How Stanford’s Vaccine Algorithm Caused a Great Controversy and Left Frontline Workers Behind the Line

To choose who would be first in line for a coronavirus vaccine, the UCSF Medical Center in San Francisco gathered a group of brain people this fall to study guidelines and prioritize staff who came in most contact with patients, including guards. Kaiser Permanente did pretty much the same thing across 21 medical centers.

Stanford Medicine officials also studied guidelines – then built a mathematical algorithm that prioritized people who on paper were at high risk for COVID-19, such as elderly staff, but not the health care workers who came into most contact with patients, such as intensive care physicians. and nurses.

It did not take long for medical residents and fellows on the front lines of the crisis to see that they were being asked to risk their lives while a shot that would protect them was given to others. Doctors caught national attention as they staged an angry protest Friday, confronting Stanford Medicine executives not only about the skewed rollout, but about what they described as a secretive approach. The leaders quickly apologized and promised to change their approach and make the vaccine distribution process more transparent.

Attention then turned to the algorithm that Stanford created to control their decision.

The private hospital has refused to discuss it. But the method that made Stanford an outlier in how it prioritized the vaccine has been confused by researchers at MIT and downplayed by math nerds on Twitter.

“What happens when you apply the age priority scheme … to a population selected on the basis of their theoretical level of * exposure *?” tweeted Whitney Robinson, an epidemiologist at the University of North Carolina’s School of Public Health. “You get ridiculous results as a wealthy 68-year-old radiologist from Stanford Health Care and executives working from home in spacious houses and prioritizing over 29-year-old residents with two roommates working in the (emergency department) every day. ”

An MIT Technology Review article on Stanford kerfuffle points to a 2019 study in the journal Science, which reveals that algorithms commonly used by hospitals to guide health decisions tend to favor white patients over black patients. In a case where patient health expenses are used as a power of attorney for how sick they are, the algorithm mistakenly concludes that black patients who tend to spend less are healthier than equally sick white patients. The study’s authors estimate that if the algorithm were fixed, the proportion of black patients receiving the care they needed would increase by 29 percentage points from 17.7% to 46.5%.

“Asking the algorithm to answer the wrong question results in all sorts of problems – racial bias in our study and the problems (highlighted) at Stanford,” said study author Ziad Obermeyer, a physician and associate professor of health policy and management at UC Berkeley.

Algorithms can help with decision making by assigning points to different variables. A graph of Stanford’s algorithm published in the MIT Technology Review shows that employees were awarded points based on age (65 years and over and 25 years and under), California Department of Public Health guidelines, and job features. Jobs and departments received scores based on the prevalence of COVID-19, percent positive for COVID-19, and percentage of coronavirus tests collected.


The Chronicle shared Stanford’s algorithm with Obermeyer, who said part of the problem seemed to be too much on the wrong variable: people at risk of getting sick (older employees) instead of people at risk of infection (employees, working with ICUs or patients in emergency rooms.)

“When resources are scarce, any priorities will always be open to debate,” Obermeyer said. “But the goal of algorithms must be to make this prioritization fair, transparent and justifiable. No one should hide behind ‘an error in the algorithm’, as if the algorithm has its own mind when they literally designed the algorithm. ”

UCSF took a different approach to the rollout by pulling doctors, administrators and “stock experts and ethics experts” together to make the decision, said Adrienne Green, the hospital’s chief executive. The group first looked at guidelines that recommended that employees at greatest risk be given priority: those working in COVID units and emergency rooms. Vaccination of them first can also protect vulnerable people that employees often come in contact with.

Dr. Jennifer Reid, a surgical fellow, is one of these employees. She has been at the helm for nine months. When patients need emergency care, their surgery can not wait until their coronavirus test results return, so Reid dresses up with an N95 mask, goggles – the works – and hopes she does not get sick. It’s a huge concern, especially because her mother, about chemotherapy, lives with her. Her husband, a medical assistant at another hospital, is also in danger. He has slept in a hotel room more than once for everyone’s safety.

On December 13, Reid received a warning that she had been selected by lottery to make one of the first appointments. Three days later, when the serum got into my arm, “I immediately felt relieved,” she said. “I think it was just a weight that was lifted from my shoulders.” Her second dose comes in January.

Dr.  Jennifer Reid, a surgical critical care and emergency care surgeon at UCSF, received her first dose of the coronavirus vaccine on December 16, 2020. She will need a second dose in a few weeks.

The UCSF also considered employees from communities with a high infection rate, such as the Mission District, Green said. The strong Latino area includes many UCSF employees, including security guards who also work in the emergency room and high-risk locations.

“We decided that everyone who worked in these areas would be in the same priority group for vaccination,” Green said.

Of the first 896 employees vaccinated with the Pfizer vaccine at UCSF, just over half were nurses and medical residents. Technicians, respiratory therapists and other health professionals were approx. 19% of the total number, and the faculty (ER and ICU doctors) was 17.5%. The remaining 8% included drivers, cleaners and others.

Reina Lopez, a UCSF housekeeper wearing heavy protection for work, was the second person to be vaccinated at UCSF on December 16th. “I felt all the emotions – happy, excited, scared – but most of all relieved that we are finally moving forward to get back to normal,” she said.

Lopez and Reid said they appreciated the UCSF’s rollout of the vaccination. But not all employees initially agreed. Brittany Howze, a nurse in the COVID unit at Mt. Zion campus, had grasped that the process was not going fast enough.

So Howze was surprised last weekend when she learned that her turn would be Monday this week and she would be among the first to be grafted.

It hardly happened. The serum has triggered bad allergic reactions in some people, so when asked if she had a history of such reactions, Howze hesitated. Then she answered no and kept to herself the serious reactions she had had after eating pistachios and cashews: “I was afraid I could not get it,” she later told a reporter. “I have literally had dreams at night about getting the vaccine. And then you wake up to remember that it’s not true. I’m so glad this day is finally here. ”

On Tuesday, UCSF received 7,900 doses of the second FDA-approved vaccine of Moderna and expects to inoculate 11,800 staff and medical practitioners at greatest risk for infection, the hospital said.

Kaiser Permanentes dr. Stephen Parodi, an infectious expert leading the hospital system’s COVID-19 response in Northern California and across the state, said Kaiser never considered using an algorithm.

“We felt that it was really important to have the appropriate clinical expertise and operational management in the room and that it could not be delegated to anyone other than our experts,” Parodi said, adding that these experts have used “a lens of equity” to ensure justice.

Parody said he is not vaccinated ”for example because I do not work in the intensive care unit. We have tried to stick to this approach. ”

Stanford changed its approach the very night of the protest.

“Stanford became a free-for-all,” said Andrea Henkel, one of the doctors who took part in the protest. “It became a first-come, first-served situation until they ran out of doses on Friday,” said Henkel, a resident who waited in line for 90 minutes before receiving her injection in the hospital’s atrium.

In fact, Stanford Medicine has been on vaccination since Friday, grafting more than 2,000 health workers throughout the weekend, spokeswoman Julie Greicius said.

A new committee for vaccine management is overseeing the revised plan, which “prioritizes health professionals in our clinical environment” from residents and nurses to housekeepers, she said.

Sharif Vakili, another protesting doctor, said he was pleased with the hospital’s response.

“It is clear that they are working hard to correct this mistake,” he said. “And I think they really feel terrible about what happened.”

Nanette Asimov is a co-author of the San Francisco Chronicle. Email: [email protected] Twitter: @NanetteAsimov

Source