Outcome Measures
|
Published: October, 2001
|
The LSQ and YLSQ are investigational instruments based on
a revised set of 30 items from the OQ-45 and Youth Outcome
Questionnaire (YOQ). The LSQ and YLSQ were developed based
on the assumption that the items from parent instruments would
exhibit similar psychometric properties as part of a 30-item
measure. Items were selected for inclusion in the new measures
based on three criteria:
- High face validity for symptoms and concerns common
to patients in mental health treatment
- Ability to differentiate treatment from non-treatment
samples
- Sensitivity to change in treatment samples while remaining
relatively stable in non-treatment samples.
Drs. Lambert and Burlingame have accumulated a large data
repository of the OQ-45 and YOQ items through collaboration
with several large managed care companies. The size and scope
of this repository permitted complex modeling of how the new
instruments could be expected to perform in real world clinical
settings. A sample, consisting of over 3200 adults and 600
children/adolescents, was utilized to model the properties
of the new tests. The sample came from a population of commercially
insured outpatients in managed care settings. This sample
serves as a benchmark against which to compare PBH’s outcomes.
A critical element of the ALERT system is the ability to
predict the expected change for individual patients using
the outcome measures. These trajectory of change formulas
for the ALERT system were derived from the data repository
of OQ-45 and YOQ questionnaires, again with the assumption
that these calculations would remain constant when the items
were presented in a 30-item scale. The formulas provide a
target outcome for each case against which to compare actual
results. This provides the basis for case mix adjustment and
benchmarking LSQ results against the norms from the data repository.
The assumption that the 30-item instruments would perform
as predicted proved accurate. An expected change score was
established at intake for each patient in the ALERT system,
using the formulas derived from the data repository. At the
end of treatment, it is possible to compare the difference
between the predicted change and the actual change for individual
patients or large groups of patients. In the case of the LSQ,
the average difference between the predicted change and actual
change in the sample for this report is .33 points (.018 effect
size). This represents less than 5% of the total change.
The YLSQ outcomes have tended to show more improvement than
the formulas predict, reducing the overall accuracy. However,
the YLSQ sample (n=479) is less than 20% the size of the LSQ
sample, so some of the variance may be due to small sample
size.
The YLSQ permits completion by either an adult or an adolescent.
The protocols completed by adults had slightly higher scores
at intake than adolescent-completed tests (43.7 versus 39.5)
and also showed slightly more change after controlling for
intake severity. However, neither difference approached statistical
significance (p>.2).
In addition to these full-length measures, three items from
the LSQ are used as part of the telephonic intake and triage
procedure. These three items are used to estimate improvement
between the first phone contact and subsequent treatment sessions.
The results using this three-item measure are discussed in
a subsequent section.
Back to Top
 
|
