ComScore’s qSearch data were used to identify the five highest-ranked search engines at the time of data collection (October, 2007). ComScore is a global Internet information provider that maintains databases on real-time use of the Internet and consumer behavior in the Internet, and its qSearch tool measures all search activity on the Internet, including major search engines, private sites such as MySpace, vertical searches on sites such as Amazon or eBay, local searches for maps or directions, cross-channel searches such as searching the Web, maps and images for the same term, and more [42]. It is a comprehensive tool measuring the search universe, and it has been applied previously in scientific research [43]. The five top-ranked search engines were Google, Yahoo, Microsoft Sites (including MSN/Windows Live), Time Warner Networks (including AOL), and Ask Networks (including Ask.com) [42].

A subgroup of the study team generated a list of search terms that a layperson might use to search for online cancer survivor stories. This group included four family members of cancer patients, representatives of three different racial or ethnic groups, and an information science specialist with expertise in Web searching. The list took into consideration synonyms and word variations. A pilot search with each term was conducted, and the final six search phrases were selected because they yielded the largest number of relevant hits. The six search phrases used in the study were the following: “cancer survivor stor,” “cancer stor,” “cancer patient stor,” “cancer testimonial,” “sharing cancer stor,” and “cancer experience.” Boolean operator “OR” was used to perform the union of the six phrases. For cancers not identified by the word “cancer,” such as leukemia, lymphoma, and melanoma, we replaced the word “cancer” in the six search phrases with each of these, resulting in 18 total search phrases for online stories of survivors diagnosed with leukemia, lymphoma, and melanoma. We chose these three cancers because they may have been missed in our search and are also prevalent in the United States [44]. While searching “stor” in Google, AOL, and Ask.com can yield results with both “story” and “stories,” Yahoo and MSN/Windows Live do not accept the truncation search. Therefore, the words “story” and “stories” replaced “stor” in these search engines and resulted in doubling the total number of search phrases for these search engines. Summing all permutations of the original search terms, Google, AOL, and Ask.com had 24 distinct search phrases, while Yahoo and MSN/Windows Live had 48.

The search was performed from October 15 to October 30, 2007. Excluding sponsored links, the URLs of all websites were recorded until duplicates or irrelevant results dominated the search hits list, which resulted in 1420 websites identified from Google, 1055 from Yahoo, 1039 from MSN/Windows Live, 1055 from AOL, and 1053 from Ask.com. After removing exact duplicates, 3738 distinct websites were identified.

To be included in the study, a website had to (1) contain cancer survivor stories in text, audio, or video form with an accompanying photo or video of the survivor; and (2) identify (or imply) presentation and/or sharing of stories as a purpose of the site. This latter criterion excluded personal blogs, news stories, and websites ending in a “PDF” extension, as these were usually reports. Websites hosted outside the United States were also excluded due to differences in cancer prevalence, racial/ethnic composition of the population, and use of different search engines. A total of 106 websites met these eligibility criteria.

While many of the websites included stories without photos or video, we coded only those stories accompanied by a photo or video in which the survivor or storyteller was represented. Uncoded types of stories included those with text only, links to personal blogs, and stories entered on forums. On the 106 websites, there were 7995 stories total, of which 1670 (20.9%) had an accompanying photo or video image of the survivor.

Characteristics of websites and survivor stories were coded over a 2-month period from November 2007 to January 2008. Members of the research team who received formal training, rehearsal, and evaluative feedback completed all coding, adhering to specific operational definitions and coding instructions. The research team coded every cancer survivor story. In rare instances when a member of the research team had difficulty coding race or ethnicity, another trained coder was consulted and consensus reached.

Descriptive statistics are provided to characterize survivor stories and websites. All stories and websites are included in the analysis.

Characteristics of websites are summarized in Table 1. Of the 106 websites included in the final sample, 56 (52.8%) were hosted by nonprofit organizations (“.org”), 7 (6.6%) by educational institutions (“.edu”), and the remaining by for-profit companies, with Web extensions including “.com,” “.html,” and “.net.” Sixty-four of the 106 websites addressed cancer in general (60.4%), while the remainder focused on a specific type of cancer (n = 34, 32.1%) or a set of related cancers, such as leukemia and lymphoma or brain and other nervous system cancers (n = 8, 7.5%). The number of stories per website ranged from 1 to 232, with a mean of 16.5 (SD 32.2).

Most stories were told by survivors (n = 1052, 63%), with the remainder told by a third person narrator (n = 390, 23.4%) or a family member or friend (n = 206, 12.3%). Most stories were told by women (n = 1073, 64.3%). When age was reported (n = 1008, 60.4% of stories), the mean age at diagnosis was 35 years. Characteristics of stories are summarized in Table 2.

Stories could be presented in more than one form. Nearly all stories were told through written text (n = 1643, 99.2%), with some told through audio (n = 337, 20.2%) and/or video (n = 264, 15.8%). Most stories were accompanied by a still photo (n = 1643, 98.4%), and half of these photos (n = 936, 56.0%) were professional grade.

One in four stories (n = 440, 26.3%) addressed breast cancer, followed by leukemia (n = 282, 16.9%), lymphoma (n = 165, 9.9%), prostate cancer (n = 142, 8.5%), and skin cancer, including melanoma, basal cell carcinoma, and squamous cell carcinoma (n = 69, 4.1%). Among stories that reported cancer stage at diagnosis (n = 318, 19.0%), later stages were most represented, with stage 4 diagnoses being most common (n = 88/318, 27.7%), followed by stage 3 (n = 85/318, 26.7%), stage 1 (n = 82/318, 25.8%), stage 2 (n = 57/318, 17.9%), and stage 0, (n = 6/318, 1.9%).

A large majority of stories in our sample (n = 1503, 90%) were told by whites. Among minority groups, blacks or African Americans were represented in 5.8% of stories (n = 97), Asians (including Native Hawaiian and Other Pacific Islanders) in 3.7% of stories (n = 62), and American Indian and Alaskan Natives in 0.3% of stories (n = 5). Race could not be determined in 0.2% of stories (n = 3). For ethnicity, most stories were from non-Hispanic or non-Latino survivors (n = 1524, 91.3%); 4.1% of survivors were identified as Hispanic or Latino (n = 68). In 4.7% of stories, the survivor’s ethnicity could not be determined (n = 78).

Most story characteristics did not differ across race or ethnicity. There was a significant difference in cancer type between races represented; however, with 33.3% of cells having counts less than five, the test statistic may not be valid.

As this discussion has already identified, there are limitations to the study. Cancer survivor stories that did not include pictures of the survivor were not part of the sample. While this was necessary to achieve the study aim (ie, to identify race and ethnicity of survivors whose stories are available online), it’s possible that stories with and without pictures varied in other ways not intended in the study. If minority survivors were less likely than other survivors to provide a picture with their story, the study findings would underestimate the proportion of such stories. We have no indication whether survivor stories with and without pictures varied systematically by race or ethnicity. However, from a practical standpoint, unless story content explicitly mentioned the survivor’s race or ethnicity, this information would not be available to an online information seeker who might value it. Thus, while any real differences might be of interest for academic purposes, they would be largely irrelevant to those consuming the stories.

We also acknowledge that making judgments of a survivor’s race and ethnicity from online photos was sometimes challenging and, like any coding, subject to misclassification. In cases when multiple coders could not determine or agree upon a survivor’s race or ethnicity, it was classified as such. But because there were so few cases where race and ethnicity could not be determined (3/1670, 0.2% for race; 78/1670, 4.7% for ethnicity), it is improbable that misclassification bias alone would account for the pattern of findings in the study, even if every survivor whose race or ethnicity could not be classified was in fact a minority group member. Finally, it is possible, but we think highly unlikely, that stories from minority survivors exist in proportionally greater numbers under different search terms than those used in the study.

Thirty-two percent of the stories (534/1670) were hosted by four large not-for-profit organizations promoting patient advocacy and research. Each of these organizations has a public face that may draw more survivors from a variety of racial/ethnic backgrounds and age groups. Another 14.7% of stories (246/1670) were on websites of prominent cancer research and treatment centers. Some websites and organizations are doing a better job than others in recruiting minority cancer survivors to share their stories of survival. Organizations providing this service can learn from the websites that have collected libraries of stories from diverse populations. We also recognize that three of the four primary cancers represented in these stories (breast, leukemia, and lymphoma) do not occur disproportionately in minorities, and, therefore, it may not be surprising that we did not find a larger proportion of stories from minorities.

Besides increasing the proportion of survivor stories from minorities, older adults, and men, the study findings also suggest that websites providing cancer survivor stories might enhance their offering in at least three other ways. First, our research team learned that finding survivor stories online was often challenging and time consuming, requiring study team members to search through a lot of other cancer content to find stories. Stories were seldom available from a single location on a website, and the location of stories varied greatly from site to site.

Second, we observed that few sites (11/106, 10.4%) offered users a means of searching available cancer survivor stories, and less than half (43/106, 40.6%) allowed users to share or post a story.

Third, although we found stories from survivors of a wide range of cancers, there were clear gaps in the distribution of cancers represented. For example, only 58/1670 (3.5%) of stories were from colon cancer survivors, despite colon cancer being the third leading cause of cancer death in United States [54]. In addition, stories from lung and bronchus cancer survivors, the leading cause of cancer death in the United States for both men and women, represented only 1.6% of stories in this sample (n = 27). It may be that those with lung and bronchus cancer do not live very long and therefore don’t contribute to survivor stories on the Internet. We did find that when stage at diagnosis was reported, most of the survivors (173 of 318, 54.4%) were diagnosed with stage 3 or 4 cancers. Websites that select stories to post may choose stories from long-term survivors and from survivors who have overcome greater odds. Making improvements in these areas would enhance the accessibility and benefit of stories to users, as would developing technology tools that facilitate story sharing.

The study also raises new questions and identifies promising avenues for future inquiry. An important next step may be determining the relative importance of technological (eg, online access, digital camera ownership, computer skills), psychological (eg, interest and willingness), and organizational (eg, website policies) factors in explaining the dearth of online survivor stories from minorities, older adults, and men. Intervention and audience research studies among cancer survivors might explore strategies to raise awareness of, interest in, and motivation to share online stories. Such work would be especially valuable when conducted among groups that were underrepresented in the online collections of stories in the current study (eg, racial and ethnic minorities, older adults, men, those with certain types of cancer). Usability research could test alternative Web designs and functionality to optimize ease and efficiency of use for sharing and accessing survivor stories.

We know from previous research that when characteristics of survivors match those of the reader or viewer, the information provided by the survivor will be more engaging, enhance information recall, and stimulate more thoughts about the story [21]. While these are important communication outcomes, they are several steps removed from the kinds of prevention, screening, or treatment adherence behaviors that would actually reduce cancer disparities. Still, matching the race/ethnicity of survivor stories to a viewer would seem to be an important and achievable first step toward these higher order outcomes.

Although there are currently no published studies reporting numbers of minority survivors seeking online survivor stories, there is considerable evidence that minority groups want and need better health information and often turn to the Internet to find it. The Internet is an important source of health information for African Americans and other minority groups, including patients diagnosed with cancer [55,56,58,59]. Yet cancer survivors as a group—including African Americans and Hispanics—are more likely than healthy adults to report wanting more information, having difficulty finding desired information, feeling frustrated during their search for information, and finding the information too hard to understand [57]. Because of differences in access to and use of the Internet in minority populations, changing the mix of stories alone won’t be enough to make an impact on cancer disparities. Changes in policy that would improve access to the Internet are also needed in addition to a better mix of survivor stories.

This study provides the first descriptive summary of online cancer survivor stories and identifies some important gaps in currently available offerings. There is a risk that the benefits these stories can confer to users might be unequally distributed across the population due to a lack of stories from members of certain groups. The fact that several of these underrepresented groups also bear a disproportionate burden of cancer suggests that the collection of survivor stories available online today is unlikely to help eliminate disparities in cancer.