Sign Up
Login
Try for Free
Login
The Mini-Cog is a three-minute cognitive screen scored 0 to 5, combining three-item recall with a pass-or-fail clock drawing, published by Soo Borson and colleagues in 2000. Primary care, geriatrics, and hospital teams use it at wellness visits and as a first-pass check when memory concerns arise. It screens; it never diagnoses. This page covers how to document and interpret Mini-Cog results in the chart, with a fictional sample note.
Any trained clinician or staff member (medical assistants, nurses, physicians) after brief training; interpreted by the treating clinician; no certification required
Primary care and geriatrics teams, memory clinics and neuropsychologists, hospital and post-acute teams, families receiving results, payers and quality reviewers
3 to 8 chart lines · administration about 3 minutes, plus time to disclose the result and plan
Clinician-administered cognitive screen (three-item recall 0 to 3 plus pass-or-fail clock 0 or 2, total 0 to 5)
Medicare annual wellness visits, first-pass screening on patient or family concern, hospital and perioperative baselines, triage before a longer instrument
Copyright Soo Borson (mini-cog.com): free for noncommercial clinical use unmodified; commercial and research use need permission; no authority mandates it by name
The Mini-Cog is a brief cognitive screen that pairs uncued three-item delayed recall (0 to 3 points) with a clock drawing scored pass or fail (2 or 0, no intermediate point), for a total of 0 to 5. Soo Borson and colleagues published it in 2000 as a cognitive vital-signs measure for dementia screening in a multilingual older population, and the scoring-algorithm paper followed in 2001. Two versions share one algorithm: the standardized numerical form (dated 01.19.16, with six numbered word-list versions so that repeat testing can use a different list) and the graphical form, which keeps the scoring of the 2000 proof-of-concept study. The rights holder recommends the standardized form for repeated or longitudinal use and lists 23 language versions as of August 2026. Administration takes about three minutes; the rights holder's own instructions add that time is needed to explain the procedure, discuss the result, and set the next step, so three minutes is the test, not the encounter.
The load-bearing fact is that the Mini-Cog is a screen, never a diagnosis. On the current scoring page, a total of 0 to 2 indicates a higher likelihood of clinically important cognitive impairment and 3 to 5 a lower likelihood of dementia that does not rule out some degree of impairment; the cut point was set to balance sensitivity and specificity for dementia detection, and it estimates no severity, no cause, no decisional capacity, and no driving fitness. A positive result means enough concern to justify a fuller cognitive and medical evaluation, which is why the chart carries the components, the version, the validity conditions, and the follow-up chain rather than a bare total. The MoCA and the MMSE are the usual longer comparators; the free SLUMS is the closer cousin in rights terms, and none of the four independently establishes a neurocognitive diagnosis, capacity, or fitness to drive.
Primary care is the main user: the Medicare annual wellness visit requires detection of cognitive impairment (by observation, patient report, and concerns raised by family or caregivers) without naming any tool, and the Mini-Cog's brevity and no-certification administration make it a common choice for that slot, for a first-pass check when a patient or family member voices concern, and for the triage step before a longer instrument. Geriatrics, hospital medicine, and perioperative teams use it for baselines and delirium-adjacent screening, where the validity conditions matter more than the number; skilled-nursing assessment uses its own CMS instrument instead. In Canada, the national task force recommends against instrument-based screening of asymptomatic older adults, so the indication (concern, not routine) leads the note; in Australia the 75-and-over health assessment must cover cognition without naming a tool, and local guidance more often reaches for the GPCOG, RUDAS, or KICA. When a screen is positive, the fuller workup belongs in a neuropsychological report or a longer bedside instrument, the bedside cognitive findings sit in the mental status exam, and any question of decisional ability goes to a capacity evaluation, because the screen decides none of those things.
No US, Canadian, or Australian authority prescribes a Mini-Cog note format, and none makes a score diagnostic. What survives review is a record that names the indication, both component scores with the form and word-list version, the language and administrator, the conditions that could invalidate the result, a screen-only interpretation, and a closed follow-up loop. Each element below carries the pitfall that most often undermines it.
Context and indication. Record why the screen was done: the cognitive component of a wellness visit, a patient or family concern, follow-up of a prior abnormal screen, a hospital or perioperative baseline, or part of another assessment. The Medicare requirement is detection of impairment; the tool is your choice. Pitfall: "Mini-Cog required by Medicare" (no regulation names it) or a score with no indication, which leaves a reviewer unable to tell routine screening from a symptomatic evaluation.
The exact result, both components. Chart recall x/3, clock x/2 (2 = normal, 0 = abnormal; inability or refusal to draw scores 0), and total x/5. The rights holder's FAQ asks for at least the word-list version and each component score in a discrete EHR field. Pitfall: A bare total. A 2/5 built from two words and a failed clock is a different picture from no words and a normal clock, and the next examiner cannot tell which without the components.
Form, version, and language. Name the standardized numerical or graphical form, the word-list version (1 to 6), the language version, and whether administration was language-concordant or through a qualified interpreter. On repeat testing use a different list and chart both. Pitfall: The recall words written into the narrative (they belong on the form, not the note), a translation or improvised interpretation charted as plain "Mini-Cog," or the same list reused at follow-up without saying so.
Administrator and conditions affecting validity. Record who administered and scored it (any trained staff member may), then the conditions: hearing and vision aids, motor limitation, pain, fatigue, acute illness, sedation, delirium or fluctuating attention, education and literacy, analog-clock familiarity, interruptions, and any departure from standard instructions. Close with a validity conclusion: standard and interpretable, interpretable with caution, or not interpretable. Pitfall: A precise total charted beside a condition that made it uninterpretable, or a motor, visual, or literacy barrier scored as a cognitive clock failure. Document the barrier; do not invent an adjusted score.
Interpretation kept to a screen. State the classification (0 to 2 positive; 3 to 5 lower likelihood, impairment not excluded) and the rule your program applies if it treats scores under 4 as a sensitivity trigger, then say what the result does not establish: mild cognitive impairment, dementia, cause, decisional capacity, or driving fitness. Note the reversible contributors considered. Pitfall: "Positive dementia screen" or a dementia diagnosis code supported only by the screen; a normal score used to close the question when the family reports decline.
Disclosure, collateral, and function. Document what was explained in plain, non-diagnostic terms, who was present with the patient's permission, and the patient's response; then the collateral history (informant, onset, course) and instrumental function (medications, finances, appointments, transportation, cooking, safety). Pitfall: No record of disclosure, a diagnostic label delivered from a screen, or a positive result with no informant sought.
Action, owner, and the closed loop. Record the reversible-contributor review, medication reconciliation, investigations, the fuller cognitive and functional assessment or referral, interim safety steps, and who owns follow-up by what date; a declined screen is charted as offered and declined with the alternative used and the re-offer plan. Pitfall: A positive screen that ends at a number, or a refusal entered as a zero or a positive result.
MINI-COG DOCUMENTATION BLOCK Date: [ ] Setting: [ ] Administered by: [ ] Interpreted by: [ ] Indication: [wellness visit / patient or family concern / follow-up of a prior screen / baseline before ...] Form: [standardized numerical / graphical] Word-list version: [1-6] Language: [ ] Interpreter: [none / qualified interpreter / language-concordant examiner] Result: recall [ ]/3 clock [ ]/2 (2 = normal, 0 = abnormal) total [ ]/5 Conditions: [hearing/vision aids, motor limits, pain, fatigue, acute illness, sedation, attention, education/literacy, analog-clock familiarity, deviations from standard administration] Validity: [standard and interpretable / interpretable with caution / not interpretable: reason] Interpretation: [screen positive (0-2) / screen negative (3-5); does not establish MCI, dementia, cause, capacity, or driving fitness; reversible contributors considered] Disclosure: [what was explained, who was present with permission, response] Collateral + function: [informant; onset/course; medications, finances, driving, appointments] Plan: [medication review / investigations / fuller cognitive and functional assessment or referral / interim safety steps] Follow-up owner + date: [ ] Prior result + list version, if repeat: [ ] Declined: [offered + declined + reason if volunteered + alternative assessment + re-offer plan; never entered as a score] 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 the word lists, clock instructions, or form.
Scenario: a positive screen at a subsequent Medicare annual wellness visit in primary care, standard administration in the patient's language, with a family informant present and the follow-up chain set in motion. All details are fictional.
Patient: M.R., 74 · Setting: Primary care, subsequent annual wellness visit · Clinician: L. Navarro, MD · Note date: 08/11/2026
Cognitive screening: Cognitive assessment component of today's subsequent annual wellness visit, with an added indication: the patient's daughter voiced concern about her memory at check-in. Mini-Cog, standardized numerical form, Spanish version, word-list version 3, administered and scored by J. Rivera, RN, in a quiet room before the visit; language-concordant, no interpreter. Patient wore her reading glasses and hearing aids, was alert with steady attention, no acute illness, no motor limitation, no sedating medication changes. Recall 1/3, clock 0/2 (abnormal), total 1/5: screen positive. Prior Mini-Cog 08/2025 at last year's visit: 5/5, word-list version 1.
Interpretation: Standard administration; result interpretable. A positive cognitive screen with a documented one-year change; it does not establish mild cognitive impairment, dementia, a cause, decisional capacity, or driving fitness. Reversible contributors considered: no features of delirium, no new anticholinergic or sedating medications on reconciliation, PHQ-2 negative today, sleep and hearing reviewed.
Disclosure and collateral: With the patient's permission the result was reviewed with her and her daughter in plain terms: the screen suggests memory and thinking should be looked at more closely, and it is not a diagnosis. Daughter describes about nine months of repeated questions, two missed utility payments, and a pill organizer found double-filled last month; patient acknowledges some forgetting, denies getting lost, still drives locally in daylight, manages cooking and self-care independently.
Plan: Medication reconciliation completed today. Laboratory evaluation for reversible contributors ordered per clinic protocol. Dedicated 40-minute cognitive and functional assessment visit scheduled 08/25/2026 with the daughter present, to include a longer cognitive instrument, an instrumental-activities inventory, and a structured driving history; memory-clinic referral to be decided at that visit. Interim: daughter to oversee the pill organizer and bill payments with the patient's agreement; patient advised to defer any unfamiliar or night driving until the assessment. Follow-up owner: Dr. Navarro; nursing call within one week to confirm the appointment.
Administrative: Result entered in the discrete cognitive-screen field with the form, word-list version, language, and both component scores; the clock drawing scanned to the chart. Wellness-visit cognitive detection documented for this encounter; no diagnosis coded from the screen.
This sample is fictional and for educational purposes. It does not describe a real patient or record; the scores, versions, dates, and details are invented to show documentation structure and are not clinical guidance.
Writing these after every session? BastionGPT drafts complete notes from bullets, dictation, or a transcript.
Generate a note from bulletsUnited States: the requirement to look is law; the tool is a convention. Section 1861(hhh) of the Social Security Act and 42 CFR 410.15 make detection of cognitive impairment a required element of the initial and subsequent Medicare annual wellness visit (G0438, G0439), defined as assessment by direct observation with due consideration of patient report and concerns raised by family, friends, or caregivers, and no instrument is named (LAW). CMS deliberately declined to name one, reasoning, as Cordell and colleagues characterized it in 2013, that "there is no single, universally accepted screen," and its 2026 wellness-visit guidance tells clinicians to consider brief tools without naming any (PAYER POLICY). Bills in the CHANGE Act lineage before the 119th Congress would require an NIA-identified detection tool at the wellness visit and documentation of the tool used; as of August 2026 they are proposed, not law, so a chart that says Medicare required the Mini-Cog is wrong today and may need a different sentence later. The cognitive assessment and care plan service (99483, which replaced G0505 in 2018) is PAYER POLICY with documented elements far beyond any screen: a cognition-focused history and examination with an independent historian, functional assessment including decision-making capacity, standardized dementia staging, medication reconciliation, neuropsychiatric symptom evaluation, home and driving safety, caregiver identification, and a written care plan. A Mini-Cog result may be the initiating finding; it is not by itself 99483 documentation. Medicare Advantage risk-adjustment validation asks whether a submitted diagnosis is supported by the record (PAYER POLICY): an abnormal screen supports cognitive concern under evaluation, never a dementia code on its own. Hospitals must keep records that support diagnosis and response (42 CFR 482.24), and skilled-nursing facilities assess cognition through the CMS resident assessment instrument, whose cognitive section is its own interview (LAW for the record and the assessment; no Mini-Cog mandate in either). Driving fitness and decisional capacity are decision-specific evaluations governed by state law and clinical standards (CONVENTION on the screen's role): a positive Mini-Cog can justify a driving history, a functional or on-road assessment, or a capacity evaluation; it decides neither.
Canada and Australia frame the same instrument differently, which is itself the documentation point. The Canadian Task Force on Preventive Health Care recommends against instrument-based screening of asymptomatic community-dwelling adults 65 and older (a strong recommendation on low-certainty evidence, reaffirmed 2024) while expressly excluding anyone with patient, family, or clinician concern or functional change; the 2020 Canadian consensus conference supports targeted assessment with any of several tools, so a Canadian chart records the indication (concern, not routine) and the tool chosen (CONVENTION). Physician billing is provincial: Ontario's July 2026 bulletin lists dementia-assessment fee codes (A071, C071, A073, C073, A074, C074, A078) without an age restriction and without naming an instrument (PAYER POLICY, one province). Australia's Medicare health assessment for people 75 and older (items 701, 703, 705, 707; MBS note AN.0.39, March 2026) must include psychological function including cognition and mood, must be recorded, and must offer the patient a written report; it names no instrument, and Australian guidance more often reaches for the locally developed GPCOG, the RUDAS for culturally and linguistically diverse or lower-education patients, and the KICA for older Aboriginal and Torres Strait Islander people, so a Mini-Cog result there is one input to a broader recorded assessment (PAYER POLICY for the item, CONVENTION for the tool). Austroads and the Canadian driver-fitness standard both use condition-, function-, and risk-based assessment, and neither treats a brief screen as a licensing determination. On the evidence: the 2000 development cohort (249 people, about half with dementia, uncertain cases excluded) produced the 99% sensitivity figure that still headlines most pages; the population-based 2003 sample gave 76% and 89%; the 2024 meta-analysis pooled 73% and 84% across eight primary-care and community studies (5,620 participants) with very high heterogeneity and high or unclear selection risk in most studies; a Cochrane review summarized by AAFP in 2022 found too little evidence to support routine Mini-Cog screening in primary care; a Brazilian study of 211 adults with four or fewer years of schooling found 60% and 65%; and USPSTF's 2020 I statement (an update research plan posted June 2025) finds insufficient evidence for or against screening asymptomatic older adults, while the Australian government expects updated dementia guidelines in late 2026. Report ranges and name the population; do not chart the 99%.
Rights and versions reward precision. The Mini-Cog is copyright of Soo Borson and distributed through mini-cog.com, whose 2026 permission terms tier use by who is using it: universities, foundations, health professionals, hospitals, clinics, and public-health institutes may use, reproduce, and distribute the unmodified instrument for clinical and educational purposes without permission or a licensing agreement, provided it stays free to patients; commercial entities and pharmaceutical companies need prior written permission and a license even when the end user pays nothing; and research use requires permission for everyone, with a licensing agreement when commercially funded. No express carve-out is published for a public self-scoring web tool, so a commercial operator that reproduces the stimuli and applies the algorithm should assume permission is required, and modification of any kind voids the no-permission position. Two forms share one algorithm (the standardized numerical form, dated 01.19.16, with six numbered word lists, and the graphical form that keeps the 2000 scoring), 23 language versions are listed, and availability is not validation for every population: the rights holder itself warns that people without basic literacy or analog-clock familiarity may be cognitively normal and still fail the clock. Mini-Cog is copyright Soo Borson; this page links to the official source and hosts nothing. BastionGPT is not affiliated with, or endorsed by, the rights holder or the instrument's authors. This page reproduces no word lists, clock instructions, forms, or scoring materials.
The numbers behind these errors are specific. The original 2000 development cohort gave 99% sensitivity in a sample where about half the participants had dementia; the 2024 meta-analysis pooled 73% sensitivity and 84% specificity in primary-care and community settings, so roughly one in four people with dementia screens negative; untrained raters agreed with experts on only about 60% of mildly impaired clocks in the 2001 rater study; and a Brazilian cohort with four or fewer years of schooling produced 60% sensitivity and 65% specificity. The BastionGPT Clinical Advisory Board sees the same errors most often in Mini-Cog documentation reviews:
BastionGPT is specifically trained, tuned, and clinically tested on behavioral health progress notes and screening documentation.
See how clinicians use it day to day on the AI therapy 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.
Two components sum to a total of 0 to 5: three-item delayed recall scored 0 to 3, and a clock drawing scored 2 (normal) or 0 (abnormal) with no intermediate point; inability or refusal to draw scores 0. On the rights holder's current scoring page, 0 to 2 indicates a higher likelihood of clinically important cognitive impairment (a positive screen) and 3 to 5 a lower likelihood of dementia that does not rule out some degree of impairment. The cut point was set empirically to balance sensitivity and specificity for dementia detection; it estimates no severity, no cause, and no capacity or driving fitness. Chart the components, not just the total: a 2/5 built from two words and a failed clock is a different picture from no words and a normal clock, and the follow-up examiner needs to know which.
Under the conventional rule, 3 is negative (3 to 5 means lower likelihood). The 2016 standardized form adds that a cut point below 4 may be used when greater sensitivity is desired, and some care pathways route anyone under 4 to a longer instrument; the current mini-cog.com page, however, states that treating 0 to 3 as positive has been used in some studies but has not been adequately tested against formal cognitive-disorder diagnoses. Document which rule your program applies and why, and chart the number itself with its components. A 3 built from one word and a normal clock, in a patient whose family reports decline, is exactly the case where the screen should not end the conversation whatever rule you use, and a normal clock in a patient with three words recalled and no concern is a routine negative.
No. The statute (Social Security Act 1861(hhh)) and 42 CFR 410.15 require detection of any cognitive impairment at the initial and subsequent wellness visit (G0438, G0439), defined as assessment by direct observation with due consideration of patient report and concerns raised by family, friends, or caregivers, and CMS deliberately named no instrument, reasoning that no single universally accepted screen exists. Using a brief validated tool is a convention CMS encourages, and the Mini-Cog is one option among several. Two cautions: bills in the CHANGE Act lineage before the 119th Congress would require an NIA-identified tool and documentation of the tool used, and as of August 2026 they are proposed, not law; and 99483 is a separate cognitive assessment and care plan service whose documented elements (independent historian, functional and capacity assessment, staging, medication reconciliation, safety, caregiver, care plan) go far beyond any screen.
The answer turns on who you are and whether the instrument stays unmodified and free to patients. Universities, foundations, health professionals, hospitals, clinics, and public-health institutes may use, reproduce, and distribute the current form for clinical and educational purposes without permission or a licensing agreement, which covers a printed form and an internal EHR build; the rights holder's FAQ actively recommends a discrete EHR field capturing the word-list version and both component scores. Commercial entities and pharmaceutical companies need prior written permission and a licensing agreement even when the tool is free to end users, research use always requires permission, and modification voids the no-permission position for everyone. No express carve-out is published for a public self-scoring web tool, so a commercial operator that reproduces the stimuli and applies the algorithm should ask first. Terms are as stated at mini-cog.com in August 2026; confirm with the rights holder before commercial or research use.
The clock is binary by design (2 or 0), which keeps scoring reliable for untrained staff on clearly normal and clearly abnormal drawings; the honest caveat is that agreement between untrained and expert raters fell to about 60% for mildly impaired clocks in the 2001 rater study, and the misses ran one way (subtly abnormal clocks called normal). When the drawing is borderline, get a second trained rating, keep the drawing (scan it to the chart), and state the uncertainty if it changes the classification. A patient who cannot hold a pen, cannot see the page, or has never used an analog clock has not demonstrated a cognitive clock failure even though the form would score 0: document the barrier, do not invent an adjusted score, call the result nonstandard or not interpretable, and use a screen that does not depend on the impaired function.
Believe the history, then test it. The rights holder states that a normal Mini-Cog does not rule out clinically relevant impairment, especially very mild impairment or a person with high prior ability, and that concern from a patient, clinician, or family member warrants additional evaluation whatever the score; pooled sensitivity in primary care runs about 73%, so roughly one in four people with dementia screens negative. Chart both the score and the collateral examples (onset, course, medication management, finances, driving), state that the normal screen does not close the question, and document the next step. The reverse case follows the same logic: a positive screen with no reported concern first gets a validity check (acute illness, delirium, sedating drugs, depression, sensory or language barrier, literacy, clock familiarity), then collateral and a fuller assessment; absence of concern does not erase an abnormal result.
Record that cognitive assessment was offered and why, that the patient declined (or which component), the reason if volunteered, how cognition was assessed instead (observation, patient report, collateral), any observed concern, the alternative plan or re-offer, and any immediate safety issue addressed. Do not enter a total, a zero, or a positive screen. For a US wellness visit the underlying requirement is detection by observation and history, which can still be met and documented without a completed tool. An invalid attempt (delirium or fluctuating attention, acute illness, an improvised family translation, a motor or visual barrier) is charted as attempted, nonstandard, and not interpretable, with the reason and the plan to reassess after stabilization or with an authorized language version and qualified interpreter; it is not charted as a positive screen.
Choose by the question and the setting. The Mini-Cog answers one question, whether there is enough concern to justify a fuller evaluation, in about three minutes, by any trained staff member, without certification, and free for noncommercial clinical use; its clock component is its weakness in low-literacy or non-analog-clock populations. The MoCA samples more domains and is generally regarded as more sensitive to mild cognitive impairment, at the cost of about 10 minutes and a training and certification regime for anything beyond total-score screening. The SLUMS is a free, longer screen with education-adjusted interpretation and its own training expectation. The MMSE is a licensed, per-form instrument chosen mainly for continuity with existing records or drug-coverage rules that still key to it. Many pathways run the Mini-Cog first and escalate positives (or scores under 4) to a longer instrument; no peer-reviewed same-cohort Mini-Cog-versus-MoCA accuracy study had been published as of August 2026, so claims of superiority either way outrun the evidence.
Yes. Give it the facts (form and word-list version, language, administrator, both component scores, validity conditions, disclosure, collateral, prior result, and the plan) and it drafts the full entry: components stated, validity conclusion, the screen-not-diagnosis line, disclosure and collateral, and the follow-up chain with owners and dates, ready for your review. It can also check a finished note for a total without components, a screen charted as a diagnosis, an invalidated attempt scored as abnormal, a declined screen entered as zero, and a positive result with no plan. 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 August 2026:
Educational content, not legal or billing advice. Sample notes are fictional. Follow your organization's policies and your board, payer, and jurisdiction requirements.