OHDSI Home | Forums | Wiki | Github

Comparative risk of the incident cancer between histamine-2 receptor antagonists


(Seng Chan You) #1

Dear all,

The new network study is launched to compare the risk of incident cancer between histamine-2 receptor antagonists.

Comparative risk of the incident cancer between histamine-2 receptor antagonists

Abstract: Dietary N-nitrosodimethylamine (NDMA) has been shown to be carcinogenic in animals, however, evidence from population-based studies is inconlusive. The U.S. Food and Drug Administration has issued a statement on ranitidine because they may contain unacceptable levels of NDMA in 2019.
To date, there have been several studies regarding association between NDMA exposure and risk of cancer, however, real-world evidence of cancer risk in relation with ranitidine is scarce. We aim to evaluate the comparative risk of incident cancer in patients exposed to various H2 receptor antagonists (H2RAs).
We will conduct systematic, multinational study to estimate the relative risk of primary outcome (overall cancer except thyroid cancer) and secondary outcomes (overall cancer, 16 types of cancer, and cancer mortality) in ranitidine cohort. We will compare the target cohort with the comparator cohort for the hazards of outcome during the time-at-risk by applying a Cox proportional hazards model after propensity score adjustment.

The package for feasibility test is available at the OHDSI-Studies Repo

You can see the more detailed protocol here.

Currently, We are searching for collaborators to join this network study and to execute feasibility test of this study. Please follow the instruction, and please send me the result from the feasibility test first before running execute function).

The current results from feasibility test will be presented at March 3rd OHDSI call.

If you have any question, comments, or suggestion, please let me know.


(Thomas Falconer) #2

Hi @SCYou, Columbia would be happy to run the feasibility test of your study and to join it if appropriate. Will send the test results soon!


(Tomas Mackevicius) #3

Hi, @SCYou, we are new to the OMOP game at CCH, currently we are finishing to build OMOP tables for the All of Us project.

Do you know what minimal OMOP structure/tables/fields we have to implement in order to participate in this and other OMOP community trials?

Thank you!


(Seng Chan You) #4

Thank you @thomasfalconer and @mtomas

This study is based on CDM V5.3 or newer

The minimal required tables include:

  • CONCEPT, CONCEPT_RELATIONSHIP, CONCEPT_ANCESTOR
    (You can download these tables from ATHENA )
  • PERSON, OBSERVATION_PERIOD, VISIT_OCCURRENCE, CONDITION_OCCURRENCE, DRUG_EXPOSURE, DRUG_ERA

Additionally, I recommend to build the tables below:

  • PROCEDURE_OCCURRENCE, OBSERVATION, DEATH

I’ll add this information to the README in the study repo, too. Thank you again :slight_smile:

Please make sure that invalid_reason of CONCEPT table for standard concepts should be NULL, do not leave them blank or white space in the database. :wink:


(Christian Reich) #5

What’s the difference between “overall cancer” and “overall cancer”? And what’s the difference between a primary and secondary outcome in observational studies anyway, @SCYou? We cannot distinguish between more direct or indirect consequences, can we?


(Seng Chan You) #6

Sorry for confusion, @Christian_Reich
We excluded thyroid cancer from the primary outcome (‘overall cancer except thyroid cancer’), due to epidemic of thyroid cancer in South Korea (ref1, ref2)

Although we will empirical calibration and publish all the results to minimize publication bias, and the bias caused by systematic or unmeasured sources, I set the primary outcome and analysis to avoid cherry-picking and question trolling in writing a paper based on substantial amount of evidences.


(Christian Reich) #7

I see. Well, two questions:

  • Not sure why excluding the overdiagnosed patients in thyroid cancer. There are many cancers that have screening problems, the most notable are breast and prostate. Thyroid cancer could still be caused by ranitidine, and since you do a comparison the overdiagnosing would equally happen in both the test and the comparator cohorts.
  • How would the cherry picking argument go away just because you run both overall cancer and specific cancers? Let’s say the overall is not significant, but some of the specific cancers are. Now what? Want to exclude them? Or the other way around, overall there is a signal but the CE of certain cancers does include 1. As long as it is all reported you are good.

I would just list all the outcomes. Primary and secondary is a concept from clinical trials to deal with sample size problems.


(Seng Chan You) #8

You have your own point, @Christian_Reich

It makes sense. Another reason would be exceptionally favorable prognosis of thyroid cancer. And incidence of thyroid cancer itself is another outcome in this study.

Again, it makes sense.
First of all, I’d like to measure available sample size in cohort and incidence of outcomes through feasibility test phase of this study. Our protocol can be changed after feasibility test (I should’ve mentioned that). If our sample size is big enough and the result for our primary endpoint was negative, then the final conclusion can be ‘There is no significant difference in incidence of cancer except thyroid cancer between ranitidine use and cimetidine use. But we found increased risk of several cancer types including thyroid, …’ after network study. :slight_smile:


t