Opens in a new tab

MMSE:
Scoring and Interpretation

The Mini-Mental State Examination (MMSE) is a brief examiner-administered test of global cognitive function with a conventional total of 0 to 30. MMSE scoring is done by the examiner, so the statistical questions come after: screening cutoffs as conditional rules, education and age effects, ceiling effects in repeated measures, SPSS and R checks on the total, and reporting wording.

ITEMSBrief task set, 30 points
SUBSCALESTask groups; total is standard
RESPONSE SCALEExaminer-scored tasks
SCORE RANGE0 to 30
DEVELOPEDFolstein, Folstein and McHugh, 1975

What the MMSE Measures

Folstein, Folstein and McHugh (1975) introduced the MMSE as a practical way for a clinician to grade a patient’s current cognitive state. It samples several areas of cognition in a short structured examination, including orientation, registration, attention and calculation, recall and language, and researchers use it as a descriptive measure, a screening result or an eligibility criterion in studies of aging, neurology and health.

The total reflects performance at one assessment and is not a neuropsychological profile. Low scores can arise from dementia, but also from limited education, language barriers, sensory impairment, acute illness or poor test conditions, and a high score does not rule out mild impairment. A review of the literature concluded that the MMSE should not by itself be used as a diagnostic tool for dementia (Tombaugh & McIntyre, 1992).

Structure of the MMSE

  • Items: a short set of standardized tasks scored by the examiner, with points awarded for correct performance. The conventional maximum is 30.
  • Task groups: orientation, registration, attention and calculation, recall and language. The total is the standard outcome, and group-level scores need separate justification.
  • Direction: higher totals indicate better performance. There are no Likert items and no reverse-scored items.
  • Administration: scoring depends on examiner training and standardized administration. Record the examiner, the setting and the test date.
  • Versions: the original MMSE and the MMSE-2 are different editions, and score equivalence is not assumed. Name the edition and language used.
  • Translations: language, education and cultural background affect performance, so use a version with evidence for the target population.

The instrument itself is available from the publisher, Psychological Assessment Resources (PAR).

How to Score the MMSE

  1. Document the edition, the language version, the examiner’s training and the test date for each assessment.
  2. Have the examiner administer and score each task by the instrument’s own scoring instructions. The tasks and point allocation are not reproduced on this page.
  3. Sum the points to the conventional 0 to 30 total only when every required task was administered.
  4. Distinguish a task that was not administered from one answered incorrectly. An unadministered task is missing, and must not be counted as zero or imputed without a prespecified rule.
  5. Analyze the total as a bounded integer score, and treat any cutoff as a conditional decision rule with a stated source. The code below checks an already scored total.
SPSS SYNTAX

* Illustrative syntax: mmse_total is the examiner-scored total, valid values 0 to 30.
FREQUENCIES VARIABLES=mmse_total
  /STATISTICS=MEAN MEDIAN STDDEV MIN MAX.
COMPUTE mmse_range_check = (mmse_total < 0 OR mmse_total > 30).
FREQUENCIES VARIABLES=mmse_range_check.
R

# data: one row per assessment; mmse_total is the examiner-scored total.
stopifnot("mmse_total" %in% names(data))
stopifnot(all(is.na(data$mmse_total) | (data$mmse_total >= 0 & data$mmse_total <= 30)))
summary(data$mmse_total)
mean(data$mmse_total == 30, na.rm = TRUE)  # proportion at the ceiling

Interpreting MMSE Scores

Score band Interpretation
30 Maximum total; does not show that every cognitive domain is intact
Lower totals Reduced performance on the sampled tasks
A split near 23 and 24 Long used as a screening convention; not optimal in every population
Other cutoffs Chosen by age, education, setting and purpose, with their source reported

Source: Tombaugh & McIntyre (1992); Mitchell (2009). The 23/24 split is a long-used convention, not a universally validated threshold.

The same total can mean different things in a memory clinic and in a community survey. In a meta-analysis of 34 dementia studies and five studies of mild cognitive impairment, accuracy depended on the setting, and the MMSE was best at ruling out dementia in community and primary care (Mitchell, 2009). If a study applies a cutoff, report its source, the target condition and the reference standard. Do not borrow sensitivity or specificity values from unrelated populations.

A total below a screening cutoff marks a person for fuller assessment and does not establish dementia. Age, education, language and cultural background affect scores (Tombaugh & McIntyre, 1992), and ceiling effects limit sensitivity to mild impairment.

Psychometric Properties

Property Published finding Source
Original method Practical bedside method for grading cognitive state Folstein et al. (1975)
Reliability and validity Judged satisfactory in a comprehensive review of the literature Tombaugh & McIntyre (1992)
Criterion validity High sensitivity for moderate to severe impairment, lower for mild impairment Tombaugh & McIntyre (1992)
Demographic effects Scores were affected by age, education and cultural background, not by gender Tombaugh & McIntyre (1992)
Pooled accuracy for dementia Sensitivity 79.8% and specificity 81.3% in memory clinics; 85.1% and 85.5% in community settings Mitchell (2009)

These are source-specific findings and cannot be assumed to hold as accuracy estimates in a new sample.

Using the MMSE in a Study

As an outcome, baseline covariate or eligibility criterion, the MMSE total is bounded between 0 and 30 and often piles up near the top. Inspect the distribution, then choose a model that fits repeated assessments and attrition, such as a linear mixed model, and state the missing-data assumption. Do not turn a continuous total into a dichotomy unless the clinical question requires it, since that discards information.

If the question is how well a cutoff identifies a condition, that is a diagnostic-accuracy design against a reference standard, with sensitivity, specificity and their confidence intervals, and not a scale-reliability question. When groups differ in education, ask whether the total differences reflect cognition or the way the test samples it, and record examiner training, test version and any steps taken against practice effects at repeat visits.

Reporting the MMSE

Global cognitive function was assessed with the [edition and language] Mini-Mental State Examination (MMSE; Folstein et al., 1975). [Examiner role] administered and scored the test, and total scores were analyzed on the 0 to 30 scale, with higher scores indicating better performance. Scores were used [as a continuous variable or with a prespecified cutoff of X, from source and population]. A task that was not administered was treated as missing, and [missing-data approach] was applied. The test version and examiner training are described in [section].

References

Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). “Mini-mental state”: A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12(3), 189-198. https://doi.org/10.1016/0022-3956(75)90026-6

Mitchell, A. J. (2009). A meta-analysis of the accuracy of the mini-mental state examination in the detection of dementia and mild cognitive impairment. Journal of Psychiatric Research, 43(4), 411-431. https://doi.org/10.1016/j.jpsychires.2008.04.014

Tombaugh, T. N., & McIntyre, N. J. (1992). The Mini-Mental State Examination: A comprehensive review. Journal of the American Geriatrics Society, 40(9), 922-935. https://doi.org/10.1111/j.1532-5415.1992.tb01992.x

Frequently Asked Questions

The conventional total ranges from 0 to 30, with higher scores indicating better performance.

No. A split near 23 and 24 is a long-used screening convention, and accuracy varies by setting and population. Confirmation needs a diagnostic assessment.

Yes, with attention to the distribution, ceiling and floor effects and the study design. A mixed model is common for repeated assessments.

No. They are different tests with different content, distributions and validation evidence, so totals do not convert one to the other.

NEXT STEP

Need MMSE Data Analyzed
or Checked?

OnlineSPSS is a statistical consultancy for researchers. Share your design and dataset, and a statistically trained Research Coordinator reviews it. The scope and one fixed project fee are confirmed before work begins.

No Commitment Required

Strictly Confidential

Fixed Project Fee

In-House PhD-Level Statisticians

Same-Day Response

Serving Researchers Since 2014