Data Sources for Postmarketing Pediatric Adverse Drug Events

DATA SOURCES FOR POSTMARKETING PEDIATRIC ADVERSE DRUG EVENTS

The main source of pediatric pharmacovigilance data is spontaneous reports which are compiled by vari-ous regulatory agencies, the largest system being the database maintained by the FDA. The FDA Adverse Event Reporting System (AERS) is a computerized information database designed to support the FDA’s postmarketing safety surveillance program for all approved drugs and therapeutic biologic products. It is comprised of mandatory reports from manufacturers as required by regulation and voluntary reports from health-care professionals and consumers through the MedWatch program. FDA codes all reported adverse events using a standardized international terminol-ogy, MedDRA (the Medical Dictionary for Regula-tory Activities). FDA staff use reports from AERS in conducting postmarketing drug surveillance and compliance activities and in responding to outside requests for information.

The reports in AERS are evaluated by FDA staff to monitor drug safety and detect new safety signals.

Results of further evaluation of the signals may form the basis for regulatory action to improve product safety and protect the public health, such as updating a product’s labelling information, sending out a ‘Dear Health Care Professional’ letter, or re-evaluating an approval decision.

The limitations of voluntary and spontaneous drug adverse event reporting systems are well recognized and adequately described elsewhere, but there are some unique aspects of use of drugs in pediatrics that contribute to these limitations, especially to under-reporting in children. First, health-care profession-als may be less likely to report suspected adverse reactions for drugs that are unlicensed or used off-label. Second, the practice of polyphar-macy in the premature and sick neonate, often with unapproved and off-label drugs adds another level of complexity that further hampers recognition and reporting by health-care practitioners and consumers. Third, adverse drug reactions arising from in-utero or breast milk exposures and manifesting during the neonatal period may be underreported because mater-nal history of pregnancy drug exposures are poorly documented or their potential contribution to neona-tal problems are under-appreciated. Fourth, delayed adverse drug reactions, especially those with a long latency such as effects on growth, development and cognition are less likely to be recognized and reported. Fifth, possible drug adverse reactions may not be recognized because young children may be unable to describe their symptoms, and intermediaries such as parents and caretakers may fail to report them. The result is relatively few pediatric reports enter-ing AERS and a longer time period needed to build an adequate case series of postmarketing reports to perform a pediatric safety analysis.

The premarket evidence base for pediatric drug safety is non-existent for most approved drugs because few clinical drug development programs have included pediatric patients. Despite the absence of data to guide dosage, dosage frequency, route of administration or appropriate formulation and evidence of clinical efficacy or safety, drugs are commonly prescribed off-label to pediatric patients. These off-label uses constitute the collection of ‘N of One trials’, outcomes of which become the single most important source of information of adverse drug reactions in pediatric patients and to a lesser extent information on the drug’s benefits.

In recent years, US legislation and financial incen-tives to sponsors have led to increased clinical drug studies in children. These studies have resulted in useful data on a drug’s pharmacokinetics, safety and efficacy for pediatric labelling. However, these stud-ies are often short in duration, include a small sample of selected patients and are typically not powered for safety. Therefore, postmarket reporting of adverse events still continues to be the primary source of safety data for pediatrics. But there is increasing inter-est in going beyond the postmarketing spontaneous adverse event reports to assess the safety and effec-tiveness of drugs in the pediatric population.

Although not a requirement yet, sponsors can play a significant role in postmarketing surveillance by submitting periodic safety update reports (PSUR) after a new drug’s approval for marketing. A separate and detailed safety analysis focused on pediatrics is currently a feature of the PSUR. The PSUR was designed to be a stand-alone document that allows a periodic but comprehensive assessment of the world-wide safety data of a marketed drug or biological product. The PSUR can be an important source of data for the identification of new safety signals, a means of determining changes in the benefit–risk profile, an effective means of risk communication to regu-latory authorities, and an indicator for the need for risk management initiatives. Incorporating a require-ment for a separate pediatric safety analysis as part of the PSUR submitted by sponsors will facilitate early detection and evaluation of possible safety issues.

In the United States, postmarketing drug adverse events surveillance data are available from sources other than spontaneous reporting systems such as emergency department-based systems and epidemi-ologic data from automated claims databases. The National Electronic Injury Surveillance System (NEISS), which collects data on all injuries from a probability sample of emergency departments (ED) in approximately 100 hospitals, recently evaluated an active drug adverse event surveillance program using ED chart reviews in six sites.¹⁹ The results indicated that although the predictive value positive for Adverse Drug Reactions (ADRs) was high, sensitivity was low suggesting the need for additional training of review-ers and coders in the recognition and reporting of suspected ADRs.

Another source of surveillance information is the Drug Abuse Warning Network (DAWN), whose data-collection procedures were modified in 2003 to include adverse drug reactions. This system collects data from a probability sample of short-term, general, non-federal hospitals and from medical examin-ers/coroners in 300 jurisdictions in 48 metropolitan areas, and it collects data on any drug-related visit irrespective of intent including drug abuse, misuse, overmedication, intentional/accidental ingestion, and drug adverse reaction. DAWN and other drug adverse reaction data sources such as the Toxic Exposure Surveillance System (TESS) run by the American Association of Poison Control Centers may benefit from an assessment similar to that done for NEISS with a focus on pediatric adverse drug events. All of the above systems have significant shortcomings (i.e., only severe cases are captured in EDs, claims databases collect information only from hospitalized patients enrolled in a particular health plan) and can only serve as a complement to existing postmarketing drug adverse event data systems.

Population-based, computerized administrative health databases linked to drug utilization data and outcomes have been gaining popularity in evalua-tion of drug safety. Among these, the largest and best known internationally is the General Prac-tice Research Database (GPRD) maintained by the Medicines and Healthcare products Regulatory Agency (MHRA).²⁰ GPRD is a longitudinal database that collects data on patient demographics, prescrip-tion drug use, diagnosis, treatment outcomes, and laboratory tests from a voluntary group of general practitioners who provide primary health care via the National Health Service throughout the United King-dom. Although GPRD has been used less in drug safety research in pediatrics than in adults, it has proven to be useful in the assessment of safety signals, drug usage patterns, quantification of population risk of drugs including those of rare outcomes. Because the data are collected prospectively and are longitudinally linked, GPRD can particularly be useful in evaluating pediatric drug adverse effects with long latency such as adverse effects on growth, cognitive development and neoplasia.

Other population-based, computerized databases commonly used for pharmacovigilance are organized at regional or health-care setting level. Examples include the Saskatchewan Health Database²¹ that contain linked data on prescription drug, hospital services, physician services and vital statistics for all residents in one province in Canada. Examples of the health care setting-based databases are TennCare (state-based Medicaid program), the Kaiser Foun-dation database and the Harvard Pilgrim database (health maintenance organizations) who were awarded contracts for drug safety research by the FDA in 2005. Important limitations of these databases include the inability to study rare drug adverse events due to the small population size, and inability to study effects of newly marketed drugs due to formulary restrictions. These databases, although designed for administrative purposes, offer many opportunities for pharmacoepi-demiology but remain underutilized for pediatric-specific drug safety evaluation and research.

The potential to identify and report suspected drug adverse events can greatly be enhanced by the imple-mentation of the requirement for electronic medical records for all patients. Electronic medical records can also be useful in preventing serious adverse events by incorporating automated reminders about previous drug reaction history, drug–drug and drug–food inter-actions, dosage adjustments and new safety alerts. Until an electronic medical record for all patients becomes a reality, postmarketing safety assessments will have to employ one or more of the available resources described above.