npg
© 2014 Nature America, Inc. All rights reserved.
NEWS that included Sirturo cleared the bacteria from their sputum faster than 81 people who received the same drug regimen plus a placebo. However, those who received Sirturo also had a greater increased risk of death: nine deaths occurred in the Sirturo group compared to two deaths in the placebo group. As part of the approval, the FDA required Janssen, the drug’s manufacturer, to conduct a phase 3 trial to confirm the drug’s efficacy, but Janssen isn’t required to complete the trial until 2022. “That seems like a very long time horizon, and I don’t see the justification,” Kesselheim says. Lisa Vaga, a spokesperson for Janssen, explains that the company is looking for ways to get the information sooner: “We’ve been in active discussions with health authorities and additional external partners to identify ways to meet our post-marketing commitments and generate additional clinical data, as well as how to accelerate the outcome.” Society may accept drug approvals based on preliminary evidence “because we want drugs to market faster,” says Ross. But “the argument then is we need to have much better data in the post-market period.” Policing drugmakers The FDAAA authorized the FDA to fine companies that fail to comply with postapproval requirements or charge them with misbranding, but the FDA seldom uses this
authority. The authors of the medical device study published in September found the FDA has never issued a warning letter, fined a company or withdrawn a device’s approval because of a manufacturer’s failure to meet deadlines or complete a post-marketing study. And Nature Medicine found only two instances of warning letters issued to pharmaceutical companies because of failure to begin or complete a post-approval study required under the FDAAA. One of those went to Shionogi Inc., the New Jersey–based arm of a Japanese pharmaceutical company. In September 2012, the FDA notified the company that it would be required to conduct two clinical trials to identify an unexpected serious risk of heart problems associated with its pain medication Rybix ODT, a drug that was approved in 2005. The company told the FDA they did not intend to comply because the costs of the study “were not commercially justified based on the sales potential for Rybix.” Shionogi proposed that the FDA allow the company to sell down its remaining stock of the drug before pulling it off the market. In a warning letter sent last spring, the agency advised Shionogi that it was in violation of the FDAAA. A month and a half after the FDA sent the letter, the company discontinued Rybix. Shionogi did not comment on its reasons for doing so when contacted by Nature Medicine.
Kesselheim suggests that drugmakers might have more incentive to complete their postapproval studies if a drug’s or device’s approval automatically expired after a certain number of years unless the required post-approval studies had been completed. “But that would require a legislative amendment,” he says. Even when manufacturers do complete their post-approval requirements, the data can be hard to come by. Two years ago, Alexander tried to gain access to the results of a post-marketing study by filing a US Freedom of Information Act request. The FDA told him his request could take several years to process and that the agency might not be able to fulfill it because the documents could contain proprietary information regarding the manufacturer’s risk mitigation efforts. The request for information remains unfulfilled, and without access to that kind of data, Alexander says, it’s impossible to assess whether the FDA’s risk mitigation strategies are helping to make drugs safer. “The jury is out not because the data isn’t there,” he says, “but because the data isn’t available for review and evaluation.” Cassandra Willyard 1. Reynolds, I.S. et al. J. Am. Med. Assoc. Intern. Med. doi:10.1001/jamainternmed.2014.4194 (29 September 2014). 2. Fain, K., Daubresse, M. & Alexander, G.C. J. Am. Med. Assoc. 310, 202–204 (2013). 3. Downing, N.S. et al. J. Am. Med. Assoc. 311, 368–377 (2014).
As funds to sync health records dwindle, research could suffer Along a stretch of road in Boston lie two hospitals within a mile of each another, Brigham and Women’s Hospital and the Beth Israel Deaconess Center. In the event that a resident nearby has a heart attack and is brought to the emergency department, the choice of hospital could determine how quickly he or she receives medications based on his or her medical history. That’s because if the patient’s electronic health record (EHR) resides at one institution, it may not be available in a format that can easily be read by the other hospital. Even if partial information is recorded about the individual in each system, “the data won’t be able to talk to each other,” says Catherine DesRoches, a senior researcher at Mathematica Policy Research, a public policy research firm headquartered in Princeton, New Jersey. A massive push toward digitizing health records came in 2009 when lawmakers in the US passed the HITECH—short for the Health Information Technology for Economic and Clinical Health—Act. To date, $547 million
has been made available by that legislation for states to set up ‘health information exchanges’, which facilitate the transfer of electronic patient records between hospitals, diagnostic centers and pharmacies. But that money is now running out fast. And with the future of these exchanges in doubt, the prospect of harnessing them for research purposes is also under threat. Hospitals across the country are building EHR systems to fulfill the criteria of different stages of federal guidelines in order to receive payments as a part of the Medicare EHR Incentive Program. As of September 2014, out of the 5,000 hospitals in the US eligible for the Medicare EHR Incentive Program, only 235 hospitals had demonstrated that they used certified systems that shared digital patient data continuously for 90 days in a manner that complied with the latest stage in federal guidelines, according to the US Centers of Medicare and Medicaid. However, proponents of electronic health record sharing say that more hospitals use digitized
NATURE MEDICINE VOLUME 20 | NUMBER 11 | NOVEMBER 2014
information than ever before. A 2014 report from the US Office of National Coordinator for Health Information Technology, the agency entrusted with spearheading the government’s health information technology efforts, stated that more than 59% of hospitals in the US use some form of electronic health records that include the results of diagnostic tests, patient summaries, admission dates and other such information. Private vendors such as Practice Fusion, Surescripts and GE Healthcare’s Centricity programs, and handful of open source solutions approved by the government, offer software packages for EHR systems. Wisconsin-based software company Epic, which sells a wide share of EHRs currently on the market, estimates that by late 2015, 51% of people in the US will have a health record stored in one of their electronic systems. But the transfer of information between hospitals can prove difficult because each vendor’s proprietary software runs differently. Even two hospitals in the same network might 1225
NEWS
npg
© 2014 Nature America, Inc. All rights reserved.
Expensive exchange Some private information exchanges charge hospitals using different pricing models, either on a subscription basis or by individual charges, to share health data. Generally, there are high costs associated with building new interfaces to share data between different vendors. Seven EHR software vendors, including Allscripts and athenahealth, formed the CommonWell Health Alliance in 2013, to make data sharing easier between alliance members. But health exchange between large vendor networks still remains a difficult and costly process. The health information exchanges established by states were envisioned to make the sharing process simpler by assimilating patient information in a format that would be useful to all medical institutions. But many hospitals don’t have plans to pay to support such exchanges out of their budgets. “Providers and hospitals don’t have a business case to do this,” says DesRoches. Julia Adler-Milstein, a health management and policy researcher at the University of Michigan–Ann Arbor, who conducts a yearly survey of health information exchanges with her colleagues, found that 52% of them listed grants and contracts as their most substantial source of support (Health Aff. 32, 1486– 1492, 2013). Because of their heavy reliance on federal funds, some have proved to be unsustainable. Adler-Milstein expects that without any new funds, some will shut down in the near future. State-run exchanges could have been the driving force for biomedical research and collecting surveillance data, but AdlerMilstein laments the lack of long-term planning for financial longevity when the federal funds were initially distributed. “Each state was required to have a plan, but [the plan] didn’t need to get vetted. That’s where things broke down,” she says. The many state-run exchanges lacked longterm plans to establish central repositories of data, which would have made it easier for multiple hospital systems to access that information. “All health information exchanges exist because of the lack of interoperability. But all exchanges don’t allow for the data to be interoperable,” AdlerMilstein says. Emerging solutions As the fate of state-run exchanges remains uncertain, alternative initiatives continue
1226
Brain light / Alamy
encounter difficulties sharing information if they have different versions of the same software.
Digital divide: Systems lack interoperability.
to strive to make patient record systems from different medical sites interoperable for the purposes of research and improving care. The eMERGE (short for Electronic Medical Records and Genomics) Network, which received funding from the US National Human Genome Research Institute in Bethesda, Maryland, was launched in 2007 to facilitate high-throughput genetic research by linking DNA biobanks with electronic medical records from specially curated data warehouses at ten hospitals and university divisions, including Vanderbilt University School of Medicine in Nashville, Tennessee, where Joshua Denny researchers bioinformatics. Denny used the eMERGE Network data to assess how often patients with a range of more than 3,000 different mutations associated with illnesses such as celiac and Alzheimer’s diseases developed those ailments (Nat. Biotechnol. 31, 1102– 1110, 2013). It’s not just government-funded initiatives that have tried tackling the challenge of interoperability. “One of the goals [of setting up exchanges] was health research and to test treatment options. The private health system is doing a great job at that,” says Andrew Phillips, an informatics researcher at the MGH Institute of Health Professionals in Boston. Researchers say large health networks such as Kaiser Permanente, headquartered in Oakland, California, have understood the advantages to such systems and have employed them for improving treatment patterns. Last year, a study showed that the implementation of their EHR system reduced emergency room visits for patients with diabetes by 5.5% (J. Am. Med. Assoc. 310, 1060–1065, 2013). On 29 July this year, Kaiser Permanente began collaborating with Johns Hopkins University in Baltimore to share electronic health records so that doctors could access
clinical information more easily. In light of such circumstances, private exchanges are currently seen as more receptive to making their systems interoperable, at least within their own networks. But these private exchanges, too, have their flaws. The dominance of private networks in the exchange of health information has made it difficult for the data to be analyzed at a larger level that includes community hospitals and smaller clinics. Furthermore, although hospitals have access to their own EHR data, individual biomedical researchers must jump through many technical hoops to access it. “In general, the data can always be removed from the system, but the technical expertise to do that doesn’t exist in every system,” says Kenneth Mandl, an informatics researcher at Boston Children’s Hospital. “Some need to hire consultants to mine that out, and EHR vendors haven’t made it particularly easy to get the data out.” Mandl and his colleagues have used a $15 million grant received in April 2010 from the US Department of Human and Health Services under the HITECH Act to develop the SMART platform, short for Substitutable Medical Apps & Reusable Technology. He explains that the open-source and free SMART platform can extract the useful bits of health data from different software systems and thus help developers build new interfaces that present the information in a format that the researcher wants. “Like the iPhone, the SMART platform gives developers a programming interface to allow them to create apps, without talking to the people at Apple,” Mandl says. The first clinical implementation of this platform was in 2012, when Mandl and his colleagues built an application that helped cardiologist Justin Zachariah at Boston Children’s Hospital analyze the blood pressure of different patients over time while adjusting for their age by percentile, something doctors found difficult to do during daily visits. Although the patient records in that case were from a single EHR system, applications that use disparate EHR systems have since been adapted to run on the SMART platform. Ultimately, exploiting health information exchanges to successfully exchange data for medical and research purposes will need stronger efforts from both the government and the private vendors. “The number one challenge is to build the underlying infrastructure for data exchange,” DesRoches says. “If we don’t get that right, we won’t be able to reap the benefits of the EHRs.” Manasi Vaidya
VOLUME 20 | NUMBER 11 | NOVEMBER 2014 NATURE MEDICINE
© 2014 Nature America, Inc. All rights reserved.
NEWS that included Sirturo cleared the bacteria from their sputum faster than 81 people who received the same drug regimen plus a placebo. However, those who received Sirturo also had a greater increased risk of death: nine deaths occurred in the Sirturo group compared to two deaths in the placebo group. As part of the approval, the FDA required Janssen, the drug’s manufacturer, to conduct a phase 3 trial to confirm the drug’s efficacy, but Janssen isn’t required to complete the trial until 2022. “That seems like a very long time horizon, and I don’t see the justification,” Kesselheim says. Lisa Vaga, a spokesperson for Janssen, explains that the company is looking for ways to get the information sooner: “We’ve been in active discussions with health authorities and additional external partners to identify ways to meet our post-marketing commitments and generate additional clinical data, as well as how to accelerate the outcome.” Society may accept drug approvals based on preliminary evidence “because we want drugs to market faster,” says Ross. But “the argument then is we need to have much better data in the post-market period.” Policing drugmakers The FDAAA authorized the FDA to fine companies that fail to comply with postapproval requirements or charge them with misbranding, but the FDA seldom uses this
authority. The authors of the medical device study published in September found the FDA has never issued a warning letter, fined a company or withdrawn a device’s approval because of a manufacturer’s failure to meet deadlines or complete a post-marketing study. And Nature Medicine found only two instances of warning letters issued to pharmaceutical companies because of failure to begin or complete a post-approval study required under the FDAAA. One of those went to Shionogi Inc., the New Jersey–based arm of a Japanese pharmaceutical company. In September 2012, the FDA notified the company that it would be required to conduct two clinical trials to identify an unexpected serious risk of heart problems associated with its pain medication Rybix ODT, a drug that was approved in 2005. The company told the FDA they did not intend to comply because the costs of the study “were not commercially justified based on the sales potential for Rybix.” Shionogi proposed that the FDA allow the company to sell down its remaining stock of the drug before pulling it off the market. In a warning letter sent last spring, the agency advised Shionogi that it was in violation of the FDAAA. A month and a half after the FDA sent the letter, the company discontinued Rybix. Shionogi did not comment on its reasons for doing so when contacted by Nature Medicine.
Kesselheim suggests that drugmakers might have more incentive to complete their postapproval studies if a drug’s or device’s approval automatically expired after a certain number of years unless the required post-approval studies had been completed. “But that would require a legislative amendment,” he says. Even when manufacturers do complete their post-approval requirements, the data can be hard to come by. Two years ago, Alexander tried to gain access to the results of a post-marketing study by filing a US Freedom of Information Act request. The FDA told him his request could take several years to process and that the agency might not be able to fulfill it because the documents could contain proprietary information regarding the manufacturer’s risk mitigation efforts. The request for information remains unfulfilled, and without access to that kind of data, Alexander says, it’s impossible to assess whether the FDA’s risk mitigation strategies are helping to make drugs safer. “The jury is out not because the data isn’t there,” he says, “but because the data isn’t available for review and evaluation.” Cassandra Willyard 1. Reynolds, I.S. et al. J. Am. Med. Assoc. Intern. Med. doi:10.1001/jamainternmed.2014.4194 (29 September 2014). 2. Fain, K., Daubresse, M. & Alexander, G.C. J. Am. Med. Assoc. 310, 202–204 (2013). 3. Downing, N.S. et al. J. Am. Med. Assoc. 311, 368–377 (2014).
As funds to sync health records dwindle, research could suffer Along a stretch of road in Boston lie two hospitals within a mile of each another, Brigham and Women’s Hospital and the Beth Israel Deaconess Center. In the event that a resident nearby has a heart attack and is brought to the emergency department, the choice of hospital could determine how quickly he or she receives medications based on his or her medical history. That’s because if the patient’s electronic health record (EHR) resides at one institution, it may not be available in a format that can easily be read by the other hospital. Even if partial information is recorded about the individual in each system, “the data won’t be able to talk to each other,” says Catherine DesRoches, a senior researcher at Mathematica Policy Research, a public policy research firm headquartered in Princeton, New Jersey. A massive push toward digitizing health records came in 2009 when lawmakers in the US passed the HITECH—short for the Health Information Technology for Economic and Clinical Health—Act. To date, $547 million
has been made available by that legislation for states to set up ‘health information exchanges’, which facilitate the transfer of electronic patient records between hospitals, diagnostic centers and pharmacies. But that money is now running out fast. And with the future of these exchanges in doubt, the prospect of harnessing them for research purposes is also under threat. Hospitals across the country are building EHR systems to fulfill the criteria of different stages of federal guidelines in order to receive payments as a part of the Medicare EHR Incentive Program. As of September 2014, out of the 5,000 hospitals in the US eligible for the Medicare EHR Incentive Program, only 235 hospitals had demonstrated that they used certified systems that shared digital patient data continuously for 90 days in a manner that complied with the latest stage in federal guidelines, according to the US Centers of Medicare and Medicaid. However, proponents of electronic health record sharing say that more hospitals use digitized
NATURE MEDICINE VOLUME 20 | NUMBER 11 | NOVEMBER 2014
information than ever before. A 2014 report from the US Office of National Coordinator for Health Information Technology, the agency entrusted with spearheading the government’s health information technology efforts, stated that more than 59% of hospitals in the US use some form of electronic health records that include the results of diagnostic tests, patient summaries, admission dates and other such information. Private vendors such as Practice Fusion, Surescripts and GE Healthcare’s Centricity programs, and handful of open source solutions approved by the government, offer software packages for EHR systems. Wisconsin-based software company Epic, which sells a wide share of EHRs currently on the market, estimates that by late 2015, 51% of people in the US will have a health record stored in one of their electronic systems. But the transfer of information between hospitals can prove difficult because each vendor’s proprietary software runs differently. Even two hospitals in the same network might 1225
NEWS
npg
© 2014 Nature America, Inc. All rights reserved.
Expensive exchange Some private information exchanges charge hospitals using different pricing models, either on a subscription basis or by individual charges, to share health data. Generally, there are high costs associated with building new interfaces to share data between different vendors. Seven EHR software vendors, including Allscripts and athenahealth, formed the CommonWell Health Alliance in 2013, to make data sharing easier between alliance members. But health exchange between large vendor networks still remains a difficult and costly process. The health information exchanges established by states were envisioned to make the sharing process simpler by assimilating patient information in a format that would be useful to all medical institutions. But many hospitals don’t have plans to pay to support such exchanges out of their budgets. “Providers and hospitals don’t have a business case to do this,” says DesRoches. Julia Adler-Milstein, a health management and policy researcher at the University of Michigan–Ann Arbor, who conducts a yearly survey of health information exchanges with her colleagues, found that 52% of them listed grants and contracts as their most substantial source of support (Health Aff. 32, 1486– 1492, 2013). Because of their heavy reliance on federal funds, some have proved to be unsustainable. Adler-Milstein expects that without any new funds, some will shut down in the near future. State-run exchanges could have been the driving force for biomedical research and collecting surveillance data, but AdlerMilstein laments the lack of long-term planning for financial longevity when the federal funds were initially distributed. “Each state was required to have a plan, but [the plan] didn’t need to get vetted. That’s where things broke down,” she says. The many state-run exchanges lacked longterm plans to establish central repositories of data, which would have made it easier for multiple hospital systems to access that information. “All health information exchanges exist because of the lack of interoperability. But all exchanges don’t allow for the data to be interoperable,” AdlerMilstein says. Emerging solutions As the fate of state-run exchanges remains uncertain, alternative initiatives continue
1226
Brain light / Alamy
encounter difficulties sharing information if they have different versions of the same software.
Digital divide: Systems lack interoperability.
to strive to make patient record systems from different medical sites interoperable for the purposes of research and improving care. The eMERGE (short for Electronic Medical Records and Genomics) Network, which received funding from the US National Human Genome Research Institute in Bethesda, Maryland, was launched in 2007 to facilitate high-throughput genetic research by linking DNA biobanks with electronic medical records from specially curated data warehouses at ten hospitals and university divisions, including Vanderbilt University School of Medicine in Nashville, Tennessee, where Joshua Denny researchers bioinformatics. Denny used the eMERGE Network data to assess how often patients with a range of more than 3,000 different mutations associated with illnesses such as celiac and Alzheimer’s diseases developed those ailments (Nat. Biotechnol. 31, 1102– 1110, 2013). It’s not just government-funded initiatives that have tried tackling the challenge of interoperability. “One of the goals [of setting up exchanges] was health research and to test treatment options. The private health system is doing a great job at that,” says Andrew Phillips, an informatics researcher at the MGH Institute of Health Professionals in Boston. Researchers say large health networks such as Kaiser Permanente, headquartered in Oakland, California, have understood the advantages to such systems and have employed them for improving treatment patterns. Last year, a study showed that the implementation of their EHR system reduced emergency room visits for patients with diabetes by 5.5% (J. Am. Med. Assoc. 310, 1060–1065, 2013). On 29 July this year, Kaiser Permanente began collaborating with Johns Hopkins University in Baltimore to share electronic health records so that doctors could access
clinical information more easily. In light of such circumstances, private exchanges are currently seen as more receptive to making their systems interoperable, at least within their own networks. But these private exchanges, too, have their flaws. The dominance of private networks in the exchange of health information has made it difficult for the data to be analyzed at a larger level that includes community hospitals and smaller clinics. Furthermore, although hospitals have access to their own EHR data, individual biomedical researchers must jump through many technical hoops to access it. “In general, the data can always be removed from the system, but the technical expertise to do that doesn’t exist in every system,” says Kenneth Mandl, an informatics researcher at Boston Children’s Hospital. “Some need to hire consultants to mine that out, and EHR vendors haven’t made it particularly easy to get the data out.” Mandl and his colleagues have used a $15 million grant received in April 2010 from the US Department of Human and Health Services under the HITECH Act to develop the SMART platform, short for Substitutable Medical Apps & Reusable Technology. He explains that the open-source and free SMART platform can extract the useful bits of health data from different software systems and thus help developers build new interfaces that present the information in a format that the researcher wants. “Like the iPhone, the SMART platform gives developers a programming interface to allow them to create apps, without talking to the people at Apple,” Mandl says. The first clinical implementation of this platform was in 2012, when Mandl and his colleagues built an application that helped cardiologist Justin Zachariah at Boston Children’s Hospital analyze the blood pressure of different patients over time while adjusting for their age by percentile, something doctors found difficult to do during daily visits. Although the patient records in that case were from a single EHR system, applications that use disparate EHR systems have since been adapted to run on the SMART platform. Ultimately, exploiting health information exchanges to successfully exchange data for medical and research purposes will need stronger efforts from both the government and the private vendors. “The number one challenge is to build the underlying infrastructure for data exchange,” DesRoches says. “If we don’t get that right, we won’t be able to reap the benefits of the EHRs.” Manasi Vaidya
VOLUME 20 | NUMBER 11 | NOVEMBER 2014 NATURE MEDICINE
