Online information on pancreatic cancer often above reading level of general population

Online health information regarding pancreatic cancer often overestimated the reading level of the general population and lacked accurate information about alternative therapeutic options, according to study results published in JAMA Surgery.

“As a pancreatic surgeon, I noticed a variable level of knowledge on the complex topic of treatment among my patients, as well as a variable level of research that patients and families have done during the treatment period,” Tara S. Kent, MD, MS, assistant professor of surgery at Harvard Medical School and program director of the general surgery residency at Beth Israel Deaconess Medical Center, told HemOnc Today. “The Internet is varied, so we wanted to see what information was most searchable, whether or not it was accurate, and whether it was written in a way that most people could understand.”

Ninety-nine percent of adults in the United States aged older than 18 years are able to read at a 5th- to 6th-grade level. By contrast, 88% are able to read at a 12th-grade level, and 29% are able to read at the level of a university graduate.

Kent and colleagues sought to compare the readability and veracity of patient-directed online resources for pancreatic cancer, evaluated by treatment modality and website affiliation.

The researchers surveyed 50 websites that highlighted five pancreatic cancer treatment modalities: alternative therapy, chemotherapy, clinical trials, radiation therapy and surgery. They identified and classified website affiliation as academic, government, media, nonprofit or privately owned.

To compare the median readability level of each website, Kent and colleagues employed nine standardized tests, then compared readability scores among treatment modality and affiliation category. Readability levels correspond to education, where a readability level of 5 indicates a 5th grade reading level, 13 indicates a first-year university level, and so on; master’s and professional degrees equate to a readability level of 17 to 18, and a doctorate degree equates to a readability level of 19.

All treatment modalities had median readability levels above the recommended 6th-grade reading level, with readability scores ranging from 3.8 to 19. Median readability scores differed significantly for all five modalities (alternative therapies, 14.2; chemotherapy, 14.5; clinical trials and radiotherapy; 15.2; surgery, 13.7; P = .03).

Websites disseminating surgical information (13.7; interquartile range [IQR], 11.9-15.6) appeared significantly easier to read than those discussing clinical trials (15.2; IQR, 12.8-17) or radiotherapy (15.2; IQR, 13-17; P = .003).

A four-person expert panel comprised of two medical specialists and two surgeons determined the accuracy of the information. They scored website accuracy on a scale of one to five (score of one, accuracy < 25%; two, 26% to 50%; three, 51% to 75%; four, 76% to 99%; five, 100%).

The researchers found that, like readability, accuracy varied widely according to treatment modality and the affiliation of the source website.

Websites disseminating information on alternative therapy had the lowest median accuracy scores (median, 2; IQR, 1-4), with scores significantly lower than websites on chemotherapy (median, 4; IQR, 3-5; P < .001), clinical trials (median, 5; IQR, 4-5; P < .001) and radiotherapy (median, 4; IQR, 4-5; P < .001).

Clinical trial websites had significantly higher accuracy scores than websites on chemotherapy (P = .003) and surgery (median, 4; IQR, 3-4; P < .001).

These results remained significant after the researchers performed a sensitivity analysis to ensure that the opinions of the surgical experts did not come from bias (P < .001).

Websites rated with an accuracy score of five had the highest median readability (16; IQR, 14-17). Readability significantly differed between websites with accuracy ratings of five and those rated at accuracy levels of one (P = .003), three (P = .003) and four (P = .002).

When quantified by affiliation, statistically significant differences in readability occurred among all categories (P < .001).

Websites owned by nonprofit entities appeared more easily readable (median, 12.9; IQR, 11.2-15) than those owned by media outlets (median, 16; IQR, 13.4-17) and academic treatment centers (median, 14.9; IQR, 12.9-17; P < .001 for both).

Media-owned websites were more difficult to understand than privately owned websites (median, 14; IQR, 12.1-16.1; P = .001).

Accuracy also significantly differed by affiliation (P < .001). Websites owned by nonprofit organizations (median, 4; IQR, 4-5), academic centers (median, 4; IQR, 3.5-5) and the government (median, 5; IQR, 4-5) were more accurate than media-owned (median, 4; IQR, 2-4) and privately owned websites (median, 3.5; IQR, 1.5-4; P = .004).

The researchers acknowledged limitations of their study. Readability level did not take into account potential outside factors, such as prior knowledge, motivation to learn, and organization of text in relation to comprehension.

Standard reading comprehension may be further confounded by the unfamiliarity of medical language.

“I don’t think we have collectively done enough in the past to ensure that the information available is understandable for our patients,” Kent said. “As we work to empower patients and their families to have meaningful discussions with doctors, nurses and treatment providers, we should work to limit inaccurate information. It would be helpful to develop a list of websites with the most accurate information that we can give to our patients and encourage them to look there.” – by Cameron Kelsall

For more information:

Tara S. Kent, MD, MS, can be reached at tkent@bidmc.harvard.edu.

Disclosure: The researchers report no relevant financial disclosures.