Google search results feed misinformation on cancer treatments

February 26, 2021
Google search might not always help. (AP Photo/Mark Lennihan)

Google search might not always help. (AP Photo/Mark Lennihan)

In a Google search for cancer supplement advice, only about one-quarter of the results considered relevant by Google algorithms were vetted by a team of nutrition professionals as actually containing high-quality information, suggesting that people using the search engine may be facing a flood of misinformation.

Using a Google search for “dietary supplements for cancer,” researchers from Tufts University applied their novel Health Information Quality Index, a quality-rating tool consisting of 12 objective criteria to score websites that appear in search results on their provided health and nutrition information, in a case study published Feb. 5 in Current Developments in Nutrition. They found that about 75% of the “relevant” results in the search scored as moderate, poor or low quality on their index.

People frequently turn to the internet to self-diagnose and treat their illnesses, ranging from common cold remedies to complex supplemental cancer treatments. But taking health care advice from a commercial business like a search engine is inadvisable, the authors said, because it is not primarily designed to be an authoritative and unbiased source of health-related information. 

In the U.S., vulnerable populations such as low-income individuals and immigrants are particularly at risk of being targets of health care misinformation, Hannah Cai, a graduate student researcher at Tufts University and lead author of the paper, told The Academic Times. Cai previously worked with the Immigrant Health and Cancer Disparities Department at Memorial Sloan-Kettering Cancer Center in New York City.

“The unequal access to health care and food witnessed among cancer patients left lasting impressions. Given language barriers, different health literacy levels and disincentives to seek health care, immigrant populations are at high risk of being misinformed or disinformed by poor quality online resources,” Cai said of the current study.

Search engines function by using proprietary algorithms that are meant to push relevant results to the top of the page, but according to the current study’s findings, “relevant results” are not likely to represent the most well-informed resources on the given search subject. And relying on automated algorithms to regulate information quality is not a substitute for good clinical judgment, the authors said.

“To our disappointment, most of the information was of mediocre quality, and there was a high volume of distracting advertisements,” Cai said. “The commercial nature of Google results contradicts the common expectation that the search engine functions solely for informational purposes.”

Google claims that its algorithms adhere to a set of internal standards for judging the expertise, authoritativeness and trustworthiness of website attributes, the authors explained, particularly for sites that have a potential impact on the happiness, health, financial stability or safety of users. These sites may give advice or sell products related to diet, nutrition and medical devices, for example.

The researchers conducted a Google search in May 2020 on “dietary supplements for cancer” and tested the resulting web pages using their Health Information Quality Index. They hypothesized that results scoring highest in quality on the index would emerge first and be most prominent on Google’s search engine results page.

This search resulted in “an uncurated mass of information spread across dozens of pages in no particular order, including some seemingly authoritative advice combined with press releases, advertisements, anecdotal reports and false health-related claims regarding disease prevention and cures,” the authors said in the paper.

“These are likely to contribute to misinformation and confusion that lead to patient skepticism of or nonadherence to conventional medical therapies, with possible adverse influences on prognosis,” they continued.

Out of 299 million possible results for the search, Google displayed 187 web pages that it deemed relevant. After disregarding duplicates, the researcher analyzed and scored 160 sites. In addition to those results, Google displayed 496 commercial advertisements that were also disregarded for the study. 

The Health Information Quality Index developed by the researchers rates the search results with a score ranging from 0, meaning lowest quality, to 12, meaning highest quality. The criteria are characteristics that can be objectively assessed from the search result, including website domain, authoritative nature of the health and nutrition information, evidence of commercial sponsorship, commissioned links, apparent conflict of interest, authorship and timeliness of information.

Eleven out of the 160 results scored a perfect 12, seven of which were reprints of peer-reviewed publications. Just one web page, an Amazon listing for a book about following a cancer-relevant diet, scored a 0. 

“While peer-reviewed papers are valuable resources for health care professionals, they are not accessible to a lay audience seeking practical advice. There is a need to improve public access to and trust in clinicians through online platforms,” Cai said.

Nine percent of the results scored in the 0-3 range, categorized as poor quality; 31% scored between 4-6, as low quality; 34% scored between 7-9, as moderate quality; and 26% were considered the highest quality, with scores between 10-12. The highest-quality results were also not found to be displayed first or near the top of the results.

The authors noted that many of the 34% of webpages of moderate quality were published by major health care institutions or well-known sites, including the Mayo Clinic, the American Cancer Society, Cancer Treatment Centers of America, Healthline, MedScape and WebMD. 

“The major shortcoming of low subscores on the authoritative nature of the health and nutrition information was lack of documentation,” the authors said. “Over half of the search results failed to disclose sources and references adequately.”

A 2018 article in The Atlantic coined the term “misinfodemic” to describe how the spread of health care misinformation online can lead to the spread of diseases in real life, citing digital anti-vaxx groups that caused the return of measles to the U.S. and social media-based rumors that added to the Ebola virus death toll in West Africa. 

In the context of COVID-19, the threat of damaging health information spreading online has become a reality, and the role of online information merits more urgent attention, Cai said. She pointed to a civil antitrust lawsuit that the U.S. Department of Justice filed against Google in late 2020, alleging that Google has unlawfully maintained monopolies in search and search advertising to the detriment of competition and consumers.

Cai and the research team recommended further evaluations of the quality of web-based information on medical and nutritional topics. Future research should explore if Google results vary depending on whether a search is completed using typed text or voice search, they said. 

“Even a ‘perfect’ score on the [index] did not guarantee that the information presented was accessible and understandable by an audience of varying health literacy levels,” the authors said. “Comprehension should be considered when evaluating online information in the future.”

The study, “Using the Google Search Engine for Health Information: Is There a Problem? Case Study: Supplements for Cancer,” was published Feb. 3 in the Current Developments in Nutrition journal. Hannah Cai, a master's candidate at the Friedman School of Nutrition Science and Policy at Tufts University and a dietetic intern at Tufts Medical Center’s Frances Stern Nutrition Center, was the lead author. Johanna Dwyer, of Tufts University, and Leanne King, of Cone Health, served as co-authors.

Saving
We use cookies to improve your experience on our site and to show you relevant advertising.