You are here:
Vanderbilt Biostatistics Wiki
>
Main Web
>
Projects
>
DivisOfGenIntMedAndPubHealth
>
PillCVD
>
PillCvdStatisticalAnalysis
(02 May 2008,
SvetlanaEden
)
(raw view)
E
dit
A
ttach
---++++[[PillCvdPreviousVersionOfStatAnalysis][Previous Versions]] ---+++ 10th of April, 2008 ------------------- ---++++Primary Outcomes Mixing actual and potential ADEs is not a good idea for two main reasons. First, these variable are defined in a different way. Second, the way of reporting actual ADEs is problematic for two reasons: 1) the patient is not blinded to the treatment, 2) the side effects and actual adverse drug events are mixed together, which will probably lead to the absence of intervention effect for actual ADEs outcome. Because of these issues, we should consider =potential serious ADEs= as a main primary outcome, and =preventable/ameliorable serious adverse drug events (ADEs)= as an additional primary outcome. 1 Number of =potential serious ADEs= (see [[PillCvdDefinition][PILL-CVD definitions]]) per patient during the first 30 days after hospital discharge. 1 Number of =preventable/ameliorable serious adverse drug events (ADEs)= (see [[PillCvdDefinition][PILL-CVD definitions]]) per patient during the first 30 days after hospital discharge. ---+++++Effect Modification (Interaction) Variables * Level of health literacy (according to the protocol) * Investigation site (It is possible that the intervention effect will be different in two sites. If it turns out that it is not different we can omit it when interpreting the results.) ---+++++Possible Confounders It is true that randomization makes sure that we have two very similar samples. Nevertheless the model would benefit from having all possible variables that the outcome might depend on. I would definitely put the following (the investigator should decide on that): * Number of Prescribed Medications * Age * Social support * Educational attainment * insurance * income * Cognitive function I would not put the following variables in the model (the investigator should decide on that): * hospital service * Primary language (%Q% don't you think that cognitive function would account for primary language differences) * Gender (%Q%) * Race (%Q%) ---+++++ Not Confounders * Number of Medication Changes During Hospitalization (Changes in prescription might be caused by intervention, therefore, this variable should not be included as a confounder) ---------------- ---++++Secondary Outcomes 1 Number of actual severe ADEs 1 Disease-Specific Quality of Life / Disease Control 1 Health Care Utilization (%Q% How is it measured ?) ---+++++Confounders Confounders listed under Primary outcomes. ------------------------- ---+++ANALYSIS ---++++Baseline Analysis The following baseline variables are summarized for the control and intervention groups and compared using Wilcoxon's rank-sum (for continuous variables) and Chi-Square test (for categorical variables): * Level of health literacy * Cognitive function * Educational attainment * Primary language * Gender * Race * Age * Number of prescribed medications ---++++Preliminary Statistical Summary The following variables are summarized for the control and intervention groups and compared using Wilcoxon's rank-sum test: * Number of =preventable/ameliorable ADEs= per subject * Number of =potential ADEs= per subject The following variables are summarized for the control and intervention groups and compared using Chi-Square test: * Number of =preventable/ameliorable ADEs= by severity ---++++Primary Analyses 1 Poisson regression (or proportional odds ordinal logistic regression) is used to assess the association between the number of =potential ADEs= and intervention. The model includes health literacy and site as effect modification variables and controls for variables listed under "Possible confounders". 1 Poisson regression (or proportional odds ordinal logistic regression) is used to assess the association between the number of =preventable/ameliorable ADEs= and intervention. The model includes health literacy and site as effect modification variables and controls for variables listed under "Possible confounders". ---++++Secondary Analysis The effects of the intervention on quality of life(/disease control) is assessed using proportional odds ordinal logistic regression, controlling for variables listed under "Possible Confounders". The effect of the intervention on health care utilization is assessed using Cox proportional hazard model, controlling for the variables listed under "Possible Confounders". ---++ Studio 1 [[PILLCVDStatAnalysisStudio][Notes for the Studio]] -------------------------------------- -------------------------------------- --------------------------------------
E
dit
|
A
ttach
|
P
rint version
|
H
istory
: r10
<
r9
<
r8
<
r7
|
B
acklinks
|
V
iew topic
|
Edit
w
iki text
|
M
ore topic actions
Topic revision: r10 - 02 May 2008,
SvetlanaEden
Main
Department Home Page
Biostatistics Graduate Program
Vanderbilt University Medical Center
Main Web
Main Web Home
Search
Recent Changes
Changes
Topic list
Biostatistics Webs
Archive
Main
Sandbox
System
Register
|
Log In
Copyright © 2013-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki?
Send feedback