ABSTRACT
BACKGROUND: Birth defects are the leading cause of infant death. While causes of most are unknown, those that might be due to medication use are among the most preventable. This study describes an approach to identifying those medications that most warrant attention by using a "screen" program that calculates odds ratios for pairs of exposures and specific birth defects. METHODS: We discuss the development of this tool and illustrate its application to two large risk factor studies, the Slone Epidemiology Center's Birth Defects Study and the Centers for Disease Control and Prevention's National Birth Defects Prevention Study, ideal settings for the systematic study of risks and relative safety of drugs in relation to birth defects while recognizing the inherent limitations of such an approach. RESULTS: Suggestions for establishing criteria for exposures and outcomes that balance the need for specific details with the practical considerations of sample size and volume of output are presented. Selection of appropriate exposure reference categories and control groups is also discussed, as well as the need to address potential confounding. An example that motivated a detailed investigation of possible associations between a medication (butalbital) and selected specific birth defects is provided. CONCLUSION: While screening programs such as the one described can be a valuable tool for exploring potential associations in large data bases, they must be applied with caution. The issue of multiple testing and chance findings is a major concern. While statistics are a necessary component, human judgment must be an integral part of the process.
Subject(s)
Abnormalities, Drug-Induced/etiology , Congenital Abnormalities/etiology , Databases, Factual , Pharmaceutical Preparations/administration & dosage , Population Surveillance , Abnormalities, Drug-Induced/epidemiology , Abnormalities, Drug-Induced/prevention & control , Case-Control Studies , Centers for Disease Control and Prevention, U.S. , Congenital Abnormalities/epidemiology , Congenital Abnormalities/prevention & control , Female , Humans , Pregnancy , Risk Factors , United States/epidemiologyABSTRACT
There has been a substantial rise in the utilization of large databases in radiation oncology research. The advantages of these datasets include a large sample size and inclusion of a diverse population of patients in a real-world setting. Such observational studies hold promise in enhancing our understanding of questions for which evidence is conflicting or absent in lung cancer radiotherapy. However, it is critical that investigators understand the strengths and limitations of large databases in order to avoid the common pitfalls that beset observational analyses. This review begins by outlining the data variables available in major registries that are used most often in observational analyses. This is followed by a discussion of the type of radiotherapy-related questions that can be addressed using such datasets, accompanied by examples from the lung cancer literature. Finally, we describe some limitations of observational research and techniques to mitigate bias and confounding. We hope that clinicians and researchers find this review helpful for designing new research studies and interpreting published analyses in the literature.
ABSTRACT
BACKGROUND: Large databases provide an efficient way to analyze patient data. A challenge with these databases is the inconsistency of ICD codes and a potential for inaccurate ascertainment of cases. The purpose of this study was to develop and validate a reliable protocol to identify cases of acute ischemic stroke (AIS) from a large national database. METHODS: Using the national Veterans Affairs electronic health-record system, Center for Medicare and Medicaid Services, and National Death Index data, we developed an algorithm to identify cases of AIS. Using a combination of inpatient and outpatient ICD9 codes, we selected cases of AIS and controls from 1992 to 2014. Diagnoses determined after medical-chart review were considered the gold standard. We used a machine-learning algorithm and a neural network approach to identify AIS from ICD9 codes and electronic health-record information and compared it with a previous rule-based stroke-classification algorithm. RESULTS: We reviewed administrative hospital data, ICD9 codes, and medical records of 268 patients in detail. Compared with the gold standard, this AIS algorithm had a sensitivity of 91%, specificity of 95%, and positive predictive value of 88%. A total of 80,508 highly likely cases of AIS were identified using the algorithm in the Veterans Affairs national cardiovascular disease-risk cohort (n=2,114,458). CONCLUSION: Our algorithm had high specificity for identifying AIS in a nationwide electronic health-record system. This approach may be utilized in other electronic health databases to accurately identify patients with AIS.
ABSTRACT
Comparative effectiveness research (CER) is a concept initiated by the Institute of Medicine and financially supported by the federal government. The primary objective of CER is to improve decision making in medicine. This research is intended to evaluate the effectiveness, benefits, and harmful effects of alternative interventions. CER studies are commonly large, simple, observational, and conducted using electronic databases. To date, there is little comparative effectiveness evidence within hand surgery to guide therapeutic decisions. To draw conclusions on effectiveness through electronic health records, databases must contain clinical information and outcomes relevant to hand surgery interventions, such as patient-related outcomes.
Subject(s)
Comparative Effectiveness Research/organization & administration , Databases, Factual , Hand/surgery , Patient Outcome Assessment , Humans , United StatesABSTRACT
People generally prefer their initials to the other letters of the alphabet, a phenomenon known as the name-letter effect. This effect, researchers have argued, makes people move to certain cities, buy particular brands of consumer products, and choose particular professions (e.g., Angela moves to Los Angeles, Phil buys a Philips TV, and Dennis becomes a dentist). In order to establish such associations between people's initials and their behavior, researchers typically carry out statistical analyses of large databases. Current methods of analysis ignore the hierarchical structure of the data, do not naturally handle order-restrictions, and are fundamentally incapable of confirming the null hypothesis. Here we outline a Bayesian hierarchical analysis that avoids these limitations and allows coherent inference both on the level of the individual and on the level of the group. To illustrate our method, we re-analyze two data sets that address the question of whether people are disproportionately likely to live in cities that resemble their name.