Note: Where available, the PDF/Word icon below is provided to view the complete and fully formatted document
Senate Select Committee on Health
Health policy, administration and expenditure

MINTZES, Dr Barbara, Senior Lecturer, Faculty of Pharmacy, University of Sydney

Committee met at 08:47

CHAIR ( Senator O'Neill ): Good morning. I declare open this public hearing of the Senate Select Committee on Health. I welcome you all here today. On behalf of the committee I would like to acknowledge the traditional owners of the land on which we meet and pay my respects to the elders both past and present. I also extend that respect to Aboriginal and Torres Strait Islander people present today. This is a public hearing and a Hansard transcript of the proceedings is being made. The hearing is also being broadcast via the Australian Parliament House website.

Before the committee starts taking evidence, I remind all witnesses that in giving evidence to the committee they are protected by parliamentary privilege. It is unlawful for anyone to threaten or disadvantage a witness on account of evidence given to a committee, and such action may be treated by the Senate as a contempt. It is also a contempt to give false or misleading evidence to a committee. The committee generally prefers evidence to be given in public, but under the Senate's resolutions witnesses have the right to request to be heard in private session. If a witness objects to answering a question, the witness should state the ground upon which the objection is taken and the committee will determine whether it will insist on an answer having regard to the ground which is claimed. If the committee determines to insist on an answer, the witness may request that the answer be given in camera. Such a request may, of course, also be made at any other time. I also ask witnesses to remain behind for a few minutes at the conclusion of their evidence in case the Hansard staff need to clarify any terms or references.

I now welcome Dr Barbara Mintzes, senior lecturer at the Faculty of Pharmacy at the University of Sydney. Can I ask you to make an opening statement, and then the committee will have some questions. Thank you.

Dr Mintzes : The main focus of my research is on the health effects of medicines, including population based drug safety research. I was asked to speak to the committee about the use of linked administrative health databases in Canada and other countries and how similar research using such databases could benefit Australia.

Linkage of electronic databases covering prescriptions, physician services and hospital care allows for medication safety research to be carried out at a population level. This research is important because harmful effects of medicines are a common cause of emergency room visits, hospitalisations and even deaths. In Australia, there are 190,000 hospitalisations per year due to harm from medicine use at a cost of around $660 million. In Canada, one in eight emergency room visits are medication related and there is an estimated 100,000 to 200,000 deaths per year from medicine use in the US—over twice the number of deaths from motor vehicle accidents.

When a medicine is first approved for marketing, we know very little about its safety, especially in the longer term. On average, drugs are approved based on the experience of around 2,000 people who have used the medicine for short periods of time. Once on the market, millions of people may use the same drug. This is what happened with the arthritis drug Vioxx, or rofecoxib. In its five years on the market, Vioxx caused up to 140,000 heart attacks in the US.

Inadequate drug safety surveillance contributed to the Vioxx disaster by delaying discovery of harm. The Canadian government responded by establishing a national Drug Safety and Effectiveness Network in 2011. A key aim was to build research capacity as well is carrying out needed studies. I have enclosed an article describing this network. A recent example of the network's research is a study that found higher risks of kidney injury with use of higher dose statins. This project used linked health databases in seven provinces, as well as in the UK and the US. Knowledge of these extra risks has helped to guide treatment decisions.

Access to data at the point of care is also important. In British Columbia, all dispensed prescriptions are recorded in the provincial PharmaNet database, which became accessible to doctors and community pharmacists in 1995. Duplicate opioid painkiller subscriptions, which can lead to overdoses or diversion for illegal use, dropped sharply as a result. The experience internationally with linked administrative health databases shows that this research is important for safeguarding patients from harms from medicines and that these databases can be used in a way that is secure and ensures full protection of privacy.

In Australia, similar research is currently possible only on a limited basis. We can track individual patients' medicine use through the 10 per cent Pharmaceutical Benefits Scheme but cannot link this information to health effects. The Department of Veterans' Affairs' database allows for full linking of medicine use to other health services and health outcomes. However, this database only covers 220,000 people with an average age of 75, so it is not able to represent the Australian population in general.

Over the last 25 years only 30 Australian studies have used administrative data to track individual patients and evaluate the health outcomes of the medicines that they were taking. In contrast, in Nordic countries with a similar population size, 263 such studies were published from 2005 to 2010. The difference likely reflects the limited access to linked data in Australia. Australia needs improved infrastructure for data linkage and oversight of medication safety. Australia has the advantage of being able to gain from the experience with linked administrative datasets in other countries. The experience to date in Canada, the US, the UK and Scandinavia makes it clear that these databases are important tools for medication safety and protection of public health. Thank you.

CHAIR: Thank you very much for that opening statement. This is the committee's first hearing into big data to try to get an idea of what the international implications are. Clearly, coming from another context, we appreciate you coming to give us your evidence. Could you just speak a little more to the decision making and the action that happened when Canada realised that they needed to do something in response to the concerns that were raised about that particular drug. What were the processes around enabling that? Is there any detail you can give us with regard to that?

Dr Mintzes : This is an academic and government joint initiative. A lot of the initiative to actually start a Drug Safety and Effectiveness network nationally came, really, from outside of government—from clinicians and medical organisations and also academic researchers—but there was a recognition in government that the resources that were devoted to post-market drug safety surveillance were quite limited and that there was a need for additional resources being put to this. So this network works closely with government—with Health Canada, the regulator—but is based in universities, so it is a virtual network throughout the country.

CHAIR: It is quite a while ago that this happened, so we are at a very different point of time in terms of technical capacity right now here in Australia. How was it facilitated, and was there much resistance?

Dr Mintzes : I do not actually know about the resistance to it. As to how it was facilitated, it has been funded through the Canadian Institutes of Health Research. Actually, I have given you a document that describes how it was formed in detail. Part of what the network is doing is working with the regulator, Health Canada—which is similar to the TGA, essentially—where there will be a particular drug safety concern that Health Canada would like to see researched at a population level. Then it is also able to initiate its own research projects as well. There has been some funding for other research projects that other researchers are involved in.

One of the aims of the network has also been training of people—newer researchers—in analysis of administrative databases to actually bring this area of research forward. I think the thinking was that most of the information in terms of drug safety surveillance once a drug is already on the market traditionally has been based on spontaneous adverse drug reaction reports. These are dependent on clinicians or the public actually recognising, if there is a harmful effect, that it is due to the medicine or may be due to the medicine and knowing that they can report to Health Canada. We know that there is a massive degree of underreporting of harmful effects of medicines.

CHAIR: In layman's terms, then, could you explain how the incident would trigger a registration of an event in a different way if the databases were linked.

Dr Mintzes : For instance, if you took the 10 per cent PBS sample and had an ongoing structure in place to link it to hospitalisations and to deaths, and right now we had another drug like either Vioxx or the diabetes drug Avandia that had been shown to cause an increased rate of heart attacks, essentially you would follow through time the population of people who are prescribed that medicine and then look at their rate of hospitalisations for heart attacks and, say, deaths due to cardiovascular disease or overall deaths. You would compare that with an otherwise similar population of people who are not using this medicine or are using an alternative medicine that is less associated with that harm. You can then look at the number of extra events that you have as a result.

CHAIR: So, in essence, because the database records so much more, instead of doing a moment in time snapshot or fishing for a needle in a haystack, you have permanent longitudinal data that you can access prospectively or retrospectively to see what has gone on.

Dr Mintzes : Exactly. In fact, you can also link it. Some of these databases have been linked to things like children's readiness for school, starting school, and school performance. Some of the area of the research that I work on is exposures to medicines in pregnancy. You can then follow not only what happened in a woman's pregnancy but whether there is a difference in health outcomes for the exposed children through time.

CHAIR: Currently, in Australia, how do you access that data, Dr Mintzes?

Dr Mintzes : You have the real expert on that sitting behind me and who is going to be speaking later on—Sallie Pearson and her group. In Australia, with the 10 per cent PBS data, you can follow people through time by looking at their prescriptions. So you might be able to combine the prescriptions. There has been research carried out combining prescriptions for one medicine and then looking at later prescriptions for another medicine that might be a marker for something that has gone wrong as a result of the medicine used. But you cannot easily link that to hospitalisations or to, say, what the person's diagnosis was in the first place. For instance, a recent study looking at statin use where there was an expected difference among people who had previous heart disease and people who did not have previous heart disease was not actually able to look at those diagnoses for heart disease. In Australia, there are all the separate databases with this administrative data because every act in a hospital and every prescription that is dispensed is recorded in a database, but they are not linkable. The actual infrastructure work that needs to be done in order to link these databases is not yet in place.

Senator MOORE: Dr Mintzes, I am fascinated by your research on pregnancy and the use of drugs. A number of other committees which I have been on have been looking at that. I just want to put that on the record. I am trying to get my head around the evidence that we have received where, of course, there are problems with linking the databases, but a number of researchers have said that it is very difficult to trace patients through the process because, if you are trying to find out the exact things which you have been talking about—the impact of particular interventions, medications or actions on people in the system—gaps occur. You can match little bits but you just have these gaps about what happens. Because of the privacy restrictions, it is very, very hard to know what happens with, say, a person who lives in a remote area and has a predisposition to cancer as they go through the system. Can you help me with how that works?

Dr Mintzes : When researchers are accessing the data, it is completely de-identified. They are not able to look into a dataset and see 'When did I go to the doctor and for what?' But each person has a unique code, which means that that same code can be linked between different datasets. So it is possible to then follow a person through time, to look at what happened in, as you say, a remote rural area with their medical care and also their previous medical history, and to follow them through, say, with prescriptions, hospitalisations and hospital discharge data. And then there is the birth and death data. So that is the whole point—

Senator MOORE: Those peak things have data on them. If we are wanting other information we have to see whether there is data available. The things you have identified are already in the system somewhere. The evidence is all of the problems about how you get them to link.

Dr Mintzes : Yes. So the question really is that there needs to be work done to develop that infrastructure. I think Australia has an advantage in terms of developing the infrastructure now, when it has already been in place in different countries. In Canada, we have different databases for each province. We also have a system where some of the health care is provided at state level and some is provided at a national level. It is a bit of a hodgepodge, you could say. Looking at the various databases, some of them would do a more complete job, for instance, of capturing everything that needs to be captured. In the British Columbia database, for instance, you can capture the diagnoses, the medical consultations, the prescriptions, the hospitalisations, but we cannot completely capture emergency room visits. They are often a very important first signal in terms of harm. Similarly, we do not have across-the-board indications as to why the person was prescribed a medicine. That could easily be worked into the database to begin with.

Where it has worked especially well is in Ontario. There is a group called ICES, an institute for clinical excellence—I always forget what that stands for—which is managing the linked health databases at a provincial level and has internal capacity to carry out a lot of the analyses that need to be carried out, working closely with the provincial government. Other researchers can also access those data for questions that they have. Part of my point is: there is a lot of international experience now in terms of how to set up these linkages. Also, there has not been the problem with loss of privacy that people are concerned about. That is other question. The Scandinavian countries are world leaders in terms of having everything in the single databases that cover all of people's medical care—the hospitalisations, births, deaths, et cetera. Again, they have not had the problems with breach of privacy. The same is with the US, where you have large health insurance databases and state Medicaid databases.

Senator MOORE: Some of the submissions have mentioned the international comparisons and the big data conferences where people with the knowledge get together and talk to each other. Is there a goodwill amongst the people in sharing knowledge and actually learning from each other? You would have read some of the submissions, I know, and there is this tension that suggests, 'Australia does some things well, but if we do not do better we are going to fall behind the international comparisons.' This theme is running through a lot of the submissions—that we really have to take the opportunity to do things better now or fall too far behind. In that arrangement, do people learn from each other and take up best practice?

Dr Mintzes : I would hope so. My experience is not with big data conferences. I have gone to conferences of the International Society for Pharmacoepidemiology. That would include many of the presentations there. Certainly, there are Australian researchers also attending those conferences. Many of the presentations there are really on honing methods in order to do this research, because there are complications with it compared to, say, clinical trial research. You have a lot of control over what happens in a clinical trial, and people can be randomised to one exposure or another. When you are looking at population level data, there are hundreds of thousands of individual clinical decisions about who gets a specific medicine or not. So the methods then are fairly complex in terms of being able to compare how people do over time without mistakenly thinking that people do worse because they started out worse—to put it in lay terms.

CHAIR: I want to just ask about the health data a discrete set of pieces of information. In one of the submissions that we have here—and, I think, we will be meeting the presenter in the next roundtable—the linkages across collections are not considered traditional health data. How important is that? I fear that that might be a point where there are points of resistance, with people asking, 'Why do the health services need to know about my social life?' These are points of concern, I expect, that would happen.

Dr Mintzes : I have not seen the other submissions because I was invited quite late to come, so I cannot really comment on the other submissions. Your social life is one thing. I do not think that tends to end up in the databases. In research that I have been involved in in British Columbia, we looked at districts of residents as a proxy for socioeconomic status. That is, again, because we know that socioeconomic status is a very strong determinant of people's health and for other things. For instance, if you are following up on how well children who have been exposed to a certain medication prebirth are doing, you also need to know about socioeconomic status. It is very general and they are large enough districts that there is not a concern about invasion of privacy or being able to recognise individuals. But there are situations where it can be helpful to have certain types of data that are not strictly medical also included or linkable into these databases. There has been a lot of work in British Colombia with the BC perinatal database, for instance, linking it in. It has much more depth of information in terms of health during pregnancy, other exposures during pregnancy and things like smoking, which often you do not get by just by using administrative data alone. You do not know who is a smoker and who is a non-smoker, for instance.

Senator MOORE: So it is a social determinants of health model?

CHAIR: Yes. I think that there are so many other integrated parts, and we will hear more about it throughout the day. From your experience in Canada and North America generally, how do you see the relationships between provinces and the nation, and the states and the nation here in Australia, working in terms of information-sharing? Is that a political problem or is that one that was overcome reasonably easily?

Dr Mintzes : The Drug Safety and Effectiveness Network is actually a very interesting example of how that is being overcome. They will join together to decide to research a specific question. Researchers in each province have separate linked databases, so they are able to do separate analyses that are then joined together into a single study. There is a real interest in being able to look at things jointly. We have some differences in that Australia has, I would say, a better managed and organised drug payment system nationally through the Pharmaceutical Benefits Scheme than we have in Canada, because each province pays for drugs separately and there is a bigger mix in terms of drug payments of private and public payments. But it is still possible to capture, at the pharmacy level when prescriptions are dispensed, every dispensed prescription regardless of who is paying for it. I would say that it just takes some planning and discussion, but there is no reason that Australia, with its mixed Commonwealth and state system, would be less able to have these sorts of linked databases than Canada, with its mixed provincial and federal system. We are dealing with some of the same issues.

CHAIR: Given the experience that you have to bring to this, if you were in a position to make a recommendation about the way in which Australia should proceed in this area, what would be the key elements of such a recommendation?

Dr Mintzes : I would say to really plan the data linkage in discussion with researchers who are already involved in this area in Australia in order to make sure that it is planned in a way that is actually able to oversee not only medication safety but also quality of healthcare services and how health care is being carried out. Understanding that this is an important platform for patient safety and health system performance is kind of a prerequisite in the sense that there is a need to put some resources into having systems set up for data linkage that would work well in the long term. So I guess I would say: involve the researchers in the planning. These are administrative data in the sense that the databases may be developed for billing purposes, for hospital financing purposes and for a number of other reasons other than for medication related research. But in planning how to link them together, it is important to also consider how they could be used in a way that would really improve the oversight of the health system's performance and could improve medication safety oversight. I cannot say anything about how much money that would cost. It is just not my area. It may be something that some of the other people who will be speaking together today will be able to speak to. But it should actually pay for itself. If you think about the billions of dollars spent on medicines and on health service provision, if there are tools in place to be able to manage that provision of healthcare funding better by being able to, in an ongoing way, examine what the effects are, then that is going to pay for itself. That would be my guess.

Senator MOORE: Could I ask one more question?


Senator MOORE: Dr Mintzes, you actually put in your submission about the 10 per cent Pharmaceutical Benefits Scheme that you had. Is there any background to the 10 per cent? Is that all you can get, or is that what you needed to have?

Dr Mintzes : No. There is a dataset that has been set up for research—

Senator MOORE: That is what I thought.

Dr Mintzes : at this point, which, I suspect, Professor Pearson is going to speak about in detail. She and her team have been very involved in researching—

Senator MOORE: I think she is just making notes there! So the 10 per cent is something that is available from the government?

Dr Mintzes : Yes.

Senator MOORE: That is what you are allowed to have?

Dr Mintzes : Yes. That is accessible for research. It is a very important tool, but it still has limitations in the sense that—

Senator MOORE: There is 90 per cent that is not there?

Dr Mintzes : Okay, so that is one limitation. For certain types of studies you can take a sample, as well, of the population. But the limitation would be that it is not easily linkable to other data. You can follow individual patients though time with a code that is used as an identifier but not follow whether they have ended up in hospital as a result or even died—I am not sure—but whether they have ended up in hospital and the other use of medical services linked to the use of that medicine. I am just saying it is a tool that is in place, but it is limited at this point. It could be developed much further.

CHAIR: Thank you very much for joining us this morning, Dr Mintzes. We appreciate the insights that you have brought. You have tabled a document from the Canadian Network for Observational Drug Effect Studies. Senator Dastyari, will you move that we received that?

Senator DASTYARI: Yes, of course.

CHAIR: Thank you, Senator Dastyari. Given the season, Merry Christmas.

Dr Mintzes : Thank you.