Research Using Emergency Department–related Data Sets: Current Status and Future Directions

The 2009 Academic Emergency Medicine consensus conference focused on "Public Health in the ED: Surveillance, Screening and Intervention." One conference breakout session discussed the significant research value of health-related data sets. This article represents the proceedings from that session, primarily focusing on emergency department (ED)-related data sets and includes examples of the use of a data set based on ED visits for research purposes. It discusses types of ED-related data sets available, highlights barriers to research use of ED-related data sets, and notes limitations of these data sets. The paper highlights future directions and challenges to using these important sources of data for research, including identification of five main needs related to enhancing the use of ED-related data sets. These are 1) electronic linkage of initial and follow-up ED visits and linkage of information about ED visits to other outcomes, including costs of care, while maintaining deidentification of the data; 2) timely data access with minimal barriers; 3) complete data collection for clinically relevant and/or historical data elements, such as the external cause-of-injury code; 4) easy access to data that can be parsed into smaller jurisdictions (such as states) for policy and/or research purposes, while maintaining confidentiality; and 5) linkages between health survey data and health claims data. ED-related data sets contain much data collected directly from health care facilities, individual patient records, and multiple other sources that have significant potential impact for studying and improving the health of individuals and the population.     

Assessing Data Adequacy for High Blood Pressure Clinical Decision Support: A Quantitative Analysis

Objective  This study examines guideline-based high blood pressure (HBP) and hypertension recommendations and evaluates the suitability and adequacy of the data and logic required for a Fast Healthcare Interoperable Resources (FHIR)-based, patient-facing clinical decision support (CDS) HBP application. HBP is a major predictor of adverse health events, including stroke, myocardial infarction, and kidney disease. Multiple guidelines recommend interventions to lower blood pressure, but implementation requires patient-centered approaches, including patient-facing CDS tools. Methods  We defined concept sets needed to measure adherence to 71 recommendations drawn from eight HBP guidelines. We measured data quality for these concepts for two cohorts (HBP screening and HBP diagnosed) from electronic health record (EHR) data, including four use cases (screening, nonpharmacologic interventions, pharmacologic interventions, and adverse events) for CDS. Results  We identified 102,443 people with diagnosed and 58,990 with undiagnosed HBP. We found that 21/35 (60%) of required concept sets were unused or inaccurate, with only 259 (25.3%) of 1,101 codes used. Use cases showed high inclusion (0.9–11.2%), low exclusion (0–0.1%), and missing patient-specific context (up to 65.6%), leading to data in 2/4 use cases being insufficient for accurate alerting. Discussion  Data quality from the EHR required to implement recommendations for HBP is highly inconsistent, reflecting a fragmented health care system and incomplete implementation of standard terminologies and workflows. Although imperfect, data were deemed adequate for two test use cases. Conclusion  Current data quality allows for further development of patient-facing FHIR HBP tools, but extensive validation and testing is required to assure precision and avoid unintended consequences.     

Detection,Prevention, and Management of Extrapyramidal Symptoms

Extrapyramidal symptoms (EPSs), such as akathisia, dystonia, psuedoparkinsonism, and dyskinesia, are drug-induced side effects that can be problematic for persons who receive antipsychotic medications (APMs) or other dopamine-blocking agents. The clinical manifestations include a number of atypical involuntary muscle contractions that influence gait, movement, and posture. The symptoms can develop acutely, be delayed, or overlap making diagnosing a challenge. Preventive interventions include selective prescribing of APMs, close monitoring of uncharacteristic movements through the use of screening instruments, prompt management of symptoms, and thorough client education. Nurse practitioners who do not practice in psychiatric mental health nursing on a regular basis or who infrequently prescribe psychotropic medications must be cautious with these potential life-threatening symptoms.     

The utility of synthetic and regression estimation. Techniques for local health planning

Health planning agencies have long had a need for local area data on issues such as health status, utilization, and preventive care. Synthetic estimation and regression techniques have been advanced for use in adjusting national or regional data from the NCHS Health Interview Survey (HIS) to the local area based on demographic and other pertinent variables. To evaluate the accuracy and utility of these techniques for local health planning, estimates of certain key HIS variables were obtained from a telephone survey of 2,500 randomly selected households located in the Central Maryland Health Systems Agency (CMHSA). This paper presents results of the comparison between these telephone estimates and the corresponding estimates derived from the national HIS. The techniques are evaluated on the basis of data generally available to local planning agencies.     

Performance management using health outcomes: in search of instrumentality

The past decade has seen the growing use of performance data in the hope of bringing about quality improvements in health care. Most recently, an emphasis on health outcomes (and especially mortality rates) has led to much activity around collecting and publishing such data. Two major problems intervene. What meanings can be ascribed to reported health outcomes? And what impacts are they likely to have on clinical performance? Much empirical work supports the assertion that reported outcomes may be poor indicators of service quality. In addition, the impact of these data may be small or even detrimental unless great care is made to connect the reporting with explicit quality-improving actions.     

Measures That Identify Prescription Medication Misuse,Abuse, and Related Events in Clinical Trials: ACTTION Critique and Recommended Considerations

Accurate assessment of inappropriate medication use events (ie, misuse, abuse, and related events) occurring in clinical trials is an important component in evaluating a medication's abuse potential. A meeting was convened to review all instruments measuring such events in clinical trials according to previously published standardized terminology and definitions. Only 2 approaches have been reported that are specifically designed to identify and classify misuse, abuse, and related events occurring in clinical trials, rather than to measure an individual's risk of using a medication inappropriately: the Self-Reported Misuse, Abuse, and Diversion (SR-MAD) instrument and the Misuse, Abuse, and Diversion Drug Event Reporting System (MADDERS). The conceptual basis, strengths, and limitations of these methods are discussed. To our knowledge, MADDERS is the only system available to comprehensively evaluate inappropriate medication use events prospectively to determine the underlying intent. MADDERS can also be applied retrospectively to completed trial data. SR-MAD can be used prospectively; additional development may be required to standardize its implementation and fully appraise the intent of inappropriate use events. Additional research is needed to further demonstrate the validity and utility of MADDERS as well as SR-MAD.

When biotoxins are tools of terror. Early recognition of intentional poisoning can attenuate effects

Toxin-mediated diseases have made humans ill for millennia. They also have been used in beneficial ways. Unfortunately, the use of biological agents as weapons of terror has now been realized, and separating naturally occurring disease from bioterroristic events has become an important public health goal. The key to timely identification of such attacks relies on education of primary care physicians, first responders, and public health officials. We must remain vigilant to unusual case presentations or clusters of similar cases and report them immediately to public health authorities.     

Mass gathering medicine: a predictive model for patient presentation and transport rates

INTRODUCTION: This paper reports on research into the influence of environmental factors (including crowd size, temperature, humidity, and venue type) on the number of patients and the patient problems presenting to first-aid services at large, public events in Australia. Regression models were developed to predict rates of patient presentation and of transportation-to-a-hospital for future mass gatherings. OBJECTIVE: To develop a data set and predictive model that can be applied across venues and types of mass gathering events that is not venue or event specific. Data collected will allow informed event planning for future mass gatherings for which health care services are required. METHODS: Mass gatherings were defined as public events attended by in excess of 25,000 people. Over a period of 12 months, 201 mass gatherings attended by a combined audience in excess of 12 million people were surveyed throughout Australia. The survey was undertaken by St. John Ambulance Australia personnel. The researchers collected data on the incidence and type of patients presenting for treatment and on the environmental factors that may influence these presentations. A standard reporting format and definition of event geography was employed to overcome the event-specific nature of many previous surveys. RESULTS: There are 11,956 patients in the sample. The patient presentation rate across all event types was 0.992/1,000 attendees, and the transportation-to-hospital rate was 0.027/1,000 persons in attendance. The rates of patient presentations declined slightly as crowd sizes increased. The weather (particularly the relative humidity) was related positively to an increase in the rates of presentations. Other factors that influenced the number and type of patients presenting were the mobility of the crowd, the availability of alcohol, the event being enclosed by a boundary, and the number of patient-care personnel on duty. Three regression models were developed to predict presentation rates at future events. CONCLUSIONS: Several features of the event environment influence patient presentation rates, and that the prediction of patient load at these events is complex and multifactorial. The use of regression modeling and close attention to existing historical data for an event can improve planning and the provision of health care services at mass gatherings.     

Linking Provider Specialty and Outpatient Diagnoses in Medicare Claims Data: Data Quality Implications

Background  With increasing use of real world data in observational health care research, data quality assessment of these data is equally gaining in importance. Electronic health record (EHR) or claims datasets can differ significantly in the spectrum of care covered by the data. Objective  In our study, we link provider specialty with diagnoses (encoded in International Classification of Diseases) with a motivation to characterize data completeness. Methods  We develop a set of measures that determine diagnostic span of a specialty (how many distinct diagnosis codes are generated by a specialty) and specialty span of a diagnosis (how many specialties diagnose a given condition). We also analyze ranked lists for both measures. As use case, we apply these measures to outpatient Medicare claims data from 2016 (3.5 billion diagnosis–specialty pairs). We analyze 82 distinct specialties present in Medicare claims (using Medicare list of specialties derived from level III Healthcare Provider Taxonomy Codes). Results  A typical specialty diagnoses on average 4,046 distinct diagnosis codes. It can range from 33 codes for medical toxicology to 25,475 codes for internal medicine. Specialties with large visit volume tend to have large diagnostic span. Median specialty span of a diagnosis code is 8 specialties with a range from 1 to 82 specialties. In total, 13.5% of all observed diagnoses are generated exclusively by a single specialty. Quantitative cumulative rankings reveal that some diagnosis codes can be dominated by few specialties. Using such diagnoses in cohort or outcome definitions may thus be vulnerable to incomplete specialty coverage of a given dataset. Conclusion  We propose specialty fingerprinting as a method to assess data completeness component of data quality. Datasets covering a full spectrum of care can be used to generate reference benchmark data that can quantify relative importance of a specialty in constructing diagnostic history elements of computable phenotype definitions.     

Reliability of homeless women's reports: concordance between hair assay and self report of cocaine use

BACKGROUND: To assess the concordance of homeless women's self-reported drug use with objective data. OBJECTIVE: To determine whether objective data (e.g., hair assays) are necessary supplements to self reports in assessing homeless women's cocaine use. METHOD: Self reports of cocaine use by 1,037 homeless women were compared to objective data based on radioimmunoassay of hair; independent correlates of cocaine use and underreporting were assessed using logistic regression. RESULTS: Forty-two percent of the women self-reported cocaine use in the past 6 months, whereas 49% had positive hair assays. Over 25% underreported cocaine use; however, underreporting decreased as hair cocaine levels increased. Predictors of underreporting included being Latino, younger and living primarily in shelters. Nevertheless, independent predictors of self-reported cocaine use and positive hair assays were identical. CONCLUSION: Homeless women's self-reports of cocaine use are fairly accurate. Objective data are particularly critical for assessing cocaine use among subgroups who are fearful of sanctions or use cocaine relatively infrequently or in smaller amounts.     

A Standard System to Study Vertebrate Embryos

Staged embryonic series are important as reference for different kinds of biological studies. I summarise problems that occur when using ‘staging tables’ of ‘model organisms’. Investigations of developmental processes in a broad scope of taxa are becoming commonplace. Beginning in the 1990s, methods were developed to quantify and analyse developmental events in a phylogenetic framework. The algorithms associated with these methods are still under development, mainly due to difficulties of using non-independent characters. Nevertheless, the principle of comparing clearly defined newly occurring morphological features in development (events) in quantifying analyses was a key innovation for comparative embryonic research. Up to date no standard was set for how to define such events in a comparative approach. As a case study I compared the external development of 23 land vertebrate species with a focus on turtles, mainly based on reference staging tables. I excluded all the characters that are only identical for a particular species or general features that were only analysed in a few species. Based on these comparisons I defined 104 developmental characters that are common either for all vertebrates (61 characters), gnathostomes (26), tetrapods (3), amniotes (7), or only for sauropsids (7). Characters concern the neural tube, somite, ear, eye, limb, maxillary and mandibular process, pharyngeal arch, eyelid or carapace development. I present an illustrated guide listing all the defined events. This guide can be used for describing developmental series of any vertebrate species or for documenting specimen variability of a particular species. The guide incorporates drawings and photographs as well as consideration of species identifying developmental features such as colouration. The simple character-code of the guide is extendable to further characters pertaining to external and internal morphological, physiological, genetic or molecular development, and also for other vertebrate groups not examined here, such as Chondrichthyes or Actinopterygii. An online database to type in developmental events for different stages and species could be a basis for further studies in comparative embryology. By documenting developmental events with the standard code, sequence heterochrony studies (i.e. Parsimov) and studies on variability can use this broad comparative data set.     

Transient lower esophageal sphincter relaxation pharmacokinetic-pharmacodynamic modeling: count model and repeated time-to-event model

Transient lower esophageal sphincter relaxation (TLESR) is the major mechanism for gastroesophageal reflux. Characterizations of candidate compounds for reduction of TLESRs are traditionally done through summary exposure and response measures and would benefit from model-based analyses of exposure-TLESR events relationships. Pharmacokinetic (PK)-pharmacodynamic (PD) modeling approaches treating TLESRs either as count data or repeated time-to-event (RTTE) data were developed and compared in terms of their ability to characterize system and drug characteristics. Vehicle data comprising 294 TLESR events were collected from nine dogs. Compound [(R)-(+)-[2,3-dihydro-5-methyl-3-(4-morpholinylmethyl)pyrrolo[1,2,3-de]-1,4-benzoxazin-6-yl]-1-naphthalenylmethanone mesylate (WIN55212-2)] data containing 66 TLESR events, as well as plasma concentrations, were obtained from four dogs. Each experiment lasted for 45 min and was initiated with a meal. Counts in equispaced 5- and 1-min intervals were modeled based on a Poisson probability distribution model. TLESR events were analyzed with the RTTE model. The PK was connected to the PD with a one-compartment model. Vehicle data were described by a baseline and a surge function; the surge peak was determined to be approximately 9.69 min by all approaches, and its width in time at half-maximal intensity was 5 min (1-min count and RTTE) or 10 min (5-min count). TLESR inhibition by WIN55212-2 was described by an I(max) model, with an IC(50) of on average 2.39 nmol · l(-1). Modeling approaches using count or RTTE data linked to a dynamic PK-PD representation of exposure are superior to using summary PK and PD measures and are associated with a higher power for detecting a statistically significant drug effect.     

Advanced Statistics:Statistical Methods for Analyzing Cluster and Cluster-randomized Data

Sometimes interventions in randomized clinical trials are not allocated to individual patients, but rather to patients in groups. This is called cluster allocation, or cluster randomization, and is particularly common in health services research. Similarly, in some types of observational studies, patients (or observations) are found in naturally occurring groups, such as neighborhoods. In either situation, observations within a cluster tend to be more alike than observations selected entirely at random. This violates the assumption of independence that is at the heart of common methods of statistical estimation and hypothesis testing. Failure to account for the dependence between individual observations and the cluster to which they belong can have profound implications on the design and analysis of such studies. Their p-values will be too small, confidence intervals too narrow, and sample size estimates too small, sometimes to a dramatic degree. This problem is similar to that caused by the more familiar "unit of analysis error" seen when observations are repeated on the same subjects, but are treated as independent. The purpose of this paper is to provide an introduction to the problem of clustered data in clinical research. It provides guidance and examples of methods for analyzing clustered data and calculating sample sizes when planning studies. The article concludes with some general comments on statistical software for cluster data and principles for planning, analyzing, and presenting such studies.     

Using data mining techniques to characterize participation in observational studies

Data mining techniques are gaining in popularity among health researchers for an array of purposes, such as improving diagnostic accuracy, identifying high‐risk patients and extracting concepts from unstructured data. In this paper, we describe how these techniques can be applied to another area in the health research domain: identifying characteristics of individuals who do and do not choose to participate in observational studies. In contrast to randomized studies where individuals have no control over their treatment assignment, participants in observational studies self‐select into the treatment arm and therefore have the potential to differ in their characteristics from those who elect not to participate. These differences may explain part, or all, of the difference in the observed outcome, making it crucial to assess whether there is differential participation based on observed characteristics. As compared to traditional approaches to this assessment, data mining offers a more precise understanding of these differences. To describe and illustrate the application of data mining in this domain, we use data from a primary care‐based medical home pilot programme and compare the performance of commonly used classification approaches – logistic regression, support vector machines, random forests and classification tree analysis (CTA) – in correctly classifying participants and non‐participants. We find that CTA is substantially more accurate than the other models. Moreover, unlike the other models, CTA offers transparency in its computational approach, ease of interpretation via the decision rules produced and provides statistical results familiar to health researchers. Beyond their application to research, data mining techniques could help administrators to identify new candidates for participation who may most benefit from the intervention.     

Do statins slow down Alzheimer's disease? A review

More than 4 million people suffer from Alzheimer's disease (AD) in the United States. The prevalence increases with age as the rate is 3% in those between 65 and 74 years compared with 47% among those over 85 years of age. Some epidemiological studies have reported a decrease in the incidence of AD with the use of 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase inhibitors (statins). Other studies have shown statins to decrease Abeta peptides, but data regarding cognitive benefits is lacking in this patient population. There are also concerns that statins, which cross the blood-brain barrier may cause more side-effects, but more information is needed. Adverse events were either infrequently noted or not reported in most of the published studies, and at this time there is insufficient evidence to suggest the use of statins for cognitive improvements in patients with AD.     

‘Nothing is really safe’: a focus group study on the processes of anonymizing and sharing of health data for research purposes

Rationale and objectives The availability of anonymized data is a keystone of medical research, yet little is known about lay views towards the process of anonymization or on the way that anonymized medical data are transferred to researchers. Methods During May and June 2009, as part of a wider consultation on methods for releasing data to researchers, three focus groups (n = 19) were conducted exploring lay attitudes towards the traditional ‘warehouse’ model commonly used in medical research for delivering anonymized National Health Service (NHS) data to researchers. The focus groups explored different processes such as the copying of data, use of programmers for linkage and anonymization, the transfer of data and governance. Results The recognition of the positive aspects of medical research and desire to support it formed the context for discussions. Nonetheless, individuals varied in their attitudes to the use of anonymized data extracts for research from their health records (without consent); although some appeared positive wanted to be asked to consent for this use. Furthermore, participants were acutely aware of security breaches of NHS information nevertheless, they continued to display a high level of trust in NHS staff. Participants were concerned about the practicalities of the warehouse model and relied on their own life experiences to make sense of the model (using analogies with ‘banks’ or ‘libraries’). The general attitude towards the processes underlying the warehouse model might best be captured by the term ‘ambivalence’. Conclusions This research (1) offers unique insights into views of anonymization of health data extracts, how it is undertaken and data are transferred and (2) adds to an increasing body of work that demonstrates that a minority of individuals are concerned about consent, even when data are anonymized although (3) those concerned about anonymization do not necessarily seek resolution through gaining consent.     

Utilities for prostate cancer health states in men aged 60 and older

PURPOSE: We sought to measure utilities for prostate cancer health states in older men. METHODS: A total of 162 men aged 60 years or older (52% of whom had been diagnosed with prostate cancer) provided standard gamble utilities for 19 health states associated with prostate cancer or its treatment using an interactive, computer-based utility assessment program. Demographics and experience with specific health states were examined as predictors of ratings using ordinary least squares regression analysis. RESULTS: Mean utilities ranged from 0.67 to 0.84 for living with symptom-free cancer under conservative management ("watchful waiting") and from 0.71 to 0.89 for symptoms occurring with treatment (prostatectomy, radiation, and hormone ablation). For long-term treatment complications, bowel problems (0.71) were rated as significantly worse than impotence (0.89), urinary difficulty (0.88), or urinary incontinence (0.83). Combinations of these conditions were rated as significantly worse than individual component states. Men who had experienced impotence or urinary incontinence rated these states as slightly better than men who had not experienced the specific problems. CONCLUSIONS: Both "watchful waiting" and treatment complications from prostate cancer treatments can have large impacts on quality of life. Mean ratings are important for use in policy-making and cost-effectiveness analyses. Variation in ratings across patients suggests that mean scores do not reflect individual preferences and that shared decision-making may be best for clinical decisions.     

他汀类药物依从性差对心血管不良事件的影响

目的探讨调查分析他汀类药物依从性差对心血管不良事件的影响。方法选取我院2015年1月~2016年1月就诊的冠状动脉粥样硬化性心脏病、缺血性脑卒中、糖尿病、高血压等心脑血管疾病的患者1248例,根据诊疗规范原则,在坚持原发病治疗用药基础上,需要长期服用他汀类药物的1020例患者作为研究对象,对患者进行随访6个月。根据患者的依从性分为依从性好组（n=502）和依从性差组（n=518）,对两组患者的心血管不良事件进行分析,并对心血管不良事件的发生率、发生时间、发病与停药（或间断用药）的时间、发病人群年龄段关系进行亚组分析。结果依从性差组发生心血管不良事件发生率明显高于依从性好组（P < 0.05）;大于60岁的患者发生心血管不良事件比例明显高于小于60岁患者（P < 0.05）;发生时间大于1年的患者发生心血管不良事件比例明显高于小于1年的患者（P < 0.05）;病程大于5年的患者发生心血管不良事件比例明显高于小于5年的患者（P < 0.05）;间断用药大于1年的患者发生心血管不良事件比例明显高于小于1年的患者（P < 0.05）;Logistic多元回归分析年龄大于60岁,停药大于1年、发病大于5年及发生时间超过1年是发生他汀类药物依从性差的心血管不良事件的危险因素。结论他汀类药物依从性差的患者更容易发生心血管不良事件,而且年龄大于60岁,停药大于1年、发病大于5年及发生时间超过1年是发生他汀类药物依从性差的心血管不良事件的危险因素。      

How a therapy-based quality improvement intervention for depression affected life events and psychological well-being over time: a 9-year longitudinal analysis

BACKGROUND: Short-term quality improvement (QI) interventions for depression can improve long-term mental health but mechanisms are unknown. We hypothesized that 1 pathway for such health benefits was an indirect effect with QI reducing risk factors for depression such as stressful life events. OBJECTIVE: To determine whether 6-12 month QI programs for depression reduce negative life events at 5-year follow-up and to model the relationship between program implementation, life events and mental health over 9 years. DESIGN: Forty-six primary care clinics in 6 managed care organizations were randomized to usual care or 1 of 2 QI interventions. We focus on the intervention that provided resources to assess and manage depression while particularly facilitating access to evidence-based psychotherapy ("QI-Therapy"). SUBJECTS: A total of 1300 enrolled patients with current depressive symptoms, who had data at any of 4 data points: baseline, or follow-up year 1, 5, or 9. MEASURES: Total and negatively-evaluated life events and psychologic well-being. RESULTS: A path model showed that QI-Therapy, in addition to improving psychologic well-being at year 1 (P = 0.0033), reduced negative life events at year 5 (P = 0.0033). This effect was not fully explained by improved psychologic well-being. Better mental health (P < 0.0001) and fewer negative life events (P = 0.0013) at year 5 were associated with improved psychologic well-being at 9 years. CONCLUSIONS: Depression QI programs that include resources for psychotherapy can reduce occurrence of life events, further protecting subsequent mental health. Implications for the design of QI programs and development of prevention interventions are discussed.