Deborah L. Birx offered then-President Donald Trump an unusual compliment in an interview last March.

“He’s been so attentive to the scientific literature and the details and the data,” the coordinator of the White House's coronavirus pandemic response said in an interview with the conservative network CBN. “I think his ability to analyze and integrate data that comes out of his long history in business has really been a real benefit during these discussions about medical issues.”

This was early in the coronavirus outbreak and shortly after Trump and the White House had endorsed new measures aimed at containing the virus. It was before Trump later pushed to scale those measures back, before he amplified dubious claims about various treatments like hydroxychloroquine and before he gave tacit approval to building immunity by letting the virus spread broadly. Birx's bold assessment of Trump's reliance on data nonetheless conflicted with other available evidence, like his insistences about the availability of not-readily-available tests or his repeated claims the virus would soon dissipate. It was an odd claim to make about a president already known to embrace whatever bit of evidence he wanted to hear and not necessarily the evidence most illustrative of reality.

But this was, in part, Birx’s job: finding a path between Trump’s political goals and the federal government’s response to the pandemic. Some parts of her comments were obviously less statements of fact than of hope. If Trump heard her praising the embrace of data, this line of thinking went, maybe he would actually embrace the data she presented.

In an interview with CBS News this weekend, Birx declared that — unsurprisingly — Trump did not actually make decisions about the virus based on reliable data.

“I think the president appreciated the gravity in March,” Birx said. “It took a while after I arrived in the White House to remove all of the ancillary data that was coming in. I mean, there was parallel data stream coming into the White House that were not transparently utilized. And I needed to stop that where people were —”

“You mean outside advisers?” interviewer Margaret Brennan interjected.

“Outside advisers, coming to inside advisers,” Birx replied. “And to this day, I mean, until the day I left, I am convinced there were parallel data streams because I saw the president presenting graphs that I never made.”

There’s no real question that this occurred. In fact, it was a feature of his presidency from the outset. One of the changes implemented soon after John Kelly took over as White House chief of staff in the summer of 2017 was to vet the information being presented to Trump in the Oval Office. No more random visitors slipping Trump nonsense from far-right websites to advance their own goals — or so Kelly hoped. By early last year, though, Kelly was long gone from the White House and, as Birx discovered, so was any effective system at limiting what was presented to Trump.

Asked by Brennan who was presenting that information, Birx suggested that it was likely coming from inside the house.

“I know now by watching some of the tapes that certainly Scott Atlas brought in parallel data streams. I don't know who else was part of it,” she said.

Atlas was the neuroradiologist who had gotten Trump’s attention through appearances on Fox News in which he minimized the threat posed by the rapidly spreading virus. Atlas explicitly argued that surging infections were a good way to build immunity, telling Trump what the president — eager to scale back economic restrictions before the 2020 election — wanted to hear. Trump added him to the White House response team, where friction quickly ensued. In an interview with the New York Times this weekend, the other leading figure on that team, Anthony S. Fauci, argued that the addition of Atlas was meant to be “a pushing out of Debbie Birx.”

In other words, it seems clear that Trump brought Atlas on specifically to embrace the sort of data Atlas was willing to provide. That Atlas’s data was iffy wasn’t a secret. At one point he tweeted a chart aimed at diminishing the virus’s death toll. At another point, Twitter removed a message from Atlas claiming that masks weren’t effective at controlling the spread of the virus.

Again, that this was Trump’s approach — find someone saying what he wants to hear, not necessarily someone saying what’s true — was not new to the pandemic, though it certainly flourished last year. He was already infamous for quizzing random people nearby for their opinions on issues of governance. As one former administration official told The Washington Post in 2017 about Trump’s private property in Florida: “At Mar-a-Lago, anyone who can get within eyesight changes the game.”

Atlas was easily able to get within eyesight at the White House.

Birx's interview with CBS was likely aimed in part at revising the history of her involvement in a pandemic response which, by her own measure, had failed.

“I think when the record goes back and people see what I was writing on a daily basis that was sent up to White House leadership,” Birx told Brennan, “that they will see that — that I was highly specific on what I was seeing and what needed to be done.”

She also touted her efforts to interact directly with states outside of the Trump communications umbrella.

“You have to figure out how to get that message out when you can’t get it out from the head of the country,” she told Brennan. “And that’s our job.”

Unfortunately, for Birx, the record, both public and private, includes numerous examples of Birx not only pushing Trump to embrace the data she provided but also of her providing data to the public which misrepresented the pandemic.

Last July, the Times explored the ways in which Trump’s response had already stumbled. One reported reason? Birx’s embrace of overly sunny assessments of the course the pandemic was likely to take.

“She routinely told colleagues that the United States was on the same trajectory as Italy, which had huge spikes before infections and deaths flattened to close to zero,” the Times reported. She would reportedly distribute diagrams aimed at supporting her case, telling people that the country had hit its peak — “and that message would find its way back to Mr. Trump.”

That the United States would surge beyond Italy was quickly apparent. Even by July, we hadn’t seen a pattern like Italy’s: It had largely suppressed the virus, but the United States never did. Italy’s case totals surged in the fall, passing the United States on a population-adjusted basis, but by late November the United States had again taken the lead.

It wasn’t just behind closed doors that Birx’s claims were questionable. Over and over, the data she presented at White House press briefings was overly rosy or obviously misleading. We documented such statements on March 27 (the day she praised Trump’s reliance on data), April 6 and April 16. A claim such as stating that there was a “low level of cases” in 40 percent of the country because 19 states had few cases is clearly misleading, for example, given that those 19 states represented only 7 percent of the country’s population.

But this, too, is how the Trump White House often worked. Eagerness to keep Trump pointed in the right direction seems to have meant making occasional adjustments to observed reality. Perhaps Birx’s data points, derived from the sorts of models of how the virus would progress about which Fauci repeatedly offered words of caution, were themselves flawed in ways that aided Trump. Or maybe at times she tweaked things a bit to comport with what Trump wanted to hear.

On May 23, for example, Birx informed reporters that the White House had determined that it was safe to return to activities like “playing golf” — something Trump, an avid golfer, hadn’t done in months but which he did again on May 24.

Trump, we assume, was simply being attentive to the scientific literature and the details and the data.