In 2015, President Obama heavily promoted one of the more overlooked changes of the Affordable Care Act: paying for results instead of the number of visits and procedures.

The idea is simple: Insurance companies and Medicare/Medicaid would pay doctors and hospitals more or less based in part on patients’ performance ratings, or patient outcomes. For example, the Affordable Care Act already imposes financial penalties on hospitals with high rates of hospital-acquired conditions, and the Medicare Advantage program explicitly takes patients’ performance ratings into consideration for plan payments, as of 2012.

We have teamed up with HealthOutcome.org, a Silicon Valley start-up collecting large-scale treatment ratings directly from patients, and crunched its data. The results are clear: Outcome-based assessment of health-care provision has the potential to improve care and decrease health-care costs in the future.

U.S. health care is more expensive and invasive than elsewhere in the world

The U.S. has one of the most expensive health-care systems in the world, with costs of $3 trillion in 2014, or $9,523 per person. Health spending accounts for a staggering 17.5 percent of the nation’s gross domestic product, more than that of any other country.

Not only is health care expensive, but it is also invasive. On average, Americans have 51.4 million surgeries a year — three to four times more per person than in the U.K. And there is a link between invasiveness and spending. On average, more invasive treatments cost much more than non-invasive treatments. That gives doctors more incentive to recommend surgery rather than, say, physical therapy. The National Academy of Medicine estimates that one-third of all treatments, or $750 billion’s worth, are unnecessary.

But does it work better?

HealthOutcome.org has collected more than 60,000 individual treatment ratings by asking patients to rate their own treatment experiences for a given condition such as lower back pain to help patients find evidence-based cures. Treatment ratings range from “Cured” to “Worsened.” The company is applying crowdsourcing techniques that have proven successful in such areas as travel and shopping  to health care.

We took the patients’ treatment ratings for two common injuries, lower back pain and plantar fasciitis, an inflammation of the ligament at the bottom of the foot. We use their data to predict the effectiveness of two treatments per injury by comparing the results from patients who got one treatment to the results from similar patients (when it comes to weight, age and gender) who got the other.

We compared outcomes for treatments for lower back pain 

Low back pain affects approximately 80 percent of all Americans; it costs around $100 billion a year to treat. We compared the satisfaction ratings of patients who got one of two treatments: back surgery and non-invasive training in modifying posture.

We estimate the probabilities of being cured for a hypothetical patient with median age and weight. If this hypothetical patient chooses postural modification instead of surgery, his or her likelihood of getting worse goes down by a factor of 1.6 — and his or her likelihood of being cured goes up by a factor of 2.7. That’s a vast variation and extremely significant, considering the high number and high cost of back surgeries, as you can see in the figure below.

And we compared outcomes for treatments for plantar fasciitis

For plantar fasciitis, we compared results from cortisone injections, on the one hand, and what’s known as the Graston technique, therapeutic exercises applied to the skeletal muscles, on the other. Around 40 percent of all plantar fasciitis patients in the HealthOutcome.org database had cortisone injections; only 4 percent were treated with the Graston technique. However, we find that the Graston technique is much more likely to cure patients. A hypothetical patient with median age and weight treated with the Graston technique is 1.6 times more likely to be cured than the person who receives cortisone injections and is 1.5 times less likely to get worse, as you can see in the figure below.

Sometimes, of course, doctors recommend more or less invasive practices based on a specific patient’s condition: Patients might get the more invasive treatment because they have been suffering for a longer time or are in more extreme pain than other patients. That’s something we cannot control for.

Overall, less invasive works better

But we do find that patients with the same aliments, with comparable weight, age and gender, have dramatically better outcomes with the less invasive procedures.

We would be very surprised if that is entirely because those who get the more invasive treatment are suffering from a worse form of the ailment. This isn’t a randomized trial, which would give us the best data but is much harder to implement. And yet these two examples — and the general trend we see in the data — clearly suggest that just because a treatment is often prescribed doesn’t mean it’s often successful.

Paying by procedure – and therefore giving medical professionals an incentive to aim for volume rather than quality – isn’t ideal. As patients, we care more about the outcome than we do about the volume of treatment. And so that’s how medicine should be both judged and paid: by quality, not quantity. As Obama said last year about the Affordable Care Act’s shift to paying for outcome: “These changes are encouraging doctors and hospitals to focus on better outcomes for their patients.”

Tobias Konitzer is a PhD candidate in communication at Stanford University.

David Rothschild is an economist at Microsoft Research. Find him on Twitter @DavMicRot.