The reliability of preclinical research, or lack thereof, has received a lot of attention recently as one of the challenges of the biomedical research enterprise. But clinical-stage research, unfortunately, suffers from rather massive problems of its own. (See BioWorld Today, April 1, 2012.)

Speaking at the National Institutes of Health recently, Robert Califf shared a sobering assessment with his audience about the U.S. health care system, how it gathers evidence and what that means for Americans.

“Most of our decisions are not supported by good evidence,” he told the audience. Califf is a cardiologist at the Duke Translational Research Institute, and said that even in his own discipline, which is often considered the kingpin of evidence-based medicine, “85 percent of our recommendations are not based on high-quality evidence.”

And that lack of evidence has serious consequences for public health.

“Our country has poor health status, and it’s getting worse compared to our economically advanced competitors,” Califf said.

Improvements in health outcomes over time, such as increased life expectancy, mask the fact that “when we compare ourselves to all the other countries, the curves are diverging – we’re getting better at a slower rate than other economically developed countries.”

His solution? Be like Walmart. Walmart is cheap, of course, and the U.S. health care system is not. But that’s not what Califf was getting at.

When you buy something at Walmart, Google or Amazon “everything you do is tracked though barcoding and pretty quickly loaded into an electronic warehouse, where it’s analyzed . . . and the experience that you have can be used to help the next person that comes along.”

“I used to sort of chafe when people said we should be as good as they are,” he added. “But I actually think it’s not all that different.”

He added that now that most people have electronic medical records of some sort, the barriers to changing the health care system into what he termed a learning-based system are cultural more than technological.

In his view, one key problem is that data are not shared readily enough.

Reasons include concerns over privacy, even though he showed data indicating that the majority of patients would be willing to share their medical data, if it were done anonymously. Another problem is that while data in international clinical trials is shared readily, in the U.S., “possession of the data” – even if the data will be made public later – “is 95 percent of getting famous.”

Another is that each clinical problem is addressed in isolation, often due to funding constraints. Ironically, those constraints in turn stem in part from how expensive it is to start each clinical trial from scratch.

At his talk, Califf presented the Patient-Centered Research Outcomes institute (PCORI) and its experimental arm, the Collaboratory, as new institutions that could address all of those problems.

Funded as part of the Affordable Care Act, PCORI is focused on patient-centered comparable effectiveness research. Part of its goal is to establish a network of clinical trial centers whose patients agree to participate in medical research, and which can be rapidly tapped for both observational and interventional trials of approved drugs.

During his talk, Califf banged the drum for NIH researchers to get involved in these networks, which he said will be able to answer critical questions at a fraction of the cost of clinical trials the way they are currently done.

For example, a third of the U.S.’ population suffers from pain, and opioids are the most commonly prescribed remedy. Drug overdoses now account for more deaths than motor vehicle accidents.

Yet there is no evidence either way on whether opioids are beneficial beyond a treatment duration of 12 weeks – a state of affairs that Califf termed “frightening” in light of how many prescriptions are written for opioids.

Designing the best interventional trials to address whether it makes sense to prescribe opioids for longer than 12 weeks, he added, is the sort of question NIH researchers could and should be involved in. “These are things where I think the NIH and its investigators would have a lot to say.”

Califf acknowledged that switching to learning-based health systems will initially be “more painful for the pioneers. But as we learn together, I think the cost of knowledge would really plummet.”

And anyway, he added, it’s the right thing to do.

“Until a few years ago, it just wasn’t plausible that you could generate reliable evidence at the needed rate. So in my view, it was OK. But now we can ­– there’s no technological limitation. So to me it’s in a different [category] of urgency. Because it’s now just possible to do it whereas it wasn’t before.”