Enhance the FDA's Postmarketing Drug Surveillance System of Adverse Effects through the Electronic Healthcare Data

In the United States, a medication or drug must be approved by the Food and Drug Administration (FDA), a Federal Agency of the United States Department of Health and Human Services, before marketing (open to the public use). The FDA has a systematic premarketing risk identification and assessment system and makes decisions to approve or not approve a drug to treat a specific condition, based on a benefit-risk analysis for the intended population [1-3]. Once approved, drugs move swiftly into the market place for use by prescribers and patients. The FDA also has a postmarketing drug surveillance and risk assessment system with intention to identify unexpected short- and long-term risks of approved drugs or products, which currently relies primarily on two methods of adverse event reporting to the FDA [1-12]. One is the direct, voluntary reporting by health professionals and consumers. The other is a mandated reporting by pharmaceutical manufacturers that is mainly based on the voluntary submission of reports from medical facilities, health professionals, and consumers, so-called MedWatch program [1-7]. The spontaneous case reports from health professionals like the MedWatch program are important, but may miss over 90% of all cases with rare toxic events [5-7].

The inadequacies of the FDA’s current postmarketing surveillance system have been addressed by the Editors of the Journal of American Medical Association (JAMA) in their editorial [13], including the underreporting of adverse outcomes with capture of only a small fraction of adverse events and difficulty in calculating true rates of adverse events because of incomplete numerator data on exposures together with unreliable denominator data on exposure. Clearly, the FDA’s postmarketing drug surveillance system should be improved [13-20]. One way to enhance the postmarketing surveillance system would be to establish regional centers to be responsible for monitoring new drugs that come on the market [17]. Agency for Healthcare Research and Quality (AHRQ) has awarded grants to support seven Centers for Education and Research on Therapeutics (CERTs) since 1999, in consultation with the FDA and agencies with the U.S. Department of Health and Human Services. A study from one of CERT centers recently concluded that there is no systematic coherent approach for evaluating the long-term safety of drugs once the drugs have been approved for marketing by the FDA [16]. This finding was supported by studies from other investigators [13,17]. A key aspect of postmarketing surveillance now should involve the use of electronic medical records [17]. In fact, the FDA has taken actions to strengthen the postmarketing surveillance of drug safety [18] and has been considering expanding its postmarketing drug surveillance and risk assessment programs, such as increasing access to the hospital or laboratory-based computerized data systems or to the databases maintained by Health Maintenance Organizations (HMOs) [3]. However, to our knowledge, no studies have previously reported using the nationwide, population-based computerized Medicare claims databases for achieving the goal of more complete information on postmarketing drug surveillance of adverse effects. Therefore, we took initiatives with funding from the Agency for Healthcare Research and Quality (AHRQ) to use the nationwide and population-based electronic Medicare claims data to identify a series of chemotherapy-related short- and long-term adverse effects or toxicities [21-26].

For example, we studied 19,478 women aged 65 years or older diagnosed with breast cancer in 1991-1997 from 16 regions in the Surveillance, Epidemiology and End Results program with followup to December 2005 [21]. The excess cumulative incidence of congestive heart failure in the year 2010, among patients receiving anthracycline-containing chemotherapy compared to patients without chemotherapy was 4.7% (31.9% versus 27.2%). After adjusting for patient and tumor characteristics, patients receiving anthracyclines were 25% more likely to have congestive heart failure compared to those without chemotherapy (hazard ratio=1.25, 95% Confidence Interval=1.07-1.46). Those receiving other agents did not have a significantly elevated risk of developing congestive heart failure. The adjusted risk of cardiomyopathy was two-fold higher in women who received anthracyclines (1.95, 1.44-2.62) and was 16% higher in those receiving other agents (1.16, 0.97-1.39) compared to those without chemotherapy. The increased risk of developing congestive heart failure, cardiomyopathy and cardiac dysrhythmias in association with chemotherapy were similar after adjusting for hypertension and diabetes. The risk of ischemic heart disease and conduction disorders were not significantly different among the three groups. This study concluded that anthracycline-containing chemotherapy regimens were associated with an increased risk of congestive heart failure, cardiomyopathy and cardiac dysrhythmias, but not significantly associated with ischemic heart disease or conduction disorders. In another example, we found no significant association between chemotherapy and the risk of developing Alzheimer’s disease and vascular dementia [25]. This study was conducted because several small scale clinical trials indicated a possible relationship between chemotherapy administration and the increased risk of cognitive impairments in patients with breast cancer, but little information was available from large population-based cohort studies with long followup time. We studied 62,565 women who were diagnosed with stages I-IV of breast cancer at age 65 years or older from 1991 through 2002 who were free of cognitive impairments at diagnosis with up to 16 years of follow-up, and also studied 9,752 matched cohort based on the propensity of receiving chemotherapy. Overall, patients who received chemotherapy were 8% more likely to have drug-induced dementia compared to those without chemotherapy, but that was not statistically significant after adjusting for patient and tumor characteristics (hazard ratio=1.08, 95% confidence interval=0.85-1.37). The risk of developing Alzheimer’s disease, vascular dementia, or other dementias was significantly lower in patients receiving chemotherapy except for cognitive disorder which was not significantly different between the two chemotherapy groups. The results were somewhat similar between the entire cohort and the matched cohort based on the probability of receiving chemotherapy. This study with long-term follow-up did not support the findings that chemotherapy was associated with an increased risk of late stage cognitive impairments (such as Alzheimer’s disease and vascular dementia).

We also used the electronic Medicare claims data to identify known or unknown short- and long-term toxicities in 41,361 men and 30,804 women who were diagnosed with Non-Small Cell Lung Cancer (NSCLC) at age 65 years or older [26]. The incidences of 50 toxicityassociated end points were calculated for 14 chemotherapy agents. The most common short-term toxicities (9.2%~60.0%) included acute anemia, nausea, and neutropenia. The most common longterm toxicities (15%~37%) included acute anemia, respiratory failure, pulmonary fibrosis, dehydration, neutropenia, nausea, and fever. Longterm toxicities increased over time and were more likely in women, minority populations, those with fewer baseline co-morbidities, and across disease stages. The study concluded that the administration of various chemotherapy agents for lung cancer was associated with a number of short- and long-term toxicities.

The use of any medication has the possibility of unintended consequences or toxicities, which include predictable and unpredictable side effects. To assess the frequency and severity of toxicity resulting from the drug therapies, randomized clinical trials with adequate sample sizes offer a unique opportunity because these trials have the most comprehensive and systematic accumulation of pertinent information in a controlled setting. However, clinical trials have strict enrollment criteria, often excluding the elderly, pregnant women, patients with multiple coexisting diseases, and those taking medications suspected of interacting with the study drug. Therefore, study participants (motivated volunteers) in clinical trials represent a much more homogeneous patient population in terms of gender, age, race/ethnicity, and co-morbidity, and are not representative of the entire population. In addition, clinical trials are often conducted in specialist centers under the “ideal world” conditions with close monitoring/ follow-up and timely control of the side effects of the drug. This may result in earlier recognition of toxicity and avoidance of subsequent hospitalization. As a result, substantial differences in toxicity rates between clinical trial settings and routine community practice settings would be expected. Furthermore, clinical trials may not be able to detect rare or long-term drug toxicities because of their small sample sizes and short follow-up periods. Hence, for complete information on drug toxicity, supplementing the data from premarketing clinical trials with data from the postmarketing drug surveillance would produce the best estimates.

Postmarketing drug surveillance is critically important for detecting rare or long-term drug toxicities. Currently, there is no systematic approach for evaluating long-term toxicity of the marketed drugs and none of the existing methods for identifying postmarketing drug toxicity is sufficient on its own [3-20]. The medical chart review is effective for this research but is too costly for routine clinical monitoring. The innovative use of the nationwide, population-based computerized Medicare claims data offers a promising and less expensive way of obtaining this information on postmarketing drug adverse-effects or toxicities. This database is especially unique for studying cancer chemotherapy and its toxicities because chemotherapy is among the few drugs that are covered by the Medicare program for past several decades. Since January 1, 2006 when the Medicare Part D was implemented for comprehensive drug coverage, it claims data should offer more benefits for the postmarketing surveillance system in identifying adverse effects or side effects related to increasingly large numbers of drugs or medications. Moreover, these claims data, together with other electronic databases, can offer the really ‘up to date’ and ‘real time’ data and sensitive information associated with the marketed drugs. The process may require the routine automated and computer-generated information frequently on the types of drugs and potentially associated adverse effects or toxicities through the SAS macro programs consisting of all potential diagnosis and procedure codes. The process may also require a group of specialists to run the analyses on a regular basis to monitor the patterns of adverse effects and to alert any unusual patterns to the appropriate agencies or offices for further actions, including FDAs. The programs or analyses can be performed at the Federal or state level healthcare agencies. All health care providers, health insurance companies, and pharmacies with electronic data at national and local levels should be encouraged or even required to report these analyses on a regular basis. By doing so, the FDA’s postmarketing drug surveillance system of adverse effects or toxicities will be significantly enhanced through the electronic healthcare data.

This study was supported by a grant entitled “Postmarketing Surveillance of Toxicities Associated with Cancer Chemotherapy” from the Agency for Healthcare Research and Quality (R01-HS016743).