What is CDR-SB, and what does a small change on it actually mean?

What it is

CDR-SB - the Clinical Dementia Rating - Sum of Boxes - is a cognitive and functional rating scale used routinely in Alzheimer's clinical trials. It is the scale the field most often cites when summarising whether a treatment slowed decline.

The scale was developed in the 1980s by Dr John Morris and colleagues at Washington University. It was originally designed as a simple global staging instrument for dementia severity and has since become the dominant quantitative endpoint in late-stage Alzheimer's drug development.

How it is scored

CDR-SB is scored by a trained clinician based on a structured interview with the patient and, critically, a caregiver or close informant - somebody who knows the patient well enough to describe their functioning across daily life.

The clinician rates the patient in six "boxes" - six areas of cognitive and functional ability:

• Memory - retention of recent and remote information.
• Orientation - awareness of time, place, and situation.
• Judgement and problem solving - handling everyday decisions.
• Community affairs - independent function outside the home.
• Home and hobbies - function within the home.
• Personal care - dressing, bathing, eating, hygiene.

Each box is scored 0 (no impairment), 0.5 (questionable), 1 (mild), 2 (moderate), or 3 (severe). The six box scores are summed to produce the total CDR-SB, which ranges from 0 (no impairment in any domain) to 18 (severe impairment in all domains).

A person early in Alzheimer's disease - the population enrolled in anti-amyloid trials - typically scores between 2 and 6 on CDR-SB at baseline. The scale is sensitive enough to distinguish meaningful differences in this range, which is why it became the workhorse endpoint for the field.

Why "sum of boxes" specifically

The older version of the CDR produces a global rating - 0, 0.5, 1, 2, or 3 - that stages the patient's dementia overall. That global rating is stable across broad ranges of decline, which makes it useful for clinical staging but insensitive as a trial endpoint.

CDR-SB - the sum of all six box scores rather than just the global stage - gives a finer-grained number that changes earlier and more reliably over the timescale of a clinical trial. That is the technical reason it became the primary cognitive/functional endpoint in contemporary Alzheimer's drug development.

What a "small" change actually means

The defining feature of anti-amyloid trial readouts is that the CDR-SB differences between treatment and placebo were, in absolute terms, fractions of a point on an 18-point scale over 18 months. The debate over whether these small population-level differences translate to something clinically meaningful for an individual is the interpretive question that has defined the field. We cover this in our Signal on the CDR-SB meaningfulness debate.

Two framings matter:

• Relative slowing. A treatment that produces a smaller increase in CDR-SB over 18 months is, by definition, slowing the pace of decline. Expressed as a percentage of the placebo's increase, the numbers look different - 20–30% slowing is a larger figure than half a point.
• Patient-level meaningfulness. Whether a fraction-of-a-point slower decline is something a patient or caregiver would notice in daily life is a harder question, and where reasonable clinicians disagree. Minimum-clinically-important-difference estimates for CDR-SB range by analysis approach.

Neither framing is wrong. They are answering different questions.

What it cannot tell you

• Biological change. CDR-SB is a clinical scale; it doesn't directly measure amyloid, tau, or brain volume. For biological evidence of treatment effect, the field uses amyloid PET and tau PET - see our explainer on amyloid PET.
• Symptom-specific changes. CDR-SB summarises cognitive and functional domains at a staging level. It is not well-suited to capturing changes in behavioural symptoms, sleep, or quality-of-life measures.
• Individual prognosis. CDR-SB is a population-level endpoint. A specific patient's trajectory is informed by, but not predicted by, the group-level CDR-SB delta reported in a trial.

Why it keeps coming up

Every late-stage Alzheimer's readout - anti-amyloid, anti-tau, GLP-1, and others - will report a CDR-SB delta. The question that follows is the same one the field has been working on since the first CLARITY-AD and TRAILBLAZER-ALZ 2 reads: how to read a small number on an 18-point scale. For the broader frame on what the field now treats as binding evidence, see our Insight on real-world data as the binding signal. For the reference context of other endpoints used alongside CDR-SB, see our snapshot of clinical trial endpoints.

This page is a plain-language primer. It is not medical advice.