Understanding What FDA is Looking for in Postmarket Safety Assessments: Going Beyond AERS Webinar

good morning and good afternoon everybody and thank you for joining our webinar i'm jeff Antos from the Weinberg group and I'll be moderating today's session on understanding what the FDA is looking for and posts market assessments going beyond ayers the Weinberg group is a regulatory and scientific consulting firm serving pharma biotech and device clients we are headquartered in Washington DC I know you're really going to enjoy our speaker today dr. Bob Roth is the Weinberg group's medical director he holds two doctorates a PhD in biochemistry from Baylor and an MD from the University of Texas at Galveston Bob provides consultation in all phases of research development and marketing of FDA related drugs biologics and medical devices a noted author and speaker bob has led numerous projects in pharmacovigilance take it away Bob thank you very much Jeff what I'd like to do today is give everybody a brief overview of options that are out there for for drug manufacturers for drug safety surveillance activities I'm sure that most of you are aware of ayres which is the FDA's adverse event database system what we'd really like to talk about today is is a little bit about errors but but mainly all the other options that can go into you know companies either routine drug safety surveillance or focused activities perhaps very often in in the context of putting together an NDA as a just a way of introduction you know the past few years I think what everyone is seen is that FDA's focus not only is for new and innovative effective drugs but but safer drugs and and with all the the major major drug toxicity problems blockbusters taken off the market FDA has they've taken a lot of heat over the years for drugs that got approved and then had to be withdrawn situations like that so what FDA is really requiring of drug manufacturers sponsors is you know an extensive safety profile to begin with for for the approval and then a drug safety post market safety surveillance program that that hopefully is is sensitive enough to pick up problems that may have been missed during development again by way of introduction a lot of what I'm going to talk about today really began to be put into public focus a few years ago when FDA published their safe use initiative back in 2009 and the focus of this particular FDA publication was actually on unnecessary injuries from drugs this is preventable medical errors and misuse the the context was that literally hundreds of thousands of injuries and fatalities a year that are associated with with drug use and misuse would actually be considered preventable and FDA has in various sources have estimated it anywheres between a low of 10% and as much as a half of injuries are actually preventable if if everyone is paying attention so the safety initiative talked about interventions that could be made across systems these are patients and providers caretakers insurers other federal agencies like the DEA and CDC and and everybody taking a part in paying attention to drug safety then identifying particular drugs and particular problems that have been identified in keeping track of them prospectively this is something that FDA does internally and and the whole idea is to be able to recognize something that that may or may not turn out to be a verifiable problem if it does be able to do something before there's a lot of harm and FDA does this routinely in-house they have a very very good and I think pretty complete safety surveillance system in-house where they look at their own data and over time and and determine if there's some sort of problems and and so they've they go to the next slide what they have FDA has done is developed a list of drug candidates this is a growing list these are drugs that are followed internally there's these lists it's there's safety alerts for human medical products drugs and therapeutic biologics and you can you the public can go and see what this list consists of these are drugs that have been identified with certain concerns safety concerns but at this point there's no there's no conclusion that there's a problem even a problem let alone a problem that needs to be fixed with something like a withdrawal or a change in a label it's simply a list of drugs and problems that everyone should know that FDA has there has their eyes on and the basic concept that comes out of the safe use initiative is the quote in the bottom bullet on the slide the the majority of harm that you get from approved drugs is actually fixable and comes from things like misuse inappropriate use medical mix-up sand and the like so I wanted to very quickly just give a case example of propoxyphene because this is one of the the first drugs that's sort of in the process of looking at the safety of this drug a lot of the post market surveillance databases I'll talk about for the rest of the stock sort of first became you know part of the the public awareness so this goes back to 2006 when Public Citizen requested that propoxyphene be take off the market and this was mostly realized related to suicide risks and a few years later FDA actually did put out a a review of the propoxyphene evaluation that they had done they've acknowledged that there was evidence for overdose situations but they did not go to the point of saying that that there was an acknowledged suicide risk at that point what they did point out – which had not been a major focus previously was was a risk of cardio toxicity and FDA's concerned main concern at that point was whether or not cardiac toxicity should be the reason for removing the drug from the market and what they recommended at that time were some safety studies a year later after further updated review they did eventually recommend that the drug be withdrawn and the reason at that time were fatal arrhythmias now the drugs safety sir sources at FDA evaluated for for this decision are the ones that we're going to be talking about today first of all there was Aires FDA's own adverse event system and what was clear from errors that there were many propoxyphene associated deaths and many suicides they also looked at mortality data that came from medical examiner data in the US and also suicide data and from some sources abroad they reviewed the the dawn' system the drug abuse warning network system of data this is this is one of the sources that we'll talk about a little bit later in the in the talk they evaluated data that came from the place the national poison control centers we'll talk about that and those data also talked about fatalities and suicide attempts and so these are sort of the the basic data sources that FDA evaluated and then ultimately came to a conclusion about risk versus benefit now that the data sources I want to talk about today are briefly Ayers and then get into an analogous system called the vigeous that's run by the WHL then get to the drug abuse warning network dawn the poison control and then a few words about safety surveillance the analysis of electronic medical records the basis for considering all of these different sources is number one that you know FDA safe use initiative brought them all up as available information sources that can have utility and actually can and don't necessarily overlap completely so you can get some independent information these sources were all identified in a number of times over the years and different groups they associated with FDA and going back to 2005 division of risk evaluation mentioned airs dawn and the poison control system there was discussion of these sources in front of advisory committees and in even a House committee I should say that there is no obligation that a drug company or anyone else has to follow or evaluate their drug safety in any of these sources certainly most people will pay some attention to the airs because that is the US FDA system there but there's certainly no no official mandate that any of these other systems be be evaluated however is FDA is said in a number of places there there really can be some very good utility to paying attention to these sources it just briefly about airs I'm assuming that most of you are somewhat familiar with with the MedWatch and air system these are it's a database of MedWatch reports that are submitted by physicians and pharmacists and and customers and other sources most of the information in airs falls into the regulatory category of us of a medically important a serious adverse event the airs system is is is a large and updated regularly updated database that's that published with quarterly updates the information the safety information that goes in is is QC than in some fashion and so there is a lag period between when a case gets into Aires and when it shows up in a quarterly publication this is generally a three to four month lag so if on January first you're looking for Aires information on on your drug of interest the last entry that you may be able to find is likely to be many many months before that the information in Aires can change many of you are aware that that there's an initial MedWatch report and then in many cases additional information comes in and and a follow-up report is submitted and so that the the record in Aires is can be a changing record over time there can be follow-up information clarifications and and and duplicate reports can be weaned out so that you know from one point of time to another a report can change should be aware that that the publicly available information in an Aires quarterly update does not provide narratives surprisingly not all FDA reviewers know this and they'll sometimes ask for narratives when new drug applications are put in and and they have we have to educate them that narratives are not available through FOI requests it is possible to get a full MedWatch report with narrative but this is if if any of you have done FOI requests previously it's it's it's a one by one request weeks or months can go by and and it's just simply not practical for any kind of routine safety surveillance should say that there are a number of ways to look within the AERS database to try to understand whether or not there are some interesting associations between drugs and side effects changes over time FDA itself uses data mining procedures routinely to look for you know strengths of drug adverse event associations and and this process is basically what allows them to to create their list of safety alerts that we talked about previously the other thing that's possible within within the Earth's system are what are called standard Meldrick queries measures the coding language for adverse events and so you may have a particular interest in your drug say of pancreatitis and you could simply go to Ayres and see how many reports there are pancreatitis but there are lots and lots of other terms that could show up in a nares report that would not necessarily say pancreatitis but would be but would represent that issue and this is what the standard measure queries do they their groups of terms that together represent the concepts of interest these can be names of disorders signs symptoms diagnosis syndromes physical findings and so I just presented here as an example if you are interested in a granular cytosis is a potential side effect of the drug here's the various preferred terms that the a given case could have been coded as so you can see if you simply look for a green lesotho psious you would miss many of many of these other cases and so the the smq process is a way to get a better view of of a disease process so that's all I was going to say about about Ayers vigeous is is a very similar system its adverse event data much like the air system but it's much more worldwide it's run by the w-h-o and maintained by the upsala monitoring center it's a voluntary system and countries around the world will submit adverse adverse events information will be submitted to a national database and pretty much every country around the world and then and then in some cases countries will also submit their information to the vigeous and so it's it's a fair number of participating countries it's basically the same information that goes into a MedWatch report contrary to the FDA process there's not really a QC step that's that's it's available in here the material if if you're interested in your particular drug and what adverse events exist in Vidhya base it's possible to obtain this information that takes anywhere from a few days to a couple weeks it there's a small cost but but it does have to be ordered and paid for via the upsala Center and what we use we use Vigi base very routinely in drug submissions and the the value that the way that we use the database is is we will extract all out all information in all cases that have originated from the US with the assumption that these cases were already reported in errors and so we end up with a database that that we call an X us safety data set and we use that to compare and contrast with with what we find independently from from an errors analysis there are some some differences between Aires and vigil base one is that Aires as I mentioned is is updated and and then published as a quarterly update report where as bigger bases is maintained regularly and every day in essence you can get a different and more updated data set if the Aires information is predominantly coming from the US whereas the vigeous data are are predominantly from countries other than the US and we found it particularly useful in looking at adverse events in Asian populations that are generally very very poorly represented in Aires but you can simply focus your your video based search on adverse events that are originated from China Japan Korea and you know specific Asian countries if you want and pick out those data that are most likely related to an Asian population specifically the other thing that we found every time that we that we do this exercise is that whereas ayres reports are generally considered serious most of the information in video base is actually non serious I'd like to next go to the drug abuse warning Network to the dawn system these are the data that are collecting collected from emergency rooms it's it's a nationally representative database of over a thousand ers the information includes drug-related deaths coming from medical examiner's and-and-and toxicities if you will that in complications that are seen in the emergency rooms the system the the data system is operated by the substance abuse and mental health services administration so it's not an FDA related database but but Samsa it's the Samsa program has gone through a number of changes over the years several years ago about a decade ago was redesigned as new dawn and then it was actually for confidentiality reasons further redesign was was made that actually closed the system for a couple of years and then it's and then it reappeared recently as as the substance abuse mental health data archive what you what you the public can can get from Dawn are number one some some general reports if you're of interest but they turn out not to be very useful for drug surveillance or you can actually obtain the what are called the public use files which are the actual data datasets of cases that are compiled from these emergency room visits and the the new dawn system that just just started up again several months ago was one in which annual reports now are being generated so at this point in time there are I guess nine annual reports that are available and you look year by year so if you're unfortunately if you're interested in and a long term view of your drug of interest and Dawn you have to evaluate and analyze each of the annual reports what's represented in dawn are prescription and OTC medications illegal drugs dietary supplements alcohol non-pharmaceutical inhalant so it's it's a broader view of toxic compounds than than what you see in airs but it does include some information that's that is related to – you know drugs are drugs of interest the information that you see in in a dawn record number one a case is categorized as by by some drug-related category name abuse addiction detox misuse things like that so you get sort of a general category that that the cases is considered there are I should mention that again like like in airs there's there are no available narratives so you this the story which I believe is actually end on some place is not not publicly available in addition if you see information that is that really is reflective of an adverse event the the presentation of this information is not like an errors there's no body system presentations or use of terminology that would be consistent so it's it's really quite a different way of presenting drug information what you see in in done evaluation if you're interested in your particular drug of interest you can get the number of reports that have been submitted over that year you get patient demographics and this does include race identification which is not present in eras or bigger base there are seven general categories of reason whether called reasons for the emergency room visit the I've listed here one of which is called adverse reaction but the others obviously are more you know drug abuse related in poisoning related categories information that falls under the the context of disposition includes what has happened to the patient and the options are that the patient was treated in the emergency room and released release could be sent home or or sent to jail or someplace else but not to a hospital and then a second second disposition category is admitted to the hospital it does the the records to give information for what kind of unit if you will within the hospital one gets admitted to and then and then there's a category of others so it is possible if you know from a record that that the patient was seen in the emergency room and then released to and admitted to a surgical unit in the hospital that that you know this would qualify as a hospitalization as you would see in errors in the mid watch situation jumping now to the the national poison data system this is information that comes from the American Association of poison control centers so it's it's independent of dawn but it's it's a somewhat analogous system but these are the poison centers this previously had the name of tests but more recently called the NPDES there's 64 major poison control centers this is a real time toxic Oh vigilant system cases when when patients are seen in a poison control center there's pretty much an instant electronic submission of the case information to the database and so this is this is really a real-time data source there are some adverse event information not it's somewhat limited but but there is some again no no narratives and if you want to evaluate the information in poison control you need to go through the AAPC see put in a request that has to be reviewed and accepted by that group and there is a there is a form of a fairly complete form that has to be filled out and there's a very high charge not to give numbers to things but but in essence you will be charged by the number of specific drugs that you're searching for and and by by the number of years that you're looking so the charges can grow quite rapidly the information that you get from a database of from the NPDES our demographic information age gender weight whether or not a woman is pregnant the quantities of exposures that are ingested and then a number of reasons for exposure on the left hand side in this in this on table is what's called adverse drug reaction which is pretty much the closest to what you know we would call you know a drug a ii should point out that the actual intent of an adverse drug reaction is to to be able to describe situations that that come from normal use exposure so normal prescription use of the drug so the intent is to exclude problems and there could be adverse events but problems that come from misuse suicide attempts and things of that nature I think as you can see from the slide the most common type of category that you see is is unintentional therapeutic error which represents things like incorrect dosing inadvertently taking someone else's medicine infant exposures via breast milk things of that nature you know sometimes this sort of information can actually be helpful when tied in to specific you know adverse event information that you might be able to glean from a case each case is defined by an outcome and these are the lists of outcomes so there's no greater definition of these outcomes and then than this but but major moderate minor potentially toxic this is the sort of information that that when we use the database for evaluating a you know a marketed drugs safety for say a an NDA submission for a 505 B tube product we'll use this information to make the case that most of the for example most of the occurrences that that show up in the in the poison control centers are perhaps minor effects associated with a stroke and then that that's not really a problem in terms of the adverse events and what you can get out of the poison control database is extremely limited as many of you are aware there are tens of thousands of adverse event terms and errors for instance so that that can describe a particular side drug side effect in the poison control database there's about a hundred there's a little over a hundred terms and these are terms that are not we're not chosen to represent drug adverse events these were chosen to reflect you know poisoning problems so like burns and acidosis and so most of the information that you see in that the clinical effects portion of a poison control record are not really related to to you know the drug drug per se but you know when you're looking up the the side effects whatever is reported the side effects of your drug there will be generally there will be one or two terms that reflect something to do with with an organ toxicity so the way so we we use these four basic data sources fairly routinely for for drugs for many drug subs submissions certainly not all they have been recommended the group of them by FDA for pediatric drug submissions when there's not a great deal of pre market safety data obviously drugs with potential abuse FDA maybe want to look at each of these these sources if a drug has a potential need for a special formulation or packaging to make it safe and not abusable obviously those those sorts of drugs would fall into this category and then and then we've seen in many times drugs that are undergoing a process of rx OTC switch FDA wants before they before FDA concludes that it's safe to mark this drug without a physician being involved they would like to know what sort of safety information exists in in each of these sources now one other fairly recent concept that that FDA has been promoting is what they call their Sentinel initiative and I'd like to just say a couple words about this these as as I mentioned the adverse event information and errors for instance is cases that were submitted to Midwest reports submitted to airs from sometimes patients themselves or physicians or pharmacists but this is generally the considered tip of the iceberg most most side effects don't don't get reported people don't take the time or don't have the interest so the the Sentinel initiative is a way that FDA is trying to get access to a more broad view of drug side effects so they're looking at a drug safety information that comes from sources like electronic health records insurance databases and and registries is a way to sort of collect much more of that of that iceberg of data so for comparison at this point FDA has what they call a pilot the mini Sentinel project where they now have greater than 100 data from more than 100 million individuals in terms of drug safety and adverse event information this would compare to airs that has a small very very small fraction of that information the last quarterly update when we looked at the drug in ears had about a hundred and fifty thousand reports obviously you know you know a hundred hundred times or greater source of information in the the set no initiative process a couple things about electronic medical records I mean it sounds like a wonderful idea for finding out all sorts of site drug side-effect information and in probably this this is the wave of the future because like I said most most adverse events don't actually get reported to FDA so this is a way to to avoid the problem of under report ability what needs to be developed or better systems to to prompt physicians to to have some sort of recognition of adverse events in the hospital notes and and in addition some way for pharmaceutical companies to use this sort of information and that's really in its infancy at this point what you what you can get are a fairly representative view of safety and safety signals that come out of lab reports you know in a hospital record progress reports physician and nursing progress reports all of this information potentially is is a way to depend point some organ toxicity or some other some other side effect that would be generally ignored you know for the most part there is a an issue that's that's sort of in progress being worked out in progress which has to do with coding of this information as I mentioned the the AERS information and also the vigil base works with you know a large number of coded terms the meddra terms for specific adverse events and when a physician writes a note in a chart this obviously does not use meddra terminology so there needs to be a way of translating what the words that show up in a medical record into adverse event term terminology that can be compared with these other sources in some way and and this is where there is quite a bit of current advance in terms of natural language processing and ways computer system ways to take information and put it into some format in which it's possible to you know compare one record to another in an actual case that we ran into a couple of years ago was was quite interesting FDA actually required a sponsor when they were putting in an nd8 to to do electronic medical record review this actually turned out to be a huge process it costs and then credible amount of money it took six months or more to accomplish protocol had to be written software had to be generated for analysis and all that but the question in this particular case had to do with a you know a drug that would mask you know some some sentence that would treat some symptoms in a situation where the symptoms themselves could be the first indication of an underlying in this case an underlying cancer so this was a particular situation where there was a potential of a renal related cancer and if renal symptoms were suppressed by the drug could that hide the cancer so that it was not diagnosed until later and the only way to answer a question like that would be with electronic medical records and look at long term long long-term follow-up of patients and it was actually possible to answer the the FDA question to their satisfaction via this this approach and and one last one last source of information I just wanted to mention very very briefly didn't put together a slide on it but there was just a very recent publication that that I noticed a week or two ago but there there are there's there's now the beginning of drug safety data mining that's being done via the Internet and there was a very interesting publication we wrote a brief blog on it on our website if you're interested but it was a way that the author's would look at internet search sites they focused on Google Yahoo and and I think a couple of the other major sites and they were trying to to answer the question of whether or not it would be possible using searches just simple you know search evaluations to look for potential drug interactions so what they did is they looked at all the Google searches over a period of time and looked for associations between two drugs of interest and a a side effect and we're able to identify that that two particular specific drugs were in a search a group of search terms associated with the side effect much more often than than other drug combinations and it was an identification of a potential drug interaction that actually in this case turned out to likely be real so this is again probably a at this point more an academic process but at some point it's it's almost certain that FDA will will start you themselves using internet search approaches to trying to find out specific drug problems so to conclude just the basic the the basic messages from this webinar are that there are a large number of data of drug safety databases that go beyond errors and FDA has in some instances asked for several – all of these all of these sources of information FDA is promoting a wide safety evaluation for risk management and so we have seen in a number of instances where they will ask that that companies prospectively follow some of some or all of these these data sources and and keep an internal record of what they're finding is similar to doing you know routine literature searches for instance there are there are currently some hot spots if you will narcotic products and pediatric products where it's much more likely that that some or all of these data sources would be followed in some way as opposed to just airs but I think over time the the groups of products that that you know will be evaluated by these these these various sources are clearly getting larger and larger the data sets are complementary you just don't see a case that's reported in in Don and also clearly you know in Aires or it's actually difficult to know necessarily but the preliminary view is that is that the the information is generally complimentary the the bigger problem is that there's no easy way to integrate the information so when if if a drug submission is made and these sources are all reviewed and analyzed for the integrated safety analysis the analyses are separate their individual last comment is that this kind of process is somewhat time consuming certainly if poison control information is is involved it can be the expense can be considerable however we found many many companies now larger drug companies are are using these sources routinely so that they are kept aware and so that and so that they can provide a convincing argument to FDA that that in fact you know they're they're doing the best that they can and with that I'd like to finish and I would be happy to take any questions thanks Bob we've got a number of questions here let me start out with one of from one of our participants and the question is why do we not normally tap the safety database of electronic medical record reviews for post marketing reporting well that the main reason is because it's it's a it's a pain there's it you have to write a protocol you the IT people have to be involved for software analysis so that you can be sure of the data that you collect and analyze there's HIPPA requirements IRB approvals so there's a number of steps that that get involved that means that this process costs a lot and takes a lot of time so it's just not practical on a routine basis and and I mean I certainly would not recommend that anyone start off with a electronic medical record analysis for anything routine certainly recommend that you wait until s FDA requires it but the main thing is is the cost in time it's it's not trivial it's not something that that you know somewhat you know a regulatory person can do at their desk okay that makes sense Thanks another question could you explain a little bit more about foi how that works well the Freedom of Information Act says that that FET you know data that it was include FDA data are available upon request there's there's a commercial there's a company called foi services that one can go to and and request information that is you know made available to the public you know sort of upon request but not necessarily you know provided by FDA or some other government agency you know it's just sort of out there on a website so if you are interested in a summary basis of approval document for a drug that's not yet available on FDA's own site that can be requested as an FOI request as I mentioned before adverse event information from specific MedWatch reports can be can be required can be those requests can be submitted as well foi Services has a large number and growing number of documents that that are it has obtained and our electric electronically available for sale to people and so many times if if you make a request of information to foi services you you simply you know pay the pay the fee and the information is available almost immediately however if they don't already have the information they have to to get it through an FOI request that's that's what can take a lot of time okay thanks hey how much does it cost to use the AERS database well actually getting getting the information is is is not costly I mean there's there's no charge specifically for you know for the for the data set although what what you get is something that is very difficult to use yourself and if you you can actually make errors request for information through commercial search services such as dialog and that doesn't cost very much in practice there is some expense to doing a an air search and analysis because unless you have the software and the IT capability yourself in-house you need to go through a vendor and so there's there some cost that has to do with with vendor expenses I guess it's a generality I would say that if you're putting together an NDA submission and you have reason to include in your safety evaluation you know in Aris analysis of some related product that's out there and you go through a vendor and and have them obtain and and go through all the QC and all that kind of stuff with the with the data set and then prove provide you with particular tables of interest you know you're probably talking about inexpensive ten or fifteen thousand dollars something of that nature that that's not the cost for the data itself but that has to do with once the data are obtained you know working with those systems and and doing all the the IT related work okay good thank you if a healthcare professional submits non-serious cases are they going to be included in Aires yeah basically anything that's anything that comes in as a MedWatch report that meets minimal criteria ends up in errors in the know minimal criteria basically means that there's a problem identified in an identifiable patient the identify to be an identifiable patient all you need is some information about gender or age or something so it's very minimal but but the idea being that there's evidence that there is a specific patient in mind with the report and so yeah so you'll see cases that are non serious as well as serious FDA is more interested in serious cases so that's really what it was intended for but probably I don't know maybe 20 25 percent of cases and here's for any given drug might be non serious obviously if you're looking for a non-toxic drug generally non-toxic drug information that there may be more of the non serious cases and and if you have a highly toxic drug that may be much more weighted toward the serious the other thing it beyond your question specifically that's important to recognize is that when a MedWatch report is submitted to heirs a serious criteria seriousness criteria is applied to the case such that the case is considered serious or non serious as a whole and there may be one more than one there may be quite a number of adverse events that come out of a case that are coded out of a case based on the narrative and if your case is is considered serious from the regulatory perspective and so a common situation would be someone that was hospitalized because of a side effect all the side effects all the adverse events that are coded out of that out of that case record are are identified as serious so it's very obvious that that this is not reality in in a particular case record someone may have been hospitalized because of of what turned out to be a heart attack but other adverse events that come out of the the narrative might include hangnail so obviously that that's not a serious medical problem but it does get categorized as an SAE as a serious adverse event because the case as a whole was okay all right thanks for just about out of time here I just want to let the audience know that we will be sending you the presentation plus the the questions and answers you should on the left side of your screen see a survey panel and I encourage you to just take a moment to fill

Leave a Reply

Your email address will not be published. Required fields are marked *