Too much information

IT'S becoming more common to head to the internet to obtain a diagnosis of a rash or headache. But according to new research from Microsoft, this isn't such a good idea. A Google search of your common symptoms could lead to anxiety, work interruption and cyberchondria.

Cyberchondria is the unfounded escalation of concerns about common health symptoms based on search results found on the web. And it's easy to imagine how it happens. In Australia, about one in every 21,000 people are diagnosed with a brain tumour every year. Yet the probability that someone searching headaches on the web will end up at a site suggesting the possibility of a brain tumour is significantly higher, the Microsoft study says.

"The internet contains a relatively large amount of content on rare and scary diseases versus what's written about more common explanations for symptoms like headaches, muscle twitches and chest pain," explains Eric Horvitz, co-author of the study and principal researcher at Microsoft Research.

"When people put common symptoms in the search engine, in many cases you are just about as likely to bring up the scary things as the common things."

The study surveyed 500 Microsoft employees in the US city of Seattle. It found employees searched the internet for health-related inquiries on average 10 times a month. Most searches (more than 85 per cent) were for benign symptoms such as headaches and stomach aches. And of those interviewed, more than 90 per cent found their medical inquires of common symptoms led to a review of content on serious illnesses.

This escalation of web queries, from googling headache to retrieving information on brain tumours, can cause anxiety and interrupt work and online activities for weeks and months. Almost 60 per cent of those surveyed admitted that the escalation of health-based inquiries interrupted their online activities, and when asked "How often do your queries for serious disease persist over weeks, months or years?" only 27per cent said never.

The study, published in ACM Transactions on Information Systems, also uncovered another problem with people and how they use the web. "With medical diagnosis people really expect to get great information because people get accustomed to the accuracy of information from the internet in other areas," Horvitz explains. "But the web is a retrieval system, not a diagnostic system."

Web search engines rank according to the popularity of a website, its credibility and the number of times a term is used within the site. Unlike an accurate diagnostic measure, search engines do not consider the person's age, genetics or family history.

"For someone under 35 who is otherwise healthy and has no family history of heart disease, there is an exceedingly small chance that their chest pain is a sign of a heart attack. But if you put in chest pain into the web, you will get lots of pages on heart attacks," says Horvitz, who also completed a medicine degree at Stanford University in California.

"People use the internet like it is an oracle," he says. This study reveals that, just like us, search engines are prone to biases in judgment. And when we are self-diagnosing we should keep this in mind.
Story
17/07/2009