identifying meaningful patient subgroups via clustering - sensitivity graphics bob obenchain...
TRANSCRIPT
![Page 1: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/1.jpg)
Identifying Meaningful Patient Subgroups
via Clustering - Sensitivity Graphics
Bob ObenchainCommercial Information Sciences
Outcomes Research, US Medical Division
![Page 2: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/2.jpg)
Rosenbaum PR. Observational Studies, Second Edition. New York: Springer-Verlag. 2002, page 5.
In its essence, to “adjust for age” is to compare smokers and nonsmokers of the same age.
…differences … in age-adjusted mortality rates cannot be attributed to differences in age.
Adjustments of this sort, for age or other variables, are central to the analysis of observational data.
![Page 3: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/3.jpg)
Cochran WG, Cox G.Cochran WG, Cox G. Experimental Designs 1957Experimental Designs 1957
• BlockingBlocking• RandomizationRandomization• ReplicationReplication
![Page 4: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/4.jpg)
• Epidemiology (case-control & cohort) studies• Post-stratification and re-weighting in surveys• Stratified, dynamic randomization to improve
balance on predictors of outcome• “Full Matching” on Propensity Score estimates • Econometric Instrumental Variables (LATEs)• Marginal Structural Models (InvProbWgt 1/PS)• Unsupervised Propensity Scoring: Nested
(Treatment within Cluster) ANOVA …with LOA, LTD and Error components
Forms of Local Controlfor Human Studies
![Page 5: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/5.jpg)
“Local” Terminology:
•Subgroups of Patients
•Subclasses…
•Strata…
•Clusters…
![Page 6: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/6.jpg)
Source Degrees-of- Freedom Interpretation
Clusters (Subgroups)
C = Number of Clusters
Local Average Treatment Effects (LATEs) are
Cluster MeansTreatment
within Cluster
Number of “Informative” Clusters C
Local Treatment Differences (LTDs)
Error Number of Patients 2C Uncertainty
Although a NESTED model can be (technically) WRONG, it is sufficiently versatile to almost
always be USEFUL as the number of “clusters” increases.
Nested ANOVA
![Page 7: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/7.jpg)
In general, subgroups of patients can be
considered meaningful only if patients are much
more similar within subgroups than between
subgroups.
![Page 8: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/8.jpg)
Notation for Variables
y = observed outcome variable(s)x = observed baseline covariate(s)t = observed treatment assignment
(usually non-random)z = unobserved explanatory
variable(s)
![Page 9: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/9.jpg)
When making head-to-head treatment comparisons,
subgroups of patients remain meaningful only if the
observed distributions of within subgroup differences in outcome due to treatment also
differ among subgroups.
![Page 10: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/10.jpg)
When making head-to-head treatment comparisons,
subgroups of patients remain meaningful only if the
observed distributions of within subgroup differences in outcome due to treatment also
differ among subgroups.
![Page 11: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/11.jpg)
Meaningful subgroups can only contain smaller patient subgroups that
are not meaningful.
Overshooting!!!
![Page 12: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/12.jpg)
16 Clusters (each containing both treated
and untreated patients) in a two dimensional X-space
Different Possible LTD Distributions of Y Outcomes will be Illustrated here in the Bottom Half of the Following Slides...
![Page 13: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/13.jpg)
These clusters may NOT be “meaningful” when the resulting distribution of
treatment differences in Y looks quite peaked:
Main Effect with Little Noise
16 Clusters (each containing both treated
and untreated patients) in a two dimensional X-space
0 Y-outcome-difference
![Page 14: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/14.jpg)
However, these clusters are “meaningful” when the resulting distribution of
treatment differences in Y is different from the
corresponding distribution using RANDOM subgroups:
16 Clusters (each containing both treated
and untreated patients) in a two dimensional X-space
0 Y-outcome-difference
![Page 15: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/15.jpg)
Artificial LTD Distributions fromRandom Subgroupings…
• Relatively Flat?, Smooth? or Unimodal?• Maximum Uncertainty and Bias because least
relevant comparisons are included !!!
LTD Distributions from SubgroupsRelatively Well-Matched in X-space…
• Shifted Mean? Skewness?• Distinct Local Modes?• Lower Local Variability? …Meaningful!
![Page 16: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/16.jpg)
These clusters may be meaningful when the
distribution of treatment differences in Y looks something like this:
16 Clusters (each containing both treated
and untreated patients) in a two dimensional X-space
0 Y-outcome-difference
![Page 17: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/17.jpg)
These 7 clusters are meaningful when the
distribution of treatment differences reveals a “local
mode” attributable to “adjacent” clusters:
16 Clusters (each containing both treated
and untreated patients) in a two dimensional X-space
0 Y-outcome-difference
![Page 18: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/18.jpg)
Clusters could fail to remain meaningful if the “local mode” in the distribution of treatment
differences corresponds to outcomes from widely
dispersed clusters:
16 Clusters (each containing both treated
and untreated patients) in a two dimensional X-space
0 Y-outcome-difference
![Page 19: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/19.jpg)
Subgroup-based approaches used in Observational Human Studies face “reality”:
1. (Non-parametric) Nested ANOVA models (treatment within subgroup) are “robust.”
2. Sometimes, it’s just not possible to make clearly fair comparisons!
Traditional Covariate Adjustment methods used in Randomized Clinical Trials (i.e. Least Square Means) make very strong assumptions:
1. Their (complex?) parametric models are correct.
2. Factors should compete for (causal) credit.
![Page 20: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/20.jpg)
Global / Marginal Inference…
Difference of Overall Averages…one average for each treatment group or a simple “contrast” (single degree-of-freedom)
Local / Conditional Inference…
Distribution of Local Differences…one treatment difference within each informative subgroup of similar patients
What is a Treatment Effect?
![Page 21: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/21.jpg)
Mortality Rates: Simpson’s Paradox
Mild Severe
W-Class 1% 6%Local 3% 9% Difference: 2% 3%
Total
4.4%3.8%
Mild Severe Total
W-Class 3/327 41/678 44/1005Local 8/258 3/33 11/291Total 11/585 44/711
Disease severity is a confounder here in the sense that it is associated with both outcome (mortality) and treatment choice (hospital.)
+0.6%
![Page 22: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/22.jpg)
i.e. not Generalized Linear Modelsand their Nonlinear extensions.
The “statistical methodology” engine ideal for making fair treatment comparisons is:
Cluster Analysis(Unsupervised Learning)
plus Nested ANOVA
![Page 23: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/23.jpg)
Is there statistical “theory” suggesting use of clustering to identify treatment
effects?
![Page 24: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/24.jpg)
Fundamental PS TheoremJoint distribution of x and t given p:
Pr( x, t | p ) = Pr( x | p ) Pr( t | x, p ) = Pr( x | p ) Pr( t | x ) = Pr( x | p ) times p or (1p) = Pr( x | p ) Pr( t | p )
...i.e x and t are conditionally independent given the propensity for new, p = Pr( t = 1 | x ).
![Page 25: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/25.jpg)
Conditioning (patient matching) on estimated Propensity Scores
implies both…
Balance: local X-covariate distributions must be the same for both treatments
and
Imbalance: Unequal local treatment
fractions unless Pr( t | p ) = p = 1p = 0.5
![Page 26: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/26.jpg)
Pr( x, t | p ) = Pr( x | p ) Pr( t | p )The unknown true propensity
score (in non-randomized studies) is the “most coarse”
possible balancing score.
The known X-vector itself is the“most detailed” balancing score…
Pr( x, t ) = Pr( x ) Pr( t | x )
![Page 27: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/27.jpg)
The known X-vector itself is the“most detailed” balancing score…
Pr( x, t ) = Pr( x ) Pr( t | x )
Pr( x, t | p ) = Pr( x | p ) Pr( t | p )The unknown true propensity score (in non-
randomized studies) is the “most coarse”possible balancing score.
![Page 28: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/28.jpg)
Conditioning upon Cluster Membership is intuitively somewhere between the two PS extremes in the limit as
individual clusters become numerous, small and compact…
The known X-vector itself is the“most detailed” balancing score…
Pr( x, t ) = Pr( x ) Pr( t | x )
Pr( x, t | C ) Pr( x | C ) Pr( t | x, C ) constant Pr( t | C )
Pr( x, t | p ) = Pr( x | p ) Pr( t | p )The unknown true propensity score (in non-
randomized studies) is the “most coarse”possible balancing score.
![Page 29: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/29.jpg)
Start by Clustering Patients in X-Space
Divisive Coefficient = 0.98
3clusters
21clusters
![Page 30: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/30.jpg)
In Observational Human Studies, fraction treated (imbalance) varies even more from subgroup to subgroup due to:
3. Treatment Selection Biases
In Randomized Experiments, fraction treated (imbalance) will vary from subgroup to subgroup due to:
1. Bad Luck (Murphy’s Law)
2. Small Subgroups
![Page 31: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/31.jpg)
Source Degrees-of- Freedom Interpretation
Clusters (Subgroups)
C = Number of Clusters
Local Average Treatment Effects (LATEs) are
Cluster MeansTreatment
within Cluster
Number of “Informative” Clusters C
Local Treatment Differences (LTDs)
Error Number of Patients 2C Uncertainty
Although a NESTED model can be (technically) WRONG, it is sufficiently versatile to almost
always be USEFUL as the number of “clusters” increases.
Nested ANOVA
![Page 32: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/32.jpg)
0 = Number untreated patients in th cluster > 0in i
Nested ANOVA Treatment Difference within ith Cluster:
(1 )iPS
1 = Number treated patients in th cluster > 0in i
1 0
outcomefor a treated patient outcomefor an untreated patient
i in n
1 0 1 1/i i i i iPS n n n n
PSi Local TreatmentImbalance!
![Page 33: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/33.jpg)
Local Control: A Subgroup / Sensitivity Graphics
approach to Robustness
• Replace covariate adjustment based upon a global model with inference based upon local clustering (sub-grouping) of patients in X-covariate space.
• Explore sensitivity by increasing the number of clusters, intentionally over-shooting, then recombining.
• Also vary distance metric and clustering method while employing computationally intensive algorithms and interactive graphical displays.
![Page 34: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/34.jpg)
Urgent need for “Up Front” Sensitivity Analyses
Common Threads:• Many possible answers !!!• No single approach nor set of
assumptions is clearly most appropriate.
![Page 35: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/35.jpg)
Survey of 1314 Whickham Women
y = 20 year mortality (yes or no) in 1995 follow-up study of a survey made in 1972-1974
x = age decade (20, 30, 40, 50, 60, 70 or 80) at the time of the initial survey
t = smoker or non-smoker at the time of the initial survey
Appleton DR, French JM, Vanderpump MPJ. “Ignoring a Covariate: An Example of Simpson’s Paradox” Amer. Statist. 1996; 50: 340-341.
![Page 36: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/36.jpg)
20 Year Mortality
Rate Difference
Overall Mean and +/-Two Sigma Limits for the Distribution of
LTD Differences:
Mortality Rate of Smokers minus
that of Nonsmokers.
Number of Clusters
![Page 37: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/37.jpg)
0.0
25.0
50.0
75.0
100.0
Y
0 20 40 60 80 100
Age
Simpson’s Paradox At Work:
Percentage of Smokers by Age Decade…and 20 Year Mortality Percentages for Smokers and Non-Smokers
![Page 38: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/38.jpg)
LTD Distribution of Heteroskedastic Estimates…
-0.2 -0.1 0 .1 .2 .3
49 Informative Clusters for 20 Year Mortality of 1314 Wickham Women: Smokers minus Nonsmokers
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Cu
m P
rob
-0.2 -0.1 0 .1 .2 .3
CDF
KEY QUESTIONS:
• Is this distribution mostly just noise around some central value?• How many local modes might this distribution really have?• Do the Xs predict the most likely LTD for some (or all) patients?
![Page 39: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/39.jpg)
0.00000
0.10000
0.20000
0.30000
0.40000
Y
-6.000 -4.000 -2.000 .000 2.000 4.000 6.000
X
2 3
1
|12|/
Mixture Joint Density:
![Page 40: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/40.jpg)
• Continuing evolution in methodologies for and attitudes about analyses of non-randomized human studies
• Postpone decisions whenever available data are insufficient to provide high confidence
• Statistical methods CAN work better-and-better in the dense data limit
Future “Needs”
![Page 41: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/41.jpg)
Hansen BB. “Full matching in an observational study of coaching for the SAT.” JASA 2004; 99: 609-618.
Fraley C, Raftery AE. “Model based clustering, discriminant analysis and density estimation.” JASA 2002; 97: 611-631.
Imbens GW, Angrist JD. “Identification and Estimation of Local Average Treatment Effects.” Econometrica 1994; 62: 467-475.
McClellan M, McNeil BJ, Newhouse JP. “Does More Intensive Treatment of Myocardial Infarction in the Elderly Reduce Mortality?: Analysis Using Instrumental Variables.” JAMA 1994; 272: 859-866.
References
![Page 42: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/42.jpg)
McEntegart D. “The Pursuit of Balance Using Stratified and Dynamic Randomization Techniques: An Overview.” Drug Information Journal 2003; 37: 293-308.
Obenchain RL. “Unsupervised Propensity Scoring: NN and IV Plots.” 2004 Proceedings of the ASA.
Obenchain RL. “Unsupervised and Supervised Propensity Scoring in R: the USPS package” March 2005. http://www.math.iupui.edu/~indyasa/download.htm.
Rosenbaum PR, Rubin RB. “The Central Role of the Propensity Score in Observational Studies for Causal Effects.” Biometrika 1983; 70: 41-55.
Rosenbaum PR. Observational Studies, Second Edition. 2002. New York: Springer-Verlag.
References …concluded
![Page 43: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/43.jpg)
Backup
![Page 44: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/44.jpg)
Current GuidelineInitiatives…
• Randomized Clinical Trials
– CONSORT: www.consort-statement.org
• Observational & Non-randomized Studies
– STROBE: www.strobe-statement.org
– TREND: www.trend-statement.org
![Page 45: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/45.jpg)
The Propensity Scoreof a Patient with Baseline Characteristic Vector x :
PS = Pr( t | x )is a vector of conditional
probabilities that sum to 1.
The length of the PS vector is the total (finite) number of different treatments.
![Page 46: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/46.jpg)
Propensity Scoresfor only 2 Treatments:t = 1 (new) or 0 (standard)
p = Propensity for New Treatment = Pr( t = 1 | X ) = E( t | X ) = a scalar valued function of X only
X = vector of baseline covariate values for patient
PS = (p, 1p)
![Page 47: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/47.jpg)
0 Y-outcome-difference
![Page 48: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/48.jpg)
![Page 49: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/49.jpg)
0 Y-outcome-difference
![Page 50: Identifying Meaningful Patient Subgroups via Clustering - Sensitivity Graphics Bob Obenchain Commercial Information Sciences Outcomes Research, US Medical](https://reader036.vdocument.in/reader036/viewer/2022062619/55151c53550346c77d8b4fe5/html5/thumbnails/50.jpg)
0 Y-outcome-difference