3.9 Tree Aggregation File Parameters

The Tree Aggregation file is optional. This file must be specified if TreeScan analysis is requested. This file defines parameters necessary to execute a TreeScan analysis. Table 3.15 contains detailed specifications for this file.

Table 3.15: TREEAGGFILE specification
Parameter Field Name Description
TreeScan analytic dataset identifier TREEANALYSISID Corresponds to analytic dataset creation for individual TreeScan execution.
Note 1: one TREEANALYSISID is required for each unique combination of TREEANALYSISGRP and LEVELID for stratified analyses.
Format: SAS character $3
Example: 001
Label for the analytic dataset TREEANALYSISGRP Standardized name used to differentiate analytic datasets for use by TreeScan.
Note 1: TREEANALYSISGRP values must remain consistent during the course of a sequential surveillance activity.
Format: SAS character $40
Example: Insulin_4_365
Run identifier RUNID RunID assigned to the run that corresponds to the specified TREEANALYSISGRP.
Format: SAS character $5
Example: r01
Stratification identifier LEVELID Stratification variable identifying each unique combination of strata. Receives a unique level value that remains consistent across requests. Strata level variables must be requested in the distributed User-defined Strata Levels Look-up file.
Format: SAS character $3
Example: 000
Numeric identifier for stratification levels LEVELNUM Numeric identifier for the number of categories within a LEVELID. There needs to be 1 row for the number of categories within a LEVELID. For example, if LEVELID = 001 (age), and there are 2 age groups (0-65, 65+), then there should be 2 LEVELNUM assignments: LEVELNUM = 01 for LEVELNUMLBL = 0-65, and LEVELNUM = 02 for LEVELNUMLBL = 65+.
Note 1: if no stratification requested, (i.e., LEVELID = 000), then LEVELNUM should be left blank.
Format: SAS character $32
Example: 01
Label for stratification levels LEVELNUMLBL Label for the number of categories within a LEVELID. Within a given strata, this is exactly the text for a particular level as seen in the CIDA output files. For example, if LEVELID = 001 (age), and there are 2 age groups (0-65, 65+), then there should be 2 LEVELNUM assignments: LEVELNUM = 01 for LEVELNUMLBL = 0-65, and LEVELNUM = 02 for LEVELNUMLBL = 65+.
Note 1: if no stratification requested, (i.e., LEVELID = 000), then LEVELNUM should be left blank.
Note 2: when there are multiple stratification variables at the specified LEVELID, then the desired values in LEVELNUMLBL must align exactly with the order of variables in the distributed User-defined Strata Levels Look-up file.
Format: SAS character $32
Example: 0-65
Risk window start RWSTART Used to define the start of a user-defined fixed risk window when performing a Bernoulli TreeScan analysis with a self-controlled risk interval design (as opposed to a Tree-Temporal).
Note 1: only relevant for Type 3 analyses and is optional based on whether a fixed window analysis is being performed (as opposed to a variable window analysis with a temporal scan).
Note 2: must be set to missing for Type 2 analyses.
Note 3: TREEANALYSISGRP can be used for multiple sets of risk and control windows but each unique analysis requires its own TREEANALYSISID.
Note 4: control windows may precede or follow risk windows and the two windows are not required to be contiguous.
Note 5: this should be specified as a “timeline” label as opposed to an algebraic quantity, assuming Day 0 = 0.
Format: Numeric
Example: 1
Risk window end RWEND Used to define the end of a user-defined fixed risk window when performing a Bernoulli TreeScan analysis with a self-controlled risk interval design (as opposed to a Tree-Temporal).
Note 1: only relevant for Type 3 analyses and is optional based on whether a fixed window analysis is being performed (as opposed to a variable window analysis with a temporal scan).
Note 2: must be set to missing for Type 2 analyses.
Note 3: TREEANALYSISGRP can be used for multiple sets of risk and control windows but each unique analysis requires its own TREEANALYSISID.
Note 4: control windows may precede or follow risk windows and the two windows are not required to be contiguous.
Note 5: this should be specified as a “timeline” label as opposed to an algebraic quantity, assuming Day 0 = 0.
Format: Numeric
Example: 28
Control window start CWSTART Used to define the start of a user-defined fixed control window when performing a Bernoulli TreeScan analysis with a self-controlled risk interval design (as opposed to a Tree-Temporal).
Note 1: only relevant for Type 3 analyses and is optional based on whether a fixed window analysis is being performed (as opposed to a variable window analysis with a temporal scan).
Note 2: must be set to missing for Type 2 analyses.
Note 3: TREEANALYSISGRP can be used for multiple sets of risk and control windows but each unique analysis requires its own TREEANALYSISID.
Note 4: control windows may precede or follow risk windows and the two windows are not required to be contiguous.
Note 5: this should be specified as a “timeline” label as opposed to an algebraic quantity, assuming Day 0 = 0.
Format: Numeric
Example: -56
Control window end CWEND Used to define the end of a user-defined fixed control window when performing a Bernoulli TreeScan analysis with a self-controlled risk interval design (as opposed to a Tree-Temporal).
Note 1: only relevant for Type 3 analyses and is optional based on whether a fixed window analysis is being performed (as opposed to a variable window analysis with a temporal scan).
Note 2: must be set to missing for Type 2 analyses.
Note 3: TREEANALYSISGRP can be used for multiple sets of risk and control windows but each unique analysis requires its own TREEANALYSISID.
Note 4: control windows may precede or follow risk windows and the two windows are not required to be contiguous.
Note 5: this should be specified as a “timeline” label as opposed to an algebraic quantity, assuming Day 0 = 0.
Format: Numeric
Example: -29