Brief Psychiatric Rating Scale

Abstract

The Brief Psychiatric Rating Scale is a type of psychiatric status rating scale: a short, clinician-rated instrument that scores the severity of psychotic and general psychiatric symptoms on a fixed set of items so that a patient's condition can be quantified and tracked. Introduced to give the first controlled trials of antipsychotic drugs a common yardstick, it condenses a psychiatric interview into eighteen symptom ratings, each scored on a seven-point severity anchor, and summed into a single total. This article treats the scale as a measurement instrument: the eighteen anchored items, the four- to five-factor structure that shows the total is not one thing, the anchoring that makes two raters agree, the twenty-four-item Expanded version, and the floor-corrected percentage change that decides treatment response. Three interactive demonstrations model the factor grouping, anchored rating, and the response calculation.

Keywords: Brief Psychiatric Rating Scale, symptom severity, psychosis, factor structure, treatment response

The Brief Psychiatric Rating Scale (BPRS) is a clinician-administered instrument that assigns a number to the current severity of a defined set of psychiatric symptoms, most of them the symptoms of psychosis. It was built to solve a specific problem: the psychopharmacology trials of the early 1960s needed a rapid, reproducible way to measure whether a drug changed a patient's condition, and a clinician's global impression — better, worse, unchanged — was too coarse and too idiosyncratic to compare across raters and studies (Overall & Gorham, 1962). The scale's answer was to decompose the clinical picture into a fixed list of symptom constructs, rate each on an ordered severity scale, and sum them, so that the same patient could be scored repeatedly and a falling total read as improvement. That design made the BPRS the template for the symptom rating scales that followed, and it remains in use six decades later precisely because its brevity — eighteen items scored from a single interview — buys a measurement that a busy clinic or a large trial can actually complete.

What the Brief Psychiatric Rating Scale Is

The BPRS is a status rating scale: it presupposes that a clinician is assessing a patient's current psychiatric state and quantifies how severe that state is, rather than deciding what disorder the patient has. Overall and Gorham built it from a set of symptom constructs distilled from earlier, longer rating systems, choosing a small number of clinically salient dimensions that a trained rater could judge from a standard interview and observation (Overall & Gorham, 1962). The original report worked with sixteen constructs; the version that became standard carries eighteen, adding excitement and disorientation, and each is rated on a seven-point scale of severity. The instrument's economy is the point. A rater can complete it in the time of a single clinical interview, which is what allowed it to be administered repeatedly across the many assessment points of a drug trial, and repeated administration is the whole purpose of a status scale — the total is only informative because it can be compared to the same patient's total last week.

Two properties distinguish the BPRS from a diagnostic instrument. First, it is severity-graded, not categorical: every item runs from absent to extreme, so the output is a position on a continuum, not a yes-or-no. Second, it is anchored: each of the seven points on each item is tied to a described threshold of severity, so that a rating of 4 is meant to denote the same moderate level of the symptom for every rater and every patient. Whether that intention holds is an empirical question about the scale's reliability, and much of the BPRS literature exists to test and improve it (Woerner et al., 1988).

The Eighteen Items and Their Anchors

The eighteen items sample a deliberately broad range of psychopathology: somatic concern, anxiety, emotional withdrawal, conceptual disorganization, guilt feelings, tension, mannerisms and posturing, grandiosity, depressive mood, hostility, suspiciousness, hallucinatory behaviour, motor retardation, uncooperativeness, unusual thought content, blunted affect, excitement, and disorientation. The list mixes symptoms a patient reports (anxiety, depressive mood, guilt) with signs a clinician observes (mannerisms, blunted affect, motor retardation), and this mixture is deliberate — a scale meant to capture the whole clinical picture of a psychotic illness must draw on both what the patient says and how the patient appears.

Each item is scored on the same seven-point severity scale: 1 (not present), 2 (very mild), 3 (mild), 4 (moderate), 5 (moderately severe), 6 (severe), and 7 (extremely severe). Because the minimum rating is 1 rather than 0, a patient with no symptoms at all still scores 18, and the maximum is 126; this floor of 18 is not a technicality but a fact that shapes how change is calculated, as the worked example below shows. The severity points are anchored — accompanied, in the trained versions of the scale, by explicit descriptions of what each level of each symptom looks like — and the quality of those anchors largely determines whether two raters watching the same interview arrive at the same number (Woerner et al., 1988).

The Factor Structure

The BPRS total treats eighteen distinct symptoms as one number, but the symptoms do not vary together as if they were one thing. Factor analyses of large BPRS datasets consistently recover several correlated but separable dimensions. A meta-analysis pooling the published factor solutions found a stable four-factor structure — an affect or anxiety–depression factor, a positive-symptom or thought-disturbance factor, a negative-symptom or withdrawal–retardation factor, and a resistance or hostility factor — with the items loading in much the same way across studies despite differences in sample and setting (Shafer, 2005). Some solutions add a fifth activation factor, splitting excitement and related items from hostility. The practical import is that two patients with the same total of, say, 45 can be clinically opposite — one withdrawn and retarded, the other agitated and hostile — because the total sums across dimensions that move independently. Reporting the factor subscales rather than the bare total recovers information the sum discards, which is why analyses of the BPRS and its Expanded form continue to argue over the correct factor count and item assignment (Dazzi et al., 2016). The first demonstration makes the grouping concrete by letting the reader assign the eighteen items to their factors.

The eighteen items and their four factors

Select a factor to see which items load on it. The total score sums all eighteen, so it collapses these separable dimensions into one number.

All eighteen items shown. A stable four-factor structure recurs across studies (Shafer, 2005); some solutions split a fifth activation factor from hostility.

Anchoring and Reliability

A total score is only as trustworthy as the agreement between the raters who produce it: if two clinicians watching the same interview assign different totals, a change between visits could reflect a change of rater rather than of patient. The single most effective way to raise that agreement is to anchor the scale — to replace the bare severity labels with explicit descriptions of what each level of each symptom looks like, so that raters are calibrated to a common standard rather than to their private sense of what counts as moderate. Providing such anchor definitions measurably improved the inter-rater reliability of the BPRS, and the anchored version became the basis for reliable multi-site use (Woerner et al., 1988). Reliability and construct validity were examined in detail for the psychosis items, confirming that the scale's core dimension behaves as a coherent measure of the severity of a schizophrenic syndrome when administered by trained raters (Andersen et al., 1989). The lesson generalizes beyond this one scale: the reproducibility of any clinician-rated instrument is manufactured, not assumed, and is bought with anchor descriptions and rater training. The second demonstration shows an anchored item being rated and how the anchor narrows the gap between two independent raters.

Anchored rating and rater agreement

Two clinicians rating the same interview disagree by 2 points. With explicit anchor descriptions the ratings are calibrated to a common standard, so the spread stays within one point (Woerner et al., 1988).

The Expanded BPRS

The most widely used modern revision is the twenty-four-item BPRS-Expanded, version 4.0, which added items and, more importantly, supplied a full set of anchor points and a written administration manual so that the scale could be given the same way by different raters at different sites (Ventura et al., 1993). Standardizing administration — the wording of probes, the definitions of each anchor, the conduct of the interview — attacks a source of error that item selection alone cannot, and the Expanded version's manual is a large part of why it became the research standard. Its factor structure has been re-examined in the same way as the original, with meta-analytic work arguing for a refined item grouping and, on that basis, a further revision of the instrument (Dazzi et al., 2016). The Expanded scale has also proved sensitive outside psychosis: applied to outpatients with unipolar depression, its factor structure held and its scores tracked treatment change, evidence that the instrument measures a broad band of psychopathology rather than psychosis alone (Zanello et al., 2013).

Measuring Change: Response and the Floor Correction

The clinical and regulatory value of the BPRS lies in reading a change in the total as a change in the patient, and the standard summary of that change is the percentage reduction from baseline — conventionally, a reduction of at least 20% marks a minimal response and 50% a substantial one. But the calculation has a trap that is specific to scales like this one. Because every item floors at 1, the lowest possible total is 18, not 0, and a naive percentage computed against the raw total treats those unavoidable 18 points as if they were reducible symptoms. The result is that raw percentage change systematically understates improvement, and the distortion is worst for the sickest patients. The correction, argued for in the analysis of what these psychosis scores mean, is to subtract the minimum possible score from both the baseline and the endpoint before taking the ratio, so that the percentage is computed over the range the scale can actually move (Leucht et al., 2005). The choice between the two formulas can move a patient across the response threshold, changing the reported outcome of a trial, which is why the correction matters beyond arithmetic pedantry. The third demonstration lets the reader set a baseline and endpoint total and watch both the raw and floor-corrected percentages resolve against the response thresholds.

Response: raw versus floor-corrected reduction

The two formulas disagree on this patient. Because the scale floors at 18, the raw percentage understates improvement and can place the same patient on opposite sides of a response threshold. The corrected figure, taken over the range above the floor, is the defensible one (Leucht et al., 2005).

Worked Example

Consider a patient admitted with acute psychosis whose baseline BPRS total is 52. After six weeks of treatment the total is 34.

The raw percentage reduction divides the change by the baseline: the absolute reduction is 52 − 34 = 18 points, and 18 ÷ 52 = 0.346, or 34.6%. By this figure the patient has not reached the conventional 50% threshold for a substantial response, and would be classed as only a partial responder.

The floor-corrected reduction first subtracts the minimum possible score of 18 from both totals, because those 18 points can never be removed. The corrected baseline is 52 − 18 = 34 and the corrected endpoint is 34 − 18 = 16, so the reduction is 34 − 16 = 18 points over a corrected baseline of 34: 18 ÷ 34 = 0.529, or 52.9%. By this figure the same patient does meet the 50% threshold and is a substantial responder.

The identical clinical improvement yields two different verdicts depending on whether the scale's floor is corrected for, and the corrected figure is the defensible one because it measures the reduction against the range the instrument can actually traverse (Leucht et al., 2005). The effect is not small and not rare: for any patient whose baseline is well above the floor, the raw percentage will read several points lower than the corrected one, which is enough to reclassify borderline responders and, in aggregate, to change a trial's reported response rate.

Discussion

The BPRS solved the problem it was built for. It gave psychopharmacology a fast, reproducible severity measure, and the evidence base for antipsychotic treatment rests in large part on the scores it and its descendants produce. Its longevity is a measure of how well the original design choices held: eighteen broadly sampled items, a single anchored severity scale, and a total short enough to administer repeatedly. But the same compression that makes the scale usable also imposes permanent costs. Summing eighteen symptoms into one number discards which symptoms are present, so that clinically opposite patients can share a total; treating that ordinal sum as an interval quantity, so that a drop from 52 to 34 is treated as a 34.6% reduction, imposes arithmetic on anchors that were never guaranteed to be equal-interval; and reading a percentage change against the raw total, ignoring the floor of 18, distorts the very quantity the trial reports. Each simplification is defensible, and each can mislead when the total is read as if it were the patient.

The scale's relationship to its successors clarifies what it is. The Positive and Negative Syndrome Scale took the BPRS items as its core and added a dedicated negative-symptom subscale, precisely because the BPRS undersampled the withdrawn, blunted end of psychopathology relative to the florid, positive end (Kay et al., 1987). That the BPRS could be extended rather than replaced is the strongest evidence that its underlying design was sound; that it needed extending is the clearest statement of its limits.

Current Directions

Work on the BPRS now runs along three lines. The first is the continued re-examination of its factor structure. Meta-analytic pooling of the many published solutions for the Expanded version has been used to argue for a specific refined item grouping and, from it, a further revision of the instrument — an attempt to settle by aggregation a question that individual studies, with their varying samples, could not (Dazzi et al., 2016). The second is the scale's use beyond psychosis. Applied to unipolar depression its structure held and its scores moved with treatment (Zanello et al., 2013), and a broader evaluation has framed the BPRS explicitly as a transdiagnostic instrument, arguing that a general psychopathology scale can quantify severity across diagnostic boundaries rather than within a single disorder (Hofmann et al., 2022). This reframing sits alongside a wider movement in psychiatric measurement toward dimensional, cross-cutting symptom assessment and away from strictly disorder-bound scales, of which the critical clinimetric re-appraisal of the legacy depression scales is a parallel example (Carrozzino et al., 2020). The third is the persistent problem of content coverage: because any fixed item set samples some symptoms and omits others, scales for overlapping conditions do not measure quite the same thing, the same heterogeneity documented in detail for the common depression scales (Fried, 2017). Digital administration and more frequent, in-the-moment symptom sampling are the emerging responses to the constraint that a scale can only be as current as its last interview.

Common Misconceptions

The BPRS diagnoses schizophrenia.

It does not. The scale measures the severity of a symptom state that the clinician has already judged to be present; a high total quantifies how ill the patient is, not what disorder they have. The instrument was never validated to establish a diagnosis, and its items span depression, anxiety, and general psychopathology as well as psychosis (Overall & Gorham, 1962).

A single BPRS total captures the patient's condition.

The total sums four to five independent symptom dimensions, so two patients with the same score can be clinically opposite — one withdrawn and retarded, the other agitated and hostile. The factor subscales, not the bare total, carry that distinction (Shafer, 2005).

A percentage reduction in the total is a percentage reduction in illness.

Only if the scale's floor is corrected for. Because the lowest possible total is 18, not 0, a raw percentage computed against the total understates improvement, and the distortion can move a patient across the response threshold. The reduction should be taken over the range above the floor (Leucht et al., 2005).

Glossary

Affect factor.

The BPRS dimension grouping anxiety, depressive mood, guilt, and somatic concern; one of the four factors recovered across studies.

Anchor point.

The described threshold attached to each of the seven severity points of an item, intended to make a given rating denote the same severity for every rater.

Anchored version.

A form of the BPRS supplying explicit anchor descriptions for each severity level, developed to raise inter-rater reliability.

BPRS-Expanded (BPRS-E).

The twenty-four-item version 4.0 with a full set of anchor points and a written administration manual, the modern research standard.

Conceptual disorganization.

A BPRS item rating the degree of disordered, disconnected, or incoherent thought as evidenced in the patient's speech.

Factor structure.

The set of correlated symptom dimensions recovered when the item scores are factor-analysed; for the BPRS, four to five factors rather than one.

Floor correction.

Subtracting the minimum possible score (18) from baseline and endpoint totals before computing percentage change, so the reduction is measured over the scale's actual range.

Hostility factor.

The BPRS dimension grouping hostility, suspiciousness, and uncooperativeness; sometimes termed the resistance factor.

Inter-rater reliability.

The degree to which two clinicians scoring the same interview produce the same total; a precondition for reading change over time as change in the patient.

Negative-symptom factor.

The BPRS dimension grouping emotional withdrawal, blunted affect, and motor retardation — the deficit, withdrawn end of psychopathology.

Positive-symptom factor.

The BPRS dimension grouping unusual thought content, conceptual disorganization, hallucinatory behaviour, and grandiosity — the florid, productive symptoms of psychosis.

Response.

A proportional improvement from baseline — conventionally a 20% reduction for a minimal response and 50% for a substantial one — used to classify treatment outcome.

Severity anchor.

The seven-point ordered scale, from 1 (not present) to 7 (extremely severe), on which every BPRS item is rated.

Status rating scale.

An instrument that quantifies the current severity of a presumed psychiatric condition, designed to be re-administered so that change can be tracked.

Total score.

The sum of the eighteen item ratings, ranging from 18 to 126, treated as a single index of overall symptom severity.

Transdiagnostic scale.

An instrument used to quantify symptom severity across diagnostic boundaries rather than within a single disorder.

Key Researchers

Stefan Leucht (contemporary). Professor of psychiatry at the Technical University of Munich; established what psychosis rating-scale totals mean clinically and clarified how percentage change should be computed. ORCID - Wikidata

Keith H. Nuechterlein (contemporary). Professor of psychiatry and psychology at UCLA; co-author of the BPRS-Expanded and a leading schizophrenia researcher known for the vulnerability-stress model. Faculty Page - Google Scholar

John E. Overall (1929-2016). Biostatistician and professor of psychiatry at the University of Texas Medical School at Houston; with Donald Gorham he created the Brief Psychiatric Rating Scale, the first widely adopted psychiatric severity measure. Memorial

Joseph Ventura (contemporary). Research psychologist at the University of California, Los Angeles; lead author of the anchored BPRS-Expanded version 4.0 and its administration manual. ORCID

Frequently Asked Questions

What is the Brief Psychiatric Rating Scale?
It is a short, clinician-administered instrument that rates the severity of eighteen psychiatric symptoms, most of them symptoms of psychosis, on a seven-point scale and sums them into a total. It was introduced to give early antipsychotic drug trials a reproducible way to measure symptom change (Overall & Gorham, 1962).

How many items does the BPRS have and how are they scored?
The standard version has eighteen items, each scored from 1 (not present) to 7 (extremely severe), giving a total between 18 and 126. The Expanded version 4.0 has twenty-four items with a full administration manual (Ventura et al., 1993).

Does the BPRS diagnose a disorder?
No. It measures the severity of a symptom state the clinician has already judged to be present; it quantifies how ill the patient is, not what disorder they have. It was not validated to make a diagnosis (Overall & Gorham, 1962).

Is the BPRS total a single dimension?
No. Factor analyses recover four to five separable dimensions (affect, positive symptoms, negative symptoms, and hostility/resistance), so patients with the same total can be clinically very different. The subscales carry information the total hides (Shafer, 2005).

How is inter-rater agreement improved?
By anchoring the scale, supplying explicit descriptions of what each severity level of each symptom looks like, and by training raters to a common standard. Adding anchor definitions measurably raised the BPRS's inter-rater reliability (Woerner et al., 1988).

Why subtract 18 when calculating percentage change?
Because every item floors at 1, the lowest possible total is 18, not 0, so a raw percentage computed against the total understates improvement. Subtracting the minimum from baseline and endpoint measures the reduction over the range the scale can actually move (Leucht et al., 2005).

How does the BPRS relate to the PANSS?
The Positive and Negative Syndrome Scale took the BPRS items as its core and added a dedicated negative-symptom subscale, extending rather than replacing the older instrument (Kay et al., 1987).

Can the BPRS be used outside psychosis?
Yes. Its factor structure held and its scores tracked treatment change in unipolar depression (Zanello et al., 2013), and it has been evaluated as a transdiagnostic measure of general psychopathology across diagnoses (Hofmann et al., 2022).

References