Bernstein, Peter S. MD, MPH; Farinelli, Christine MD; Merkatz, Irwin R. MD
An Institute of Medicine report in November 2002 called for implementing state-of-the-art information technology in all health care settings to improve communication among providers. The objective was to limit errors and costly duplication of effort, with the ultimate goal of eliminating paper-based processes.1 Obstetrics is an ideal specialty for both the implementation and evaluation of the effectiveness of a paperless record. It is a well-defined field with a relatively standard course of care for a common condition, and it focuses on events that occur over a set interval of time lasting approximately10 months (including postpartum care). Communication among providers is a key element in ensuring quality care, because all patients have their care transferred from outpatient to inpatient settings and back, and patients may be cared for by different providers.
Despite this large need for new medical record systems, there is almost no obstetric literature to demonstrate that the quality of care is improved with the use of electronic medical record systems. Many providers are loath to adopt such records because of their cost, their reticence to work with computers, and their fear that confronting a computer will diminish their productivity.
The objective of this study was to demonstrate that a network-based electronic prenatal record could improve communication within Montefiore Medical Center among one of its largest outpatient obstetric offices, its antepartum testing and prenatal ultrasonography unit, and its labor and delivery suite.
MATERIALS AND METHODS
We undertook a retrospective, multiple-time-series study before and after implementation of an electronic prenatal record. After receiving approval from our Institutional Review Board, we reviewed the inpatient charts of all women who delivered at the Jack D. Weiler Hospital of Montefiore Medical Center during 2 separate periods 1 year apart: the first before implementation of a paperless prenatal record and the second after the system was put into operation. Patients studied all received their prenatal care at one of the largest outpatient sites in the Montefiore network, Comprehensive Family Care Center. This office is located in a self-contained building 4 blocks from the hospital. Comparisons were made between the group of Comprehensive Family Care Center patients who delivered in August 2002 and those who delivered August 2003.
Montefiore Medical Center, the University Hospital for the Albert Einstein College of Medicine is ideally situated to benefit from the use of a network-based computerized information system. It is the largest health care system in the borough of the Bronx, in New York City. Montefiore Medical Center has an established health care delivery system that incorporates full-time physician employees at 16 ambulatory care sites located throughout the Bronx and lower Westchester County. Unlike many academic medical centers, Montefiore Medical Center has experienced a substantial growth in its obstetric patient base during the past 6 years and now performs approximately 4,600 deliveries each year. The full-time faculty of the Department of Obstetrics & Gynecology and Women's Health of Montefiore Medical Center is directly responsible for the total prenatal, intrapartum, and postpartum care for more than 4,000 (> 90%) of Montefiore Medical Center's annual deliveries.
All the offices in the Montefiore Medical Center network are connected by a secure hospital computer network. This network supports scheduling and registration systems, as well as a sophisticated clinical information system, which incorporates patient laboratory and radiology data, as well as basic information about patient admissions.
Comprehensive Family Care Center is one of the larger offices operated by Montefiore Medical Center. Comprehensive Family Care Center primarily serves an inner city group of women who are either uninsured or receive Medicaid. Approximately 45% of the patients are Hispanic, 30% are African American, and the remainder are non-Hispanic whites and Asian. Comprehensive Family Care Center houses a maternal-fetal medicine practice that receives referrals from community health centers throughout the Bronx, both from within and outside of the Montefiore Medical Center network.
On July 1, 2003, we began the roll-out of a computerized prenatal record that was developed by AS Software, Inc, (Englewood, NJ) in design consultation with our Department of Obstetrics & Gynecology and Women's Health. This system was installed on a server that was integrated into the Montefiore Medical Center intranet. The system is a comprehensive prenatal record encompassing all elements of prenatal care, and access to individual patient records is simultaneously available in the outpatient office, the hospital's antepartum testing unit, and the labor suite. All prenatal contacts with the patient (eg, office visits, telephone calls, and consultations with other providers) are recorded in the record, as are ultrasound and antepartum fetal testing reports.
Providers, nursing staff, and clerical staff were trained in the use of the software by one of the authors (P.S.B.) in a single 1-hour session. Use of the software is relatively self-explanatory, and providers with any experience using computers encountered little difficulty in learning the system. Clerical staff were trained to enter demographic data, nursing staff to enter vital signs, and obstetric providers were trained to enter all of the clinical data that they typically would enter into their patients’ paper charts.
This computerized record replaced our paper-based system that had used the POPRAS (Problem Oriented Patient Risk Assessment System) forms distributed by Perinatal Health, Inc., (Citrus Heights, CA). In the process of completing these forms, 3 carbon copies are made. The design of the forms allows one copy to be sent to the antepartum testing unit, one to remain in the ambulatory office, and another to be sent to labor and delivery suite at the hospital in anticipation of the patient's admission. This last copy of the prenatal record is typically sent between 32 and 35 weeks of gestation.
Inpatient charts were reviewed for the presence of any outpatient prenatal record. If a prenatal record was noted to be present in the inpatient chart, the record was further inspected to determine how long before the patient's admission for delivery was the last documented outpatient prenatal visit. Study of the presence or absence of relevant laboratory data could not be undertaken as an evaluation of the electronic medical record because of the preexistence of the medical center's comprehensive laboratory information system.
Other data collected included the presence of obstetric or medical complications either during the course of prenatal care or during the course of labor. Complications of interest were those medical and obstetric issues that would be relevant to caregivers on labor and delivery. In particular, we also reviewed the inpatient charts for any documentation of an ultrasound examination that had been performed during the course of prenatal care. This documentation could be the actual ultrasound report or a note by a provider that documented the results of the ultrasound examination. Records of the ultrasonography unit were also reviewed to determine whether, in fact, the patient had undergone an ultrasound examination at any point during the pregnancy; this allowed us to determine whether documentation of an ultrasound examination was missing from the inpatient chart at the time of admission for delivery. All patient identifiers were removed from the data set once the data had been abstracted from the charts to maintain patient confidentiality.
Our primary outcome variable was the presence of a prenatal chart in the hospital record at the time of admission for delivery. A preliminary chart review suggested that approximately 30% of the time before the implementation of the computer-based system, no prenatal chart was available when the laboring patients were admitted to the delivery suite. We anticipated that use of the computer-based system would increase the percentage of the time that the record was available to nearly 100%. Thus, assuming a Type II error of 20% and an alpha error of 5%, we would need to review a minimum of 30 patient charts in each of the 2 time periods. Approximately, 40–50 patients deliver each month at Weiler Hospital who have received their prenatal care at Comprehensive Family Care Center. Thus, we chose to review the inpatient charts of all Comprehensive Family Care Center patients who delivered in 2 comparable months before and after the implementation of the computerized prenatal record. A Fisher exact test was used to compare categorical variables and Student t test for normally distributed continuous variables. The Mann-Whitney U test was used to compare nonnormally distributed continuous variables. We determined whether continuous data were normally or nonnormally distributed based on a graphical plot of the data and visual inspection.
The total number of women cared for at Comprehensive Family Care Center who delivered at Montefiore Medical Center during the 2 time periods selected were 49 in August 2002 and 45 in August 2003. All but 8 inpatient charts were available to review from the 2 study periods combined, 6 from patients who delivered in August 2002 and 2 from August 2003. That left 43 charts from each time period for review.
Demographic and clinical characteristics of the populations were similar (Table 1). Complications noted during the chart abstraction included obstetric issues such as prior cesarean delivery, breech presentation, preeclampsia, twin gestation, and suspected macrosomia. Medical complications noted included chronic hypertension, diabetes (gestational and pregestational), hepatitis B and C, sickle cell disease, and asthma.
The prenatal chart was absent from the inpatient chart in 7 (16%) of the cases of patients from August 2002 compared with 1 (2%) in August 2003 (P < .05) (Table 2). Among patients with the prenatal record missing from the inpatient chart, the complications noted in the admission history were pregestational diabetes mellitus, asthma (2), prior cesarean delivery (2), and prior classical cesarean delivery, preterm premature rupture of membranes and preterm labor, and chronic hypertension.
Among charts with prenatal records available, the median length of time between the last documented prenatal visit and delivery was significantly greater for August 2002 patients compared with August 2003 patients (36 compared with 4 days, respectively, P < .001). All patients had received prenatal ultrasonograms, but documentation of the ultrasonogram was missing from the inpatient records in 7 (16%) of the August 2002 charts compared with none of the August 2003 charts (P = .01).
We have found that the implementation of an electronic prenatal record improves communication among an outpatient office, the labor floor, and the antepartum testing unit of a busy metropolitan medical center. Not only were records more likely to be available when the patient was admitted for labor, but the records were also more likely to be up to date and contain information from the most recent office visits. Thus, providers in the inpatient setting had all data available to permit them to provide the best quality care to their patients in labor after we began using the computerized prenatal record. Implementation of such a prenatal record system across a complex and busy network of primary care offices, such as the one run by Montefiore Medical Center, has the potential to vastly improve the quality of care provided to a large underserved population of pregnant women.
More specifically, we found that the use of a computerized prenatal record significantly improved the availability of important data, such as the most recent blood pressures measured in the office, which in turn may be vital to the care of a patient suspected of having preeclampsia. Additionally, providers on the labor floor had better access to prenatal ultrasound reports, something that may be critical to the management of a patient in labor. In fact, we noted that providers caring for patients whose charts were unavailable at the time they were admitted were at a distinct disadvantage: They did not have access to information about a range of conditions potentially affecting the course of their patients’ labor and delivery, conditions which were addressed during the course of prenatal care.
Providers and nursing staff seemed to readily accept the change to using a computerized prenatal chart. They were particularly happy because of their improved access to information both in the office and when patients were admitted to the labor floor. We noted that they had a relatively easy time learning to use the software. More difficulty was encountered in adapting office routines to the new system of charting, but these too were well tolerated when balanced against the benefits accrued from using the new system. The most telling piece of data is the fact that many of the other offices that provide prenatal care at Montefiore Medical Center are impatiently waiting for their turn to have the electronic system implemented. At this point in time the only impediments to a full rollout are time and financial limitations.
Given the potential to improve the quality of care, we were surprised that findings such as ours have not been frequently reported or discussed in the medical literature. Although computer networks are more often being used by health care organizations, there is very little literature demonstrating their benefits in the clinical setting, particularly in obstetrics. Most publications have been descriptive in nature. Studney and colleagues2 reported in 1977 on a system used by a large health maintenance organization. In this system, providers entered data onto a form at the time of an encounter with a patient. The data on the form was later entered into the computerized system. At subsequent visits, the provider was given computer-generated summaries and flow sheets instead of a traditional chart. The database was subsequently purported to be useful for quality assurance studies. Studney3 also reported that a computerized prenatal record also improved the validity of claims data.
Gonzalez and Fox4 in 1989 reported on their experiences in the development and implementation of a computerized prenatal record. Although the authors provide no hard data, they reported that time was saved because there was no searching for misplaced records. In addition, because laboratory data were directly imported into their system, time was also saved because no filing was required. The authors also noted that illegible handwriting was no longer an issue. They did report some difficulties in training staff to use the system, particularly those with limited typing skills.
Computerized medical records have shown usefulness in other outpatient settings. Whiting-O'Keefe and colleagues5 found that a computerized outpatient medical record in an arthritis clinic resulted in improved flow of information and allowed physicians to better predict their patients’ future symptom changes. Margolis and colleagues6 limited their computerized record to a problem list and standard database components of a problem-oriented record in pediatric clinics. They noted that a more complete problem list was created with the computerized system and hypothesized that this led to better care.
Given these prior articles, our expectation was that the computerized prenatal chart would perform better than the traditional paper-based prenatal record. Our current multiple-copy, carbonized, paper-based prenatal record system has multiple pitfalls that exist even when the system is employed under ideal circumstances. Because our protocol for the paper-based prenatal record is that a copy of the prenatal record is sent to the delivery suite at a point between 32–35 weeks of gestation, the record is rarely up to date with the most recent information when a patient presents in labor. Thus, important information, such as the most recent blood pressures, may be missing in the middle of the night when a patient is admitted with potential complications, such as suspected preeclampsia, for example.
Most important information will not be available for a patient who presents to the labor floor at a gestational age before the 32nd week of gestation when using a paper-based system. This is almost by definition a high-risk patient with medical complications or suspected preterm labor. Her care is automatically compromised by the absence of a prenatal record. Even for the patient presenting at term to the labor floor, the reality is that given the hundreds of records being sent by various practices to our labor floor each month, it is not uncommon for an individual patient's record to become misplaced somewhere between the outpatient office and the labor floor.
A computer-based prenatal record that is available over a network of computers can solve all of these problems. Additionally, our electronic medical record has additional important benefits such as improved legibility of the charts, a problem-oriented design in the organization of the chart, suggested plans of management, and reporting functions that allow for ease in auditing charts prospectively instead of manually abstracting data from a sample of charts.
Although the computerized chart in this study offers enhanced communication among providers, there is always a concern that the improved availability of medical records as a result of this technology will lead to breaches in patient confidentiality. This issue has been addressed in several ways at our institution. First, access to patient records is available only by a secure network operated by the medical center. Second, access to the patient chart requires that a provider sign in with a user identification and password. This makes it possible to audit who has accessed any given patient chart. Third, because efficient users tend not to want to shut the software down when they leave a workstation, our software package has a safeguard feature that allows a user to lock the program. The software can only be unlocked when the provider reenters their user ID and password. In this way, time lost in restarting the program is avoided. And finally, different types of users can be limited in how much access they can have to patient charts. For example, front desk clerks can be limited to only seeing patient demographics when they log in to a chart, whereas physicians can view the entire chart. These features in many ways actually offer superior levels of protection to the confidentiality of patient information than does the existing paper-based system.
Another potential drawback of an electronic medical record is that there may be times when the software is unavailable due to a malfunction, computer virus infection, need for a software update, or the like. During the period of this study, the electronic record was nearly always available. It even continued to function during the blackout that engulfed the Northeast United States in August 2003 (the medical center computer network was maintained using generator power). Nevertheless, we have had several scheduled and unscheduled downtimes in the availability of our server since we began using this software. Scheduled downtimes have been brief (never more than an hour) and occur after office hours are completed to minimize impact. Paper versions of the electronic chart are printed out at specific times during the pregnancy as a back up that can be sent to the labor floor if needed, and policies exist to convert to using paper record for the occasions when the computerized system is unexpectedly not available. Data collected on paper can subsequently be entered into the electronic version of the chart.
Electronic medical records such as ours hold the potential to improve communication of other data collected during the course of prenatal care, such as laboratory data and consultant reports. At our institution, most laboratory test results are sent to our medical center and are reported to the referring sites and the labor floor by a computerized information system that predates our electronic medical record and is therefore widely available within our network. Laboratory data received from outside centers can be manually entered into our prenatal software, and ultimately, both the need and potential exists to create electronic interfaces to allow the automatic importation of the received data into our system. Similarly, it is our plan to incorporate the ability to scan consultant reports that were prepared for by providers without access to our software into the system so that images of these reports can be viewed in the electronic medical record at all locations where the electronic record is available.
In conclusion, this study has taken an important first step to demonstrate some of the potential benefits of electronic medical records in obstetric care. We hypothesize that the ultimate benefit of a paperless record may be in its implementation across a large network of providers. In this setting, auditing of all patient records, rather than the cumbersome process of abstracting a small sample of charts, may be easily accomplished in a timely, even prospective fashion, and quality of care initiatives can be implemented and thoroughly monitored.
Beyond improving the quality of care for prenatal patients, given that obstetrics is such a disproportionately large contributor to the current medical malpractice crisis, we believe that improving the quality of care and of record keeping could potentially mitigate medicolegal risks. It is quite possible that the costs of implementing such a system might be offset by reduced malpractice costs as well as by preventing the incurred costs that result from treating patients harmed by medical errors. This small study suggests an example of how patient care could be improved and legal liability reduced: Consider the patient who had an ultrasonogram the day before presenting in labor that suggests that her fetus may be macrosomic. The provider who does not have access to this report runs a greater risk of encountering a bad newborn outcome than a provider who has immediate access to that ultrasound report.
Future studies of this and similar information systems should begin to look at whether the benefits of these systems actually translate into evidence of improved health care outcomes for patients. We hope that such evidence will lead to a nationwide investment in the development and replication of evidence-based information technology initiatives.