Every diversion program has a long list of controls. Only a few of them change how soon you find out. The question this guide asks is narrow: of the controls a hospital or pharmacy can run, which three most shorten the gap between the first anomalous transaction and the day someone acts on it? The answer rests on one measured study and on federal rules, and it carries one honest limit: nobody has run a head-to-head trial of the controls against each other. Where we are giving an editorial ranking rather than a finding, we say so.

What the evidence can and cannot tell you

The best timing evidence we found is a 2022 paper in the American Journal of Health-System Pharmacy by Knight and colleagues. Ten acute care hospitals in four health systems supplied their medication transaction data, 27.9 million medication movement transactions by 19,037 nursing, 1,047 pharmacy and 712 anesthesia clinicians, covering 8 to 24 months. The data held 22 diversion cases that the hospitals had already detected by their own methods. The authors then asked, in blinded data, when their analytics would have flagged each case. The analytics flagged the known cases a mean of 160 days and a median of 74 days earlier than the hospitals' existing methods did, with a range of 7 to 579 days (Knight et al., AJHP 2022).

Three cautions come with those numbers. First, the study is retrospective: it measures when a model would have flagged a case, not when a hospital using it actually did. Second, it only includes cases the hospitals eventually caught, so it cannot speak to diversion nobody found. Third, the paper's first two authors list Invistics Corporation, a company that sells diversion analytics, as their affiliation, so the study tested a product its authors work for. None of this makes the timing result wrong, but it means one study should not carry the weight of a law of nature. A scoping review of 312 articles in the Journal of Hospital Medicine found that careful configuration of technologies and processes can reduce the opportunity for diversion, and also concluded that further evidence is urgently needed (Fan et al., J Hosp Med 2019, abstract).

The ranking below is therefore editorial. For each control we give the evidence, the rule or document it connects to, and what to measure.

Control 1: Reconcile across systems, on a short cycle

The Knight study describes what most hospitals do today. Detection relies on "historical reporting from automated dispensing cabinets (ADCs), such as monthly anomalous usage reports or daily discrepancy reports," and the authors note that these approaches "typically look back historically at data over several months, causing them to be slow to detect diversion" (Knight et al.). The method that did better was consolidation. The team pulled five kinds of data into one dataset: medication administration records, cabinet dispense and waste transactions, internal inventory movements into and within the central pharmacy, wholesaler purchases, and employee time-clock records.

Consolidation let them run reconciliations no single system can run. They compared wholesaler shipments with receipts in the inventory system, inventory moves from the central pharmacy with restocking of the cabinet, cabinet dispenses with the documented administration and any later waste, and clock-in times with the times of medication transactions. Each comparison corresponds to a stage where drugs can leave: in transit, in the vault, at the bedside, and off the clock.

You do not need a vendor to apply the principle. A community pharmacy can reconcile wholesaler invoices against received counts and the perpetual inventory every week rather than at the biennial count that 21 CFR 1304.11 requires at a minimum. A hospital can reconcile cabinet withdrawals against administration documentation on every shift in high-volume areas. What to measure: the interval between a transaction and the first time anyone compares it with a second record of the same event. If the answer is a month, that is your detection floor.

Control 2: Resolve every discrepancy quickly, by someone other than the person involved

The same paper says that, once suspected diversion was flagged, all participating hospitals began the same way: the investigator manually reconciled cabinet and medical record data, which took between 4 and 20 hours depending on the number of transactions by the clinician. With a consolidated dataset, users reported the same audit took 10 to 30 minutes (Knight et al.). The time cost matters because a discrepancy report that takes half a day to work will, in a busy week, wait. The longer a discrepancy waits, the harder it is to reconstruct what happened on the shift it came from.

Two practices follow. Resolve each discrepancy on a fixed clock, and assign it to someone who was not the person who dispensed, counted or administered. Our witnessed waste guide covers the second half of the same logic. Then track how many discrepancies remain open past your own deadline, because a rising count is an earlier warning than any single case.

Federal rules also put a clock on the other side of detection. A registrant must notify its DEA Field Division Office in writing of a theft or significant loss within one business day of discovery and file a complete DEA Form 106 within 45 days (21 CFR 1301.76(b)). The rule counts from discovery, so every day a discrepancy sits unresolved is a day the discovery date cannot be fixed. See our Form 106 guide.

Control 3: A reporting route that staff will actually use

In the Knight study the hospitals' existing detection relied on two techniques: the cabinet reports above, and asking clinicians to report impaired behavior by colleagues. People notice things data does not show. DEA's own rule says so in plain terms: an employee who knows of diversion by a fellow employee "has an obligation to report such information to a responsible security official of the employer," the employer must treat it as confidential and protect the employee's identity, and the employer must inform all employees of the policy (21 CFR 1301.91).

The rule makes the channel an obligation, but it does not make it work. What to measure: whether every employee has signed an acknowledgment of the policy, whether staff can name the person they would tell, and how many reports in the past year led to a record review. A channel that has received no reports in three years may mean there is nothing to report or that staff do not trust it, and the numbers alone cannot tell you which, so treat the silence as a question to investigate. See how to report drug diversion for staff-facing wording.

What we left off the list, and why

Locks, cameras, alarms and access control did not make the three, and that is not a claim they are unimportant. 21 CFR 1301.71 requires effective controls against theft and diversion, and its factors include supervision of employees with access and the adequacy of the registrant's system for monitoring the receipt, manufacture, distribution and disposition of controlled substances (21 CFR 1301.71). Physical controls reduce opportunity. The three above are about speed of discovery, which is a different goal. A program needs both, and our factor-by-factor guide maps the physical side.

Putting a number on your own gap

The study's method is one any program can copy on a small scale. Take your last confirmed diversion case, or the last significant discrepancy. Find the date of the first anomalous transaction in your records and the date someone acted on it. The difference is your time to detect. Do the same for each case and track the median, as the study did. If you have no cases, run the exercise on a drill: pick a drug, introduce a known gap in a test count, and time how long the reconciliation takes to surface it.

To see where your program stands on all three controls, run the hospital self-assessment.

Sources: Knight T, May B, Tyson D, McAuley S, Letzkus P, Enright SM. Detecting drug diversion in health-system data using machine learning and advanced analytics. Am J Health Syst Pharm 2022 (PMC9353695, open access); Fan M, Tscheng D, Hamilton M, Hyland B, Reding R, Trbovich P. Diversion of controlled drugs in hospitals: a scoping review of contributors and safeguards. J Hosp Med 2019 (PubMed 31251158; abstract read); 21 CFR 1301.71, 1301.76, 1301.91 (eCFR, current); 21 CFR 1304.11. Educational content, not legal advice.