Knowledge Item:
CA-Medical Outcomes-26
More Complex Hierarchical Linear
Models Showing Differential Change in CD4 Counts for Groups Formed by
Demographic Categories and Drug Abuse Status
Hierarchical Linear Modeling (or HLM)
methods are used to study change in CD4 counts over time. For each
person, an individual trend is line fit to that patient's CD4 counts over time. The method uses all observations for individual
patients and takes into account the fact that patients received the
assessments at different times in their treatment histories. After
fitting "individual" curves for each patient, overall trends
in the "average curves" or "trajectories" are
studied and related to other factors. HLM modeling
can be considered the "definitive" method for studying change
in the kinds of naturally-occurring treatment assessment data from these
projects.
The hierarchical models show that on average patients experience increasing CD4 counts over the
course of their involvement with the programs represented here. This
Knowledge Item studies the rate of change as a function of different
service needs and vulnerabilities of the patient. A
number of service needs-vulnerabilities are related to both the
initial CD4 count status of the patient at the time of enrollment
into the program and to the rate of change throughout the program.
The following can be concluded from
these models.
-
CD4 counts
increase over time for all patients (p < .001).
-
There is
no statistically significant differential change in CD4 counts over time by
Gender.
-
There is
no statistically significant differential change in CD4 counts over time by
Project Type.
-
There is a
Gender by Project Type interaction such that females from
University-based Comprehensive Healthcare projects showed less increase in CD4 counts
over time than other groups (p
< .002). Note that these patients start with higher CD4
counts at enrollment.
-
The positive
change in CD4 counts over time is largely unrelated to patient
demographic factors, including drug abuse.
This difference
is not statistically significant.
This difference
is not statistically significant.
This difference
is not statistically significant.
Knowledge Item Citation: Huba, G. J., Melchior, L. A., Panter, A. T., and the HRSA/HAB SPNS Cooperative Agreement Steering Committee (1998-2001). Knowledge Item:
CA-Medical Outcomes-26 from
HRSA/HAB's SPNS Cooperative Agreements on Innovative Models of Care, The Measurement Group Knowledge Base on HIV/AIDS Care, Online at www.TheMeasurementGroup.com.
Last Updated:
March 25, 2005; data through June 15,
1999; analyses conducted March - May 2000.
|
