Emergency Department Access to a Longitudinal Medical Record

Abstract

Our goal is to assess how clinical information from previous visits is used in the emergency department. We used detailed user audit logs to measure access to different data types. We found that clinician-authored notes and laboratory and radiology data were used most often (common data types were used up to 5% to 20% of the time). Data were accessed less than half the time (up to 20% to 50%) even when the user was alerted to the presence of data. Our access rate indicates that health information exchange projects should be conservative in estimating how often shared data will be used and the wide breadth of data accessed indicates that although a clinical summary is likely to be useful, an ideal solution will supply a broad variety of data.

Introduction

The emergency department (ED) represents an important transition of care for patients, where understanding the longitudinal patient health condition (e.g., problems, allergies, medications, diagnoses, recent procedures, recent laboratory tests) is critical to forming an appropriate plan of care. ^1,2 Because many emergency department visits are unplanned and urgent, this information may not be conveyed in advance to ED physicians. Even when such information may be available to physicians, time constraints can restrict their availability to pursue it. For example, a recent survey found that even though many ED physicians believe a majority of their patients would benefit from longitudinal patient health information, they attempt to obtain such data less than 10% of the time. ³ Patients and caregivers are asked to provide such information, but studies have consistently shown that they are ill-prepared for this role. ⁴ The result is information gaps for ED physicians in approximately one third of patients, of which half are considered essential to patient care. ⁵ These information gaps can lead not only to decreased care quality, ⁵ but also inefficiencies in care, in the form of redundant testing, care delays, and less-effective treatments prescribed rather than more tailored interventions as a precaution when trusted data are not available. ⁶

Researchers have studied the information needs of emergency department physicians by surveying them ³ and by reviewing physician behavior in a constrained laboratory environment. ² Stiell and colleagues surveyed physicians for information gaps, ⁵ determining what physicians needed beyond what was available in their clinical information system. Our study complements theirs by determining how physicians use the information that is actually available to them in a longitudinal medical record.

Case Description

The New York-Presbyterian Hospital/Columbia University Medical Center has a longitudinal medical record system called WebCIS ⁷ that contains data on over 2,500,000 patients and that has been in operation for 17 years. WebCIS contains all laboratory data (including microbiology tests), discharge summaries, operative reports, radiology reports, pathology reports, and similar reports from ancillary departments. It serves as the primary legal source for all such reports, which are not stored in the paper record. In the ED, WebCIS is the sole means of obtaining these reports, other than directly calling the appropriate department. WebCIS includes electrocardiogram images, but not radiological images, which are reviewed on a PACS system. It also includes administrative data for all inpatient, emergency department, and outpatient visits. It contains approximately 2,000,000 clinician notes such as admit notes, progress notes, and visit notes, although most of these notes are still written on paper at the institution. Resident signout notes are generally used to document a patient’s problems and plan for residents that cover a service overnight or over a weekend.

The ED staff comprises ED attendings, emergency medicine residents and residents from other departments who rotate through the ED, nurses, social workers, clerks, other permanent personnel, and visitors such as consultants, private physicians, residents who are admitting a patient, and others. Nearly all staff members have access to the longitudinal record. ED attendings sometimes review patient data directly on WebCIS, residents sometimes review data and present it to the attendings, and clerks sometimes print out data and place it on a paper chart. WebCIS access is available throughout the ED at clinical stations and in work rooms; there is about one workstation per three patient beds. Clinicians use WebCIS for all data review, other than direct viewing of radiographic images and direct viewing of the current electrocardiogram. Most documentation is done on paper, although the clinicians sometimes write notes in WebCIS when access to the ED information is perceived to be especially important (e.g., if the plan of care is complex).

Methods

We measured all access to patient data within a day of an individual ED visit by any user at any location. We included all ED patients, adult and pediatric, from surgery and medicine. This included ED staff as well as consultants, the inpatient admit team, the private physician, etc. While this inclusive approach might have included data that were not essential to the ED mission of stabilization and triage, we used the approach to allow us to recognize any data that might have been useful to the ED staff even if it came through an indirect route, and to account for delayed registration and delayed documentation.

We tallied access by data type. As soon as a user logs onto a patient record in WebCIS, it shows the user the availability of data for most data types. The date of the most recent data is shown, and the user can review the data by selecting its link. WebCIS categorizes data types in varying levels of granularity. For example, WebCIS puts notes such as admit notes, progress notes, and visit (clinic and private office) notes, which are authored in its general-purpose clinical note writing tool, in a single group (“clinician notes”), but resident signout notes, discharge summaries, operative reports, pediatric resident notes, dictated consult notes, and eye clinic notes are authored in separate functions or systems (but still viewed in WebCIS). For a few data types (indicated in the results), WebCIS does not show a corresponding date, so the user must click the link to find out if there are data available.

We focused on review of “old data,” which we defined as data that were at least three days old at the time of access, based on the date that a specimen was obtained, or that a test was performed, or that a clinician note or order was written. We tallied the number and proportion of visits in which the users reviewed old data, stratifed by data type. As a denominator, we used all visits and we used the number of visits for which old data of that type were available. We broke several aggregate data types down by individual reports or notes.

We analyzed audit logs to ascertain user access; the logs record all user access at the granularity of individual reports for individual patients. We analyzed the clinical records to determine the availability of old data. Patients who visited the ED twice in one day were counted as having a single visit. All analysis was retrospective and was carried out under Institutional Review Board approval.

Results

From June 1, 2005 to December 31, 2005, there were 68,337 ED visits. There were 40,818 ED visits (59.7%) that had any WebCIS activity (review, update, etc.) by any WebCIS user within a day of the visit. ▶ shows the frequency of access to old data (as defined in Methods) as a percentage of ED visits. Clinical laboratory reports were viewed most frequently. For some clinical laboratory test displays, old and new data were shown simultaneously. If those displays are eliminated from the analysis, then the percentage drops from 30.5% to 21.2%. The last column shows the frequency that users accessed old data given that they were shown that it was available. Even given this information, users still only reviewed the data up to about half the time. ▶ illustrates what the three most common aggregate data types contained.

Table 1.

Table 1 Frequency of Review of Old Data

Data Type	% of Visits in Which Users Reviewed Old Data (of 68,337 Total Visits)	% of Visits in Which Users Reviewed Old Data, Given that They Were Shown that Old Data of That Type Were Available
Clinical laboratory tests§	30.5	42.5
Radiology reports§	14.4	25.7
Clinician notes§	11.0	47.4
Visits	10.4	∗
Outpatient meds	10.1	∗
Diagnoses (ICD9)	9.2	∗
Resident signout notes	8.9	36.5
Pathology reports	7.0	20.9
Obstetrics and gynecology ultrasound	6.9	49.2
Discharge summaries	6.7	31.7
Cardiology reports	5.5	20.0
Operative reports	4.3	24.7
Providers	3.1	∗
Inpatient medications	2.3	†
Electrocardiogram images	2.1	†
Nursing and pediatric resident notes	0.8	10.6
GI endoscopy reports	0.7	13.9
Neurophysiology reports	0.5	13.2
Pulmonary function test reports	0.1	7.1
Dictated consult notes	0.1	15.5
Eye clinic notes	0.0	3.0

^∗ For these data types, users were never shown whether old data were available; they had to pursue a link to find old data.

^† Although users were shown whether old data were available for these data types, the number of times that occurred was unavailable to this study.

^§ These aggregate data types are broken down in ▶.

Table 2.

Table 2 Breakdown of Most Common Reports for Aggregate Data Types

Data Type: Report	Distribution of Reports for That Data Type (Percent)
Clinical laboratory reports
Complete blood count	13.6
Basic metabolic panel	11.3
Urinalysis	10.5
Hepatic function panel	8.1
Sexually transmitted disease (culture, serology)	6.2
Urine culture	3.7
Alcohol or drug screen	3.6
Blood type	3.4
Coagulation profile	3.3
Radiology reports
Chest radiograph	19.5
Head CT or MRI	13.4
Pelvic CT or MRI	8.0
Abdominal CT or MRI	6.7
Spine imaging (X-ray, CT, MRI)	6.6
Abdominal ultrasound	5.8
Clinician notes
Generic untitled note	31.3
Consult note	12.4
Clinic note	11.0
Resident progress note	10.0
Off service or discharge note	8.6
Admit note	6.3
Hospitalist attending note	4.8
Emergency department note	4.5
Event note	4.5

Discussion

Our results indicate that data generated before the current visit are accessed often from the ED, but not a majority of the time, even when the user is notified of the availability of data. The frequency of access is consistent across data types, with users accessing data 5% to 30% of the time for most general data types, and 20% to 50% of the time for most data types when those data are known to be available. There may be several reasons why old data are not accessed more often. It may be known that another team will review the results when a patient is admitted, for example. Some data may be rendered irrelevant by current data. The clinical condition may not require a detailed review of the patient’s history. Depending on the circumstances in the ED, there may simply be insufficient time to review old data for some patients.

It is surprising that despite our inclusive definition of ED data access, only about 60% of visits involved any use of WebCIS (for current or old results, or for data entry). A large proportion of our ED visits are for minor trauma, and many patients use the ED for primary care, arriving with symptoms indicative of upper respiratory infections and mild gastrointestinal illness. Clinicians may not review data on all these patients. Sometimes testing is done with the intention that the primary care provider will check the results at a later date. Furthermore, radiological images and electrocardiograms from the current visit are generally reviewed directly, so the current radiological and electrocardiogram reports may not be accessed via WebCIS. In addition, providers simply may not take the time to do it. ³

Our results provide lessons about health information exchange: what to exchange, what standards should cover, and how to evaluate it. Initiatives like the Continuity of Care Record (CCR) ^{8, 9} strive to define documents that “contain the most relevant and timely core health information about a patient.” ¹⁰ Our results can be used to confirm the importance of laboratory data, clinician notes, visit and diagnosis information, and key ancillaries such as radiology and cardiology. As ▶ demonstrates, however, an ideal system will supply a broad variety of data.

The relatively low rate of access implies that data exchange projects should be conservative in estimating the effect of the data exchange, so such studies may require a larger sample size to achieve sufficient power. This is especially true because we provided a best-case scenario for access: data were likely to be available on most patients, many types of data were available, and the data were accessible in the clinical information system that clinicians used daily. Data exchange projects in which less clinical information is available and in which the information must be accessed via a separate system are likely to have even lower rates.

The limitations of our study include the fact that this was a single center with a single longitudinal record. Although we would have liked to measure user intent, we only had user behavior in the form of audit logs. Furthermore, although we can determine what users looked at, we cannot determine the utility of that information. Users may fail to access old data because they can tell from the first screen that the data are too old to be useful; this may lead to a minor underestimation of the rate of access to old data. Most of our inpatient and ED notes are still written on paper, and this may reduce ED users’ attempts to access electronic notes, resulting in an underestimation of the value of text notes.

In summary, we used detailed audit logs to determine how often users in the ED access clinical information that comes from previous visits. We found that clinician-authored notes and laboratory and radiology data were used most frequently. Even when the user was shown that old data were available, and even though the data display was integrated into the user’s normal workflow, he or she generally accessed the data less than half the time.