Nationwide study finds COVID-19 patients triple risk of death in small hospitals

Studies in Boston and across the country have shown that severely ill COVID-19 patients are very likely to survive if treated in a larger hospital.

Drastic Investigation JAMA Internal Medicine is the first journal to examine hospital mortality rates for COVID-19 patients nationwide. It contains detailed data for more than 2,200 patients from 65 hospitals.

“Patients admitted to hospitals with less than 50 ICU beds, or smaller hospitals, had more than three times the risk of death than those admitted to large hospitals,” said David, senior author at Harvard Medical School and Brigham. Dr. Leaf says Women’s Hospital. Below is an edited excerpt of our conversation.

How would you summarize the content of this white paper?

We conducted a multicenter study at 65 hospitals in the United States, collecting detailed data on more than 2,000 severely adult patients with COVID-19. Within 28 days of admission to the ICU, the patient’s risk of death was found to be 35%.

We also identified several independent patient- and hospital-level mortality risk factors. Patient-level risk factors included characteristics such as older age, male sex, obesity, and cancer.

Regarding hospital-level risk factors, we found that patients admitted to hospitals with less than 50 ICU beds, or smaller hospitals, were more than three times more likely to die than patients admitted to large hospitals.

Whether they were talking about medications such as hydroxychloroquine or steroids, or specific interventions such as proning, we found that treatments differ dramatically from hospital to hospital.

Many people hear statistics that the risk is three times higher and say, “If I get a COVID, I want to go to a big hospital.” Would you like to soften the reaction with some warning?

Thank you. In this study, we collected very detailed data. There were over 800 data points per patient, each manually chart-reviewed. So for more than 2,000 patients, there were over 1 million individual data. Adjusted a very comprehensive set of potential confounders.

Still, I couldn’t capture everything I could imagine. So, for example, there was no data for doctors or nursing staff. There was no data on the burden on the hospital. In addition, there were no data on the socioeconomic status of patients emerging as an important risk factor for COVID-19 patients. Therefore, there is a possibility that confounding remains. There may be factors we did not measure that could explain some of these findings.

But do you think it’s a rational takeaway and better to go to a hospital with more than 50 ICU beds?

It’s really hard to answer. Larger hospitals have more resources. And with settings other than COVID-19—for example, when looking at patients who need ventilators because of respiratory failure due to causes other than COVID-19—larger hospitals seem to give better results.

Could you comment on your finding that race does not seem to correlate with mortality?

In fact, it turns out that blacks are far more included in our cohort of critically ill patients than in the country. About 30% of the population was black, but black is only about 13% of the United States population. Thus, blacks were over-represented nearly three-fold in our severely ill cohort.

However, among patients who participated in our study, mortality rates did not differ by race. This is similar to the findings reported in the Louisiana study in the New England Journal of Medicine. If you’re black, you’re more likely to be admitted to a hospital or ICU, but once you’re there, you’ll have similar results compared to patients of other races.

How are our US mortality statistics assessed against other countries?

It’s very difficult to compare studies because each study has a different follow-up period. In some studies, patients are followed for only one week, and patients often do not die in the first week. Inpatient care varies by country and by hospital, even within the same country. I think our deaths are the same as those reported by other countries.

Are your findings workable for medical staff and families?

We have not developed a prognostic scoring system that incorporates patient age, gender, or other factors. For example, it turns out that there is a 90% or 10% chance of death. I did that in a subsequent paper currently under review.

But I think just understanding what the risk factors for death may be to help determine how aggressive a patient or family will be.

Was there a systematic red light for outlier hospitals that looked unusual?

It doesn’t go into a specific name, but the simple answer is yes. Risk-adjusted mortality in this study ranged from 7% in the lowest risk hospitals to 80% in the highest risk hospitals. In fact, this was one of the key findings of the study. Taking into account many of the differences in patient characteristics across hospitals, the risk of death has changed more than 10-fold.

So isn’t it a desire to go to a larger hospital if possible?

But you may not have a choice. You may need to go to the nearest hospital.

What’s next?

What we are really interested in is identifying treatments that can improve outcomes. Of course, randomized controlled trials are underway and they will always be the gold standard, but randomized controlled trials are time consuming and cannot answer all the questions.

So, for example, there are ongoing randomized controlled trials of certain drugs, but they are not specifically targeted at critically ill patients. Or, they may be enrolled in a far smaller number of patients than those enrolled here, and therefore lack sufficient statistical power to assess hard outcomes such as mortality.

So I’m trying to answer some of these questions about which treatments improve outcomes, using a database of over 5,000 patients. There are several papers under review that are considering this.