There is no national rule for turning verified visit time into billable Medicaid units. Each state writes its own — sometimes several, inside one state. As a working example, this brief walks through four: Delaware, New Jersey, Pennsylvania, and Maryland.
Rounding is set three levels down. First by state — each Medicaid agency writes its own conversion rule. Then by program office within a state — Pennsylvania runs different rules for ODP, OLTL, and OMAP, and has said plainly it does not intend to align them. Then by payer within a program — Delaware applies one rule to fee-for-service hourly codes and a different rule to the same codes under managed care.
Across these four states alone there are four distinct regimes:
| State | Regime | Seconds | Published table? |
|---|---|---|---|
| Delaware | 8-minute rule, computed in seconds | Counted to the second | Yes — full seconds table |
| New Jersey | 8-minute rule, in whole minutes | Not addressed | Yes — minute thresholds |
| Pennsylvania (ODP) | Truncation — rounding prohibited | Discarded | Yes — minute bands |
| Pennsylvania (OLTL / CHC) | Set by program office + MCO | Not published | No |
| Maryland | State system calculates units | Not published | No |
The practical consequence: an agency operating in two of these states cannot run one billing configuration. And an agency using an alternate EVV vendor with a single global rounding setting is producing wrong units in at least one state.
Delaware (DMMA) is the only state of the four that computes units in seconds. For 15-minute unit codes, the aggregator applies an 8-minute rule in 900-second steps:
| Seconds | Elapsed time | Units |
|---|---|---|
| 0 – 479 | < 8:00 | 0 |
| 480 – 1,379 | 8:00 – 22:59 | 1 |
| 1,380 – 2,279 | 23:00 – 37:59 | 2 |
| 2,280 – 3,179 | 38:00 – 52:59 | 3 |
| 3,180 – 4,079 | 53:00 – 67:59 | 4 |
Because the boundary is 479/480 seconds, a visit of 7:59 pays zero units and a visit of 8:00 pays one. There is no minute-level rounding to soften that edge — a caregiver who clocks out four seconds early on a short visit loses the entire unit.
Hourly codes (T2013, S9123, S9124) split by payer. Under fee-for-service, the same seconds bands pay quarter-hours: 8:00–22:59 pays 0.25, 23:00–37:59 pays 0.50, up to 1.00 at 53 minutes. Under managed care, there are no partial units: under 53 minutes pays zero, 53–113 pays 1, 113–173 pays 2. Same code, same visit — a 40-minute T2013 visit pays 0.50 units under FFS and zero under an MCO. If billing isn't split by payer type, that difference is invisible.
Two more Delaware rules interact with rounding: units on the claim must match units the aggregator calculated, or the line denies; and a visit spanning midnight bills as one line, with all units on the date the visit began. Note the enforcement ("hard edit") start date has been published inconsistently across DMMA documents — confirm it with the payer before relying on it.
New Jersey (DMAHS, Newsletter Vol. 33 No. 11, effective October 2023) states its rule as minimum minutes per unit: 8 minutes for the 1st unit, 23 for the 2nd, 38 for the 3rd. Expressed as a table, it is the same 8-minute rule as Delaware — under 8 minutes pays 0, 8–22 pays 1, 23–37 pays 2, and so on. For one-hour units: 53 minutes earns the 1st unit, 113 the 2nd; 112 minutes rounds down to one unit. That matches Delaware's managed care hourly rule — New Jersey has no quarter-unit partial-hour billing.
What New Jersey does not publish is a seconds rule. The newsletter is written entirely in whole minutes; whether 7:50 counts as 7 minutes or 8 is an open question, not a settled rule. For an agency on the DE/NJ border, two payers whose written rules look identical can produce different unit counts on the same visit.
One more wrinkle: for DDD's Supports Program and Community Care Program, non-continuous sessions within a calendar day may be added together — the first unit must be a full 15 minutes, then remainders of half a unit or more round up. 53 minutes makes 4 units; 52 minutes makes 3.
Pennsylvania has no statewide table, and DHS has said directly it will not align OLTL and ODP billing rules. The only published table belongs to ODP, whose position is that rounding is not permitted at all: the aggregator truncates to completed units. For 15-minute codes, 14 minutes pays 0; 15–29 pays 1; 30–44 pays 2; 45–59 pays 3. For one-hour codes, 59 minutes pays 0. Seconds are discarded entirely — the exact opposite of Delaware.
Truncation makes span billing the recovery lever. ODP allows one claim line to accumulate continuous or non-continuous service time across up to 31 days; the aggregator totals the minutes, then converts. Three days of 25, 0, and 40 minutes billed as three lines pay 1 + 0 + 2 = 3 units; billed as one spanned line, 65 minutes pays 4. Under truncation, per-day billing forfeits partial units that spanning recovers — and nothing on any exception report will ever flag it, because no exception occurred. When claim units exceed verified units, the line denies outright (error codes 927/937); PROMISe does not cut units back.
Separate from rounding but on the same workflow: DHS monitors manual-edit rates quarterly, with recent vendor and MCO communications citing an 85%-verified-without-edits expectation and corrective action plans after two consecutive out-of-range quarters. The figure has moved before — verify against the live bulletin.
Maryland runs neither Sandata nor HHAeXchange. Its state-built system, LTSSMaryland with the ISAS component, generates the claim itself from the matched clock-in/clock-out pair — agencies do not independently compute units, and personal assistance claims may not be filed through eMedicaid. The failure modes are upstream: missed clocks, calls from phone numbers not authorized in the client profile, overlapping times, missing-time requests past the 30-day window.
What is published: the unit is 15 minutes (W5519, W5527 and related codes), capped at 48 units — 12 hours — per day, above which service converts to a daily rate; claims must be received within 12 months. What is not published: how ISAS converts elapsed time to units, or how it treats seconds. That question goes to the LTSSMaryland billing support office, not to a table.
Units produced by a 15-minute-unit visit of a given length. Bold marks where Pennsylvania ODP pays fewer units than Delaware and New Jersey for the identical visit.
| Visit duration | DE | NJ | PA — ODP | MD |
|---|---|---|---|---|
| 7 min | 0 | 0 | 0 | Confirm |
| 8 min | 1 | 1 | 0 | Confirm |
| 10 min | 1 | 1 | 0 | Confirm |
| 22 min | 1 | 1 | 1 | Confirm |
| 29 min | 2 | 2 | 1 | Confirm |
| 37 min | 2 | 2 | 2 | Confirm |
| 44 min | 3 | 3 | 2 | Confirm |
| 52 min | 3 | 3 | 3 | Confirm |
| 59 min | 4 | 4 | 3 | Confirm |
| 67 min | 4 | 4 | 4 | Confirm |
Delaware and New Jersey agree on 15-minute units — both are 8-minute-rule states — and diverge on hourly units and seconds. Pennsylvania ODP is materially harsher: every visit ending in the first seven minutes of a quarter-hour forfeits that unit, with no 8-minute grace. At a nominal $8 per unit, an agency running 3,000 visits a month where 15% land in the forfeited band is looking at a five-figure annual difference driven purely by which state's table applies.
Four failure modes, in order of how hard they are to see:
— Vendor pre-rounding. The EVV system rounds clock times before transmitting. Visit and claim then disagree by construction. The fix is configuration, not process.
— One configuration, multiple states. Delaware counts seconds; PA ODP discards them. Whichever way a shared vendor is set, one state is wrong on every visit.
— FFS versus managed care on the same code. The claim pays less than it should, or denies for reasons that look unrelated — nothing appears on the exception report.
— Per-day billing where span billing is permitted. Under truncation, every partial unit on every day is forfeited. The units were earned; they were never claimed. This is the rounded-unit loss in its purest form: revenue lost on visits that billed and paid cleanly.
Rules like these move — enforcement dates shift, thresholds change, program offices issue new guidance. Everything above is drawn from state-published documents current as of mid-2026; verify against the live payer documents before relying on any figure. Key sources: DMMA EVV program page, DMAHS Newsletter Vol. 33 No. 11, PA-DHS EVV FAQ and the ODP Comprehensive EVV Guide, and LTSSMaryland EVV resources.