Knee replacements are booming. Between 2005 and 2015, the number of knee replacement procedures in the U.S. doubled, to more than one million. Experts think the figure might rise another sixfold in the next couple of decades because of our aging population.

Since many people receiving knee replacements are elderly, Medicare picks up most of the cost of such procedures. It shouldn’t be surprising, then, that the program is experimenting with ways to reduce the cost of each procedure.

The problem is, if health care providers make less money on each knee replacement they perform, they might start replacing more people’s knees than they should.

Does that sound crazy? Well, consider what happened when Medicare began experimenting with a new way of paying for knee replacements — something called bundled payment. Under such reimbursement, Medicare pays one lump-sum for the total cost of a knee replacement – not just the cost of the operation but also the cost of post-operative X-rays, physical therapy, even time in nursing homes or rehab hospitals.

Before bundled payment, providers received separate payments for each of these services. As a result, inefficient providers would take more X-rays than necessary or keep patients in rehab hospitals longer than needed, and they would be rewarded for such inefficiency. Under bundled payment, providers cannot send separate bills to Medicare for hospital charges, physician fees, outpatient X-rays and the like. Instead, they get a lump sum payment to cover all these expenses.

Moreover, Medicare tracks all the knee-replacement costs for a given patient over a 90-day period. If a patient incurs lower expenditures than expected, Medicare gives the providers part of these savings back as a reward (Warning: This is a way oversimplified description of bundling.)

Early evidence suggests that bundled payments reduce the cost of knee replacements by an average of almost $1,200 per patient. Save that much money on a few million such procedures in a year, and we are looking at billions of dollars of savings.

Moreover, research to date suggests that these savings don’t come at the expense of quality, at least as far as we can tell. (Quality measurement in health care is notoriously difficult.) For example, when knee replacements were paid for through bundled payments, there was no subsequent increase in readmission to the hospital or emergency room visits among patients whose procedures were reimbursed according to bundled payments. Same quality at a lower price — who could be against that?

Well, caution is in order. Health care systems that enrolled in the bundled payment system might have saved money on each procedure, but they more than made up for that by increasing the number of procedures they performed — about three procedures more per hospital compared with hospitals not receiving bundled payments. This finding — indeed all these findings — are tremendously preliminary. Bundled payments are still in their infancy. Quality measurement still doesn’t capture everything we’d like it to.

Nevertheless, the increase in procedures is concerning. People don’t like to see their income decline. And surgeons are people, too! Thus, we should not be surprised if surgeons and hospitals perform more procedures when they receive less money per procedure.

In trying to reduce the cost of health care, we should work on increasing the efficiency of specific services. But if we don’t also pay attention to the volume of services, we may be creating unintended consequences.

Bundled payments are not going to go away under the Trump administration, although the Department of Health and Human Services, under the leadership of orthopedic surgeon Tom Price, does appear to be slowing down bundled payment efforts. In assessing such payments going forward, policymakers should keep a few things in mind:

1. The federal government should continue to fund efforts to improve the science of quality measurement while reducing the burden of collecting such measures. These two goals potentially compete with each other, but we need to pay attention to both of them.

2. The federal government should increase the budget for health policy, health outcomes, and patient-reported outcome research. We need to understand how payment reform influences the quality and quantity of medical care.

3. Medicare and private payers need to assess the appropriateness of health care services, so doctors and hospitals don’t respond to reduced reimbursement rates by providing unnecessary services.

No one said bending the cost curve was easy.

Peter Ubel is a physician and behavioral scientist who blogs at his self-titled site, Peter Ubel and can be reached on Twitter @PeterUbel. He is the author of Critical Decisions: How You and Your Doctor Can Make the Right Medical Choices Together. This article originally appeared in Forbes.

Image credit: Shutterstock.com