Sign Up
Login
Try for Free
Login
The OASIS (Outcome and Assessment Information Set) is the assessment item set CMS requires of Medicare home health agencies, in force as OASIS-E2 since April 1, 2026 and collected at start of care, resumption, recertification, transfer, and discharge. Home health nurses and therapists complete it, and its responses drive PDGM payment, star ratings, HHVBP, and survey findings. This page covers how to document a defensible OASIS, with a fictional sample.
The assessing clinician: a registered nurse when skilled nursing is ordered, a physical therapist or speech-language pathologist in a therapy-only case, and since January 1, 2022 an occupational therapist when OT is ordered with PT or SLP and no nursing; licensed practical and vocational nurses, therapy assistants, social workers, and aides may contribute through collaboration but cannot own the assessment
Agency clinical managers and OASIS review or quality assurance staff, coders and billers who derive the HIPPS code, the physician or allowed practitioner who signs the plan of care, CMS and state surveyors, Medicare Administrative Contractors and other medical reviewers, iQIES and the quality measures built from it, and families reading Care Compare
400 to 900 words of supporting visit narrative beside the item set at start of care, resumption, and recertification, a few lines at transfer and death at home · 60 to 120 minutes in the home at start of care (clinical team estimate)
Federally required structured assessment item set (CMS data specifications, encoded and transmitted through iQIES) collected at fixed time points and read together with the clinician's narrative and the plan of care; a data set with coding conventions, not a scored scale
Start of care within 5 calendar days of the SOC date, resumption of care within 2 calendar days of return home, recertification on days 56 to 60 of each 60-day period, other follow-up after a significant change in condition, and transfer, death at home, and discharge within 2 calendar days; all payers since July 1, 2025, except patients under 18, maternity-only cases, and personal care or chore-only cases
Required by the Medicare home health Conditions of Participation (42 CFR 484.55 and 484.45) and interpreted through the CMS OASIS-E2 Guidance Manual and quarterly Q&As, effective April 1, 2026; a CMS instrument in the public domain, United States only; described here for documentation, item text and response options not reproduced
The Outcome and Assessment Information Set is the standardized set of patient assessment items that every Medicare home health agency must collect, encode, and transmit to CMS as part of the comprehensive assessment required by the Conditions of Participation. It was developed in the 1990s by the Center for Health Services and Policy Research at the University of Colorado for the agency now called CMS, became a condition of participation in 1999, and has been revised in numbered versions since: OASIS-C (2010), C1, C2, D (2019), D1 (2020), E (January 1, 2023), E1 (January 1, 2025), and OASIS-E2, which took effect on April 1, 2026 for every assessment with a start of care, resumption of care, recertification, transfer, or discharge date on or after that day and is the version in force as of September 2026 (OMB control number 0938-1279, expiration December 31, 2028); no later version has been announced. Named here by item number and topic only, the E2 item set opens with administrative and time point items (M0080 assessing discipline, M0090 date the assessment was completed, M0100 reason for assessment, M0102 and M0104 ordered and referral dates) and then runs through sensory status (B0200 hearing, B1000 vision, A1110 language), cognition and mood (the C0200 to C0500 mental status interview, C1310 signs of delirium, the D0150 mood interview), functional abilities and goals (GG0130 self-care and GG0170 mobility), health conditions (M1400 shortness of breath, M1033 risk of hospitalization), skin (M1306 and M1311 pressure ulcer and injury items), medications (M2001 drug regimen review, M2003 follow-up, M2005 at transfer or discharge, M2020 oral medication management, N0415 high-risk drug classes), the legacy activity items M1800 to M1870, and the transfer and discharge items J1800 and J1900 (falls) and M2401 (intervention synopsis). The E2 revision replaced the gender item with A0810, replaced the A1250 transportation items with A1255, removed the O0350 vaccination item, added the hearing, vision, and language items at resumption of care, and dropped four social-need items finalized a year earlier before they were ever collected. The item text, response options, and skip logic are CMS's, are read from the instrument and the Guidance Manual, and are not reproduced on this page.
The load-bearing fact for documentation is that OASIS is a data set with coding conventions, not a scale, and every response has to be traceable to something the assessing clinician observed or gathered within the assessment window. Two conventions do most of the work and are routinely confused. The legacy items M1800 to M1860 and M1033 are rated on the patient's ability to perform the task safely on the day of assessment, which the Guidance Manual defines as the 24 hours before the visit plus the time in the home, with the usual status rule (what is true more than half of that time) when ability varies; these eight items set the PDGM functional impairment level for payment. The GG0130 and GG0170 items code usual performance, based on a functional assessment at or soon after the start or resumption of care and reflecting the patient's baseline before agency services could improve it; these items feed the Discharge Function Score used in the quality reporting program and the expanded HHVBP model, and they do not drive PDGM payment. Layered on both is the one clinician convention: one qualified clinician is responsible for the comprehensive assessment and signs it, but since January 1, 2018 that clinician may gather information from the patient, caregivers, the physician, the pharmacist, and other agency staff who had direct contact with the patient, provided agency policy allows it, the collaboration happens inside the assessment window, and M0090 records the last date information was gathered. Everything else on this page follows from those rules: the windows that bound each time point, the narrative that must support each response, the correction procedures, and the payment, star rating, value-based purchasing, and survey consequences of getting a response wrong. Neighbors own the clinical content the items summarize. The pressure injury staging page owns how a wound is staged and described, while this page owns how M1306 and M1311 must agree with that wound note; the Braden Scale and Morse Fall Scale pages own the risk tools that the current item set no longer names (the M1300 and M1910 risk-tool items were retired in OASIS-D), and the medication management note owns the reconciliation record behind M2001 to M2005 and N0415.
Registered nurses complete the start of care and resumption of care assessments whenever skilled nursing is ordered, physical therapists and speech-language pathologists complete them in therapy-only cases, occupational therapists complete them for Medicare patients when OT is ordered together with PT or SLP and nursing is not, and any of those four disciplines may complete the recertification, other follow-up, transfer, death at home, and discharge assessments once the case is open; licensed practical and vocational nurses, physical therapist assistants, occupational therapy assistants, social workers, and aides contribute observations through collaboration and cannot be the assessing clinician. Agency clinical managers and OASIS review staff read the item set against the visit narrative and the plan of care before the record is locked and transmitted, coders and billers take the eight functional items and the diagnoses into the HIPPS code with occurrence code 50 carrying the M0090 date on the claim, and the physician or allowed practitioner signs a plan of care that 42 CFR 484.60 requires to answer the needs the assessment found. Downstream, CMS builds the PDGM functional level, the Quality of Patient Care star rating, the HH QRP measures, and the expanded HHVBP scores from the transmitted responses; state surveyors reconcile M0030, M0090, M0102, M0104, the visit dates, and the transmission date against the Conditions of Participation; Medicare Administrative Contractors and the Review Choice Demonstration contractors compare functional responses with the notes in medical review; and the Office of Inspector General compares OASIS with claims. The OASIS page owns one job: the time point and its window, who assessed and who collaborated, the responses and the narrative that supports each, the two functional conventions, the medication items, and the correction and transmission trail. Neighbors take over from there: the care coordination note carries the interdisciplinary plan the assessment feeds, the transfer of care summary and the discharge summary carry the patient out of the episode alongside the transfer and discharge OASIS, a fall during the episode is reported through the incident report while J1800 and J1900 record it at transfer or discharge, and the medication management note holds the reconciled list the drug regimen review items summarize.
CMS prescribes the item set, the data specifications, the time points, and the coding conventions; it does not prescribe the narrative that sits beside them, and that narrative is what a surveyor, a medical reviewer, or a quality auditor reads when a response is questioned. What survives review is a record in which the time point and its window are provable from dates, the assessing clinician and any collaborators are named and the M0090 date is set correctly, every response that drives payment or a measure traces to a dated observation, the day-of-assessment and usual-performance conventions are kept apart, the drug regimen review timing is documented, and the plan of care, the QA review, any correction, and the transmission all reconcile. Each element below carries the pitfall that most often undermines it.
Time point, window, and who assessed. Open with the time point (M0100) and the dates that prove its window: the referral date (M0104), any physician-ordered start date (M0102), the start of care date (M0030), the visit dates, and M0090. Under 42 CFR 484.55 the initial assessment visit falls within 48 hours of the referral, within 48 hours of the patient's return home, or on the ordered date, and the comprehensive assessment is finished no later than 5 calendar days after the start of care; the Guidance Manual sets resumption of care within 2 calendar days of the facility discharge or of learning the patient is home, recertification on days 56 to 60 of the 60-day period, other follow-up within 2 calendar days of a significant change, and transfer, death at home, and discharge within 2 calendar days of the event or of learning about it, with no visit required for transfer or death at home. Name the assessing discipline (M0080) and, at start of care, the reason that discipline was eligible: a registered nurse when nursing is ordered, PT or SLP in a therapy-only case, OT only when PT or SLP is also ordered and nursing is not. Pitfall: A start of care with no referral date, so the 48-hour visit cannot be shown, or "OASIS must be done on the first visit" written as a rule when the CoP allows 5 calendar days.
One clinician, collaboration, and M0090. State who owns the assessment and who contributed. One qualified clinician is responsible for the comprehensive assessment and signs it, and the signature attests that the responses reflect the patient's status as assessed and supported in the record. Since January 1, 2018 that clinician may gather information from the patient, caregivers, the physician, the pharmacist, and other agency staff who had direct contact with the patient, when agency policy allows it and the input arrives inside the assessment window; agencies may also keep the older practice of limiting the OASIS to the assessing clinician's own findings. Record the contributors and their dates the way your policy requires, and set M0090 to the last date information used to complete the assessment was gathered and documented, which may be a later visit than the first; an assessment may span more than one visit inside the window. A correction made later under the agency's correction policy does not by itself move M0090. Pitfall: A therapist's evaluation pasted in as the functional items with no assessing clinician named, or M0090 set to the first visit while the mobility observations came two days later.
Item response and the narrative that supports it. For every response that a measure, the payment grouping, or a survey tag can turn on, the visit narrative must carry the observation behind it: what the patient did, with how much help, using what device, and when. The manual's convention is that current-status items rest on observation and report at the time of the assessment, not on documentation from the hospital or the last certification period. Make the item families agree with the clinical notes that describe the same thing: the M1306 and M1311 skin items with the wound note that stages and measures the injury under the staging conventions, the falls items J1800 and J1900 at transfer and discharge with the fall documentation and the safety report, the M1400 shortness of breath item with the observed activity that produced it, and the interview items (the C0200 to C0500 mental status interview, the D0150 mood interview) with a note that the patient, not a family member, answered. Use a dash only when no information is available, which CMS expects to be rare; where an item carries its own code for a refused, not attempted, or not applicable activity, use that code and write why. Pitfall: A response coded from the hospital discharge summary, a wound item that disagrees with the wound note, or a dash where the item offered a refused or not-attempted code.
Functional items M1800 to M1860 and M1033: ability on the day of assessment. These items rate what the patient is able to do safely, not what the patient chose to do or did with a caregiver's habitual help, over the day of assessment: the 24 hours before the visit plus the time in the home. When ability varied across that day, the general convention applies and the response reflects what was true more than half of the time, so write both observations and say which one the response reflects. Because these eight items set the PDGM functional impairment level for both 30-day periods of the certification, the narrative should show the task attempted, the assistance actually required for safety, the device, and any symptom that limited it, in ordinary clinical words rather than in the language of the response options; where the patient refused to attempt a task, record the refusal and the basis for the ability judgment. Pitfall: A patient scored as needing assistance because a daughter always helps, or "independent" for a transfer the clinician never watched.
Section GG items: usual performance at baseline. GG0130 and GG0170 code the patient's usual performance across the assessment period, based on a functional assessment at or soon after the start or resumption of care and reflecting the baseline before agency services could improve it; a single best or worst attempt is not the code. Have the patient attempt the activity before instruction that would make the performance more independent, record each attempt with the type and amount of help, and let therapy observations feed the code through collaboration when policy allows. When an activity was not attempted, use the item's own reason code and write the reason (safety, refusal, the environment) rather than a dash, because a dashed item can remove the patient from the Discharge Function Score, the measure that these items feed in the HH QRP and the expanded HHVBP model. Keep the GG narrative separate from the M1800-series narrative: the two conventions can legitimately produce different pictures of the same patient on the same day. Pitfall: GG coded from one assisted walk down the hall, or the M1800-series ability judgment copied into GG as if the conventions were the same.
Drug regimen review and the medication items. Document the M2001 drug regimen review at start and resumption of care against the reconciled list in the medication management note: the potential or actual clinically significant issues found, the date and time the physician or allowed practitioner was contacted, and when orders or recommendations were received. The Guidance Manual's standard for M2003 is that the contact and the prescribed or recommended actions were completed by midnight of the next calendar day at the latest, so the timestamps are the evidence, and M2005 carries the same question across the episode at transfer and discharge. N0415 records the high-risk drug classes in the patient's current reconciled regimen even if a dose was held or missed on the day of assessment, which is an exception to the day-of-assessment convention; M2020 rates the patient's ability to take oral medications reliably and safely, and the teaching that follows belongs in the narrative and the plan of care. Pitfall: "Med rec done" with no issue named and no contact time, or an anticoagulant left off N0415 because the patient skipped it yesterday.
Plan of care, QA review, corrections, and transmission. Close the loop. The plan of care under 42 CFR 484.60 must specify the care and services that meet the needs identified in the comprehensive assessment, with measurable outcomes and goals, so the problems the responses describe should each have an intervention and a goal in the plan; agency QA review before the record is locked checks item against narrative against plan and returns queries to the assessing clinician. The agency must encode and transmit the completed assessment to CMS within 30 days of completing it, counted from M0090, and the record can be corrected afterward: Chapter 2 of the Guidance Manual provides a modification record for clinical errors and errors in non-key fields and an inactivation record when the event did not occur or a key field is wrong, with 24 months from the assessment target date to submit, modify, or inactivate. Corrections follow the agency's written correction policy, normally by the assessing clinician; a true change in the patient's status after the window closed is new information for the next time point, not a correction. For a late start of care, record the actual dates, the reason the window was missed, and any agency attestation your policy requires; CMS publishes no late-attestation form. Pitfall: A plan of care with a wound order and no wound response, or "the OASIS was transmitted so it cannot be fixed" written into a QA note when the 24-month correction window is open.
OASIS DOCUMENTATION BLOCK (supporting record for one time point)
Date / time: [ ] Agency and setting: [ ] Assessing clinician + discipline (M0080): [ ]
Item set version: [OASIS-E2, effective April 1, 2026] Payer: [Medicare / Medicaid /
other payer; all-payer collection since July 1, 2025 unless exempt]
Time point (M0100): [SOC / ROC / recertification / other follow-up / transfer, not
discharged / transfer, discharged / death at home / discharge]
Window evidence: referral date (M0104) [ ] ordered start date (M0102) [ ]
SOC date (M0030) [ ] visit dates [ ] M0090 date completed [ ]
window met: [yes / late: reason and agency attestation per policy]
Eligibility of the assessing discipline at SOC: [RN, nursing ordered / PT or SLP,
therapy only / OT with PT or SLP ordered and no nursing]
Collaboration (agency policy allows: yes / no): [contributor, discipline, date,
what was contributed; input received inside the window; M0090 reflects last date]
Item responses with supporting narrative (plain clinical words, dated observation):
sensory, cognition, mood interviews: [patient answered / refused: item code used]
M1400 shortness of breath: [activity observed and response]
M1800 to M1860 and M1033 (ability, day of assessment = 24 hours before the
visit plus the visit; usual status if it varied): [task, assist, device, symptom]
GG0130 self-care and GG0170 mobility (usual performance at baseline, before
agency services; each attempt recorded; not-attempted reason code, no dash): [ ]
skin M1306 / M1311 agree with the wound note dated: [ ]
falls J1800 / J1900 (transfer and discharge) agree with the fall record: [ ]
Drug regimen review: M2001 issues found [ ] contact date/time [ ] orders received [ ]
(by midnight of the next calendar day) N0415 classes in the reconciled regimen [ ]
M2020 oral medication ability [ ] M2005 at transfer or discharge [ ]
Dashes used: [none / item and reason; CMS expects dashes to be rare]
Plan of care linkage (42 CFR 484.60): [each identified need has an intervention and
a measurable goal; sent for signature on ...]
QA review: [date; queries and how resolved; corrected before lock under the agency
correction policy; M0090 unchanged unless new information was gathered]
Transmission: [iQIES date, within 30 days of M0090] Correction after acceptance:
[modification / inactivation record, date, reason; 24-month window]
Next time point due: [recertification days 56 to 60 / other]
Clinician signature / credentials: Date:Free to use and share, no signup. The PDF includes a one-page cheat sheet with element-by-element pitfalls and a pre-sign checklist; the DOCX is the blank documentation block, ready to adapt. Neither reproduces OASIS item text or response options.
Scenario: a US home health agency's start of care for a 79-year-old woman discharged from a hospital heart failure admission, with skilled nursing and physical therapy ordered; the registered nurse documents the window with dates, completes the assessment across two visits with the physical therapist's collaborative input, sets M0090 to the last date information was gathered, supports every functional response with an observation under the correct convention, times the drug regimen review, and records the QA review, the plan of care linkage, and the transmission. All details are fictional.
Patient: H.T., 79 · Setting: Home health agency, start of care after hospital discharge (heart failure) · Clinician: A. Brennan, RN · Note date: 09/17/2026
Time point, window, and assessing clinician: Start of care time point, OASIS-E2 item set, traditional Medicare. Referral received from the hospital case manager on 09/14/2026 at 13:20 (M0104), the day of discharge; no physician-ordered start date (M0102 not applicable). Initial assessment visit 09/15/2026 at 10:30 by A. Brennan, RN, within 48 hours of the 09/14/2026 return home; start of care date (M0030) 09/15/2026. Skilled nursing and physical therapy are ordered, so a registered nurse is the assessing clinician (M0080). The comprehensive assessment was completed across the 09/15/2026 and 09/17/2026 visits, inside the 5-calendar-day window that closes 09/20/2026. Face-to-face encounter documented by the hospitalist on 09/13/2026, within 90 days before the start of care.
Collaboration and M0090: Collaboration is permitted under agency policy HH-OASIS-04. J. Okoro, PT, evaluated H.T. on 09/16/2026 and reported gait distance, stair attempt, and transfer observations to the assessing RN by secure message the same day; the daughter (primary caregiver) described the home routine on 09/15/2026; the reconciled medication list was confirmed with the pharmacy by telephone on 09/15/2026. The RN considered that input, re-observed transfers and a walk on 09/17/2026, and selected the responses. Contributors are logged in the assessment record per policy. M0090 (date assessment completed) is 09/17/2026, the last date information used to complete the assessment was gathered and documented; the RN's signature attests that the responses reflect H.T.'s status as assessed and supported in this record.
Functional items M1800 to M1860 and M1033: ability on the day of assessment: Observed on 09/17/2026 between 09:40 and 10:50, with the preceding 24 hours by report. Ambulation (M1860): walked 25 feet from the recliner to the bathroom with a rolling walker and contact-guard assist, two standing rest breaks, short of breath while talking by the end; she reports the same distance and breaks on 09/16/2026. Bathing (M1830): needs a person present to get in and out of the tub shower with a shower chair and to wash below the knees; washes upper body herself once seated. Lower body dressing (M1820): needs help with compression stockings, socks, and shoes; upper body dressing (M1810): independent, slow. Toilet transferring (M1840): uses the grab bar and the walker, no hands-on help, observed once. Bed to chair transfer (M1850): stands with one-person standby assist and a verbal cue to lock the walker before turning. Grooming (M1800): independent once supplies are within reach. Shortness of breath (M1400): occurs with walking about 20 feet and with lower body dressing, not at rest, not while seated talking. Risk of hospitalization (M1033): supported in the history by two hospital admissions since March 2026, nine scheduled medications, and the exhaustion she describes since discharge. Each response entered on the item set matches the observation written here, and no response was taken from the hospital discharge summary.
Section GG: usual performance at baseline: GG0130 self-care and GG0170 mobility were coded on usual performance across the two visits and the PT's 09/16/2026 observations, before the benefit of agency services: no teaching or new technique was introduced before the attempts that were coded. Where attempts differed, usual performance was used and each attempt is listed: on 09/16/2026 she walked 40 feet once with the PT before needing to sit, while the other three observed walks across the two visits stopped at 20 to 25 feet with contact guard, so the walking distance items reflect the shorter distance with contact-guard assistance. Sit to stand and chair to bed transfers needed the same standby assistance and verbal cue on every attempt. Stairs were not attempted for safety on either visit (four exterior steps, no rail; dyspnea at 20 feet), and the stair items carry the item's own not-attempted reason code rather than a dash, with the reason written here. Tub transfer and shower use were observed with the daughter present at the 09/15/2026 visit. These baseline codes are what the Discharge Function Score will be measured against at discharge, so nothing was rounded toward independence.
Drug regimen review and medication items: Drug regimen review (M2001) completed 09/15/2026 against the reconciled list of nine scheduled medications (medication management note dated 09/15/2026). Potential clinically significant issue: two loop diuretics, the furosemide started in the hospital and the torsemide from before admission, both present in the home and both taken on 09/14/2026 evening. Cardiology office contacted 09/15/2026 at 11:50; order received 16:40 the same day to stop torsemide and continue furosemide, and the daughter removed the torsemide from the organizer at 17:10 with the RN on the phone. M2003 is therefore documented as contact and prescribed actions completed well before midnight of the next calendar day. N0415 records the anticoagulant (apixaban) in her reconciled regimen although the 09/16/2026 evening dose was missed; the class is coded from the current regimen, not from the day's intake. Oral medication management (M2020): she needs the daughter to fill the weekly organizer and a reminder for evening doses; she takes morning doses correctly from the organizer. Teaching on the single diuretic, daily weights, and the evening-dose alarm is in the plan of care.
Other items, plan of care, QA review, and transmission: Skin: head-to-toe inspection on 09/15/2026 and 09/17/2026 found intact skin at every pressure point and device site, so the M1306 pressure injury item and the integumentary narrative agree; a Braden assessment was completed under agency policy since the item set names no risk tool. Falls: one fall in July 2026 without injury, from the history and the daughter's account, recorded in the hospitalization risk factors and in the agency's multifactor fall risk assessment; J1800 and J1900 will be completed at transfer or discharge. Interview items: H.T. answered the mental status and mood interviews herself on 09/15/2026 with the daughter out of the room. Plan of care drafted 09/17/2026 from these findings under 42 CFR 484.60, with goals for walking distance, independent lower body dressing, daily weight logging, and diuretic adherence, each tied to a nursing or therapy intervention, and sent to the cardiologist for signature on 09/18/2026. Agency OASIS review on 09/18/2026 (M. Salcedo, RN, QA) compared items, narrative, and plan and returned three queries (the walking distance code, the stair reason code, and the missing pharmacy contact time); the assessing RN corrected the two documentation gaps and confirmed the code before the record was locked, under agency correction policy, and M0090 stayed 09/17/2026 because no new patient information was gathered. Transmitted to iQIES on 09/21/2026, four days after M0090 and inside the 30-day limit; the transmission confirmation is filed with the assessment. Recertification window falls on days 56 to 60 of the period and is flagged on the agency calendar.
This sample is fictional and for educational purposes. It does not describe a real patient or record; the dates, observations, and details are invented to show documentation structure and are not clinical guidance. No OASIS item text or response options are reproduced.
Writing these after every session? BastionGPT drafts complete notes from bullets, dictation, or a transcript.
Generate a note from bulletsUnited States: the item set is LAW, and most of what happens to a response afterward is PAYER POLICY. Under LAW, 42 CFR 484.55 requires the initial assessment visit within 48 hours of the referral or of the patient's return home, or on the physician-ordered start date; a comprehensive assessment completed no later than 5 calendar days after the start of care; a review of all medications; and updates at least in the last 5 days of every 60 days, within 48 hours of the patient's return from a hospital admission of 24 hours or more, and at discharge. It names the registered nurse as the initial assessor when nursing is ordered and lets the appropriate rehabilitation professional conduct the visit and complete the assessment in a therapy-only case, with the occupational therapy allowance that took effect January 1, 2022. 42 CFR 484.45 requires the agency to encode and electronically transmit each completed assessment to the CMS system within 30 days of completing it, and 42 CFR 484.60 requires an individualized plan of care that specifies the care and services needed to meet the needs identified in the comprehensive assessment, with measurable outcomes and goals. The OASIS-E2 Guidance Manual and CMS's quarterly Q&As carry the sub-regulatory interpretation (the time point table with its 2-calendar-day windows for resumption, transfer, death at home, and discharge and its recertification days 56 to 60, the day-of-assessment and usual-performance conventions, the one clinician convention and its 2018 collaboration expansion, and the Chapter 2 correction policy), and surveyors cite comprehensive-assessment deficiencies under the 484.55 tags. All-payer collection is LAW by rule: since July 1, 2025, OASIS is collected and submitted for every patient beginning skilled home health services regardless of payer, except patients under 18, maternity-only cases, and personal care or chore-only cases, and non-Medicare data enter the annual payment update calculation for patients with a start of care M0090 date on or after January 1, 2027. Under PAYER POLICY, the HH QRP requires a 90 percent submission compliance rate or the agency loses 2 percentage points from its market basket update; PDGM builds the functional impairment level for each 30-day period from M1800, M1810, M1820, M1830, M1840, M1850, M1860, and M1033 alone, with the claim's occurrence code 50 carrying the M0090 date of the assessment that supplied them, and the CY 2027 proposed rule (fact sheet released July 1, 2026) updates the functional point tables, proposes no new OASIS items, and proposes revised HH QRP submission deadlines and calendar-year reporting beginning with the CY 2027 program year; the Quality of Patient Care star rating uses seven measures (timely initiation of care, improvement in ambulation, bed transferring, bathing, shortness of breath, and management of oral medications, and the claims-based within-stay potentially preventable hospitalization measure that replaced the acute care hospitalization measure), needs 20 complete quality episodes per measure and 5 of the 7 measures to report, and refreshes quarterly; the expanded HHVBP model adjusts payment by up to 5 percent in either direction, paid its first adjustment in 2025 on 2023 performance, and for CY 2026 uses six OASIS-based measures (discharge function score, improvement in dyspnea, improvement in management of oral medications, and the newly added improvement in bathing, upper body dressing, and lower body dressing) beside claims-based potentially preventable hospitalization, discharge to community, and Medicare spending per beneficiary measures and the HHCAHPS survey, the total normalized composite measures having ended after 2024. Medical review is PAYER POLICY with teeth: the Review Choice Demonstration runs in Illinois, Ohio, Texas, North Carolina, Florida, and Oklahoma (added December 1, 2023), was extended five years from June 1, 2024 with the minimal-review choice removed, and its reviewers, like the Targeted Probe and Educate and Supplemental Medical Review contractors, read the functional responses against the visit notes; the face-to-face encounter under 42 CFR 424.22 (LAW) must occur within 90 days before or 30 days after the start of care, and the comprehensive assessment is where homebound status and skilled need are usually evidenced. CONVENTION covers the rest: the visit narrative that supports each response, the QA review before the record is locked, an agency correction policy that names who corrects when the assessing clinician is unavailable, and the late-start attestation some agencies require; none of these is a CMS form, and CMS publishes none.
The version line and the borders are part of the record, and the evidence explains why the narrative matters. OASIS-E2 is the item set and Guidance Manual in force since April 1, 2026, an off-cycle change from OASIS-E1 (January 1, 2025) and OASIS-E (January 1, 2023); the E2 instrument carries OMB control number 0938-1279 with an expiration date of December 31, 2028, the CMS OASIS Data Sets and User Manuals pages list no later version, and the CY 2027 proposed rule proposes no new items, so a note written in late 2026 names OASIS-E2 and watches the CMS pages for a draft successor rather than assuming one. The instrument is United States only: Canada's home care programs assess with interRAI Home Care and Australia's aged care system uses its own national assessment tools, and OASIS has no standing in either, so this page and its templates apply to US home health agencies alone. On the evidence: in the original reliability study, two clinicians independently assessed 66 home health patients and inter-rater reliability was excellent (kappa above 0.80) for many items and substantial (kappa above 0.60) for most (Hittle and colleagues, Home Health Care Services Quarterly, 2003); Madigan and Fortinsky (The Gerontologist, 2004) examined inter-rater reliability of the item set, Tullai-McGuinness, Madigan, and Fortinsky (Home Health Care Services Quarterly, 2009) tested the validity of OASIS items against established instruments, and O'Connor and Davitt (Home Health Care Services Quarterly, 2012) reviewed the validity and reliability literature; neither research pass for this page located peer-reviewed inter-rater reliability data for the Section GG items in home health, a gap to remember when GG codes are audited. The Office of Inspector General found that 55 percent of the falls with major injury and hospitalization it identified in Medicare claims were not reported on the associated OASIS assessments, with worse reporting for Black, Hispanic, and Asian patients and by for-profit agencies (OEI-05-22-00290, September 5, 2023), and CMS concurred with all four recommendations; MedPAC's March 2026 report to Congress repeats the Commission's concern about the validity of the provider-reported patient assessment information used for case mix, reports a 21.2 percent fee-for-service Medicare margin for freestanding agencies in 2024, and recommends a 7 percent reduction to the 2026 base payment rate for 2027. Those findings are why a response with no dated observation behind it is the defect every reviewer looks for first.
OASIS is a Centers for Medicare & Medicaid Services instrument. The item set, the data specifications, and the Guidance Manual are published by CMS as works of the United States government, which federal copyright law places in the public domain, and the instrument's original developers at the Center for Health Services and Policy Research in Denver stated, in a permission notice that CMS now archives on its OASIS Data Sets page, that OASIS is their intellectual property, may not be copyrighted by any other party, and may be freely copied and incorporated into assessment forms and software by home care providers, organizations, and individuals with no royalty or use fee, with an acknowledgement of authorship expected. Reproducing the items in an agency's own forms or electronic record is therefore permitted, while submission to CMS must follow the CMS data specifications through iQIES, and the screening tools embedded in the item set carry their own status: the PHQ-2 to 9 mood interview is free to use, the mental status interview is a government-developed instrument, and the delirium item derives from the Confusion Assessment Method, whose rights belong to its author. BastionGPT is not affiliated with, or endorsed by, the Centers for Medicare & Medicaid Services. This page reproduces no test items, stimuli, norms, or scoring materials.
The numbers behind these errors are specific. In the reliability study run by the instrument's developers, two clinicians independently assessed 66 home health patients and inter-rater reliability was excellent (kappa above 0.80) for many items and substantial (above 0.60) for most, which means a minority of items disagree between trained raters before any workflow pressure is applied (Hittle and colleagues, 2003). The Office of Inspector General found that 55 percent of the falls with major injury and hospitalization it identified in Medicare claims were missing from the associated OASIS assessments, that for-profit agencies and the care of Black, Hispanic, and Asian patients showed worse reporting, and that the agencies with the lowest published fall rates reported falls least often (OEI-05-22-00290, September 5, 2023). MedPAC's March 2026 report to Congress repeats its concern about the validity of provider-reported patient assessment information used for case mix, beside a 21.2 percent fee-for-service Medicare margin for freestanding agencies in 2024 and a recommended 7 percent rate reduction for 2027. The HH QRP threshold is 90 percent submission compliance with a 2 percentage point penalty, the Review Choice Demonstration in six states reads functional responses against visit notes claim by claim, and every accepted record can be modified or inactivated for 24 months. No published audit counts how often a chart carries an OASIS response with no supporting observation; the survey, payment, and medical review rules described under compliance considerations are what turn that omission into a deficiency or a denial. The BastionGPT Clinical Advisory Board sees the same errors most often in OASIS documentation reviews:
BastionGPT is specifically trained, tuned, and clinically tested on home health assessments and plan of care documentation.
See how clinicians use it day to day on the AI medical notes page.
Many BastionGPT users report saving more than 90 minutes per day on documentation.
HIPAA-compliant with a signed BAA on every plan. Your data is never used to train models. BastionGPT drafts, you review and sign.
OASIS-E2, effective April 1, 2026 for every assessment with a start of care, resumption of care, recertification, transfer, or discharge date on or after that day, an off-cycle change from OASIS-E1 (January 1, 2025) and OASIS-E (January 1, 2023). The CMS OASIS Data Sets and User Manuals pages carry the E2 instruments, the final E2 Guidance Manual, and the E2 Q&As, and the instrument carries OMB control number 0938-1279 with an expiration date of December 31, 2028; no later version has been announced, and the CY 2027 proposed rule proposes no new items. Described at the topic level, E2 replaced the gender item with A0810, replaced the A1250 transportation items with A1255, removed the O0350 vaccination item when its measure was retired, added the B0200 hearing, B1000 vision, and A1110 language items at resumption of care, revised the J1900 fall-injury descriptions, and dropped four social-need items that had been finalized a year earlier before they were ever collected. Many training materials and vendor pages still call E1 current; name E2 in your policy and your record, and read the E2 manual, not a cached E1 copy, for item guidance.
Eight reasons for assessment sit under M0100 in six time points: start of care, resumption of care, follow-up (recertification, and other follow-up after a significant change), transfer to an inpatient facility (with or without discharge from the agency), death at home, and discharge. The Conditions of Participation set the outer limits: the initial assessment visit within 48 hours of the referral or of the patient's return home, or on the physician-ordered date; the comprehensive assessment completed no later than 5 calendar days after the start of care; updates in the last 5 days of every 60 days, within 48 hours of return from a hospital admission of 24 hours or more, and at discharge. The Guidance Manual's time point table fills in the rest: resumption of care within 2 calendar days of the facility discharge date or of learning the patient is home, recertification on days 56 to 60 of the current 60-day period, other follow-up within 2 calendar days of a significant change in condition, and transfer, death at home, and discharge within 2 calendar days of the event or of learning about it. Transfer and death at home are completed from the record and do not require an in-person visit; every other time point does. The 30-day transmission clock under 42 CFR 484.45 runs from M0090, the completion date, not from the visit. Write the dates that prove each window into the record, because a surveyor reconciles M0104, M0102, M0030, the visit dates, M0090, and the iQIES date.
The assessing clinician must be a registered nurse, physical therapist, speech-language pathologist, or occupational therapist, and which one depends on the orders. When skilled nursing is ordered, the registered nurse conducts the initial assessment visit and completes the start of care comprehensive assessment. In a Medicare therapy-only case the physical therapist or speech-language pathologist may do both, and since January 1, 2022 an occupational therapist may do both when the referral includes PT or SLP along with OT and no nursing; OT alone does not establish initial Medicare eligibility, so an OT-only referral cannot open a Medicare case, although a patient who qualified through nursing, PT, or SLP keeps eligibility if OT becomes the only remaining discipline. Once the case is open, any of the four qualified disciplines may complete the recertification, other follow-up, resumption, transfer, death at home, and discharge assessments. Licensed practical and vocational nurses, physical therapist assistants, occupational therapy assistants, social workers, and aides may not be responsible for the comprehensive assessment or sign it, but under the collaboration convention their direct observations may inform the assessing clinician's responses when agency policy allows and the input arrives inside the window. Write the assessing discipline in M0080 and, at start of care, the reason it was eligible.
One qualified clinician is responsible for the comprehensive assessment and signs it, and the signature attests that the responses reflect the patient's status as assessed and supported in the record. What changed on January 1, 2018 is that the assessing clinician may collect data for any or all OASIS items with input from the patient, caregivers, and other health care personnel, including the physician, the pharmacist, and other agency staff who had direct in-person contact with the patient or another means of gathering information (a monitoring device, a photograph, a phone call), provided agency policy allows it and the collaboration happens inside the assessment timeframe and consistent with item guidance; the clinician weighs the input and selects the response. Agencies may instead keep limiting the OASIS to the assessing clinician's own findings, and either way the agency is responsible for tracking who contributed. M0090 is the last date information used to complete the assessment was gathered by the assessing clinician and documented, so a two-visit start of care carries the second visit's date, and a later correction under the agency's policy does not by itself change it. Document the policy that permits collaboration, each contributor with the date and what they contributed, the assessing clinician's own re-observation where one occurred, and the M0090 rule applied, as the sample on this page does.
Under two different rules, and the record should show both. The legacy items M1800 to M1860 rate the patient's ability to perform the task safely on the day of assessment, defined as the 24 hours before the visit plus the time in the home; the general convention for status that varies is to report what is true more than half of that period, and ability is not willingness or the help a caregiver habitually gives. The GG0130 and GG0170 items code usual performance across the assessment period, from a functional assessment at or soon after the start or resumption of care that reflects the baseline before agency services could improve it, so a single best attempt and a single worst attempt are both wrong codes; if the only help needed was retrieving a device, the manual treats that as setup assistance. Write each attempt with the help required, name the convention behind each response, and expect the two sections to describe the same walk differently when the rules make them differ. When an activity was not attempted, use the item's own reason code and write the reason rather than a dash, and record a refusal as a refusal. Collaboration lets the therapist's observations inform both sections when policy allows.
Different items for different programs, which is why a myth persists that GG drives payment. PDGM's functional impairment level for each 30-day period comes from eight items, M1800, M1810, M1820, M1830, M1840, M1850, M1860, and M1033, taken from the most recent start of care, resumption, recertification, or other follow-up assessment before the claim period, with occurrence code 50 on the claim carrying that assessment's M0090 date; the GG items play no part in the payment grouping, and the CY 2027 proposed rule keeps that design while updating the point tables. The Quality of Patient Care star rating uses seven measures: timely initiation of care, improvement in ambulation, bed transferring, bathing, shortness of breath, and management of oral medications, and the claims-based within-stay potentially preventable hospitalization measure that replaced acute care hospitalization; an agency needs 20 complete quality episodes per measure and 5 of the 7 measures, refreshed quarterly. The expanded HHVBP model, which adjusts payment by up to 5 percent in either direction with the first adjustment paid in 2025 on 2023 performance, uses six OASIS-based measures for CY 2026: the discharge function score built from the GG items, improvement in dyspnea, improvement in management of oral medications, and the newly added improvement in bathing, upper body dressing, and lower body dressing, beside the claims-based and HHCAHPS measures. The practical point is that the same functional responses reach three programs by different routes, and a response that cannot be supported is exposed in all of them.
Yes, for 24 months, and the myth that a submitted OASIS is frozen costs agencies real corrections. Chapter 2 of the Guidance Manual describes the automated correction policy in iQIES: a modification record corrects clinical errors and errors in non-key fields in a record that has been accepted, and an inactivation record is used when the event did not occur or any key field is wrong, because key fields cannot be edited and the record must be inactivated and resubmitted. Since January 1, 2020 agencies have had 24 months from the assessment target date to submit, modify, or inactivate. Corrections follow the agency's written correction policy and are normally made by the assessing clinician, with a supervisor or QA staff correcting under the policy when that clinician is unavailable and the reason documented; a genuine coding error is corrected whenever it is found, while a true change in the patient's status after the window closed is information for the next time point, not a correction of the last one, and M0090 does not necessarily change when an error is fixed. Before public reporting, agencies also get a preview period to review and correct. For a late start of care, record the actual visit and completion dates, set M0090 to the real completion date, document why the 5-day window was missed, and add whatever attestation your agency policy requires; CMS publishes no late-attestation form, and a backdated M0090 is a falsified record rather than a late one.
Yes to both. Since July 1, 2025, after a penalty-free phase-in from January 1 to June 30, 2025, OASIS is collected and submitted for every patient who begins skilled home health services regardless of payer, with three exclusions: patients under 18, patients receiving maternity services only, and patients receiving only personal care, housekeeping, or chore services; non-Medicare data begin to count toward the annual payment update for patients whose start of care M0090 date is January 1, 2027 or later. On rights, OASIS is a CMS instrument published as a work of the United States government, and its original developers at the Center for Health Services and Policy Research in Denver granted everyone a free, nonexclusive right to copy the items into printed forms, software, and other products with no royalty, expecting an acknowledgement of authorship; CMS archives that permission notice on its OASIS Data Sets page. So an agency may build the item set into its own assessment form and electronic record, provided the transmitted data follow the CMS data specifications through iQIES, and a vendor's build should carry the current E2 specifications. Two cautions: the embedded screening tools (the PHQ-2 to 9 mood interview, the mental status interview, the delirium item derived from the Confusion Assessment Method) have their own rights positions, and this page and its templates reproduce none of the item text because the documentation guidance, not the instrument, is the point.
Yes. Give it the facts (the time point and the dates that prove its window, the assessing discipline and why it was eligible, each collaborator with a date and what they contributed, the observations behind each functional response under the convention that applies, the drug regimen review issue with its contact and order times, the skin and fall findings, the plan of care goals, the QA queries, and the transmission date) and it drafts the supporting record: the window statement, the collaboration and M0090 line, the item-by-item narrative in plain clinical words, the Section GG baseline with attempts listed, the medication items with timestamps, and the plan of care, QA, and transmission lines, ready for your review. It can also check a finished record for a response with no observation behind it, ability and usual performance confused, a window that cannot be proven, M0090 on the wrong visit, a collaborator unnamed, a dash where a code exists, or a plan of care that does not answer the assessment. BastionGPT is HIPAA-compliant with a signed BAA on every plan, and your data is never used to train models.
The instrument facts and compliance claims on this page trace to these sources, last verified September 2026:
Educational content, not legal or billing advice. Sample notes are fictional. Follow your organization's policies and your board, payer, and jurisdiction requirements.