It's a retrospective Kaiser Health System cohort of 728,312 visits across two years, and the authors calculated the base rate of 50 per 100,000, as well as looking at other features and discharge diagnoses that increased the OR for death within 7 days. And, even the sickest, most elderly have OR that are low enough that you're still going to have good outcomes the overwhelming preponderance of the time. Age greater than 80 gives an OR of 10.6 and a score >3 on the Charlson Comorbidity Index gives an OR of 6.7. As for the diagnoses they found that are most highly associated with bad outcomes - the only two with OR great than 5 are noninfectious lung disease (OR 7.1) and renal disease (OR 5.6). These are kind of interesting buckets of diagnoses, specifically in the sense regarding how nonspecific they are - which the authors attribute to diagnostic uncertainty. I.e., the reason why patients had bad outcomes with "noninfectious lung disease" is because clinicians missed finding the specific morbid diagnosis in these patients.
I don't think this is practice-changing news, since these rates are so low in general that additional testing and hospitalization will harm more people than these missed diagnoses - but it's an interesting number crunch article.
"Patterns and Predictors of Short-Term Death after Emergency Department Discharge"