The first two or three diagnosis codes on a chart get real scrutiny. They're the ones driving the primary clinical picture, and everyone knows to check them carefully. Further down the list, the fifth, sixth, seventh secondary code, the scrutiny tends to quietly disappear, even though those codes can carry just as much financial weight.

Here's why that happens, and why it's worth fixing.

Why attention drops off further down the list

PDGM allows a lot of secondary diagnosis codes on a single chart, often well over twenty, but not every one of them carries equal weight in the moment they're being coded. The primary diagnosis sets the clinical group. The first few secondary codes often capture the most obvious comorbidities. By the time a coder is working through the sixth, seventh, or eighth secondary diagnosis, there's a natural drop-off in scrutiny, not from carelessness, but because human attention doesn't distribute evenly across a long list, especially under real chart volume pressure.

The problem is that comorbidity adjustments don't work the same way attention does. A code sitting sixth or seventh on the list can be exactly the one that triggers a comorbidity adjustment tier, or that pairs with another code elsewhere on the chart to trigger a high adjustment through interaction, and it's just as easy to under-specify or miss entirely as any other code.

Where this actually costs money

The most common version is a code carried forward from a referral packet without being revisited for specificity, sitting far enough down the list that nobody circles back to check whether it's coded as precisely as the documentation actually supports. An unspecified version of a code that's sitting sixth on the list is just as capable of losing a comorbidity adjustment as an unspecified primary diagnosis, it's just far less likely to get a second look.

The other version is simpler: a real, clinically relevant condition that's documented somewhere in the chart but never makes it onto the coded list at all, because by the time a coder reaches that part of the review, the obvious diagnoses are already accounted for and the process starts to feel complete before it actually is.

Why this is an easy problem to miss and a straightforward one to fix

This isn't really a training issue, it's an attention-distribution issue, and it responds well to a structured process rather than relying on a coder's focus holding steady across every single line. A review step that checks every secondary diagnosis with the same rigor, regardless of where it sits on the list, catches what naturally gets missed when scrutiny fades further down.

See how this shows up on your own charts

We'd like to walk through how our coding process gives every secondary diagnosis the same level of review, not just the first few. Olli's clinical AI is built to read through an entire referral packet, hundreds of pages if that's what it takes, and surface every relevant secondary diagnosis it finds, not just the ones near the top. It doesn't just find them either, it helps sequence them in the order that's both compliant and actually maximizes appropriate revenue, so nothing valuable gets buried simply because of where it happens to land on the list.