Tuesday, April 13, 2010

If Patients Know Best, then Patient Social Networks Can Help Capture and Report AEs

A recent article published in the New York Times caught my attention. The article -- "In Reporting Symptoms, Don’t Patients Know Best?" (find it here) -- suggested that doctors tend to "downgrade symptoms" (ie, adverse effects of drugs) reported to them by their patients because of "wishful thinking" or, worse, because they do not want to document problems they failed to act on for fear of being sued.

That being the case, there is little hope that widespread use of electronic medical records will capture useful adverse events that the FDA can analyze. Garbage in, garbage out as they say.

Speaking of garbage, the article criticized FDA's MedWatch program that allows doctors and patients to report problems that they think are adverse events from drugs already on the market. "But it’s a passive system that waits for reports instead of actively surveying patients," says the article. "Many people don’t know about it, and it has failed to catch some important adverse events," said Dr. Ethan Basch, an oncologist who treats men with prostate cancer and does research at Memorial Sloan-Kettering Cancer Center in New York.

"A better approach," says Basch, "would be to have large numbers of patients filling out questionnaires before and after drugs are marketed. For example, in the postmarket setting we could ask 5,000 selected patients starting Bextra to report monthly."

Of course, Bextra has been taken off the market by the FDA. Dr. Bach's point is that if post marketing surveys of patients were done soon after Bextra entered the market, the problems may have been noticed much earlier.

Meanwhile, Dr. Basch is working on his idea: "Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events" is National Cancer Institute funded outcomes Research project that began in October 2008 (see here). As part of that research, Dr. Basch and his co-investigators initiated development of an electronic patient-reported system for monitoring and reporting symptomatic AEs that patients may experience during treatment.

I imagine that this system would be pretty expensive to implement on a large scale across many different drug treatments. But there is a cheap (ie, FREE) electronic system already available for capturing adverse events from patients: patient-focused social media sites such as PatientsLikeMe (PLM) and others.

PLM found that 7% of 500 randomly selected posts from the 364,000 posts contributed by patients within the PatientsLikeMe Forum during 2009 incorporated all four elements required for reporting an adverse event (an identifiable patient, a specific medication, an identifiable reporter and a reaction). This is 35 times the rate in the frequently cited Nielson study, which found only one such message in its random sample of 500 posts. (See "PatientsLikeMe Reports High Rate of Adverse Event Reporting Among Its Members".)

I'm not sure how Dr. Basch is getting his questionnaire out to patients, but he might look to social media sites like PatientsLikeMe to perform a public service and place the form on its site so that patients can easily access it. Just a thought on how social media can help expedite outcomes research!

NOTE: I will be speaking on this topic next week at the World Drug Safety Congress. The title of my presentation is "Web based AE reporting; is it the future?" I will specifically address how social media can be incorporated into a drug’s safety strategy and the challenges involved. If you have any solutions, ideas, or comments I can incorporate into my presentation, please tell me about them (you can email me at editor@news.pharma-mkting.com).


  1. John,

    one more element in our VERY old conversation about the misunderstood power of patient communities!

    I can't believe we said the same things 13 years ago, long before anybody heard of PLM or social media.Almost nothing has changed since then with one notable exception: the amount of patients communicating through social media has grown exponentially.

    It is high time for the FDA to stop acting like if this is a "business as usual" period. And it is high time for us to develop the tools to make sense of the invaluable narrative streams generated from all corners of the Internet.

  2. John:

    This is something we have started working on - looking for patterns in the data - independent of whether or not they meet the FDA's 4 criteria.

    The data, by the way, is unstructured text from the web. The vast majority of this conversation is on condition-centric forums - few of which are Pharma owned properties.

    We are analyzing millions of patient & caregiver posts - a subset of which (around 10%) mention any drug, a subset of which could be considered AE's.

    My guess is that one HUGE difference in the Nielsen and PLM numbers is one of the 4 AE criteria - identifiable patient/reporter. PLM requires somewhat "full" registration which means they can always check the box on this criteria. This is almost never true in most other condition based communities where all we have to go on is highly ambiguous user names/handles.

    I think smart Pharma companies will use this source of feedback as an early signal detection mechanism to avoid huge recalls/issues.

    (Ford has successfully used similar techniques for predictive recall/warranty work - saving billions.)

    Tom O'Brien
    Motivequest LLC

  3. Tom,

    Thanks for your comment and insight about PLM v Nielsen. How can I find out more about your work in this area?


Related Posts Plugin for WordPress, Blogger...