We conducted a cross-sectional survey using a modified Dillman [18] survey methodology, following the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) reporting standards for cross-sectional surveys [19]. Survey packages included an addressed, stamped, return envelope, which was to be returned after the survey was completed. After 4 weeks, a second survey package was sent to nonrespondents.

We surveyed men diagnosed with prostate cancer in four Canadian provinces—British Columbia, Alberta, Saskatchewan, and Ontario—in 2014-2015 using their respective provincial cancer registries. We sought to obtain responses from 10% of the provincial patients and expected a 30% response rate. Thus, to achieve responses from 10% of the target population in each province, each registry invited a random selection of 55%-60% of men diagnosed with prostate cancer in the last 6 months of 2012 to participate in the study. We selected 2012, as it was the latest year for which the registries had complete data. The only inclusion criteria were that the patient had a prostate cancer diagnosis and lived in that province that year.

Three registries (British Columbia, Alberta, and Saskatchewan) used an “opt-out” recruitment strategy, whereby the registry provided a cover letter introducing the study in the survey package, making clear that the recipient could choose whether to complete the survey. The fourth registry (Ontario) used an “opt-in” recruitment strategy, providing a letter introducing the study and required the recipient to phone the registry to express their interest in participating. The names of interested patients from the opt-out province were then forwarded to the Ontario Lead at the study central office from where the survey was mailed out.

Each survey had a unique study identification code and did not include any identifying information. For the opt-out provinces, the provincial registries identified nonresponders and sent out the second survey 4 weeks later. For the opt-in province, the Ontario lead identified nonresponders and sent out the second survey 4 weeks later. All completed surveys were eventually mailed to the Ontario Lead at the study central office to be entered in an electronic database for analysis. Ethical approval was attained from each province’s respective university and cancer agency and from the Ontario lead’s university.

The questionnaire contained five sections with 40 questions. It was developed by a team of researchers and health care professionals and piloted with five patients prior to implementation [13]. This paper reports on questions pertaining to patterns of internet access and use (n=7), confidence using the internet as a source of health information (n=1), factors associated with health-related internet use including participant demographics (n=8), and what men with prostate cancer want in a website (n=2). All responses were self-reported.

Questions pertaining to internet access and use included use of the internet (yes/no), type of internet access (dial-up, broadband, cellular network, wireless network, or not sure), devices used to access the internet (desktop/laptop computer, tablet, cellphone/smartphone, or other), internet access from home (yes/no), frequency of internet access (at least once a day, at least once a week, at least once a month, or less than once a month), and websites used for information on prostate cancer or its treatments (open response option).

Health-related internet use was measured by asking respondents to indicate “how easy or difficult it was to obtain information about prostate cancer and/or its treatments from each of the sources below” of which, “internet (other than personal email or online support groups)” was an option. Response categories included very easy, easy, difficult, very difficult (which were all coded as “yes”), and did not try to use this source (which was coded as “no”).

Confidence in using the internet as a health resource was measured with one item from the eHealth Literacy Scale (eHEALS). The eHEALS is a validated measure of electronic health (eHealth) literacy, which is intended to capture people’s perceived skills at using information technology for health [20]. Specifically, we adapted the final question in the eHEALS scale for prostate cancer: “I feel confident in using information from the Internet to help make health decisions [related to my prostate cancer].” We used the same response options as the eHEALS scale: a 5-point Likert scale (1=strongly disagree, 5=strongly agree) with the addition of, “Not applicable, I do not use information from the Internet to help make decisions related to my prostate cancer.”

Lastly, preferences for using the internet as a health resource were investigated by asking participants to select information and features wanted in a website, from a list.

We examined associations between several variables and health-related information use. Sociodemographic variables included age (43-65, 66-75, ≥76 years), education completed (primary or secondary school/college or university), income (≤CAD 40,000, CAD 40,001-80,000, ≥CAD 80,001), area of residence (urban or suburban/rural, town or country), and broadband internet use (yes/no). Internet experience variables included frequency of internet use (every day/not every day), number of devices used to access the internet (one device/more than one device), and confidence in using information from the internet to inform health decisions (as described above). Clinical variables included treatments (active/active surveillance or watchful waiting). Information seeking and decision making variables included an information-seeking role (I did all the looking, I did some of the looking along with someone else/someone else did most or all of the looking, or I did want any information), and role in the decision making process (I made the decision by myself or with my doctor, family/my doctor or a family member made the decision). Health status variables included overall health (very good or good/poor or very poor) and number of unmet supportive care needs, as measured by the validated 34-item Supportive Care Needs Survey and the 8-item prostate cancer–specific module [21]. Responses were no need or not applicable, yes and the need is met (met need), and yes but the need is not met (unmet need). Unmet need items were summed to create a total unmet need score.

A total of 1362 patients returned partially or fully completed surveys across provinces and 46 returned blank surveys. Blank surveys and missing responses from partially completed surveys were excluded from the analysis. The survey response rate for the opt-out provinces was 46%-55% and for the opt-in province, it was 13%. Table 1 shows respondents’ sociodemographic, treatment, and general health characteristics. The age range of respondents was 43-95 years. Most were on follow-up after treatment (63%) and in good or very good health (93.6%).

Table 2 shows the patterns of internet access and use. A total of 82% of respondents were internet users. The majority of respondents (70.7%) used the internet at least once a day, had home internet access (84.3%), and accessed the internet from a desktop or laptop computer (75.7%) through broadband or wireless (47%) internet connection.

As shown in Table 2, 65% (n=784) reported that they used the internet as a source of information about prostate cancer. With respect to confidence in using information from the internet to help make decisions related to prostate cancer, 40.2% (n=521) of the respondents agreed or strongly agreed with the statement, “I feel confident in using information from the Internet to help make health decisions related to my prostate cancer,” 33.5% (n=386) were undecided or disagreed, and about one-quarter (26.3%; n=341) reported that they did not use information from the internet to make decisions related to their prostate cancer. Findings from hypothesis testing provide support of the construct validity of the single item eHealth confidence variable. eHealth confidence was positively correlated with the frequency of internet use (r=0.1, P=.003) and negatively correlated with “not being comfortable using a computer or mobile device” (r=–0.15, P<.001). In addition, eHealth confidence was negatively correlated with known components of eHealth literacy when framed as barriers. These include “not knowing how to judge the quality of the information or what information to trust” (r=–0.2, P<.001), “not knowing how what information applied to me” (r=–0.2, P<.001), “not knowing how or where to search for information” (r=–0.2, P<.001), and “having difficulty finding information that I could understand” (r=–0.2, P<.001).

As shown in Table 3, factors that correlated significantly (P<.05) with health-related internet use were younger age, higher education, higher income, urban residence, broadband internet access, frequency of internet use, accessing the internet from multiple devices, confidence in using information from the internet to help make health decisions related to prostate cancer, active information seeking, active decision making role, and greater number of unmet supportive care needs.

All factors found to be significantly associated with health-related internet use in the univariable analyses were entered into the model. Table 4 shows the results of the multivariable logistic regression analysis. Our results indicate that the odds of using the internet as a health resource were higher among respondents who were active information seekers (OR: 4.57, 95% CI 2.5-8.3), were confident using information from the internet to make health decisions (OR: 3.56, 95% CI 2.27-5.59), had broadband internet access (OR: 1.76, 95% CI 1.14-2.7), and had more unmet supportive care needs (OR: 1.05, 95%CI 1.02-1.09).

A total of 22% of respondents (n=308) provided 359 valid responses to the open-ended question “Are there any specific Internet site(s) that you like to go to for information related to prostate cancer or its treatments? If Yes, please specify.” Responses that were not valid consisted of unspecified sites (eg, “health website”, search engines etc). As shown in Table 5, of the 10 categories of websites that were identified, the most commonly reported websites were national cancer agencies, government, or hospitals. The top three most commonly mentioned national cancer agencies or government websites were the Canadian Cancer Society (n=40), Prostate Cancer Canada (n=23), and the American Cancer Society (n=10). The top three most commonly reported hospital affiliated websites were Mayo Clinic (n=51) BC Cancer Agency (n=14) and Johns Hopkins (n=8). American hospital-affiliated websites accounted for over 50% of all responses in this latter category.

Respondents were also asked to indicate what features and information they would want in a website for men with prostate cancer and their families. As shown in Table 6, the top three features that over 50% of respondents wanted in such a website were a library of topics (n=893, 65.6%), tools to help select treatment options (n=815, 59.8%), and tools to help navigate the prostate cancer journey (n=698, 51.2%). As shown in Table 7, the information that ≥50% of respondents wanted in such a website included information on prostate cancer treatments (n=916, 65%), what is prostate cancer and its natural progression (n=904, 64%), how to manage side effects (n=858, 61%), personally relevant information (n=769, 55%), and latest research (n=746, 53%).

This study examined the patterns of, and factors associated with, the use of the internet as a source of health information among Canadian men with prostate cancer, and the features and information wanted in a website. A total of 82% of respondents in this sample were internet users. The prevalence of internet use in this sample is consistent with the 2016 internet use rates among Canadians within this age group, which showed that 81% of people aged 65-74 were internet users [12]. Although the majority of the respondents in our sample used the internet, less than half (47%) accessed the internet through broadband and less than 30% accessed the internet with a mobile device, with the most using desktop computers to access the internet. These access patterns are also consistent with current broadband and mobile phone usage rates among Canadian seniors, which are still considerably lower than those in younger age groups [12,25].

Our findings suggest that sociodemographic factors, namely age, education, and income, may not play a significant role in determining health-related internet use among prostate cancer survivors in Canada when other factors are considered. Similarly, a previous analysis [15] of the US National Health Information Trends Survey from 2003 to 2008 showed no statistically significant differences in age, education, or racial/ethnic aspects in health-related internet use patterns among cancer survivors, with the exception of emailing providers [15]. Survivors who were younger and had higher education were more likely to email their providers. Likewise, another analysis [11] of the 2000 Canadian National household internet survey found no association between health-related internet use and age, education, or income. Other studies suggest that there is still a gap in health-related internet use based on age and socioeconomic factors in North America [10,17]. However, these studies focused their analyses on internet users [10,17]. Chou et al [15] reported an independent association with educational attainment in their sample of cancer survivors. Similarly, our univariable analysis showed that age, education, and income were associated with health-related internet use among our sample of prostate cancer survivors.

However, our findings suggest that there is a gap in health-related internet use among prostate cancer survivors in Canada based on broadband access, which may reflect rurality [26]. Previous research has shown that urban Canadians are 1.5 times more likely to use the internet than rural Canadians [26]. In our model, respondents who had broadband access were 2.74 times more likely to use the internet as a health resource than those who did not. Broadband, as opposed to traditional dial-up, provides improved internet services, including faster browsing and downloading as well as telephone, radio, and videoconferencing. Not having access to broadband considerably limits one’s quality of internet and access to essential services. In 2000, Johnson et al [11] did not find an association between high-speed internet and health-related internet use in the Canadian national household internet survey. However, longer duration of internet use was associated with decreased likelihood of using the internet as a health resource [11]. Residential broadband was launched in Canada in 1998; hence, most Canadians would have been using dial up then, which would have been costly and unstable over longer periods of use [27]. In a 2015 qualitative study, patients with prostate cancer living in remote areas of British Columbia explained that they did not like to use the internet as a health resource because it was not reliable [28]. In 2016, Canada’s telecom agency declared broadband a basic service [29], along with a commitment to increase targets for download and upload speeds and extend access to regions without access. Our findings provide further evidence in favor of universal broadband internet access. Future research should investigate the impact of this policy.

In our model, the factor associated with the highest odds of using the internet as a health resource was being an active information seeker. Respondents who were information seekers were 4.6 times more likely to use the internet as a health resource than passive information seekers. A 2018 systematic review [30] of men’s health-seeking behavior that largely focused on prostate cancer found that the internet was the primary source of information for active information seekers. We assessed active information seeking by asking respondents to indicate who looked for information about prostate cancer and its treatment for them. The options included (1) “I did not want any information,” (2) “someone else did most or all of the looking,” (3) “I did some of the looking and someone else did some of it for me,” and (4) “I did most or all of the looking myself.” We dichotomized the responses to create a binary variable by combining (1) and (2) to reflect passive information seekers and (3) and (4) to reflect active information seekers. This variable reflects the difference between two coping styles categorized by Miller as “Monitors” and “Blunters” [31]. Monitors focus on acquiring information to help them problem solve and reduce uncertainty and are less satisfied with the information they receive from health care providers [31]. Blunters use distraction to avoid threatening information and are typically satisfied with the amount of information they receive from their health care providers [31]. Research suggests that psychoeducational interventions for cancer patients are most effective when the level and type of information are consistent with the individual's monitoring style and the demands of the health threat [32].

The factor next most strongly correlated with health-related internet use was eHealth confidence. Respondents who were confident using information from the internet to help make health decisions related to their prostate cancer were used 3.6 times more likely to use the internet as a health resource. eHealth confidence was measured using the final item of the eHEALS eHealth literacy measure [20]. eHealth literacy is defined as the ability to seek, find, understand, and appraise health information from electronic sources and apply that knowledge to solve a health problem or make a health-related decision [20]. The final item of the eHEALs scale is intended to capture one’s confidence in using information from the internet to help make health decisions. The level of agreement with this item in this sample (40.2%) is similar to that reported in a sample of older adult internet users in the United States (43.1%) [16]. Other studies have found an association between eHealth literacy, confidence, and health-related internet use. In 2003, Mead et al [33] found that positive outcome expectancy (eg, patients’ belief that it would enable them to better deal with their health), previous use of health websites, and positive self-efficacy (patients’ confidence in their ability to use the technology) were the strongest predictors of patient-reported interest in getting health information from the internet in the United Kingdom [33]. Ten years later, Tennant et al [16] reported that greater eHealth literacy was associated with greater use of Web 2.0 and social media for health information among a sample of Americans.

As shown in other studies of cancer patients’ use of online resources [34], unmet supportive care needs were also associated with health-related internet use. For each unit increase in unmet supportive care needs, the odds ratio of using the internet as a health resource increased by a factor of 1.05. This may suggest that use of the internet by patients with prostate cancer is problem focused, driven by a need to find information to address a specific issue or problem. This finding has important implications for the design and evaluation of health information websites, as it suggests that websites should be designed to address the specific supportive care needs of patients. Respondents in our sample indicated that they want prostate cancer websites to include information on treatment options and side effects and how to manage them, as well as features that help them make treatment decisions and navigate the prostate cancer journey. This type of task-oriented internet use that is motivated by specific needs also suggests that once prostate cancer patients’ needs are met by a Web resource, they may be unlikely to keep using it, unless a new need arises. Therefore, discontinued use of a Web resource may not necessarily be a failing of the site’s design but rather a potential logical reaction to changing needs and circumstances [35]. Hence, website evaluation metrics should focus on assessing whether users’ specific needs have been met.

Problematically, our findings revealed that the majority of prostate cancer patients’ in our sample lacked confidence in using information from the internet to make health decisions related to their prostate cancer. Research suggests that seniors have lower eHealth literacy than their younger counterparts [36]. One study found that seniors felt confident in their ability to use the internet to search for health information, but less confident in their ability to assess the quality of that information [16]. eHealth literacy interventions may help. An intervention study by Xie [37] showed that a 2-week eHealth literacy training program increased seniors knowledge, skills, and eHealth literacy efficacy [37]. At the same time, there is a need to improve the quality of prostate cancer websites. When systematically reviewed by Black and Penson in 2006 [38] and again by Kobes et al in 2018 [39], websites containing information on prostate cancer were found to be lacking in currency, attribution, balance of evidence, and comprehensiveness. In another study, only 3 of 62 websites containing information on prostate cancer treatment options were written below the recommended high school reading level [40]. In yet another study, Genova and Bender [41] found that websites containing information on prostate cancer treatment did not present information in a useful or credible way for patients. Of the 35 websites examined, the average communication quality score was 24 of 50, and less than 50% included content on risk communication, usefulness, and scientific value.

This study has certain limitations. First, as this study was cross-sectional, only association and not causation can be inferred. We aimed to obtain a representative sample of prostate cancer survivors in Canada by recruiting a random sample of patients with prostate cancer from four provincial cancer registries. It is possible that the individuals who chose not to participate in the survey had different characteristics, experiences, and attitudes regarding using the internet as a health resource. It is also possible that patterns of internet use among prostate cancer survivors have increased since then. Nonetheless, our findings are comparable to other studies involving patients with prostate cancer and current Canadian population norms. Our study was also limited by the use of the opt-in methodology used by one provincial cancer registry, which reduced the overall response rate. We encourage researchers to argue in favor of opt-out recruitment methods for future research. We also assessed only one parameter of the validity of the eHealth confidence measure. Further work is needed to establish the validity of this single item as a measure of eHealth confidence. Lastly, we did not ask for respondents’ views on patient portals or personal health records, which are growing in demand. Future research should also explore the internet use trends among French-speaking Canadian patients with prostate cancer and those from diverse ethnic backgrounds.

The internet has the potential to serve as a health information resource for the majority of Canadian men with prostate cancer. In our sample, over two-thirds of patients with prostate cancer used the internet as a health resource. However, one-third of patients with prostate cancer in our sample did not want to use the internet as a health resource, and more than half were not confident in using the internet to make health decisions. Clinicians and educators should not assume that because of their age, education, or income, men with prostate cancer are not interested in using the internet as a health resource. Rather, they should be aware of the importance of being an active information seeker, having confidence in using information from the internet to make health decisions, and having broadband access and unmet supportive care needs as determinants of health-related internet use. These findings also show that men are looking for information on the internet about treatment options, disease progression, and management of side effects, and want websites to include features that help them make treatment decisions and navigate the prostate cancer journey. Future work should examine eHealth literacy interventions as a means to boost men’s confidence in using information from the internet for prostate cancer decision making and design websites that include the information and features that patients with prostate cancer want most.