Editorial correction — September 20, 2026. The earlier framing could imply that naloxone should be withheld when xylazine is suspected. That is incorrect. We also remove an undated seizure percentage and the claim that CDC tracks every overdose in real time.
If an overdose is suspected
Give naloxone if available, call 911, and stay with the person. Follow the dispatcher’s instructions. CDC guidance on xylazine explains that naloxone reverses opioid effects, although it does not reverse xylazine itself. Because xylazine may be mixed with fentanyl or other opioids, naloxone should still be administered. Continuing symptoms require emergency medical care.
Xylazine is a non-opioid veterinary sedative not approved for human use. It can slow breathing and is associated with serious skin wounds. Someone with concerning wounds should seek medical care. A person’s illness or drug exposure is not a measure of their moral worth.
The encouraging number, with its date attached
In its August 31, 2026 announcement, CDC reported a provisional predicted 67,798 overdose deaths during the twelve months ending March 2026, down 12.3% from the corresponding previous period. Provisional estimates can be revised; this is a dated estimate rather than a permanent final count.
The announcement introduced a resource bringing together fatal and nonfatal overdose surveillance and prevention information. It did not claim complete, real-time observation of every overdose. Different data systems answer different questions and operate on different reporting schedules.
What progress should ask of us
Fewer deaths should be welcomed without qualification about whether the people saved deserve our concern. My moral argument is that progress creates a responsibility to preserve what helps and investigate the harms that remain. An improving national total is not permission to overlook a family facing an emergency today.
It is equally important to communicate clearly. A headline about the limits of an antidote can become dangerous if readers mistake a limitation for a reason to do nothing. Public-health commentary should put the actionable instruction before the dramatic contrast.
How to read the next update
Check the period covered, whether the figure is provisional, and whether it counts deaths, nonfatal events, or substances detected. Do not compare those measures as if they described the same population. National improvement also cannot establish that every locality or group improved.
For a community discussion, ask which local services can provide overdose-response training, treatment referrals, and wound care. For an emergency, use emergency services rather than a trend article. The test of institutional care is not simply whether a dashboard looks better. It is whether someone in danger can reach effective help.