One of the key features of the ALERT system is the ability
to track at-risk cases based upon trajectory of change projections.
Once the LSQ/YLSQ has been administered at least twice, the
system calculates the most likely outcome using trajectory
of change projections. These are based on formulas derived
from cases in the data repository. The system is designed
to target approximately 10% of cases at highest risk for a
poor outcome. In order to test the validity of this logic,
a sample of 2,343 cases (adults and children combined) with
assessments at the first and third sessions was used to identify
at-risk cases and track the eventual outcome of these cases.
This analysis used calculations identical to those employed
every day by the ALERT system in generating the daily High
Risk Report. The result was that 204 (9%) of the cases were
targeted at the third session as high risk for a poor
outcome. This high risk group has scores averaging well into
the severe range at the third session, and they have worsened
an average of 16 points (-.85 effect size) since the first
session.
The most probable outcome for these cases is simple—members
are likely to end treatment based on the fact that they are
feeling worse after three visits. Only 31% of these cases
have a subsequent assessment at the fifth or later session,
compared to 42% of members not identified as being at-risk.
Those at-risk cases that fortunately do continue in treatment
to at least the fifth session show significant improvement
beyond the third session, averaging 13 points of improvement
(.7 effect size) between the third session and last assessment
point.
This analysis indicates that the logic for targeting at-risk
cases is valid. The percentage of cases targeted in this analysis
is similar to the percentage targeted by the algorithms built
into the daily High Risk Report (9% versus 10%). These cases
appear to leave treatment prematurely at a higher rate than
the comparison group. Those members who continue in treatment
show significant improvement from that point forward. (Nonetheless,
it should be noted the gains do not fully offset the deterioration
seen in the first 3 sessions.) These findings suggest that
quality improvement efforts should focus on identifying methods
to keep these at-risk patients engaged in treatment for a
sufficient duration to experience benefits.
