90833 Is Not a Growth Lever
Three boxes decide whether the add-on exists at all. Any no, and it's off the table.
- BOX 1
- SEPARATE THERAPY HAPPENED?
- BOX 2
- 16 MINUTES OR MORE?
- BOX 3
- IN THE NOTE?
- ANY NO
- → OFF THE TABLE
Composite. Not your chart. Not billing or legal advice, your MAC and your payer are the source.
Feasible set
Decision analysisBefore optimising anything, establish which options legitimately exist.
Optimising has two stages, and almost everybody skips the first. Stage one is: which options actually exist? That set has a name — the feasible set. Stage two is picking the best one in it.
Skip stage one and you do not waste time. You produce confident, wrong answers.
Composite. The add-on was worth about $1,680 a month on paper. That is real money, and it is why the argument keeps happening. So I wrote the conditions down as three boxes. Did the separate work happen? Was it long enough? Is it in the record? Three boxes. Tick, or do not.
Constraints stack. Every one has to be true at the same time. Three yeses and the option exists. One no and it does not, however good the other two look. There is no partial credit. $1,680 × 0 is 0.
This is why everybody does it is not evidence. Popularity tells you what people chose. It tells you nothing about whether the option was ever in the set. A whole industry can optimise outside its own rules for years, right up until somebody checks.
The fix is cheap and unglamorous. Write the conditions as boxes. Mark each one honestly. Then argue about the money. The most expensive mistakes I have watched were all stage two arguments, held by people quietly failing stage one.
Go deeper: three things a school would add
- Write constraints as tests with evidence attached, not as beliefs. "It happened and is documented" is a test; "it is generally accepted" is not.
- Sort hard constraints (rules, law, physics) from soft ones (preference, habit) before arguing. Soft constraints are negotiable; hard ones are the boundary.
- When an option would be very valuable if feasible, attack the binding constraint directly — do not quietly assume it away.
Work the case
Two decisions. Choose before you read the reasoning. The wrong answer is the one worth understanding.
An option satisfies two of three required conditions, and it would be worth a great deal.
Choose one — the reasoning opens after you commit.
Everyone in your peer group does the thing routinely, and none of them has ever had a problem with it.
Choose one — the reasoning opens after you commit.
Box 1: separate therapy happened? Not "we talked about her mom." A psychotherapy service, something you'd recognize as therapy if the med part was deleted.
Box 2: sixteen minutes or more? 90833 is the thirty-minute add-on; the floor is sixteen.
And those sixteen are therapy minutes. The med review, the refill, the side-effect check, those minutes don't count.
You don't get to count them twice.
Box 3: in the note? Time, what you did, separate from the medical part, so a stranger can find both.
Then the red line: any no, off the table. Not "probably fine." Off the table.
Schools call this the feasible set: the wall you optimize inside. You don't move the wall. You don't vote on the wall.
Growth is what you do inside the boxes; the boxes aren't the growth. You're treating a billing rule like a marketing tactic, but a lever is something you pull harder, and a rule is something you're inside of.
The question is never "can I add it." The question is "did it happen?"
Now the math they showed you. Eighty visits, seventy dollars, fifty-six hundred. Audit it: composite, but this thread runs every week, twenty-four of those charts pass all three boxes.
24 × 70 = 1,680. That's the lever. It was always 1,680. The other 3,920 isn't revenue.
It's exposure with a recoupment letter attached, and that sentence belongs to a lawyer.
And mind the slot. A thirty-minute slot barely holds a medical visit and sixteen-plus minutes of a separate therapy service, fourteen minutes for the medical part, every patient, every time.
It's legitimately billable when it happened (MDM has no time floor). But if every slot says both, the schedule is heroic or lying.
The question, as it actually gets asked
Composite questions — blended from the operator rooms, never one person's words.Read these for the method, not the answer. The next question you have will not be on this list — the point is that the concept above answers it anyway.
Which concept this really isAn add-on isn't a lever you switch on; it's an option that either exists for this encounter or doesn't. Every condition has to hold at once.
How you work it outRun the boxes before the arithmetic: was the service actually furnished, was it separate and identifiable from the medical work, is it timed and documented as such. Any single no and the option isn't in the set, whatever the potential revenue.
The mental model that makes it hardA code as a growth strategy. This is the highest-consequence pattern in the whole archive: confident universals travel further than the corrections do.
The Monday rule, what do you actually run this week?
The question is never "can I add it." The question is "did it happen."
Connected lessons
- two-scoreboards Once the add-on exists, the medical part changes scoreboards — that is the next page.
This is one of eleven lessons, free to read. One arrives by email each week, on composite numbers you can check.