Risk Effects Model Template

Todo

To begin documenting a risk effects in GBD 2019 for a Vivarium simulation, start by following these steps (after you have created a new git branch to work in):

  1. Make a subdirectory risk_name/ in the folder gbd2019/risk_effects/, where risk_name is replaced with the (potentially shortened) name of the risk you are modeling, such as lbwsg/ or iron_deficiency/. This subdirectory is where you will put all the files for your risk effects model documentation, including this document, image files, .csv files, etc.

  2. Copy this template into your subdirectory and rename it index.rst.

  3. Replace the internal hyperlink target .. _2019_risk_effect_template: at the top of this file with a unique reference label for your cause. The reference label should have the form .. _2019_risk_effect_{risk_name}:, where {risk_name} is replaced with a unique descriptive name or abbreviation for your cause, e.g. .. _2019_risk_effect_lbwsg:.

  4. Create the document title: Replace the above title

    ===========================
Risk Effects Model Template
===========================

    with a title that includes the appropriate risk name, i.e.

    ========================
{Risk Name} Risk Effects
========================

    where {Risk Name} is the full name of your risk in GBD 2019.

    Note: Be sure to adjust the length of the title’s underline ====== and overline ====== to match the length of your new document title, or you will get errors in the section structure when Sphinx builds HTML from the index.rst file.

  5. Once you complete these steps, delete this .. todo:: directive from index.rst.

Risk Overview

Todo

Provide a brief description of the risk, including potential opportunities for confounding (factors that may cause or be associated with the risk exposure), effect modification/generalizability, etc. by any relevant variables. Note that literature reviews and speaking with the GBD risk modeler will be good resources for this.

GBD 2019 Modeling Strategy

Note

This section will describe the Vivarium modeling strategy for risk effects. For a description of Vivarium modeling strategy for risk exposure, see the {RISK_EXPOSURE_PAGE_LINK} page.

Todo

Replace {RISK_EXPOSURE_PAGE_LINK} with a reference to the appropriate risk exposure page in the above note.

Todo

Provide a brief overview of how the risk affects different outcomes, including data sources used by GBD, GBD assumptions, etc. Note that the [GBD-2019-Risk-Factors-Appendix-Risk-Effects-Model-Template] is a good source for this information in addition to the GBD risk modeler.

Todo

Fill out the following table so that it reflects all entities affected by the risk in GBD 2019.

Affected Entities

Outcome

Outcome type

Outcome ID

Affected measure

Note

Vivarium Modeling Strategy

Note

This section will describe the Vivarium modeling strategy for risk effects. For a description of Vivarium modeling strategy for risk exposure, see the {RISK_EXPOSURE_PAGE_LINK} page.

Todo

Replace {RISK_EXPOSURE_PAGE_LINK} with a reference to the appropriate risk exposure page in the above note.

Todo

List the risk-outcome relationships that will be included in the risk effects model for this risk factor. Note whether the outcome in a risk-outcome relationship is a standard GBD risk-outcome relationship or is a custom relationship we are modeling for our simulation.

Risk Outcome Relationships for Vivarium

Outcome

Outcome type

Outcome ID

Affected measure

Note

Risk Outcome Pair #1

Todo

Replace “Risk Outcome Pair #1” with the name of an affected entity for which a modeling strategy will be detailed. For additional risk outcome pairs, copy this section as many times as necessary and update the titles accordingly.

Todo

Link to existing cause model document or other documentation of the outcome in the risk outcome pair.

Todo

Describe which entitity the relative risks apply to (incidence rate, prevalence, excess mortality rate, etc.) and how to apply them (e.g. affected_measure * (1 - PAF) * RR).

Be sure to specify the exact PAF that should be used in the above equation and either how to calculate it (see the Population Attributable Fraction section of the Modeling Risk Factors document) or pull it (vivarium_inputs.interface.get_measure(risk_factor.{risk_name}, 'population_attributable_fraction'), noting which affected entity and measure should be used)

Todo

Complete the following table to list the relative risks for each risk exposure category on the outcome. Note that if there are many exposure categories, another format may be preferable.

Relative risks for a risk factor may be pulled from GBD at the draw-level using vivarium_inputs.interface.get_measure(risk_factor.{risk_name}, 'relative_risk'). You can then calculate the mean value as well as 2.5th, and 97.5th percentiles across draws.

The relative risks in the table below should be included for easy reference and should match the relative risks pulled from GBD using the above code. In this case, update the Note below to include the appropriate {risk_name}.

If for any reason the modeling strategy uses non-GBD relative risks, update the Note below to explain that the relative risks in the table are a custom, non-GBD data source and include a sampling strategy.

Note

The following relative risks are displayed below for convenient reference. The relative risks in the table below should match the relative risks that can be pulled at the draw level using vivarium_inputs.interface.get_measure(risk_factor.{risk_name}, 'relative_risk').

Relative Risks

Exposure Category

Relative Risk

Note

Validation and Verification Criteria

Todo

List validation and verification criteria, including a list of variables that will need to be tracked and reported in the Vivarium simulation to ensure that the risk outcome relationship is modeled correctly

Assumptions and Limitations

Todo

List assumptions and limitations of this modeling strategy, including any potential issues regarding confounding, mediation, effect modification, and/or generalizability with the risk-outcome pair.

Bias in the Population Attributable Fraction

As noted in the Population Attributable Fraction section of the Modeling Risk Factors document, using a relative risk adjusted for confounding to compute a population attributable fraction at the population level will introduce bias.

Todo

Outline the potential direction and magnitude of the potential PAF bias in GBD based on what is understood about the relationship of confounding between the risk and outcome pair using the framework discussed in the Population Attributable Fraction section of the Modeling Risk Factors document.

References

Todo

Update the GBD 2019 Risk Factor Methods appendix citation to be unique to your risk effects page (replace ‘Risk-Effects-Model-Template’ with ‘{Risk Name}-Effects’)

Update the appropriate page numbers in the GBD risk factors methods appendix below

Add additional references as necessary

GBD-2019-Risk-Factors-Appendix-Risk-Effects-Model-Template

Pages ???-??? in Supplementary appendix 1 to the GBD 2019 Risk Factors Capstone:

(GBD 2019 Risk Factors Capstone) GBD 2019 Risk Factors Collaborators. Global burden of 87 risk factors in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet 2020; 396: 1223–49. DOI: https://doi.org/10.1016/S0140-6736(20)30752-2