Early Model-based Provisional Estimates of Drug Overdose, Suicide, and Transportation-related Deaths

This visualization displays model-based provisional estimates of the weekly numbers of drug overdose, suicide, and transportation-related deaths using “nowcasting” methods to account for the normal lag between the occurrence and reporting of these deaths. These early model-based provisional estimates were generated using a multi-stage hierarchical Bayesian modeling process to generate smoothed estimates of the weekly numbers of death, accounting for reporting lags. These estimates are based on several assumptions about how the reporting lags have changed in recent months across different jurisdictions, and the resulting estimates differ from other sources of provisional mortality data.  For now, these estimates should be considered highly uncertain until further evaluations can be done to determine the validity of these assumptions about timeliness. The true patterns in reporting lags will not be known until data are finalized, typically 11–12 months after the end of the calendar year. Importantly, these estimates are not a replacement for monthly provisional drug overdose death counts, or quarterly provisional mortality estimates. For more detail about the nowcasting methods and models, see:

Rossen LM, Hedegaard H, Warner M, Ahmad FB, Sutton PD. Early provisional estimates of drug overdose, suicide, and transportation-related deaths: Nowcasting methods to account for reporting lags. Vital Statistics Rapid Release; no 11. Hyattsville, MD: National Center for Health Statistics. February 2021. DOI: http://dx.doi.org/10.15620/cdc:101132.

In this visualization, users can select the outcome of interest (drug overdose, suicide, or transportation-related deaths) and the jurisdiction to display the weekly predicted numbers of deaths from 2016 through the most recent week available. Estimates for the United States exclude Connecticut and North Carolina, because those jurisdictions have historically had longer reporting lags. The estimates are therefore not nationally representative. The figures show four-week rolling averages. Figure 1 shows the trends over time. Figure 2 shows the numbers in 2020 and 2021 relative to the numbers in the same weeks from 2016–2019.

Suggested citation:

Rossen LM, Hedegaard H, Warner M, Ahmad FB, Sutton PD. Early model-based provisional estimates of drug overdose, suicide, and transportation-related deaths. Available from: https://www.cdc.gov/nchs/data/vsrr/vsrr011-508.pdf