Resources: Data

HHS hospital mortality and readmission data

Updated August 2018

The U.S. Health and Human Services' Centers for Medicare & Medicaid Services "Outcome of Care Measures" is an attempt to show what happens after patients with certain conditions receive hospital care.

The AHCJ version of the data attempt to help you understand outcome measures, namely 30-day mortality rates and readmission rates. This information helps you understand, for particular conditions, whether a hospital's rates are in line with national averages, significantly better or significantly worse.

Latest release can reveal ongoing issues

Now that CMS has been reporting information on mortality and readmissions for several years for most utcome measures, it’s possible to use the data to identify both hospitals that repeatedly excel and those that have ongoing problems. It’s easier for a hospital to explain away a one-year blip, but much more difficult to explain why it has turned in subpar performance for three consecutive years, using formulas that give hospitals the benefit of the doubt to ensure they aren’t unfairly labeled worse than others.

ProPublica senior reporter Charles Ornstein has written an article showing how the data can be used. When the data were originally released, AHCJ held a webinar that provides an introduction to this data, including ideas on how to use the data in your own area. An MP3 recording of the webinar is available.

CMS describes "outcomes of care" in this way:

"Outcome of Care Measures" show what happened after patients with certain conditions received hospital care. The death rates focus on whether patients died within 30 days of their hospitalization. The rates of readmission focus on whether patients were hospitalized again within 30 days. Death rates and rates of readmission show whether a hospital is doing its best to prevent complications, teach patients at discharge, and ensure patients make a smooth transition to their home or another setting such as a nursing home.

The hospital death rates and rates of readmission are based on people with Medicare who are 65 and older. These rates are calculated using Medicare enrollment and claims records, and a complex statistical procedure. The death rates and rates of readmission are "risk-adjusted" - meaning that the calculations take into account how sick patients were when they went in for their initial hospitalization. When the rates are risk-adjusted, it helps make comparisons fair and meaningful. ...

The 30-day period is used because this is the time period when deaths are most likely to be related to the care patients received in the hospital.

Risk-adjusted mortality and readmission rates are available for individual hospitals in the areas of heart attack, heart failure, pneumonia, stroke, chronic obstructive pulmonary disease (COPD), coronary artery bypass grafting (CABG). CMS notes that, "Patients may have been readmitted back to the same hospital or to a different hospital or acute care facility. They may have been readmitted for the same condition as their recent hospital stay, or for a different reason."

The CMS numbers are based on statistical predictions. Because of this, the actual performance of smaller hospitals contributes less to their predicted rates than does the performance of larger hospitals. Comparing hospitals on rates that are not significantly different is not recommended. Thus, AHCJ's version of this data simply notes whether a hospital is identified as an outlier - that is, their rate is "Worse than U.S. National Rate" or "Better than U.S. National Rate" - it's reasonable to say that a hospitals' performance is significantly different than the national average.

The accompanying Microsoft Excel files are distilled from a larger database. You can also look up individual hospitals online via the government Hospital Compare site. The Excel files we've compiled are designed to allow you to compare more than three hospitals at a time, using spreadsheet or database software to filter, sort and use other analysis tools with precision.

THE BASICS

Mortality: One way to tell whether a hospital is doing a good job is to find out whether patients admitted to the hospital have death (mortality) rates that are lower (better) than the U.S. national rate, about the same as the national rate, or higher (worse) than the national rate, given how sick they were when they were admitted to the hospital. The information in the database shows how the 30-day risk-adjusted death rates for heart attack, heart failure and pneumonia at different hospitals compared to the national rate. For some hospitals, the number of cases is too small (fewer than 25) to reliably tell how well the hospital is performing, so no comparison to the national rate is shown.

Readmission: The information shows how often patients are readmitted within 30 days of discharge from a previous hospital stay for heart attack, heart failure or pneumonia. Readmission is when patients who have had a recent stay in the hospital go back into a hospital again. Patients may have been readmitted back to the same hospital or to a different hospital or acute care facility. They may have been readmitted for the same condition as their recent hospital stay, or for a different reason. HHS data show how different hospitals’ rates of readmission for heart attack, heart failure, and pneumonia patients compared to the U.S. national rate. You can see whether the 30-day risk-adjusted rate of readmission for a hospital is lower (better) than the national rate, no different than the national rate, or higher (worse) than the national rate, given how sick patients were when they were admitted to the hospital. For some hospitals, the number of cases is too small (fewer than 25) to reliably tell how well the hospital is performing, so no comparison to the national rate is shown.

Even if you've used this data before, check the data details page for changes and additional explanation.

A few caveats:

• Some hospitals treat too few Medicare patients with these conditions for the data to be meaningful. So if the number of cases is fewer than 25, you should be wary of trying to compare it to other hospitals.

• Readmission and mortality data are compiled using claims and enrollment data for patients in fee-for-service (original) Medicare only. It doesn't include those enrolled in Medicare Advantage plans, the uninsured, people in Medicaid or patients insured by private insurers. Be cautious about drawing broader conclusions.

• Hospitals may fare better than the U.S. average in one category and below it in another so be wary about drawing broad conclusions about the overall quality of the hospital from one figure. For example, a hospital with a better mortality rate might pay the price of that positive mark with higher readmission rates. Also, consider community factors. In locations with few primary care physicians or less access to ambulatory care, hospitals might have higher readmission rates for those reasons.

• A hospital's raw score - available here - may be below the national average, but when accounting for the possibility of chance, the results may not be statistically significant. That is why CMS notes which hospitals have rates significantly above or below the national averages.

• Results are for one period in time. While poor performance is noteworthy, sustained poor performance warrants greater attention.

On this AHCJ page, you will find journalism-friendly data in comparing different time periods. You will see results for every hospital in the country, as well as historical information that allow you to compare hospitals over time. 

HOW TO USE THE DATA

What types of stories can you write using this data? Every local reporter can find stories in this data: Which of your hospital rates high and which low?

Using another page on our site, you can add another layer of comparison for hospitals. For example, do local hospitals with the highest satisfaction have better mortality or readmissions outcomes? Then continue to add layers of information: If any hospitals have improved, how did they do it? Collect hospital inspection reports (known as 2567 forms) on your local hospitals and see how those with substantiated deficiencies fare on quality measures? Do any of your hospitals perform very well in one area and not so well in another area?

Sometimes the data provide a mixed picture. The Cleveland Clinic and Barnes Jewish Hospital in St. Louis, both of which have good reputations, are worth considering. A few years ago, the clinic's death rate from heart failure is lower than the national average but its readmission rate for the same condition was higher than the national average. Barnes Jewish had lower mortality rates for heart attack and heart failure and higher readmission rates in both categories.

Here's a story about the Cleveland Clinic in The Plain Dealer.

Using this data, you can identify hospitals in your state, community or ZIP code with worse-than-expected hospital outcomes for one or multiple years. By filtering and sorting with spreadsheet software, you can quickly compare hospitals. Numbers can be a start — not the end — of a story. Remember that these numbers only mean so much. Using data can give you a valuable tip sheet to generate ideas and questions in your pursuit of a story.

If you write a story based on these data, we’d love to see a copy. Please send an e-mail to jeff@healthjournalism.org.

Need help in analyzing Excel files? Check out the AHCJ tutorials about investigating health data using spreadsheets.

OTHER RESOURCES

Here are some background stories that are worth reading, from U.S. News and Slate.

A number of states and foundations have coalesced around reducing hospital readmissions. They can be good resources. Here is a link to the California Health Care Foundation's efforts.

HHS offers statistics about patient satisfaction measures for hospitals across the United States. The survey tool, known as HCAHPS, is the first attempt by HHS to standardize survey results of patients' perspectives of hospital care. AHCJ has compiled Excel files that allow you to compare hospitals at a time, using spreadsheet or database software to filter, sort and use other analysis tools with precision.