A controlled substance discrepancy is any difference between what the records say should be on hand and what is physically there. Most programs handle each one well enough: someone recounts, finds the cause or doesn't, documents it and closes it. The weak point is what happens next. A discrepancy closed on its own tells you nothing about the last ten, and someone taking small amounts relies on exactly that.

This guide covers both halves: resolving a single discrepancy, and reviewing all of them together so that a pattern can be seen.

When a count does not match

Record the discrepancy before anyone adjusts the balance. Then work through the usual causes, in roughly this order:

  • a miscount, or a count of the wrong strength or form;
  • a dispense, return, waste or transfer that happened but was not posted, or was posted to the wrong patient or location;
  • a waste that was documented without the quantity, or without a witness;
  • a receipt that was put away but not entered, or entered but not put away.

If one of these explains the difference, document which one, with the transaction or record that proves it. “Count error” on its own is not a resolution. If none of them explains it, the discrepancy is unresolved and goes to whoever your policy names, usually the pharmacist in charge or the diversion program lead. It should not be adjusted and closed.

How quickly a discrepancy must be resolved is set by your state's rules and your own policy, not by a single federal deadline. A common policy is resolution before the end of the shift, while the people involved are still on site. Whatever your deadline is, write it down and measure against it. For the shift-by-shift count itself, see why controlled substance count reconciliation stops diversion early.

Why resolving each one is not enough

Someone diverting has every reason to keep each loss small. A small loss rarely gets escalated, and it can be put down to a miscount, a spill or a missed entry. Reviewing each discrepancy separately confirms that the record was corrected. It does not show whether the same thing has happened before, on the same shift, with the same people, on the same drug.

The DEA's own test for a reportable loss looks at the pattern too. Among the factors a registrant should consider when deciding whether a loss is significant, 21 CFR 1301.76(b) lists “a pattern of losses over a specific time period, whether the losses appear to be random, and the results of efforts taken to resolve the losses.” You cannot answer that without looking at discrepancies together.

Recording discrepancies so they can be trended

Free-text notes cannot be sorted. Record every discrepancy in the same structured format, with at least:

  • drug, strength and form;
  • location, cabinet or vault;
  • date, time and shift;
  • who counted, who witnessed, and who last accessed the stock;
  • size and direction of the difference (short or over);
  • the cause found, or “unresolved”.

If your dispensing cabinet or pharmacy system already logs discrepancies, check that the resolution reason is a fixed list rather than free text, and that “unresolved” is one of the options.

Reviewing the pattern

Set a schedule, put it in policy, and have the review done by someone who does not do the counts. Monthly is a common choice. Higher-risk drugs or areas may justify a shorter interval. Sort the same records at least three ways: by drug, by shift or day of week, and by the people involved. Compare against volume. A drug that moves a lot will produce more discrepancies than one that rarely leaves the shelf, so look at rates, not totals.

Things that should make you look closer:

  • The same shift. Discrepancies that keep appearing after one shift has handled a cabinet or vault.
  • The same person. One name that keeps appearing as the counter, the witness or the last user.
  • The same drug. Repeated small shortages in one product, especially an opioid or benzodiazepine.
  • Always short. Random counting errors go both ways. Shortages that are never offset by overages point to loss.
  • Vague resolutions. Discrepancies closed as “count error” or “resolved” with no record behind them.
  • The same pair. Counts that are always done by the same two people, or done unusually fast.

None of these proves diversion. Each is a reason to pull the transaction history, override and waste records and access logs for the people and period involved. Treat staff fairly while you do: a pattern can have an innocent cause, such as a workflow that one shift does differently.

Keep the review itself

Keep a record of each periodic review: the date, the reviewer, the period and data covered, what was flagged and what was decided. Keep it even when nothing was flagged. A record showing that you looked and found nothing is evidence that the review happens. DEA requires inventories and other required records to be kept and available for inspection for at least two years (21 CFR 1304.04(a)); your state may require longer. Keeping the trend reviews with the count records makes them easy to produce together.

When a discrepancy becomes a report

A registrant must notify the DEA Field Division Office in writing of a theft or significant loss “within one business day of discovery”, and file a complete and accurate DEA Form 106 “within 45 days after discovery” (21 CFR 1301.76(b)). There is no fixed quantity that makes a loss significant. The regulation lists factors to weigh, including the quantity in relation to the type of business, the specific drug, whether the loss can be tied to particular people, and the pattern quoted above.

In a hospital, losses of controlled substances must also be reported internally to the person responsible for pharmacy services and, as appropriate, to the chief executive officer (42 CFR 482.25(b)(7)). State boards of pharmacy set their own reporting rules, and some are stricter than the federal ones. Check yours. The DEA Form 106 guide covers the filing itself.

Frequently asked questions

Is every discrepancy a reportable loss?
No. A discrepancy you resolve with a documented cause is not a loss. An unresolved one may be, depending on the factors in 21 CFR 1301.76(b). If your review cannot rule out theft, involve your compliance lead or counsel promptly. The one-business-day clock runs from discovery.

How often should discrepancies be trended?
No federal rule sets an interval. Monthly is common, with a shorter interval for high-risk drugs or areas. The important part is that the interval is written down and kept to.

Who should do the trend review?
Someone who does not count or dispense the drugs being reviewed, so nobody is checking their own work. In a hospital that is often the diversion program lead or a pharmacist outside the area. In a small pharmacy it may be the owner or a pharmacist from another site.

Sources