The CDC has acknowledged its yearly estimates of nonfatal gunshot injuries are unreliable, due in part to the low number of hospitals in its data set, according to FiveThirtyEight, which publishes data-driven news and analysis.
The CDC must "improve the precision and accuracy of [its] non-fatal firearm injury estimates," CDC Director Robert Redfield wrote in May, when 11 senators inquired into the estimates' accuracy. The agency has now hidden the 2016 and 2017 gun injury figures on its public data portal because they are "unstable." The portal will also start providing statistical information about data reliability.
The changes come after FiveThirtyEight and The Trace, a nonprofit news organization covering gun violence, highlighted the estimates' unreliability. They showed the CDC's current model is highly sensitive to changes in its sample of hospitals. Adding just one new hospital to the database increased the number of nonfatal gun injuries by 22,000 in the 2015 national estimate, for example, accounting for over one-fourth of total estimated gunshot injuries for that year.
"The influence of a gradually changing roster of participating hospitals does not translate to poor data quality, but rather reflects the varying characteristics of these hospitals," a CDC spokesperson told FiveThirtyEight. The CDC is exploring ways to improve its estimates' reliability, including collecting data from more hospitals, the spokesperson said.