How assessment scoring works

A validated assessment turns a handful of item responses into one number — and that number only means something once you read it against severity bands, clinical cutoffs, and a reliable-change threshold. Here is how to read the score, not just the arithmetic.

From raw score to severity band

Every validated measure sums its items into a total, and that total maps onto a severity band the authors validated against clinical samples. A PHQ-9 of 14 is not just “14” — it is moderate depression. The band, not the raw number, is what carries clinical meaning, because it has been calibrated against real outcomes.

Common cutoffs and reliable-change thresholds

A few of the most-used measures, with their severity bands, screen-positive cutoff, and the change needed to count as real (see the next section):

Measure	Severity bands	Screen-positive	Reliable change
PHQ-9 (depression)	0–4 minimal · 5–9 mild · 10–14 moderate · 15–19 mod. severe · 20–27 severe	≥10 screens positive	~5 points
GAD-7 (anxiety)	0–4 minimal · 5–9 mild · 10–14 moderate · 15–21 severe	≥10 screens positive	~4 points
PCL-5 (PTSD)	0–80 total	~31–33 suggests probable PTSD	~10 points
ISI (insomnia)	0–7 none · 8–14 subthreshold · 15–21 moderate · 22–28 severe	≥15 clinical insomnia	~6 points

These are screening cutoffs, not diagnoses. A positive screen means “look closer,” not “diagnose.”

When is a change “real”? Reliable change

Scores wobble from week to week. A one- or two-point move is usually measurement noise, not progress. The Reliable Change Index (RCI) is the threshold a score must cross for the change to be statistically unlikely to be error. A client whose PHQ-9 drops from 18 to 11 has moved 7 points — past the ~5-point threshold — so that is real improvement, not a wobble. A drop from 14 to 12 is not.

Read the trend, not the dot

One score is a snapshot; the trend is the story. Direction across three or more administrations tells you more than any single value. Watch for sustained direction, plateaus worth raising in session, and reversals that warrant attention before the next appointment.

Talking to clients about scores

Sharing scores well strengthens the alliance instead of feeling like a report card:

Normalize early. “We’ll check in with a few quick questions each time so we can both see what’s working.”
When scores improve, reinforce the client’s effort without overpromising a straight line.
When scores plateau, treat it as clinical information to explore, not a failure.
When scores worsen, be direct and compassionate — a rising score is data to work with, not a verdict.

Let the scoring happen for you

Theracharts scores every assessment automatically, applies the validated severity bands, flags reliable change, and charts the trend — so you read the clinical story instead of doing the arithmetic.

Get started free Browse assessments

Frequently asked questions

What is a clinical cutoff on an assessment?

A cutoff is the score the instrument's authors validated as the point where a condition becomes likely enough to warrant further evaluation. For example, a PHQ-9 of 10 or above is the common screen-positive cutoff for major depression. A cutoff signals screening positive — it is not a diagnosis.

Does a higher score mean a diagnosis?

No. Validated symptom measures are screening and monitoring tools, not diagnostic instruments. A high score tells you symptoms are elevated and warrant clinical attention; the diagnosis is a clinical judgment made in context.

How much does a score have to change to be meaningful?

Enough to clear the measure's reliable change threshold — the point past which a change is unlikely to be measurement error. Roughly 5 points on the PHQ-9, 4 on the GAD-7, and 10 on the PCL-5. Smaller moves are usually noise, not progress.

These pages are educational. Validated assessments are screening and monitoring tools, not diagnoses — interpret every score in clinical context. If a client is in crisis, call or text 988 (Suicide & Crisis Lifeline, US).