@Rijnbeek, @jon_duke, @Christian_Reich, @j_vanderlei and others are participating this week in a EMA workshop on Common Data Models (this was subject of a community meeting a couple weeks ago). During yesterday's session, it was highlighted by Dr. June Raine that there are currently several outstanding safety issues with substantial uncertainty that require better evidence to support regulatory decision-making. One recent example she mentioned was the potential risk of fractures and mortality associated with Radium-223 in patients with prostate cancer. The PRAC announcement is here: http://www.ema.europa.eu/ema/index.jsp?curl=pages/medicines/human/referrals/Xofigo/human_referral_prac_000071.jsp&mid=WC0b01ac05805c516f
The questions that Dr. Raine mentioned in her talk:
-clinical characterization (treatment utilization): how many patients are exposed to Radium-223? what are the baseline characteristics of these patients? what proportion of abiraterone users have concomitant radium-223?
-clinical characterization (incidence rate): what is the rate of fractures in patients exposed to Radium-223?
-population-level effect estimation: what is the drug-attributable risk of fractures for radium-223 concomitant with abiraterone vs. abiraterone alone?
Dr. Stephen Evans closed the first day of the EMA workshop but reminding everyone that, while this meeting is about common data models, we mustn't lose sight of our shared objective, which is to generate evidence that is sufficiently reliable to support regulatory decisions and improve public health. He argued that we also mustn't let the perfect be the enemy of the good, as an approximate answer quickly is much more powerful than no information at all. He went back to Dr. Raine's example, saying that if someone can shed light on the use of Radium-223 and it's potential effects, then that would be an important contribution to the EMA. It sounded to me like a direct call-to-action to the OHDSI community!
To heed that call, I have created the following two cohort definitions:
1) Help the EMA 1: New users of radium-223: http://www.ohdsi.org/web/atlas/#/cohortdefinition/1662807
This cohort identifies the persons who have a drug record or device record for radium-223. I ran this cohort internally on my data, and found that I have several hundred patients exposed in various US databases, but didn't have any exposures outside US - I suspect that has to do with the intravenous injection codes outside-of-US and how they may be getting mapped. I used the new ATLAS feature to 'Generate with features' on the larger cohorts I produced, and that allowed me to confirm that indeed, >98% of patients on radium-223 have a prior diagnosis of prostate cancer, and also identify that ~35% have prior exposure to abiraterone, including ~15% concomitantly.
- Help the EMA 2: New users of abiraterone, with use of Radium-223
This cohort starts with new users of abiraterone who started after May2013 (FDA approval date for radium-223), and then using the attrition feature, identifies the subset of those patients with exposure to Radium-223. To keep this broad initially, I've looked for any radium-223 exposure, prior to or after abiraterone exposure. Across 4 US claims databases, I'm seeing about its about 5-8% of new users of abiraterone have exposure to radium-223.
(I also created a cohort, 'Help the EMA: New fractures' and provide a template for how you can use the Incidence Rate tool to summarize the rate of fractures in the new users of radium-223 here: http://www.ohdsi.org/web/atlas/#/iranalysis/1662810)
I'd be keen to hear if others in the OHDSI community have access to exposures on Radium-223, and if so, what types of patterns you are observing? If together as a community, we have sufficient data, perhaps we can collaborate on a OHDSI network study to get Dr. Raine, Dr. Evans, the PRAC, and all the patients who are waiting, the reliable evidence they deserve.