OHDSI Home | Forums | Wiki | Github

Research questions that the OHDSI community can potentially answer to suport the COVID-19 response

(Nigel Hughes) #42

Is there any utility in looking for any patterns with concomitant medications, e.g. statins, and outcomes with COVID-19 (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4244940/pdf/BMRI2014-872370.pdf)

Would be useful to know if any other existing therapies people are taking may have an impact on e.g. severity of disease, or indeed mortality.

There have been some submissions to IMI along the same lines.

(Seng Chan You) #43

In Korea, we’re trying to make a dataset with previous medical history of patients with the COVID-19 :slight_smile:

(Ray Chen) #44

Another question that has come up in the context of clearing providers who have been exposed or have URI symptoms to ensure we have enough clinical staff to support care: how long does it take for these tests turn positive after exposure? (or with symptoms; but presuming with symptoms, we would expect there to be few false negatives)

And how generalizable are the experiences/data around COVID testing across institutions or countries? I presume they’re all PCR-based but seems like everyone created their own test. Can we use data from around the world to figure out when we can safely rule out an exposed individual?

(Neus Valveny) #45

In case no detailed prior medication history is available, association of outcomes vs presence of autoimmune diseases like systemic lupus erythematosus (SLE) that use hydroxychloroquine (HCQ) could be interesting…

(Vojtech Huser) #46

I presume they’re all PCR-based but seems like everyone created their own test.

This is indeed an interesting question. Looking at temp LOINC codes for it(I think the CDC test; LOINC does not say this (it should)), the genes tested are specified in the component of the test.

Inventory of tests by country is indeed a nice goal.

(Seng Chan You) #47

Interesting! @Vojtech_Huser

I don’t know which vocab will be most suitable for Korean test… (https://github.com/OHDSI/Covid-19/wiki/Novel-vocabulary-for-COVID-19)

(Vojtech Huser) #48

Since you find this interesting, read more on LOINC forum here: https://loinc.org/forums/topic/sars-cov-2-codes-discussion/

If other countries need different codes, LOINC seems to encourage that (on their main landing page)

(Ed) #49

Please note, we have now launched an ECMO OHDSI study with further details here

(Julianna Kohler) #50

Has anyone suggested looking into the France MOH’s assertion that NIAIDs are counterindicated for covid? He seems to be basing it on the idea that NIAIDs are immunosuppressants, but I haven’t seen any case studies or data around covid, specifically.

(Ana Szarfman) #51

Heavy smoking seems to be a risk factor

(Ana Szarfman) #52


(Julianna Kohler) #53

Sorry, NSAIDS. Clearly thinking about too many things!

(Jason) #54

Great to see how active the forum is here. I am infectious disease doctor and the lack of data is a huge problem. We are making clinical decisions off of extremely limited data and so any help from the OHDSI community would be overwhelmingly appreciated! Some of the things we have been debating in clinical meetings (many of which have already appropriately been mentioned already) include;

ACE-I and ARBs harmful or protective
The role of Chloroquine/Hydroxychloroquine as they have direct anti-viral effects in addition to immunomodulatory ones
The role of steroids in COVID-19 ARDS
The potential for NSAIDs to be harmful
Vent management for patients

Cases are sadly expected to rise and whatever the community can do to put the best data possible in the hands of clinicians would be appreciated.


A really basic (but non-trivial) question of importance to public health officials will be to estimate the number of deaths due to Covid-19 in different countries since we know that these are likely under-reported to varying degrees.

For example, the 2009 H1N1 pandemic had 18,500 laboratory confirmed deaths worldwide, but the study below estimated actual mortalities to be between 151,700 to 575,400 deaths.


The methods in this work relies on a lot of assumptions of about baseline characteristics of populations etc.

Can we develop methods that make greater use of claims/EHRs data to inform and improve these types of mortality estimates?

OHDSI community could develop and test approaches using historical claims/EHR data to study H1N1 mortality and compare to published literature examining this question for H1N1. The approaches could be ready in the OHDSI community by the time there is sufficient data on Covid-19 to apply the approaches.

(Dale Penndorf) #56

Has Athena been updated with all of the new LOINC, SNOMED, CPT, AND ICD10 codes for COVID-19?

(Vojtech Huser) #57

Thinking about consent and research infrastructure - I was researching informatics questions about deceased patients few years back. In this COVID time, it is tiny bit relevant perhaps. I know it is imperfect, but if we create a subset of COVID data from US that only contains deceased patients), than a special provision (in terms of IRB review [not needed]) for use of data on deceased for research apply. See details in this article here:
Title: Don’t take your EHR to heaven, donate it to science: legal and research policies for EHR post mortem

(Ru Cheng) #58

I applaud @Patrick_Ryan and community for this effort! As a physician researcher in women’s health, my team and I have some women specific research questions to propose with regard to pregnancy and lactation. There are guidelines from professional societies (ACOG, SMFMF’s joint statement here: https://www.acog.org/clinical/clinical-guidance/practice-advisory/articles/2020/03/novel-coronavirus-2019) but there is not enough data to really inform these guidelines. Any additional clarity on these topics would be of tremendous public health interest.

~Pregnancy and lactation:
• rate of infection in pregnancy. By trimester if possible.
• complication rates for pregnant infected women
o severity of illness (versus non pregnant women, versus men)
o comorbidities
o length of illness
o transmission to newborn
o pregnancy complications: premature birth rates, fetal loss
• does cesarean section v vaginal delivery influence transmission to newborn?
• any transmission through breast milk to infant

I echo @SCYou’s earlier suggestion for analysis by sex for any and all research questions. There may be important sex differences that could drive different approaches to triage, diagnosis, and treatment. Additional research questions for women are listed below.

• are women tested at the same rate? If not, why?
• Do women have access to testing?

~Clinical presentation
• Infection rates (known, will not be fully possible until after the epidemic and serology surveillance is conducted)
• Infections with and without symptoms, any difference when compared with men?
• Prevalence of presenting symptoms (not just respiratory)
• Severity of illness
• Length of illness
• Time from symptoms to hospitalization_and Impact of comorbidities
• Radiology findings – CT, CXR – any differences between men and women? _
• “silent spreaders” – any difference in rates between men and women?

~Health outcomes:
- Response to specific medications/treatments ie antivirals, and to vaccines when testing begins
• Required hospitalization
• ICU admissions
• Ventilator requirement
• Length of stay
• Death rates?
• Impact of comorbidities on sequelae of infection ie co-infection- flu , bacterial infections

  • what is impact of COVID-19 on existing co-morbidities

~Social impact:
• Transmission rates-woman exposed versus man exposed. There may be differences in how women come into contact w virus compared with men. Ie women are more likely to take public transportation than men. Women are more likely in supermarkets.
• Differences in access to care
• Engagement of women versus men in healthcare related to outbreak, Women are more frequently front line health care workers (nurses, phlebotomists, aids, etc) so may experience more occupational exposure than men

~Can a Risk prediction model be developed for men and women?
o severity of illness
o need for ICU
o need for ventilatory support
o increased length of stay

Thank you all for your consideration!


I agree as well.
And what are the effects on the fetus or newborn baby ?
so many things need to be looked in the case of mother and newborn baby .

(Sturkenboom) #60

@ru-cheng @rajeev1 Great to hear this interest. We tried to set up the maternal and child health group in OHDSI some time ago, but there was minimal interest back then. In the IMI funded ConcePTION consortium https://www.imi-conception.eu we have developed a protocol for studies responding to these questions and it would be great to explore how to all collaborate. As we know pregnancy outcomes take some time to reveal especially if we wish to look at the impact on different trimesters, good thing: that buys us some time. Happy to organise a meeting to discuss.

(Michael Kallfelz) #61

I probably missed that, but do we have already a post looking into the role of co-infection with other viruses in more severe cases? I was thinking reactivated sleeping viruses such as Varicella Zoster or Epstein Barr. Varicella Zoster could be consistent with liver damage and the fact that children experience a severe course of disease less often, maybe because they had been routinely vaccinated.
I feel it would be good to review records of childhood diseases, vaccination status as well as looking into endemic prevalence of common viruses in hot spots that have a higher count of severe cases. With a quick lookup for VZV I found only this. Here Italy has a slightly higher incidence of cases in the age group 40+