Various studies have estimated that between 1 million and 2 million people in the U.S. are hospitalized each year due to drug side effects. As personalized medicine research continues to look into the effectiveness of tailored therapies, it should also aim to identify which patients are most apt to suffer dangerous side effects from treatment.
Some progressive organizations are taking us in this direction already. For instance, when the Cleveland Clinic first collaborated on an IBM translational medicine platform several years ago to help study the correlation of treatments and outcomes of specific patients, researchers at the clinic also predicted that those kinds of efforts could help identify patients most at risk for some adverse drug events.
[ The more detailed the EHR, the fewer the side effects? See HIMSS Survey: Advanced EMRs Make A Difference. ]
For instance, pattern recognition and data mining could have helped discover the correlation between heart attacks, strokes, and use of the prescription arthritis drug Vioxx, which was pulled from the market in 2004 by Merck & Co.
Also, it should be noted that many adverse drug reactions that occur nationally each year happen in hospitals, often because of mistakes in dosing, or prescribing clinicians overlooking drug interactions or patient allergies. In the famous Institute of Medicine 1999 study, To Err is Human, it was estimated that between 44,000 and 98,000 patients die each year in hospitals due to preventable errors.
Health information technologies, such as medication bar coding, computerized physician order entry, and e-medical record systems, all are aimed at reducing those sorts of preventable errors. But new and better insights about drug side effects learned through personalized medicine research that involves data mining and analysis could also be useful in hospital settings.
Just last week, IBM announced that a cancer research and treatment center in Italy, the Fondazione IRCCS Istituto Nazionale dei Tumori, is testing Clinical Genomics, IBM's new decision support tools designed to help physicians personalize treatments based on automated interpretation of pathology guidelines and intelligence from past clinical cases documented by the hospital.
Those tools also are helping AIDS researchers in Europe analyze genomic and clinical data "for making better decisions about the drug cocktails most effective for treating HIV patients," said Chalapathy Neti, director of global healthcare transformation at IBM Research in an interview with InformationWeek Healthcare.
Although IBM isn't ready to talk about any data analysis initiatives underway to help identify and prevent potential cases of adverse drug reactions, researchers using IBM's analysis tools have found that some geriatric patients with certain diseases sometimes do better without treatment. IBM's research team in Israel, for instance, found that doctors are less inclined to order chemotherapy for older patients because the side effects can be difficult for them.
Other findings might help younger patients avoid treatments that could cause serious complications. About a month ago, a seemingly healthy young relative of mine had a serious stroke soon after starting medication for a fairly benign condition. The poor girl--she's only 18--faces a lot of rehab and might not fully recover. Increased risk for clotting and strokes is one of the many side effects of the drug she was taking.
Lists of side effects might be more relevant to some people than others, and knowing more about the pros and cons of a medical treatment before it's prescribed could have life-long implications for many patients. The FDA's Adverse Event Reporting System (AERS) helps the agency with post-market drug safety surveillance by collecting information voluntarily reported by healthcare providers and consumers. When drug companies receive adverse drug reports, it's mandatory that they report them to the FDA.
The FDA also posts quarterly reports so consumers and clinicians can learn about what adverse events and potential new risks are being reported to and identified by AERS. But that kind of investigation by patients and clinicians takes some digging around on FDA's website, so it's not exactly a bedside decision-support tool, and the information isn't exactly real-time, either.
Social media, including sites such as PatientsLikeMe, can also be helpful for consumers sharing experiences about their bad--and good--outcomes related to medical treatments. The problem is that unless patient-reported information on social media sites is somehow monitored, vetted for accuracy, and analyzed by medical experts or researchers, potentially relevant correlations between different treatments, side effects, and patient types aren't likely to be "officially" discovered or disseminated to the clinicians prescribing those treatments. It's widely acknowledged that even formal evidence-based research can take years to reach and be embraced by doctors.
Technologies being deployed for personalized medicine efforts today, such as clinical and genomic data analysis and decision-support tools, will likely help more patients get more individualized and effective treatments for the diseases that ail them. But I'm hoping those same tools, perhaps combined with sentiment-analysis software to glean insight into social media postings, can be used to improve post-market surveillance of drugs and help doctors and patients make safer treatment decisions. Then doctors can tailor their treatments to sick patients--and avoid the unintentional harm that comes from bad drug reactions, whether their patients are 80 or 18.
The 2012 InformationWeek Healthcare IT Priorities Survey finds that grabbing federal incentive dollars and meeting pay-for-performance mandates are the top issues facing IT execs. Find out more in the new, all-digital Time To Deliver issue of InformationWeek Healthcare. (Free registration required.)