Abstract
This article describes the rationale, processes, technology, and results of creating of a paper-based rounds report that is now used by our entire institution for efficient inpatient work rounds and checkout rounds that are routinely done in virtually every hospital, both academic and private, in the US. The results of a survey of clinicians suggests that printed rounds reports have markedly improved rounding efficiency, saved substantial amounts of physician time, standardized checkout processes, and improved patient safety.
Introduction
Electronic medical records (EMRs), among other things, are touted as improving patient safety, providing coded data for quality and other database uses, and increasing staff efficiency. 1 The new federal administration is undertaking a large and expensive effort to increase the use of EMRs by hospitals and office-based practices. A study of the use of full EMRs via a survey in April 2009 showed that only 1.5 percent of hospitals in the US use complete EMRs. 2 Good, complete EMRs contain a wealth of clinical information including problem lists, medication lists, disease prevention and disease management modules, laboratory, radiology, and other ancillary service data, multiple clinic notes and ways for the practitioner to order all services, and streamlined methods for creation of their clinical notes. All of these data present the problem of how to create summaries that can be used as a ‘snapshot’ of patient data that can be quickly perused without scrolling bars and multiple mouse clicks and screen changes. In addition, many large institutions have purchased ‘best of breed’ systems which leads to the problem of integrating disparate vendor applications so that they appear transparent to the end user.
The ergonomics of the use of computers and of moving from patient to patient, particularly in an inpatient ‘rounding’ environment or from exam room to exam room in the outpatient setting, also present a major challenge involving the types of devices used as well as security issues when multiple users of single devices are involved. Though PDAs, cellular telephones, and other small devices are being used to try to solve the challenges of data viewing and gathering, only a small amount of summarized data can easily be presented on these devices. A standard sized computer in every hospital or exam room, or computers on wheels (COWS) and tablet computers that are carried from room to room by hand, can present a larger volume of summarized data reasonably quickly, but still are not as efficient as a well-designed piece of paper with summarized data in the hand of the provider.
The medical literature suggests, in some cases based on observation of workflow in clinical environments, that both patient-centered and clinician-centered views of data improve efficiency in care in both the inpatient and outpatient settings. Therefore, though a detailed record of laboratory, reports, medications etc. viewable on a single patient record is a necessary feature of clinical care workflow, a succinct summary of the same data for groups of patients, either on a given outpatient clinic day or on a ward service, is just as important for efficiency for a clinician caring for that grouping of patients. 3, 4
This article describes the process, satisfaction of staff members, and average time saved per day by our house staff after creation of an inpatient paper-based rounds report that was designed by our institution’s house staff, and provides a concise presentation of data gathered from our complete EMR on several sheets of paper (for an average of a 20 patient service).
Background
We have developed a comprehensive EMR named WebCIS at the University of North Carolina Health Care System (UNCHCS) over the last 20 years. 5,6 The EMR (which has continuous improvements released approximately every 4 months) has made our practice of medicine almost ‘paperless’ in both inpatient and outpatient environments. Over the years we have added commercial applications that are seamlessly integrated into the user interface by a global session manager (GSM) which adheres closely to the Clinical Content Object Workgroup (CCOW) standards. We used to our advantage the fact that we have a single unique number that is used by all the disparate applications to identify a specific patient. This avoids having a master patient index to provide a crosswalk table if different identifiers were used in each of the applications that comprise our entire system. The main vendor applications include a complete electronic nursing suite of applications that record, among other things: shift-based vital signs, fluid balances, data from monitors and ventilators in the ICU settings, nursing notes, care plans, ancillary service notes, and a complete computerized physician order entry system (CPOE). The CPOE has been in use for almost 7 years and is mandatory in all areas of the hospital excepting chemotherapy orders which are still done on paper. The overall benefits to a large academic hospital in terms of reduced costs of care and improvements in throughput of patients within the system have been well reported. 7, 8
Though we have not formally proven that our health system has realized these benefits, mainly because we have had our EMR for almost 20 years, making a comparative study quite difficult, the consensus of our clinicians and administrators is that we have realized the gains that a full EMR provides. However, proof that the EMR actually improves patient safety has been very problematic. Most of the literature on improvement in patient safety either has been flawed, 9 or has been based on users’ perceptions that certain applications within an EMR have improved patient safety. 10 More methodologically robust studies on improvement in patient safety from use of the EMR are in progress, 11 but the jury is still out as to whether use of the EMR actually improves morbidity and mortality in large population groups.
These benefits, however, have come at a cost of increased time for daily inpatient rounds compared to the paper world. In the paper world (for us in the late 1980s and early 1990s), morning and evening rounds were made with a rolling chart rack which contained the orders, notes, and ancillary reports, and every room had a ‘door sheet’ with a 24 hour total of vital signs, fluid balances and, in the ICU setting, a large paper flow sheet with handwritten graphical representation of ventilator settings, cardiac and central line data, and other acute care parameters. Orders would be handwritten immediately into the paper chart and the team would move to the next room and repeat the process. In the computerized world, rounds are conducted usually with one or more COWS or one large monitor over the patient bed in some units. In this environment the rounding team must gather around the computers, with one person navigating to the different sections of the EMR that provide the needed data in electronic format, and then enter orders using the CPOE module. Our older faculty who have rounded in the paper world as well as the computer world and could compare the time spent on team rounds in both environments have unanimously felt that ward and ICU rounds take half an hour to one hour longer in the computerized environment. 12
In addition, we have had problems in both the paper era and the computerized era of insuring that due diligence is paid to detailed checkout rounds at times of house staff changeover, usually in the evenings and in the early morning. There has been no standard for how these rounds have been conducted at our institution. Checkout rounds on our different services range from a telephone conversation between house staff, to various handwritten or word processor generated sheets passed between outgoing and incoming staff. We believe that this presents a significant patient safety issue.
To solve the problem of both the extended time to round and the inefficiency of the checkout process, we created an electronically generated PDF printed paper rounds report that serves as a means both to streamline the time for rounding and to standardize the checkout process. Below we describe how we accomplished this, and the resulting time savings and perceived improvement in patient safety.
Developing the paper rounds report
We began the planning process in the manner we have used for years in the development of our EMR by involving the clinicians most affected by this process change, in this case our house staff. A group of 15 volunteer interns and residents representing all of our specialties was established and co-chaired by our Chief Medical Information Officer and our Chief of Staff, both of whom are practicing clinicians in the inpatient areas. This group decided what minimum set of data was required to adequately provide enough summary information from our EMR for safe rounding and checkout while not supplying so much information that the paper-based report for a large service became unwieldy. We created mockups of the report and continued an iterative process until a final mockup draft was agreed upon.
The main technical issue to be overcome was that we had to generate a method of simultaneously running and retrieving data from multiple databases. This had to be done in the most efficient way to prevent degradation of response time of the entire system when reports were run. When running the application we had to gather data pertinent to each patient on the list and to make sure that the data pulled back were correct.
The first database needed was a nursing documentation system. Our data were stored in multiple rows and tables all provided by the vendor application. There were five tables with a billion rows in some. As this was a production system we could not extract data directly from it. We had to come up with a method of copying this amount of data without affecting our production environment. We tried various ways but were hindered by the amount of data that needed to be moved. We also tried to replicate the data on each write to the production database, using industry standards, but again with the amount of data that needed to be moved, this too was unsuccessful.
To solve the problem, we created a net new database (a mirror database) by restoring the production database at a given point in time. This gave us the bulk of preexisting data, but the challenge was to copy to the separate ‘mirrored database’ the ongoing incremental writes to the production database. This challenge was more intrusive. The current production system backup schedule is a nightly backup, a 4 hour differential backup, and a 5 minute log file. In order for us to capture current data we had to create a method of incrementally capturing the 5 minute log file. We wrote scripts to go to the production server to see if there was a log file to be inserted (making sure it was the next sequential log file) and used this to update the mirrored database.
After having the data replicated, we then had to create calls into the database to extract the pertinent detail we required. This included the vital signs and fluid balances that were documented by the nurse. We needed a method that called on four tables that house the vitals data, medical record number, and visit number. The second database that we needed data from, DB2, was where all the lab results were kept. This scenario was the simplest to achieve: it was a straight call to the tables we used for our EMR.
The last was more complicated in that these data were in an older (VSDM) file system. This is where that particular vendor stored all the CPOE data. We wrote a COBOL stored procedure to capture the data direct from that production database.
One of the challenges with the data extraction was that we needed the data pertinent to each patient. Some of the hospital units could have upwards of 60 patients. The program was written so that each patient in the list had multiple calls to each of the three stored procedures simultaneously. A standard ‘percent time to completion’ sliding bar was presented to the user to allow them to follow the progress of the report generation. The final report was in XML format on screen, and printed as a PDF.
We were very concerned that the process of generating the rounds reports on a demand basis at any time of the day would result in a total production slowdown of our entire EMR (which averages 3 million database queries/writes daily and an average of 3000 simultaneous users during the day), or that the generation of the rounds report itself would take an inordinate amount of time. In fact, the average time to generate a rounds report for a 20 patient inpatient service was around 30 seconds, with no apparent slowdown of the EMR itself from simultaneous generation of multiple rounds reports.
The design team of residents purposefully had the report printed in 6 point font in order to get as many patients and as many data on each page of the report as possible. The average number of patients per sheet was four (eight if double-sided printing was chosen). A representative single page containing several patients’ data is shown in Figure 1, and the ‘legend’ for the data which was printed as the last page of the report is shown in Figure 2. The only complaints about the font size on the report predictably came from senior clinicians over 50 years old; however, the report was not designed with them in mind, though many are using it themselves to quickly review data on patients on their service when not rounding with the house staff.
Single page of printed rounds report with data
egend at bottom of printed rounds report
The report on each patient consisted of mostly prefilled data from our systems that included most recent vital signs and the range over the last 24 hours, total fluid balances including all drainage tubes, IV fluids, and oral fluids, daily weights as well as admission weights, current diet, resuscitation status, isolation, restraints and other risks (such as flight risk), length of stay, all current medications with antibiotics separated out with start dates and cumulative number of days on a given antibiotic, and many laboratory values which were chosen by the development group of residents that they felt were most important to see day to day. The labs included the most recent lab with its date and time stamp as well as the previous value, if present, for that lab, again with a date and time stamp. We made use of the special characters ( ), [ ], { } to enclose lab values and put at the top of the lab section the date and time stamps associated with each special character so we could save a great deal of space in the lab display in the printed PDF file. We also made use of the traditional fishbone display for the lab results. The only areas the clinicians had to enter data themselves were in the diagnosis, procedures and culture results, and plan sections.
Impact and evaluation of the printed rounds report
We also observed that right after the initial release, other members of the care team, including nursing, social work, physical therapy and others, used the rounds report to enter their specific plans within the plan section of the report by giving them a heading such as ‘nursing plans’ to discriminate them from the clinicians’ plan. Similar to the clinicians, these other team members used these paper-based reports for their daily rounds and for communication among each other at shift changes.
Two months after the rounds report function was released, we averaged 19,154 generations of the report monthly. This number may be actually lower than the usage during hospital rounding by ward teams and consulting teams, as one member of the team may have generated one report and done multiple prints for the rest of the members of the team, which we know by report from the residents was a common occurrence. At this time, we also sent out a questionnaire to all of our residents (we did not include nurses and other team members in this questionnaire as the application was designed only with the resident clinicians in mind). The response rate was approximately 25 percent with 112 residents responding. The questions and responses are shown in Figure3.
Overall usage (including at shift change) and perceived improvement in patient safety using rounds reports
Discussion
Though others have reported developing rounds reports in the medical literature, these reports were either viewed on a PDA, one screen per patient, or printed in PDF format, again with one sheet per patient. 13,14 Other groups have developed electronic ‘hand-off’ sheets designed for use at the time of shift changes. 15 These have been described to improve patient safety but, as discussed above, that conclusion has been based on the perception of users of such systems, not on rigorous data-based methodologies. None of them contain the richness of data in such a concise format as ours does. The few rounds reports described in the literature were generated from a single database structure, while ours is unique in the literature as it overcame the hurdle of generating data from multiple disparate databases, giving it the ability to auto-generate not just laboratory but also medications, nursing data, alerts and risk data, and other data felt crucial for care of the patient.
The rapidity with which this report could be generated and printed without affecting the performance of the entire EMR further increased the efficiency of our clinicians. The time savings per day per house staff officer (ranging from 15 minutes to more than 1 hour) allowed them to spend less time looking at computer screens and more time on their other duties. Since the reports also contained a plan section that was continuously updated, it has begun to standardize our checkout process (per the results section, 68 percent of house staff use the report during discussions at shift changes) which previously was done in multiple different ways depending on the specialty involved. The large number of reports generated per month indicates that the application is being used not just by ward teams, but also by consulting teams and other caregivers throughout our facilities. As stated earlier, the reports were also used by nurses and others on the care team, both for rounding purposes and for their own checkout rounds.
Our survey and counts of reports printed was done only two months after the application was released. The response rate to the survey was only 25 percent, which makes any statistical analysis of efficiency improvements and other parameters of usage impossible. However, the numbers of reports printed per month have continued to rise, indicating further adoption of rounds reports for daily care. We have recently expanded the data on the reports to include additional auto-generated data on ICU patients such as ventilator settings and data from indwelling catheters such as Swan Ganz lines, arterial lines, and other ICU monitors. Feedback of the system has included requests that medication doses be included in the report (we only report the medication itself but not the dosing), that stop and start dates for antibiotics be automatically calculated from our CPOE system, and that the reports should be printable from handheld devices. These requests have to be balanced against the initial design decisions that dictated that the summaries be as concise as possible and take up as little room as possible on the printed sheets of paper.
Conclusions
We created a paper-based report generated from our EMR that was designed by our house staff to provide for concise, data-rich documents that are used by almost every hospital ward and consulting service for the purposes of efficiently rounding on inpatients. The reports are only several pages long for fairly large services and the data are auto-generated from our existing disparate databases. They provide the same information that can be viewed from our EMR on a patient-by-patient basis, but markedly improve the ergonomics of rounding and save substantial amounts of time over the use of the EMR as teams move from room to room. The reports have also begun to standardize the checkout process at the end of the day for our ward and consulting teams.
Evaluation by clinicians (though limited by the low survey response rate) suggests that paper-based reports are more effective and efficient in some situations, compared with viewing computer screens. Our findings are consistent with previous work that shows that concise information based on a clinician-centric view of groups of patients can be as effective as the more detailed, complete information (in some cases redundant) that is seen in a patient-centric view in the full EMR. Our findings also suggest that there may be other patient care duties now performed by viewing EMR data that might benefit from a move back to paper.
Footnotes
None of the authors have a conflict of interest on this work.