Custom home care software in Baltimore: the clock starts at the pill

The woman who made me write this article runs a residential service agency out of a converted rowhouse office in Towson. Twenty‑six aides, four people in the office, fifty‑odd clients spread from Parkville down to Brooklyn and out to Ellicott City. She has been doing it for nine years and she is very good at it.

She called me about invoicing. We spent about four minutes on invoicing. Then she said the thing that turned into this piece, which was that she had recently failed a survey finding on nursing supervision, and that she had failed it on a client whose file, as far as she could tell, was perfect.

The nurse had been to that house. There were notes. There was a signed care plan. What had happened was that eleven months earlier the client's daughter had moved to Delaware, and the aide — quietly, kindly, without telling anybody, because it was obviously the right thing to do — had started popping the blister pack and handing over the pills instead of standing there while he fumbled at them for ten minutes.

“Nobody made a decision. She just started helping more. And that moved my deadline by seventy‑five days, backwards, without anybody typing anything.”

That is the sentence I have not been able to put down. This is the thirty‑first trade we have taken apart in this series and it is the first one where the regulated interval is set by a verb. In commercial cleaning the tax attached to a room. In towing it attached to a boundary. In landscaping it attached to a label written by a manufacturer. Here it attaches to a gesture — to whether a hand opens or a hand receives — and there is no field for a gesture in any product in this market.

Let us start with the market, because home care in Maryland has a structure that almost nobody outside it knows about, and that structure is the reason the software question is strange here.

Seven hundred and forty-three businesses, fifty-one licenses

Home care is much larger in this state than its public profile suggests, and it is concentrated in a way that surprises people who assume it is a Baltimore City industry. It is not. It is a Baltimore County industry with a Baltimore City edge.

The six jurisdictions that make up the Baltimore metro — the City, Baltimore County, Anne Arundel, Howard, Harford and Carroll — hold 365 establishments and 10,978 employees between them, which works out at 30.1 people per business. That is the agency I priced everything against in this article: thirty on payroll, around fifty‑five active clients. It is a real shape and it is a hard one to buy software for, because thirty people is far too many to run on a spreadsheet and far too few to interest an enterprise vendor.

Baltimore City is the interesting row. It has the smallest average agency in the metro at 21.5 people, and the highest pay per employee in the entire state at $48,253 — twenty‑three percent above the Maryland average of $39,110 and sixty‑two percent above Harford's $29,754. That is not a story about generosity. It is a story about mix. The city is where the hospital‑affiliated skilled agencies sit, and a payroll dominated by registered nurses and therapists reads very differently from a payroll dominated by aide hours. If you run an aide‑hours agency in Baltimore City, the average in that row is not your business and you should not let a vendor benchmark you against it.

Now the number that reframes everything. Those 743 establishments are businesses in the ordinary sense. Here is how many of them Medicare will actually pay:

Fifty‑one certified agencies for seven hundred and forty‑three businesses — about fifteen to one. Twenty‑four of the fifty‑one are located in the Baltimore metro, against 365 establishments here, which is the same ratio to within a rounding error. The median certification year is 1985. Fifty‑seven percent were certified before 1990. The oldest, VNA of Maryland in Owings Mills, has been certified since 1 July 1966. And the newest certification in the entire state was issued on 10 February 2017 — nine and a half years ago, with nothing since.

That is not an accident and it is not a backlog. It is the Certificate of Need working exactly as designed, and it is the single most important commercial fact about this industry in Maryland. Under COMAR 10.24.16, the State Health Plan for home health agency services, you need a Certificate of Need from the Maryland Health Care Commission to establish a home health agency — and, separately, to expand an existing one into a jurisdiction it is not already authorized to serve. An acquisition, by contrast, does not require a CON as long as the type and scope of services do not change, and the buyer typically keeps the seller's license and Medicare certification number.

Read those two rules next to each other and the histogram above stops being a mystery. In Maryland you do not start a Medicare home health agency. You buy one. The 1966 certification is still in service because certifications here are inherited, not issued.

The residential service agency license is the other door, and it is wide open by comparison. An RSA is licensed by the Office of Health Care Quality under COMAR 10.07.05, requires no Certificate of Need, costs nothing to apply for and $1,000 for a three‑year term once granted. That is why the establishment count is fifteen times the certified count: the overwhelming majority of home care businesses in this state are RSAs doing personal care and private‑pay work, structurally outside the Medicare side of the market. OHCQ, for scale, oversaw 23,095 providers across 47 industries as of 1 July 2025.

So when a national platform's marketing talks to you about Medicare episode management and PDGM revenue optimization, understand what has happened: it is talking to about seven percent of the businesses in your industry, and you are probably not one of them. That is not the vendor being dishonest. It is a product built for a market whose shape does not match Maryland's.

The vendors, and what they are honest about

Before I take the pricing apart I want to be fair, because the products in this category are genuinely good and several of them do things I would not attempt to rebuild for anybody.

WellSky is the incumbent of incumbents and its Personal Care product, which most of the market still calls ClearCare, has more real‑world scheduling edge cases baked into it than any greenfield build would find in three years. HHAeXchange has done something structurally useful that deserves credit: it built the connective tissue between agencies and Medicaid managed care payers, and in states where that integration is mandated it is close to indispensable. AlayaCare has the best clinical documentation model I have looked at in this category. Axxess is the one I would point a Medicare‑certified agency at first, because its OASIS and PDGM tooling is deep and current in a way that only sustained investment produces. And Aaniie, which used to be Smartcare, deserves particular credit for something almost nobody else in this market does, which I will come to in a moment.

None of that is faint praise. If your problem is “I need to schedule two hundred visits a week, run EVV, and get a clean claim out of the door,” these products solve it, they solve it today, and you should rent one. I say that at the top because the rest of this section is going to be unflattering and I do not want the unflattering part mistaken for a recommendation to leave.

Here is the unflattering part. On 19 August 2026 I went to sixteen home care platforms and tried to find out what they cost.

Every meter in this market counts a noun

Six of the sixteen returned an HTTP 404 on their pricing URL: WellSky (both the main path and the personal‑care path), HHAeXchange, KanTime, CareVoyant, Medflyt and Careficient. Not a page saying “contact us” — no page at all. That matters more than it sounds, because the aggregator sites confidently quote prices for several of those products. WellSky Personal Care is variously listed at $12 per user per month, $40 per user per month and, for the Home Health product, $50 per user per month. For a thirty‑person agency those three third‑party numbers work out at $4,320, $14,400 and $18,000 a year — a spread of more than four to one, invented entirely by people who do not sell the software.

Of the ten that did serve a pricing page, most publish a philosophy rather than a price. Axxess shows features and a demo button. AlayaCare offers “value‑based pricing designed to adapt as you grow.” Alora states, disarmingly, that “pricing simply depends on how many people in your agency will use Alora software, or how many patients you serve,” which is a description of a meter rather than a rate. CareSmartz360 publishes four tier names banded by active client count — 0 to 100, 101 to 249, 250 and up, and franchisor — with a quote request behind each. AxisCare shows no figure. Rosemark publishes exactly one number, a $250 non‑refundable account setup and onboarding fee for Standard and Enterprise accounts, and nothing about the subscription that fee attaches to.

Two publish real, computable prices. Here is the whole market, applied to the same thirty‑person, fifty‑five‑client Baltimore agency:

Among the platforms that will tell you a number, the same agency pays anywhere from $2,880 to $8,580 a year — a three‑fold spread produced not by capability but by the choice of unit. ShiftCare counts staff, so its bill tracks your hiring. Aaniie counts clients, so its bill tracks your census. Those two numbers move in opposite directions in the situation every agency owner in Baltimore is currently living through, which is rising client demand against an aide shortage. Grow your census by ten and Aaniie charges you $1,560 more a year while ShiftCare charges you nothing. Hire five more aides to cover it and ShiftCare charges you $480 more while Aaniie charges you nothing.

Aaniie deserves the credit I promised it earlier. Its pricing page not only publishes rates, it explicitly says that providers can choose to be metered on clients, named users, average daily census, or service hours. That is the most honest thing anyone in this category does, because it admits out loud that the meter is arbitrary — that there is no natural unit of home care and the vendor is picking one.

Which is the whole problem in one line. Every meter in this market counts a noun: a client, a seat, a user, a census. Maryland's clock counts a verb.

Set that against what the work is actually worth. Genworth's 2024 Cost of Care Survey put the median annual cost of homemaker services in Maryland at $68,600, up 7.1 percent in a year, and Baltimore families budgeting private‑pay home care are generally working in a range of about $22 to $30 an hour. A fifty‑five‑client agency is moving somewhere north of two million dollars of care a year. Against that, the difference between $2,880 and $8,580 of software is a rounding error, and I want to say clearly that the subscription is not your problem. I am not going to argue you out of $8,580. I am going to argue that the $8,580 buys you a system that cannot see the thing that will actually cost you money.

The part no national platform models

Everything below is Maryland. Every one of these is a rule that changes what your schedule is allowed to contain, and not one of them is expressible in a product designed for fifty states at once.

Forty-five days, three months, four months

This is the one that started the article, and it is worth quoting rather than summarizing. COMAR 10.07.05.12E(2) requires that the registered nurse provide periodic, on‑site supervision of care:

“At least every 45 days if the staff administers medications to the client; at least every 3 months if the staff assists the client with self‑administration of medications; at least every 4 months if the staff does not administer medications or assist the client with medication self‑administration; or at a greater frequency established by the registered nurse due to the client's medical condition or clinical status.”

Four settings. Read the first two again slowly, because the distance between them is the whole point. Administers and assists with self‑administration describe the same eight o'clock, the same kitchen, the same two people and the same pill. What separates them is whose hand closes around it. And that distinction is worth seventy‑five days of legal deadline — forty‑five against one hundred and twenty if you slide all the way from the third setting to the first.

Now notice what is absent. There is no notification requirement attached to the transition. Nothing in the chapter says the aide must report that she has begun handing over the tablets, nothing says the office must be told, nothing generates a document. The care plan is supposed to say how and by whom services are provided, under .12C(4)(c), and a good agency updates it. But the regulation does not make the deadline contingent on the care plan; it makes the deadline contingent on what the staff does. So the plan can say one thing while the practice does another, and it is the practice that governs. The clock moves in the client's kitchen, silently, and the first anyone hears about it is a survey finding.

Every product I looked at models supervisory visits as a recurring task with an interval. Some let you set the interval per client. None of them let the interval be a function of the medication task recorded on the visit — which is what the regulation actually describes. The correct data model here is not a field called “supervision interval.” It is a derived value: look at what the aides on this client's roster have been documenting, pick the shortest applicable interval, and recompute the due date against the last completed on‑site supervision every time a visit note is filed. That is about a day of work if the visit notes carry a structured medication task. It is impossible if they carry free text.

Forty-eight hours, unless a nurse writes down why it is seven days

Assessment timing has the same shape and it is sharper still. Under .12C(2)(c) a client assessment must be completed within 48 hours of the start of services when the client requires any of nine listed things — wound and catheter care, treatment of stage three or stage four skin ulcers, ventilator services, skilled monitoring and aggressive medication adjustment where a fluctuating acute condition is present or risked, monitoring of a chronic condition not readily controllable, infusion therapy, specialized intravenous therapies or nutrition support, monitoring for high risk of complications that cannot be adequately managed, or a different level of care after notification of a significant change of condition.

Then .12C(3) does something I have not seen in any other Maryland chapter we have worked through. If the registered nurse determines, in the nurse's clinical judgment, that the client does not need the 48‑hour assessment, the nurse must document that determination in the client's record and ensure an assessment happens within 7 calendar days instead.

So the deadline is forty‑eight hours, unless a specific person writes down a specific opinion, in which case it becomes seven days. The document does not merely record the decision; it is the mechanism that changes the deadline. A compliance calendar that stores “assessment due” as a date cannot represent this. What it needs to store is a small state machine: which trigger fired, whether a nurse's override exists, who wrote it, when, and which of the two clocks is consequently running. That is not a hard thing to build. It is simply not a thing anybody has built, because in forty‑nine other states it is not a rule.

The waiver is only good if the client signed it

This is the one I would check first if I were you, because I think it is quietly the most common compliance defect in Maryland home care and it hides inside a perfectly reasonable assumption.

Under .10D and .10E, an agency may not provide or refer an uncertified caregiver to a client who needs help with activities of daily living or who needs medication administered, unless one of three conditions holds — and the third is that the client has signed a waiver of skilled services under .12D. Fine. Now read who is allowed to sign it. .12D(2) says a cognitively capable adult client, “but not a client representative with legal authority to make health care decisions,” may waive the recommendation of certified care services. .12D(3) says a cognitively capable adult client may waive recommended skilled care. And the definition at .02B(30) closes it: a waiver of services means an informed agreement by a competent adult client.

Maryland recognizes a long list of people who can make health care decisions for someone else — a guardian of the person, a guardian of the property, an advance directive, a surrogate decision maker, a power of attorney, a representative payee, and anyone the client designated while competent, all of them set out at .13A. Any of them can consent to a plan of care. None of them can sign this particular form.

A daughter with a fully valid power of attorney can agree to almost everything in her father's care and cannot agree to this. If she signs the waiver, the waiver is void, and every subsequent visit by an uncertified aide is an unlawful assignment.

The software consequence is precise. It is not enough to store a boolean for “waiver on file.” The system has to store who signed it, check that signatory against the client's own identity rather than against the representative list, and then let that answer gate the scheduler — because a waiver signed by the right person unlocks a whole class of assignment and a waiver signed by the wrong person silently does not. I have never seen a home care platform that models the signatory of a consent document as a scheduling constraint. I have seen a great many that store the PDF.

A cold does not count, unless it keeps coming back

“Significant change of condition” is a load‑bearing phrase in this chapter. It triggers a care note under .14D(1)(b), it can trigger the 48‑hour assessment under .12C(2)(c)(ix), and it has to be communicated to the client, the representative and the whole care team under .14C. So the definition matters, and the definition at .02B(27) is unusually interesting:

“Significant change of condition does not include any ordinary, day‑to‑day fluctuations in health status, function, or behavior or an acute short‑term illness, such as a cold, unless these fluctuations continue to reoccur.”

Read the last four words. Whether today's event is reportable is not a property of today's event. A cold in March is not a significant change of condition. The same cold in March, April and May retrospectively makes all three of them one. The definition can only be evaluated by looking backwards across the record, which means the correct implementation is not a checkbox on a visit form — it is a recurrence query over the client's history, run every time something minor is logged.

Here is the thing that makes me slightly evangelical about it: your agency already has this data. Every aide who has ever written “coughing again today” in a visit note has recorded an instance. Nobody has ever counted them, because counting them is not a feature anyone shipped. A system that noticed “this is the third respiratory note for this client in eleven weeks” and put it in front of the nurse would be doing exactly what the regulation contemplates, and it would be doing it from records you are already keeping.

One hour, any hour, with a log

On‑call is a regulated service in Maryland, not a courtesy. .12G requires the agency to be reachable by a communication device 24 hours a day, seven days a week, to route inquiries appropriately, and to respond not later than one hour after the inquiry is received — or sooner where medical standards require it. It also requires a log, and the log has three mandatory columns: the identity of the response personnel, the content of the inquiry, and the time of each inquiry.

That is a specification, and it is a specification that a shared mobile number and a group chat cannot satisfy. What it describes is a small ticketing system with an SLA clock attached: inbound record, assignment, response timestamp, and a durable record of who answered. Most of the agencies I have talked to in Maryland run this on a personal cell phone belonging to whoever is on the rotation, which means the log exists in that person's call history and nowhere else, and disappears when they leave. It is the single cheapest thing on this list to fix properly and the one most likely to be cited.

The county line is the license

If you hold only an RSA license, geography is a business decision and you can serve whoever you can reach. If you are on the Medicare side, geography is a licensed asset. Under COMAR 10.24.16 a home health agency's Certificate of Need authorizes it to serve named jurisdictions, and expanding into one more requires a fresh CON — for which the Commission first requires the applicant to demonstrate that its existing agency has hit the required quality performance levels. The Commission's stated policy position is that each jurisdiction should have no fewer than three operating agencies, which is the trigger that opens a review cycle at all.

For a Baltimore agency this is sharper than it is anywhere else in Maryland, for a reason that has nothing to do with health policy. Baltimore City is an independent city. It is not inside Baltimore County and never has been; they are two separate jurisdictions with two separate governments, and the CON qualifying lists name them separately. So an agency authorized for Baltimore County and not Baltimore City can take a referral in Cockeysville, twenty‑odd miles out, and must decline one three blocks the other side of Northern Parkway.

What that demands from software is a jurisdiction check at intake, run against the client's address, resolved to the actual civil jurisdiction rather than to the mailing city — and the mailing city is precisely the trap, because a very large number of addresses in Baltimore County are addressed “Baltimore, MD” with a 212xx ZIP. The word “Baltimore” on an intake form is not evidence of anything. A system that validates service area from a ZIP list will be wrong on the county line, and the county line is where a third of this metro lives.

Four minimum wages inside one dispatch radius

Maryland's minimum wage is $15.00 an hour for all employers. That is the floor in Baltimore City, Baltimore County, Anne Arundel, Harford and Carroll. Howard County reaches $16.00 for all employers on 1 July 2026. Prince George's County has been $15.30 since 1 January 2026. Montgomery County reaches as high as $18.00 for large employers on 1 July 2026.

In most industries this is a nuisance and no more, because most businesses have a worksite. Home care does not. The worksite is the client's living room, and a single aide can work a morning in Catonsville and an afternoon in Columbia. The wage floor is therefore a property of the visit, not of the employee — and payroll settings in national platforms are almost universally a property of the employee, or at best of a branch. An agency working the Howard County line and paying its people one rate all day is either overpaying on half its book or underpaying on the other half, and only one of those two errors is merely expensive.

The letter the regulator wrote about your payroll

In November 2022 the Office of Health Care Quality published a guidance document with a title that does not leave much to interpretation: Understanding how Maryland's employee protection laws apply to residential service agencies and personal care aides. Its opening line is that Maryland's RSAs “sometimes wrongly classify PCAs… as independent contractors rather than employees,” and that when this happens “it is called worker misclassification and it is illegal.”

The document then works through three separate legal tests and reaches the same conclusion under each. Under the wage and sick leave laws, because the agency controls the work and pays hourly, aides are employees. Under the unemployment insurance law, an aide is presumed an employee unless the agency can satisfy the ABC test, which it generally cannot because personal care is the type of work the agency is in the business of performing. Under workers' compensation, the common‑law test points the same way. And it is explicit that the IRS having accepted a classification does not help you, because Maryland's statutes are, in its words, “humanitarian statutes… more favorable to employees.”

The stakes are set out just as plainly: a court may order treble damages — up to three times the wages that should have been paid — plus the aides' attorneys' fees, and individual owners of the corporation may be held personally liable, putting personal assets at risk. This is not theoretical enforcement risk. It is a regulator telling an entire industry, in writing, what it expects to find.

And it has already reached the license. The current RSA application requires that “an individual with authority over the RSA's pay or employment practices” complete an online RSA Certification form about the classification of personal care aides — and it requires it of all applicants, including those who do not plan to hire aides at all. Worker classification is now a licensing condition, not merely a labor matter. Which means the way your scheduling system assigns work — whether it directs, whether it controls, whether it sets the hours — is producing evidence about a question your license depends on.

What Maryland charges you to be licensed

The licensing arithmetic is worth having in one place, because it is unusually front‑loaded in effort and unusually back‑loaded in cost.

Applying is free. There is no application fee for an RSA license at all. What the application asks for instead is work: a business plan demonstrating financial or administrative ability, containing a one‑year operating budget, a marketing plan identifying the populations to be served and a detailed description of services; an organizational chart; a full set of policies and procedures covering administration, personnel, patient care, informed consent, and environment and safety; a letter of good standing from the Maryland Department of Assessments and Taxation; and workers' compensation evidence, or a certificate of compliance, or a letter of exemption. Applications are held 180 days from initial receipt and then administratively closed, so a stalled application has a shelf life.

Once granted, the license costs $1,000 for a three‑year term and $1,000 again to renew for another three, with an annual data collection survey required in the form and manner prescribed by the Secretary. And then there is the line that people miss, at COMAR 10.07.05.04K: if an agency fails to comply with the chapter and the Department needs more than one on‑site prelicensure visit, the Department may charge $250 per additional visit — or deny the license. Those are the two options the regulation gives it, in that order, in one sentence. Not being ready when the surveyor arrives the first time is a priced event, and the price is the smaller of the two things that can happen to you.

Set that against the compliance rules above and the economics become clear. The license is cheap and the evidence is expensive. Every item in the previous section — the 45‑day clock, the 48‑hour assessment, the signatory of the waiver, the recurrence test, the one‑hour on‑call log, the weekly care notes required by .14D(1) on admission and at least weekly and on any significant change and whenever the plan is modified — is a documentation obligation. You are not paying Maryland for permission. You are paying, in staff hours, for the ability to prove things afterwards. That is a software problem wearing a compliance costume.

What custom actually costs

We publish our prices, so here they are against the problem described above.

The comparison I would actually think about is not against the subscription. It is against the survey. An Operations System is $12,000 once, and you own the source code. A single repeat visit at $250 is trivial next to it; a plan of correction that consumes your director of nursing for three weeks is not, and a licensing action is in a different category altogether. I am not claiming software prevents deficiencies. I am claiming that most of the deficiencies in this chapter are failures to notice a date, and noticing dates is the one thing computers have always been good at.

What we would actually build

Not a scheduling platform. I want to be emphatic about this, because it is where money goes to die in this industry. ShiftCare is $8 per staff member per month and Aaniie is $13 per client, and both of them have years of scheduling edge cases in them — overnight shifts, split shifts, mileage, EVV, live‑in arrangements, cancellations at six in the morning. Rebuilding that against products at that price would be indefensible and I would tell you so on the call.

Nor would we rebuild EVV. Maryland's Medicaid personal supports program runs on the In‑Home Supports Assurance System, a telephony‑based check‑in and check‑out system where the caregiver clocks in from the participant's own landline or cell phone, or from a one‑time‑password token assigned to that home. That is a state system with a state call‑in number and you connect to it; you do not reimplement it. It is also, incidentally, another instance of the same pattern this whole article is about: the verification is anchored to the place rather than to the worker, which is why generic mobile clock‑in apps do not satisfy it.

What we would build is the layer above the scheduler that knows the Maryland‑shaped facts, and there are five of them. A supervision engine that derives each client's interval from the medication task recorded on recent visits rather than from a setting, and recomputes the due date on every note. An assessment state machine that models the 48‑hour trigger list, the nurse's documented override and the 7‑day fallback as one object with an author and a timestamp. A consent register keyed to the signatory, not the document, wired into the scheduler so an assignment an uncertified aide is not permitted to take simply cannot be made. A recurrence watcher that reads care notes and raises the “continues to reoccur” question before a surveyor does. And a jurisdiction resolver at intake that turns an address into a civil jurisdiction, checks it against your authorization, and attaches the right wage floor to every visit that follows.

All five read from the scheduler you already pay for and write back to it. None of them replaces it. That is deliberate: the parts of this business that are the same in Maryland as in Michigan should be rented, and the parts that are not should be owned.

Build, buy, or both

If I had to compress nine years of somebody else's operational experience and a week of reading COMAR into advice, it would be this:

• Keep renting the scheduler and the EVV. At $8 per staff member or $13 per client this is the best value in your entire cost base, and the connection to Maryland's ISAS is not something to home‑build.
• Own the clocks. The 45/90/120‑day interval, the 48‑hour and 7‑day assessment pair, and the weekly care note cadence are the three things a survey will actually test, and no product sold in fifty states will ever derive them correctly for you.
• Own the consent register. Store the signatory, not the signature, and let it gate assignments — because a waiver signed by a power of attorney looks identical in a filing cabinet and is worthless in a file review.
• Resolve jurisdiction at intake, once. Address to civil jurisdiction, then service authorization and wage floor both hang off it. Never infer a county from a mailing city in this metro.
• Put the on‑call log somewhere that is not a phone. It is the cheapest fix on this list and the one that vanishes the day the person holding the rotation resigns.

If you take one thing from this article, make it the supervision engine, because it is the only item here where the deadline can move without anybody deciding anything — and a deadline that moves by itself is the definition of a thing software should be watching. If you take two, make the second the consent register, because it is the failure that looks most like success right up until someone reads the signature line. And if you take three, make the third the question that agency owner in Towson asked me, which I still think is the most precise description of an operations problem anyone has given me in this series: nobody made a decision, she just started helping more. Maryland has an opinion about that. Your software should have one too.

Who we are

We are a small studio in Baltimore. We build custom web apps, online stores and operations systems at fixed prices, we hand over the source code, and we tell people to keep their existing software when keeping it is the right answer — which, in home care, it almost always is for scheduling and EVV, and almost never is for the compliance calendar.

If you run a residential service agency or a home health agency anywhere in the metro and any of the above sounded like your last survey, book a free call. Bring your client list and your last plan of correction. We will work out which of your clients are on the wrong supervision interval today, how many of your waivers were signed by the wrong person, and what we would build — with the fixed price attached before you decide anything.

This article describes Maryland law and regulation as we read it in August 2026 and is not legal, regulatory or clinical advice. COMAR 10.07.05, COMAR 10.24.16, Md. Health‑General Article Title 19, and Maryland's wage, sick leave, unemployment insurance and workers' compensation statutes all change, and county minimum wage rates are indexed and move annually. The classification of any particular worker depends on facts specific to your agency. Establishment data is County Business Patterns 2023 (NAICS 621610) and counts only establishments with paid employees, which understates an industry containing many sole proprietors; the Maryland total includes five establishments the Census Bureau reports statewide without allocating them to a county. Certification data is the CMS Provider Data Catalog home health agency file read on 19 August 2026, and a Medicare certification date follows the certification number through changes of ownership, so it records when a certification was first issued rather than when a business was founded. Vendor prices were read from public vendor pages on 19 August 2026 and change without notice; where a vendor published no figure we have said so rather than quoting an aggregator. ShiftCare publishes its rates in dollars without a currency code and serves several markets, so the figures above are those shown to a US visitor on that date. Verify your own position with the Office of Health Care Quality, the Maryland Health Care Commission, the Maryland Department of Labor, or your own advisers before relying on anything here.