Pioneering Researcher at 95 Returns to Lifelong Quest in Drug Interactions
By Crystal Phend, Staff Writer, MedPage Today
Published: December 24, 2008
Reviewed by Zalman S. Agus, MD; Emeritus Professor
University of Pennsylvania School of Medicine.
OAKLAND, Calif., Dec. 24 -- When Morris Collen, M.D., peers into his medicine cabinet at the assisted-living facility apartment where he lives, his thoughts are far from the typical musings of a 95-year-old. He is thinking of adverse drug interactions.
He considers his eight daily prescription drugs, the eyedrops, and the multivitamins he takes, and he wonders how each affects the others, and what this hodgepodge is doing in his body.
"If I threw them into a cup of hot coffee, what would happen?" he asks. "We have no idea. But what I do is take my eight pills and throw them into my stomach where there's hydrochloric acid."
As a founding physician of Kaiser Permanente and pioneer in medical informatics, Dr. Collen has been asking these questions for decades.
Now, as a valedictory to his career, he is helping to resurrect a project Kaiser had to abandon in 1971 because of a recession. Kaiser and IBM are seeking to unravel the impact of adverse reactions to polypharmacy for millions of older Americans -- a massive undertaking -- and a protocol to deal with it.
Dr. Collen is instrumental in a research project to dig through Kaiser's rich medical record database to determine an evidence-based protocol for dealing with polypharmacy interactions. The collaboration with IBM and its major league computer power aims to sort out the possible associations, which rise exponentially with the number of drugs considered together.
Dr. Collen expects to start seeing results in 2009.
Joe Terdiman, M.D., Ph.D., who oversees the project as director of information technology in Kaiser's research division, points out that although the scope of the effort is unprecedented, not every aspect of is new.
"Things that we think are new ideas, Dr. Collen will say, 'Oh, we did that 30 years ago,'" Dr. Terdiman said. Dr. Collen serves as a consultant to the Kaiser division of research he directed before his retirement in 1983.
A Lifetime Effort
The monumental undertaking represents the culmination of Dr. Collen's life work, said Donald A.B. Lindberg, M.D., director of the NIH National Library of Medicine and, like Dr. Collen, a pioneer in applying computer technology to healthcare. Under Dr. Lindberg's direction, MEDLINEplus and PubMed databases were instituted.
In Dr. Collen's case, all clinical, laboratory, and pharmacy records for Kaiser Permanente members feed into the research database developed during the years he led the division of research.
When Dr. Collen graduated in 1934 as an electrical engineer, the Great Depression was in full swing, making an advanced degree more appealing than trying to find employment. His wife, nurse Frances Bobbie Diner, convinced him to try for an M.D rather than a Ph.D.
Then a few years later, his life changed dramatically when Pearl Harbor launched the country into World War II. When bronchial asthma kept him from the draft, Dr. Collen joined the small medical team in the San Francisco Bay-area shipyards of steel-tycoon Henry Kaiser.
When what had been essentially a first-aid unit shifted to full medical care for workers and their families, Dr. Collen became chief of Kaiser's medical service here. Then when Kaiser grew into a multimillion-member health maintenance organization, he became director of the research division, and of technology assessment.
His collaboration there with Kaiser Permanente founder Sidney Garfield, M.D., put him at the forefront of early efforts to incorporate computers into well-patient health screening.
Patients would come in for screening. Data were entered into an automated system using punch cards read by computers.
Physicians go through "pretty much a routine of multiple procedures" for health screenings, Dr. Collen said. "Routine repetitive procedures are ideal for computer applications."
With this data on thousands of patients, Dr. Collen led the research division in pioneering the development of electronic medical records, which they expanded into a research database that by 1970 had reached a million patient records linked across Kaiser centers.
"It became obvious that with the pharmacy data and physician diagnoses entering the computer, we could monitor for drug-drug interactions," Dr. Collen recalled.
FDA gets involved
At that time, the FDA and Congress were particularly interested in adverse drug reactions. The agency's first forays into voluntary event reporting in 1955 expanded into a large-scale reporting system by 1963.
But because of Kaiser's unique database, the agency approached Dr. Collen and colleagues to study rates of adverse events in their population under a contract started in 1966.
Although these studies were terminated in 1971 when an economic recession diminished funding, Dr. Collen's interest in drug interactions persisted.
"Multiple drug reactions are not really studied to the extent that I feel our public deserves," he said. "To me, it's the most important thing we can do since our population is aging."
Dr. Collen said the project with IBM on drug interactions is likely to bring Kaiser back into the spotlight as one of the leading research institutions in the world. "What we're doing now is hard to compare."
Ironically, while it was the multiphasic well-patient screening that gained Kaiser Permanente the world's respect as a research institution, the medical community ultimately dismissed this model of care, Dr. Lindberg said.
The resurrection has been rejuvenating for Dr. Collen personally too.
Once or twice a week, Dr. Collen leaves the office that dominates his small studio at the senior living facility and heads down to the Kaiser research facility for what he calls his survivor therapy.
"When I spend an hour or two with that brilliant team there, I drop off some 40 years," he said. "I feel much younger, exhilarated by the discussions that go on, so I will do this until they carry me out so to speak."
A Unique Opportunity
In an interview with Medpage Today, Dr. Collen cited the commonly quoted statistic that patients in their 80s and 90s take about one prescription per decade of life.
One study in 2002 found that up to 40% of ambulatory adults ages 65 and older take five or more prescriptions. For nursing home residents, the average is six to eight drugs per day. Regimens of up to 24 medications have been reported for older adults.
But studies into interactions among multiple agents have been limited. Pharmaceutical companies typically evaluate only their own drugs, with little incentive to monitor adverse events or interactions with agents of competitors. The FDA, he added, has a wealth of data from volunteer reporting on adverse reactions. But the agency has little data on the population at risk to determine rates of events.
European countries with universal healthcare systems have the denominators to be able to do these kinds of studies but by and large haven't had the expertise or computing power to do so, Dr. Collen said.
In the U.S., some centers have attempted to mine their databases for this kind of information. Roughly half a dozen institutions including Brigham and Women's Hospital in Boston have developed programs to provide computer-based alerts to physicians or determine interactions.
The major barrier is the setting, said Dr. Lindberg. "What's important is to have a healthcare setting where the institution agrees on the importance of a study and the patients agree it is permitted," he said. "Then, of course, you have got to keep it up for four or five years minimum or you don't have long-term follow-up in a large number of patients that give it importance."
This kind of longitudinal data is a unique feature of Kaiser Permanente's records, said Marion J. Ball, Ed.D., of the IBM Center for Healthcare Management in Baltimore and professor emeritus at Johns Hopkins, who helped broker the collaboration.
"They have data on the same patients for 20 to 30 years," she said. "Most of us jump from one healthcare provider to another so we don't have the continuity of the collection of data, which is available at Kaiser."
Dr. Collen agreed that the size of the Kaiser Permanente patient database is an advantage. With about 8.5 million members, it is the world's largest civilian electronic health record and spans decades.
However, without IBM even they would not have been able to undertake such a project, he said.
"Up to now," Dr. Ball said, "we haven't had tools to be able to do such a mass data mining exercise, which takes an enormous amount of computer power, enormous knowledge of mathematics, and modeling techniques."
Mining the Data
With such a large dataset, searching for meaningful associations without a preset hypothesis requires a different approach than typically used in medicine, said Bill Cody, Ph.D., who is leading the project at IBM's Almaden Research Center in San Jose, Calif.
IBM's algorithms and techniques scour through databases taking all comers as statistical associations, whereas studies of adverse events tend to be smaller scale investigations looking for a specific event with a specific drug.
Combining these two statistical philosophies is fairly novel and technically challenging, Dr. Cody said.
Their data-mining tools have typically been applied in retail and financial settings, such as to determine products consumers often purchase together so corporations can optimize their store layout, he said.
The first attempt at healthcare applications was a broad look at the Virginia Commonwealth University Health System's clinical database in 2004, which demonstrated feasibility by finding relationships such as co-occurrence of paralysis, peptic ulcer conditions, and rheumatoid arthritis.
Since then, the Kaiser-IBM group has been refining the statistical tools for use on drug interactions by running them on a subset of the de-identified clinical, pharmacy, and laboratory data from the roughly 3.5 million-member Northern California cohort within the Kaiser Permanente databases.
Once the decision rules have been beefed up enough to keep from generating more potential associations that anyone could confirm as potentially interesting, the project will start generating results and expand to the national cohort, Dr. Cody said.
The first application will be to notify the FDA of potential true positives for validation, Dr. Collen said. "It probably is going to take five or more years to get answers that can be used in patient care."
"You need a large healthcare system in order to observe rare complications or observe interactions between the drugs," Dr. Lindberg said. "Collen saw all this 30 to 35 years ago and tried to do it."
© 2004-2009 MedPage Today, LLC.
http://www.medpagetoday.com/tbindex.cfm?tbid=12170&pfc=101&spc=222