Standardized Data File - User Guide

Version 2021

Updated 12/22/2022

Table of Contents

1. Introduction

2. Phenotypes: Quality and Updates

3. Case Inclusion Criteria

4. Table Descriptions

5. Included Variables

2. Quality Measure Specifications (separate document)

1. Introduction

The Multicenter Perioperative Outcomes Group (MPOG) is a consortium of hospitals across

the United States, which seeks to improve experiences of patients receiving care by

anesthesia clinicians. Since its inception in 2008, MPOG has developed policies,

procedures, and the technical infrastructure required to conduct large scale research, create

quality improvement initiatives, educate caregivers, and guide healthcare administration.

The goal of MPOG research is to systematically transform real-world perioperative health

File to help MPOG researchers conduct studies as efficiently as possible.

The standardized data file is a pre-packaged MPOG dataset of cases within a specified time

frame that is ready to distribute quickly after an approved PCRC proposal. The data file

contains commonly requested, validated case-level phenotypes (i.e. summary variables

computed or derived from raw data) for each included case, as well as raw administrative

data associated with the patient. A new standardized data file will be released annually with

additional data from new MPOG centers, more recent cases, and additional validated

phenotypes.

This user guide provides information on the included study population and phenotypes, the

structure of the data, and the process for accessing the data.

2. Phenotypes: Quality and Updates

Phenotypes help transform messy, real-world electronic health record data into structured,

clinically useful inferences about the case and the course of clinical care. Each phenotype

may be a computation, recode, or other combination of variables in the raw dataset that

summarize information through one variable. Examples include the BMI phenotype

(computed from weight and height) and the Last Known Alive (which is derived from the

MPOG, these phenotypes are subjected to rigorous development and validation processes

data file remains fixed for each version release, if inconsistencies are found in the underlying

update notes outlining the change(s) under the FAQ section of this document.

Additional information on MPOG phenotypes can be found by accessing the MPOG

research website → Tips & Tricks → “Transforming Raw Data into Clinical Inferences:

Phenotypes”

MPOG’s full collection of phenotypes can be found by exploring the phenotype browser,

https://phenotypes.mpog.org/.

3. Case Inclusion Criteria

participating MPOG medical centers throughout the United States which meet the

2021. Please note that participating sites must be contributing cases to MPOG at the time of

the data pull in order for their data to be included in that year's standardized data file.

data file release. Version 2021 contains four separate tables that can be merged and

analyzed together if desired as described in Section 4: Table Descriptions.

defined as the intraoperative research standard, to ensure it is a case suitable for research.

The intraoperative research standard was developed by members of the research and

quality teams at the MPOG coordinating center and is continually re-evaluated to ensure

reliability.

To meet the intraoperative research standard, a case must satisfy all of the following

criteria:

● There must be a date and time noted for the beginning and end of the anesthetic

procedure. If multiple start times exist, the earliest is used, and the start time must be

not more than 30 minutes prior to the listed anesthesia start time. If a patient in room

be greater than or equal to 5 minutes.

● There must be an age listed for the patient in the data.

● There must be data regarding the patient’s sex (either male or female). This

phenotype does not reflect the patient’s gender identity.

● The data must include an American Society of Anesthesiology Physical Status

classification score (ASA), between 1 and 6, which gives information on the patient’s

overall health and potential risks in anesthetic management. A case is excluded if

there are multiple ASA statuses, the ASA status is missing, or an unused

classification number that is present.

● There must be at least one blood pressure value recorded in the case data, which

4. Table Descriptions

linkable across one another.

5. Included Variables

Below is the list of variables included in the standardized data file version 2021. Full

definitions and specifications for phenotypes and measures are available in Appendix 1 and

Appendix 2.

9. Frequently Asked Questions

Q: How does the CPT prediction tool work?

the first two codes.

● If the predicted codes are for OB cases and the patient’s age makes that prediction

unlikely (age less than 10 or greater than 55), those codes are not returned.

data table?

days before to 365 days after the date of service along with case date and the admit and

discharge dates associated with those codes. Therefore, each researcher will need to decide

per project, what relevant timeframe to look for associated ICD 9/10 codes. For example, if a

researcher is trying to define a comorbidity using ICD 9/10 codes, then they may look for

relevant codes 90 days prior to 7 days after the date of surgery (similar to how the Elixhauser

comorbidity phenotypes are defined). In contrast, if a researcher is attempting to define an

outcome based on ICD 9/10 codes, then they may only look for relevant codes occurring on or

after the date of surgery.

Q: Can phenotype definitions change over time?

taken a “snapshot” of the data and included the specification for how each phenotype was

have changed. While the set of phenotypes included in the standardized data file remains fixed

Q: How should a researcher utilize the physiologic parameters that have both a set and

actual value (for example: median positive end-expiratory pressure (PEEP) set and

actual, and median tidal volume set and actual)?

covariates in their analysis. In some instances, at the discretion of the research team, it may be

useful to combine the actual and set parameters to have more complete data.