Measuring change from phone intake to treatment follow-up
|
Published: October, 2001
|
The ALERT system assesses patient change from the
point of initial phone contact to a follow-up assessment six
months after treatment. There are of course many challenges
to the systematic collection of so many repeated measures
in a naturalistic setting across a long period of time. Missing
data at the various assessment points compound to limit the
number of cases with complete data across all measurement
points. Despite these limitations, these data are unique in
permitting a glimpse at patient change across an entire episode
of care.
The initial measure of severity is captured during the telephone
intake by the use of three items from the LSQ. The member
is asked to respond to the items using a scale from 0 (or
"Never") to 4 (or "Almost Always"). The
three items were selected due to their ability to differentiate
a treatment from a non-treatment sample. A fourth item is
used as a self-report screening item for substance abuse problems.
- I am satisfied with my life
- I feel hopeless about the future
- I feel sad. (Note: wording changed from "blue"
to "sad" after feedback from customer service
associates.)
In order to permit ready comparison of the three-item scale
to the full-scale LSQ scores, a regression analysis was conducted
to provide the formula for converting the 3-item scale score
to its 30-item equivalent. A sample of 442 cases was identified
with data at triage and the first session. The mean score
at triage for this sample (using the 30-item equivalent score)
was 54.5, and the standard deviation was 18. The mean score
for the first session was 53.2 for this sample. The mean score
at the first session for all members in the PBH data repository
is 52.5, and this non-significant difference supports the
premise that this triage sample is comparable in severity
to the entire PBH patient population. The total average change
from the point of phone triage to the first session is a nominal
1.3 points.
The follow-up measure was collected by systematically mailing
out LSQs to all adult members in the ALERT system after no
additional data had been received for six months. These follow-up
mailings did not begin until Q1 2000. Of the first 2700 questionnaires
mailed out, approximately 10% were returned as undeliverable
due to an address change. A total of 393 were completed and
returned, for a return rate of 16%, which is comparable to
return rates for large HMOs using a similar method.
This follow-up sample provides an estimate of the rate of
change during the post-treatment period. A total of 286 follow-up
LSQ protocols were matched with complete LSQ scores at the
first session. Of these 286 cases, 139 also contained a second
data point during the treatment episode.
In virtually all respects this sample appears comparable
to the PBH data repository. The mean LSQ score at intake was
53.6, with a standard deviation of 18. Forty eight percent
(48%) contained two data points, compared to the overall PBH
average of 42% with at least two data points. Those cases
with two data points displayed virtually identical patterns
of change to the PBH data repository. The patterns are so
close that the regression formulas to predict change, when
rounded to two decimal places, are identical. The status at
follow-up for patients with a single measurement at the first
session was comparable to those with multiple measurement
points in treatment.
The following graph shows the regression lines for change
at the end of treatment and at six-month follow-up. Across
the range of severity, members tended to show continued improvement
after treatment. The amount of change is 3.4 points on average,
or an effect size of .19.
Taken together, these samples provide a unique opportunity
to look at patterns of change across the entire treatment
episode, from initial phone contact, to the end of treatment,
to the post-treatment consolidation of treatment gains. The
following graph displays the percentage of overall change
that occurs during each phase of the treatment process.
Back to Top
 
|
