regulation and compliance
How do I bill Medicare correctly for a pelvic floor evaluation and biofeedback in the same visit?
Evaluation complexity tiers, the 90911 and 97112 distinction, NCCI edits between manual therapy and biofeedback, and when a modifier is legitimate rather than a red flag.
You bill the evaluation once, at the complexity level your history and exam actually support, and you bill the biofeedback separately using the timed perineal biofeedback codes, provided the visit meets Medicare's coverage conditions for biofeedback and your documentation shows two distinct services rather than one service described twice. On an evaluation day, most pelvic therapists in an independent clinic end up with one untimed evaluation code plus one or two units of timed treatment, and the whole question of whether that claim survives is settled by three things: the code pair itself, the minute math, and the note.
The first correction to make is the code number in your fee sheet. CPT 90911 was deleted after 2019. Since January 2020 the perineal biofeedback service has been reported with 90912 for the initial 15 minutes of one on one contact and 90913 for each additional 15 minutes. If 90911 is still sitting in your practice management system, it is either being rejected or being crosswalked by your clearinghouse, and neither is a good way to run a clinic.
The second correction is conceptual. Biofeedback is not simply "we used a sensor." It is a distinct therapeutic modality with its own national coverage rules, and Medicare's expectation is that it follows a documented, unsuccessful trial of pelvic muscle exercise training. If the sensor is being used to give the patient a signal during an exercise you would have taught anyway, you are usually describing neuromuscular reeducation, not biofeedback.
What Medicare actually pays for under pelvic floor rehabilitation
There is no CPT code called "pelvic floor therapy." Medicare pays for a set of ordinary outpatient therapy services that happen to be delivered to the pelvic floor, plus one modality with its own coverage policy. In a small clinic your working set is short.
| Code | Service | Timed? |
|---|---|---|
| 97161 / 97162 / 97163 | PT evaluation, low / moderate / high complexity | No, untimed |
| 97164 | PT re-evaluation | No, untimed |
| 97110 | Therapeutic exercise | Yes, 15 minute units |
| 97112 | Neuromuscular reeducation | Yes, 15 minute units |
| 97140 | Manual therapy, one or more regions | Yes, 15 minute units |
| 97530 | Therapeutic activities | Yes, 15 minute units |
| 90912 / 90913 | Biofeedback, perineal muscles, anorectal or urethral sphincter | Yes, initial 15 then each additional 15 |
Everything on that list travels with the GP modifier, because it is furnished under a physical therapy plan of care. That plan of care needs physician or nonphysician practitioner certification, and recertification at least every 90 days. Independent clinics lose more money to unsigned certifications than to coding disputes, because the coding dispute is a fight you can win and the missing signature is not.
There is also an annual therapy threshold at which you attach the KX modifier to attest that continued care is medically necessary. The dollar figure is republished each year, so pull the current amount from your MAC rather than from a blog post, including this one.
Keep reading: What should I charge for a sixty minute pelvic floor visit if I go out of network?
Choosing among the three PT evaluation complexity levels
The three levels are not a judgment about how hard the patient was. They are driven by four defined components: the patient's history and comorbidities, the number of body systems examined, whether the presentation is stable or evolving, and the clinical decision making required. Time is a typical descriptor, not the determinant.
In pelvic health, a large share of genuine evaluations sit at moderate complexity. A postpartum patient with urgency, a second degree tear, one comorbidity and a stable presentation examined across musculoskeletal and neuromuscular systems is a defensible 97162. Reflexively billing 97161 because it feels safer is not conservative, it is inaccurate, and a chart full of low complexity evaluations paired with twelve visit plans of care reads strangely to anyone reviewing it.
When high complexity is real
97163 belongs to the patient whose presentation is unstable or evolving, with three or more personal factors or comorbidities that affect the plan. Chronic pelvic pain with a history of endometriosis surgery, central sensitization features, bowel and bladder involvement and a medication list that changes the exercise plan is a real 97163. Write the comorbidities into the history section so the level is visible on the page, not just in your head.
90912 biofeedback versus 97112 neuromuscular reeducation
This is where most small clinic audits start. Use a two question rule at the end of the session, before you code.
- Was an instrument used to display a physiologic signal back to the patient so she could learn voluntary control of it? If no, it is not biofeedback.
- Is there documentation, from you or elsewhere, of a trial of pelvic muscle exercise training that did not achieve the goal? If no, expect the biofeedback line to fail Medicare's coverage condition even when the technique was real.
97112 covers reeducating movement, balance, coordination, kinesthetic sense, posture and proprioception. Teaching a patient to find a contraction with tactile cueing, verbal cueing and a mirror is 97112 all day. Attaching a surface EMG sensor and using the trace as the teaching instrument, after exercise training has been tried and fallen short, moves that portion of the visit to 90912.
What you cannot do is bill both codes for the same minutes. If the sensor was on for ten minutes of a thirty minute reeducation block, then 90912 owns ten minutes and 97112 owns twenty. Split the minutes in the note before you split them on the claim.
Keep reading: Why do my postpartum patients stop doing their home program after the second week?
NCCI procedure to procedure edits that hit pelvic visits
The National Correct Coding Initiative publishes a quarterly file of code pairs. Each pair has a Column One code, a Column Two code and a modifier indicator. Indicator 0 means the pair can never be unbundled: if you report both, only Column One pays, and no modifier rescues it. Indicator 1 means a modifier may be used when the services were genuinely separate. Indicator 9 means the edit does not apply.
The practical instruction is not to memorize pairs. It is to download the current PTP edit file for practitioner services once a quarter, filter to the seven codes in the table above, and keep the result taped inside your billing binder. Edits change, and the version your billing software shipped with may be two years old. Also check the Medically Unlikely Edit value for each code, which caps units per day per patient and quietly denies the fourth unit you thought you had earned.
When modifier 59 or XU is defensible and when it is not
Modifier 59 says the two services were distinct. The X subset says how they were distinct: XE for a separate encounter, XS for a separate structure, XP for a separate practitioner, XU for an unusual non overlapping service. Many payers now prefer the specific X modifier, and using it makes your intent auditable rather than generic.
Defensible: the biofeedback block ran from 10:05 to 10:22 with the sensor in place, and the manual therapy for the abdominal wall ran from 10:25 to 10:40 with no instrumentation. Different minutes, different technique, different target. XU is honest here.
Not defensible: the two services overlapped, or you appended the modifier because a denial came back and adding it made the denial go away. A modifier applied after a rejection, with no change to the underlying note, is the single clearest pattern a payer looks for.
See how PelvicPath handles this for pelvic floor physical therapy
The eight minute rule applied to a mixed internal and external session
For Medicare, total the minutes of all timed codes, then convert the total to units: 8 to 22 minutes is one unit, 23 to 37 is two, 38 to 52 is three, 53 to 67 is four. Assign the units to the codes with the most minutes, and remember that the untimed evaluation contributes zero minutes to this total.
Worked example. A 53 minute treatment portion following the evaluation:
- 90912 biofeedback, sensor in place: 18 minutes
- 97112 neuromuscular reeducation, external cueing and breath coordination: 21 minutes
- 97140 manual therapy, internal soft tissue work: 14 minutes
Total timed minutes: 53. That is four units. CMS allocates by giving one unit to each code that has at least 15 minutes, then awarding any remaining unit to the largest leftover block. Each of the three codes clears 15 minutes, so each takes one unit, using three. The leftovers are 3 minutes of 90912, 6 minutes of 97112 and 14 minutes of 97140, so the fourth unit goes to 97140.
The claim reads 97162, 90912 x1, 97112 x1, 97140 x2, with GP on the therapy lines and the appropriate distinct service modifier on whichever pair your current PTP file flags. Note also that multiple procedure payment reduction cuts the practice expense portion of the lower valued units, so four units does not pay four times one unit. Build your schedule template around that reality rather than around the gross fee schedule amounts.
Documentation that supports the code you selected
Three habits carry almost all of the compliance weight in a small pelvic clinic.
Record start and stop times per intervention, not a single session total. This is what makes a distinct service modifier survive review. Record the failed exercise trial that justifies biofeedback, with dates, in the plan of care rather than buried in a daily note. And record objective change over time: sensor amplitude values, contraction hold seconds, leak episodes per day from the patient's own log. Skilled care is demonstrated by measurement, not by adjective.
Where the between visit record fits
The weakest link is usually the home program. You need evidence that pelvic muscle exercise training was genuinely tried before biofeedback, and you need objective interval data to justify continued visits at recertification. A patient who says "I did them most days" gives you neither.
PelvicPath was built for exactly that gap: it delivers the home program, logs each day the patient completes it, captures symptom flags between visits, and hands you a summary before she walks back in. That log is your exercise trial evidence, your progress measure and your defense of medical necessity, produced without adding a minute to the visit. Set it up on the evaluation day, and the documentation you need three months from now is already written.
Want to see what she actually did between visits?
PelvicPath sends the home program to your patient's phone, logs each completed session and flags the days that hurt, so the summary is waiting for you before she sits down.