Brief Symptom Inventory 18

By: Derogatis, Leonard R, 20010101, Vol. 15

Mental Measurements Yearbook

Review of the Brief Symptom Inventory 18 by ROGER A. BOOTHROYD, Associate Professor, Department
of Mental Health Law and Policy, Louis de la Parte Florida Mental Health Institute, University of South
Florida, Tampa, FL:

DESCRIPTION. The Brief Symptom Inventory 18 (BSI 18), as its name implies, is an 18-item "self-report
symptom inventory designed to serve as a screen for psychological distress and psychiatric disorders"
(manual, p. 1). According to the administration manual the measure was designed for use "with a broad
spectrum of adult medical patients 18 or older and adult individuals in the community who are not
currently assigned patient status" (p. 3). Patients rate their level of distress during the past week on
each of the 18 symptoms using a 5-point Likert-type scale ranging from 0 (not at all) to 4 (extremely).
The author indicates that the items assess three symptom dimensions: Somatization (6 items),
Depression (6 items), and Anxiety (6 items), as well as a Global Severity Index (GSI) based on all 18 items.
The BSI 18 is recommended for use by health and behavioral health professionals as a psychological
screen, to support clinical decisions, for monitoring treatment progress, and to assess treatment
outcomes. The BSI can be completed by most respondents in 4 minutes and is purportedly written at a
sixth grade reading level.

DEVELOPMENT. The BSI 18 has an extensive developmental history. It is a reduced version of the 53-
item Brief Symptom Inventory (BSI; Derogatis & Melisaratos, 1983) that was developed from the
Symptom Checklist-90 (SCL-90; Derogatis, Rickels, & Rock, 1976) that originally evolved from the
Hopkins Symptoms Checklist (HSCL; Derogatis, Lipman, Rickels, Uhlenhuth, & Covi, 1974). The BSI 18
focuses on three symptom dimensions in contrast to the nine dimensions assessed by its predecessors.
The author indicates that the three symptom dimensions of the BSI 18 were selected because they
represent about 80% of the psychiatric disorders that occur in primary care practice. The author also
states that although multiple criteria were used to determine which items to retain, the high prevalence
of the specific symptoms in clinical disorders was the most significant selection factor.

TECHNICAL.

Scoring and standardization. In the absence of missing responses, raw scores are simply the sum of the
item responses within a symptom domain. The manual includes scoring instructions for dealing with
missing responses. Raw scores are converted to standardized T scores using the norm tables provided.
Gender specific normative data are provided on two samples: a community sample of 1,134 adults and
an oncology sample of 1,543 adults being treated for cancer. Although the age distributions of these
samples are provided, no information is presented regarding the racial/ethnic composition of either
sample, raising some question about the ethnic/racial diversity of the norming samples. Scores can then
be plotted on the appropriate profile sheet. Computerized scoring is available through the purchase of
MICROTEST Q 5.04 Assessment System from NCS Assessments. The software supports scoring of the BSI
18 and other NCS assessments, reporting results, and storing and exporting data. Optical scanning is also
available; however, the time and effort required for hand scoring and interpretation is minimal.

Reliability. Internal consistency reliability estimates were derived from the community sample. Alpha
coefficients for the three symptom dimensions and GSI are .74 (Somatization), .84 (Depression), .79
(Anxiety), and .89 (GSI), and are certainly very acceptable. Additionally, these reliability estimates
compare favorably with those derived from the longer BSI on a sample of 719 psychiatric outpatients.
Although no test-retest reliability studies are reported on the BSI 18, the author provides test-retest
estimates ranging from .68 to .84 on the symptom dimensions over an unspecified time interval based
on a sample of 60 nonpatients who completed the BSI. GSI test-retest estimate was .90.

Validity. As with many newly developed measures, evidence of validity is limited. The construct validity
of the BSI 18 was assessed by correlating the three symptom dimension scores and GSI with the
corresponding scores on the SCL-90-R. All correlations were high ranging from .91 on the Somatization
dimension to .96 on Anxiety (Depression and GSI were both .93) suggesting little information was lost
with the reduced number of items.

The factor structure of the BSI 18 was examined using data from the community sample as a means of
validating the hypothesized symptom dimensions. Although the results of a principal component
analysis support a four-factor solution, the author argues that the findings are not "fundamentally
inconsistent with the hypothesized structure of the BSI 18 test" (manual, p. 14). His rationale is that the
items loading on the fourth factor, representing panic, are subsumed under anxiety disorders in the
DSM-IV (American Psychiatric Association, 1994).

Although no specific studies using the BSI 18 are reported, evidence of the measure's convergent-
discriminant validity and criterion-related validity is inferred on the basis of studies conducted with its
predecessors the BSI and SCL-90.

COMMENTARY. The BSI 18 is the newest incarnation of the Hopkins Symptoms Checklist that has been
evolving over a span of nearly 30 years. The measurement foundation of the BSI 18 is quite strong even
though the psychometric properties of this specific rendition are not well understood. The measure is
brief and easy to score. The three symptom dimensions of the BSI 18 should identify individuals with the
most common mental health problems. The dimensions are highly correlated to those in the more
extensive SCL-90-R, supporting the validity of scores from the BSI 18. However, it would be a worthwhile
effort to compare scores from the BSI 18 to other independently developed brief patient self-report
symptomatology measures such as the Colorado Symptom Index (Shern, et al., 1994) as was done with
the BSI by Conrad et al. (2001). This would provide additional validity evidence for the BSI 18. The
administration manual is well written and contains information frequently omitted such as how to treat
missing data in scoring. Over a quarter of the manual is devoted to "Specific Application of the BSI 18
Test" (p. 15); however, all of the studies summarized in this section were conducted using the longer 53-
item BSI.

In Peterson's (1989) review of the BSI, he questioned whether the reduction in administration time from
15-20 minutes for the SCL-90 to the 7-10 minutes of the BSI was meaningful in light of the potential loss
of clinical sensitivity. His concern certainly remains apropos with the BSI 18 given the number of
symptom domains has been reduced from nine to three and that administration time is now a mere 4
minutes. Perhaps as Peterson suggested the time is approaching when we will just ask people "Do you
feel depressed?"

SUMMARY. The BSI 18 appears to be a useful measure for assessing anxiety, depression, and
somatization as well as for obtaining an overall level of psychological distress. Although few studies have
been conducted assessing the psychometric properties of the BSI 18, it is an abbreviated version of a
frequently used and psychometrically tested measure of mental health symptomatology.

REVIEWER'S REFERENCES

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.).
Washington, DC: Author.

Conrad, K. J., Yagelka, J. R., Matters, M. D., Rich, A. R., Williams, V., & Buchanan, M. (2001). Reliability
and validity of a modified Colorado Symptom Index in a national homeless sample. Mental Health
Services Research, 3, 141-153.

Derogatis, L. R., & Melisaratos, N. (1983). The Brief Symptom Inventory: An introductory report.
Psychological Medicine, 13, 595-605.

Derogatis, L. R., Rickels, K., & Rock, A. (1976). The SCL-90 and the MMPI: A step in the validation of a
new self-report scale. British Journal of Psychiatry, 128, 280-289.

Derogatis, L. R., Lipman, R. S., Rickels, K., Uhlenhuth, E. H., & Covi, L. (1974). The Hopkins Symptom
Checklist (HSCL): A measure of primary symptom dimensions. In P. Pichot (Ed.), Psychological
measurement in psychopharmacology (pp. 79-111). Basel, Switzerland: Karger.

Peterson, C. A. (1989). [Review of the Brief Symptom Inventory.] In J. C. Conoley & J. J. Kramer (Eds.),
The tenth mental measurement yearbook (pp. 112-113). Lincoln, NE: Buros Institute of Mental
Measurements.

Shern, D. L., Wilson, N. Z., Coen, A. S., Patrick, D. C. et al. (1994). Client outcomes II: Longitudinal client
data from the Colorado Treatment Outcome Study. Milbank Quarterly, 72, 123-148.

Review of the Brief Symptom Inventory 18 by WILLIAM E. HANSON, Assistant Professor, Department of
Educational Psychology, University of Nebraska-Lincoln, Lincoln, NE:

DESCRIPTION. The Brief Symptom Inventory 18 (BSI 18) is a norm-referenced, self-report instrument
composed of, as its namesake suggests, 18 items. According to the manual, it is first and foremost a
screening instrument, "developed primarily as a highly sensitive screen for psychiatric disorders and
psychological disintegration and secondarily as an instrument to measure treatment outcomes"
(administration, scoring, and procedures manual, p. 1). The manual indicates that it may be useful in
most clinical and research settings and may be used with a wide range of medical and community
populations, including, to name a few, people (18 years old or older) who have been diagnosed with
cancer, who have a compromised immune system (e.g., HIV/AIDS), and/or who are experiencing chronic
pain or sexual difficulties. The manual did not, however, indicate the minimum reading level required to
complete the instrument.

DEVELOPMENT. The BSI 18 is the fourth iteration in a family of well-known and widely used symptom-
based instruments. Its parent instrument, the Brief Symptom Inventory (BSI; Derogatis, 1993; Derogatis
& Spencer, 1982), is a derivative of the Symptom Checklist-90-Revised (SCL-90-R; Derogatis, 1977, 1994),
which, in turn, is a derivative of the Hopkins Symptom Checklist (HSCL; Derogatis, Lipman, Rickels,
Uhlenhuth, & Covi, 1974). It was developed with the following considerations/assumptions in mind:
Mood and anxiety disorders are common, though difficult to detect/diagnose appropriately, especially in
medical and community settings; somatic symptoms co-occur frequently with Mood and anxiety
disorders, further complicating the diagnostic picture; and administration and scoring of the instrument
should be brief (<4-5 minutes), straightforward, and cost-effective.

The instrument has 18 nonoverlapping items, all of which were taken directly from the BSI (Derogatis,
1993; Derogatis & Spencer, 1982). They were selected based on "multiple considerations, including
prevalence of the symptom, item analysis characteristics, and loading saturations in factor analyses of
the BSI and SCL-90-R" (manual, p. 2). No other details related to the development of the instrument
were reported (e.g., specific results of item analyses, factor loadings, or pilot testing). It is difficult,
therefore, to evaluate the appropriateness of the selection criteria and the decision rules that were used
to choose the 18 items.

Nevertheless, the BSI 18 has three, six-item subscales: Somatization (SOM), Depression (DEP), and
Anxiety (ANX); and a Total, or Global Severity Index (GSI), score. The subscale and total scores measure
constructs identical to the like-named BSI scores (Derogatis, 1993; Derogatis & Spencer, 1982).
Specifically, the SOM subscale score measures "distress caused by the perception of bodily dysfunction,
focusing on symptoms arising from cardiovascular, gastrointestinal, and other physiologic systems"
(manual, p. 5). The DEP subscale score measures "core symptoms of various syndromes of clinical
depression" (e.g., disaffection, dysphoric mood, suicidal ideation; manual, p. 5). The ANX subscale score
measures "symptoms that are prevalent in most major anxiety disorders" (e.g., nervousness, tension,
apprehension; manual, p. 5). The GSI score is a composite measure of psychological distress and is "the
single best indicator of the respondent's overall emotional adjustment and psychopathologic status"
(manual, p. 6). Each subscale is scored on a 5-point Likert-type scale, ranging from 0 not at all to 4
extremely. Subscale scores can range from 0-24 and can, if desired, be summed to obtain a total score,
which can range from 0-72.

TECHNICAL INFORMATION.

Standardization procedures: Norming. Normative information is available for two separate samples: an
adult nonclient, or "community" sample; and an adult nonclient "oncology" sample. The community
sample consisted of 1,134 adult employees (605 men and 517 women; 12 did not report their sex) of an
unspecified U.S. corporation. The employees were of diverse age (reported range: 18-69), with the
majority being between the ages of 40-59. No other characteristics of this sample were reported (e.g.,
race/ethnicity, education level, or SES), making it difficult to determine its representativeness. Details
related to how these individuals were identified and/or recruited to participate in the original norming
study were also not reported.

The oncology sample consisted of 1,543 adults (802 men and 741 women) who had been diagnosed
with cancer and who were patients at an unspecified U.S. east coast cancer center. The adult cancer
patients were of diverse age (reported range: <30-80+), with the majority being between the ages of
50-69. At least 20 different types/manifestations of cancer were represented. Similar to the community
sample, no other characteristics of this sample were reported.
Of note, both the community and the oncology norms are gender-keyed. Separate norms are available
for men and women. Combined norms are also available. However, test users are "strongly
recommended" to use the separate, gender-keyed norms (manual, p. 37).

Administration and scoring. Administration and scoring procedures are straightforward and easy to
understand. The BSI 18 may be administered by hand or by computer. It may also be scored by hand,
using a preprinted scoring sheet that includes detailed scoring directions, or by computer, using scoring
software that may be purchased from the publisher. Computer-based progress reports are also available
for purchase. The availability of progress reports is appealing and, if used, may prove to be a useful
feature of the scoring software.

If administered by hand, the instructions, a sample test item, and the 18 test items are printed on one
side of a single sheet of paper. Test takers are instructed to read a list of "problems," or symptoms, and
to indicate how much each symptom has distressed or bothered them over the past week (i.e., "the past
7 days including today").

If scored by hand, the scoring directions and either a community- or oncology-based blank profile graph
(one for men and one for women) are printed on one side of a single sheet of paper. Nine specific, step-
by-step scoring directions are provided, including directions for determining a profile's validity, for
calculating estimated values of omitted items, and for plotting raw score totals on the blank profile
graph. As a general rule of thumb, a test taker may omit two items per subscale without jeopardizing the
validity of the BSI 18's scores. If, however, three or more items are omitted from any of the three
subscales, the scores should be considered invalid. Also, test users are reminded to always calculate
estimated values of omitted items, as these estimates are included in raw score totals.

Interpretation. To facilitate interpretation, raw score totals are converted to area, or uniform, T-scores,
with a mean of 50 and a standard deviation of 10. Conversion tables for the two normative samples
(community and oncology) are in the manual. Percentile rank equivalents of the raw scores are also in
the manual.

The manual recommends that interpretation of the BSI 18 occur at three interrelated levels: the global
level; the dimensional level; and the symptom, or item, level. Basically, it involves three steps. The first
step occurs at the global level. It involves determining "caseness" (manual, p. 23), that is, whether or not
the test taker's scores meet predetermined, empirically based criteria for identification/positive risk of
psychological distress-stated differently, that they fall within the clinical range. If so, then the test user is
encouraged to evaluate the test taker further.

The second step occurs at the dimensional level. It involves considering each BSI 18 subscale score
independently, in the following recommended order: DEP, ANX, and SOM. If any of these subscale
scores fall within the clinical range (T score >63), then the test user is also encouraged to evaluate the
test taker further.

The third and final step occurs at the symptom, or item, level. It involves considering individual BSI 18
items. For example, Item 17 (an item related to suicidal ideation) and Items 9, 12, and 18 (items related
to panic attacks) should be examined closely to determine if further evaluation is necessary in these
clinically important areas. Test users are referred to Derogatis and Savitz (1999) for additional
information related to interpretation.
RELIABILITY AND VALIDITY. Reliability estimates of the BSI 18 subscale and total scores are, based on the
adult nonclient "community" sample mentioned earlier, satisfactory and meet traditional professional
standards of acceptability. Estimates of internal consistency range, in this sample, from "fair"
(Somatization [.74], Anxiety [.79]) to "good" (Depression [.84], Total [.89]; cf. Cicchetti, 1994). Test-retest
reliability estimates are not reported. However, test-retest reliability estimates of the BSI (Derogatis,
1993; Derogatis & Spencer, 1982) are reported. These estimates range, in a different sample of 60
nonpatients, from .68 (Somatization) to .90 (Total). By reporting these estimates, it appears that the
author is relying on reliability induction, whereby test-retest reliability of the subscale and total scores is
generalized from one sample (and, in this case, a different test, the BSI) and assumed to be an
appropriate estimate for another sample (and the BSI 18). BSI 18 test users are, therefore, encouraged
to compute reliability estimates, including estimates of internal consistency and test-retest reliability,
for their "data in hand."

Standard error of measurement (SEM), a common method of estimating the reliability of a test taker's
score, is also not reported for the BSI 18 subscale or total scores. Subscale and total score
intercorrelations are not reported either, which limits appropriate and responsible interpretation of the
subscale and total scores and precludes profile interpretation altogether (Anastasi, 1985).

Preliminary evidence of equivalence, or correspondence, between BSI 18 scores and SCL-90-R scores is
provided in the manual. These correlations, which are based on the community sample, ranged from .91
(SOM) to .96 (ANX). The manual states that "basic considerations concerning such issues as face and
content validity have been addressed previously in the context of the development of the parent
instrument" (pp. 13-14). Preliminary evidence of criterion-related validity of BSI 18 scores is also
provided in the manual. This evidence is based on a selective review of published studies that used the
BSI-not the BSI 18. The studies were related to eight different clinical areas: screening studies, cancer
populations, pain assessment/management, military populations, HIV/AIDS research, immune system
functioning, human sexuality, and medical and law students.

Finally, preliminary evidence of construct validity, in particular, convergent validity, is provided in the
manual. This evidence is based on correlations between BSI and SCL-90-R scores and MMPI clinical,
content, and Tryon cluster scores that measure similar constructs. Reported correlation coefficients
ranged from .40-.72 and were generally in the expected direction. Specific evidence of discriminant
validity was not reported. Of relevance here, factor analyses of the BSI 18 resulted in a four-factor
solution that accounted for a respectable 57.2% of the total variance. The four identified factors and
corresponding item loadings are more-or-less consistent with the author's hypothesized, a priori
dimensional structure of the instrument and its scores.

CONCLUSIONS AND RECOMMENDATIONS. The BSI 18 is an intriguing new, commercially available

screening instrument. Given the popularity and track records of its parent instruments (e.g., BSI, SCL-90-
R), it likely has a promising future. Because of how it was developed, its overall strengths and
weaknesses parallel those of its predecessors (for MMY reviews of the BSI, see Cundick, 1989, and
Peterson, 1989; for MMY reviews of the SCL-90-R, see Pauker, 1985, and Payne, 1985). Its most obvious
strengths include: professionally developed, user-friendly testing materials and scoring software;
brevity; straightforward administration, scoring, and interpretation procedures; availability of computer-
based progress reports; and acceptable estimates of internal consistency. Its most obvious weaknesses
include: limited normative, reliability, and validity data, including data related to score sensitivity and
specificity; lack of profile interpretation capabilities; and, similar to other brief, self-report instruments,
susceptibility to distortion and "faking" of responses.

That said, the manual, in its current form, makes it virtually impossible to determine the BSI 18's true
merits. Though well organized and well written, it is largely uninformative. Too few details are included.
Test users who refer to it for information regarding: the reading level required to complete the
instrument; specific details related to how/why the test items were chosen; characteristics of the two
normative samples; test-retest reliability estimates; SEM estimates; subscale and total score
intercorrelations; and, perhaps most importantly, adequate evidence of construct and predictive
validity, will be disappointed. The omission of these types of details is significant and, in this reviewer's
opinion, should be addressed in future editions of the manual.

All things considered, use of the BSI 18 may, quite frankly, be premature. Clearly, additional normative,
reliability, and validity data are needed on the BSI 18 to justify its use, especially for clinical purposes.
Until that occurs, it is recommended that the BSI 18 be used only for research purposes and as an
adjunct, or supplement, to traditional, more well-established screening instruments/interview methods
in clinical applications and settings. Also, it is recommend that, at this time, only the Total, or GSI, score
be used.

Prospective test users looking for a suitable, though slightly longer, alternative to the BSI 18 may find
the OQ-45.2 (Lambert et al., 1996) to be a potentially satisfactory, psychometrically sound option.

REVIEWER'S REFERENCES

Anastasi, A. (1985). Interpreting results from multiscore batteries. Journal of Counseling and
Development, 64, 84-86.

Cicchetti, D. V. (1994). Guidelines, criteria, and rules of thumb for evaluating normed and standardized
assessment instruments in psychology. Psychological Assessment, 6, 284-290.

Cundick, B. P. (1989). [Review of the Brief Symptom Inventory]. In J. C. Conoley & J. J. Kramer (Eds.), The
tenth mental measurements yearbook (pp. 111-112). Lincoln, NE: Buros Institute of Mental
Measurements.

Derogatis, L. R. (1977). Symptom Checklist-90-R (SCL-90-R) administration, scoring, and procedures

manual I. Baltimore, MD: Clinical Psychometric Research.

Derogatis, L. R. (1993). Brief Symptom Inventory (BSI) administration, scoring, and procedures manual
(3rd ed.). Minneapolis: NCS Pearson, Inc.

Derogatis, L. R. (1994). Symptom Checklist-90-R (SCL-90-R) administration, scoring, and procedures

manual (3rd ed.). Minneapolis: NCS Pearson, Inc.

Derogatis, L. R., Lipman, R. S., Rickels, K., Uhlenhuth, E. H., & Covi, L. (1974). The Hopkins Symptom
Checklist (HSCL): A measure of primary symptom dimensions. In P. Pichot (Ed.), Psychological
measurements in psychopharmacology (pp. 79-111). Basel, Switzerland: Karger.

Derogatis, L. R., & Savitz, K. L. (1999). The SCL-90-R, Brief Symptom Inventory, and matching clinical
ratings scales. In M. E. Maruish (Ed.), The use of psychological testing for treatment planning and
outcomes assessment (2nd ed.; pp. 679-724). Mahwah, NJ: Lawrence Erlbaum Associates.
Derogatis, L. R., & Spencer, P. (1982). Brief Symptom Inventory (BSI) administration, scoring, and
procedures manual I. Baltimore, MD: Clinical Psychometric Research.

Lambert, M. J., Hansen, N. B., Umphress, V., Lunnen, K., Okiishi, J., Burlingame, G. M., & Reisinger, C. W.
(1996). Administration and scoring manual for the OQ-45.2. Stevenson, MD: American Professional
Credentialing Services LLC.

Pauker, J. D. (1985). [Review of the SCL-90-R]. In J. V. Mitchell, Jr. (Ed.), The ninth mental measurements
yearbook (pp. 1325-1326). Lincoln, NE: Buros Institute of Mental Measurements.

Payne, R. W. (1985). [Review of the SCL-90-R]. In J. V. Mitchell, Jr. (Ed.), The ninth mental measurements
yearbook (pp. 1326-1329). Lincoln, NE: Buros Institute of Mental Measurements.

Peterson, C. D. (1989). [Review of the Brief Symptom Inventory]. In J. C. Conoley & J. J. Kramer (Eds.),
The tenth mental measurements yearbook (pp. 112-113). Lincoln, NE: Buros Institute of Mental
Measurements.

The Brief Symptom Inventory (BSI) is a 53 item scale scored on a 5-point Likert like scale from
"not at all" to "very much" (Derogatis, 1975). The BSI is the short version of the SCL-R-90
(Derogatis, 1983), measuring the same factors and consisting of those items with the highest
loadings in a factor analysis. The 9 dimensions of the BSI are: somatization, obsessive-
compulsion, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid
ideation, and psychoticism. The Hebrew version of the BSI has been widely used in clinical
work and in research (Bachar, 1997). In this study the alpha of the total scale was 0.96.

Derogatis, L. (1983). Symptom Checklist 90-R: Administration, Scoring and Procedures Manual.
Baltimore: Clinical Psychometric Research.
Derogatis, L. R. B. (1993). Brief Symptom Inventory AdministrationScoring Procedure Manual (
4th edition ed.). Minneapolis MN: National Computer Systems Research.
Bachar, E., Cannetti, L., Bonne, O., DeNour, A. K., & Shalev, A. (1997). Physical punishment
and signs of mental distress in normal adolescents. Adolescence, 32, 945-958.