The number that hides the work
"Patient produced /s/ in the initial position of words with 16 of 20 correct." It reads as solid progress. Whether it is depends on something the sentence leaves out: how much help each of those 20 attempts took.
Sixteen correct with no prompting and sixteen correct after a model of every word are different clinical facts. Pooled into one figure, they look the same, and the note can no longer say whether the patient is any closer to doing it on their own. The same problem, in a more dangerous form, shows up in swallowing, where a calm bedside exam is easy to write up as a safe swallow.
This article covers how a speech-language pathology record can keep both honest, and how ArkCloud Rehab enforces it.
Trial data: keep the cueing visible
A block of practice has a handful of facts: the target skill, what was presented, how many attempts were made, how many were correct, and how much support each attempt needed. The last one is the one that matters most and gets dropped first.
The principle ArkCloud is built on is simple: a response produced after a cue is not evidence of independent mastery. So the product treats it that way:
- Accuracy is reported for each cue level separately, and never pooled into a single headline number that hides the cueing.
- Independent accuracy comes only from unprompted trials. If no unprompted trial was run, independent accuracy is left blank, not shown as zero - because "independent performance was not measured" and "the patient got none right on their own" are different facts.
- A set where every attempt was cued is reported as cueing-dependent, and cannot be displayed as mastery.
The cue levels run from independent, through indirect and direct verbal cues, phonemic cues, model or imitation, visual or gestural cues, and tactile or phonetic placement, to maximal cueing. The clinician chooses the level for each block; nothing infers it from the target, the diagnosis or the last session. And the count of correct responses is the clinician's count. The product does not judge whether a response was correct.
Speech sound probes: a clinician listens
Articulation and phonology work produce a different kind of data: which sounds, in which positions of the word, and what the patient said instead.
In ArkCloud the clinician records the target sound, where it sat in the word, and their judgement of what was produced, with accuracy kept separate for each position. Speech recognition is not allowed to decide whether a sound was produced. A clinician listening is.
When the clinician records the sound heard in place of the target, the note shows the phonological patterns behind the errors, each counted against every opportunity the probe tested. An incomplete probe is not partly analyzed; the note says what is missing instead of drawing a pattern from half the data. The sound inventory today is General American English consonants.
Bedside swallow notes: what the exam cannot say
The most important limit in dysphagia documentation is this: a bedside swallow examination can neither establish nor rule out aspiration.
Silent aspiration - material entering the airway with no cough, no throat clearing and no other outward sign - is common in people with dysphagia, and a bedside exam sees none of it. A bedside exam with no overt signs is evidence of exactly one thing: no overt sign was seen. It is not evidence that the patient is not aspirating, and a record that reads that way can tell the next clinician a patient is safe when they may not be.
ArkCloud's bedside swallow record is built around that:
- Overt signs the clinician saw can be recorded: coughing or throat clearing, choking, gagging, a wet or gurgly voice after the swallow, a change in breathing, a fall in oxygen saturation, repeated swallows for one bolus, residue or pocketing, and loss of food or fluid from the mouth.
- Suspicion signs have their own place - a low-grade fever, rattling or rales on breathing, recurrent pneumonia - because they matter precisely when there is no overt sign at all.
- An aspiration finding, in either direction, requires an instrumental study. "Aspirates" and "does not aspirate" can each be recorded only with a videofluoroscopic swallow study (VFSS) or a fiberoptic endoscopic evaluation of swallowing (FEES). No number of bedside signs, no severity and no confidence level produces either one.
- Audio is not evidence. A recording or transcript of a swallow cannot support a swallowing finding, and the product has no way to use one as such.
It also does not decide a diet, a texture level or a nil-by-mouth order, and it does not grade severity. Those are clinical decisions with a named owner. The record holds the evidence they are made from.
The Medicare threshold for speech-language pathology
Speech-language pathology shares Medicare's yearly therapy threshold with physical therapy: the two are counted together, and once a Medicare patient's combined charges for the year reach it, further lines need the KX modifier. ArkCloud counts SLP with PT, suggests KX at that point, and never adds it automatically. How that works, and how timed units are checked, is in the 8-minute rule and the KX threshold.
Where the wording comes from
The pick-list wording follows common practice and has not yet had clinical sign-off for this product. Anything it does not cover is written in the note as usual.
That applies to the trial, cueing, speech sound and swallow pick-lists described above. The rules - per-cue accuracy, independent accuracy from unprompted trials only, and aspiration only from an instrumental study - are enforced by the product regardless of the wording.
ArkCloud Rehab is generally available. ArkCloud Rehab is $49.99 a month per organization, not per clinician. A clinic that also wants ArkCloud to build and track its insurance claims adds RCM Essentials at $499 a month, so the two together are $548.99 a month. Both prices are published on the pricing page.
See the full speech-language pathology page, read how visit notes are signed and locked, or request a demo to record an SLP session with sample patients. Related reading: documenting occupational therapy.