Understanding Death Data
Updated August 31, 2019
When we release provisional data on drug overdose deaths, we provide an additional set of numbers called “predicted counts.”
Drug overdose death records are sometimes first submitted with the manner of death reported as ‘pending investigation,’ due in part to the time needed to get toxicology results. Because it can take a longer time to determine the cause of death in these cases, we can look at historical patterns to arrive at a likely estimate of the final numbers. These predicted counts represent our best estimate of the number of drug overdose deaths that will eventually be included in the final data, after the records reported as ‘pending investigation’ have been updated.
While predicted counts often differ from both the provisional and final data, they can help us more accurately understand this shifting and unfolding public health crisis.
Mortality data answers critical questions like these, helping us understand how many Americans are dying and – importantly – why. These data are relied on by researchers, epidemiologists, clinicians, policymakers, and many others working to identify problems, find solutions, and save lives.
Where death data come from
CDC’s National Vital Statistics System (NVSS) captures all deaths from all causes across every state in the nation. These mortality data help us track the characteristics of those dying in the United States, help determine life expectancy, and allow comparisons of death trends with other countries.
Today, CDC is moving toward a strategy that allows us to collect and share these data faster for real-time surveillance. Work being done to modernize the vital statistics system across our country has allowed CDC to release data in new ways, including through quarterly and special reports that are made available in advance of our final data releases.
The descriptions below are intended to answer common questions about our data releases and offer links to help you find vital statistics data.
Provisional vs. final data releases
What are provisional data?
To use our mortality data accurately and effectively, it’s important to understand the differences between the two main types of data we release: provisional data and final data.
Provisional data are preliminary data that may not yet be complete. Provisional data reports are based on a current flow of new and updated records received from states. These data may be released monthly or quarterly and are subject to change as information continues to be collected and analyzed – as such, they are estimates that may differ from the final count.
Provisional counts and estimates are released through NCHS Vital Statistics Rapid Release Program and currently include:
- Provisional Death Counts for Coronavirus Disease 2019 (COVID-19)
- Quarterly Provisional Estimates
- Monthly reports of Provisional Drug Overdose Death Counts, with a detailed descriptionof how provisional counts of drug overdose deaths are adjusted for underreporting
- Weekly Pneumonia and Influenza Mortality Surveillance
- Online Reports from the Vital Statistics Rapid Release Program provide additional methodological information or analyses to help readers understand and interpret provisional mortality and natality data.
Final (annual) data are available in:
- Public use files
- Online query tools such as:
- National Center for Health Statistics (NCHS) publications: National Vital Statistics Reports (NVSR) and various NCHS Data Briefs
- Restricted use files, and through the NCHS RDC. (see section on “Customized files available under restricted conditions”)
- The National Death Index (NDI), which is available to epidemiologists and other health and medical investigators solely for statistical purposes in medical and health research. Because it may contain personally identifying information, the NDI service is not accessible to the general public.
Provisional data are not final and must always be used with the understanding that numbers and information may change as the data becomes more complete.
What are final data?
Final (annual) data are released only after we have received all death records from the states and have fully reviewed the data for completeness and quality.
Final data contain the most accurate and complete information we provide. These official records are used in publications and data visualizations, and for investigation and research, among other uses.
Why does CDC release provisional data?
The benefit of releasing provisional data is timeliness. Final data takes time to ensure accuracy and completeness. Faster and more frequent data releases may signal changes in trends and offer investigators clues about emerging public health events. Preliminary data allows us to track and monitor the ongoing impact of a crisis and take smarter action to save lives.
Timing is everything
The timing of data releases can be complex. Different sets of vital statistics data are released at different intervals — usually monthly, quarterly, or annually. Additionally, it can take varying lengths of time (often between six months and a year, or sometimes longer) for CDC to collect and review different sets of data.
For example, certain types of data (like some causes of death) may require added investigative work and take more time to confirm. Final data always takes longer to release than provisional data.
These differences in timing, as well as some differences in the ways the data are tabulated, can result in variation between the numbers found in the final and provisional data. Understanding the types of data we release and their purposes can help when deciding which data to use.
- Data Visualization Gallery
This gallery showcases useful visual representations of valuable CDC data, including mortality data dashboards and provisional data dashboards
- National Center for Health Statistics (NCHS) product and report release schedule for birth and death data
This page tracks the release of upcoming reports on mortality
- Upcoming Releases of Note
This page is the most up-to-date information on releases by week, updated on Fridays for the next week