- Email: [email protected]
Getting Doctors to Report Medical Errors: Project DISCLOSE
Reporting Systems
Getting Doctors to Report Medical Errors: Project DISCLOSE ince the 1999 Institute of Medicine (IOM) report To Err is Human1 there has been an increased recognition of the importance of reporting errors and a focus on improving patient safety in hospitals. Yet a recent status report on patient safety systems indicates that progress, including adverse event and error reporting, is slower than expected.2 At our academic medical center—the Temple University Hospital (Philadelphia)—an incident, also referred to as an “event” or an “occurrence,” was defined as “any untoward event or potential event, which is not consistent with the routine operation of the hospital and/or which affects a patient, visitor, or staff member.”3 This definition encompasses a broad array of processes from minor to life threatening. Incident reports are used by health care workers in hospitals to identify problems, faulty processes, and undesired outcomes that occur.4 The focus is on gathering data on adverse events (defined as an injury caused by medical management rather than the underlying condition of the patient) and near misses (defined as any process variation that did not affect the outcome, but for which a recurrence carries a significant chance of serious adverse outcome).5 The goal is to analyze the data collected, identify flaws in hospital systems, and make changes to prevent future occurrences.4 The critical importance of effective data collection on incidents has been emphasized by the investigators in the VHA hospitals’ patient safety system, who have adopted the mantra “you can’t fix what you don’t know about.”5 Unfortunately, earlier data suggest that only 1.5% of adverse events result in a physician incident report and that most hospital-based events actually go unreported.6
S
382
July 2006
Emmanuel S. King, M.D. Darilyn V. Moyer, M.D. Michael J. Couturie, M.D. John P. Gaughan, Ph.D. David J. Shulkin, M.D.
Article-at-a-Glance Background: Despite the number of patient safety incidents that occur in hospitals, physicians currently may not have the ideal incident reporting tools for easy disclosure. A study was undertaken to assess the effectiveness of a simplified paper incident reporting process for internal medicine physicians on uncovering patient safety incidents. Design: Thirty-nine internal medicine attending physicians were instructed to incorporate the use of a simplified paper incident reporting tool (DISCLOSE) into daily patient rounds during a three-month period. All physicians were surveyed at the conclusion of the three months. Results: Compared with physician reporting via the hospital’s traditional incident reports from the same time period, a higher number (98 incidents versus 37; a 2.6-fold increase) of incidents were uncovered using the DISCLOSE reporting tool in a larger number of error categories (58 versus 14, a 4.1-fold increase). When reviewed and classified with a five-point harm scale, 41% of events were judged to have reached patients but not caused harm, 33% to have resulted in temporary harm, and 9% of reports, though not considered events, were to indicate a “risky situation.” Surveyed physicians were more satisfied with the process of submitting incident reports using the new DISCLOSE tool. Discussion: A simplified incident reporting process at the point of care generated a larger number and breadth of physician disclosed error categories, and increased physician satisfaction with the process.
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
In many hospitals, incident reporting is paper based and requires the health care professional to locate an incident report form at the nursing station, complete a series of questions, obtain additional signatures, and then forward the form to the risk management or patient safety operations office within the hospital. Computer-based reporting systems also exist, including our state’s Pennsylvania Patient Safety Reporting System (PAPSRS). Health care organizations in the state of Pennsylvania have submitted mandatory error reports via this system since June 2004, in accordance with changes stemming from the Medical Care Availability and Reduction of Error (MCARE) Act.7,8 Recent evidence seems to indicate that phasing in such electronic systems may be a worthwhile patient safety intervention and may have distinct advantages over paper-based systems.9 At Temple University Hospital, a traditional paperbased incident report has been the primary event reporting process. Event report forms, available at nurses’ stations throughout the hospital, are intended for use by physicians, nurses, and other health care professionals to fulfill mandatory internal reporting requirements. Staff are also required to report events that have caused (or have significant potential to cause) patient injury within 24 hours by telephone to risk management. Risk managers are available 24 hours a day to all staff, although hospital policy prohibits such verbal disclosure from replacing the need for a physical incident report.3 Medical staff and housestaff are periodically reminded about their duty to report incidents, with particular emphasis on hospital and state policies, at hospital-wide forums held by risk management and legal counsel. In 2004, the hospital received 10,020 incident reports, of which only 96 (approximately 1%) came from physicians. The vast majority of incident reports come from nursing staff and other ancillary health professionals. In a recent revealing study, it appears that even patients may report more frequently than physicians. Hospitalized patients reported a number of adverse events and near misses, many of which were confirmed on chart review, but none of which had been disclosed by physicians in the hospital incident reporting system.10 Clearly, current efforts toward increasing physicians’ incident reporting and error disclosure may not be paying off.
July 2006
The exact reason why physician reporting is so poor remains unclear and strong data is lacking. Physicians have traditionally been reluctant to shed light on such issues as iatrogenic complications or system problems for fear that such information may lead to lawsuits or professional embarrassment—despite evidence that disclosure may actually decrease the likelihood of lawsuits.11 Attending physicians may be particularly reluctant to disclose, except when bad outcomes or clear-cut “protocol violations” are present.12 Yet barriers to reporting may simply reflect such issues as uncertainty about what to report, lack of time, or forgetting to report.13 In addition, many physicians may not define error disclosure as their primary responsibility.14 The problem may even start in training.15–17 In Fall 2004, as part of ongoing quality improvement (QI) activities, we sought to evaluate whether a simplified, user-friendly process for paper incident reporting could improve the frequency and quality of physician reporting. We also attempted to assess the clinical importance and impact on patients of events disclosed by physicians. Most importantly, we hoped to identify physician-specific barriers to incident reporting that might be amenable to further intervention.
Methods Disclose Tool As a QI initiative, the new pilot paper incident reporting form, the DISCLOSE (Drugs, Iatrogenic complications, System errors, Communication errors, Lab and test issues, Oversights, Staffing issues, and Equipment issues) tool, was developed by two of the co-investigators [D.S., D.M.] It was formatted into pads of paper that fit easily in the physician’s white coat (Appendix 1, page 392). Physicians were instructed to circle a category and if appropriate, a subcategory, for each event. In cases where no category was chosen but rather only a written description of the event was recorded, incidents were classified into one or more categories of error by data entry personnel. The DISCLOSE mnemonic prompted the user to look for some of the most commonly encountered errors in hospital care in the specified areas.
Participants All physicians who participated in this QI project were full-time academic physicians in Temple University
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
383
Table 1. Harm Scale Categories
Hospital and Temple University School of Medicine. As part of their academic duties, generalists and specialists serve as teaching physicians on general medical services in the internal medicine department. During a threemonth period (September–December 2004), these general medical service attending physicians were introduced to the use of this alternatively formatted incident reporting form at monthly departmental in-services. Physicians were instructed to incorporate the use of the tool into daily patient rounds with residents and students as seamlessly as possible. They were told to use the DISCLOSE tool as if they were submitting reports of medical errors or incidents that they or their housestaff encountered on the inpatient service. All participants were reminded throughout the three months, however, that the DISCLOSE tool was for our pilot project only and did not fulfill hospital mandatory reporting requirements via incident reports. We emphasized that traditional hospital incident reports still needed to be submitted as mandated by hospital protocol or state law and that submitting a DISCLOSE form did not adequately fulfill those reporting requirements.
patient) were immediately brought to the attention of risk management; all submitted reports are examined and then forwarded to appropriate hospital committees.
Instructions
Data Collection
The concept of DISCLOSE was briefly introduced at the in-service and in a departmental memorandum that was circulated before the start of the project. Formal attempts at more extensive instruction and education regarding patient safety, medical errors, definitions of what needed to be reported, and the tool’s specific categories were not undertaken to better assess true changes in reporting patterns from baseline. Participants were urged to provide detailed narrative information regarding the incident on both the form and all concurrent hospital incident reports. Although we instructed participating physicians to submit traditional incident reports whenever they submitted DISCLOSE forms, any events reported with the DISCLOSE tool that qualified as potential “sentinel events” (defined as an unexpected occurrence involving death or serious physical or psychological injury) or “serious events” (defined as an event, occurrence, or situation involving the clinical care of a patient in a medical facility that compromises patient safety and results in an unanticipated injury requiring the delivery of additional health services to the
At the conclusion of the three months, the number and categories of various incidents reported using the DISCLOSE reporting tool were compared with the data from the hospital’s existing paper incident reporting database for the same time period. The narrative portion of all reports was analyzed, and the incidents were recategorized if necessary to conform as much as possible to the categories listed in the DISCLOSE tool. Because the patient’s name and other identifying information was frequently not available on DISCLOSE submissions, the narrative descriptions of the events reported on DISCLOSE forms were compared against the hospital’s risk management database to determine if these events appeared to have been previously reported. At the conclusion of the three months, incidents reported via the DISCLOSE tool were also retrospectively classified into harm categories by one of the physician investigators [E.K.] and two other physicians who had not been involved in reporting during the project period. The five categories, recently validated by another group of investigators,18 are described in Table 1 (above).
384
July 2006
1.
An event occurred and resulted in death to the patient. 2A. An event occurred, patient was harmed, but harm did not result in death (temporary harm/complete recovery expected). 2B. An event occurred, patient was harmed, but harm did not result in death (permanent harm/complete recovery not expected). 3. An event occurred and reached the patient, but the patient was not harmed. 4. An event occurred but was caught before affecting/reaching the patient. 5. No event occurred, but a risky situation was found that increased the risk for an event to occur. N/A No event occurred, and no risky situation in place that increased risk for events.
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
Table 2. Summary of Errors Reported by Physicians, by Category with DISCLOSE Tool Category
Description Dosage, timing, route, rate, omitted doses, allergies, reactions, wrong Drugs drug, wrong patient Iatrogenic Complications from procedures, treatments System Issues Delays, missed treatments, order entry Communication Identification issues, confusion over orders, failure to give information Lost/mislabeled specimens/films, results not reported, delays, Lab/Tests improper studies Oversights Judgment issues, missed diagnoses, deviations from standard of care Staff Nursing, respiratory, consultants, transport Equipment Delays in equipment, wrong equipment, failures, supply problems Total
Number Proportion of Total 26
15%
6 25 24
3% 14% 14%
14
8%
15 59 7 176 *
9% 34% 4%
* The 98 discrete incident reports were categorized into 176 categorized incidents.
The three physicians reviewed the data twice, with cases in random order each time. Median percentages were calculated in each harm category, for individual raters, as well as an overall median percentage for all three raters. A statistician then calculated the inter- and intrarater variability based on an Analysis of Variance on ranks, followed by calculation of the intraclass coefficient, and confidence intervals.19 Because of restructuring in the risk management department, we were unable to access their data regarding the severity of incidents reported in the traditional incident reporting system for purposes of comparison. We surveyed physicians at the end of the data collection period regarding their reporting experience.
ple categories after review. Staff-related events were the most frequently reported category (which did not involve physicians), comprising 59 incidents (34%). Examples of incidents from all categories reported using the DISCLOSE tool are shown in Table 3 (page 386). Table 4 (page 387) breaks down reported events in further detail (subcategory) under each category of error in the DISCLOSE tool; only 10 of the common subcategories are shown. The most commonly reported subcategorized events were delays attributed to staff, accounting for 20 (11%) of all reports. Table 5 (page 388) contains some examples of DISCLOSE incidents in the staff category, and Table 6 (page 389) lists examples for the communication category.
Results Categories of Events
Harm of Events
Given that none of the DISCLOSE events could be definitively linked to a concurrent incident report by a physician or other healthcare worker, the errors identified by physicians using the DISCLOSE tool were presumed to have not been previously reported. The 39 attending physicians on general internal medical service from September 15, 2004 to December 15, 2004 reported 98 discrete incidents using the DISCLOSE tool. Thirteen physicians identified themselves and were responsible for 67 (69%) of the reports. Table 2 (above) summarizes the categories of errors reported, which totaled 176 because many errors fit under multi-
Retrospective application of the five-point harm scale to the DISCLOSE database indicated the following results: ■ 41% of events were felt to have reached patients but not caused harm ■ 33% of events were felt to have resulted in temporary harm ■ 9% of reports, though not considered events, were felt to indicate a “risky situation” ■ 2% of events were felt to have been caught before reaching the patient ■ 1.4% of events were felt to have contributed to permanent harm not resulting in death
July 2006
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
385
Table 3. Examples of Various Incidents Reported with DISCLOSE Tool for Each Category Harm Scale Rating* Drugs Patient given 9,000 units/hour of heparin instead of 900 units/hour. No bleeding resulted. 3 Central line thought to be in femoral vein discovered at time of removal one week later Iatrogenic 2A to be in femoral artery. No complications, patient tolerated removal. Took extended time to get “stat” abdominal radiograph to confirm nasogastric tube placement. Patient misses scheduled cardiac medications and despite intravenous Systems 2A administration of furosemide and antihypertensive agents, develops severe pulmonary edema, hypertensive emergency. Requires transfer to intensive care unit. Patient transferred from outside hospital to emergency department (ED), then hospital room. Transfer center notified ED staff, but neither ED staff nor transfer 2B Communications center notifies internal medicine housestaff. Patient, in room four hours without evaluation, is later found to have had sustained stroke and rhabdomyolysis (exact timing undetermined). Urinalysis on febrile patient sent, received in lab but no result for two days, and test then 2A Labs/Tests “auto-cancelled,” leading to treatment delay. Patient’s condition did not deteriorate. Patient with altered mental status falls twice. After first fall, fall precautions were 2A Oversights ordered by physician. Second fall results in minor injury. Staff Team waited three days for surgical consult. 5 Equipment Air mattress spontaneously deflates 1–2 times daily, causing significant patient discomfort. 2A Category
Description
* Shown in Table 1 (page 384).
■ 11.6% of reports were felt to have not actually constituted events or patient safety incidents. None of the events were felt to have directly resulted in death to a patient. Based on the results of the analysis of variance on ranks (ANOVA), the ICC as a measure of inter-rater variability was 0.8413 (95% confidence interval, 0.7432–0.8991). For intra-rater variability, the ICC was 0.8687 (95% confidence interval, 0.8165–0.9079).
Traditional Incident Reports For the total of 1,123 reports received via the traditional hospital incident reporting system received for the three-month study period, only 37 (3.3%) were submitted by physicians. Staff nurses (731, 65.1%), laboratory personnel (108, 9.6%), and pharmacy staff (54, 4.8%) were the top reporters by professional category. Reporters submitted in a total of 14 categories. Drug errors (13, 35% of physician reports) and delays (9, 24%) constituted the categories most frequently reported by physicians, who also cited iatrogenic complications (5, 14%), oversights (2, 5%), and equipment failure (2, 5%).
386
July 2006
Nurses and other nonphysician health care professionals submitted a total of 1,123 incident reports. Nurses submitted 731 incident reports, including patient falls (108, 15%), compromised skin integrity (104, 14%), oversights (96, 13%), iatrogenic events (91, 12%), and drug errors (80, 11%).
Experience of Reporting All 39 physicians who had attended on general internal medicine service were surveyed on their experiences regarding incident reporting at the conclusion of the three months. Twenty-five (64%) returned the questionnaires. More than half of the respondents (13; 52%) felt that the DISCLOSE tool was easier to use than the traditional incident report, and 10 (40%) found the DISCLOSE tool less threatening than existing reports. The most common reasons limiting incident reporting were time constraints (15, 60%), difficulty locating forms (13, 52%), lack of closure/feedback (8, 32%), and “did not find it important” (3, 12%). Only one physician (4%) cited “fear of disclosure to risk management” as a limiting factor.
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
Table 4. Summary of Most Common Subcategories of Error Reported by Physicians with DISCLOSE Tool
in addition to risk management, was less punitive and more likely to result in reform. Physicians, who Category Subcategory Number Proportion of Total frequently cite lack of time and lack Missed dose 4 2% of faith that any change will be Drugs Wrong dose 5 3% made based on their incident Wrong patient 3 2% reports, may not realize that the Iatrogenic Procedure 4 2% errors they feel are too minor to System Issues Delays 17 10% report that are often the best targets Communication Failure to give information 12 7% for systemwide or more focused Labs/Tests Delays 8 5% intervention.22 Oversights Judgment 8 5% The fact that the majority of the Staff Delays 20 11% DISCLOSE events that had either Equipment Supply 5 3% harmed or had the potential to harm our patients had presumably not Discussion been disclosed by any other means underscores the Physicians working on inpatient medical services are importance of our findings. Physicians who used the likely to have a unique perspective on patient care. They DISCLOSE tool seem much more apt to report all events, are often more privy to problems and errors than other from minor to life threatening. Increasing the pool of health care workers because of their primary role in iniincident reports is the first, though obviously not the tiating patient care plans, requesting procedures and only, step toward recognizing patterns of medical errors consultations, and their close working relationship with and identifying systems errors for intervention. Whether housestaff. This makes physicians a key but often adequate measures are in place once reporting signifiunderused source of incident reports. The challenge, cantly increases, and how to sustain initial increases in however, lies in finding a way to overcome traditional reporting, however, will be the most important factors in barriers to reporting such as time constraints, difficulty ensuring success.23 locating forms, fear of disclosure to risk management, Physicians using the DISCLOSE tool also tended to ignorance of reporting requirements, and dissatisfaction report on broader and distinct types of incidents than with lack of feedback or closure.20 Moreover, despite did other health care workers. Of particular interest is growing evidence that interventions to decrease medthe fact that they frequently reported errors by houseical error may be effective, many physicians are still staff in the staff category, even when housestaff were 21 skeptical. The importance of increasing physician inciexcluded from the category. In the communications catdent reporting, especially with mandatory institutional, egory, a number of reports highlighted physician frustrastate, and federal reporting on the rise, cannot be tion at orders not being filled properly or at not being overemphasized. notified of unexpected delays. Because of these findings, The finding that none of the 98 incidents reported via in subsequent forums, we have explicitly instructed the DISCLOSE tool could be linked to reports filed using physicians to address all staff-related incidents to approthe traditional reporting mechanism suggests that they priate supervisors or department heads in addition to represent new disclosure of events that would not othsubmitting incident reports. erwise have been uncovered. We believe that the Meanwhile, the 37 incident reports submitted by increase in reporting arises largely from the simplifiphysicians via the hospital’s incident reporting syscation of the error reporting process, reflecting the tem were significantly outnumbered by the number DISCLOSE tool’s portability, with minimal disruption of of reports from nonphysician healthcare workers. the patient care routine. We also believe that physicians Nurses, in particular, reported nearly 20-fold greater than physicians during the three-month study period, a perceived that a form going to our physician investigators,
July 2006
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
387
Table 5. Examples of DISCLOSE Incident Reports in Staff Category* Description
Other Categories
Intern substituted 4 mg hydromorphone IV for 4 mg PO. Patient difficult to arouse. Attending physician never called to notify of patient intubated and transferred to ICU. Patient has chest tube placed for pneumothorax by emergency department staff. When admitting housestaff arrive, chest tube noted to be disconnected (timing uncertain) and patient has recurrent pneumothorax. Strict In’s and Out’s not recorded; patient given “bathroom privileges” with diagnosis of congestive heart failure. IV nurse fails to replace IV after pulled out by patient, causing missed drug dose. No medications sent from pharmacy four hours after ordered. Not notified by nursing of systolic blood pressure 240. 24-hour urine collection ordered, not started by nurse until 3 days later.
Drug—wrong dose Oversight—judgment Communication Oversight—patient Observation Communication— confusion over ordering System—missed treatment System—delay Oversight—judgment Lab/Test—delay
Harm Scale Rating† 2A N/A 2A 5 2A 5 5 N/A
* IV, intravenous; PO, orally; ICU, intensive care unit. †
Shown in Table 1 (page 384).
rate that has been stable for some time. They tended to report more on protocol-driven issues such as falls, skin integrity, and patients who signed out against medical advice than physicians. This dramatic difference in reporting behavior illustrates a fundamental difference between physicians’ and nurses’ attitudes and perhaps the expectations of their supervisors and peers.12 In a review of nursing incident report data, 88% of surveyed nurses thought it was “important to report all incidents” and 72% stated that they “always fill out incident reports,” even though 17% believed that doing so would be used against them in an evaluation.24 It would be encouraging if more physicians embraced this dedication to reporting, especially if they continued to report on issues that are visible only from their unique vantage point. Other institutions have likewise reported on their efforts to address incident reporting. For example, Nast and colleagues introduced a simplified patient safety event reporting tool in a cardiothoracic surgery intensive care unit. Of the 163 reports submitted over six months, at least 69% were from nurses and only 19% were from physicians. The majority (54%) of events caused no harm despite “reaching the patient.” The remainder ranged from “risky situations” (15%), all of which were reported by nurses, to near misses (10%), to
388
July 2006
temporary harm (22%). No events reported caused permanent harm or death. These results are largely in concert with our findings that many events either reach patients or have significant potential for harm. Many different harm scales exist, but perhaps the most clinically useful are those that stratify not only injury but the sometimes intangible potential to predispose to injury. Some may contend that the potential to cause harm is in the eye of the beholder, but we believe that the more frequently this potential is investigated, the safer a system will become. Perhaps this project’s most telling findings lie in the participating physicians’ comments themselves. Most of those surveyed had never previously used the traditional incident reporting mechanism. Their reasons included time constraints, difficulty locating forms, lack of closure/follow-up, and “did not find it important.” Among those who submitted DISCLOSE forms, most found the new tool easier to use. Asked why, one physician commented, “Because I had the forms. I knew exactly where they were.” She also appreciated that the new tool enabled her to report events that “did not necessarily lead to adverse health or clinical outcomes… I do not call risk management unless the patient suffered or deteriorated…” Thirty-two percent of our surveyed physicians also cited lack of feedback
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
Table 6. Examples of DISCLOSE Incident Reports in Communication Category* Description Initial plan to discharge patient, order placed. Based on study result, plan changed and patient was to stay for further workup. Despite order being cancelled, nurse continues discharge protocol and patient sent home. Discharge medication prescribed at wrong dose contributing to readmission. Patient with new diagnosis of diabetes, discharged with insulin prescription but neither prescription nor discharge instruction has dose, causing patient confusion. Patient admitted to hospital room at 7 P.M. No housestaff notified, and patient not assessed until 9:30 A.M. the next day. Attending physician given verbal report that transesophageal echocardiogram was “negative.” Official dictated report mentions cardiac thrombus, without any follow-up call to medicine attending by cardiologist. Stress test delayed, cancelled. Team not notified, no reason given by department. Patient kept NPO* for prolonged period.
Other Harm Scale Categories Rating†
None
3
Drugs— Wrong dose
2A
Drugs— Erroneous prescription Oversight
3
Oversight
3
Staff— Consultants
3
System— Delay
3
* NPO, nothing by mouth. †
Shown in Table 1 (page 384).
from risk management as a deterrent to active error reporting. Where physician reports go and how they are handled may be an important component of designing future effective strategies to improve physician reporting.
July 2006
The project had some significant limitations. First, the study period was limited to three months and to one institution. We wanted to quickly assess the feasibility and benefits of a simple QI intervention and then adopt it soon after. Although we met our goals, the brevity of our project may limit extrapolation to larger forums. Second, given the lack of adequate patient identifiers and event-specific data (that is, date, time, place) on DISCLOSE submissions, the “matching” process and comparison with the risk management incident reporting database may not appear ideal. However, given the reasonable number of physician reports in both databases, we were able to review all reports individually rather than relying on merely querying a large database. On the basis of the narrative descriptions, we felt confident that all DISCLOSE-submitted incidents were in fact distinct, newly reported events. If someone other than a physician filled out an incident report, however, this may not have been captured in our review because of the high volume of nonphysician reports. The difficulty we encountered in obtaining highly specific event data was mirrored in the high number of submissions from a relatively small subset (13 out of 39) physicians, the remainder of which chose to remain anonymous. It is not surprising, given the baseline institutional milieu, that these obstacles emerged. In fact, much of our current effort is focused on emphasizing to the medical staff that incident reports are a key internal reporting mechanism that require not only physician self-identification but also patient and event-specific information to fulfill our current institutional and state disclosure and reporting policies. In the future, we expect that our educational efforts will significantly improve the quality and consistency of reporting data. Third, the retrospective application of our harm scale to assess the impact of the occurrences on our patients was limited by the level of detail in submitted reports. Previous studies allowed reporters to designate the suspected level of risk on their form in a prospective fashion, which would certainly reduce recall bias.18 We were also not always able to discern the specific outcome and level of harm, making this analysis imperfect at times. However, as shown, the reviewing physicians, two of whom lacked specific training in patient safety activities, were able to classify the events with some
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
389
degree of consistency. As the culture of patient safety in our institution grows, we hope that reporters will include much more event-specific data, especially regarding outcomes. In the future, we plan to prospectively capture and apply our harm scale, with medical record review as needed for clarification, in order to yield even more useful data.
Postscript Following the data collection for the study, we presented an interim progress report in a hospitalwide forum open to medical staff, residents, and medical students. The positive comments and questions we received suggest widespread interest in both improving the current incident reporting system and developing effective interventions to improve patient safety. In fact, system errors and discrete incidents uncovered by the DISCLOSE tool have already served as an impetus for improved medical education in our residency program in peer-protected environments. In the short term, a number of the events uncovered by the DISCLOSE tool have already been used in patient safety and morbidity and mortality conferences to further encourage open discussion on medical error. We are in the process of implementing a new patient safety curriculum for our residents to introduce basic patient safety and medical error terminology, root cause analysis, and open discussion of cases of error uncovered by the DISCLOSE tool in small groups as well as in a revised, nonpunitive, departmentwide morbidity and mortality conference. With growing emphasis on identifying and attacking areas of preventable error, such as the Institute for Healthcare Improvement’s 100K Lives Campaign,25 the importance of equipping physicians with tools such as DISCLOSE to uncover and facilitate open discussion about medical errors cannot be overemphasized. In the near future, the health system’s new computer-based reporting system, which will likely eventually replace our paper-based system, is set to be unveiled, and its potential to facilitate both reporting and dissemination of event data and to expedite processing
390
July 2006
and enable quick intervention seems promising. Recent successful efforts at “going paperless” by implementing electronic and Web-based incident reporting have been published.26,27 We are confident that the use of a more user-friendly error reporting tool, whether it is ultimately paper or computer-based, combined with the increased awareness of the importance and duty to disclose errors, will sustain our efforts.
Conclusion Project DISCLOSE’s findings suggest that strategies can be successfully established to increase physician incident reporting. Simple reporting mechanisms such as the DISCLOSE tool are likely to lead to improved physician reporting of clinically significant incidents with potential for major patient harm and increased satisfaction with the process. J The authors gratefully acknowledge the help of Raymond Gathiuni, M.D., Kathleen Coppola, M.D., Mary Kraemer, M.D., Rebecca Horvath, M.A.S., J.D., and Marie Swety, R.N., J.D. Dr. Gathiuni helped with the initial literature review. Drs. Coppola and Kraemer generously devoted time to reviewing the data and applying the five-point harm scale. Ms. Horvath helped plan the initial study. Ms. Swety was instrumental in coordinating the review of risk management data.
Emmanuel S. King, M.D., is Assistant Professor of Medicine and Assistant Program Director for Patient Safety Education in Internal Medicine Residency, Department of Internal Medicine, Section of Hospital Medicine, Temple University Hospital, Temple University School of Medicine, Philadelphia. Darilyn V. Moyer, M.D., is Associate Professor of Medicine and Program Director, Internal Medicine Residency, and Michael J. Couturie, M.D., is a Resident, Department of Internal Medicine. John P. Gaughan, Ph.D., is Associate Professor, Epidemiology and Biostatistics, Temple University School of Medicine. David J. Shulkin, M.D., formerly Chief Medical Officer and Patient Safety Officer, Temple University Hospital, is President and Chief Executive Officer, Beth Israel Medical Center, New York City, and a member of the Joint Commission Journal on Quality and Patient Safety’s Editorial Advisory Board. Please address reprint requests to David J. Shulkin, M.D., [email protected].
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
References 1. Kohn L.T., Corrigan J.M., Donaldson M.S.: To Err Is Human: Building a Safer Health Care System. Washington, D.C.: National Academy Press, 1999. 2. Longo D.R., et al.: The long road to patient safety: A status report on patient safety systems. JAMA 294:2858–2865, Dec. 14, 2005. Erratum in: JAMA 295:164, Jan. 11, 2006. 3. Temple University Hospital Administrative Policies and Procedures: Patient Safety Plan (internal document). No. 950.140, effective Dec. 2001. 4. Shojania K.G., Wald H., Gross R.: Understanding medical error and improving patient safety in the inpatient setting. Med Clin N Am 86:847–867, Jul. 2002. 5. Bagian J.P., et al.: Developing and deploying a patient safety program in a large health care delivery system: You can’t fix what you don’t know about. Jt Comm J Qual Improv 27:522–532, Oct. 2001. 6. Milch C.E., et al.: Voluntary electronic reporting of medical errors and adverse events: An analysis of 92,547 reports from 26 acute care hospitals. J Gen Intern Med 21:165–170, Feb. 2006. Epub Dec. 2005. 7. Pennsylvania Patient Safety Authority. http://www.psa.state.pa.us (last accessed May 3, 2006). 8. Medical Care Availability and Reduction of Error (MCARE) Act: P.L.154, No. 13. Mar. 20, 2002. 9. Tighe C.M., et al.: Incident reporting in one UK accident and emergency department. Accid Emerg Nurs 14:27-37, Jan. 2006. 10. Weingart S.N., et al.: What can hospitalized patients tell us about adverse events? Learning from patient-reported incidents. J Gen Intern Med 20:830–836, Sep. 2005. 11. Guadognino C.: Effects of medical error disclosure and apology [interview with Albert Wu, M.D., M.P.H.]. Physician’s News Digest, Feb. 2005. http://www.physiciansnews.com/spotlight/205.html (last accessed May 3, 2006). 12. Lawton R., Parker D.: Barriers to incident reporting in a healthcare system. Qual Saf Health Care 11:15–18, Mar. 2002.
July 2006
13. Nast P.A., et al.: Reporting and classification of patient safety events in a cardiothoracic intensive care unit and cardiothoracic postoperative care unit. J Thorac Cardiovasc Surg 130:1137, Oct. 2005. 14. Robinson A.R., et al.: Physician and public opinions on quality of health care and the problem of medical errors. Arch Intern Med 162:2186–2190, Oct. 28, 2002. 15. Wu A., et al.: Do house officers learn from their mistakes? JAMA 265:2089–2094, Apr. 24, 1991. 16. Aron D.C., Hendrick L.A.: Educating physicians prepared to improve care and safety is no accident: It requires a systematic approach. Qual Saf Health Care 11:168–173, Jun. 2002. 17. Waring J.J.: A qualitative study of the intra-hospital variations in incident reporting. Int J Qual Health Care 16:347–352, Oct. 2004. 18. Osmon S., et al.: Reporting of medical errors: An intensive care unit experience. Crit Care Med 32:727–733, Mar. 2004. 19. Shrout P.E., Fleiss J.I.: Intraclass Correlations: Uses in assessing rater reliability. Psychol Bull 86:420–428, Mar. 1979. 20. Hebert P.C., Levin A.V., Robertson G.: Bioethics for Clinicians: 23. Disclosure of medical error. CMAJ 164:509–513, Feb. 20, 2001. 21. Rosen A.B., et al.: Physicians’ views of interventions to reduce medical errors: Does evidence of effectiveness matter? Acad Med 80:189–192, Feb. 2005. 22. Mills P.D., et al.: Using aggregate root cause analysis to reduce falls. Jt Comm J Qual Patient Saf 31:21–31, Jan. 2005. 23. Brennan T.A., et al.: Accidental deaths, saved lives, and improved quality. N Engl J Med 353:1405–1409, Sep. 29, 2005. 24. Elnitsky C., Nichols B., Palmer K.: Are Hospital Incidents Being Reported? J Nurs Adm 27:40–46, Nov. 1997. 25. Institute for Healthcare Improvement: 100K Lives Campaign. http://www.ihi.org/IHI/Programs/campaign (last accessed May 3, 2006). 26. Dixon J.F.: Going paperless with custom-built Web-based patient occurrence reporting. Jt Comm J Qual Improv 28:387–395, Jul. 2002. 27. Kivlahan C., et al.: Developing a comprehensive electronic adverse event reporting system in an academic medical center. Jt Comm J Qual Improv 28:583–594, Nov. 2002.
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
391
Appendix 1. DISCLOSE Incident Report
392
July 2006
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
Getting Doctors to Report Medical Errors: Project DISCLOSE ince the 1999 Institute of Medicine (IOM) report To Err is Human1 there has been an increased recognition of the importance of reporting errors and a focus on improving patient safety in hospitals. Yet a recent status report on patient safety systems indicates that progress, including adverse event and error reporting, is slower than expected.2 At our academic medical center—the Temple University Hospital (Philadelphia)—an incident, also referred to as an “event” or an “occurrence,” was defined as “any untoward event or potential event, which is not consistent with the routine operation of the hospital and/or which affects a patient, visitor, or staff member.”3 This definition encompasses a broad array of processes from minor to life threatening. Incident reports are used by health care workers in hospitals to identify problems, faulty processes, and undesired outcomes that occur.4 The focus is on gathering data on adverse events (defined as an injury caused by medical management rather than the underlying condition of the patient) and near misses (defined as any process variation that did not affect the outcome, but for which a recurrence carries a significant chance of serious adverse outcome).5 The goal is to analyze the data collected, identify flaws in hospital systems, and make changes to prevent future occurrences.4 The critical importance of effective data collection on incidents has been emphasized by the investigators in the VHA hospitals’ patient safety system, who have adopted the mantra “you can’t fix what you don’t know about.”5 Unfortunately, earlier data suggest that only 1.5% of adverse events result in a physician incident report and that most hospital-based events actually go unreported.6
S
382
July 2006
Emmanuel S. King, M.D. Darilyn V. Moyer, M.D. Michael J. Couturie, M.D. John P. Gaughan, Ph.D. David J. Shulkin, M.D.
Article-at-a-Glance Background: Despite the number of patient safety incidents that occur in hospitals, physicians currently may not have the ideal incident reporting tools for easy disclosure. A study was undertaken to assess the effectiveness of a simplified paper incident reporting process for internal medicine physicians on uncovering patient safety incidents. Design: Thirty-nine internal medicine attending physicians were instructed to incorporate the use of a simplified paper incident reporting tool (DISCLOSE) into daily patient rounds during a three-month period. All physicians were surveyed at the conclusion of the three months. Results: Compared with physician reporting via the hospital’s traditional incident reports from the same time period, a higher number (98 incidents versus 37; a 2.6-fold increase) of incidents were uncovered using the DISCLOSE reporting tool in a larger number of error categories (58 versus 14, a 4.1-fold increase). When reviewed and classified with a five-point harm scale, 41% of events were judged to have reached patients but not caused harm, 33% to have resulted in temporary harm, and 9% of reports, though not considered events, were to indicate a “risky situation.” Surveyed physicians were more satisfied with the process of submitting incident reports using the new DISCLOSE tool. Discussion: A simplified incident reporting process at the point of care generated a larger number and breadth of physician disclosed error categories, and increased physician satisfaction with the process.
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
In many hospitals, incident reporting is paper based and requires the health care professional to locate an incident report form at the nursing station, complete a series of questions, obtain additional signatures, and then forward the form to the risk management or patient safety operations office within the hospital. Computer-based reporting systems also exist, including our state’s Pennsylvania Patient Safety Reporting System (PAPSRS). Health care organizations in the state of Pennsylvania have submitted mandatory error reports via this system since June 2004, in accordance with changes stemming from the Medical Care Availability and Reduction of Error (MCARE) Act.7,8 Recent evidence seems to indicate that phasing in such electronic systems may be a worthwhile patient safety intervention and may have distinct advantages over paper-based systems.9 At Temple University Hospital, a traditional paperbased incident report has been the primary event reporting process. Event report forms, available at nurses’ stations throughout the hospital, are intended for use by physicians, nurses, and other health care professionals to fulfill mandatory internal reporting requirements. Staff are also required to report events that have caused (or have significant potential to cause) patient injury within 24 hours by telephone to risk management. Risk managers are available 24 hours a day to all staff, although hospital policy prohibits such verbal disclosure from replacing the need for a physical incident report.3 Medical staff and housestaff are periodically reminded about their duty to report incidents, with particular emphasis on hospital and state policies, at hospital-wide forums held by risk management and legal counsel. In 2004, the hospital received 10,020 incident reports, of which only 96 (approximately 1%) came from physicians. The vast majority of incident reports come from nursing staff and other ancillary health professionals. In a recent revealing study, it appears that even patients may report more frequently than physicians. Hospitalized patients reported a number of adverse events and near misses, many of which were confirmed on chart review, but none of which had been disclosed by physicians in the hospital incident reporting system.10 Clearly, current efforts toward increasing physicians’ incident reporting and error disclosure may not be paying off.
July 2006
The exact reason why physician reporting is so poor remains unclear and strong data is lacking. Physicians have traditionally been reluctant to shed light on such issues as iatrogenic complications or system problems for fear that such information may lead to lawsuits or professional embarrassment—despite evidence that disclosure may actually decrease the likelihood of lawsuits.11 Attending physicians may be particularly reluctant to disclose, except when bad outcomes or clear-cut “protocol violations” are present.12 Yet barriers to reporting may simply reflect such issues as uncertainty about what to report, lack of time, or forgetting to report.13 In addition, many physicians may not define error disclosure as their primary responsibility.14 The problem may even start in training.15–17 In Fall 2004, as part of ongoing quality improvement (QI) activities, we sought to evaluate whether a simplified, user-friendly process for paper incident reporting could improve the frequency and quality of physician reporting. We also attempted to assess the clinical importance and impact on patients of events disclosed by physicians. Most importantly, we hoped to identify physician-specific barriers to incident reporting that might be amenable to further intervention.
Methods Disclose Tool As a QI initiative, the new pilot paper incident reporting form, the DISCLOSE (Drugs, Iatrogenic complications, System errors, Communication errors, Lab and test issues, Oversights, Staffing issues, and Equipment issues) tool, was developed by two of the co-investigators [D.S., D.M.] It was formatted into pads of paper that fit easily in the physician’s white coat (Appendix 1, page 392). Physicians were instructed to circle a category and if appropriate, a subcategory, for each event. In cases where no category was chosen but rather only a written description of the event was recorded, incidents were classified into one or more categories of error by data entry personnel. The DISCLOSE mnemonic prompted the user to look for some of the most commonly encountered errors in hospital care in the specified areas.
Participants All physicians who participated in this QI project were full-time academic physicians in Temple University
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
383
Table 1. Harm Scale Categories
Hospital and Temple University School of Medicine. As part of their academic duties, generalists and specialists serve as teaching physicians on general medical services in the internal medicine department. During a threemonth period (September–December 2004), these general medical service attending physicians were introduced to the use of this alternatively formatted incident reporting form at monthly departmental in-services. Physicians were instructed to incorporate the use of the tool into daily patient rounds with residents and students as seamlessly as possible. They were told to use the DISCLOSE tool as if they were submitting reports of medical errors or incidents that they or their housestaff encountered on the inpatient service. All participants were reminded throughout the three months, however, that the DISCLOSE tool was for our pilot project only and did not fulfill hospital mandatory reporting requirements via incident reports. We emphasized that traditional hospital incident reports still needed to be submitted as mandated by hospital protocol or state law and that submitting a DISCLOSE form did not adequately fulfill those reporting requirements.
patient) were immediately brought to the attention of risk management; all submitted reports are examined and then forwarded to appropriate hospital committees.
Instructions
Data Collection
The concept of DISCLOSE was briefly introduced at the in-service and in a departmental memorandum that was circulated before the start of the project. Formal attempts at more extensive instruction and education regarding patient safety, medical errors, definitions of what needed to be reported, and the tool’s specific categories were not undertaken to better assess true changes in reporting patterns from baseline. Participants were urged to provide detailed narrative information regarding the incident on both the form and all concurrent hospital incident reports. Although we instructed participating physicians to submit traditional incident reports whenever they submitted DISCLOSE forms, any events reported with the DISCLOSE tool that qualified as potential “sentinel events” (defined as an unexpected occurrence involving death or serious physical or psychological injury) or “serious events” (defined as an event, occurrence, or situation involving the clinical care of a patient in a medical facility that compromises patient safety and results in an unanticipated injury requiring the delivery of additional health services to the
At the conclusion of the three months, the number and categories of various incidents reported using the DISCLOSE reporting tool were compared with the data from the hospital’s existing paper incident reporting database for the same time period. The narrative portion of all reports was analyzed, and the incidents were recategorized if necessary to conform as much as possible to the categories listed in the DISCLOSE tool. Because the patient’s name and other identifying information was frequently not available on DISCLOSE submissions, the narrative descriptions of the events reported on DISCLOSE forms were compared against the hospital’s risk management database to determine if these events appeared to have been previously reported. At the conclusion of the three months, incidents reported via the DISCLOSE tool were also retrospectively classified into harm categories by one of the physician investigators [E.K.] and two other physicians who had not been involved in reporting during the project period. The five categories, recently validated by another group of investigators,18 are described in Table 1 (above).
384
July 2006
1.
An event occurred and resulted in death to the patient. 2A. An event occurred, patient was harmed, but harm did not result in death (temporary harm/complete recovery expected). 2B. An event occurred, patient was harmed, but harm did not result in death (permanent harm/complete recovery not expected). 3. An event occurred and reached the patient, but the patient was not harmed. 4. An event occurred but was caught before affecting/reaching the patient. 5. No event occurred, but a risky situation was found that increased the risk for an event to occur. N/A No event occurred, and no risky situation in place that increased risk for events.
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
Table 2. Summary of Errors Reported by Physicians, by Category with DISCLOSE Tool Category
Description Dosage, timing, route, rate, omitted doses, allergies, reactions, wrong Drugs drug, wrong patient Iatrogenic Complications from procedures, treatments System Issues Delays, missed treatments, order entry Communication Identification issues, confusion over orders, failure to give information Lost/mislabeled specimens/films, results not reported, delays, Lab/Tests improper studies Oversights Judgment issues, missed diagnoses, deviations from standard of care Staff Nursing, respiratory, consultants, transport Equipment Delays in equipment, wrong equipment, failures, supply problems Total
Number Proportion of Total 26
15%
6 25 24
3% 14% 14%
14
8%
15 59 7 176 *
9% 34% 4%
* The 98 discrete incident reports were categorized into 176 categorized incidents.
The three physicians reviewed the data twice, with cases in random order each time. Median percentages were calculated in each harm category, for individual raters, as well as an overall median percentage for all three raters. A statistician then calculated the inter- and intrarater variability based on an Analysis of Variance on ranks, followed by calculation of the intraclass coefficient, and confidence intervals.19 Because of restructuring in the risk management department, we were unable to access their data regarding the severity of incidents reported in the traditional incident reporting system for purposes of comparison. We surveyed physicians at the end of the data collection period regarding their reporting experience.
ple categories after review. Staff-related events were the most frequently reported category (which did not involve physicians), comprising 59 incidents (34%). Examples of incidents from all categories reported using the DISCLOSE tool are shown in Table 3 (page 386). Table 4 (page 387) breaks down reported events in further detail (subcategory) under each category of error in the DISCLOSE tool; only 10 of the common subcategories are shown. The most commonly reported subcategorized events were delays attributed to staff, accounting for 20 (11%) of all reports. Table 5 (page 388) contains some examples of DISCLOSE incidents in the staff category, and Table 6 (page 389) lists examples for the communication category.
Results Categories of Events
Harm of Events
Given that none of the DISCLOSE events could be definitively linked to a concurrent incident report by a physician or other healthcare worker, the errors identified by physicians using the DISCLOSE tool were presumed to have not been previously reported. The 39 attending physicians on general internal medical service from September 15, 2004 to December 15, 2004 reported 98 discrete incidents using the DISCLOSE tool. Thirteen physicians identified themselves and were responsible for 67 (69%) of the reports. Table 2 (above) summarizes the categories of errors reported, which totaled 176 because many errors fit under multi-
Retrospective application of the five-point harm scale to the DISCLOSE database indicated the following results: ■ 41% of events were felt to have reached patients but not caused harm ■ 33% of events were felt to have resulted in temporary harm ■ 9% of reports, though not considered events, were felt to indicate a “risky situation” ■ 2% of events were felt to have been caught before reaching the patient ■ 1.4% of events were felt to have contributed to permanent harm not resulting in death
July 2006
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
385
Table 3. Examples of Various Incidents Reported with DISCLOSE Tool for Each Category Harm Scale Rating* Drugs Patient given 9,000 units/hour of heparin instead of 900 units/hour. No bleeding resulted. 3 Central line thought to be in femoral vein discovered at time of removal one week later Iatrogenic 2A to be in femoral artery. No complications, patient tolerated removal. Took extended time to get “stat” abdominal radiograph to confirm nasogastric tube placement. Patient misses scheduled cardiac medications and despite intravenous Systems 2A administration of furosemide and antihypertensive agents, develops severe pulmonary edema, hypertensive emergency. Requires transfer to intensive care unit. Patient transferred from outside hospital to emergency department (ED), then hospital room. Transfer center notified ED staff, but neither ED staff nor transfer 2B Communications center notifies internal medicine housestaff. Patient, in room four hours without evaluation, is later found to have had sustained stroke and rhabdomyolysis (exact timing undetermined). Urinalysis on febrile patient sent, received in lab but no result for two days, and test then 2A Labs/Tests “auto-cancelled,” leading to treatment delay. Patient’s condition did not deteriorate. Patient with altered mental status falls twice. After first fall, fall precautions were 2A Oversights ordered by physician. Second fall results in minor injury. Staff Team waited three days for surgical consult. 5 Equipment Air mattress spontaneously deflates 1–2 times daily, causing significant patient discomfort. 2A Category
Description
* Shown in Table 1 (page 384).
■ 11.6% of reports were felt to have not actually constituted events or patient safety incidents. None of the events were felt to have directly resulted in death to a patient. Based on the results of the analysis of variance on ranks (ANOVA), the ICC as a measure of inter-rater variability was 0.8413 (95% confidence interval, 0.7432–0.8991). For intra-rater variability, the ICC was 0.8687 (95% confidence interval, 0.8165–0.9079).
Traditional Incident Reports For the total of 1,123 reports received via the traditional hospital incident reporting system received for the three-month study period, only 37 (3.3%) were submitted by physicians. Staff nurses (731, 65.1%), laboratory personnel (108, 9.6%), and pharmacy staff (54, 4.8%) were the top reporters by professional category. Reporters submitted in a total of 14 categories. Drug errors (13, 35% of physician reports) and delays (9, 24%) constituted the categories most frequently reported by physicians, who also cited iatrogenic complications (5, 14%), oversights (2, 5%), and equipment failure (2, 5%).
386
July 2006
Nurses and other nonphysician health care professionals submitted a total of 1,123 incident reports. Nurses submitted 731 incident reports, including patient falls (108, 15%), compromised skin integrity (104, 14%), oversights (96, 13%), iatrogenic events (91, 12%), and drug errors (80, 11%).
Experience of Reporting All 39 physicians who had attended on general internal medicine service were surveyed on their experiences regarding incident reporting at the conclusion of the three months. Twenty-five (64%) returned the questionnaires. More than half of the respondents (13; 52%) felt that the DISCLOSE tool was easier to use than the traditional incident report, and 10 (40%) found the DISCLOSE tool less threatening than existing reports. The most common reasons limiting incident reporting were time constraints (15, 60%), difficulty locating forms (13, 52%), lack of closure/feedback (8, 32%), and “did not find it important” (3, 12%). Only one physician (4%) cited “fear of disclosure to risk management” as a limiting factor.
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
Table 4. Summary of Most Common Subcategories of Error Reported by Physicians with DISCLOSE Tool
in addition to risk management, was less punitive and more likely to result in reform. Physicians, who Category Subcategory Number Proportion of Total frequently cite lack of time and lack Missed dose 4 2% of faith that any change will be Drugs Wrong dose 5 3% made based on their incident Wrong patient 3 2% reports, may not realize that the Iatrogenic Procedure 4 2% errors they feel are too minor to System Issues Delays 17 10% report that are often the best targets Communication Failure to give information 12 7% for systemwide or more focused Labs/Tests Delays 8 5% intervention.22 Oversights Judgment 8 5% The fact that the majority of the Staff Delays 20 11% DISCLOSE events that had either Equipment Supply 5 3% harmed or had the potential to harm our patients had presumably not Discussion been disclosed by any other means underscores the Physicians working on inpatient medical services are importance of our findings. Physicians who used the likely to have a unique perspective on patient care. They DISCLOSE tool seem much more apt to report all events, are often more privy to problems and errors than other from minor to life threatening. Increasing the pool of health care workers because of their primary role in iniincident reports is the first, though obviously not the tiating patient care plans, requesting procedures and only, step toward recognizing patterns of medical errors consultations, and their close working relationship with and identifying systems errors for intervention. Whether housestaff. This makes physicians a key but often adequate measures are in place once reporting signifiunderused source of incident reports. The challenge, cantly increases, and how to sustain initial increases in however, lies in finding a way to overcome traditional reporting, however, will be the most important factors in barriers to reporting such as time constraints, difficulty ensuring success.23 locating forms, fear of disclosure to risk management, Physicians using the DISCLOSE tool also tended to ignorance of reporting requirements, and dissatisfaction report on broader and distinct types of incidents than with lack of feedback or closure.20 Moreover, despite did other health care workers. Of particular interest is growing evidence that interventions to decrease medthe fact that they frequently reported errors by houseical error may be effective, many physicians are still staff in the staff category, even when housestaff were 21 skeptical. The importance of increasing physician inciexcluded from the category. In the communications catdent reporting, especially with mandatory institutional, egory, a number of reports highlighted physician frustrastate, and federal reporting on the rise, cannot be tion at orders not being filled properly or at not being overemphasized. notified of unexpected delays. Because of these findings, The finding that none of the 98 incidents reported via in subsequent forums, we have explicitly instructed the DISCLOSE tool could be linked to reports filed using physicians to address all staff-related incidents to approthe traditional reporting mechanism suggests that they priate supervisors or department heads in addition to represent new disclosure of events that would not othsubmitting incident reports. erwise have been uncovered. We believe that the Meanwhile, the 37 incident reports submitted by increase in reporting arises largely from the simplifiphysicians via the hospital’s incident reporting syscation of the error reporting process, reflecting the tem were significantly outnumbered by the number DISCLOSE tool’s portability, with minimal disruption of of reports from nonphysician healthcare workers. the patient care routine. We also believe that physicians Nurses, in particular, reported nearly 20-fold greater than physicians during the three-month study period, a perceived that a form going to our physician investigators,
July 2006
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
387
Table 5. Examples of DISCLOSE Incident Reports in Staff Category* Description
Other Categories
Intern substituted 4 mg hydromorphone IV for 4 mg PO. Patient difficult to arouse. Attending physician never called to notify of patient intubated and transferred to ICU. Patient has chest tube placed for pneumothorax by emergency department staff. When admitting housestaff arrive, chest tube noted to be disconnected (timing uncertain) and patient has recurrent pneumothorax. Strict In’s and Out’s not recorded; patient given “bathroom privileges” with diagnosis of congestive heart failure. IV nurse fails to replace IV after pulled out by patient, causing missed drug dose. No medications sent from pharmacy four hours after ordered. Not notified by nursing of systolic blood pressure 240. 24-hour urine collection ordered, not started by nurse until 3 days later.
Drug—wrong dose Oversight—judgment Communication Oversight—patient Observation Communication— confusion over ordering System—missed treatment System—delay Oversight—judgment Lab/Test—delay
Harm Scale Rating† 2A N/A 2A 5 2A 5 5 N/A
* IV, intravenous; PO, orally; ICU, intensive care unit. †
Shown in Table 1 (page 384).
rate that has been stable for some time. They tended to report more on protocol-driven issues such as falls, skin integrity, and patients who signed out against medical advice than physicians. This dramatic difference in reporting behavior illustrates a fundamental difference between physicians’ and nurses’ attitudes and perhaps the expectations of their supervisors and peers.12 In a review of nursing incident report data, 88% of surveyed nurses thought it was “important to report all incidents” and 72% stated that they “always fill out incident reports,” even though 17% believed that doing so would be used against them in an evaluation.24 It would be encouraging if more physicians embraced this dedication to reporting, especially if they continued to report on issues that are visible only from their unique vantage point. Other institutions have likewise reported on their efforts to address incident reporting. For example, Nast and colleagues introduced a simplified patient safety event reporting tool in a cardiothoracic surgery intensive care unit. Of the 163 reports submitted over six months, at least 69% were from nurses and only 19% were from physicians. The majority (54%) of events caused no harm despite “reaching the patient.” The remainder ranged from “risky situations” (15%), all of which were reported by nurses, to near misses (10%), to
388
July 2006
temporary harm (22%). No events reported caused permanent harm or death. These results are largely in concert with our findings that many events either reach patients or have significant potential for harm. Many different harm scales exist, but perhaps the most clinically useful are those that stratify not only injury but the sometimes intangible potential to predispose to injury. Some may contend that the potential to cause harm is in the eye of the beholder, but we believe that the more frequently this potential is investigated, the safer a system will become. Perhaps this project’s most telling findings lie in the participating physicians’ comments themselves. Most of those surveyed had never previously used the traditional incident reporting mechanism. Their reasons included time constraints, difficulty locating forms, lack of closure/follow-up, and “did not find it important.” Among those who submitted DISCLOSE forms, most found the new tool easier to use. Asked why, one physician commented, “Because I had the forms. I knew exactly where they were.” She also appreciated that the new tool enabled her to report events that “did not necessarily lead to adverse health or clinical outcomes… I do not call risk management unless the patient suffered or deteriorated…” Thirty-two percent of our surveyed physicians also cited lack of feedback
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
Table 6. Examples of DISCLOSE Incident Reports in Communication Category* Description Initial plan to discharge patient, order placed. Based on study result, plan changed and patient was to stay for further workup. Despite order being cancelled, nurse continues discharge protocol and patient sent home. Discharge medication prescribed at wrong dose contributing to readmission. Patient with new diagnosis of diabetes, discharged with insulin prescription but neither prescription nor discharge instruction has dose, causing patient confusion. Patient admitted to hospital room at 7 P.M. No housestaff notified, and patient not assessed until 9:30 A.M. the next day. Attending physician given verbal report that transesophageal echocardiogram was “negative.” Official dictated report mentions cardiac thrombus, without any follow-up call to medicine attending by cardiologist. Stress test delayed, cancelled. Team not notified, no reason given by department. Patient kept NPO* for prolonged period.
Other Harm Scale Categories Rating†
None
3
Drugs— Wrong dose
2A
Drugs— Erroneous prescription Oversight
3
Oversight
3
Staff— Consultants
3
System— Delay
3
* NPO, nothing by mouth. †
Shown in Table 1 (page 384).
from risk management as a deterrent to active error reporting. Where physician reports go and how they are handled may be an important component of designing future effective strategies to improve physician reporting.
July 2006
The project had some significant limitations. First, the study period was limited to three months and to one institution. We wanted to quickly assess the feasibility and benefits of a simple QI intervention and then adopt it soon after. Although we met our goals, the brevity of our project may limit extrapolation to larger forums. Second, given the lack of adequate patient identifiers and event-specific data (that is, date, time, place) on DISCLOSE submissions, the “matching” process and comparison with the risk management incident reporting database may not appear ideal. However, given the reasonable number of physician reports in both databases, we were able to review all reports individually rather than relying on merely querying a large database. On the basis of the narrative descriptions, we felt confident that all DISCLOSE-submitted incidents were in fact distinct, newly reported events. If someone other than a physician filled out an incident report, however, this may not have been captured in our review because of the high volume of nonphysician reports. The difficulty we encountered in obtaining highly specific event data was mirrored in the high number of submissions from a relatively small subset (13 out of 39) physicians, the remainder of which chose to remain anonymous. It is not surprising, given the baseline institutional milieu, that these obstacles emerged. In fact, much of our current effort is focused on emphasizing to the medical staff that incident reports are a key internal reporting mechanism that require not only physician self-identification but also patient and event-specific information to fulfill our current institutional and state disclosure and reporting policies. In the future, we expect that our educational efforts will significantly improve the quality and consistency of reporting data. Third, the retrospective application of our harm scale to assess the impact of the occurrences on our patients was limited by the level of detail in submitted reports. Previous studies allowed reporters to designate the suspected level of risk on their form in a prospective fashion, which would certainly reduce recall bias.18 We were also not always able to discern the specific outcome and level of harm, making this analysis imperfect at times. However, as shown, the reviewing physicians, two of whom lacked specific training in patient safety activities, were able to classify the events with some
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
389
degree of consistency. As the culture of patient safety in our institution grows, we hope that reporters will include much more event-specific data, especially regarding outcomes. In the future, we plan to prospectively capture and apply our harm scale, with medical record review as needed for clarification, in order to yield even more useful data.
Postscript Following the data collection for the study, we presented an interim progress report in a hospitalwide forum open to medical staff, residents, and medical students. The positive comments and questions we received suggest widespread interest in both improving the current incident reporting system and developing effective interventions to improve patient safety. In fact, system errors and discrete incidents uncovered by the DISCLOSE tool have already served as an impetus for improved medical education in our residency program in peer-protected environments. In the short term, a number of the events uncovered by the DISCLOSE tool have already been used in patient safety and morbidity and mortality conferences to further encourage open discussion on medical error. We are in the process of implementing a new patient safety curriculum for our residents to introduce basic patient safety and medical error terminology, root cause analysis, and open discussion of cases of error uncovered by the DISCLOSE tool in small groups as well as in a revised, nonpunitive, departmentwide morbidity and mortality conference. With growing emphasis on identifying and attacking areas of preventable error, such as the Institute for Healthcare Improvement’s 100K Lives Campaign,25 the importance of equipping physicians with tools such as DISCLOSE to uncover and facilitate open discussion about medical errors cannot be overemphasized. In the near future, the health system’s new computer-based reporting system, which will likely eventually replace our paper-based system, is set to be unveiled, and its potential to facilitate both reporting and dissemination of event data and to expedite processing
390
July 2006
and enable quick intervention seems promising. Recent successful efforts at “going paperless” by implementing electronic and Web-based incident reporting have been published.26,27 We are confident that the use of a more user-friendly error reporting tool, whether it is ultimately paper or computer-based, combined with the increased awareness of the importance and duty to disclose errors, will sustain our efforts.
Conclusion Project DISCLOSE’s findings suggest that strategies can be successfully established to increase physician incident reporting. Simple reporting mechanisms such as the DISCLOSE tool are likely to lead to improved physician reporting of clinically significant incidents with potential for major patient harm and increased satisfaction with the process. J The authors gratefully acknowledge the help of Raymond Gathiuni, M.D., Kathleen Coppola, M.D., Mary Kraemer, M.D., Rebecca Horvath, M.A.S., J.D., and Marie Swety, R.N., J.D. Dr. Gathiuni helped with the initial literature review. Drs. Coppola and Kraemer generously devoted time to reviewing the data and applying the five-point harm scale. Ms. Horvath helped plan the initial study. Ms. Swety was instrumental in coordinating the review of risk management data.
Emmanuel S. King, M.D., is Assistant Professor of Medicine and Assistant Program Director for Patient Safety Education in Internal Medicine Residency, Department of Internal Medicine, Section of Hospital Medicine, Temple University Hospital, Temple University School of Medicine, Philadelphia. Darilyn V. Moyer, M.D., is Associate Professor of Medicine and Program Director, Internal Medicine Residency, and Michael J. Couturie, M.D., is a Resident, Department of Internal Medicine. John P. Gaughan, Ph.D., is Associate Professor, Epidemiology and Biostatistics, Temple University School of Medicine. David J. Shulkin, M.D., formerly Chief Medical Officer and Patient Safety Officer, Temple University Hospital, is President and Chief Executive Officer, Beth Israel Medical Center, New York City, and a member of the Joint Commission Journal on Quality and Patient Safety’s Editorial Advisory Board. Please address reprint requests to David J. Shulkin, M.D., [email protected].
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
References 1. Kohn L.T., Corrigan J.M., Donaldson M.S.: To Err Is Human: Building a Safer Health Care System. Washington, D.C.: National Academy Press, 1999. 2. Longo D.R., et al.: The long road to patient safety: A status report on patient safety systems. JAMA 294:2858–2865, Dec. 14, 2005. Erratum in: JAMA 295:164, Jan. 11, 2006. 3. Temple University Hospital Administrative Policies and Procedures: Patient Safety Plan (internal document). No. 950.140, effective Dec. 2001. 4. Shojania K.G., Wald H., Gross R.: Understanding medical error and improving patient safety in the inpatient setting. Med Clin N Am 86:847–867, Jul. 2002. 5. Bagian J.P., et al.: Developing and deploying a patient safety program in a large health care delivery system: You can’t fix what you don’t know about. Jt Comm J Qual Improv 27:522–532, Oct. 2001. 6. Milch C.E., et al.: Voluntary electronic reporting of medical errors and adverse events: An analysis of 92,547 reports from 26 acute care hospitals. J Gen Intern Med 21:165–170, Feb. 2006. Epub Dec. 2005. 7. Pennsylvania Patient Safety Authority. http://www.psa.state.pa.us (last accessed May 3, 2006). 8. Medical Care Availability and Reduction of Error (MCARE) Act: P.L.154, No. 13. Mar. 20, 2002. 9. Tighe C.M., et al.: Incident reporting in one UK accident and emergency department. Accid Emerg Nurs 14:27-37, Jan. 2006. 10. Weingart S.N., et al.: What can hospitalized patients tell us about adverse events? Learning from patient-reported incidents. J Gen Intern Med 20:830–836, Sep. 2005. 11. Guadognino C.: Effects of medical error disclosure and apology [interview with Albert Wu, M.D., M.P.H.]. Physician’s News Digest, Feb. 2005. http://www.physiciansnews.com/spotlight/205.html (last accessed May 3, 2006). 12. Lawton R., Parker D.: Barriers to incident reporting in a healthcare system. Qual Saf Health Care 11:15–18, Mar. 2002.
July 2006
13. Nast P.A., et al.: Reporting and classification of patient safety events in a cardiothoracic intensive care unit and cardiothoracic postoperative care unit. J Thorac Cardiovasc Surg 130:1137, Oct. 2005. 14. Robinson A.R., et al.: Physician and public opinions on quality of health care and the problem of medical errors. Arch Intern Med 162:2186–2190, Oct. 28, 2002. 15. Wu A., et al.: Do house officers learn from their mistakes? JAMA 265:2089–2094, Apr. 24, 1991. 16. Aron D.C., Hendrick L.A.: Educating physicians prepared to improve care and safety is no accident: It requires a systematic approach. Qual Saf Health Care 11:168–173, Jun. 2002. 17. Waring J.J.: A qualitative study of the intra-hospital variations in incident reporting. Int J Qual Health Care 16:347–352, Oct. 2004. 18. Osmon S., et al.: Reporting of medical errors: An intensive care unit experience. Crit Care Med 32:727–733, Mar. 2004. 19. Shrout P.E., Fleiss J.I.: Intraclass Correlations: Uses in assessing rater reliability. Psychol Bull 86:420–428, Mar. 1979. 20. Hebert P.C., Levin A.V., Robertson G.: Bioethics for Clinicians: 23. Disclosure of medical error. CMAJ 164:509–513, Feb. 20, 2001. 21. Rosen A.B., et al.: Physicians’ views of interventions to reduce medical errors: Does evidence of effectiveness matter? Acad Med 80:189–192, Feb. 2005. 22. Mills P.D., et al.: Using aggregate root cause analysis to reduce falls. Jt Comm J Qual Patient Saf 31:21–31, Jan. 2005. 23. Brennan T.A., et al.: Accidental deaths, saved lives, and improved quality. N Engl J Med 353:1405–1409, Sep. 29, 2005. 24. Elnitsky C., Nichols B., Palmer K.: Are Hospital Incidents Being Reported? J Nurs Adm 27:40–46, Nov. 1997. 25. Institute for Healthcare Improvement: 100K Lives Campaign. http://www.ihi.org/IHI/Programs/campaign (last accessed May 3, 2006). 26. Dixon J.F.: Going paperless with custom-built Web-based patient occurrence reporting. Jt Comm J Qual Improv 28:387–395, Jul. 2002. 27. Kivlahan C., et al.: Developing a comprehensive electronic adverse event reporting system in an academic medical center. Jt Comm J Qual Improv 28:583–594, Nov. 2002.
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations
391
Appendix 1. DISCLOSE Incident Report
392
July 2006
Volume 32 Number 7
Copyright 2006 Joint Commission on Accreditation of Healthcare Organizations