Structured Data Fields Research Articles

Using computer-extracted data from electronic health records to measure the quality of adolescent well-care.

To determine whether quality measures based on computer-extracted EHR data can reproduce findings based on data manually extracted by reviewers. We studied 12 measures of care indicated for adolescent well-care visits for 597 patients in three pediatric health systems. Observational study. Manual reviewers collected quality data from the EHR. Site personnel programmed their EHR systems to extract the same data from structured fields in the EHR according to national health IT standards. Overall performance measured via computer-extracted data was 21.9 percent, compared with 53.2 percent for manual data. Agreement measures were high for immunizations. Otherwise, agreement between computer extraction and manual review was modest (Kappa=0.36) because computer-extracted data frequently missed care events (sensitivity=39.5 percent). Measure validity varied by health care domain and setting. A limitation of our findings is that we studied only three domains and three sites. The accuracy of computer-extracted EHR quality reporting depends on the use of structured data fields, with the highest agreement found for measures and in the setting that had the greatest concentration of structured fields. We need to improve documentation of care, data extraction, and adaptation of EHR systems to practice workflow.

Natural language processing in biomedicine: a unified system architecture overview.

In contemporary electronic medical records much of the clinically important data-signs and symptoms, symptom severity, disease status, etc.-are not provided in structured data fields but rather are encoded in clinician-generated narrative text. Natural language processing (NLP) provides a means of unlocking this important data source for applications in clinical decision support, quality assurance, and public health. This chapter provides an overview of representative NLP systems in biomedicine based on a unified architectural view. A general architecture in an NLP system consists of two main components: background knowledge that includes biomedical knowledge resources and a framework that integrates NLP tools to process text. Systems differ in both components, which we review briefly. Additionally, the challenge facing current research efforts in biomedical NLP includes the paucity of large, publicly available annotated corpora, although initiatives that facilitate data sharing, system evaluation, and collaborative work between researchers in clinical NLP are starting to emerge.

Informing the design of clinical decision support services for evaluation of children with minor blunt head trauma in the emergency department: A sociotechnical analysis

Integration of clinical decision support services (CDSS) into electronic health records (EHRs) may be integral to widespread dissemination and use of clinical prediction rules in the emergency department (ED). However, the best way to design such services to maximize their usefulness in such a complex setting is poorly understood. We conducted a multi-site cross-sectional qualitative study whose aim was to describe the sociotechnical environment in the ED to inform the design of a CDSS intervention to implement the Pediatric Emergency Care Applied Research Network (PECARN) clinical prediction rules for children with minor blunt head trauma. Informed by a sociotechnical model consisting of eight dimensions, we conducted focus groups, individual interviews and workflow observations in 11 EDs, of which 5 were located in academic medical centers and 6 were in community hospitals. A total of 126 ED clinicians, information technology specialists, and administrators participated. We clustered data into 19 categories of sociotechnical factors through a process of thematic analysis and subsequently organized the categories into a sociotechnical matrix consisting of three high-level sociotechnical dimensions (workflow and communication, organizational factors, human factors) and three themes (interdisciplinary assessment processes, clinical practices related to prediction rules, EHR as a decision support tool). Design challenges that emerged from the analysis included the need to use structured data fields to support data capture and re-use while maintaining efficient care processes, supporting interdisciplinary communication, and facilitating family-clinician interaction for decision-making.

Outside the Box

I did my GP trainee placement in 1989. When I started, we were using fountain pens on Lloyd George cards. By the time I left, there was a hard disc and printer on every consulting room desk and we were just discovering the model-reality gap between the structured data fields on the software and the messy reality of real-world illness and its treatment.1 One particularly dramatic change was in repeat prescribing. In the pre-computer days, the doctors all gathered round a table in the conservatory after morning surgery for coffee and cake. In the middle of the table were several boxes of patient notes, each with a handwritten prescription affixed to it with a rubber band. We worked through them, initiating …

Why is warfarin underused for stroke prevention in atrial fibrillation? A detailed review of electronic medical records

Objective:Automated electronic queries of structured data fields in electronic medical records (EMR) found no barriers to warfarin in 42% of patients with atrial fibrillation or atrial flutter (AF) with moderate or high risk of stroke and no warfarin. A thorough manual review of records (including text reports) from the same EMR may better identify physicians’ reasons for not using warfarin.Methods:This was a cross-sectional, retrospective, manual EMR review. Patients identified in a previous automated EMR study with a CHADS2 (Chronic heart failure, Hypertension, Age >75 years, Diabetes mellitus, Stroke) score ≥2, no record of warfarin, no barrier to warfarin use, and (in the present study) confirmation of AF diagnosis were included in the manual EMR review. A structured chart abstraction form was used to extract data visible in the clinicians’ EMR user interface. Reasons why warfarin had not been prescribed were reported using descriptive statistics.Results:Among 408 patients with ‘no barriers’ to warfarin in the automated EMR review, AF diagnosis was confirmed in 319 patients (mean age 74.8; 65% female). Forty-one percent (n = 132) did not have chart records explaining why they were not on warfarin. Among the 59% (187) with a rationale against warfarin found in the records, the most common category (52%) was indicative of the risk of bleeding, either risk of fall or history of recent bleeding. The second most common category (16%) reflected that the patient was back in sinus rhythm. These findings are subject to inherent limitations of retrospective chart reviews.Conclusions:Many patients with AF and moderate-to-high risk of stroke are not treated with warfarin, and reasons for not using warfarin could not always be identified in patient records. Among patients with documented reasons, risk of bleeding (risk of fall or recent bleeding) was the most common category.

Can structured data fields accurately measure quality of care? The example of falls

By automating collection of data elements, electronic health records may simplify the process of measuring the quality of medical care. Using data from a quality improvement initiative in primary care medical groups, we sought to determine whether the quality of care for falls and fear of falling in outpatients aged 75 and older could be accurately measured solely from codable (non-free-text) data in a structured visit note. A traditional medical record review by trained abstractors served as the criterion standard. Among 215 patient records reviewed, we found a structured visit note in 54% of charts within 3 mo of the date patients had been identified as having falls or fear of falling. The reliability of an algorithm based on codable data was at least good (kappa of at least 0.61) compared with full medical record review for three care processes recommended for patients with two falls or one fall with injury in the past year: orthostatic vital signs, vision test/eye examination, and home safety evaluation. However, the automated algorithm routinely underestimated quality of care. Performance standards based on automated measurement of quality of care from electronic health records need to account for documentation occurring in nonstructured form.

Proven Processes

Proven Processes