Many of the 17,000 people currently waiting for liver transplants in the U.S. have had bags  packed for weeks, months, or even years, ready for the moment the phone rings and the voice on the other end tells them to head to the hospital. Once an organ becomes available, they know, there will only be a small window in which they can receive it. This year, an estimated 6,000 of those people will undergo a transplant; another 1,500 will die waiting for that call. There just aren’t enough livers to go around.

The task of determining who on the waiting list should receive a new organ falls to the United Network for Organ Sharing (UNOS), a non-profit organization that coordinates transplants across the U.S. When a liver becomes available, someone from the organization enters the donor’s information—weight, blood type, geographical location—into a computer system, which spits out a list of patients who would be a good match. The program then ranks the patients based on an objective measure of illness, called a Model End-Stage Liver Disease (MELD) score. The patient most likely to die gets put at the top of the list, with first dibs on the organ. “It’s balancing equity in regards to patient access,” explained David Klassen, the chief medical officer of UNOS, “[while] making sure that the outcomes from the transplant surgeries are as good as possible.”

But access to organs is far from equitable. One of the biggest challenges UNOS faces is geographic disparity in access to organs—some parts of the country have far more organ donors than others, and organs can only be preserved and shipped for a limited time before transplantation. The country is divided into 11 Organ Procurement and Transplantation Network (OPTN) regions; across those 11 regions, typical wait times for organs across the country range from one year to more than six.

Researchers have been studying ways to tackle this issue for years, but the disparity also has another, more fine-grained element. A study recently published in the Journal of Hepatology identified a substantial, yet often overlooked, barrier to access: the likelihood that a particular transplant center will choose to use the organs when they are available.

For the study, a team of researchers at the University of Pennsylvania School of Medicine requested data on 23,000 liver offers to patients across all 11 OPTN regions. Only 37 percent of the organs included in the study were accepted for the first (sickest) patient. The researchers found dramatic differences in the percentage of organ offers that each center accepted, even after they took the quality of the organs and the size of each facility into account. Some transplant centers accepted up to 58 percent of liver offers; others accepted as few as 16 percent and turned down the rest, which were then offered to the next patient on the list.

David Goldberg, an assistant professor of medicine at the University of Pennsylvania who worked on the study, said he was surprised by “the magnitude of the variability among centers, and especially among centers in the same little geographic area.” In some cases, a rejection ultimately led to the organ not going to anyone at all—in 2014, around 10 percent of livers that were recovered for transplants went unused, despite a growing organ shortage in the United States, according to the annual report published that year by the Scientific Registry of Transplant Recipients (SRTR). In many cases, patients won’t even know they were offered a liver, as it’s up to the doctor to decide whether the organ is suitable for transplant.

When doctors reject an organ, they must submit a code to UNOS indicating the reason for the rejection. They may say they believed the size of the organ was inappropriate for their patient, or that the patient may have told their surgeon that they don’t want an organ from a donor over a certain age or who has used intravenous drugs, even if the liver has been cleared for transplant. They may say they decided that the patient is likely to live long enough to receive an offer of a higher-quality liver. But many researchers suspect there is another factor driving the rejection of offers, one that that has no code: a center’s reluctance to perform riskier surgeries (meaning procedures with either less-than ideal livers or less-than ideal patients) for fear it could damage their statistics.

Every six months, the SRTR releases a publicly available assessment of each transplant center in the country, using one-year post-transplant survival rates as a key metric. A poor evaluation means a center gets “flagged,” triggering an investigation into why its one-year survival rates aren’t what they should be. Consultants come in. Medicare and Medicaid may refuse to pay for their patients to be transplanted at these centers. It can cost the centers thousands or even millions of dollars. On rare occasions, they may even be shut down.

But one-year survival may not be the most useful metric for judging a transplant center’s quality. “It’s hard to argue against patient survival as an important measure,” Klassen said, but he also explained that because liver-transplant outcomes have improved so much in recent years, many surgeons and organ-procurement organizations wonder if the difference between top-performing centers and flagged centers is really clinically significant. The differences can be extremely small, often just percentage point or two.

The current system “puts pressure on people to try to make sure that they’re using the best organs that they can,” said John Roberts, the chief of transplant surgery at the University of California, San Francisco. This comes with its own drawbacks: “Not listing patients who may otherwise benefit from transplants because they’re too old or too sick ... It’s a problem that no one has a great solution to.”

In 2014, three larges transplant centers were flagged for having unacceptable one-year survival rates. The following year, all three transplant centers became more conservative, conducting about 30 percent fewer transplants. The centers may have increased post-transplant survival rates and thus stayed in business, but cutting down on the number of surgeries means they also likely denied potentially lifesaving transplants to patients that they would have treated a year before. “You’re only accountable for the patients that you transplant,” said Kevin Cmunt, the president and CEO of the organ-procurement organization Gift of Life. “So if you don’t transplant somebody, you can’t get in trouble.”

UNOS is currently working on a pilot program to identify alternative measures for assessing transplant centers, like program structure and access to resources. Next year, it hopes to test the measures in a small group of hospitals; in 2017, the organization hopes to apply the measures nationally, for a system that more accurately reflects a transplant center’s success and allows more leeway for high-risk procedures.

In the meantime, Goldberg is hoping for a smaller step: that transplant centers’ organ-acceptance rates will become publicly available information. “Patients, providers, and insurers really don’t know anything about this stuff,” he said. “This is an area that needs to have some more transparency.”