PEDSnet Common Data Model, v5.9

Submission Date

Data Collection

Data Source

Tags


Abstract

This document provides the ETL processing assumptions and conventions developed by the PEDSnet data partners that a data partner should use to ensure common ETL business rules.

Provenance

Description

The PEDSnet Common Data Model is an evolving specification, based in structure on the OMOP Common Data Model, but expanded to accommodate requirements of both the PCORnet Common Data Model and the primary research cohorts established in PEDSnet. This version of the PEDSnet CDM reflects the ETL processes developed after several iterations of network development. As such, it proposes to align with the current version of the PCORnet CDM.

This document provides the ETL processing assumptions and conventions developed by the PEDSnet data partners that a data partner should use to ensure common ETL business rules. This document will be modified as new situations are identified, incorrect business rules are identified and replaced, as new analytic use cases impose new/different ETL rules, and as the PEDSnet CDM continues to evolve.

**** NEW in PEDSnet CDM v5.9 ****

(1) Additional ETL Guidance for 1.09 Observation:

In PEDSnet version 5.9, we are expanding the list of Social Determinants of Health (SDOH) and Patient Reported Outcome (PRO) surveys to include in the observation table. Please see the linked csv files below for the full lists of survey questions:

Sites are not expected to have representation for all surveys and questions in their source EHR system.

In order to ease the burden of identifying and extracting this additional survey information (names, questions, and answers), sites are encouraged to only extract and submit the raw, unmapped survey information. When received, the PEDSnet DCC will map the survey information to standard pedsnet concepts where applicable. The PEDSnet DCC will attempt to map any observation record that meets the following conditions:

  1. observation_type_concept_id = 32862 (Patient filled survey)
  2. observation_concept_id = 0 (Unmapped)

Please refer below on how to model raw survey data such that the DCC can map to standard concepts:

Field Site Responsibility PEDSnet DCC Responsibility
observation_id Populate with typical primary key value.
person_id Populate with the person_id who answered the survey question.
visit_occurrence_id Populate with the visit_occurrence_id that the survey was taken during (if available).
provider_id Populate with the provider_id who administered the survey (if available).
observation_date / observation_datetime Set Date fields equal to the date/datetime that the survey question was answered or completed.
observation_concept_id Set observation_concept_id = 0 DCC will use observation_source_value to map the raw question name to a standard LOINC concept (if applicable).
observation_source_concept_id Leave NULL DCC will use observation_source_value to map the raw survey name to a standard LOINC concept (if available and applicable).
observation_source_value Please include the question text and the name of the survey that the question belongs to (if available), separated by a pipe delimiter when concatenating. For Example: Question Text | Survey Name
observation_type_concept_id Set observation_type_concept_id = 32862 (Patient filled survey)
value_source_value Raw answer to the survey question.
  • If a numerical response is expected (such as number of days, raw score, t-score, etc.), then populate with the numerical value cast as a string.
  • If a text response is expected, then populate with the corresponding text value.
  • If a text response is expected, but only an integer answer exists, there may be a ZC category list table that can be joined to in order to obtain the integer’s corresponding text value. In such cases, concatenate the text response with the integer response cast to a string and separate with a pipe delimeter. For Example: Text Answer | Integer Answer
value_as_concept_id Leave NULL DCC will use the values from value_source_value to map the raw answer text to a standard LOINC concept (if applicable).
value_as_string Leave NULL DCC will extract text from value_source_value to populate.
value_as_number Leave NULL DCC will extract any numerical values from value_source_value to populate.

Additional Notes on Surveys:

  1. Prioritize including surveys for Depression, Anxiety, Food and Housing Stability, and Alcohol Use.

  2. Sites are encouraged not to modify any code for surveys that have already been mapped to standard concepts for previous PEDSnet versions (PHQ-2, PHQ-9, Hunger Vital Signs, Food Insecurity).

  3. For additional help on where and how to extract these surveys, the SDOH Surveys and PRO PROMIS Surveys files also include columns indicating where pilot sites found these surveys in their source system.

(2) Additional Guidance for 1.11 Drug Exposure:

In PEDSnet version 5.9, we are adding two new drug_type_concept_id values to model additional external drug records sourced from the Data Exchange Repository (DXR).

  • For drug records from a Health Information Exchange (HIE):

    • Set drug_type_concept_id = 32849 (Health Information Exchange Record)
    • Examples include unverified Epic Care Everywhere records or records sourced from Regional/State HIEs.
  • For non-dispensed drug records from an external claims source:

    • Set drug_type_concept_id = 32810 (Claim)
    • If the claim indicates that the drug has been dispensed at a pharmacy, please use drug_type_concept_id = 38000175 (Prescription Dispensed in Pharmacy) instead.

(3) Guidance for Modeling Mother-Child Relationships:

If maternal and delivery information exists in your source EHR system, then:

  1. Ensure that both the Mother and Child have a record in the person table​ (in case the mother is not currently included in your PEDSnet Cohort).

  2. Create a record in the fact_relationship to model the relationship​ as follows:

Field Value
domain_concept_id_1 56 (person)​
fact_id_1 person_id for the Child
domain_concept_id_2 56 (person)​
fact_id_2 person_id for the Mother
Relationship_concept_id 581437 (Child to Parent Measurement)​

(4) Finalized Guidance for Modeling MAR Actions for Deriving Continuous Infusion Volumes

Based on lessons learned from the Continuous Infusion Volume Derivation Pilot taking place in PEDSnet version 5.7 and 5.8, we are now finalizing guidance for extracting MAR actions and loading them into the drug_exposure table. The PEDSnet DCC will then use the individual fluid administration MAR actions to derive the total volume administered when an infusion rate and rate unit are present.

Data Extraction

Please extract all MAR Actions (including stops and pauses) for intravenous fluid administrations, particularly for continuous infusions and boluses.

Please be sure to extract the following data for each MAR action:

  1. Infusion rate (clarity source MAR_ADMIN_INFO.INFUSION_RATE)
  2. Infusion rate units (clarity source MAR_ADMIN_INFO.MAR_INF_RATE_UNIT_C or zc_med_unit.name)
  3. Local identifier for the drug order (clarity source MAR_ADMIN_INFO.ORDER_MED_ID)
  4. MAR Action Name (clarity source MAR_ACTION_C_NAME)
  5. Datetime of MAR action event (clarity source MAR_ADMIN_INFO.TAKEN_TIME)

Transformation & Loading

Starting in PEDSnet version 5.9, we have removed the drug_iv_pilot table from the DDL. Fluid MAR Actions should now be loaded into the drug_exposure table such that there is one record for each MAR action.

  • To specify which drug_exposure records represent a MAR action of an intravenous fluid administration versus other inpatient administrations, we have created a new concept_id - 2000001594 “MAR Action of inpatient intravenous fluid administration”.

  • For each MAR Action record, please set drug_type_concept_id = 2000001594 so that the DCC can identify which MAR actions should be grouped to derive total adminstered fluid volumes.

The below table specifies where to insert each extracted MAR Action field:

Extracted Field Target Drug_Exposure Field(s) Notes
Infusion Rate effective_drug_dose, eff_drug_dose_source_value
Infusion Rate Units dose_unit_source_value, dose_unit_concept_id dose_unit_concept_id populated with corresponding concept_id for the infusion rate.
Local identifier for the drug order drug_source_value Include the text med_id= concatenated with the local identifier. Use a pipe delimeter to separate from other data in the source value.
MAR Action Name drug_source_value Use a pipe delimeter to separate from other data in the source value
Datetime of MAR action event drug_exposure_start_date, drug_exposure_start_datetime
  • drug_exposure fields other than the target fields listed in the table above should be populated in the typical way that inpatient medication administration records are populated.

  • The drug_source_value should be formatted as follows:

    • Local identifier for the drug order | MAR action name | Local Drug Code | Local Drug Name
    • Ex: med_id=200100074|Rate Change|372000|FUROSEMIDE IV INFUSION-FUROSEMIDE 10 MG/ML (UNDILUTED) INJECTION

Please refer to the Continuous IV Fluid Volume Guidance documentation for for further details.

Vocabulary

Related Publications

Creative Commons license

Except where otherwised noted, this item's license is described as a CC-BY Attribution 4.0 License.

Search Results

Version History

Now showing 1 - 8 of 8
VersionDateSummary
9*
2025-07-16 12:29:21
Version 5.9 of the PEDSnet CDM ETL Conventions
2025-07-16 12:22:41
Version 5.8 of the PEDSnet CDM ETL Conventions
2025-03-11 10:39:21
Version 5.7 of the PEDSnet CDM ETL Conventions
2024-11-15 10:52:12
Version 5.6 of the PEDSnet CDM ETL Conventions
2024-08-13 15:01:37
Version 5.5 of the PEDSnet CDM ETL Conventions
2024-08-13 14:58:18
Version 5.4 of the PEDSnet CDM ETL Conventions
2024-08-13 14:53:47
Version 5.3 of the PEDSnet CDM ETL Conventions
2024-08-01 11:39:29
Version 5.2 of the PEDSnet CDM ETL Conventions
* Selected version