Schizophrenia Bulletin, Vol. 28, No. 4, 2002
4. To improve coverage of disorganized symptoms, an item for bizarre behavior was added.
5. In place of separate ratings for syndromal depres- sion and mania, severity ratings > 2 were permitted only when a full mood syndrome had ever been present. Suici- dal thoughts or attempt (DE-li), subjective report of per- vasively depressed mood (DE-3), classical manic features (M-3), and bipolar course (M-4) were eliminated as redundant.
6. Item MP-3 (concurrent psychotic and mood symp- toms) was deleted because it can be derived from high scores on psychosis items with absence of prolonged non- affective psychosis.
7. A single rating of atypical/complicating features was substituted for multiple items.
Version 2 of the LDPS is shown in table 5. Note that for some items there is a checklist to indicate presence of specific symptoms. These have been included to permit using the form to rate all DSM-IV or RDC criteria for schizophrenia and schizoaffective disorders, but their use may be considered optional, particularly when a more detailed inventory of symptoms such as the SAPS and SANS will be completed.
Second Reliability Study. ICCs for the second reliability study (table 6, for LDPS version 2) demonstrated good- excellent reliability for most items. Reliability was less adequate for psychosis without prominent mood symp- toms (whose duration must be judged to differentiate DSM-IV schizoaffective disorder and schizophrenia), poverty of speech, and bizarre behavior. For some items (mania, concurrent manic mood + delusions or hallucina- tions, maximum number of manic features, and compli- cating factors), low variation among subjects (mean squares subjects) prevented meaningful interpretation.
The preliminary version of the LDPS demonstrated ade- quate interrater reliability, a factor structure consistent with previous research, and high intercorrelations among some items, which permitted abbreviation of the scale. This has the advantage of making the scale easier to apply, although a reduced item set can have the drawback of restricting the variability of scores. LDPS version 2 con- tains 20 items plus a rating of quality of information, and optional presence/absence ratings of specific features rele- vant to categorical diagnostic criteria. Most items showed good-excellent reliability, but reliability was poor for poverty of speech and modest for bizarre behavior, and too few subjects had manic symptoms in this exercise to per- mit meaningful conclusions. The unique aspects of the scale are that all of the relevant dimensions of psychotic
D.F. Levinson et al.
symptoms, plus mood and mood psychosis symptoms and course of illness variables, can be rated on a lifetime rather than cross-sectional basis using a relatively brief instru- ment that experienced research clinicians can use after only minimal training.
Further studies of samples with a broader range of diagnoses will be needed to determine whether additional training increases reliability. The data collected in most studies may prove insufficient to achieve high reliability for items such as poverty of speech, bizarre behavior, and the relative timing of mood and psychotic symptoms. One benefit of lifetime-dimensional scales may be to identify dimensions for which reliability tends to remain low and for which better assessment methods may be needed.
While the LDPS can be used to study any type of psy- chotic disorder, it may prove to be most useful for studies in which a broad range of psychotic subjects are inter- viewed and the clustering of these symptoms can be con- sidered in a noncategorical fashion. We would suggest that the current focus on categorical diagnoses has produced a narrow view of the spectrum of psychotic disorders. For example, if one excludes all subjects without an unequivo- cal major diagnosis, one ignores the many subjects who have ambiguous, mild, or partial syndromes. It may be more fruitful to capture the full range of psychotic subjects on the basis of dimensional ratings, and then to perform quantitative analyses and/or to select subgroups for cate- gorical analyses after considering the full distribution and clustering of scores. Examples of efforts to develop a dimensional model include a thought-provoking series of studies by van Os and colleagues (1999a, 19996, 2000) and a longitudinal study by Arndt et al. (1995).
Dimensional ratings could also play a useful role in clinical practice. Within a categorical system, clinicians often reach rapid conclusions based on a few acute symp- toms. Lifetime dimensional ratings could promote more comprehensive assessment in training and practice. We would note, however, that in developing a dimensional instrument, we do not make any assumption about whether it will ultimately be more productive to base etiologic studies on dimensional ratings, clusters based on these rat- ings, or traditional categories. The development of rating strategies like the LDPS will facilitate study of this issue.
The scale has a number of shortcomings. Its reliability and factor structure have not been studied with enough subjects or with a sufficiently diverse sample. A few items have not yet been shown to be highly reliable, and it will have to be investigated whether this is due to inadequate training, the nature of the case material, or the wording of the items. It does not provide a full profile of mood symp- toms and their longitudinal course, which might be accom- plished by improving this scale or by supplementing it when necessary. Its concurrent and external validity must