Health equity

The patients who need monitoring most are the hardest to monitor

Remote monitoring quietly assumes a connected patient: a smartphone, home broadband, an app store account, English, steady hands, and the patience to pair Bluetooth. Chronic disease burden runs in the other direction. A program whose default kit assumes any of that has selected its patients before the first enrollment — so the default kit assumes none of it. Equity here is not a pledge page. It is a set of architecture decisions, and this page lists them.

The exclusion audit

How monitoring programs exclude without meaning to

Each assumption below is a filter. None of them appear in program brochures, and every one of them shows up in enrollment data.

Smartphone required
Drops: no-phone, prepaid-plan, shared-device households
The default kit is cellular. The device transmits on its own radio from the box. A patient app exists for those who want it; nothing clinical or billable depends on it.
Broadband and WiFi assumed
Drops: rural, tribal lands, unstable housing
LTE-M coverage with store-and-forward: a dead zone costs latency, not the transmitted day. Coverage is checked against the patient's address at enrollment, not discovered at the first missed month.
Accounts, portals, pairing
Drops: low digital literacy, the very old
Take the device out of the box and take the reading. Setup and education are the 99453 visit, done with a human. No password is ever between a patient and a transmitted day.
English only
Drops: LEP patients — 1 in 12 US adults
Enrollment materials, device instructions, adherence nudges, and patient-reported outcome collection run in the patient's language, written to plain-language reading level. Language is a configuration field, not a special request.
Vision, dexterity, hearing assumed
Drops: disabled patients — who carry the most chronic disease
Kit selection per program includes accessibility criteria: large-type instructions, voice guidance where devices support it, cuff and sensor alternatives for limited dexterity. The accessible option is a first-class kit, not an exception process.
Cost sharing disclosed late
Drops: fixed-income Medicare — after enrollment, at collections
Part B coinsurance on monitoring is real money on a fixed income, and a program that discovers this for the patient at billing time loses them and deserves to. Expected out-of-pocket is stated at consent, hardship is screened at enrollment, and the number is on the first page, not the last.
Structural, again

Built into the platform, not the brochure

Personal baselines instead of population references

Thresholds evaluate against the patient's own baseline wherever absolute numbers are unreliable. This is an equity control disguised as a clinical one: population reference ranges import population bias. The known case is pulse oximetry overestimating saturation in darker skin — so SpO2 alerts weight trend over absolute value, and every program using oximetry carries the documented limitation. How thresholds work.

The kit is the policy

Every equity commitment that costs money lives in the kit decision: cellular radio instead of an app dependency, accessible device variants stocked instead of special-ordered, multilingual materials printed instead of promised. The device fleet is where this is enforced — a program cannot activate with a kit that contradicts its enrollment population.

The funnel is stratified

Offered → enrolled → transmitting → alerts answered in clock, per program, stratified by consented demographics. Programs discover their exclusions in the data instead of in a complaint. Demographic capture is optional and refusal carries no penalty; unknowns are an explicit category on the dashboard, never a silent drop from the denominator.

Escalation does not triage by fluency

A threshold firing routes to a named owner with a clock regardless of whose chart it is. The response-clock report is one of the stratified views, because the failure mode worth catching is not enrollment bias — it is the program that enrolls everyone and answers some of them slower.

Where the gaps still are

Hardware measurement bias cannot be fully corrected in software, only mitigated and documented. Camera-derived measurement is labeled by the populations it has been validated in, and treated as context where evidence is thin. Cellular coverage is not universal, and store-and-forward narrows the gap without closing it. And stratified dashboards are only as complete as the demographics patients choose to share. A platform that claimed to have solved these would be lying; this one surfaces them.

The business case, stated plainly

Serving the excluded population is not charity economics

The patients these filters drop are the ones with the highest disease burden, the most avoidable admissions, and therefore the largest measurable deltas a monitoring program can produce. CMS is moving payment toward equity measurement, not away from it. And for sponsors, evidence generated in a real-world population that actually resembles the disease demographics is worth more than another homogeneous cohort. Building for the hardest-to-monitor patient is the durable strategy, not the concession.

Does a patient need a smartphone or home internet to be monitored?

No. The default kit is cellular: the device ships provisioned and transmits on its own radio from first use. No smartphone, no app account, no home broadband, no Bluetooth pairing. A patient app exists for those who want it, but nothing clinical or billable depends on the patient operating software.

What does remote monitoring cost the patient?

Under Medicare Part B, monitoring services carry the standard coinsurance, roughly 20 percent after the deductible, unless supplemental coverage picks it up. The device itself is covered by the device-supply codes, so there is no equipment purchase. Programs on Gathermed state the expected out-of-pocket cost at consent, before enrollment, and screen for hardship then, not at collections.

How does the platform handle pulse oximeter accuracy across skin tones?

Imperfectly, and it says so. Pulse oximetry can overestimate oxygen saturation in patients with darker skin, and no software fully corrects a hardware measurement bias. The platform mitigates: thresholds evaluate against the patient's own baseline rather than population reference ranges, alerts weight trend over absolute value for SpO2, and every program using oximetry documents the limitation for the clinical team.

Can a program see whether it serves all of its patients equally?

Yes. The same event spine that produces the billing audit trail produces a stratified program funnel: who was offered monitoring, who enrolled, who is transmitting, and whose alerts were answered inside the response clock. Demographic capture is consented and optional, declining carries no penalty, and patients who decline appear as an explicit unknown category rather than being silently dropped from the denominator.