While well intended, a crude percent of patients who receive naloxone is an extremely poor measure of opioid adverse events; it's use and potential cross-hospital comparisons have a large potential for unintended consequences. First, the administration of naloxone is not by itself reflective of harm from an opioid. Fairview Health Services has had a system in place for over a decade in which every instance of naloxone administration is reviewed by a pharmacist. We have found that under 30% were truly opioid-related adverse events. The others proved to be situations where the patient was not doing well, naloxone was tried in an attempt to rule out narcotics as a culprit, and once the naloxone failed the physician moved on to something else. So at best, naloxone is a screen; if we want to understand the rates of harm, one needs to perform chart review. Using naloxone as a crude screen for public reporting and comparison purposes sets up several unintended consequences. One of this is that it would provide a perverse incentive for physicians to NOT use naloxone to assist with a differential diagnosis because each time they did, it would count as an adverse drug event even if the naloxone had no impact. That would set up a hospital to provide poorer care. Another unintended consequence relates to the clinical care of patients who indeed did have an opioid-related adverse event but which was minor. At Fairview, we have constructed our order sets in a way that allows for a relatively liberal use of naloxone; this is to help ensure that a minor adverse event doesn't escalate to a more serious one. If all of these events would be "counted against us" there would be an incentive against using naloxone early on and instead treat the patient without the use of the reversal agent. Not only is this sub-optimal care, but it increases the risk of delayed and more serious adverse events. Further, since every hospital has different thresholds and protocols for administering naloxone, cross-organization benchmarking and comparisons will be futile and inappropriate; yet hospital leaders will attempt to do so. If a different hospital has a lower rate, it may be because they are providing better care. But it also may be because they have different thresholds and could arguably be providing poorer care. In addition, the data criteria are unnecessarily complex. To require evidence of a narcotic dose having been administered only if the naloxone dose was given within the 1st 24 hours of admission but not afterwards creates complexity without value. At Fairview, we exclude doses administered in the Emergency Department with the assumption that almost all of those are related to patients having presented with the problem. For all others, we require evidence that an opioid was administered within a plausible timeframe of the dose of naloxone. For example, if the last dose of opioid was at 10:00 on Monday and it was a short-acting IV dose and the naloxone was given at 14:00 on Tuesday, it is not plausible that the issue was related to a dose of narcotic that was legitimately given by the hospital, even if the naloxone appeared to work. In such cases, it is possible that there was illicit use of an opioid and we do not count that as a hospital-based ADE. I have been an active member and past-chair of the Minnesota Hospital Association's (MHA) Medication Safety Committee, formed to implement the HEN/HIIN work as part of the Partnership for Patients Initiative. MHA has used a form of the measure that has been proposed (non-ED administrations of naloxone/1,000 patient days) as its standard metric for this topic. This was selected because the data are relatively easy to retrieve and report. But I can attest that the committee and participants have found little value in the data and acknowledge that a simple count of "triggers" is a poor and misleading surrogate for understanding and quantifying adverse drug events. Fairview believes that measuring opioid-related adverse drug events is vital. We applaud the efforts exemplified in this draft. But we strongly believe that, as with infection measures reported to NHSN, we need to be measuring true adverse events and not surrogate markers. We would be happy to discuss this at length at your pleasure.