AI-assisted refactor of opioid use disorder documentation into a cause model by aflaxman · Pull Request #1852 · ihmeuw/vivarium_research

aflaxman · 2025-11-18T18:08:42Z

Move cause model to its own page, update based on GBD 2023 docs.

This commit refactors the Opioid Use Disorder (OUD) documentation to align with GBD 2021 concepts and best practices for modeling chemical dependency: - Created new OUD cause model at docs/source/models/causes/opioid_use_disorder/ - Comprehensive documentation includes: * Three-state model: susceptible, with_condition, and on_treatment * Detailed state definitions and transitions aligned with GBD 2021 * Clinical course and natural history of OUD * GBD 2021 modeling strategy using DisMod-MR 2.1 * NumPyro/DisMod-AT methodology for parameter estimation * Data sources, validation criteria, and limitations * Extensions and risk factor interactions - Updated MOUD concept model to reference the new OUD cause model - Model aligned with: * GBD 2021 definitions and diagnostic criteria (DSM-5, ICD-10) * Literature on compartmental modeling of substance use disorders * RESPOND model and other contemporary OUD modeling approaches The refactored model provides a comprehensive foundation for simulating MOUD interventions and understanding the opioid epidemic dynamics.

… extensions This commit updates the Opioid Use Disorder cause model documentation to reflect GBD 2023 concepts and adds comprehensive extensions for modeling polysubstance use and casual drug use: Major Updates: - Updated all references from GBD 2021 to GBD 2023 - Updated diagnostic criteria from DSM-IV-TR to DSM-5 - Added note on evolution of diagnostic criteria (DSM-IV vs DSM-5) - Updated reference label from 2021_cause_opioid_use_disorder to 2023_cause_opioid_use_disorder - Updated temporal coverage from 1990-2021 to 1990-2023 - Updated all GBD data source references (COMO, CoDCorrect, Demography, DisMod-MR 2.1) New Extensions Added: 1. Polysubstance Use Modeling: - Joint state space for co-occurring opioid and other substance use disorders - Risk stratification by polysubstance patterns (opioid+stimulant, opioid+sedative) - Modified transition rates reflecting polysubstance effects on treatment outcomes - Epidemiological context (opioid-methamphetamine co-use increased 10.1% from 1992-2017) - Modeling framework options (compartmental, Markov, system dynamics, agent-based) 2. Casual Use and Subclinical States: - Four-state model extension (susceptible, casual use, OUD untreated, OUD on treatment) - Transition pathways from casual use to disorder - Frequency-dependent modeling using sigmoid patterns and Hill functions - Clinical significance and data requirements - Methodological considerations for DSM-5 severity gradations New References Added: - Jalal et al. 2022 (systematic review of opioid simulation models) - Ciccarone 2019 (triple wave epidemic) - Jones et al. 2020 (methamphetamine resurgence) - Ellis et al. 2018 (opioid-methamphetamine twin epidemics) - Anthony et al. 1994, Lopez-Quintero et al. 2011 (transition to dependence) Key Research Findings Incorporated: - GBD 2023 uses DisMod-MR 2.1 (consistent with GBD 2021) - DSM-5 criteria combine abuse/dependence into unified disorder with severity levels - Polysubstance overdose deaths (opioids+stimulants) increased from 187 (1999) to 14,777 (2020) - Only 15-68% of substance users develop dependence (substance-specific) - Compartmental models most common (36%) for opioid epidemic modeling The documentation now provides comprehensive guidance for extending the base three-state OUD model to capture heterogeneity in substance use patterns, treatment responses, and progression to disorder.

CORRECTION: Previous commit incorrectly stated GBD 2023 updated to DSM-5 criteria. This commit corrects the documentation to reflect the actual GBD 2023 methodology from the official supplementary appendix. Major Corrections: 1. Diagnostic Criteria: - GBD 2023 continues to use DSM-IV-TR (NOT DSM-5) - Requires ≥3 of 7 symptoms for opioid dependence - Full DSM-IV-TR criteria now documented with ICD codes - Added note that clinical practice has evolved to DSM-5, but GBD maintains DSM-IV-TR 2. Severity Distribution (from GBD 2023 appendix): - Asymptomatic: 16% (13-19%), DW = 0 - Mild: 37% (20-55%), DW = 0.335 (0.221-0.473) - Moderate/Severe: 47% (29-64%), DW = 0.697 (0.510-0.843) - Based on NESARC and Comorbidity and Trauma Study - Previous version incorrectly showed only 2 severity levels 3. DisMod Methodology Details: - Added prior settings: no incidence/EMR before age 15, no incidence after 64 - Remission upper limit of 0.2 - Country-level covariates: IDU prevalence, SDDD of prescribed opioids - EMR covariate: IDU with restricted influence (0-2 range) 4. Data Sources and Adjustment: - Documented IHME-indirect data creation methodology - MR-BRT crosswalk adjustment factor: 0.25 for direct surveys - Direct surveys adjusted upward (logit beta = -1.07, gamma = 0.24) - Explained multiplier method using treatment data and ST-GPR 5. EMR Methodology: - MR-BRT method with HAQ Index prior (though not supported by data) - IDU covariate for EMR: exponentiated beta = 6.84 (6.12-7.36) - EMR consistent across HAQ levels due to data-driven results 6. Casual Use Extension: - Updated to reference both DSM-IV-TR (≥3 symptoms) and DSM-5 (≥2 symptoms) - Clarified that GBD uses DSM-IV-TR but extensions may use DSM-5 - Added guidance on specifying diagnostic criteria in extensions Updated References: - Abbreviations: DSM-5 → DSM-IV-TR - Disability weights in data tables - Sequelae descriptions (asymptomatic, mild, moderate/severe) Sources: - GBD 2023 Supplementary Appendix: Opioid Use Disorders - GBD 2023 Supplementary Appendix: Drug Use (risk factor) - Direct quotes from official IHME methodology documentation This ensures documentation accurately reflects GBD 2023 methodology for consistency with official estimates and proper model parameterization.

- Corrected all heading level decorators to match document template - Changed top-level sections from `---` to `+++` (Section Level 1) - Changed subsections from `+++` to `---` (Section Level 2) - Fixed all CRITICAL title level inconsistency errors - Documentation now builds with ZERO warnings and ZERO errors - Verified with: make html -C docs/ SPHINXOPTS="-W --keep-going -n"

…9FaK4GYMbtsaogFUQrU8Nx

- Added 14 citation references throughout the document - GBD-2023-Capstone-Opioid: cited when defining GBD 2023 OUD - DisMod-Methods: cited when discussing DisMod-MR methodology - Degenhardt-2019: cited for OUD recovery pathways - SAMHSA-MOUD: cited for MOUD medications - Wakeman-2020: cited for treatment effectiveness - Sordo-2017, Santo-2021: cited for mortality reduction from MOUD - Friedman-2022: cited in disease overview - Ciccarone-2019: cited for opioid epidemic evolution - Jalal-2020: cited for simulation modeling frameworks - Jones-2020, Ellis-2018: cited for polysubstance use trends - Anthony-2005: cited for dependence prevalence among users - Lopez-Quintero-2011: cited for transition probabilities Resolves Sphinx warnings about unreferenced citations

References fixed: - Replace [Friedman-2022] (hallucinated reference) with [DSM-5] for OUD definition - Replace [Jalal-2020] with [Cerda-2022] (correct authors: Cerdá et al.; correct journal: Epidemiol Rev) - Upgrade [GBD-2023-Capstone-Opioid] to cite 2025 Lancet GBD 2023 capstone paper - Fix [SAMHSA-MOUD] publication number to official HHS PEP21-02-01-002 - Rename [Anthony-2005] to [Anthony-1994] to match actual publication year ICD-10 codes corrected: - Remove incorrect T14.0 (superficial injury) and Z79.8 (other drug therapy) - Use F11.x for opioid use disorders - Clarify T40.x codes are for overdose models, not chronic OUD cause Modeling assumptions clarified: - MR-BRT crosswalk: removed unpublished specific parameter values, made qualitative - EMR covariates: removed unpublished beta coefficients, kept qualitative description - Age 64 incidence cutoff: clarified as modeling assumption, not hard EMCDDA rule - Treatment EMR/disability: clarified these are modeling choices with sensitivity analyses, not clinical facts that MOUD eliminates all disability/mortality All changes ensure documentation reflects verifiable sources and distinguishes modeling choices from empirical findings.

Added DOI links and URLs to all bibliography entries: - GBD-2023-Overview: Added DOI link to 2025 Lancet paper - DisMod-Methods: Added DOI link to 2018 Lancet GBD 2017 paper - Degenhardt-2019: Added DOI link - SAMHSA-MOUD: Added SAMHSA library link and NCBI Bookshelf full text link - Wakeman-2020: Added DOI link - Sordo-2017: Added DOI link - Santo-2021: Added DOI link - DSM-5: Added overview link to APA website - Cerda-2022: Added DOI link - Ciccarone-2019: Added DOI link - Jones-2020: Added DOI link - Ellis-2018: Added DOI link - Anthony-1994: Added DOI link - Lopez-Quintero-2011: Added DOI and PubMed link All references now include hyperlinks for easy access to source documents.

- Converted all reference URLs to use RST anonymous link syntax - Changed from single underscore `text <URL>_` to double underscore `text <URL>__` - This prevents "duplicate explicit target name" warnings for repeated link text - All "Available at" links now render as clickable hyperlinks - Also made NCBI Bookshelf link clickable for SAMHSA-MOUD reference - Build now passes with only 1 warning (graphviz - environment specific) All 14 references now have properly clickable hyperlinks to their source documents.

Changed all link text from generic "Available at" to descriptive labels: - Journal articles: "(full text)" for DOI links - DSM-5: "(overview)" for the APA website - SAMHSA-MOUD: "(publication page)" and "(full text)" for two different links - Lopez-Quintero: "(PubMed)" for PubMed link Now readers can see at a glance what type of resource each link provides. All link text is clickable and uses RST anonymous link syntax.

- Change treatment discontinuation symbol from 'tf' to 'td' throughout document - Fix LaTeX underscore escaping in math formulas (treatment_coverage, emr_multiplier, disability_weight) - Update all references to use 'td' in diagrams, tables, and text

…ng approaches section - Change 'treatment discontinuation/failure' to 'treatment discontinuation' throughout - Move modeling framework discussion to end of Disease Overview section - Add expanded framing about simulation modeling approaches based on Cerda et al systematic review - Contextualize compartmental model approach within broader landscape of OUD modeling frameworks

Copilot

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

aflaxman

concept_model.rst should not have some of the data notes anymore.

docs/source/models/causes/opioid_use_disorder/index.rst

…MbtsaogFUQrU8Nx

- Remove section 4.1 Core Disease Model Parameters (now covered in cause model documentation) - Add reference to OUD cause model for parameter details - Update transition symbol from 'tf' to 'td' throughout - Update 'treatment failure' to 'treatment discontinuation' language - Renumber remaining sections in Data Notes

Copilot

Pull Request Overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

docs/source/models/causes/opioid_use_disorder/index.rst

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

docs/source/models/causes/opioid_use_disorder/index.rst

SylLutze · 2025-11-21T16:50:44Z

docs/source/models/causes/opioid_use_disorder/index.rst

+
+5. **Age Patterns**: Age-specific prevalence and incidence should follow observed patterns from GBD and epidemiological studies
+
+6. **Temporal Trends**: Trends in prevalence, incidence, and mortality should match observed temporal patterns


How are you doing this? Usually we take 2023 GBD data and hold steady

I'm going to use my DisMod-AT clone to extrapolate for now; time trends are really important for this.

How does it extrapolate? That wasn't described in the section about it above.

docs/source/models/causes/opioid_use_disorder/index.rst

aflaxman · 2025-11-21T19:39:18Z

docs/source/models/causes/opioid_use_disorder/index.rst

+
+Model validation should compare simulated outputs to reference data:
+
+1. **Prevalence**: Total OUD prevalence in the simulation (C + T states) should match GBD 2023 prevalence estimates within uncertainty bounds


Suggested change

1. **Prevalence**: Total OUD prevalence in the simulation (C + T states) should match GBD 2023 prevalence estimates within uncertainty bounds

1. **Prevalence**: Total OUD prevalence in the simulation (C + T states) should match GBD 2023 prevalence estimates within uncertainty bounds; also check YLDs, which will be more of a validation than verification because we have assumed severity in T is all mild.

Are we assigning nobody to the asymptomatic sequela?

aflaxman · 2025-11-21T19:41:00Z

docs/source/models/causes/opioid_use_disorder/index.rst

+
+2. **Treatment Coverage**: The proportion of individuals with OUD in the treatment state (T / (C + T)) should match observed treatment coverage ratios
+
+3. **Cause-Specific Mortality**: Deaths attributed to OUD in the simulation should match GBD 2023 CSMR estimates


Suggested change

3. **Cause-Specific Mortality**: Deaths attributed to OUD in the simulation should match GBD 2023 CSMR estimates

3. **Cause-Specific Mortality**: Deaths attributed to OUD in the simulation should match GBD 2023 CSMR estimates; also check EMR and YLLs

EMR isn't really estimated by GBD so I don't understand that piece; I guess it's just a combo check for CSMR and prevalence? Or are you planning to use DisMod EMR? This was unclear above.

aflaxman · 2025-11-21T19:41:50Z

docs/source/models/causes/opioid_use_disorder/index.rst

+
+5. **Age Patterns**: Age-specific prevalence and incidence should follow observed patterns from GBD and epidemiological studies
+
+6. **Temporal Trends**: Trends in prevalence, incidence, and mortality should match observed temporal patterns


I'm going to use my DisMod-AT clone to extrapolate for now; time trends are really important for this.

aflaxman · 2025-11-24T03:52:17Z

docs/source/models/causes/opioid_use_disorder/index.rst

+
+4. **Incidence**: Population incidence rate (transitions from S to C) should be consistent with GBD 2023 incidence estimates
+
+5. **Age Patterns**: Age-specific prevalence and incidence should follow observed patterns from GBD and epidemiological studies


Suggested change

5. **Age Patterns**: Age-specific prevalence and incidence should follow observed patterns from GBD and epidemiological studies

5. Remission, Treatment initiation, Treatment Discontinuation, Treatment-associated Recovery

I don't understand this suggestion

zmbc

Interesting stuff!

Sorry I made so many comments, happy to discuss live if that would be helpful.

docs/source/models/causes/opioid_use_disorder/index.rst

zmbc · 2025-11-25T18:46:10Z

docs/source/models/causes/opioid_use_disorder/index.rst

+
+4. **Incidence**: Population incidence rate (transitions from S to C) should be consistent with GBD 2023 incidence estimates
+
+5. **Age Patterns**: Age-specific prevalence and incidence should follow observed patterns from GBD and epidemiological studies


I don't understand this suggestion

zmbc · 2025-11-25T18:46:39Z

docs/source/models/causes/opioid_use_disorder/index.rst

+
+5. **Age Patterns**: Age-specific prevalence and incidence should follow observed patterns from GBD and epidemiological studies
+
+6. **Temporal Trends**: Trends in prevalence, incidence, and mortality should match observed temporal patterns


How does it extrapolate? That wasn't described in the section about it above.

zmbc · 2025-11-25T18:47:37Z

docs/source/models/causes/opioid_use_disorder/index.rst

+
+6. **Temporal Trends**: Trends in prevalence, incidence, and mortality should match observed temporal patterns
+
+Extensions and Modifications


This section is a bit unusual for us, we usually don't include notes for the future like this; might want to label them as such a bit more clearly.

zmbc · 2025-11-25T18:48:53Z

docs/source/models/causes/opioid_use_disorder/index.rst

+
+1. **Three-State Model**: The model simplifies the complex heterogeneity of OUD into three discrete states. In reality, OUD severity exists on a continuum, and individuals may have varying levels of use, dependence, and treatment engagement.
+
+2. **MOUD as Single Treatment State**: The "on_treatment" state aggregates all forms of MOUD (methadone, buprenorphine, naltrexone) despite important differences in effectiveness, retention, and accessibility across these medications.


Are we also assuming that non-medication treatments have no effect at all? That seems like a big one!

I guess that they are wrapped up in the transition from OUD to Recovery "without treatment"; I will change that language to "without medication"

Interesting, we also miss heterogeneity within the "without medication" category (because some people in that category are receiving non-medication treatments) and also assume that non-medication treatment and medication treatment are totally independent which seems unlikely.

zmbc · 2025-11-25T18:50:02Z

docs/source/models/concept_models/vivarium_moud/concept_model.rst

+- **Three states**: susceptible, with_condition (untreated OUD), and on_treatment (receiving MOUD)
+- **Five transitions**: incidence (i), natural remission (r), treatment initiation (ti), treatment discontinuation (td), and treatment-associated recovery (ts)
+- **GBD 2023 alignment**: Parameterization consistent with Global Burden of Disease 2023 estimates for OUD prevalence, incidence, remission, and excess mortality
+- **Treatment representation**: Explicit modeling of MOUD engagement and outcomes
+- **DisMod-AT methodology**: Transition rates estimated using a NumPyro implementation of DisMod-AT-like Bayesian inference to ensure internal consistency


I think if you include this summary here, it should be shorter

…MbtsaogFUQrU8Nx

docs/source/models/causes/opioid_use_disorder/index.rst

Co-authored-by: Zeb Burke-Conte <zmbc@users.noreply.github.com>

claude added 3 commits November 18, 2025 04:43

aflaxman marked this pull request as draft November 18, 2025 18:08

aflaxman requested a review from Copilot November 18, 2025 18:08

Copilot started reviewing on behalf of aflaxman November 18, 2025 18:09 View session

Copilot finished reviewing on behalf of aflaxman November 18, 2025 18:10

This comment was marked as outdated.

Sign in to view

claude and others added 9 commits November 18, 2025 18:31

Merge branch 'ihmeuw:main' into claude/review-opioid-documentation-01…

e7755eb

…9FaK4GYMbtsaogFUQrU8Nx

aflaxman requested a review from Copilot November 18, 2025 22:27

Copilot started reviewing on behalf of aflaxman November 18, 2025 22:27 View session

Copilot finished reviewing on behalf of aflaxman November 18, 2025 22:31

Copilot AI reviewed Nov 18, 2025

View reviewed changes

aflaxman commented Nov 19, 2025

View reviewed changes

aflaxman added 8 commits November 18, 2025 21:37

Update docs/source/models/causes/opioid_use_disorder/index.rst

d3d8eb6

Update docs/source/models/causes/opioid_use_disorder/index.rst

8c51dc1

Update docs/source/models/causes/opioid_use_disorder/index.rst

5485729

Update docs/source/models/causes/opioid_use_disorder/index.rst

ea78f21

Update docs/source/models/causes/opioid_use_disorder/index.rst

e406fd6

Update docs/source/models/causes/opioid_use_disorder/index.rst

108124d

Update docs/source/models/causes/opioid_use_disorder/index.rst

38b61e9

Merge branch 'main' into claude/review-opioid-documentation-019FaK4GY…

1427bf7

…MbtsaogFUQrU8Nx

aflaxman requested a review from Copilot November 19, 2025 05:57

aflaxman marked this pull request as ready for review November 19, 2025 05:57

aflaxman self-assigned this Nov 19, 2025

Copilot started reviewing on behalf of aflaxman November 19, 2025 05:58 View session

aflaxman requested review from NathanielBlairStahn, SylLutze, alibow, tylerdy and zmbc November 19, 2025 05:58

Copilot finished reviewing on behalf of aflaxman November 19, 2025 06:00

Copilot AI reviewed Nov 19, 2025

View reviewed changes

docs/source/models/causes/opioid_use_disorder/index.rst Outdated Show resolved Hide resolved

docs/source/models/causes/opioid_use_disorder/index.rst Show resolved Hide resolved

Update docs/source/models/causes/opioid_use_disorder/index.rst

3ee7202

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

SylLutze approved these changes Nov 21, 2025

View reviewed changes

aflaxman commented Nov 21, 2025

View reviewed changes

aflaxman commented Nov 24, 2025

View reviewed changes

zmbc reviewed Nov 25, 2025

View reviewed changes

Merge branch 'main' into claude/review-opioid-documentation-019FaK4GY…

9501eab

…MbtsaogFUQrU8Nx

aflaxman commented Dec 30, 2025

View reviewed changes

docs/source/models/causes/opioid_use_disorder/index.rst Outdated Show resolved Hide resolved

aflaxman commented Dec 30, 2025

View reviewed changes

docs/source/models/causes/opioid_use_disorder/index.rst Show resolved Hide resolved

aflaxman commented Dec 30, 2025

View reviewed changes

docs/source/models/causes/opioid_use_disorder/index.rst Outdated Show resolved Hide resolved

aflaxman commented Dec 30, 2025

View reviewed changes

docs/source/models/causes/opioid_use_disorder/index.rst Outdated Show resolved Hide resolved

aflaxman and others added 6 commits December 30, 2025 15:31

Update docs/source/models/causes/opioid_use_disorder/index.rst

1e7fdf3

Co-authored-by: Zeb Burke-Conte <zmbc@users.noreply.github.com>

Update docs/source/models/causes/opioid_use_disorder/index.rst

ae5ea57

Co-authored-by: Zeb Burke-Conte <zmbc@users.noreply.github.com>

Update docs/source/models/causes/opioid_use_disorder/index.rst

f727461

Update docs/source/models/causes/opioid_use_disorder/index.rst

c638429

Update docs/source/models/causes/opioid_use_disorder/index.rst

2cd66f8

Apply suggestions from code review

fe4b0f2

aflaxman mentioned this pull request Jan 6, 2026

AI-assisted refactor of opioid use disorder documentation into a cause model, take 2 #1881

Open

aflaxman closed this Jan 7, 2026


		5. Age Patterns: Age-specific prevalence and incidence should follow observed patterns from GBD and epidemiological studies

		6. Temporal Trends: Trends in prevalence, incidence, and mortality should match observed temporal patterns


		Model validation should compare simulated outputs to reference data:

		1. Prevalence: Total OUD prevalence in the simulation (C + T states) should match GBD 2023 prevalence estimates within uncertainty bounds


		2. Treatment Coverage: The proportion of individuals with OUD in the treatment state (T / (C + T)) should match observed treatment coverage ratios

		3. Cause-Specific Mortality: Deaths attributed to OUD in the simulation should match GBD 2023 CSMR estimates


		4. Incidence: Population incidence rate (transitions from S to C) should be consistent with GBD 2023 incidence estimates

		5. Age Patterns: Age-specific prevalence and incidence should follow observed patterns from GBD and epidemiological studies

	5. Age Patterns: Age-specific prevalence and incidence should follow observed patterns from GBD and epidemiological studies
	5. Remission, Treatment initiation, Treatment Discontinuation, Treatment-associated Recovery


		6. Temporal Trends: Trends in prevalence, incidence, and mortality should match observed temporal patterns

		Extensions and Modifications


		1. Three-State Model: The model simplifies the complex heterogeneity of OUD into three discrete states. In reality, OUD severity exists on a continuum, and individuals may have varying levels of use, dependence, and treatment engagement.

		2. MOUD as Single Treatment State: The "on_treatment" state aggregates all forms of MOUD (methadone, buprenorphine, naltrexone) despite important differences in effectiveness, retention, and accessibility across these medications.

Conversation

aflaxman commented Nov 18, 2025

Uh oh!

This comment was marked as outdated.

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

aflaxman left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zmbc left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!