Denials and underpayment in drug-heavy practice

Standard denial management assumes claims are roughly comparable in value. In an infusion practice they are not, and that single fact changes how the function should be run.

Rank by dollars, always

A denial report sorted by count will put office-visit rejections at the top and a denied high-cost drug claim somewhere in the middle. Sorted by dollars, the ranking inverts entirely. Run both, and let the dollar ranking drive the work.

Hard denials are inventory losses

In a buy-and-bill model the drug was purchased before it was billed. A hard denial is therefore not deferred revenue — it is a realised loss on goods already consumed. Track hard-denial dollars separately and treat the figure as what it is: the direct cost of upstream process failure.

Root-cause categories specific to this work

• Authorisation absent or expired — including the extension case, where the initial approval lapsed mid-course.
• Site-of-care mismatch — therapy delivered in a setting the payer did not approve.
• Drug unit or wastage reporting — units billed not matching documentation or payer rules.
• Medical necessity on the antimicrobial itself — where the payer's criteria for a specific agent were not evidenced.
• Administration coding — hierarchy, initial versus subsequent, missing documented times.
• Standard front-end causes — eligibility, coordination of benefits, demographics.

The silent one: underpayment

A drug claim paid at less than the contracted amount produces no rejection and no alert. Without loaded fee schedules the error is not rare — it is invisible. Given the value of individual drug claims, this is the highest-leverage single control an infusion practice can implement, and it is a systems configuration task rather than a labour one.

Running it

Monthly: hard-denial dollars by root cause, one category assigned to the process step that owns it, with a target and a date. Quarterly: a payment-variance review against contracted rates across your highest-value drug claims. That combination finds both the failures that shout and the ones that do not.