br Usefulness assessments were completed
Usefulness assessments were completed using a 33-item checklist of cancer caregivers’ unmet supportive care needs based on a systematic review by Lambert et al. (2012) . The six domains evaluated include: comprehensive cancer care, emotional and psychological, caregiver impact and daily activities, relationship, information, and spirituality. Items were rated from = unmet need not addressed, to 1 = unmet
3.1 Description of Resources Seven hundred online resources were retrieved from the Google Search. Resources were excluded if they E-64-c (1) were redundant (i.e. retrieved by the Google search using different search terms or locale) (n=462); (2) did not address the caregiver explicitly (n=54); (3) were not cancer specific (n=1); (4) were in the form of news articles, online magazines, questionnaires, blogs, testimonials, videos, podcasts, books, and journal articles (n=107); (5) were mainly advertisements (n=33); (6) were for children and youth with cancer (n=4); (7) were about end-of-life care or palliative care (n=8); and (8) required a password or payment for access (n=6). These criteria excluded 675 resources of the original list of resources, leaving a final sample of 25 resources. The Environmental Scan yielded 125 online resources. Resources (n=95) were excluded if they (1) were redundant (n=8); (2) did not address the caregiver explicitly (n=8); (3) were not cancer specific (n=4); (4) were in the form of news articles, online magazines, questionnaires, blogs, testimonials, videos, podcasts, and journal articles (n=42); (5) were mainly advertisement (n=3); (6) were only available in French (n=17); (7) were for children and youth with cancer (n=2); (8) were about end-of-life care, palliative care, or grief (n=5); (9) required a password or payment for access (n=1); (10) were dead links (n=5). This left a final sample of 30 resources.
2.4 Inter-Rater Reliability
Descriptive statistics were used to summarize the SAM, DISCERN, and unmet needs assessment item and total scores. The total scores were compared and tested across the following website characteristics using T- test : type of cancer resource, resource location, resource host, resource funding, year last updated, presence of advertisements, presence of audiovisual aids, MANUSCRIPTandresourceformat.oidentifygroupsof‘high’ scoring websites on the SAM and DISCERN, hierarchical cluster analysis was performed [16, 17]. The clusters identified were compared with each website characteristics; Chi-square test and Fisher exact test was performed to identify the significant associations at alpha 0.05. The SAM and the DISCERN were included in the cluster analysis because they are validated assessment measures. In this study, readability was assessed as part of the SAM, and was therefore not independently included in the cluster analysis. The unmet needs assessment was not included in the cluster analysis as it is not a validated assessment measure. Inter-rater reliability was established by having two raters independently rate the resources until an Intraclass Correlation Coefficient (ICC) of 0.70 was achieved. total of 26 resources were evaluated to achieve this cut-off. Caregivers were not involved in the evaluation of resources.
need addressed, with higher ratings reflecting more detailed descriptions of unmet care needs. A total score was obtained by adding the rating given to each 33 unmet needs (maximum possible total score = 33) and converted to a percentage. 2.3 Data Analysis
Of the 55 resources (see supplementary material, Appendix B, Table B.1. for list of all websites evaluated) included in the final sample, most resources were from the United States (n=34), followed by Australia (n=7), Canada (n=5) and the United Kingdom (n=5), Ireland (n=2), and New Zealand (n=1). The majority of the resources were published by national and community-based organizations (n=42), followed by major cancer centers (n=8), and accredited universities (n=5). Fifty-two (95%) resources did not contain audiovisual aids and 48 (87%) resources did not contain advertisements. Finally, 30 (55%) resources were in the form of webpages, 21 (38%) were in the form of electronic booklets, 3 (5%) were in the form of fact sheets, and 1 (2%) was an e-book.
3.2 Assessment Measures
The mean total SAM score for the sample was within the “adequate” range at 48.1% (SD 8.6), with (0%) resources in the “superior” range, 48 (87%) in the “adequate” range, and 7 (13%) resources were deemed “not suitable”. Of the 6 factors assessed by the SAM, layout and typography (mean=1.4, SD=0.3), content (mean=1.3, SD=0.2), literacy demand (mean=1.3, SD=0.2) and cultural appropriateness (mean=1.2, SD=0.3) rated the highest, whereas graphics (mean=0.3, SD=0.3) and learning stimulation and motivation (mean=0.4, SD=0.3) rated the lowest (see Table 1).The highest suitability scores were for items related to cultural appropriateness, information scope, headers and topic captions, and writing style (see Table 1). The lowest suitability scores were for items related to reading grade level, type and relevance of illustrations, as well as the presence of lists, tables, graphs, and charts with explanations and captions (see Table 1). Resources in the format of electronic booklets, fact sheets and e-books scored higher (n=25, mean=51.9, SD=8.9) than resources in the webpage format (n=30, mean=44.9, S =7.0, t-test p=0.002).