Life sciences converging with healthcare: Dinesh Kasthuril, Lead Manager, Clinical and Drug Safety, Partha Chakraborty, Head of Life Sciences Solutions, and Rick Kite, Practice Manager for Life sciences, from Cognizant Technology Solutions, review the need for pharmacovigilance and integrated safety platforms.
In a bid to improve the treatment of patients, the pharmaceutical industry is likely to move closer to the healthcare industry in the near future. The exchange of patient health information will aid the search for, and development of, new targeted therapies and improved post-marketing drug safety surveillance. Patient safety has been at the front of patients,' regulators' and pharmaceutical companies' minds in recent years. High profile withdrawals of drugs have led to increased scrutiny and to regulatory authorities raising the bar, and the sheer number of ADRs (Adverse Drug Reactions) reported to pharmaceutical companies and regulators has also increased, resulting in a need for better systems to effectively process large volumes of data.
Pharmaceutical companies are also increasingly turning to leverage external sources of information to make better and more informed decisions regarding the safety of their drugs. In this context, they are looking at patient health information to aid both in the development of new targeted therapies and in improving post-marketing drug safety surveillance. The intent of all these efforts is to proactively attempt to detect safety signals earlier in the product lifecycle, which we believe is the way forward and the future direction for drug safety. The US President's Information Technology Advisory Committee said: "The same EHR (Electronic Health Records) systems that are critical for improving patient care can also help to accelerate clinical research and its impact on practice, and improve pharmaceutical safety (pharmacovigilance) and biosurveillance for public health; that is, a dual use of EHR systems that could reduce total system costs." (1)
Proactive pharmacovigilance can be greatly facilitated by access to an integrated safety infrastructure that combines biomarker data, clinical trial data, post-marketing safety data and EMRs. Currently, safety scientists and epidemiologists have limited access to the individual data sources and rarely in an integrated manner. Currently, the reporting, processing and analysis of ADRs by physicians occurs on the basis of skeletal information about the patient, the symptom and the drug. In some cases, pharmaceutical companies attempt to retrieve the patient's heath history, which then involves a time consuming process of attempting to contact the patient's physician, accessing their medical history and then evaluating the drug related event in the light of the patient's history.
The ability to access a patient's EMR while reporting an ADR would enable the pharmaceutical companies' safety physicians to make the right decision early by providing them with access to the requisite information. This would improve not only the quality of data captured and time for assessment, but could also potentially positively impact the reporting rate, which has historically been very low. The ability to spend less time and effort in reporting an ADR could potentially drive up the reporting, which has historically been pegged at 10% of overall ADRs. The detection of patterns in drug reactions using signal detection algorithms run against an integrated data repository, which combines internal and external data, allows drug safety scientists to perform analyses hitherto not possible. Scientists can carefully select and target treatment therapies based on patient data and can also test hypotheses on potential signals. Large pharmaceutical companies and public heath consortiums are increasingly focusing on this area.
Companies have already integrated internal clinical systems with external data sources such as biomarker data, FDA AERS, clinical trial and post-marketing safety data. The next step in this evolution is the extension of this infrastructure to include patient health information. The best approach for achieving this integration is to use a shared data content labelling standard (like XML) to automatically transfer patient health information. Such a standard is universally acceptable, has practical feasibility, and is supported by enabling standards such as ICH E2B, CDSIC and HL7. The creation of such an integrated infrastructure is also within reach, and can be built using existing data warehousing and data mining technologies. Existing integrated safety platforms, such as the proprietary ASPIRE framework we have deployed for our clients, can be logically extended to include EMR data sources.
[FIGURE 1 OMITTED]
Integrated Safety Infrastructure
The following section lays out the outline of the integrated safety infrastructure based on our experience of implementing similar systems. Blueprint of the Integrated Safety Platform An integrated safety platform should consist of the following logical components (Figure 1):
* Staging Layer to clean and process data from multiple sources, including FDA AERS, EMR, regulators such as MHRA, etc. This component would receive and exchange data in an ICH-compliant XML format. The staging layer would allow data stewards to perform data reviews and take remedial measures to improve data quality.
* Repository Layer to store the integrated data. The data model for storage would follow CDISC ODM (or Janus) and SDTM structure to facilitate pooled analysis and regulatory submission.
* Distribution Layer to present the data to different sets of users, such as clinical scientists, statisticians and pharmaco-epidemiologists. Each of these users would need access to different slices of the data and would be provided access through an intuitive, easy-to-use interface.
The platform would offer the following key features:
* The ability to identify signals using statistical methods or non-statistical methods across multiple data sources: Bayesian Methods, PRR, OR, Reporting Rate, Syndrome, Trial Drug and Comparator.
* Be a collaborative and secure platform to share and delegate the signals.
* Techniques to reduce the false positive signals.
* Facility to record analysis of the signals.
The infrastructure can be built on current technologies that are available. The data repository can be built using any Relational Database or SAS 9.1.3. An intuitive user interface will need a Microsoft ASP.Net or a J2EE based implementation. In our experience, the best approach for combining the integrated safety data along with predefined signal detection criteria to enable the signal detection is as Figure 2. The platform is envisioned to allow the following:
* Integrating Clinical Safety Records (E2B) and Clinical Trial Data (CDISC).
* Safety Signal Detection and Signal Management System.
* Integrate Electronic Health Record (HL7) with Clinical Trial Records (CDISC).
[FIGURE 2 OMITTED]
Such data integration can increase patient safety in multiple ways, from increasing the quality of signal detection to enabling the quicker capture of adverse event data. Great advances in clinical research and drug safety surveillance would be possible through shared patient health information and an integrated drug safety infrastructure. Using this integrated architecture, healthcare professionals can more easily report adverse events using the patient data available in EMRs. This would significantly reduce the time required for processing and assessment of a single case by minimizing the need for follow-ups and ensuring a "first time right" policy.
Another business benefit of the creation of a combined safety infrastructure with preclinical, clinical and post-marketing data along with patient health records is that it would enable pharmaco-epidemiologists to test hypotheses and evaluate patterns of drug effects across a wide range of patients. Such a repository could even be shared with regulatory bodies. An example of this is integrating sales data into this safety infrastructure, which would allow the estimation of patient exposure and allow signal detection using patient exposure parameters. Integrating genotypic and phenotypic data into this safety infrastructure could help to make clinical research more effective. This data can be used to identify and validate biomarkers, which are gaining increasing importance in clinical research. It can also assist in patient stratification and better selection of patients for clinical trials, thus helping to reduce the cycle time and associated costs. It is therefore possible to foresee a scenario wherein integrated data can pave the way for targeted medicine--where patients can be prescribed medication with a higher probability of success, based on their EMRs and genetic data.
The Road Ahead
The biggest challenge to this future lies not from technology constraints but around defining policies to prevent misuse of patient data and to define consistent standards to allow interoperability.
One of the key challenges has been the quality and quantity of data available for signal detection. The safety signal is only as good as the underlying data. Physicians and other healthcare professionals need to be incentivized to increase the reporting of adverse events and to also report them with high quality supporting patient health information.
The ownership of the patient data and its governance is a key question that remains to be addressed. Would the ownership reside with pharmaceutical companies or with insurance companies ... or with hospitals or with other independent organizations? There needs to be strict controls around who can aggregate this data and to clearly identify and track the usage of this data. The potential for misuse of the data exists, which needs to be closely monitored to prevent this.
There is a need for regulatory acceptance to allow pharmaceutical companies and regulators to share adverse event data to allow them access to a rich set of data.
Making patient data anonymous does address some concerns regarding patient privacy. However, there needs to be a process for seeking explicit informed consent from patients (similar to clinical trials) to seek an individual's permission to have their health records included in the patient health information.
There is also a need to converge existing standards (CDISC, HL7) to facilitate the interchange of such information between systems. Currently, different source systems store data using different standards. Unless these various source systems support consistent terminology, the feasibility of truly integrating the data is remote.
(1.) The terms Electronic Health Records (EHR) and Electronic Medical Records (EMR) are used interchangeably in this article to refer to patient heath information in general. As per a 2003 IOM Patient Safety Report, these records refer to a longitudinal collection of electronic health information for and about persons and include immediate electronic access to persons and population level information by authorized persons.
For more information
London SW1Y 4SP, UK.
T +44 207 321 4888
About the Authors
Dinesh Kasthuril is a Lead in the Clinical and Drug Safety, Solutions and Consulting Group of the Life Sciences Practice of Cognizant. Dinesh has more than 10 years of experience of working with leading pharmaceuticals across the US, Europe and Asia. He has extensive experience in driving and delivering R&D strategy and business capability initiatives across the pharmaceutical product development value chain.
Partha Chakraborty is Head of Life Sciences Solutions and Innovation Champion of Cognizant Life Sciences Practice. He has worked with multiple large global pharmaceuticals in the US and led IT transformation in Clinical and Safety, as well as the implementation of critical engagements. He has architected ASPIRE, Cognizant's Drug Safety Framework.
Rick Kite is Cognizant's Practice Manager for Life Sciences in the UK. He is a senior member of the Cognizant team with more than 10 years of specialized experience in Information Management.