OHDSI Home | Forums | Wiki | Github

OHDSI Study #4: Investigating Birth Month-Disease Risk Relationships Across the OHDSI Network

Hi all,

A new proposed OHDSI study protocol No. 4 has been posted on the Wiki and on Github.

This study is based on the original SeaWAS study developed and performed at Columbia University Medical Center (full details are available here). In this study, we want to perform the same analyses across multiple institutions throughout the United States of America and the world. We then hope to integrate the results to delve into the relationship between environment at birth (based on season and climate) and lifetime disease risk. We plan to use established climate models (we have some climatologist collaborators) to probe these aspects into more depth.

The code is already written and tested on the OMOP CDM v.4. Thorough testing on v.5 of the CDM has not been performed although the tables used in this study should not have changed across versions and we are in the process of performing more thorough testing on this.

Please provide any comments or suggestions. If you would like to join as a contributing investigator, contributing to the analysis and write-up of this work, please let me know.

I plan on discussing this project at the next OHDSI meeting on August 18th, 2015 to provide some more details and to answer any questions you may have.

Thanks.

And since this is OHDSI I feel compelled to say Bon Voyage!

Mary

Our AUSOM data have birth months for all patients enrolled in it. Thus we can join this study. However, as we are in shortage of hands, passive participation such as administrative works for IRB exemption, running the analysis code and returning the results only will be possible.

@rwpark: That would be great!
Running the code and returning the results is great, I’ll update the Wiki page and list you as a collaborator.

We will be happy to share our findings back with you once the analysis is complete, but it should not require work from you at that stage.

thanks @schillil and @nigam for your interest in running SeaWAS, let me know if you have any questions
I’ll be working with some administrators here at Columbia to get V.5 working especially for @nigam
thanks @Vojtech_Huser, @Patrick_Ryan for your comments as well
if anyone else wants to be involved just let me know!

@nigam and other v.5 users: we are pleased to report that testing of SeaWAS code on v.5 is complete and the code is fully functional on v.5. Code is available on Github: https://github.com/maryreginaboland/SeaWAS

Thank you for warm welcome.
I am so impressed with the research topic (phenome-wide approach to analysis of the association between birth-month and disease risks) and the paper published in JAMIA.

The IRB has approved this study in Ajou university, Korea.

Now, I do work with the code you shared.
I find small part of codes, which doesn’t work with our CDM data.

(1)Upper case error
in SeaWAS_demographics_OHDSI.R
on 139th rows, " age_data$age = age_data$year_of_service-age_data$year_of_birth "
->this sentence should be replace with "age_data$age = age_data$YEAR_OF_SERVICE-AGE_DATA$YEAR_OF_BIRTH "

The error occurred because column names of our DB were in upper cases.

(2)an error in SQL sentence
SeaWAS_RR_allconditions_OHDSI.R
on 65th row,
it works after I replace "GROUP BY (a.condition_concept_id) " with "GROUP BY a.condition_concept_id,c.concept_name ".

(3)
SeaWAS_main_OHDSI.R
on 134th row,
it works after I replace “query_2$condition_source_value[j]” with “’”,query_2$condition_source_value[j],"’" , because the values in condition_source_value were not integers in our DB.(They were in characters)

I do appreciate that I can have a chance to work with you on this fabulous research topic. I really hope that I can add little help to this project.
I’ll send the data from my hospital ASAP after a brief review for possible mistakes I made

Sincerely,
SC You.

All:

I want to congratulate @Mary_Regina_Boland. Her 2015 paper on birth month (http://jamia.oxfordjournals.org/content/22/5/1042) was highlighted as one of the best papers of the year during Russ Altman’s Year-in-Review at the AMIA Joint Summits meeting. This paper is also currently the ‘most read’ article published by JAMIA.

Also, @Mary_Regina_Boland did a great job with her presentation during the AMIA student paper competition, showing how we can link the birth-month disease relationships to seasonal changes in compound variation via gene polymorphisms to identify novel associations, following structured ontology information across multiple sources. It was a terrific demonstration of straddling the TBI->CRI continuum.

Just to reinforce the opportunity for the OHDSI community: I think Mary’s research study proposal represents a tremendous opportunity to generate large-scale evidence across our global network as a means of confirming hypotheses that have been initially derived at Columbia. I would encourage everyone to participate in this study, essentially all you need is PERSON.MONTH_OF_BIRTH and the CONDITION_OCCURRENCE table to produce aggregate summary results that can be shared for a collaborative publication.

For those with CDM data from administrative claims but without birth month: recently I discovered (the obvious) that I could infer birth month for children in claims using a combination of observation period start (often, if a child has insurance coverage during their first year of birth, there’s a good chance the insurance coverage started in the birth month) as well as timestamped procedures (e.g. services associated directly with delivery and birth, as well as other other bounded markers, like 2-wk pediatric checkup). So, whereas initially, I didn’t think we could contribute to this study with our available, we now will be able to participate using the subset of persons (children) for which we can infer birth month. This may be a trick that others find helpful in enabling you to join in on this study as well.

Hi Seng Chan You:

Thanks so much for your interest!
We are excited to have you on board.
If the code ran properly with your modifications then that is great - I don’t see anything that should change the results from an algorithmic perspective. I anticipate that some of the differences are due to differences on the hardware/software side.
Please let me know if the code ran properly with your modifications. Also feel free to email me at mb3402@cumc.columbia.edu with any additional questions.

Best,
~Mary

Hi Mary,

I recently joined @Patrick_Ryan’s group in Janssen. I am interested in participating in this study using our data sets. I will begin trying the SeaWAS code in the next week and provide some feedback on how it went for data sets that don’t require any formal protocol submission.

Thanks,

Ajit Londhe

@Mary_Regina_Boland
I cannot send you the SeaWAS result from Ajou Univ, Korea (I don’t know why the email doesn’t work. I’ve tried several times but I failed to send you an email every time). So I send an e-mail to you, via OHDSI. Please check the e-mail of OHDSI
Sorry to bother you.

t