I'm Jeremy Warner. I'm faculty at Vanderbilt University Medical Center in Nashville, Tennessee, and I am a practicing malignant hematologist, although I'm also an informatics faculty here. And I spend a lot of my time doing research related to data from or about patients with cancer.
And so when the COVID-19 pandemic really came along, I pivoted in this research and started to focus on building a registry to collect information about patients with cancer who'd been infected with SARS-CoV-2, or had presumptive COVID-19. And so that's the COVID-19 and Cancer Consortium. And at ASCO, I'll be presenting on this data, and presenting on behalf of the entire consortium, as well as my co-authors on the ASCO abstract.
And what started as five institutions on March 15th quickly expanded to 14 two days later, at which point we launched the registry. And at the latest count, we've now expanded to 105 institutions. We are open to site-level participation in the US and Canada, and we're also open to anonymous reports from individual health care practitioners in Argentina, Canada, the European Union, and the United Kingdom.
The survey that we built is designed on the REDCap platform. And it's designed for both real-time reports, so a practitioner or their proxy could go into the survey and report as an infection is happening, or it can be reported retrospectively, after the course of the COVID-19 illness. So for this first analysis, which was performed after one month of the survey being open, we had 928 cases included for analysis.
So now, I'll just tell you a little bit about the demographics of the patients that were included in this initial analysis. So they were almost exactly 50% male, and the median age, as we suspected, was older, with a median age of 66 years. And of note, almost a third of the patients in this cohort are over the age of 75.
The cohort is 50% non-Hispanic white, 16% non-Hispanic black or African-American, and 16% Hispanic. I'll just say a quick word about that. If you compare these percentages to what you might expect to see based on census reports, you see that there's almost twice as many black or African-American patients in this cohort than you'd expect from the census alone.
In terms of the cancer characteristics, the most common cancer in this cohort at this point is breast cancer, representing over 21% of the cancers reported, followed by prostate cancer, GI cancers, lymphoma, and then thoracic cancers, including lung cancer accounts for 10% of the cancers reported. Should note that about 12% of these patients have more than one cancer diagnosis.
39% of the patients were on active anti-cancer treatment. And we define that as treatment given within four weeks of the COVID-19 diagnosis. 43% of the patients had active cancer, which can be variably defined, but essentially means they had measurable cancer in their body at the time of diagnosis. And we'll see additionally that we asked about the status of that cancer. Was it responding to treatment, was it stable, or was it progressing. And that'll become important when I get to the results.
We also asked about the ECOG performance status. And the majority of this cohort was ECOG 0 or 1, accounting for 2/3, and about 8% had ECOG performance status of 2. And then the final demographic I'll mention is smoking status. About half of the cohort were never smokers, one third were former smokers, and a small percentage, 5%, were current smokers.
So getting into the results a bit, first of all, this cohort was collected quite recently, quite rapidly. Obviously, this is an urgent situation that we're dealing with, and as a result, the median follow-up is rather short, at 21 days. But even so, what we see is that 121 out of the 920 patients have died. And that calculates out to a 13% death rate.
It's hard to put that in context, perhaps, but if you compare that to the global averages, which are reported by Johns Hopkins, it's about double the global average of about 6.6%. So that's sort of our global mortality. If we start to look at subsets, we see that there are some major differences. And I'll go through a few of these.
So if you look at male sex, the death rate was 17%. For former smokers, it was 20%. For those over the age of 75, it was 25%. If we look at the cancer status-- like I said before, we asked about that-- if the patient has an active cancer that's stable or responding, their death rate was still a bit higher than the average for all patients, at 14%. And if they were progressing, the death rate was 25%. Looking at ECOG performance status, if it was 2 or higher, the death rate was 35%.
We didn't see any statistical differences in the death rate based on race or ethnicity. Really, the only population that we found had a lower crude death rate than others-- and this didn't have statistical significance because the numbers are small, but the only population we found that perhaps had a lower mortality overall were those with no comorbidities. And in that population, the death rate was only 2%.
So we looked at other outcomes beyond death. So we looked at hospitalizations. So we found that 50%, or half of the patients in this cohort were hospitalized. And of those patients, 23% died. We found that 14% were admitted to the intensive care unit, and of those, 38% died. I think a silver lining, if there is any here, is that we have seen that a few of the ICU admitted patients have survived and recovered. It's about 30% or so. Again, our follow-up is short. We found that actually a quarter of the patients who started a report in the ICU remained in the ICU at the subsequent follow-up. So perhaps those numbers will go up over time.
But when you start to look at some of the higher risk sub-populations, the numbers are very concerning. So if you look at patients who are age 75 and above who are admitted to the ICU, the death rate was 54%. And if you look at those with an ECOG performance status of 2 who were admitted to the ICU, the death rate-- sorry, that should be ECOG performance status of 2 or greater-- the death rate was 68%.
And then finally, I'll just tell you about intubation, which is a separate question from intensive care unit. It's clear that some hospitals in some regions are actually intubating patients outside of the ICU. And likewise, if you're admitted to the ICU does not necessarily mean you you'll end up on a mechanical ventilator. So we looked at that separately. 116, or 12% of the patients were intubated at some point, and of those, 43% died. The flip side of that coin, of course, is that 57% survived, but again, our median follow-up is short.
Looking at those subsets again, though, if you look at age 75 and older, those who were intubated, 59% of those patients died. And then if you look at the ECOG performance status of 2 or higher, there's only 13 patients who met that criteria, but 11 of those, or 85% of them died. The mortality was high in this cohort, and was associated with general population risk factors, which have been described before, as well as those unique to patients with cancer.
And there are a few additional independent factors which we found in our in our logistic regression model which were associated with increased mortality. And in addition to the ones I really spoke of, which were age, male sex, former smoking, ECOG performance status, and active cancer, we also found that having a very high number of actively treated comorbidities was independently associated with mortality, and we found that the receipt of the combination of azithromycin and hydroxychloroquine was independently associated with increased mortality. Like I said, older age, poor performance status, and progressing cancer were strongly associated with increased mortality, and especially when you look into those subsets admitted to the ICU or on mechanical ventilation.
Our major limitation is, even with a sample size of 928, we were not able to perform a fully adjusted statistical analysis, and neither were we able to look at a survival type analysis because the median follow-up was so short. So a larger sample size and longer follow-up are needed to more completely understand the impact of COVID-19 on patients with cancer overall, as well as specific subsets, such as those with hematologic malignancy or those with prostate cancer.
We've just had our second data lock, and we're actually over close to 2,000 patients now that have been recorded in. And we're still in the process of cleaning that before we repeat any further analyses, but that's the pace we've seen. So literally, the database has doubled in the past month. I believe that we'll see some slowing because a lot of sites hopefully have seen the peak of their infections, and are basically catching up by recording cases into the registry.
But for now, we're planning to have a data lock more or less every month. And we're still working out the details, but the idea is that the consortium members will have the opportunity to analyze this data internally for a certain period of time, probably about six months or so, and then we do plan to release an aggregated version of the data for the public through some sort of online sharing platform, such as the cBioPortal system, so that really anybody can come and take a look, and perhaps generate hypotheses that we didn't even think of.
We did this first analysis, and then we've opened to the consortium. And there's close to 300 participants in this consortium at this point, 105 sites, and many sites have multiple members. And as a member of the steering committee, I'm very interested in making the consortium democratic and the process open. So we really wanted to encourage anybody to propose a research idea. And we actually got 45 research ideas from the first round of proposals. And those range, you know, there's just so many interesting questions that could be asked. And I think we'll see another 45 the next time we collect proposals.
Some of the questions that we want to answer sooner than later would be, again, looking at specific subtypes of cancer, trying to tease out whether there are increased risks for certain patients over others, getting into that treatment question more. So not only trying to look at treatments not just in these major categories, but even looking at individual drugs, but also looking at the timing of treatment administration. We just couldn't do that with the sample size we have.
But there are some real questions about, for example, myelosuppression from certain regimens. That usually is brief, it's in the two-week period from the delivery of the chemotherapy, so that's a very short time point, whereas other drugs, such as the drug rituximab, can cause immunosuppression over a very long period of time, six months or longer. So it really matters which drug is delivered, when it was delivered. And these are the kind of details that we are hoping to get as the cohort grows larger.