. Given that past TSE practice was likely to influence intention to practice TSE in the future, this variable was entered into an ANOVA analyses as a fixed factor. Prior TSE practice moderated the influence of the information on intention to do TSE in the future. A significant group-by-prior TSE practice intervention, F(2,302) = 4.10, p = 0.018, revealed that for the TC and control groups, intention was significantly higher in the participants who had ever practiced compared with those who had not, while there was no significant difference between the two practice groups for the TC + TSE participants.
Longer-term impact of information exposure (T3)
Anxiety
State anxiety scores and TC anxiety were not significantly different across the three groups.
Threat and efficacy perceptions
Perceived severity remained higher in the TC group, 5.38 (5.17–5.60), than the control group 4.99 (4.80–5.19): (F(1,264) = 2.62, p = 0.009) and the TC + TSE group, 4.98 (4.75–5.21): F(1,264) = 6.91, p = 0.009). By follow-up, TSE efficacy perceptions no longer significantly differentiated the groups.
TSE intention
At T3, intention to self-examine remained significantly higher in the TC + TSE group, 26.99 (25.23–28.75), compared with the control group, 23.97 (22.47–25.47): (F(1,264) = 6.63, p = 0.011). The moderating effect of prior TSE practice was no longer significant.
Discussion
The intention of this study was to assess whether exposure to TC information led to an increase in anxiety. We hypothesized that the composition of the information, specifically whether it included efficacy (TSE) information, may lead to a comparatively lower level of anxiety because it might encourage cognitive style ‘danger control’ processing. The results indicated that neither TC information or TC + TSE led to increased anxiety in general, or specifically about TC, compared with the control group. This suggests that TC information is not, at least among students who choose to read it, anxiety-inducing. Additional exposure to testicular self-examination information did not produce lower anxiety than disease information alone and this was true for both general and TC anxiety between the two information groups.
As predicted, the TC + TSE group reported higher self efficacy than the other information group. According to the EPPM model (Witte, 1998), upon perceiving a threat such as risk of TC, if efficacy is perceived to be high the person will follow the ‘danger control’ pathway; if it is low they follow the ‘fear control’ pathway. The TC + TSE group reported higher levels of intention to do TSE in the future were, more likely to appraise the message positively, and were less likely to derogate the message than those who had TC alone. This pattern supports the predicted relationship, with higher perceived efficacy, being associated with following the ‘danger control’ processing pathway. It mirrors results reported by Morman (2000), where men who read a message that included efficacy information scored significantly higher on a TSE intention scale and reported better understanding of the message.
We did not have evidence that TC information without TSE advice activated the ‘fear control’ pathway. Although the TC group reported higher perceived severity of testicular cancer and lower perceived TSE self efficacy, the absolute level of perceived efficacy in the TC group was still higher than perceived threat; the condition specified for ‘danger control’ processing. Consistent with this, intention to do TSE in the future (an indicator of ‘danger control’ processing), was still relatively high in the TC group. This suggests that the TC group was also following the ‘danger control’ processing pathway but less strongly. If the TC information had activated the fear control pathway we would have expected the TC group to have low intentions and larger differences in message appraisal. This would be consistent with participants rejecting the health message as a means of emotional self-preservation (ie, if they don’t believe the message is valid they no longer need to fear it). Coupled with high intentions in both groups to carry out TSE and relatively equivalent knowledge gains, this suggests that both versions of the TC information are effective and acceptable.
The influence of TC + TSE information on message appraisal and intention to carry out TSE may be explained by several factors, such as the additional quantity of the information in the TC + TSE group, or the generally reassuring nature of the TSE information (ie, that early detection leads to a better outcome), rather than the specific reference to the TSE procedure. It is possible that the same outcomes could have been achieved by provision of other reassuring information such as the excellent survival rates, rather than specific provision of TSE procedural information. Further work is needed to elucidate the exact components of the information responsible for the findings.
The participants used in this study were unusually high on prior awareness and practice of TSE compared with other results (Evans et al., 2006; Khadra and Oakeshott, 2002; Wardle et al., 1994). This might explain the relatively high efficacy perceptions at baseline and the limited impact of either variety of information. The TC-only group may have been able to draw on their prior knowledge and past experience to form efficacy perceptions and therefore be less vulnerable to following the ‘fear control’ pathway. There is evidence that prior knowledge influences how people respond to health information that is potentially threatening (Nabi et al., 2008; Steffen, 1990). In line with this, we found that individuals with low prior knowledge and who had never practiced TSE were more likely to derogate the message.
The high level of prior awareness and TSE practice probably reflects the fact that it was a self-selected, university-based sample and from a country that has been active in increasing TC awareness in the last decade. The study was explained in the recruitment email, which may have resulted in individuals with lower TC anxiety choosing to take part. It would be useful to ascertain whether contemporary population representative samples match the level of reported awareness here, and to explore the differential impact of TC and TC + TSE information in groups with less prior knowledge and experience.
A further limitation of our study is that the number of days between exposure to the information and subsequent assessments of the variables of interest varied widely at both Time 2 and Time 3. Although we controlled in our analyses for the number of days since exposure to the information, this variation may still have important implications. For example, as time from exposure increases there may be recall biases or other time effects that could confound the relationship between the information and the variables that we measured.
The findings from this study have implications for the debate surrounding promotion of TC and TSE awareness, as well as for the broader issue of delivery of public health information. They suggest that TC information, with or without inclusion of TSE advice, is acceptable and does not lead to increased anxiety, at least in populations with high initial cancer knowledge. This is in line with other experimental studies with community samples that may be more representative of the general population with respect to their cancer knowledge (although this was not explicitly measured), which have found no increased anxiety after exposure to interventions aimed at promoting awareness and early detection of colorectal cancer (Robb et al., 2006; Wardle et al., 1999). Concerns that providing TC information will cause anxiety are likely to be unfounded. The findings also indicate that inclusion of TSE advice increases positive appraisal of TC information. Therefore, although there is no direct evidence for the clinical benefit of promoting TSE, there may be some psychological benefit of including it in terms of message acceptance and satisfaction. This has implications for the provision of cancer information generally, highlighting the value of including efficacy information within health education. Future research exploring the impact of including efficacy information relevant to other cancer prevention detection and control behaviours is warranted.