OHDSI Home | Forums | Wiki | Github

[OHDSI Cov19] Community Update 24March2020: Logistics for OHDSI COVID19 Study-a-thon

(Patrick Ryan) #1


T minus 2 days until the kick-off of the OHDSI COVID-19 study-a-thon. This update provides you details of the logistics for those who registered about how you can participate in the activities. Thanks to the ~300 individuals in ~30 countries who have signed up to join this journey with us.

We will be using Microsoft Teams as our collaboration platform for the OHDSI COVID-19 study-a-thon. Thanks to Peter and ErasmusMC for ‘hosting’ this event on their account. Either today or tomorrow, everyone who registered for the study-a-thon will receive an email from Microsoft Teams OHDSI-COVID-19 with instructions for how to log on. You can operate Teams from your web browser, but found it works much better if you download the application. All activities during the Study-a-thon - web conference live events, subteam meetings, synchronous chat sessions - will be contained within this MSTeams platform. Our OHDSI core team has been piloting the tool for the last couple of weeks in preparation for the study-a-thon event, and I’ve been impressed with how it’s held up, so I’m optimistic it’ll work well for all of us.

When you first log-in, MSTeams will prompt you to follow a quick tutorial to familiarize yourself with the platform. Several people in the community are expert in MSTeams and can reply to this point with helpful links to other tutorials or reference materials.

When you log in, if you click on ‘Teams’ on the lefthand side menu, you’ll see a series of ‘Channels’ under OHDSI-COVID-19 that represent the streams of activities. You are welcome and encouraged to monitor all the activities across all the channels. But, you’ll receiving a separate email from me with a link to a Google form about the activities below, and based on your responses about preferences, I’ll be ‘assigning’ you to a specific subgroup to focus on as a priority (so that can balance the workload across the participants).

Go to the channel: ‘Collaborator introductions’ as your starting point. We ask as your first step to go here and introduce yourself to the rest of the community. Name, organization, areas of expertise, level of familiarity with OMOP CDM and OHDSI tools, whether you have access to any database(s), what you’re looking to get out of the study-a-thon, best coping strategy for self-isolation, whatever you’d like to share to get started.

There’s a ‘General’ channel, which the Core team will use to make general announcements throughout the 4 days. This is also where you’ll see the ability to ‘Join’ the scheduled Events that’ll take place.

We plan to host Live Events 3 times a day. Our kick-off Live Event will be 26March2020 at 4pmKST/8amCET/7amGMT/3amEST. At the kick-off, Dani, George, Peter and I will provide an overview of the study-a-thon activities, the research questions, community tasks, and next steps to get started. Then, every 8 hours for the next 4 days (at 12amKST/4pmCET/3pmGMT/11amEST, then at 8amKST/12amCET/11pmGMT/7pmEST, then again at 4pmKST/8amCET/7amGMT/3amEST), we’ll hold ‘check-in’ Live Events to share updates on progress and identify areas of focus. All of these Live Events will be recorded and posted immediately afterwards, so you if you miss them and want to catch up, you’ll be able to watch them or read a summary.

Then, you’ll see a series of channels titled "Study- " and "Competency- ".

We have developed a matrix organization for the activities we want to progress during the study-a-thon. The rows in the matrix are the ‘studies’ - specific clinical topics that we want to drive from idea to evidence. The columns in the matrix are the ‘competencies’ - specific skills that need to be performed across all studies. Each ‘study’ and ‘competency’ has group leads that will help coordinate activities, primarily through posts on these channels as well as Teams meetings. Additionally, there will be additional kickoff Events for the Characterization, Prediction, and Estimation efforts to provide on overview of activities and try to get folks organized.

We are prioritizing the following ‘studies’:

Study- Characterization: The majority of our understanding of the disease natural history of Covid-19 is limited to a discrete set of questions asked on case reports. We aim to design a characterization study that will descriptively summarize cohorts of COVID-19-positive patients across the OHDSI community as these data become available. Among patients that are Covid-19 positive, we will describe patient demographics (age, gender, index month/year), prior conditions and prior drugs in the medical history, as well as treatments and procedures in the 30d following initial presentation. When possible, we will also perform the characterization against stratified subpopulations of interest including: Gender (Male/Female), AGE<18, AGE>70, Hypertension, Diabetes, COPD, Heart Disease, Cancer, Pregnant. We will summarize incidence of selected outcomes (hospitalization with pneumonia, hospitalization requiring invasive care - ICU, ventilator, intubation, death) within the cohort and subpopulations. To provide greater background context for the COVID-19 characterization results, we will also perform characterization on historical data from prior viral seasons to characterize influenza, virus-related symptoms (fever, cough, myalgia, dypsnea), complications (pneumonia, acute respiratory distress syndrome), and invasive treatments for respiratory distress (ventilation, ECMO, tracheostomy).

Study- Prediction: Across the world, towns, states, and countries are all aiming to ‘flatten the curve’, by employing a series of public health measures aimed at delaying the communal transmission of COVID-19 such that the number of infected patients are sufficiently spread out over time that the capacity of our healthcare system can accommodate the needs of the ill. In situations where demand for health services exceed supply, prioritization tools based on disease severity become valuable and education for the public about why they shouldn’t seek unnecessary care can make a difference. We are proposing 3 patient-level prediction studies that can inform this discussion: 1) Amongst [Target cohort (T)]: persons with an outpatient visit (GP, urgent care, ER) who have flu or flu-like symptoms, who are [Outcome cohort (O)]: persons who are admitted to hospital with flu or pneumonia, in [time-at-risk (TAR)]: 0d-30d from outpatient visit? The goal of this prediction is to identify, based on the medical history prior to the first encounter, which patients are likely going to need hospitalization. 2) Amongst T: persons with an outpatient visit (GP or urgent care) who have flu or flu-like symptoms but do not have pneumonia and who are not admitted to hospital on the same day or next day, who are O: persons who are admitted to hospital with pneumonia, in TAR: 2d-30d from outpatient visit? The goal with this analysis is to help with assurance of the public that if they are sent home, the likelihood of something bad happening is low so they should follow doctor’s advice and CDC recommendations for self-care. 3) Amongst T: persons with inpatient admission with pneumonia due to viral origin, who are O1: persons requiring intensive care services or O2: persons who die, TAR: during the hospital stay? The goal here is to use medical history information prior to admission to help with triaging those who arrive at hospital to determine which cases are likely to be more severe.

Study- Estimation - hydroxychloroquine: Hydroxychloroquine is a synthetic DMARDs indicated for rheumatoid arthritis which are being investigated as potential prophylaxis and treatment for COVID-19, for its antiviral properties. We propose to conduct self-controlled case series analyses and compare hydroxychloroquine against other conventional DMARDs, methotrexate and sulfasalazine, to better understand its safety profile and determine if there are differential risks in incidence of viral disease onset and complications. We will also explore the safety of concomitant exposure to hydroxychloroquine and azithromycin. Initial analyses will use influenza as a viral model, while subsequent analyses will use COVID-19 when data are available.

Study- Estimation - IL6 and JAK inhibitors: Tocilizumab, sarilumab, and baricitinib are biologic DMARDs indicated for rheumatoid arthritis which are being investigated as a potential treatment for COVID-19-positive patients for their immunosupressive effects which are posited to reduce viral complications. We propose to conduct self-controlled case series analyses and compare these drugs against other RA biologics, specifically the TNF inhibitor adalimumab, to better understand their safety profile and determine if there are differential risks in incidence of viral disease onset and complications. Initial analyses will use influenza as a viral model, while subsequent analyses will use COVID-19 when data are available.

Study- Estimation - HIV protease inhibitors: Lopinavir and ritonavir are protease inhibitors indicated for HIV, which are being investigated as potential prophylaxis treatment for COVID-19 due to their antiviral properties. We propose to conduct self-controlled case series analyses and compare these drugs against integrase inhibitors and non-nucleoside reverse transcriptase inhibitors to better understand their safety profile and determine if there are differential risks in incidence of viral disease onset and complications. Initial analyses will use influenza as a viral model, while subsequent analyses will use COVID-19 when data are available.

Study- Estimation - HepC protease inhibitors: NS3/4A protease inhibitors (including boceprevir) are indicated for Hepatitis C, as a potential treatment for COVID-19-positive patients. We propose to conduct self-controlled case series analyses and compare these drugs against peginterferon and rivavirin to better understand their safety profile and determine if there are differential risks in incidence of viral disease onset and complications. Initial analyses will use influenza as a viral model, while subsequent analyses will use COVID-19 when data are available.

Study- Estimation - ACE inhibitors: We seek to understand implications of the ACE-2 pathway, which can serve as an entry point for COVID-19 and is also upregulated by ACE inhibitors and Angiotensin Receptor Blockers. We will compare ACE and ARB vs. other anti-hypertensive drugs (calcium channel blockers, thiazide diuretics) to evaluate: 1) ‘susceptability’: amongst new users, what is the risk of Covid-19 positive? and 2) ‘severity’: amongst Covid-19 cases who are prevalent users, what is the risk of viral complication?

To support these ‘studies’, we are establishing the following cross-cutting ‘competencies’:

Competency - literature review and protocol development : Review the current literature and synthesize the available evidence to provide background context and motivate the evidence gap that a study aims to fill. Draft Background section text that can be used for protocols and final study reports. Write Methods section text based on analysis specification defined in study package.

Competency - Phenotype development and evaluation : Review the current literature for prior phenotype algorithms and validation. Design cohort definitions in ATLAS. Evaluate cohort definitions using CohortDiagnostics and PheValuator.

Competency - Study package development : Prepare and test study packages that are fully self-contained to run across a distributed network, post on github.com/ohdsi-studies. Write execution instructions.

Competency - Study execution against network : Data partners will install and build study packages from github.com/ohdsi-studies, execute functions to generate results, and export resultset to central repository.

Competency - Results consolidation and preparation : Aggregate results from across data network. Create and post R Shiny app for moderated viewing of results on data.ohdsi.org

Competency - Clinical review: Review consolidated results and interpret findings. Write ‘Results’ and ‘Discussion’ sections of study reports.

You’ll also see a couple other channels that are opened for other topics that may come up which aren’t directly aligned to the main body of work:

Support - New study design: A place for researchers to propose additional research questions and solicit for community participation to start a collaborative design and implementation. Use the OHDSI ‘Mad Libs’ as a template for structuring a research question for characterization, prediction, or estimation, see the document posted in the channel for guidance.

Support - Analytics: A place for statistical programmers and data partners to support each other in the implementation, testing, and deployment of OHDSI analytics tools, including troubleshooting errors in study packages when running across distributed network.

Support - Data: A place for data partners who have questions about OMOP Common Data Model and vocabularies and ETL conventions to ask questions or help resolve issues that can enable them to make their data ‘network-ready’ for distributed analytics

Teams support: Open discussion for any issues with using the MSTeams platform.

So, to recap next steps for those who registered:

  1. If you registered for the study-a-thon, you receive an invitation today or tomorrow to join the OHDSI-COVID-19 MSTeams. Download the app, log in, and introduce yourself on the ‘Collaborator introductions’ channel.

  2. Complete the Google form that you’re received from me in an email this afternoon to declare your preference for which ‘study’ and ‘competency’ you would like to focus on. Submissions are due by Wed at 12pmET, so that I can ‘assign’ you to your subteam.

  3. Join us for the Kickoff Team Event on Mar29 at 4pmKST/8amCET/7amGMT/3amET.

If anybody who registered has any technical difficulties logging on to their MSTeams account, or doesn’t receive a registration email in the next two days, please email: ohdsi-support@mi-erasmusmc.nl. We hope everyone will be logged on and ready to go on Thursday for the kick-off.

For everyone, whether you are registered or not, remember, this OHDSI study-a-thon is just the start of a journey that we’ll be on together for awhile until we get this pandemic under control. The power of our effort rests with the strength of our community and the spirit of open collaboration. Beyond the internal MSTeams site, we will be posting regular updates on the OHDSI forums, Twitter, and LinkedIn, including links to recordings of the Live Event meetings, and encourage you to follow along and share our work with others who you think can benefit from our collective efforts.

Thank you all for joining the journey!

Study-A-Thon Registration Has Closed; Microsoft Teams tutorial link
OHDSI virtual study-a-thon to support COVID-19 response, to take place 26-29Mar2020...Collaborators wanted!
(Jeff Hammerbacher) #2

Hey @Patrick_Ryan, the COVID-19 Host Genetics Initiative is very interested in stratifying COVID-19 patients by ABO blood group. They have a preprint up with more information about why: Review and Analysis of COVID-19 Host Genetics and Associated Phenotypes.

If possible could your team add blood group to the list of features used for characterization of stratified subpopulations of interest? And if possible, please upload the summary statistics to https://tinyurl.com/abo-covid19.

(Hung Do) #3

Hi all,
I know that I might be a bit too late for registration, but I am very eager to participate. I can possibly join the Support- Analytics or Support-Data Team. I am currently working with clinical EHR and mapping them to the OMOP standard model. My competency is in data analytics and software development: SQL/ Python/ Linux/ Java Script …

Please let me know if you need an extra helping hand.

I will keep following the forum posts!

(Mees Mosseveld) #4

Hi Hung,

Please send an email to ohdsi-support@mi-erasmusmc.nl

Kind regards,


(Benjamin Skov Kaas Hansen) #5

Hi Mees,

I know I’m late, but I’ve been so swamped that I didn’t realise this was happening. I sent an email to the address you provided, hoping I can help out in this effort. If not, I definitely understand! I have worked with OMOP for a while so could jump in fairly easily, I believe.


(Guy Tsafnat) #6

Thanks Mees. I did send an email to the email you provided. I hope I can still participate, as I think we can provide unique contributions given our software and partner network.


(Ben Hughes) #7

FYI that full sequencing of COVID patients in UK seems like a distinct possibility.

(Andrew S. Kanter, MD MPH FACMI FAMIA) #8

Hung, how are you doing the mapping? Have you seen the recent release of the CIEL terminology for COVID-19 with maps to SNOMED, ICD-10, LOINC, etc.? https://openconceptlab.org/orgs/CIEL/collections/COVID-19-Starter-Set/concepts/

(Hung Do) #9

Hi @Andy_Kanter,
Thank you for pointing me to the vocabulary.
We are currently working with 2 hospitals in Australia to map from their legacy databases (CERNER) to OMOP. Our project is not specifically about Covid19. From our last meeting, the hospital said that the admin for Covid19 is still largely paper-based.

(Jarrel Seah) #10

Hi Hung Do,

I am a doctor and data scientist at the Alfred Hospital in Melbourne, Australia. We also use Cerner and we are starting to see covid 19 patients. Can I see what kind of data you are exporting from cerner and how? I have limited access to some of our Cerner exports and could see what I can export.


(Hung Do) #11

Hi @Jarrel_Seah,
Good evening from Sydney. Every hospital is very different. One hospital gave us very raw data directly from CERNER (25 tables), while the other hospital has a third-party layer that can extract more concise information (only 9 tables) and save us a lot of time. We usually start by asking the hospital to give us their relational schema (a list of tables and columns) so that we can check against the OMOP CDM requirements (hospital registration, ward movement, theatre, surgeries, diagnosis, allergies, pathology, medication). After finalizing the useful columns, we will ask the hospital to give us a small amount of test data so that we can check and map again.

The process is usually very slow because of 2 major problems:

  1. A hurdle of admin work, including ethical approval (may take several months) and security requirements (We are using a secure version of AWS to store the data).
  2. We need to check for data consistency. Data cleaning takes 80% of the time. (In contrast, I feel that software development is the simpler part of the project. )
    If you want to collaborate, please drop me a message and maybe I can send you some examples from the hospitals we are working with.

(Jarrel Seah) #12

Thanks @hungdo1129 - yes this is quite similar to my experience. From my position - I can expedite 1. I am also working with some hospitals in Sydney so I’m quite keen to see what you’re working on so we can synchronize our efforts.

(Hung Do) #13

Hi @Jarrel_Seah, I sent you a message with my email address. Please drop me an email so that we can exchange further details.

(Andrew S. Kanter, MD MPH FACMI FAMIA) #14

If you are an Intelligent Medical Objects customer in Australia (or anywhere) be sure to make sure that you are getting the cross maps from the IMO codes to all the SNOMED CT, ICD-10-AM, LOINC, etc. codes. Cerner may not be exporting them all. You will probably need them for the ETL.